Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Outcomes of antibiotic use among children with acute respiratory tract infections
(USC Thesis Other)
Outcomes of antibiotic use among children with acute respiratory tract infections
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Outcomes of antibiotic use among children with acute respiratory tract
infections
Kinpritma Sangha
Doctor of Philosophy
(HEALTH ECONOMICS)
Faculty of the USC Graduate School
University of Southern California
August 2019
2
Table of Contents
Page
Chapter 1: Introduction 3
BACKGROUND 3
OBJECTIVES OF THE DISSERTATION RESEARCH 8
Chapter 2: Methods 14
DATA 14
OUTCOMES 17
STATISTICAL METHODS 17
KEY EXPLANATORY VARIABLES 18
Chapter 3: The Impact of Time to Treatment 21
INTRODUCTION 21
METHODS 23
OUTCOMES 27
RESULTS 28
DISCUSSION 32
LIMITATIONS 34
Chapter 4: The Impact of Time to Treatment and Antibiotic Class 46
INTRODUCTION 46
METHODS 47
OUTCOMES 51
RESULTS 52
DISCUSSION 56
LIMITATIONS 58
Chapter 5: The Impact of Adherence to HEDIS Guidelines 71
INTRODUCTION 71
METHODS 73
OUTCOMES 76
RESULTS 77
DISCUSSION 81
LIMITATIONS 83
Chapter 6: Discussion 97
CONCLUSION 98
3
Chapter 1: Introduction
Background
Prescribing antibiotics in an outpatient setting is often a complex
interaction between patients and physicians. Antibiotic prescribing is primarily
contingent upon the pathogen causing the infection. However, patient beliefs and
desires, clinician preferences to treat with an antibiotic, judgments about what is
medically necessary and what is needed to assure patient satisfaction often
impact the antibiotic treatment decision. The role of these non-clinical factors is
magnified in pediatrics where parents act as medical proxies for their children.
Studies have shown that physicians may prescribe antibiotics in children simply
to allay the parents’ fears and unease regarding the infection, or to satisfy their
desire for expedient treatment.
1,2
Physicians may also prescribe antibiotics out of
habit as primary treatment for infections outside of clinical guidelines to avert
patient dissatisfaction.
2,3
Treatment with an antibiotic confers no health benefit for infections of viral
etiology.
4
Precise diagnostic tests are required to differentiate between viral and
bacterial infections; however, the capacity to test for specific bacterial infections
and subsequently guide antibiotic prescribing is limited. Most definitive diagnostic
tests require 2-3 days to culture. In addition, the tests may not provide clear
guidance on which specific antibiotic is appropriate, incur additional costs and
may delay time to diagnosis and treatment. Not surprisingly, diagnostic tests are
not employed routinely in outpatient settings. Clinical practice depends almost
entirely on the clinical judgment of the physician. For example, in children with
selected clinical features or in severe cases, clinical guidelines recommend
antibiotics for treatment of acute respiratory tract infections (ARTIs) without
laboratory confirmation of the infection’s etiology.
5,6,7,8,9
As a result, antibiotics
may be over-prescribed in the outpatient setting because they are inexpensive,
carry few side-effects, and have minimal short-term consequences for the
individual patient when misused.
4
It is widely accepted that over-prescribing of antibiotics are common for
outpatients with acute respiratory tract infections (ARTIs). Over-prescribing
creates several public health problems. Over-prescribing is thought to contribute
to the development of antibiotic-resistance and causes unnecessary use of
antibiotics and medical resources.
10
A long-term consequence of such use of
antibiotics may reduce the overall effectiveness of antibiotic therapies.
11
Antibiotic over-prescribing may lead to antibiotic resistant organisms which infect
at least two million people annually, cause at least 23,000 deaths, and result in
$20 billion in excess direct healthcare costs in the US each year.
12,13
On the
other hand, the underuse of antibiotics for treating outpatients with bacterial
respiratory infections delays recovery and may increase the use of additional
physician visits. Underuse may also be a result of uncertain diagnosis, cost
concerns and antibiotic resistance concerns.
38
Another study found that 14% of
pediatric patients with pneumonia and 46.2% of patients with ARTI requiring
antibiotics did not receive the antibiotic treatment.
39
A 2016 CDC report states
that underuse of antibiotics should be considered in conditions for which
antibiotics are underused or the need for timely antibiotics is not recognized.
40
Antibiotics are among the most commonly prescribed medications for
children.
14,15,16
Nearly 21% of ambulatory pediatric visits lead to antibiotic
prescription.
16,17
In pediatric practices, prescription rates also vary by infections,
whereby 60-80% of the upper respiratory infection visits receive an antibiotic
treatment.
18,19,20
However, national antibiotic prescription data shows that
outpatient antibiotic prescription rates decreased by 4% between 2011 and
2015—from 877 prescriptions per 1000 population in 2011 to 838 prescriptions
per 1000 population in 2015. This decline is largely driven by reduction in
antibiotic prescriptions in children while the rate of antibiotic prescriptions among
adults remains stable. The antibiotic prescription rates in children decreased by
13% from 2011 to 2015.
21
Antibiotic stewardship programs [ASPs] have been created to guide
proper use and thus reduce overuse of antibiotics. ASPs are necessary because
use of antibiotics is associated with resistance both at the population and the
5
individual level. Antibiotic resistance patterns have been shown to change with
shifting antibiotic consumption patterns in the population.
22,23
Further, at the
individual level, people with recent antibiotic exposure display twice the likelihood
of antibiotic resistant bacteria than those who have not been recently
exposed.
24,25
Antimicrobial stewardship programs systematically implement and
support evidence based prescribing education for providers to curb antibiotic
overuse and also antimicrobial resistance.
Multiple other strategies have also been implemented over time to reduce
inappropriate use of antibiotics while recognizing the need for starting treatment
in a timely manner. One such strategy is to provide a prescription for an antibiotic
medication, but with guidance for the family on whether to fill the prescription
immediately or delay the fill.
26,27
The physician may decide to delay prescription
for 2-3 days based on child’s age, infection severity, diagnosis uncertainty and
the effectiveness of antibiotic for the ARTI.
28
The latter commonly requires that
the child’s parents evaluate if the infection has resolved spontaneously over time
or needs antibiotic treatment. A systematic review examining the effects of
immediate and delayed prescriptions shows mixed findings about the use of
antibiotic timing strategies in treating acute respiratory tract infections.
27
Therefore, while adequate antibiotics are necessary to treat bacterial ARTIs, the
timing of medication receipt is also important. Examining the antibiotic treatment
time, using real world data, can provide broader understanding of the appropriate
time of receipt to ensure proper antibiotic treatment for ARTIs and improve
patient outcomes.
Another strategy to curb promote adequate antibiotic use is to decrease
use of broad-spectrum antibiotics to treat infections for which narrow-spectrum
antibiotics are recommended. The overuse of broad-spectrum antibiotics
contributes to antibiotic-resistance. Antibiotic agents that act against a specific
set of bacteria are classified as narrow spectrum, while antibiotics that act
against a broad range of bacteria (gram-positive and gram-negative) are
classified as broad-spectrum antibiotics. Narrow-spectrum antibiotics can treat an
infection and target bacteria that have been identified.
8
Broad-spectrum
6
antibiotics are known to contribute to antibiotic resistance and higher rates of
adverse events and have not been associated with better clinical outcomes.
29
Finally, examining the real-world treatment patterns for pharyngitis may
provide insight into the appropriate role of testing, treatment and the use of
antibiotics. Unlike other acute respiratory tract infections, a diagnostic test exists
for pharyngitis that allows identification of group A beta-hemolytic streptococcus
(GABHS)—which accounts for one in four children with acute sore throat.
30
Once identified, the bacterial infection can be properly treated with specific
antibiotics. Further, the HEDIS measure for quality of care for pharyngitis
recommends the application of group A streptococcus test for children age 3
months to 18 years before dispensing an antibiotic.
31
While HEDIS measures
establish the recommendations for testing and treating for pharyngitis, the
outcomes of following these guidelines have not been studied. Particularly, there
is a lack of evidence in the effects of following the HEDIS measure of “test and
treat” on revisits in children with pharyngitis.
Antibiotic stewardship programs (ASPs) and numerous strategies to curb
antibiotic use are associated with a reduction in pediatric antibiotic prescription in
the last decade.
32
Many public health advocates argue that the variation in
treatment patterns by geography, medical specialty, race, age, etc. are evidence
that a continued need to improve antibiotic prescribing is needed to curb the
development of antibiotic-resistant bacteria.
33,34
However, better data are
needed on antibiotic use rates and on the patient outcomes associated with the
use of antibiotic before society moves forward to further restrict antibiotic
prescribing. The pediatric literature accounting for the patient outcomes
achieved by using antibiotics to treat acute respiratory tract infections is
particularly sparse.
Outpatient ARTIs rarely lead to hospitalizations or the use of other
extensive health care resources. This makes an analysis of the impact of drug
therapy on patient outcomes difficult. One measure of treatment success or
failure is the likelihood that the patient will revisit the physician. Revisit rates will
also vary by infection and age groups. ARTIs may not be defined similarly across
7
all research studies. While some studies include lower and upper respiratory
infections, others have focused on one or more infections, yet others may include
different diagnoses under respiratory infections.
Nonetheless, a limited number of
database studies evaluate revisits after antibiotics are prescribed for ARTIs in
pediatric outpatient settings. Li, et al. found that pediatric and adult patients
treated with antibiotics for acute otitis media (AOM), upper respiratory infections
(URI), and pharyngitis were more likely to return with a respiratory tract infection
within 30 days than patients who did not receive antibiotics.
3
In another study, a
chart review of children between 2 to 17 years of age with outpatient acute
respiratory infection (ARI) with a diagnosis of AOM, pharyngitis, influenza,
influenza like illness during the influenza season. Influenza and influenza-like
illness has higher rates of treatment and return rates than other respiratory
infections, especially during the flu season. This study found that 27% had follow-
up visits in one of two outpatient clinics in a city related to acute respiratory
infection within 30 days, while of the children who were not prescribed an
antibiotic, 17% had follow-up visits related to acute respiratory infection within 30
days.
35
However, contrasting this with studies that include a narrow selection
criteria based on infection and age. An Israeli study focusing specifically on AOM
pediatric patients found recurring AOM episodes at a range of 2.8-6.5% among
children age 6-36 months visiting a pediatric emergency department.
41
Another
study of pediatric otitis media patients compared return visits after being seen in
physician offices and retail clinics. These studies found that pediatric acute otitis
media patients visiting primary care offices were more likely to return than those
using retail medicine.
36
Other research has found antibiotic prescribing for
respiratory illness to be associated with small reductions in return visits, namely
for bronchitis and sinusitis, but not for patients with other respiratory illnesses
such as AOM, pharyngitis and upper respiratory infection.
37
Thus, the evidence
remains sparse on the risk of revisits after antibiotic use for ARTIs in children.
8
Objectives of the Dissertation Research
This dissertation research has three main objectives which will be
investigated in analyses of pediatric ARTI patients, a subset of ARTI patients with
acute otitis media [AOM] and pediatric patients with pharyngitis. These
objectives are:
1. Document treatment rates and the timing of antibiotic treatment.
Treatment is defined as whether or not an antibiotic prescription was filled
during the 28-day period following the index visit. This is followed by an
analysis that employs a precise definition of treatment that takes into
account the timing of treatment based on the day of antibiotic was filled
measured as the time between the index ARTI visit and the antibiotic fill
date. The delay in filling the prescription cannot be verified as
implemented by physician recommendation or patient preference. The
decision to delay an antibiotic can be recommended by the physician as
part of the watchful waiting strategy or it can occur simply because the
parent/child are unable to fill the prescription immediately.
2. Examine the factors that correlate with the antibiotic prescription decision
and the use of strategies used to improve the appropriateness of the
treatment decision. These strategies include tailoring the time to
treatment, the use of broad-spectrum antibiotics, and employing Group A
strep diagnostic test when appropriate.
3. Examine the effects of antibiotic treatment patterns on patient outcomes
as measured by revisits for pediatric ARTIs in outpatient settings.
The dissertation consists of five chapters that investigate issues pertaining to
antibiotic treatment patterns and revisits for ARTIs in children:
! Chapter 2 presents the research methodology used in this dissertation. It
also details the data source, research approach and the analyses used to
answer the three questions that comprise this dissertation.
! Chapter 3 presents treatment patterns for pediatric patients with ARTIs
and examines time to the first fill for an antibiotic and the impact of filling
9
an antibiotic within 3 days on the likelihood of a revisit. Using a logistic
regression model to control for clinical and socio-demographic factors
associated with treatment and with revisits to the outpatient provider within
28 days. Sensitivity analyses will be conducted to examine the timeframe
of a revisit. Revisit will be defined from 28 days to 7 days in an attempt to
tighten the temporal relationship of the revisit to the index visit.
! Chapter 4 focuses solely on children with Acute Otitis Media [AOM]. A
multivariable regression model is used to identify demographic and clinical
factors associated with time to treatment following diagnosis of AOM. A
similar model will then examine the class of antibiotic used and the impact
of filling an antibiotic for AOM within 3 days on revisit risk. Sensitivity
analyses will be conducted to examine revisit within 7 days.
! Chapter 5 reports results from analyses designed to evaluate the
diagnosis and treatment of pharyngitis in a manner consistent with the
Healthcare Effectiveness Data and Information Set (HEDIS) performance
measures for pharyngitis. A logistic regression models is used to estimate
the factors associated with a child being diagnosed and treated according
to HEDIS guidelines and the effect of guideline adherence on likelihood
that the child experienced a revisit within 28 days.
! Chapter 6 presents the conclusions of the research. In addition, this
chapter will discuss the need for further study and additional ways to build
upon this dissertation.
10
References
1. Lucas, Patricia J., et al. A systematic review of parent and clinician views and
perceptions that influence prescribing decisions in relation to acute childhood
infections in primary care. Scandinavian Journal of primary health care. 2015;
33(1): 11-20.
2. Mangione-Smith R, McGlynn EA, Elliott MN, McDonald L, Franz CE, Kravitz
RL. Parent Expectations for Antibiotics, Physician-Parent Communication, and
Satisfaction. Archives Pediatric Adolescent Medicine. 2001;155(7):800–806
3. Li, John, et al. Antimicrobial prescribing for upper respiratory infections and its
effect on return visits. Family Medicine. 2009; 41(3): 182-7.
4. Centers for Disease Control and Prevention. Get smart: know when antibiotics
work. http://www.cdc.gov/getsmart. Accessed Jan 12, 2018.
5. Shulman ST, Bisno AL, Clegg HW, et al. Clinical practice guideline for the
diagnosis and management of group A streptococcal pharyngitis: 2012 update by
the Infectious Diseases Society of America [published correction appears in Clin
Infect Dis. 2014;58(10):1496]. Clinical Infectious Diseases. 2012;55 (10):1279-
1282.
6. Lieberthal AS, Carroll AE, Chonmaitree T, et al. The diagnosis and
management of acute otitis media [published correction appears in Pediatrics.
2014;133(2):346]. Pediatrics. 2013;131(3):e964-e999.
7. Chow AW, Benninger MS, Brook I, et al; Infectious Diseases Society of
America. IDSA clinical practice guideline for acute bacterial rhinosinusitis in
children and adults. Clinical Infectious Diseases. 2012; 54(8): e72-e112.
8. Melander RJ, Zurawski DV, Melander C. Narrow-spectrum antibacterial
agents. Medicinal Chemistry Communications. 2018; 9(1): 12-21.
9. Rosenfeld RM, Piccirillo JF, Chandrasekhar SS, et al. Clinical practice
guideline (update): adult sinusitis. Otolaryngoly Head and Neck Surgery.
2015;152(2) (suppl):S1-S39.
10. Davies J, Davies D. Origins and evolution of antibiotic resistance.
Microbiology and Molecular Biology Reviews. 2010; 74(3): 417-433.
11. Gonzalez R, Malone DC, Maselli JH, Sande MA. Excessive Antibiotic Use for
Acute Respiratory Infections in the United States. Clinical Infectious Diseases.
2001; 33:757–62.
11
12. Centers for Disease Control and Prevention. Antibiotic resistance threats in
the United States, 2013. US Department of Health and Human Services, CDC,
2013. https://www.cdc.gov/drugresistance/pdf/ ar-threats-2013-508.pdf.
13. King L, Fleming-Dutra KE, Hicks LA. Advances in optimizing the prescriptions
of antibiotics in outpatient settings. British Medical Journal. 2018; 363: k3047.
14. Nash DR, Harman J, Wald ER, et al. Antibiotic prescribing by primary care
physicians for children with upper respiratory tract infections. Archives of
Pediatric and Adolescent Medicine. 2002; 156:1114–1119.
15. Hersh AL, Shapiro DJ, Pavia AT, et al. Antibiotic prescribing in ambulatory
pediatrics in the United States. Pediatrics. 2011;128:1053–1061.
16. Hersh AL, Jackson MA, Hicks LA and the Committee on Infectious Diseases.
Principles of Judicious Antibiotic Prescribing for Upper Respiratory Tract
Infections in Pediatrics. Pediatrics. 2013; 132 (6): 1146-1154.
17. Chai G, Governale L,McMahon AW, et al. Trends of outpatient prescription
drug utilization in US children, 2002–2010. Pediatrics. 2012;130:23–31.
18. Fleming-Dutra KE, Hersh AL, Shapiro DJ, et al. Prevalence of Inappropriate
Antibiotic Prescriptions Among US Ambulatory Care Visits, 2010-2011. The
Journal of American Medical Association. 2016;315(17):1864–1873.
19. Finkelstein JA, Stille C, Nordin J, Davis R, Raebel MA, Roblin D, Go AS,
Smith D, Johnson CC, Kleinman K, Chan KA, Platt R. Reduction in Antibiotic Use
Among US Children, 1996−2000. Pediatrics. 2003; 112; 620.
20. Watson J, Wang L, Klima J et al. Healthcare Claims Data: An Underutilized
Tool for Pediatric Outpatient Antimicrobial Stewardship. Clinical Infectious
Diseases. 2017; 64(11): 1479–1485
21. Centers for Disease Control and Prevention. Antibiotic resistance patient
safety atlas 2017. US Department of Health and Human Services, CDC, 2017.
https://gis.cdc.gov/grasp/PSA/indexAU.htm.
22. Van de Sande-Bruinsma N, Grundmann H, Verloo D, et al. European
Antimicrobial Resistance Surveillance System Group European Surveillance of
Antimicrobial Consumption Project Group. Antimicrobial drug use and resistance
in Europe. Emerging Infectious Disease. 2008;14:1722-30.
23. Dingle KE, Didelot X, Quan TP, et al. Modernising Medical Microbiology
Informatics Group. Effects of control interventions on Clostridium difficile infection
in England: an observational study. Lancet Infectious Diseases. 2017;17:411-
421.
12
24. Blaettler L, Mertz D, Frei R, et al. Secular trend and risk factors for
antimicrobial resistance in Escherichia coli isolates in Switzerland 1997-2007.
Infection 2009; 37:534-539.
25. Costelloe C, Metcalfe C, Lovering A, Mant D, Hay AD. Effect of antibiotic
prescribing in primary care on antimicrobial resistance in individual patients:
systematic review and meta-analysis. British Medical Journal. 2010;340:c2096.
26. American Academy of Pediatrics, Subcommittee on Manage- ment of Acute
Otitis Media. Diagnosis and management of acute otitis media. Pediatrics.
2004;113:1451–1465
27. Spurling, Geoffrey KP, et al. Delayed antibiotics for respiratory infections.
Cochrane Database of Systematic Reviews. 2013; 4(4): CD004417-1.
28. Kutty, Narayanan. Treating children without antibiotics in primary healthcare.
Oman Medical Journal. 2011; 26.5: 303.
29. Gerber JS, Ross RK, Bryan M, et al. Association of Broad- vs Narrow-
Spectrum Antibiotics With Treatment Failure, Adverse Events, and Quality of Life
in Children With Acute Respiratory Tract Infections. The Journal of American
Medical Association. 2017;318(23):2325–2336.
30. Choby, Beth A. Diagnosis and treatment of streptococcal pharyngitis.
American Family Physician. 2009;79(5): 383-390.
31. NCQA. HEDIS 2016.
http://www.ncqa.org/HEDISQualityMeasurement/HEDISMeasures/HEDIS2016.a
spx. Accessed October 20, 2015.
32. Barlam, T. F., Cosgrove, S. E., Abbo, L. M., MacDougall, C., Schuetz, A. N.,
Septimus, E. J., & Hamilton, C. W. Implementing an antibiotic stewardship
program: guidelines by the Infectious Diseases Society of America and the
Society for Healthcare Epidemiology of America. Clinical Infectious Diseases.
2016: 62(10), e51-e77.
33. Klepser, M. E., Dobson, E. L., Pogue, J. M., Labreche, M. J., Adams, A. J.,
Gauthier, T. P. et al A call to action for outpatient antibiotic stewardship. Journal
of the American Pharmacists Association. 2017; 57(4): 457-463.
34. Reddy, Sujan C., Jesse T. Jacob, Jay B. Varkey, and Robert P. Gaynes.
Antibiotic use in US hospitals: quantification, quality measures and stewardship.
Expert review of anti-infective therapy. 2015;13 (7): 843-854.
13
35. Zhao SR, Griffin MR, Patterson BL, et al. Risk Factors for Outpatient Use of
Antibiotics in Children with Acute Respiratory Illnesses. Southern Medical
Journal. 2017;110(3):172–180.
36. Rohrer JE , Garrison GM , Angstman KB . Early return visits by pediatric
primary care patients with otitis media: a retail nurse practitioner clinic versus
standard medical office care . Quality Management in Health Care. 2012; 21 (1):
44 – 7
37. Pan, Q., Ornstein, S., Gross, A.J., Hueston, W.J., Jenkins, R.G., Mainous,
A.G., III, et al. Antibiotics and return visits for respiratory illness: A comparison of
pooled versus hierarchical statistical methods. American Journal of Medical
Science. 2000; 319: 360-365.
38. McKellar MR, Fendrick AM. Innovation of Novel Antibiotics: An Economic
Perspective. Clinical Infectious Diseases. 2014. 59(S3): S104-7.
39. Kornblith AE, Fahimi J, Kanzaria HK, Wang RC. Predictors for under-
prescribing antibiotics in children with respiratory infections requiring antibiotics.
American Journal of Emergency Medicine. 2018. 36: 218-225.
40. Sanchez GV, Fleming-Dutra KE, Roberts RM, Hicks LA. Core Elements of
Outpatient Antibiotic Stewardship. CDC MMWR Recommendation Report. 2016;
65 (#6)
41.Rothman S, Pitaro J, Hackett A, Kozer E, Gavriel H, Muallem-Kalmovich L,
Eviatar E, Marom T. Treatment of Acute Otitis Media in the Pediatric Emergency
Department. The Pediatric infectious Disease Journal. 2018. 37(6): 520-525.
14
CHAPTER 2: Methods
Data
Healthcare claims databases are used in health services research to
answer questions about health care utilization patterns and patient outcomes in
real-world clinical practice.
1,2
A real-world database is particularly relevant for the
treatment of pediatric ARTIs given the current scarcity of evidence regarding
antibiotic use and revisits. Even the fundamental question of the rate of antibiotic
use in children and its impact on the likelihood of an ARTI related revisit are
limited for pediatric ARTIs. Current research has used univariate factor analyses
to analyze antibiotic treatment patterns but rarely involve an analysis of the
outcomes associated with antibiotic use.
3,4,5
This dissertation uses multivariate
analyses of treatment patterns and patient outcomes of antibiotic therapy using
data from a claims database.
The data for this study were derived from Optum Insight Clinformatics
claims database. The Optum data include claims of commercially insured
patients with both medical and pharmacy benefits from 2011-2013. These
individuals are privately insured through a large health plan which either fully
insures or administers claims for the patients. Study patients will be required to
have Optum data for both medical and prescription drug coverage, thus allowing
evaluation of both their medical and pharmacy claims. Optum claims database
contains several datasets: medical, member, SES (socioeconomic status) and
pharmacy.
The medical dataset includes administrative claims for outpatient visits to
ambulatory clinics, physician offices, and urgent and emergency care facilities.
Diagnostic data within the medical dataset are based on International
Classification of Diseases, Ninth Revision, ICD-9 codes. CPT (current
procedural terminology) codes are included for any tests and procedures that
were received by the patient.
15
The Optum member dataset includes data for the type of insurance plan,
initiation and end date of member enrollment in the plan, US census based
geographic location of the member, and gender.
The SES dataset includes the socio-economic information about the
member. These indicators include race, household income and education level.
Race includes White, Black, Asian and Hispanics. Education is categorized as
less than 12th grade, high school diploma, less than Bachelors degree and
higher than a Bachelors degree. Income categories are advised by the federal
guidelines for a family of four.
6
A low-income family is defined as earning
<$40,000 – $50,000; middle-income family earns >$50,000 – <$75,000 and a
high-income family earns $75,000 and above. Household income in Optum is
obtained by using algorithms based on census block groups.
The pharmacy administrative claims dataset identifies the medications
received by the member. This dataset specifies the date of prescription filling,
whether it was a refill, NDC (National Drug Codes) and AHFSCLSS (American
Hospital Formulary Service Classification System) codes for the medication.
7
Unit of Observation
Acute respiratory tract infections (ARTIs) are short-term acute conditions
with treatment and follow-up typically lasting 7 to 28 days. Treatment with an
antibiotic is typically provided for 7 to 10 days and patients whose ARTI fails to
resolve with either antibiotic therapy or watchful waiting usually return within 28
days of their initial visit, if not sooner. The following ICD-9 codes identify patients
with ARTIs included in the study but do not distinguish between viral and
bacterial infections:
1) Acute otitis media (ICD-9: 381.0, 381.4, 382.0, 382.4, 382.9)
2) Pharyngitis (ICD-9: 462, 463, 034.0)
3) Bronchitis (ICD-9: 466, 490)
4) Sinusitis (ICD-9: 461)
5) Pneumonia (ICD-9: 481-483, 484.8, 485–487, 033.0, 033.9)
16
6) Cold/URI (ICD-9: 460, 464 [excluding 464.3], 465, 487 [excluding
487.0], 786.2)
An episode of care was defined around the date of the first ARTI visit (+/-
6 months), called an index visit, which is identified for every patient initiating an
outpatient ARTI visit at an ambulatory clinic, physician office, emergency and
urgent care facilities. The International Classification of Diseases, Ninth Revision
(ICD-9) codes were used to document diagnosis associated with these index
visits. To qualify as an index visit, the patient must have an ARTI diagnosis and
at least a six-month enrollment in a health insurance plan before and following
the index visit. When a patient had multiple outpatient visits with ARTI diagnosis
within the 5-day period prior to their initial ARTI visit, the earliest of these prior
visit is used as the index visit. This assumes that the visit was for a pre-existing
infection because children are most likely to return within 5 days when the
existing infection/symptoms worsen or remain unresolved.
All patients under age 18 diagnosed with an ARTI for the 36-month period
from January 1, 2011- Dec 31, 2013 were included in the study. The medical
paid claims file was linked with the pharmacy claims. A physician visit was linked
with an associated prescription, if any. Patient history of previous acute
respiratory tract infection visits is also recorded. Any ARTI visits that occurred 90
days prior to the index visit were documented as prior visits. The time to visit was
recorded for the first subsequent visit for an ARTI following the index visit and
used to define alternative definitions for revisits for both the treated and the
untreated groups.
Using the patient identifier, medical visits were linked to the member
details and the SES file to determine insurance and socio-demographic
characteristics of the patients.
Inclusion and Exclusion Criteria for ARTI Episode Observations
Eligible episodes of treatment for this analysis were initiated by children
less than 18 years of age, who were diagnosed with an ARTI and treated in an
17
outpatient setting selected for inclusion in the analysis. This study excludes
patient episodes with less than 6 months of continuous health insurance
enrollment before or after the index date.
Outcomes
The outcomes for analysis are whether or not the patient received an
antibiotic, time to treatment and time to any revisit following the initial ARTI visit.
A time to revisit variable was used to define a dichotomous variable for patients
with revisits in 7 and 28 days. Revisits for an ARTI diagnosis may indicate a
clinical failure or the preferences of the physician or patient for a follow-up visit.
Regardless, a revisit is correlated with the total resources used to treat an ARTI
patient. Revisit is defined using both a 7 day and 28 day definition to differentiate
between short term and longer term outcomes
Revisits are not the perfect outcome measure as the reason for a revisit is
difficult to determine using paid claims data. A revisit may be a follow-up visit
routinely recommended by a treating physician or due to a concerned
parent/patient suspecting worsening infection. Nonetheless, revisit is an obvious
indicator of health resource use that occurs as a function of ARTI treatment
practices. Return visits are defined as “early revisit” when occurring at least 7
days from the initial visit, and “late revisit” when occurring within 28 days of the
diagnosis visit. Only return visits with a recorded diagnosis for an ARTI are
considered as recurrent episodes.
Statistical Methods
Data on time to antibiotic fill is measured as the time between index visit
and the date of fill for an ARTI related antibiotic. To provide flexibility in treatment
timing, 2 classes of treated patients were investigated: a) filled antibiotics on day
0-3 of the index visit, and b) filled antibiotics on day 4-28 of the index visit. These
time dependent definitions of treatment were then compared to no treatment (i.e.
children who did not receive an antibiotic).
18
A multivariable logistic regression is used to identify demographic and
clinical factors associated with filling an antibiotic within 3 days To measure the
effect of time to treatment on revisits, multivariable logistic regression analyses
were used to estimate the likelihood that the child experienced a revisit within 28
days and to identify demographic and clinical factors associated with revisits
following antibiotic treatment. Sensitivity analysis was conducted with revisit
within 7 days, to measure the effect of treatment timing on revisit in 7 days.
All analyses were completed using STATA/IC 13 (StataCorp, College
Station, TX).
Key explanatory variables
Numerous demographic and clinical variables about each patient are
included in the study to account for observed factors that affect the antibiotic
treatment decision. Data on patient age, race, gender, household income,
geographic location, practice setting type, provider specialty and type of
insurance are available from the claims file. Practice setting includes outpatient
physician offices, health centers and clinics, outpatient hospitals, emergency
department, urgent care, and ambulatory care centers. Providers include primary
care physicians, specialty care physicians, and non-physician providers. Patient
household income is obtained directly from the Optum dataset, which they
estimated using algorithms based on census block groups. Income categories
were informed by the federal guidelines for a family of four. A low-income family
is defined as earning <$40,000 – <$50,000; middle-income family earns $50,000
– <$75,000 and a high-income family earns $75,000 and above.
6
We also defined seasons of the year to account for time periods during
which children are most likely to visit a physician for respiratory infections, which
typically are fall and winter.
5
The months of December through February
constitute winter, March through May are included as spring, June through
August are summer and September through November are fall. Patient insurance
type is obtained from the Optum data and categorized as HMO (Health
19
Maintenance Organization), EPO (Exclusive Provider Organization), POS (Point
of Service), and PPO (Preferred Provider Organization). Race is defined as
White, Black, Hispanic, and Asian. Patient location is obtained from Optum and
defined according to national regions of Northeast Central, Southeast Central,
Mid Atlantic, Mountain, New England, Pacific, South Atlantic, Northwest Central
and Southwest Central. Previous visits variable was defined as whether the child
had ARTI visits in 3 months prior to the index visit.
Up to 3 ARTI diagnoses may be recorded for any outpatient visit. The first
recorded ARTI diagnosis is assumed to be the primary diagnosis and the other
ARTI diagnoses are designated as co-morbidities and are included to examine
the effects of ARTI-related co-morbidities on the primary infection, propensity for
antibiotic treatment, and any effect on revisits. ARTI related secondary diagnoses
include acute otitis media, pharyngitis, sinusitis, bronchitis, pneumonia and URI.
Non-ARTI related co-morbidities were defined as the following categories: mental
disorders, nervous system disorders, respiratory symptoms, fever, injury and
poisoning, other or no diseases. Previous visits variable was defined as whether
the child had ARTI visits in 3 months prior to the index visit.
Antibiotic Classification
Antibiotics were grouped into classes based on classification codes from
the AHFSCLSS (American Hospital Formulary Service Classification System).
7
These classes include penicillins, macrolides, cephalosporins, tetracyclines,
sulfonamides, quinolones, lincomycin, carbapenems and beta-lactam antibiotics.
We also defined broad-spectrum antibiotics as broad spectrum penicillins,
second to fifth line cephalosprins, macrolides, carbapenems and beta-lactam
antibiotics. As mentioned earlier, penicillin and amoxicillin are the common
recommended treatment and alternatives include penicillin allergy including a
first-generation cephalosporin, clindamycin, clarithromycin (Biaxin), or
azithromycin (Zithromax)
References
1. Streigel-Moore RH, Leslie D, Petrill SA, Garvin V, Rosenheck RA. One-year
use and cost of inpatient and outpatient services among female and male
patients with an eating disorder: evidence from a national database of health
insurance claims. International Journal of Eating Disorders. 2000;27: 381- 389
2. Quam L, Ellis LBM, Venus P, Clouse J, Taylor CG, Leatherman S. Using
claims data for epidemiologic research. Medical Care. 1993; 6: 498-507.
3. Chai G, Governale L,McMahon AW, et al. Trends of outpatient prescription
drug utilization in US children, 2002–2010. Pediatrics. 2012;130:23–31.
4. Hersh AL, Shapiro DJ, Pavia AT, et al. Antibiotic prescribing in ambulatory
pediatrics in the United States. Pediatrics. 2011;128:1053–1061.
5. Alsan, Marcella, et al. Antibiotic use in cold and Flu season and prescribing
quality: a retrospective cohort study. Medical care. 2015. 53(12): 1066.
6. Crimmel, Beth Levin. Health Insurance Coverage and Income Levels for the
US Noninstitutionalized Population Under Age 65, 2001. Medical Expenditure
Panel Survey, Agency for Healthcare Research and Quality. 2004.
7. AHFS/ASHP. American Hospital Formulary Service Drug Information. 2012.
AHFS drug information. www.ahfsdruginformation.com. Accessed January 20,
2018.
21
CHAPTER 3: The Impact of Time to Treatment
THE IMPACT OF DAY OF ANTIBIOTIC RECEIPT FOR CHILDREN WITH
ACUTE RESPIRATORY TRACT INFECTIONS ON THE RISK OF REVISITS
Introduction
Antibiotics are the standard treatment for acute bacterial respiratory
infections in children. Clinical guidelines recommend antibiotics for treatment of
acute respiratory tract infections (ARTIs) in children with selected clinical features
or in severe cases.
1,2,3,4
Previous studies have shown that patients who received
antibiotics for acute otitis media (AOM), URI (Upper respiratory infections) and
pharyngitis were more likely to return with an unresolved respiratory tract
infection than patients who did not receive antibiotics.
5
In another study of 2 to 17
years old patients with outpatient acute respiratory infections who were
prescribed an antibiotic, 27% of patients had follow-up visits related to ARTI
within 30 days, while of the children who were not prescribed an antibiotic only
17% had follow-up visits.
6
The evidence on when antibiotic was received and
revisit risk after antibiotic use is minimal in the pediatrics literature.
Delaying antibiotic is an alternative strategy to providing an antibiotic
prescription for immediate fill. Evidence on delaying antibiotic treatment by 24-72
hours as a way to reduce unnecessary antibiotic prescriptions is mixed. Some
studies found immediate antibiotics to be effective in relieving pain, fever, and
runny nose for sore throat, while delayed antibiotics led to a small reduction in
the duration that pain, fever, and cough persisted in people with colds.
7
Delaying
antibiotics reduces antibiotic use compared to immediate antibiotics but
symptoms of acute otitis media and sore throat were slightly improved by
immediate antibiotics when compared with delayed antibiotics. Delayed
antibiotics may, thus be an effective strategy in some situations to reduce overall
antibiotic use while immediate antibiotics may be the optimal treatment option in
other situations. The delay in filling the prescription cannot be verified as
implemented by physician recommendation or patient preference. The decision
to delay an antibiotic can be recommended by the physician as part of the
22
watchful waiting strategy or it can occur simply because the parent/child are
unable to fill the prescription immediately.
More research is needed to provide evidence-based data in pediatric
clinical decision-making regarding antibiotic prescribing for ARTI, in particular,
time to initiating antibiotic therapy.
8, 9, 10, 11
The first objective of this study is to
document the factors correlated with the use of an antibiotic to treat an ARTI
within 3 days. Next, we will document correlation between time of antibiotic
initiation and the likelihood of revisits for an ARTI within 28 days. Several
sensitivity analyses will be conducted exploring alternative classifications of time
to treatment, the time frame over which revisits are measured [7 vs. 28 days],
and alternative specifications of the statistical model for revisit risk.
Methods
Data Source
The data for this study were derived from Optum Insight Clinformatics
claims database. The Optum data include claims of commercially insured
patients with both medical and pharmacy benefits from 2011-2013. These
individuals are privately insured through a large healthcare company that either
fully insures or administers claims for the patients. Optum data are used only for
those individuals who have both medical and prescription drug coverage, thus
allowing evaluation of both their medical and pharmacy claims.
Study Design
Acute respiratory tract infections (ARTIs) are short-term acute conditions
with treatment and follow-up typically lasting 10 to 28 days. Treatment is typically
provided for 7 to 10 days and the appropriate use of antibiotics is often unclear
given the difficulty of determining the exact etiology of the infection in a timely
manner. Whether treated with an antibiotic or not, patients whose ARTI
symptoms fail to resolve usually return within 28 days of their initial visit.
An episode of care was defined around the date of the first ARTI visit (+/-
6 months). The first visit, called an index visit, is identified for every patient
diagnosed with an ARTI from the examined outpatient settings—ambulatory
clinics, physician offices, emergency and urgent care facilities, and the
International Classification of Diseases, Ninth Revision (ICD-9) codes were used
to document these index visits.
To qualify as an index visit, the patient must have an ARTI diagnosis and
at least a six-month enrollment in an Optum health insurance plan prior and
subsequent to that visit. When a patient had multiple ARTI related visits within
the 5-day period prior to their initial ARTI visit, the earliest visit is used to define
the index visit. This assumes that the first ‘initial ARTI’ visit was for a pre-existing
infection because children are most likely to return within 5 days when the
24
existing infection/symptoms worsen or remain unresolved. It also follows that the
initial index visit was then classified as a revisit for an ARTI diagnosis.
All patients diagnosed with an ARTI during the 36-month period, from
January 1, 2011- Dec 31, 2013 were included in the study. The medical claims
file was linked with the pharmacy claims to determine if the index medical visit
was associated with an antibiotic prescription. Patient history of previous acute
respiratory tract infection visits is also recorded. Any ARTI visits that occurred
within 90 days prior to the index visit were documented as prior visits. Any
subsequent visits for an ARTI within 28 days of the index visit were recorded as
revisits to the initial visit for both the treated and the untreated patients.
Using the patient identifier, medical visits were linked to the member
details and the SES (socioeconomic status) and member files in the Optum
dataset to determine insurance and socio-demographic characteristics of the
patients.
Sample Population
Eligible individuals consist of children less than 18 years of age, who were
diagnosed with an ARTI and treated in an outpatient setting. The following ICD-9
codes identify the ARTIs included in the study but do not distinguish between
viral and bacterial infections:
1) Acute otitis media (ICD-9: 381.0, 381.4, 382.0, 382.4, 382.9)
2) Pharyngitis (ICD-9: 462, 463, 034.0)
3) Bronchitis (ICD-9: 466, 490)
4) Sinusitis (ICD-9: 461)
5) Pneumonia (ICD-9: 481-483, 484.8, 485–487, 033.0, 033.9)
6) Cold/URI (ICD-9: 460, 464 [excluding 464.3], 465, 487 [excluding 487.0],
786.2)
Up to 3 ARTI diagnoses may be recorded for any outpatient visit. The first
recorded ARTI diagnosis on the index visit is assumed to be the primary
diagnosis for the treatment episode. Any additional ARTI diagnoses recorded on
25
the index visit are assumed to be secondary diagnoses. Secondary diagnoses
are included to examine the effects of co-morbidities on the primary infection,
propensity for antibiotic treatment, and any effect on revisit rates.
Time to fill the prescription
The pharmacy fill date was used to record the time to fill the antibiotic
prescription, if any, from the date of the index physician visit. We created two
initial classes of treatment timing for patients who filled an antibiotic at any time
within 28 days of the index visit: a) filled antibiotics within 3 days of the index
visit, b) filled antibiotics on days 4-28 of the index visit. These time dependent
definitions of treatment were compared to no treatment within 28 days. This
initial classification of treated patients into classes based on a 3-day definition
was subjected to a sensitivity analysis using alternative specifications for time-to-
treatment groups.
Exclusion criteria
This study excludes patients with less than 6 months of continuous health
insurance enrollment before and after their index visit.
Key explanatory variables
Numerous demographic and clinical variables about each patient are
included in the study to account for observed factors that affect the antibiotic
treatment decision and the patient’s risk of revisiting the physician. Data on
patient age, race, gender, household income, geographic location, practice
setting type, provider specialty and type of insurance are available from the
claims file. Practice setting includes outpatient physician offices, health centers
and clinics, outpatient hospitals, emergency department, urgent care, and
ambulatory care centers. Providers include primary care physicians, specialty
care physicians, and non-physician providers. Patient household income is
obtained directly from the Optum dataset, which they estimated using algorithms
based on census block groups. Income categories were informed by the federal
26
guidelines for a family of four. A low-income family is defined as earning
<$40,000 – $50,000; middle-income family earns >$50,000 – <$75,000 and a
high-income family earns $75,000 and above.
12
We also defined seasons of the year to account for time periods during
which children are most likely to visit a physician for respiratory infections, which
typically are fall and winter.
13
The months of December through February
constitute winter, March through May are included as spring, June through
August are summer and September through November are fall. Patient insurance
type is obtained from the Optum data and categorized as HMO (Health
Maintenance Organization), EPO (Exclusive Provider Organization), POS (Point
of Service), and PPO (Preferred Provider Organization). Race is defined as
White, Black, Hispanic, and Asian. Patient location is obtained from Optum and
defined according to national regions of Northeast Central, Southeast Central,
Mid Atlantic, Mountain, New England, Pacific, South Atlantic, Northwest Central
and Southwest Central. Previous visits variable was defined as whether the child
had ARTI visits within 3 months prior to the index visit.
ARTI-related co-morbidities were defined using the secondary diagnoses
listed on the paid claim for the initial visit. Non-ARTI related co-morbidities listed
on the claim for the initial visit were defined as the following categories: mental
disorders, nervous system disorders, respiratory symptoms, fever, injury and
poisoning, other or no diseases.
Provider proclivity to treat could be a key factor related to the likelihood of
a patient revisit, either because of variations in quality of care or style of practice
with respect to recommending a return visit to check the progress of the patient.
This variable was defined as the proportion of visits during which a patient
received treatment out of all the ARTI visits to the provider. It was calculated by
dividing the number of treated patients with the number of total visits for each
provider/practice. The proportion was divided into 4 categories:
a) Low propensity to treat: proportion treated=0 - 0.4
b) Medium propensity to treat: proportion treated = 0.4 - 0.6
c) Medium-High propensity to treat: proportion treated = 0.6 - 0.8
27
d) High propensity to treat: proportion treated = 0.8 - 1.0
The provider proclivity to treat variable was also use in a sensitivity analysis.
Outcomes
Revisits are used here as the measure of patient outcomes related to
treatment for acute respiratory tract infections (ARTIs). Revisits can be due to
patient and/or physician concern over verifying whether or not the child is
improving. Thus, revisits are considered an outcome not simply as a return for a
worsening infection and the need for better treatment but also as the total
resources used to treat an ARTI patient.
This study measures undertakes two major analyses. First, we derived
estimates of the rate of antibiotic prescriptions filled within 3 days among children
who were provided an antibiotic prescription following an ARTI diagnosis.
Second, we created a dichotomous outcome variable which documents those
patients who had a revisit after the initial treatment. While a host of patient
demographic and clinical factors impact revisit risk, this analysis focuses on the
potential influence of treatment timing—time to fill the antibiotic prescription
following the index visit--within 0-3 days and 4-28 days.
Analytic Methods
A multivariable logistic regression analysis was used to identify
demographic and clinical factors associated with filling an antibiotic within 3 days.
Next, we estimate a multivariable logistic regression model of the likelihood that a
child experienced a revisit within the 28 days follow-up period as a function of
when the medication was filled, if any. The initial time to treatment categories use
in this analysis were: a) filled antibiotics on days 0-3 of the index visit, and b)
filled antibiotics on days 4-28 of the index visit. A sensitivity analysis was then
run with alternate definition of time-to-treatment categories. We also reset the
timeframe for defining a revisit from 28 days to 7 days in an attempt to tighten the
temporal relationship of the revisit to the index visit. Next, we estimated a revisit
risk model using proclivity to treat categories.
28
We tested for statistically significant differences across the treated and
untreated groups. T-tests were conducted for continuous variables and chi-
square tests for categorical variables. The results are shown in Table 1.
All analyses were completed using STATA/IC 13 (StataCorp, College
Station, TX).
Results
The descriptive statistics for patients who filled within 28 days and those
who didn’t are presented in Table 1. Overall, 19,882 pediatric patients diagnosed
with ARTIs, 48.61% filled the antibiotic and 51.38% did not fill within 28 days. Of
the 9,666 patients who filled in 28 days, 8634 patients (89.3%) filled the antibiotic
within 3 days while 1032 patients (10.7%) filled the antibiotic within 4-28 days of
the index visit.
A total of 662 patients had a revisit within 28 days of their index ARTI visit
(3.32%). The unadjusted revisit rate differed for patients who were treated with
an antibiotic within 28 days (2.70%) and patients who were not treated with
antibiotics (3.93%). Of the patients who were treated with an antibiotic within
three days, 235 patients (2.72%) had a return visit, 26 patients (2.52%) who were
treated within 4-28 days had a revisit and 401 untreated patients (3.93%) had a
revisit within 28 days. Among those who received antibiotic treatment in 3 days
(235 patients), 87% (206 patients) filled the antibiotic prior to initiating a revisit.
Figure 1 presents the time to fill, by day, an antibiotic for ARTIs. Nearly
80% of the antibiotics are filled within the first day and 89% are filled within first
three days of the index visit. The fill rate tapers off by the 28
th
day.
Factors associated with filling an antibiotic in 3 days of the index ARTI visit
The significant factors associated with the fill of antibiotic treatment in 3
days of the index visit derived from the constructed multivariate model are
depicted in Table 2, with the odds ratio and 95% confidence intervals as shown.
Children diagnosed with pharyngitis were 54% less likely to receive an
antibiotic than the reference ARTI, bronchitis. ARTIs coded as upper respiratory
29
infection were treated least frequently with antibiotics (Adjusted OR = 0.292),
compared to bronchitis. Interestingly, children coded as having AOM as a
secondary diagnosis, have double the likelihood of receiving an antibiotic
prescription in 3 days, relative to those with no co-morbidities. Children with
pharyngitis, bronchitis and sinusitis as secondary diagnoses also have 44%, 75%
and 48% higher odds of receiving an antibiotic in 3 days, respectively. Children
diagnosed with co-morbid mental disorders have 67% lower odds while those
with nervous system disorders, respiratory symptoms and fever have 60%, 34%
and 45% higher likelihood of receiving the antibiotic in 3 days.
Having an ARTI visit 90 days prior to the index visit decreases the
likelihood of antibiotics being received in 3 days by 52%. Patients seen by
outpatient facility providers and non-physician providers were 16% and 35%
more likely to receive an antibiotic prescription in 3 days than those treated by
primary care physicians (p = 0.008, p=0.000).
Compared to the summer season, winter (Adjusted OR=0.666, p=0.000)
and spring (Adjusted OR=0.604, p=0.000) had lower likelihood of filling the
antibiotic within 3 days while children diagnosed with an ARTI in the fall had a
higher likelihood of filling the antibiotic in 3 days (Adjusted OR=1.255, p=0.001).
The likelihood of receiving an antibiotic is lower among infants of ages 1-5 years
(15%) while 5 to 12 years have 15% higher odds of antibiotic fill in 3 days than
the 12 to 18 year old group.
Regional differences in antibiotic receipt in 3 days were observed in this
study. In the multivariate regression, the Mid Atlantic (32%) and S Atlantic (14%)
regions had significantly lower propensity for filling within 3 days compared to the
reference Pacific region.
12
Compared to white children, Hispanic children were
25% more likely to receive the antibiotic within 3 days of the ARTI visit.
The care setting for the index visit also showed significant trends. The
outpatient hospital, emergency department and other areas of initial care showed
lower antibiotic receipt compared to the office setting. The odds of filling an
antibiotic in 3 days were 0.638 (p=0.000), 0.588 (p =0.000), and 0.580 (p
30
=0.000), respectively. Likewise, insurance and household income bracket did not
significantly influence antibiotic receipt propensity.
The Effects of Treatment on the Risk of a Revisit to the Physician
Table 3 presents the results of the multivariate analysis of the risk of a
revisit to the physician within 28 days as a function of treatment timing compared
to no treatment. Antibiotic receipt within Day 0-3 of the index visit reduces the
likelihood of a revisit by 51% compared to patients who did not fill an antibiotic.
Those who filled their antibiotic within 4-28 days of the index visit were 42% less
likely to have a revisit compared to children who never received an antibiotic
prescription.
The estimated impact of several additional variables used in the analysis
to generate an unconfounded estimate of the effect of time to fill a prescription is
of interest. Relative to bronchitis, statistically significant odds of a revisit within 28
days were lowest for URI (Adjusted OR = 0.382), pharyngitis (Adjusted OR =
0.402) and sinusitis (Adjusted OR = 0.667), and highest for pneumonia (Adjusted
OR = 2.696). Children with pneumonia as secondary diagnosis also have four
times higher odds of having a revisit compared with children with no co-
morbidities (Adjusted OR = 4.186).
Having an ARTI visit 90 days prior to the index visit had a fourfold
increase in the likelihood of initiating a revisit within 28 days of the ARTI visit.
Compared to the summer season, winter (Adjusted OR=1.419, p=0.085) and
spring (Adjusted OR=1.442, p=0.082) had higher likelihood of a revisit.
Patients seen by outpatient facility providers were 77% less likely to return
for a revisit compared to those treated by primary care physicians (p = 0.000).
Females are 15% less likely to have a revisit than males. Similarly, infants less
than 1 year old have more than double the likelihood of a revisit relative to
children over age 12. Children 1-5 years old have 1.70 times higher likelihood of
a revisit than children above the age of 12 years.
Children with EPO, HMO, and PPO plans were less likely to have a revisit
than those with POS plans. Statistically significant odds (p=0.000) of a revisit
31
within 28 days were lowest for EPO (Adjusted OR = 0.192), HMO (Adjusted OR
= 0.268) and PPO (Adjusted OR = 0.260). Regional differences in revisit risk
were observed in this study. In the multivariate regression, the Mid Atlantic
(74%), Mountain (46%) and S Atlantic (37%) regions had higher propensity for
initiating a revisit compared to the reference Pacific region.
12
Compared to white
children, only Hispanic children were 33% less likely to have a revisit in 28 days
of the ARTI visit.
The care setting also showed significant trends. The outpatient hospital,
and other areas of initial care showed higher revisit risk compared to the office
setting. The revisit odds were 2.061 (p=0.002) and 2.328 (p =0.000) for
outpatient and other areas of care, respectively. Household income bracket did
not significantly influence the risk for a revisit.
Sensitivity Analyses
Effects of Alternative Definitions of Time to Treatment on Revisit Risk
We also estimated the effects of treatment timing among the ARTI
population. (Table 4) The day of treatment receipt is a determining factor in
revisit risk. Children receiving an antibiotic within a day of the index visit had 43%
lower risk of initiating a revisit (p=0.000) within 28 days than the children who
were not treated. Children receiving an antibiotic on Day 2-3 of the index visit had
34% reduction in the likelihood of initiating a revisit compared to children who did
not get the antibiotics. However, effects on the risk of a revisit is not significant
when antibiotic is filled after day 3. Therefore, filling the antibiotic prescription
within 3 days significantly reduces the risk of a revisit.
We conducted sensitivity analyses which jointly investigate how an
alternative definition of revisits [7 days] and the inclusion of variables capturing
the physician proclivity to treat impact our core estimates of how treatment timing
effect revisit risk (Table 5). When we estimated the effects of time to fill an
antibiotic on revisit within 7 days, we found that the estimate impact of antibiotic
32
timing on revisit risk was consistent with results for 28-day revisits. Antibiotic
receipt within Day 0-3 of the index visit reduces the likelihood of a revisit by 42%
compared to patients who did not get an antibiotic. Those who filled their
antibiotic within 4-28 days of the index visit were 36% less likely to have a revisit
compared to children who never received an antibiotic prescription. The results
for revisit within 7 days and 28 days are similar in that both models showed that
the group filling the antibiotic within 3 days had a greater reduction in revisit risk
than those who filled after 3 days.
We also included provider proclivity to treat in these sensitivity models of
the risk of a revisit. These analyses found that the revisit risk is not correlated
with provider proclivity to treat with the exception that patients treated by high –
prescribers, who are most likely to treat an ARTI patient, are more likely to have
a revisit within 28 days (Adjusted OR=1.561, p=0.006).
Discussion
The 28-day revisit rate for treated (2.72%) and untreated patients (3.93%)
is fairly similar and small in magnitude. Other studies have found revisit rate
within the range of 2.8% to 6.5%.
22,23,24
A study focusing on AOM found revisit
rate 3.2% to 6.4%
22
, and another study of AOM found recurrent visit to be 6.5%
and 2.8% in two AOM groups.
23
Treatment timing is a crucial factor for children who are properly
diagnosed with an ARTI infection. We found that the time to fill an antibiotic plays
a significant role in averting revisits for children diagnosed with ARTIs. When
evaluated among both treated and untreated population, filling an antibiotic within
3 days of the ARTI visit is associated with a 51% reduction in revisit risk.
There are several factors affecting the likelihood of returning for a revisit.
The risk of revisit decreases with age. Infants <1years old are twice as likely to
have a revisit than the 12-18 years old children. Infants are less likely to fill an
antibiotic within 3 days but have higher rate of revisits than the children over the
age of 12 years. Infants may also benefit from immediate antibiotics as they are
33
more likely to return for a revisit when the antibiotics are not received within 3
days after the index visit.
Children who have pneumonia as co-morbid with an ARTI diagnosis are
more likely to have a revisit than children who have no comorbidities. Our
findings also suggest that children seen by outpatient facility providers have
lower revisit risk than those seen by primary care physicians. Primary care
physicians may be more inclined to check the progress of their patients and thus
recommend a revisit than those who are not designated providers for a child,
such as non-physician providers and outpatient facility providers. Patients treated
by their primary provider may demand more revisits because going back to their
usual source of care, or having a usual source of care, may be lower, may lower
the ‘opportunity’ cost of a revisit. We also find that children seen in outpatient
hospitals have higher revisit risk than those seen in office settings. Patients seen
by providers in outpatient hospitals may remain unsatisfied or be recommended
to return for a revisit. The outpatient facilities and providers may also readily
recommend a return visit for patients to their primary care physicians after visiting
an outpatient facility or as a follow-up to the outpatient hospital as a result of an
intensive initial visit and diagnosis.
There are regional differences in rates of risk for revisit. This is evident in
our findings that children in Mid Atlantic and South Atlantic have 74% and 37%
higher likelihood of revisits than those in the Pacific while children in both of
these regions are significantly less likely (32% and 24%) to fill the antibiotic in 3
days. These regional differences may reflect factors associated with hospital
density, provider practice style etc. that affect prescription behavior in different
parts of the country.
17, 25
Regional differences may also be related to ARTI risk
factors associated with the region, such as weather, which are not accounted for
in this analysis.
The risk of a revisit was not impacted by provider proclivity to treat (Table
5B). We found that the revisit risk significantly is higher for high prescribers
using a 28-day revisit window. High prescribers may be more inclined to
recommend a revisit after prescribing an antibiotic for ARTIs. This is suggestive
34
of provider prescribing behavior and practice style that allots for a revisit following
ARTI diagnosis.
Limitations
Our estimates represent data from a large U.S. health plan and contains a
greater representation of high-income households which is not representative of
the national U.S. population. We assume that all ARTI diagnoses in our study
population fulfilled reasonable criteria for that diagnosis. However, since the
diagnosis of ARTI usually depends on clinical criteria rather than a diagnostic
test, there is significant heterogeneity in how these diagnoses are assigned.
Infection severity and clinical symptoms are not available via ICD-9 codes,
thus it is not apparent whether a treated child presented with severe symptoms to
warrant treatment. It should be noted that lack of severity data and heterogeneity
in physician clinical treatment judgment affect patient outcomes.
Revisits are identified using diagnosis categories as defined by ICD-9
codes. We are unable to measure whether the revisit suggests resolution of
infection or worsening infection symptoms. We evaluated CPT codes to develop
and identify a revisit however CPT codes appear to suggest the financial
reimbursement schemes that may affect how a visit is coded. They do not clarify
whether the infection worsened/resolved or provide information about treatment
failure.
Table 1: Descriptive Statistics
Total, N=19,882 Filled in 3
days N=8634
(43.42%)
Filled in 4-28
days N=1032
(5.19%)
Not treated
N=10,216
(51.38%)
p-value
Revisit 28days 235(2.72%) 26 (2.52%) 401 (3.93%) 0.000
Revisit 7 days 127 (1.47%) 14 (1.36%) 205 (2.01%) 0.000
Filled antibiotic prior to
revisit (only among
treated)
206 (2.38%) 0 0 0.000
Primary ARTI Diagnosis: 0.000
AOM 2173 (25.17%) 205 (19.86%) 1369 (13.40%)
Pharyngitis 2831 (32.79%) 355 (34.40%) 4076 (39.90%)
Bronchitis 691 (8%) 53 (5.14%) 429 (4.20%)
Sinusitis 1039 (12.03%) 80 (7.75%) 790 (7.73%)
Pneumonia 322 (3.73%) 24 (2.33%) 203 (1.99%)
URI 1578 (18.28%) 315 (30.52%) 3349 (32.78%)
Secondary ARTI
Diagnoses
0.000
AOM 5444 (63.05%) 717 (69.48%) 7252 (70.99%)
Pharyngitis 368 (4.26%) 36 (3.49%) 239 (2.34%)
Bronchitis 548 (6.35%) 44 (4.26%) 602 (5.89%)
Sinusitis 103 (1.19%) 11 (1.07%) 68 (0.67%)
Pneumonia 170 (1.97%) 8 (0.78%) 139 (1.36%)
URI 19 (0.22%) 2 (0.19%) 19 (0.19%)
Baseline Medical
Diagnoses
Mental disorders 558 (6.46%) 46 (0.46%) 530 (5.19%)
Nervous system disorders 9 (0.10%) 2 (0.19%) 36 (0.35%)
Respiratory symptoms 369 (4.27%) 31 (3%) 245 (2.40%)
Fever 556 (6.44%) 74 (7.17%) 546 (5.34%)
Injury and poisoning 458 (5.30%) 55 (5.33%) 493 (4.83%)
Other/None 32 (0.37%) 6 (0.58%) 47 (0.46%)
Had previous visits (in
the prior 90 days)
36 (0.41%) 25 (2.42%) 62 (0.60%) 0.000
Season of the visit 0.000
Winter (Dec- Feb) 3787 (43.86%) 494 (47.87%) 5179 (50.69%)
Spring (Mar-May) 1862 (21.57%) 143 (13.86%) 2839 (27.79%)
Summer (Jun- Aug) 655 (7.59%) 50 (4.84%) 572 (5.60%)
Fall (Sept –Nov) 2330 (26.99%) 345 (33.43%) 1626 (15.92%)
Provider 0.000
Primary Physician 6089 (70.52%) 731 (70.83%) 6970 (68.23%)
Specialty physician 331 (3.83%) 39 (3.78%) 443 (4.34%)
36
Non-physician providers 617 (7.15%) 54 (5.23%) 510 (4.99%)
Outpatient facility provider 1591 (18.43%) 207 (20.06%) 2285 (22.37%)
Other/unknown 6 (0.07%) 1 (0.10%) 8 (0.08%)
Gender 0.000
Male 4386 (50.80%) 531 (51.45%) 5023 (49.17%)
Female 4248 (49.20%) 501 (48.55%) 5193 (50.83%)
Age (in years) 0.000
<1 153 (1.77%) 65 (6.30%) 108 (1.06%)
1- ≤5 1484 (17.19%) 268 (25.97%) 1803 (17.65%)
5 - ≤12 3879 (44.93%) 403 (39.05%) 4172 (40.84%)
12 - ≤18 3118 (36.11%) 296 (28.68%) 4133 (40.46%)
Household Income 0.003
Low 1133 (13.12%) 139 (13.47%) 1242 (12.16%)
Middle 1269 (14.70%) 154 (14.92%) 1335 (13.07%)
High 4664 (54.02%) 553 (53.59%) 5829 (57.06%)
Unknown 1568 (18.16%) 186 (18.02%) 1810 (17.72%)
Insurance Type 0.000
EPO 1023 (11.85%) 128 (12.40%) 1330 (13.02%)
HMO 589 (6.82%) 70 (6.78%) 884 (8.65%)
POS 6865 (79.51%) 816 (79.07%) 7767 (76.03%)
PPO 133 (1.54%) 15 (1.45%) 209 (2.05%)
Race 0.000
White 5959 (69.02%) 721 (69.86%) 7288 (71.34%)
Black 776 (8.99%) 89 (8.62%) 908 (8.89%)
Hispanic 1065 (12.33%) 108 (10.47%) 1043 (10.21%)
Asian 427 (4.95%) 58 (5.62%) 470 (4.60%)
Unknown 407 (4.71%) 56 (5.43%) 507 (4.96%)
Region 0.000
E N Central 1445 (16.74%) 168 (16.28%) 1576 (15.43%)
E S Central 334 (3.87%) 28 (2.71%) 391 (3.83%)
Mid Atlantic 403 (4.67%) 87 (8.43%) 700 (6.85%)
Mountain 1007 (11.66%) 119 (11.53%) 1130 (11.06%)
New England 227 (2.63%) 33 (3.20%) 281 (2.75%)
Pacific 630 (7.30%) 76 (7.36%) 696 (6.81%)
S Atlantic 2016 (23.35%) 275 (26.65%) 2574 (25.20%)
W N Central 1100 (12.74%) 102 (9.88%) 1354 (13.25%)
W S Central 1472 (17.05%) 144 (13.95%) 1514 (14.82%)
Setting 0.000
Office 7492 (86.77%) 870 (84.30%) 8234 (80.60%)
Clinic 19 (0.22%) 3 (0.29%) 43 (0.42%)
Outpatient hospital 564 (6.53%) 69 (6.69%) 789 (7.72%)
ER 89 (1.03%) 9 (0.87%) 156 (1.53%)
Other 470 (5.44%) 81 (7.85%) 994 (9.73%)
Fig 1: Time to fill an antibiotic for ARTIs
0 20 40 60 80
Percent
0 10 20 30
Time to fill
38
Table 2: Multivariable model of factors associated with filling an antibiotic
prescription within 3 days
Filling antibiotics within 3
days (N=8634)
Odds Ratio p-value 95% CI
Primary ARTI Diagnosis
AOM 0.988 0.870 0.860 1.135
Pharyngitis 0.467 0.000 0.409 0.532
Bronchitis 1 (Reference)
Sinusitis 0.821 0.011 0.705 0.956
Pneumonia 1.041 0.709 0.841 1.288
URI 0.292 0.000 0.256 0.335
Secondary ARTI
Diagnosis
AOM 2.323 0.000 1.961 2.752
Pharyngitis 1.442 0.000 1.271 1.623
Bronchitis 1.755 0.000 1.289 2.389
Sinusitis 1.484 0.001 1.174 1.875
Pneumonia 1.082 0.811 0.564 2.076
URI 1.170 0.017 1.028 1.331
Baseline Medical
Diagnoses
Mental disorders 0.339 0.005 0.159 0.724
Nervous system disorders 1.608 0.000 1.357 1.905
Respiratory symptoms 1.345 0.000 1.184 1.528
Fever 1.452 0.000 1.266 1.666
Injury and poisoning 0.873 0.569 0.549 1.389
Other/None 1 (Reference)
Had previous visits in
90days
0.488 0.001 0.324 1.734
Season
Winter 0.666 0.000 0.589 0.753
Spring 0.604 0.000 0.531 0.687
Summer 1 (Reference)
Fall 1.255 0.001 1.101 1.432
Provider
Primary Physician 1( Reference)
Specialty physician 0.916 0.282 0.781 1.074
Non-physician provider 1.352 0.000 1.192 1.533
Outpatient facility provider 1.169 0.008 1.042 1.311
Other facility provider 1.071 0.905 0.343 3.347
Gender
Male 1 (Reference)
Female 0.960 0.185 0.905 1.019
39
Age (in years)
<1 0.905 0.420 0.712 1.151
1- ≤5 0.854 0.001 0.780 0.935
5 - ≤12 1.159 0.000 1.084 1.240
12 - ≤18 1 (Reference)
Insurance Type
EPO 0.859 0.611 0.478 1.542
HMO 0.799 0.451 0.446 1.431
POS 1 (Reference)
PPO 0.701 0.263 0.377 1.304
Region
E N Central 0.987 0.853 0.862 1.130
E S Central 1.012 0.898 0.837 1.224
Mid Atlantic 0.684 0.000 0.578 0.809
Mountain 1.017 0.811 0.882 1.173
New England 0.889 0.278 0.719 1.099
Pacific 1 (Reference)
S Atlantic 0.866 0.030 0.761 0.986
W N Central 0.971 0.695 0.840 1.123
W S Central 1.092 0.200 0.954 1.249
Race
White 1 (Reference)
Black 1.092 0.109 0.980 1.216
Hispanic 1.250 0.000 1.134 1.378
Asian 1.164 0.035 1.010 1.340
Unknown 1.041 0.572 0.905 1.196
Household Income
Low 1 (Reference)
Middle 1.022 0.698 0.912 1.146
High 0.914 0.059 0.832 1.003
Unknown 0.980 0.722 0.879 1.092
Setting
Office 1 (Reference)
Clinic 0.745 0.323 0.417 1.334
Outpatient hospital 0.638 0.000 0.546 0.744
ER 0.588 0.000 0.440 0.786
Other 0.580 0.000 0.496 0.678
40
Table 3: Multivariable model of factors associated with revisits after filling an
antibiotic within 3 days of the index ARTI visit
Revisit 28 days (N=662) Odds Ratio p-value 95% CI
Treatment timing
Not treated 1 (Reference)
Day 0-3 0.494 0.000 0.325 0.751
Day 4-28 0.582 0.001 0.488 0.694
Primary ARTI Diagnosis
AOM 0.861 0.342 0.634 1.171
Pharyngitis 0.402 0.000 0.293 0.553
Bronchitis 1 (Reference)
Sinusitis 0.667 0.028 0.465 0.957
Pneumonia 2.696 0.000 1.868 3.891
URI 0.382 0.000 0.279 0.524
Secondary ARTI
Diagnoses
AOM 1.091 0.684 0.716 1.663
Pharyngitis 0.866 0.498 0.572 1.311
Bronchitis 1.086 0.834 0.499 2.364
Sinusitis 0.553 0.159 0.243 1.260
Pneumonia 4.186 0.001 1.751 10.00
URI 0.755 0.138 0.521 1.094
Baseline Medical
Diagnoses
Mental disorders 1.071 0.925 0.250 4.578
Nervous system disorders 0.711 0.186 0.429 1.177
Respiratory symptoms 1.099 0.554 0.803 1.504
Fever 1.129 0.517 0.781 1.633
Injury and poisoning 0.341 0.290 0.046 2.495
Other/None 1 (Reference)
Previous visits in 90 days 4.952 0.000 3.018 8.123
Season
Winter 1.419 0.085 0.953 2.115
Spring 1.442 0.082 0.954 2.180
Summer 1 (Reference)
Fall 1.008 0.969 0.656 1.549
Provider
Primary Physician 1 (Reference)
Specialty physician 1.165 0.409 0.810 1.677
Non-physician provider 0.878 0.500 0.602 1.279
Outpatient facility provider 0.238 0.000 0.157 0.361
Other 0.188 0.130 0.021 1.632
Gender
41
Male 1 (Reference)
Female 0.851 0.048 0.726 0.998
Age (in years)
<1 2.778 0.000 1.744 4.423
1- ≤5 1.708 0.000 1.370 2.128
5 - ≤12 0.948 0.599 0.778 1.155
12 - ≤18 1 (Reference)
Insurance Type
EPO 0.192 0.001 0.075 0.492
HMO 0.268 0.005 0.106 0.675
POS 1 (Reference)
PPO 0.260 0.012 0.091 0.747
Region
E N Central 1.033 0.868 0.700 1.525
E S Central 1.468 0.137 0.885 2.436
Mid Atlantic 1.741 0.010 1.143 2.650
Mountain 1.460 0.061 0.982 2.171
New England 1.459 0.177 0.843 2.527
Pacific 1 (Reference)
S Atlantic 1.379 0.084 0.958 1.984
W N Central 1.059 0.780 0.704 1.593
W S Central 1.335 0.136 0.913 1.953
Race
White 1 (Reference)
Black 0.818 0.193 0.604 1.107
Hispanic 0.678 0.011 0.503 0.914
Asian 0.873 0.482 0.598 1.273
Unknown 1.169 0.357 0.838 1.631
Household Income
Low 1 (Reference)
Middle 0.839 0.259 0.620 1.137
High 0.849 0.190 0.666 1.084
Unknown 0.959 0.776 0.724 1.271
Setting
Office 1 (Reference)
Clinic 0.551 0.444 0.120 2.527
Outpatient hospital 2.061 0.002 1.304 3.258
ER 1.772 0.127 0.849 3.696
Other 2.328 0.000 1.449 3.740
42
Table 4: Sensitivity analysis: Revisit 28 day model with alternative definitions of
time to treatment
Revisits 28 days (N=662)
Adjusted OR
Treatment timing
Not treated 1 (Reference)
Day 0-1 0.574 (p=0.000)
Day 2-3 0.767 (p=0.009)
Day 4-5 0.625 (p=0.365)
Day 6-7 0.430 (p=0.111)
Day 8-10 0.244 (p=0.051)
Day >11 0.555 (p=0.028)
Table 5: Sensitivity analysis: Revisit 7 day model with antibiotic treatment by day
of receipt and provider proclivity to treat
Revisits 7 days
(N=346)
Adjusted OR Revisits 28 days
(N=662)
Adjusted OR
Treatment timing Treatment timing
Not treated 1 (Reference) Not treated 1 (Reference)
Day 0-3 0.589 (p=0.073) Day 0-3 0.518 (p=0.000)
Day 4-28 0.648 (p=0.003) Day 4-28 0.625 (p=0.003)
Provider
proclivity to treat
Provider proclivity
to treat
Low prescribers 1.364 (p=0.126) Low prescribers 1.278 (p=0.107)
Medium
prescribers
1.078 (p=0.742) Medium prescribers 1.148 (p=0.415)
Medium-high
prescribers
0.981 (p=0.929) Medium-high
prescribers
1.082 (p=0.605)
High prescribers 1.190 (p=0.453) High prescribers 1.561 (p=0.006)
43
References
1. Shulman ST, Bisno AL, Clegg HW, et al. Clinical practice guideline for the
diagnosis and management of group A streptococcal pharyngitis: 2012 update by
the Infectious Diseases Society of America [published correction appears in Clin
Infect Dis. 2014;58(10):1496]. Clinical Infectious Diseases. 2012;55 (10):1279-
1282.
2. Lieberthal AS, Carroll AE, Chonmaitree T, et al. The diagnosis and
management of acute otitis media [published correction appears in Pediatrics.
2014;133(2):346]. Pediatrics. 2013;131(3):e964-e999.
3. Chow AW, Benninger MS, Brook I, et al; Infectious Diseases Society of
America. IDSA clinical practice guideline for acute bacterial rhinosinusitis in
children and adults. Clinical Infectious Diseases. 2012;54(8): e72-e112.
4. Rosenfeld RM, Piccirillo JF, Chandrasekhar SS, et al. Clinical practice
guideline (update): adult sinusitis. Otolaryngology Head and Neck Surgery.
2015;152(2) (suppl):S1-S39.
5. Li, John, et al. Antimicrobial prescribing for upper respiratory infections and its
effect on return visits. Family Medicine. 2009. 41(3): 182-7.
6. Zhao SR, Griffin MR, Patterson BL, et al. Risk Factors for Outpatient Use of
Antibiotics in Children with Acute Respiratory Illnesses. Southern Medical
Journal. 2017;110(3):172–180.
7. Spurling, Geoffrey KP, et al. Delayed antibiotics for respiratory infections.
Cochrane Database of Systematic Reviews. 2013; 4(4): CD004417-1.
24.
8. Finkelstein JA, Stille C, Nordin J, Davis R, Raebel MA, Roblin D, Go AS, Smith
D, Johnson CC, Kleinman K, Chan KA, Platt R. Reduction in Antibiotic Use
Among US Children, 1996−2000. Pediatrics. 2003; 112: 620.
9. Watson RL, Dowell SF, Jayaraman M, Keyserling H, Kolczak M, Schwartz B.
A Antimicrobial Use for Pediatric Upper Respiratory Infections: Reported
Practice, Actual Practice, and Parent Beliefs. Pediatrics. 1999; 104:1251.
10. Grijalva CG, Nuorti JP, Griffin MR. Antibiotic prescription rates for acute
respiratory tract infections in US ambulatory settings. The Journal of American
Medical Association. 2009;302(7):758–766
11. Fleming-Dutra, Katherine E., et al. Prevalence of inappropriate antibiotic
prescriptions among US ambulatory care visits, 2010-2011. The Journal of
American Medical Association. 2016; 315(17): 1864-1873.
44
12. Crimmel, Beth Levin. Health Insurance Coverage and Income Levels for the
US Noninstitutionalized Population Under Age 65, 2001. Medical Expenditure
Panel Survey, Agency for Healthcare Research and Quality. 2004.
13. Alsan, Marcella, et al. Antibiotic use in cold and Flu season and prescribing
quality: a retrospective cohort study. Medical Care. 2015. 53(12): 1066.
14. Gonzales R, Bartlett JG, Besser RE, Cooper RJ, Hickner JM, Hoffman JR, et
al. Principles of Appropriate Antibiotic Use for Treatment of Acute Respiratory
Tract Infections in Adults: Background, Specific Aims, and Methods. Annals of
Internal Medicine. 2001;134: 479–486.
15. Gonzalez R, Malone DC, Maselli JH, Sande MA. Excessive Antibiotic Use for
Acute Respiratory Infections in the United States. Clinical Infectious Diseases
2001; 33:757–62.
16. Fleming-Dutra, Katherine E., et al. Variations in antibiotic and azithromycin
prescribing for children by geography and specialty—United States, 2013. The
Pediatric Infectious Disease Journal. 2018; 37(1): 52-58.
17. Hicks, Lauri A., et al. US outpatient antibiotic prescribing variation according
to geography, patient population, and provider specialty in 2011. Clinical
Infectious Diseases. 2015; 60(9): 1308-1316.
18. Linder JA. Editorial commentary: antibiotics for treatment of acute respiratory
tract infections: decreasing benefit, increasing risk, and the irrelevance of
antimicrobial resistance. Clinical Infectious Diseases. 2008;47(6):744-746.
19. Centers for Disease Control and Prevention. Get smart: know when
antibiotics work. http://www.cdc.gov/getsmart. Accessed Jan 12, 2018.
20. Shehab, Nadine, et al. Emergency department visits for antibiotic-associated
adverse events. Clinical Infectious Diseases. 2008. 47(6): 735-743.
21. NCQA. HEDIS 2016.
http://www.ncqa.org/HEDISQualityMeasurement/HEDISMeasures/HEDIS2016.a
spx. Accessed October 20, 2015.
22. McGrath LJ, Becker-Dreps S, Pate V, Brookhart MA. Trends in Antibiotic
Treatment of Acute Otitis Media and Treatment Failure in Children, 2000–2011.
PLoS ONE. 2013, 8(12): e81210.
23. Rothman S, Pitaro J, Hackett A, Kozer E, Gavriel H, Muallem-Kalmovich L,
Eviatar E, Marom T. Treatment of Acute Otitis Media in the Pediatric Emergency
Department. The Pediatric infectious Disease Journal. 2018. 37(6): 520-525.
45
24. Pan Q, Ornstein S, Gross AJ, et al. Antibiotics and return visits for respiratory
illness: a comparison of pooled versus hierarchical statistical methods. The
American Journal of Medical Sciences. 2000;319(6):360-5.
25. Hersh AL, Jackson MA, Hicks LA and the Committee on Infectious Diseases.
Principles of Judicious Antibiotic Prescribing for Upper Respiratory Tract
Infections in Pediatrics. Pediatrics. 2013; 132 (6): 1146-1154.
46
CHAPTER 4: The Impact of Time to Treatment and Antibiotic Class
THE IMPACT OF ANTIBIOTIC TREATMENT TIME AND CLASS OF
ANTIBIOTIC FOR ACUTE OTITIS MEDIA INFECTIONS ON THE RISK OF
REVISITS
Introduction
Antibiotics are recommended treatment for acute otitis media (AOM)
infections that are thought to be bacterial in etiology, particularly in children.
1,2
A
delayed antibiotic treatment strategy is also recommended as a safety net for
avoiding rare but significant complications that can result from uncomplicated
AOM, and reducing antibiotic use, while insuring adequate treatment and
providing high levels of patient satisfaction.
The delayed antibiotic strategy involves providing a prescription for an
antibiotic medication, but with advice on whether to fill the prescription
immediately or delay the fill.
3,4,5,6
The decision to provide a delayed antibiotic is
based on child’s age, infection severity, diagnosis uncertainty and the
effectiveness of an antibiotic for AOM.
7,8
The potential for the infection to resolve
spontaneously over time, and mildness may not warrant immediate antibiotic
treatment. Furthermore, the delay to fill the prescription can be intended by the
physician or patient and the parents being unable to fill the antibiotic immediately.
However, antibiotic treatment becomes increasing important if the patient fails to
improve after 48-72 hours of observation.
9,10
Children less than 2 years of age may benefit from immediate
antibiotics.
9,10
Young children are more likely to suffer worsening infection and
experience symptoms such as fever, ear discomfort, irritation etc. due to delay in
treatment. Those patients with acute otitis media [AOM] can progress to
mastoiditis and tympanostomy tube placement.
11,12
While delayed antibiotics are
helpful in reducing the use of antibiotics and lowering healthcare costs, health
outcomes for young children with delayed treatment may be compromised, thus
implying immediate antibiotics as the superior strategy.
5
Evidence of possible effects on patient outcomes from delaying antibiotic
treatment and reduce unnecessary antibiotic prescriptions is mixed. Some
47
studies found immediate antibiotics to be effective in relieving pain, fever, and
runny nose associated with sore throat, while delayed antibiotics led to a small
reduction in how long pain, fever, and cough persisted in people with colds.
5
Delaying antibiotics reduces antibiotic use compared to immediate antibiotics but
symptoms of acute otitis media and sore throat were slightly improved by
immediate antibiotics when compared with delayed antibiotics. The delay in filling
the prescription cannot be verified as implemented by physician recommendation
or patient preference. The decision to delay an antibiotic can be recommended
by the physician as part of the watchful waiting strategy or it can occur simply
because the parent/child are unable to fill the prescription immediately.
Documenting revisits after antibiotic use in children with AOM under real
world clinical conditions is needed to guide prescribing practice and antibiotic use
decisions.
13,14,15
Thus, we document filling patterns of antibiotic prescription for
acute otitis media and identify factors correlated with filling an antibiotic within 3
days of the index visit. The main objective of this study is to estimate the effects
of filling the antibiotic prescription within 3 days and the class of antibiotic used
on the likelihood of revisits for an AOM within 28 days.
Methods
Data Source
The data for this study were derived from Optum Insight Clinformatics
claims database. The Optum data include claims of commercially insured
patients with both medical and pharmacy benefits from 2011-2013. These
individuals are privately insured through a large company, which either fully
insures or administers claims for the patients. Optum data are used only for
those individuals who have both medical and prescription drug coverage, thus
allowing evaluation of both their medical and pharmacy claims.
Study Design
Acute otitis media (AOM) is an acute infection of that ear which may
require treatment and follow-up visits over the episode of illness which typically
48
lasts 10 to 28 days. Treatment with an antibiotic, if prescribed, is typically
provided for 7 to 10 days. The appropriate use of antibiotics is often unclear
given the difficulty of determining the etiology of an infection in a timely manner.
Whether treated with an antibiotic or not, patients whose AOM symptoms fail to
resolve usually initiate a return visit to their provider within 28 days of their initial
visit.
An episode of care was defined around the date of the first AOM visit (+/-
6 months). The first visit, called an index visit, is identified for every patient
diagnosed with AOM during a visit to an outpatient setting [ambulatory clinics,
physician offices, emergency and urgent care facilities]. Diagnostic codes
[International Classification of Diseases, Ninth Revision (ICD-9)] recorded on the
paid claim for the initial outpatient visit were used to classify these index visits.
To qualify as an index visit, the patient must have AOM as primary
diagnosis and at least a six-month enrollment in a health insurance plan prior and
subsequent to that visit. When a patient had multiple outpatient visits within the
5-day period prior to the initial AOM visit, the earliest visit is used to define the
index visit. This assumes that the first ‘initial AOM’ visit was a revisit for a pre-
existing AOM infection because children are most likely to return within 5 days
when the existing infection/symptoms worsen or remain unresolved.
All patients diagnosed with AOM for the 36-month period, from January 1,
2011- Dec 31, 2013 were included in the study. The medical claims file was
linked with the pharmacy claims to determine if an antibiotic prescription was
temporally associated with the index medical visit. Patient history of previous
acute respiratory tract infection visits is also recorded. Any respiratory infection
visits that occurred 90 days prior to the index visit were documented as prior
visits. Any subsequent visits for any acute respiratory infection within 28 days of
the index visit were recorded as revisits to the initial visit for both the treated and
the untreated groups.
Using the patient identifier, medical visits were linked to the member
details and the SES (socioeconomic status) and member files in the Optum
49
dataset to determine insurance and socio-demographic characteristics of the
patients.
Sample Population
Eligible individuals consist of children and adolescents less than 18 years
of age, who had an outpatient visit for acute otitis media (AOM). The ICD-9
codes (ICD-9: 381.0, 381.4, 382.0, 382.4, 382.9) for acute otitis media were used
to identify the infection. Up to 3 diagnoses may be recorded for an AOM visit that
resulted in a prescription treatment. The first diagnosis is assumed to be the
primary diagnosis in all AOM visits. Secondary diagnoses were included to
examine the effects of co-morbidities on the primary infection and its treatment.
Antibiotic Classification
Antibiotics were grouped into classes based on classification codes from
the American Hospital Formulary Service.
23
These classes include penicillins,
cephalosporins, macrolides (Biaxin, Zithromax, Z-pack), tetracyclines,
sulfonamides, fluoroquinolones (Cipro, Levaquin, Avelox), lincomycin,
carbapenems and beta-lactam antibiotics (Amoxicillin, Amoxicillin/clavulanate
(Augmentin), Keflex, Ceftin, Omnicef).
Time to fill the prescription
The pharmacy fill date was used to identify the time to fill the antibiotic
prescription from the date of the index AOM visit. The time-to-fill data provided
flexibility in the classification of when an antibiotic is received by treated AOM
patients. The primary definition of treatment used in this analysis was a
categorical variable which identifies children who filled an antibiotic prescription
within 3 days or at any time in the post 28 day follow-up period. These ‘treated’
patients were then compared to patients who did not fill a prescription for an
antibiotic within 28 days of the index visit.
50
Several sensitivity analyses were conducted using alternative
classifications for time to treatment. For example, we also created 3 classes of
treatment timing in a separate model for patients who filled an antibiotic at any
time within 28 days of the index visit: a) filled antibiotics on the same day of the
index visit, b) filled antibiotics on day 1-2 of the index visit and c) filled antibiotics
on or after day 3.
Exclusion criteria
This study excludes patients with less than 6 months of continuous health
insurance enrollment before and after the index date.
Key explanatory variables
Numerous demographic and clinical variables about each patient are
included in the study to account for observed factors that affect the antibiotic
treatment decision. Data on patient age, race, gender, household income,
geographic location, practice setting type, provider specialty and type of
insurance are available from the claims file. Practice setting includes outpatient
physician offices, health centers and clinics, outpatient hospitals, emergency
department, urgent care, and ambulatory care centers. Providers include primary
care physicians, specialty care physicians, and non-physician providers. Patient
household income is obtained directly from the Optum dataset, which they
estimated using algorithms based on census block groups. Income categories
were informed by the federal guidelines for a family of four. A low-income family
is defined as earning <$40,000 – $50,000; middle-income family earns >$50,000
– <$75,000 and a high-income family earns $75,000 and above.
20
We also defined seasons of the year to account for time periods during
which children are most likely to visit a physician for respiratory infections, which
typically are fall and winter.
24
The months of December through February
constitute winter, March through May are included as spring, June through
August are summer and September through November are fall. Patient insurance
type is obtained from the Optum data and categorized as HMO (Health
51
Maintenance Organization), EPO (Exclusive Provider Organization), POS (Point
of Service), and PPO (Preferred Provider Organization). Race is defined as
White, Black, Hispanic, and Asian. Patient location is obtained from Optum and
defined according to national regions of Northeast Central, Southeast Central,
Mid Atlantic, Mountain, New England, Pacific, South Atlantic, Northwest Central
and Southwest Central. Previous visits variable was defined as whether the child
had ARTI visits in 3 months prior to the index visit.
ARTI-related secondary diagnoses as co-morbidities were defined as the
following categories: acute otitis media, pharyngitis, bronchitis, sinusitis,
pneumonia, and URI. Other co-morbid medical diagnoses were defined as the
following categories: mental disorders, nervous system disorders, respiratory
symptoms, fever, injury and poisoning, other or no diseases.
Outcomes
Revisits are used here as the measure of patient outcomes related to
treatment for AOM. Revisits can be due to patient and/or physician concern over
verifying whether or not the child is improving. Thus, revisits are considered an
outcome not simply as a return for a worsening infection and the need for better
treatment but also as the total resources used to treat an AOM patient.
This study undertakes two major analyses. First, we derived estimates of
the factors correlated with a pediatric AOM patient filling an antibiotic prescription
within 3 days among children who were provided an antibiotic prescription
following an AOM diagnosis. Second, we document the factors correlated with a
patient’s initiation of a revisit after the initial treatment. While a host of patient
demographic and clinical factors impact revisit risk, this analysis focuses on the
impact of the potential influence of delays in filling the antibiotic prescription on
revisit risk. For treated patients, we also investigate the impact of specific
antibiotics on revisit risk, including whether or not the use of a class of antibiotic
exhibit any beneficial impact of revisit risk.
52
Analytic Methods
A multivariable logistic regression analysis was used to identify
demographic and clinical factors associated with filling an antibiotic within 3 days.
Next, we estimated a logistic regression model of the likelihood that a child
revisited a medical provider within 28 days as a function of whether or not the
patient filled an antibiotic within 3 days. We also reset the timeframe for defining
a revisit from 28 days to 7 days in an attempt to tighten the temporal relationship
of the revisit to the index visit.
A sensitivity analysis of the revisit model was also conducted using
alternative classifications of when the medication was filled: a) filled antibiotics on
day 0 of the index visit, b) filled antibiotics on day 1-2 of the index visit and c)
filled antibiotics on or after day 3.
We investigated the impact of specific antibiotic class on revisit risk using
only treated patients. The key variables in this analysis were the antibiotic class
and whether the patient filled the antibiotic within 3 days or 4-28 days after the
initial diagnosis visit. We tested for statistically significant differences across the
treated and untreated groups. T-tests were conducted for continuous variables
and chi-square tests for categorical variables. The results are shown in Table 1.
All analyses were completed using STATA/IC 13 (StataCorp, College
Station, TX).
Results
The descriptive statistics for those who filled and did not fill the antibiotic
within 28 days of AOM index visit are presented in Table 1. There were a total of
3747 pediatric patients with acute otitis media (AOM) in an outpatient setting of
which 2378 [64%] filled an antibiotic and 1369 [36.53%] did not fill an antibiotic in
the 28 days following the index visit. Of the 2378 patients who filled an antibiotic,
91.4% filled antibiotics within 3 days and 8.6% filled antibiotics in 4-28 days of
the index visit. Additionally, penicillins were the largest class of antibiotics filled
(18%). While the overall rate of revisit to the medical provider was relatively low
53
overall, the rate of revisits to the provider within 7-days or 28-days were
significantly lower for patients who filled within 3-days.
Figure 1 presents the time to fill, by day, an antibiotic for ARTIs. Nearly
85% of the antibiotics are filled within the first day and 91% are filled within first
three days of the index visit. The fill rate tapers off by the 28
th
day.
Factors associated with filling an antibiotic in 3 days of the index AOM visit
The significant factors associated with filling the antibiotic treatment fill in 3
days of the index visit derived from the constructed multivariate model are
depicted in Table 2, with the odds ratio and 95% confidence intervals as shown.
Children coded as having upper respiratory infection as ARTI related
secondary diagnosis have 23% higher odds for filling an antibiotic prescription in
3 days relative to those with no co-morbidities. Similarly, those with fever as a
comorbid condition have 49% higher likelihood of receiving an antibiotic
prescription in 3 days relative to those with no co-morbidities. Interestingly,
children who had an acute respiratory infection visit 90 days prior to the index
visit had 67% lower likelihood of filling antibiotics in 3 days.
Compared to the summer season, children diagnosed with AOM in the
winter (Adjusted OR=0.337, p=0.000), spring (Adjusted OR=0.345, p=0.000) and
fall (Adjusted OR=0.427, p=0.000) had lower likelihood of filling the antibiotic
within 3 days. Children seen by outpatient facility providers and non-physician
providers were 31% and 25% more likely to fill an antibiotic within 3 days of the
index visit while children seen by specialty physicians were 54% less likely to fill
an antibiotic prescription in 3 days than those treated by primary care physicians.
The likelihood of receiving an antibiotic in 3 days is lower among AOM
patients less than 12 years of age groups: infants less than 1 years of age, 1 to
5-years and 5 to 12-year age groups have 34%, 55% and 20% lower odds of
filling an antibiotic in 3 days, respectively, than the 12 to 18 year old group.
Regional differences in antibiotic receipt in 3 days were observed in this
study. In the multivariate regression, the Northeast Central (27%), Southeast
Central (51%), Mid Atlantic (54%), New England (28%), South Atlantic (47%) and
54
Southwest central (45%) regions had significantly lower propensity for filling an
antibiotic within 3 days compared to the reference Pacific region.
The race of the children also plays a role in antibiotic receipt. Black,
Hispanic and Asian children were 21%, 22%, 26% more likely to fill an antibiotic
within 3 days respectively, compared to white children.
The care setting for the index visit also showed significant trends.
Compared to the physician office, ER and outpatient facilities showed lower odds
of antibiotic receipt. The revisit odds for ER and outpatient facilities were 0.514
(p=0.038) and 0.778 (p=0.059) respectively. Insurance and household income
bracket did not significantly influence the risk for a revisit.
The Effects of Treatment on the Risk of a Revisit to the Physician
The results from the core multivariate analysis of the factors associated
with patient revisits within 28 days in children with AOM are presented in Table 3.
Children who filled the antibiotic within 3 days have a 57% reduced risk of a
revisit within 28 days (Adjusted OR=0.439, p=0.000) than those who filled the
antibiotic within 4-28 days or were not treated.
While clinical guidelines recommend timely antibiotic treatment for children
diagnosed with AOM, research has shown that infants are a unique age group
who can benefit from immediate antibiotics regardless of infection severity and
diagnosis uncertainty.
7
We found that children less than 1 years old have 2.44
times higher odds of initiating a revisit within 28 days of the index AOM visit than
patient over age 12. The gap between younger AOM patients and patients over
age 12 narrows as patients age.
Additional variables were found to impact the risk of a revisit within 28
days. Having an upper respiratory infection as a secondary diagnosis with AOM
has a 37% reduction in the likelihood of initiating a revisit than having no
comorbid condition. Children with pneumonia as co-morbid to AOM are 2.5 times
more likely to return for a revisit. However, those with nervous system disorders
and fever as comorbid conditions to AOM have 28% and 32% lower likelihood of
initiating a revisit than those with no comorbidities. Children who had a visit for an
55
acute respiratory infection in the previous 90 days have 3.19 times higher odds of
a revisit risk within 28 days.
Treatment by specialty physicians (Adjusted OR=1.455, p=0.001) was
associated with significantly higher odds of revisits relative to patients treated by
primary care physicians, while AOM patients seen by non-physician providers
(Adjusted OR=0.718, p=0.022), and outpatient facility providers (Adjusted
OR=0.287, p=0.000) have lower likelihood of a revisit in comparison with primary
care physicians.
Geographically, children in New England have 41% (p=0.068) and South
Atlantic have 32% (p=0.040) higher likelihood of returning for a revisit than
children in the Pacific. Compared to white children, Black children are 24% less
likely to return for a revisit while Asian children have 30% higher odds of initiating
a revisit within 28 days of the initial AOM visit.
The care setting for the index visit and the return visit also showed
significant trends. Patients who visited an outpatient hospital were 77% more
likely to return for a revisit (p =0.001) than those who had an office visit. Likewise,
season, insurance and household income bracket did not significantly influence
antibiotic receipt propensity.
Sensitivity Analysis: Revisit within 7 days
We conducted sensitivity analysis around the time to revisit (Table 4A).
When we estimated the effects of filling antibiotics within 3 days on revisit within
7 days, we found that those who filled within 3 days had 47% lower likelihood of
initiating a revisit (Adjusted OR=0.533, p=0.002). The results for revisit within 7
days and 28 days are similar in that both models showed that the group who
filled the antibiotic within 3 days had a significant reduction in likelihood of a
revisit than those who filled after 3 days.
Effects of Alternative Specifications of Time to Fill categories
We also estimated the effects of fill timing among the AOM population.
(Table 4B) The day of antibiotic fill is a determining factor in revisit risk. Children
56
receiving an antibiotic on the same day of the index visit had 62% lower risk of
initiating a revisit (p=0.000) within 28 days than the children who were not
treated. Children receiving an antibiotic on Day 1-2 of the index visit had 21%
reduction in the likelihood of initiating a revisit (but it is not significant) compared
to children who did not receive antibiotics. However, the risk of a revisit
decreases when antibiotic is filled on or after day 3. Children who filled antibiotics
on or after Day 3 were 26% less likely to have a revisit (p=0.003) relative to
children who did not fill antibiotics. Therefore, filling the antibiotic prescription
within 3 days significantly reduces the risk of a revisit.
Effects of antibiotic class among those who filled a prescription
The results of the logistic regression analysis of revisit risk among AOM
patients in Table 4C. In a model of treated and untreated AOM patients, children
filling the antibiotic within 3 days have a 60% reduced risk and those filling within
4-28 days have a 55% lower likelihood of a revisit within 28 days compared to
children who remained untreated within 28 days. The antibiotic class used does
not have a significant impact on the revisit risk. The majority of patients [58%]
were treated with penicillin for AOM and treatment with alternative antibiotics had
no statistical significant impact on the risk of a revisit.
Discussion
The 28-day revisit rate for those who filled in 3 days (3.40%) and who
remained untreated (7.16%) is small in magnitude. This finding is similar to
others: a study focusing on AOM found revisit rate 3.2% to 6.4%
25
, and another
study of AOM found recurrent visit to be 6.5% and 2.8% in two AOM groups.
26
In spite of the relatively low rate of revisits, we find that fill time plays a role
in reducing revisit risk. With imprecise and lack of diagnostics, when deciding
whether an antibiotic is necessary, providers can choose to utilize the “wait and
see” approach whereby prescribing an antibiotic to be filled within 1-3 days of the
initial visit. An antibiotic for AOM filled within 3 days had a significant reduction in
revisit risk, both for 7 and 28-day revisits relative to treating the patient after Day
57
3 or not treating the patient. Delayed antibiotics serves as a favorable alternative
strategy to prescribing antibiotics immediately, when the diagnosis or the benefits
of prescribing are unclear. This study confirms this in the particular case of acute
otitis media whereby filling an antibiotic within 3 days reduces the risk of a revisit
while delays in filling a prescription after 3 days leads to a much lower reduction
in revisit risk. Antibiotic class does not have a significant effect on the risk for a
revisit. Therefore, time to fill the antibiotic prescription is an important determining
factor in predicting revisit risk for children with AOM.
The value of delayed antibiotics is reserved for situations when, in
provider’s clinical judgment, the child is not severely ill, doesn't need immediate
treatment and there is uncertainty surrounding diagnosis or benefits of treatment.
Our results suggest that there are several factors affecting the likelihood of a
child receiving delayed antibiotics as well as subsequently returning for a revisit.
Overall, our results show that the risk of revisits decreases with age. Infants
<1years old are twice as likely to have a revisit than the 12-18 years old children.
Infants are less likely to fill an antibiotic within 3 days and have higher rate of
revisits than the children over the age of 12 years. Hence, infants may benefit
from immediate antibiotics as they are more likely to return for a revisit when the
antibiotics are not filled within 3 days of the index visit.
The type of antibiotic used in treatment for AOM had no significant effect
on the likelihood of a revisit. Penicillins are typically used as the first line
antibiotic treatment and are the largest category of antibiotics prescribed for
AOM. However, in comparison with penicillins, treatment with other antibiotics
(cephalosporins, macrolides, sulfonamides and others) does not significantly
impact the risk of a revisit.
Furthermore, our findings also suggest that children seen by specialty
physicians are more likely to have a revisit while children visiting non-physician
providers and outpatient facility providers have lower revisit risk than those seen
by primary care physicians. Specialty physicians may be seeing more severely ill
patients who needed intensive follow-up. On the other hand, primary care
physicians may be more inclined to check the progress of their patients and thus
58
recommend a revisit than those who are not designated providers for a child,
such as non-physician providers and outpatient facility providers.
We also found that the setting of the visit is also important in affecting a
revisit. The risk of revisit is 1.7 times higher when a child visits an outpatient
hospital than a physician’s office. Outpatient hospitals largely consist of acute-
care outpatient facilities where children who visit may be severely ill and need
immediate healthcare services. Additionally, the outpatient facilities and providers
may also readily recommend a return visit for patients to their primary care
physicians after visiting an outpatient facility. The revisit may be a follow-up to
the outpatient hospital as a result of an intensive initial visit and diagnosis.
There are significant regional differences in rates of risk for revisits. This is
evident in our findings that children in New England are 28% less likely to fill their
prescription within 3 days which is consistent with our observation that children in
New England also have a 41% higher likelihood of revisits than those in the
Pacific. Similarly, children in the South Atlantic region are 47% less likely to fill
their prescription within 3 days but have a 32% higher likelihood of revisits than
those in the Pacific region. These regional differences may reflect factors
associated with hospital density, provider practice style etc. that affect
prescription behavior in different parts of the country.
22
Lastly, delayed antibiotics is a viable strategy in promoting judicious use of
antibiotics without compromising health outcomes. Incidentally, delaying an
antibiotic fill by 1-3 days may serve as a cautious strategy to reduce antibiotic
use and potentially provide adequate treatment and patient satisfaction. As seen
in the case of AOM, not filling the antibiotic within 3 days leads to a rise in the
odds of a revisit which may lead to higher use of healthcare resources.
Limitations
We assume that the AOM diagnoses in our study population fulfilled
reasonable criteria for the diagnosis. However, since the diagnosis of AOM
usually depends on clinical criteria rather than a diagnostic test, there is
significant heterogeneity in how these diagnoses are assigned. The Optum
59
claims dataset only contains ICD-9 codes for diagnoses. It does not contain
infection severity and clinical symptoms, thus it is not apparent whether a treated
child presented with severe symptoms to warrant treatment. It should be noted
that lack of severity data and heterogeneity in physician treatment decisions may
affect patient outcomes.
The Optum claims dataset also does not contain physician notes with
prescriptions. Thus, we are unable to verify whether the prescription was
intended to be filled immediately or delayed for 24-72 hours by the physician.
Often, delays in filling the prescription are not intended by the physician but occur
simply because parents are unable to fill the antibiotic immediately. We
acknowledge that delays in filling the prescription can be due to both physician
intentions and patient preferences. Nonetheless, the aim of this study is to
identify the difference in outcomes for children treated with antibiotics, filled
within different time intervals.
Revisits are identified using diagnosis categories as defined by ICD-9
codes. We are unable to measure whether the revisit suggests resolution of
infection or worsening infection symptoms. We evaluated CPT codes to develop
and identify a revisit however CPT codes appear to suggest the financial
reimbursement schemes that may affect how a visit is coded. They do not clarify
whether the infection worsened/resolved or provide information about treatment
failure.
Table 1: Descriptive Statistics
Total, N=3747
Filled in 3
days N=2173
(57.99%)
Filled in 4-28
days N=205
(5.47%)
Not treated
N=1369 (36.53%)
p-value
Revisit 28days 72 (3.31%) 9 (4.39%) 98 (7.16%) 0.000
Revisit 7 days 32 (1.47%) 6 (2.93%) 38 (2.78%) 0.000
Secondary ARTI
diagnoses
0.000
AOM 35 (1.61%) 1 (0.49%) 24 (1.75%)
Pharyngitis 67 (3.08%) 7 (3.41%) 50 (3.65%)
Bronchitis 23 (1.06%) 1 (0.49%) 16 (1.17%)
Sinusitis 46 (2.12%) 1 (0.49%) 44 (3.21%)
Pneumonia 5 (0.23%) 2 (0.98%) 5 (0.37%)
URI 314 (14.45%) 24 (11.71 %) 167 (12.20%)
Co-morbid medical
diagnoses
Mental disorders 2 (0.09%) 0 3 (0.22%)
Nervous system disorders 225 (10.35%) 16 (7.80%) 101 (7.38%)
Respiratory symptoms 97 (4.46%) 15 (7.32%) 61 (4.46%)
Fever 79 (3.64%) 9 (4.39%) 34 (2.48%)
Injury and poisoning 7 (0.32%) 2 (0.98%) 8 (0.58%)
Other/None 1273 (58.58%) 127 (61.95%) 856 (62.53%)
Had previous visits (in
the prior 90 days)
14 (0.64%) 5 (2.44%) 11 (0.80%) 0.000
Season of the visit 0.000
Winter (Dec- Feb) 991 (45.61%) 110 (53.66%) 692 (50.55%)
Spring (Mar-May) 486 (22.37%) 24 (11.71%) 449 (32.80%)
Summer (Jun- Aug) 191 (8.79%) 10 (4.88%) 58 (4.24%)
Fall (Sept –Nov) 505 (23.24%) 61 (29.76%) 170 (12.42%)
Provider 0.000
Primary Physician 1623 (75.10%) 155 (75.61%) 976 (71.29%)
Specialty physician 66 (3.04%) 6 (2.93%) 116 (8.47%)
Non-physician providers 185 (8.51%) 8 (3.90%) 88 (6.43%)
Outpt facility provider 290 (13.35%) 36 (17.56%) 189 (13.81%)
Gender 0.000
Male 1095 (50.39%) 121 (59.02%) 731 (53.40%)
Female 1078 (49.61%) 84 (40.98%) 638 (46.60%)
Antibiotic classes
Penicillins 1312 (73.54%) 60 (49.18%) 0
1
st
line Cephalosporins,
sulfonamides and
macrolides
17 (0.95%) 6(4.92%) 0
61
2
nd
–5
th
line cephalosporins 227 (12.72%) 20 (16.39%) 0
Other (Lincomycin,
Quinolones, Tetracyclines)
228 (12.78%) 36 (29.51%) 0
Age (in years) 0.000
<1 74 (3.41%) 25 (12.20%) 22 (1.61%)
1- ≤5 689 (31.71%) 92 (44.88%) 435 (31.78%)
5 - ≤12 995 (45.79%) 65 (31.71%) 603 (44.05%)
12 - ≤18 415 (19.10%) 23 (11.22%) 309 (22.57%)
Household Income 0.003
Low 270 (12.43%) 24 (11.71%) 172 (12.56%0
Middle 326 (15%) 46 (22.44%) 205 (14.97%)
High 1151 (52.97%) 100 (48.78%) 751 (54.86%)
Unknown 426 (19.60%) 35 (17.07%) 241 (17.60%)
Insurance Type 0.000
EPO 242 (11.14%) 30 (14.63%) 176 (12.86%)
HMO 145 (6.67%) 12 (5.85%) 137 (10.01%)
POS 1749 (80.49%) 162 (79.02%) 1025 (74.87%)
PPO 37 (1.38%) 1 (0.49%) 31 (1.75%)
Race 0.000
White 1527 (70.27%) 148 (72.20%) 998 (72.90%)
Black 168 (7.73%) 16 (7.80%) 123 (8.98%)
Hispanic 256 (11.78%) 18 (8.78%) 133 (9.72%)
Asian 117 (5.38%) 7 (3.41%) 59 (4.31%)
Unknown 105 (4.83%) 16 (7.80%) 56 (4.09%)
Region 0.000
E N Central 416 (19.14%) 35 (17.07%) 234 (17.09%)
E S Central 58 (2.67%) 3 (1.46%) 46 (3.36%)
Mid Atlantic 100 (4.60%) 19 (9.27%) 102 (7.45%)
Mountain 278 (12.79%) 17 (8.29%) 113 (8.25%)
New England 70 (3.22%) 6 (2.93%) 45 (3.29%)
Pacific 198 (9.11%) 20 (9.76%) 98 (7.16%)
S Atlantic 425 (19.56%) 57 (27.80%) 316 (23.08%)
W N Central 315 (14.50%) 23 (11.22%) 214 (15.63%)
W S Central 313 (14.40%) 25 (12.20%) 201 (14.68%)
Setting 0.000
Office 2001 (92.08%) 189 (92.20%) 1216 (88.82%)
Clinic 3 (0.14%) 1 (0.49%) 8 (0.58%)
Outpatient hospital 127 (5.84%) 11 (5.37%) 99 (7.23%)
ER 26 (1.20%) 4 (1.95%) 42 (3.07%)
Other 16 (0.74%) 0 4 (0.29%)
Fig 1: Time to fill an antibiotic for AOM
0 20 40 60 80
Percent
0 10 20 30
Time to fill
63
Table 2: Multivariable model of factors associated with filling an antibiotic
prescription within 3 days
Filling antibiotics within 3
days (N=2173)
Odds Ratio p-value 95% CI
Secondary ARTI
diagnoses
AOM 0.962 0.842 0.659 1.404
Pharyngitis 0.967 0.812 0.737 1.268
Bronchitis 1.064 0.785 0.679 1.666
Sinusitis 0.867 0.384 0.629 1.194
Pneumonia 0.685 0.425 0.270 1.734
URI 1.233 0.003 1.075 1.413
Co-morbid medical
diagnoses
Mental disorders 0.436 0.265 0.101 1.875
Nervous system disorders 1.200 0.029 1.019 1.414
Respiratory symptoms 1.094 0.438 0.871 1.378
Fever 1.492 0.002 1.153 1.931
Injury and poisoning 0.603 0.208 0.274 1.325
Other/None 1 (Reference)
Had previous visits in
90days
0.332 0.000 0.193 0.570
Season
Winter 0.337 0.000 0.278 0.409
Spring 0.345 0.000 0.281 0.423
Summer 1 (Reference)
Fall 0.427 0.000 0.347 0.525
Provider
Primary Physician 1( Reference)
Specialty physician 0.463 0.000 0.354 0.605
Non-physician provider 1.255 0.011 1.053 1.495
Outpatient facility provider 1.316 0.003 1.096 1.581
Gender
Male 1 (Reference)
Female 1.116 0.020 1.017 1.225
Age (in years)
<1 0.664 0.004 0.503 0.876
1- ≤5 0.450 0.000 0.391 0.517
5 - ≤12 0.807 0.001 0.708 0.920
12 - ≤18 1 (Reference)
Insurance Type
EPO 0.737 0.482 0.315 1.723
HMO 0.821 0.649 0.351 1.917
64
POS 1 (Reference)
PPO 0.500 0.140 0.200 1.253
Region
E N Central 0.737 0.002 0.607 0.895
E S Central 0.494 0.000 0.358 0.683
Mid Atlantic 0.464 0.000 0.357 0.603
Mountain 0.963 0.728 0.781 1.187
New England 0.727 0.038 0.537 0.983
Pacific 1 (Reference)
S Atlantic 0.538 0.000 0.444 0.653
W N Central 0.845 0.116 0.685 1.042
W S Central 0.551 0.000 0.451 0.673
Race
White 1 (Reference)
Black 1.217 0.032 1.017 1.456
Hispanic 1.222 0.010 1.049 1.423
Asian 1.266 0.029 1.024 1.565
Unknown 1.158 0.188 0.930 1.441
Household Income
Low 1 (Reference)
Middle 1.054 0.558 0.882 1.260
High 0.932 0.352 0.803 1.081
Unknown 1.156 0.093 0.975 1.369
Setting
Office 1 (Reference)
Clinic 0.655 0.498 0.193 2.219
Outpatient hospital 0.778 0.059 0.599 1.009
ER 0.514 0.003 0.330 0.800
Other 0.963 0.895 0.552 1.679
Table 3: Multivariable model of factors associated with revisits after diagnosis
with Acute Otitis Media
Revisits (N=179) Odds Ratio p-value 95% CI
Antibiotic filled within 3 days 0.439 0.000 0.338 0.571
Age (in years)
<1 2.438 0.000 1.782 3.335
1- ≤5 1.629 0.000 1.324 2.004
5 - ≤12 1.107 0.351 0.893 1.372
12 - ≤18 1(Reference)
Secondary ARTI diagnoses
AOM 1.064 0.782 0.683 1.659
Pharyngitis 0.636 0.036 0.417 0.971
Bronchitis 1.105 0.681 0.684 1.786
Sinusitis 0.658 0.069 0.419 1.033
Pneumonia 2.579 0.004 1.364 4.876
URI 0.637 0.000 0.525 0.774
Co-morbid medical diagnoses:
Mental disorders 1.898 0.208 0.699 5.155
Nervous system disorders 0.721 0.005 0.574 0.905
Respiratory symptoms 0.741 0.053 0.547 1.003
Fever 0.681 0.050 0.464 0.999
Injury and poisoning 2.189 0.012 1.188 4.031
Other/None 1 (Reference)
Had previous visits in
90days
3.191 0.000 2.471 4.121
Season
Winter 1.507 0.039 1.021 2.223
Spring 1.192 0.388 0.799 1.780
Summer 1 (Reference)
Fall 0.872 0.511 0.581 1.309
Provider
Primary Physician 1 (Reference)
Specialty physician 1.455 0.001 1.172 1.805
Non-physician provider 0.718 0.022 0.541 0.954
Outpatient facility provider 0.287 0.000 0.203 0.405
Gender
Male 1 (Reference)
Female 0.989 0.851 0.882 1.108
Insurance Type
EPO 3.527 0.217 1.475 5.442
HMO 4.151 0.163 2.560 6.811
POS 1 (Reference)
66
PPO 4.259 0.161 3.560 6.175
Region
E N Central 1.224 0.151 0.928 1.612
E S Central 1.114 0.595 0.747 1.663
Mid Atlantic 1.220 0.213 0.891 1.671
Mountain 1.047 0.772 0.764 1.435
New England 1.410 0.068 0.975 2.041
Pacific 1 (Reference)
S Atlantic 1.320 0.040 1.013 1.721
W N Central 1.208 0.204 0.901 1.620
W S Central 1.053 0.715 0.797 1.391
Race
White 1 (Reference)
Black 0.768 0.030 0.606 0.974
Hispanic 0.922 0.455 0.747 1.139
Asian 1.301 0.029 1.026 1.650
Unknown 0.894 0.438 0.674 1.185
Household Income
Low 1 (Reference)
Middle 1.094 0.430 0.874 1.369
High 1.072 0.460 0.890 1.292
Unknown 1.167 0.158 0.941 1.446
Setting
Office 1 (Reference)
Clinic 0.225 0.046 0.052 0.974
Outpatient hospital 1.775 0.001 1.253 2.515
ER 1.625 0.027 1.057 2.500
Other 0.793 0.601 0.334 1.887
67
Table 4A: Sensitivity analysis: Revisit within 7 days of initial visit
Revisits (N=76) Adjusted OR for antibiotic
prescription
Revisit time
Revisit in 7 days (Did not fill within 3 days relative to
filled within 3 days)
0.533 (p=0.002)
Table 4B: Time to treatment categories among AOM patients who filled a
prescription
Revisits (N=179) Odds Ratio p-value 95% CI
Time to treatment
No Treatment 1 (Reference)
Day 0 0.381 0.000 0.295 0.493
Day 1-2 0.797 0.599 0.342 1.855
Day 3-28 0.745 0.003 0.615 0.903
Table 4C: Time to treatment and antibiotic class categories among AOM patients
who filled a prescription
Revisits (N=179) Odds Ratio p-value 95% CI
Time to treatment
No treatment 1 (Reference)
Within 3 days 0.407 0.000 0.318 0.520
4-28 days 0.453 0.022 0.230 0.893
Antibiotic classes
Penicillins 1 (Reference)
1
st
line Cephalosporins,
Sulfonamides and 2
nd
line
Macrolides
1.964 0.091 0.897 4.297
2
nd
–5
th
line cephalosporins 1.103 0.377 0.887 1.371
Other (Lincomycin,
Quinolones, Tetracyclines)
0.913 0.434 0.727 1.146
References
1. Hersh AL, Jackson MA, Hicks, LA et al. Principles of judicious antibiotic
prescribing for bacterial upper respiratory tract infections in
pediatrics. Pediatrics. 2013: peds-2013.
2. Hersh AL, Shapiro DJ, Pavia AT, et al. Antibiotic prescribing in ambulatory
pediatrics in the United States. Pediatrics. 2011;128:1053–1061.
3. Alzahrani, Mohammad S., et al. Factors Associated with Prescribing Broad-
Spectrum Antibiotics for Children with Upper Respiratory Tract Infections in
Ambulatory Care Settings. Clinical Medicine Insights: Pediatrics. 2018:12:
1179556518784300.
4. Fleming-Dutra, Katherine E., et al. Prevalence of inappropriate antibiotic
prescriptions among US ambulatory care visits, 2010-2011. The Journal of
American Medical Association. 2016; 315(17): 1864-1873.
5. Spurling, G KP, et al. Delayed antibiotics for respiratory infections. Cochrane
Database of Systematic Reviews. 2013; 4(4): CD004417-1.
6. Kutty, Narayanan. Treating children without antibiotics in primary healthcare.
Oman Medical Journal. 2011; 26.5: 303.
7. Lieberthal AS, Carroll AE, Chonmaitree T, et al. The diagnosis and
management of acute otitis media [published correction appears in Pediatrics.
2014;133(2):346]. Pediatrics. 2013;131(3):e964-e999.
8. American Academy of Pediatrics, Subcommittee on Management of Acute
Otitis Media. Diagnosis and management of acute otitis media. Pediatrics.
2004;113:1451–1465.
9. Broides, Arnon, et al. Parental acceptability of the watchful waiting approach in
pediatric acute otitis media. World Journal of Clinical Pediatrics. 2016; 5(2):198.
10. Grossman, Zachi, et al. Implementing the delayed antibiotic therapy
approach significantly reduced antibiotics consumption in Israeli children with first
documented acute otitis media. The Pediatric Infectious Disease Journal. 2010;
29(7): 595-599.
11. Sun, Di, T. J. McCarthy, and Danica B. Liberman. Cost-effectiveness of
watchful waiting in acute otitis media. Pediatrics. 2017: e20163086.
12. McCormick, David P., et al. Nonsevere acute otitis media: a clinical trial
comparing outcomes of watchful waiting versus immediate antibiotic treatment.
Pediatrics. 2005; 115(6): 1455-1465.
69
13. Rohrer JE , Garrison GM , Angstman KB . Early return visits by pediatric
primary care patients with otitis media: a retail nurse practitioner clinic versus
standard medical office care . Quality Management in Health Care. 2012; 21 (1):
44 – 7.
14. Pan, Q., Ornstein, S., Gross, A.J., Hueston, W.J., Jenkins, R.G., Mainous,
A.G., III, et al. Antibiotics and return visits for respiratory illness: A comparison of
pooled versus hierarchical statistical methods. The American Journal of Medical
Sciences. 2000; 319, 360-365.
15. Zhao SR, Griffin MR, Patterson BL, et al. Risk Factors for Outpatient Use of
Antibiotics in Children with Acute Respiratory Illnesses. Southern Medical
Journal. 2017;110(3):172–180.
16. Grijalva CG, Nuorti JP, Griffin MR. Antibiotic prescription rates for acute
respiratory tract infections in US ambulatory settings. The Journal of American
Medical Association. 2009;302(7):758-766.
17. NCQA. HEDIS 2016.
http://www.ncqa.org/HEDISQualityMeasurement/HEDISMeasures/HEDIS2016.
aspx. Accessed October 20, 2015.
18. Mangione-Smith, Rita, et al. The relationship between perceived parental
expectations and pediatrician antimicrobial prescribing behavior. Pediatrics.1999;
103(4): 711-718.
19. Mangione-Smith R, McGlynn EA, Elliott MN, McDonald L, Franz CE, Kravitz
RL. Parent Expectations for Antibiotics, Physician-Parent Communication, and
Satisfaction. Archives of Pediatric and Adolescent Medicine. 2001;155(7):800–
806
20. Crimmel, Beth Levin. Health Insurance Coverage and Income Levels for the
US Noninstitutionalized Population Under Age 65, 2001. Medical Expenditure
Panel Survey, Agency for Healthcare Research and Quality. 2004.
21. Centers for Disease Control and Prevention. Get smart: know when
antibiotics work. http://www.cdc.gov/getsmart. Accessed Jan 12, 2018.
22. Hicks LA, Bartoces MG, Roberts RM, et al. US Outpatient Antibiotic
Prescribing Variation According to Geography, Patient Population, and Provider
Specialty in 2011.Clinical Infectious Diseases. 2015; 60 (9):1308–1316
23. AHFS/ASHP. American Hospital Formulary Service Drug Information. 2012.
AHFS drug information. www.ahfsdruginformation.com. Accessed January 20,
2018.
70
24. Alsan, Marcella, et al. Antibiotic use in cold and Flu season and prescribing
quality: a retrospective cohort study. Medical Care. 2015. 53(12): 1066.
25. McGrath LJ, Becker-Dreps S, Pate V, Brookhart MA. Trends in Antibiotic
Treatment of Acute Otitis Media and Treatment Failure in Children, 2000–2011.
PLoS ONE. 2013, 8(12): e81210.
26. Rothman S, Pitaro J, Hackett A, Kozer E, Gavriel H, Muallem-Kalmovich L,
Eviatar E, Marom T. Treatment of Acute Otitis Media in the Pediatric Emergency
Department. The Pediatric infectious Disease Journal. 2018. 37(6): 520-525.
71
CHAPTER 5: The Impact of Adherence to HEDIS Guidelines
IMPACT OF ADHERENCE TO HEDIS MEASURES ON THE QUALITY OF
CARE FOR PEDIATRIC PATIENTS WITH PHARYNGITIS: Evidence from a
Large Health Plan
Introduction
Acute pharyngitis is among the most commonly diagnosed and treated
pediatric acute respiratory tract infections. In our previous research, pharyngitis
constituted 34% of outpatient visits among children with acute respiratory tract
infections. Group A beta-hemolytic streptococcus (GABHS) is the most common
bacterial etiology for pediatric pharyngitis, accounting for 15 to 30 percent of
cases.
1
The capacity to employ diagnostic testing to differentiate between
bacterial and viral infections and to subsequently guide appropriate antibiotic
prescribing for patients with an ARTI is limited.
2,3
Most definitive diagnostic tests
require 2-3 days to culture and incur additional costs.
4
In addition, these tests do
not provide clear guidance on which specific antibiotic is appropriate for ARTI
patients. However, this is not the case for patients with pharyngitis.
5,6,7
Unlike other acute respiratory tract infections, a rapid diagnostic test exists
for pharyngitis that allows identification of group A beta-hemolytic streptococcus
(GABHS)—which accounts for one in four children with acute sore throat.
1,4,6
Once identified, the bacterial infection can be adequately treated with specific
antibiotics. Both the American Academy of Pediatrics and the Infectious
Diseases Society of America recommend that antibiotics be prescribed for
children and adolescents (less than 18 years of age) who have a positive test for
group A streptococcus (GAS).
8,9
Testing and antibiotic treatment for group A
strep is also included in the Healthcare Effectiveness Data and Information Set
(HEDIS) standards for pediatric pharyngitis. Specifically, the HEDIS measures
for pharyngitis recommends the application of group A streptococcus test for
children age 3 months to 18 years before dispensing an antibiotic.
10
72
Treatment for group A streptococcal patients include an antibiotic likely to
eradicate the organism within 10 days. Penicillin and amoxicillin are commonly
recommended because of their narrow spectrum of activity, few adverse effects,
and modest cost. Alternative antibiotics for those with penicillin allergy include a
first-generation cephalosporin, clindamycin, clarithromycin (Biaxin), or
azithromycin (Zithromax).
1,8,11
While HEDIS measures establish the
recommendations for testing and treating for pharyngitis, the outcomes of
following these guidelines have not been studied. Particularly, there is a lack of
evidence in the effects of following the HEDIS measure of “test and treat” on
revisits in children with pharyngitis.
There are four groups of pediatric pharyngitis patients which can be
classified based on their adherence to the HEDIS measures of ‘test and treat’:
1. Patients who are not tested yet receive antibiotic prescription, apparently
based on clinician’s judgment and patient’s symptoms. Untested pharyngitis
patients may be treated with antibiotics since they are inexpensive and can
be used in a pediatric population with minimal risk to the patient. However,
treatment without establishing the etiology of the infection contributes to the
growing societal problem of antibiotic resistance.
2. Patients who are not tested and do not receive antibiotic prescription, again
apparently based on clinician judgment, parent involvement and patient
symptoms such as the infection is deemed not severe to warrant treatment
and the clinical presentation do not necessitate the group A strep test.
3. Patients who are tested and receive antibiotic prescription, apparently
because the test was positive for group A strep bacteria.
4. Patients who are tested and do not receive antibiotic prescription.
Given these four groups of patients, the main objective of this study is to
document the extent to which clinicians use diagnostic testing and antibiotic
treatment of pharyngitis according to HEDIS guidelines. We then identify factors
that are associated with following the HEDIS guidelines of testing and antibiotic
treatment. Lastly, we estimate the effects of adherence to HEDIS guidelines for
73
antibiotic use on the likelihood of a revisit for an acute respiratory tract infection
within 28 days, controlling for factors such as patient characteristics, season,
geographic region, insurance, antibiotic prescribed, provider and the
characteristics of the clinical practice treating the patient.
Methods
Data Source
The data for this study were derived from Optum Insight Clinformatics
claims database. The Optum data include claims of commercially insured
patients with both medical and pharmacy benefits from 2011-2013. These
individuals are privately insured through a large healthcare plan which either fully
insures or administers claims for the patients. Optum data are only for those
individuals who have both medical and prescription drug coverage, thus allowing
evaluation of both their medical and pharmacy claims. Optum claims database
contains several datasets: medical, member, SES (socioeconomic status) and
pharmacy.
Study Design
An episode of care was defined around the date of the first [index] visit for
a pharyngitis diagnosis (+/- 6 months). The first visit, called an index visit, is
identified for every patient diagnosed with pharyngitis from an outpatient
setting—ambulatory clinics, physician offices, emergency and urgent care
facilities. All patients diagnosed with pharyngitis, regardless of the group A strep
test application, were included in the study.
Each pharyngitis treatment episode was screened for at least a six-month
enrollment in a health insurance plan prior and subsequent to the index visit. All
patients diagnosed with pharyngitis for the 36 month period, from January 1,
2011- Dec 31, 2013 were included in the study. The medical file was linked with
the pharmacy claims. Based on whether an antibiotic was prescribed, the rate of
74
treated and untreated was calculated. As recommended by the HEDIS measure,
we also identified patients who received a group A streptococcus (GAS) test.
Using the patient identifier, medical visits were linked to the member
details and the SES file to determine insurance and socio-demographic
characteristics of the patients. The patient’s medical file was linked with the
pharmacy claims to determine antibiotic prescription associated with outpatient
pharyngitis visits. The pharmacy administrative claims dataset identifies the
medications received, specifies the date of prescription filling, NDC (National
Drug Codes) and AHFSCLSS (American Hospital Formulary Service
Classification System) codes for each medication.
Prior visits for other respiratory infections in the previous 3 months were
identified for patients with an index visit for pharyngitis. Similarly, any subsequent
visits, within 28 days of the index visit, were also recorded to measure the health
outcome for analysis. Using the unique patient identifier, medical visits were then
linked to the member details and the SES file to determine insurance and socio-
demographic characteristics of the patients.
Sample Population
Eligible individuals consist of children less than 18 years of age, who had
an index outpatient visit for a primary diagnosis of acute pharyngitis (ICD-9: 462,
463, 034.0). Up to 3 diagnoses may be recorded for any outpatient visit and the
first diagnosis recorded for the patients index ARTI-related visit is assumed to be
the primary diagnosis for the episode. Secondary diagnoses recorded on the
index visit were used to define co-morbidities present at the index visit and were
used as explanatory variables in all analyses.
Group A streptococcus (strep) test
The HEDIS measures for pharyngitis recommend using the group A
streptococcus tests for children to identify the etiology of the infection. Patients
who received the test were identified based on the CPT codes of 87070-71,
87081, 87430, 87650-52, 87880.
10
75
Antibiotic Classification
Antibiotics were grouped into classes based on classification codes from
the American Hospital Formulary Service.
19
These classes include penicillins,
cephalosporins, macrolides (Biaxin, Zithromax, Z-pack), tetracyclines,
sulfonamides, fluoroquinolones (Cipro, Levaquin, Avelox), lincomycin,
carbapenems and beta-lactam antibiotics (Amoxicillin, Amoxicillin/clavulanate
(Augmentin), Keflex, Ceftin, Omnicef).
We also defined broad-spectrum antibiotics as broad-spectrum penicillins,
second to fifth line cephalosprins, macrolides, carbapenems and beta-lactam
antibiotics. As mentioned earlier, penicillin and amoxicillin are the common
recommended treatment and alternatives include penicillin allergy include a first-
generation cephalosporin, clindamycin, clarithromycin (Biaxin), or azithromycin
(Zithromax)
Exclusion criteria
This study excludes patients with less than 6 months of continuous health
insurance enrollment before or after the index date.
Key explanatory variables
Numerous demographic and clinical variables about each patient are
included in the study to account for observed factors that affect the antibiotic
treatment decision. Data on patient age, race, gender, household income,
geographic location, practice setting type, provider specialty and type of
insurance are available from the claims file. Practice setting includes outpatient
physician offices, health centers and clinics, outpatient hospitals, emergency
department, urgent care, and ambulatory care centers. Providers include primary
care physicians, specialty care physicians, and non-physician providers. Patient
household income is obtained directly from the Optum dataset, which they
estimated using algorithms based on census block groups. Income categories
were informed by the federal guidelines for a family of four. A low-income family
76
is defined as earning <$40,000 – $50,000; middle-income family earns >$50,000
– <$75,000 and a high-income family earns $75,000 and above.
24
We also defined seasons of the year to account for time periods during
which children are most likely to visit a physician for respiratory infections, which
typically are fall and winter.
25
The months of December through February
constitute winter, March through May are included as spring, June through
August are summer and September through November are fall. Patient insurance
type is obtained from the Optum data and categorized as HMO (Health
Maintenance Organization), EPO (Exclusive Provider Organization), POS (Point
of Service), and PPO (Preferred Provider Organization). Race is defined as
White, Black, Hispanic, and Asian. Patient location is obtained from Optum and
defined according to national regions of Northeast Central, Southeast Central,
Mid Atlantic, Mountain, New England, Pacific, South Atlantic, Northwest Central
and Southwest Central. Previous visits variable was defined as whether the child
had ARTI visits in 3 months prior to the index visit.
ARTI-related co-morbidities were defined as the following categories:
acute otitis media, pharyngitis, bronchitis, sinusitis, pneumonia, and URI. Other
co-morbid medical diagnoses were defined as the following categories: mental
disorders, nervous system disorders, respiratory symptoms, fever, injury and
poisoning, other or no diseases.
Outcomes
Revisits are used here as the measure of patient outcomes related to
testing and antibiotic prescription for acute pharyngitis. Revisits can be due to
patient and/or physician concern over verifying whether or not the child is
improving. Thus, revisits are considered an outcome not simply as a return for a
worsening infection and the need for better treatment but also as the total
resources used to treat an ARTI patient.
This study measures undertakes three major analyses. First, we
document the patient and prescribing physician characteristic that impact the
decision to use GAS testing for pharyngitis. Second, we document those factors
77
that impact the decision to use antibiotic prescriptions among children who were
diagnosed with pharyngitis. Third, we identified patients who had a revisit after
the testing and prescription. While a host of patient demographic and clinical
factors impact revisit risk, this analysis focuses on adherence to HEDIS
guidelines of testing and treatment and the antibiotic class.
Analytic Methods
A multivariable logistic regression analysis was used to identify
demographic and clinical factors associated with the decisions to test and to treat
patients with pharyngitis.
Next, we estimated a multivariable logistic regression model of the
likelihood that the child experienced a revisit within 28 days and to document the
impact of adherence to testing and treatment guidelines and demographic and
clinical factors on the patient’s likelihood of initiating a revisits following starting
treatment for pharyngitis. We also conducted a sensitivity analysis with revisit
within 7 days, entered into the analysis to measure the effect of testing and
treatment on revisit.
We tested for statistically significant differences across the treated and
untreated groups. T-tests were conducted for continuous variables and chi-
square tests for categorical variables.
All analyses were completed using STATA/IC 13 (StataCorp, College
Station, TX).
Results
There were 24,685 children diagnosed with pharyngitis, of which 47%
received the group A strep test and 48% were prescribed an antibiotic
prescription for the infection. The four groups identified for analysis here were
evenly distributed (Fig 1): untested and no Rx [27%], untested and Rx [26%],
tested and Rx [22%], and tested and no Rx [24%]. Table 1 presents the
descriptive statistics for these four patient groups. The two tested groups have a
lower revisit rate than the untested groups: the tested and treated have a revisit
78
rate of 3.3%, and the tested and untreated have a revisit rate of 2.4%, while both
the untested groups have a revisit rate of nearly 5%. (Table 1)
Factors associated with receiving a Group A strep (GAS) test (Table 2)
Children with ARTI related co-morbidities of AOM, bronchitis, sinusitis,
pneumonia, and URI have 49%, 41%, 37%, 64%, 14% lower likelihood of
receiving the GAS test respectively than those with no comorbidities. Children
with fever and respiratory symptoms have 26% and 8% higher likelihood of
receiving the GAS test respectively than children with no comorbidities.
Relative to outpatient facility providers, primary care physicians have 23%
higher and specialty physicians have 61% higher odds of employing the GAS
test. Child’s age plays a significant role in receipt of the Group A strep test.
Children 1-5 years old and 5-1 2years old children are 14% and 13% more likely
to receive the test compared to children older than 12 years.
Pharyngitis patients have disproportionately higher odds of receiving a
GAS test in all regions compared to the Pacific. For instance, children in the Mid-
Atlantic have 51% higher odds of receiving a GAS test while children in New
England have 80% higher odds of receiving the same test.
Black children have 11% lower odds of receiving the GAS test compared
to white children. Both middle income and high income children have 12% and
31% higher odds of receiving the test compared to low income children.
Compared to office based visits, children visiting a clinic are twice as likely to
receive a GAS test while those seen in the ER have 44% lower odds of receiving
a GAS test. Lastly, insurance and season had no significant impact of receipt of
GAS test.
Factors associated with receiving antibiotic prescription (Table 3)
Oddly, we find that not receiving the GAS test does not have an impact on
receiving antibiotic prescription (Adjusted OR=1.055, p=0.067).
Children with ARTI related comorbidities of AOM and sinusitis as
comorbidities have 25% and 46% higher likelihood of being prescribed an
79
antibiotic respectively compared to children with no comorbidities. Children with
URI have 18% lower likelihood of being prescribed an antibiotic respectively
compared to children with no comorbidities. Additionally, relative to primary care
physicians, children visiting non-physician providers for pharyngitis have 12%
higher odds of being prescribed an antibiotic.
Child’s age plays a role in receiving antibiotics. Children 1-5 years old and
5-1 2years old children are 57% and 68% more likely to being prescribed an
antibiotic compared to children older than 12 years. Geographically, there is
regional variation in antibiotic prescription as well. Children in South Atlantic and
Southeast Central have 24% and 18% lower odds of being prescribed an
antibiotic respectively than pharyngitis patients in the Pacific.
Black children have 10% lower odds of being prescribed an antibiotic
compared to white children. Compared to office based visits, children visiting a
clinic are 39% less likely to being prescribed an antibiotic. Likewise, household
income, insurance type, and season have no significant impact on revisit risk.
Effects of Group A strep test and antibiotic prescription on revisits (Table 4)
Receiving a GAS test reduces the odds of a revisit by 39% (p=0.000)
while being prescribed an antibiotic had no significant impact on revisit rates,
controlling for comorbidities, antibiotic prescribed, age, gender, race, household
income, season, geographic region, insurance, provider and the characteristics
of the clinical practice treating the patient.
Pharyngitis patients with upper respiratory infection as comorbidity have
31% (p=0.021) lower likelihood of a revisit than those with no comorbidities.
Patients visiting non-physician providers and outpatient facility providers are 25%
(p=0.066) and 72%(p=0.000) less likely to have a revisit than those visiting
primary care physicians.
Age is a risk factor in likelihood of initiating a revisit. Children 1-5 years old
have 39% higher odds of a revisit than children aged over 12 years. Children
80
visiting in the fall are 24% less likely to have a revisit than those who have a
pharyngitis diagnosis in the summer.
Relative to the Pacific, children in the North Eastern Central region have
29% (p=0.07) lower odds of a revisit. Compared to whites, black children are
25% (p=0.03) less likely to have a revisit. The care setting also has a significant
impact on revisit risk. Children visiting outpatient hospital and other care settings
have 63% (p=0.019) and 71% (p=0.008) higher likelihood of a revisit risk than
those visiting a physician’s office. Lastly, household income bracket and
insurance type has no significant impact on revisit risk.
Sensitivity Analysis: Revisit within 7 days
When the outcome related to time to revisit was redefined to visits within 7
days of the initial visit, the impact of GAS testing became more significant. (Table
5) Children testing for group A strep have a 73% lower likelihood of returning for
a revisit within 7 days (Adjusted OR=0.278, 0=0.000) while those receiving a
group A strep test have 39% lower risk of a revisit in 28 days (Adjusted
OR=0.612, p=0.000). Thus, the 7-day revisit model has a larger impact in
reduction of revisit risk than the 28-day revisit model.
The effect of antibiotic prescription remains not significant in both the 28-day and
7-day revisit models.
Effects of Antibiotic class on Revisit
The effect of antibiotic class was also examined in Table 6. Patients
treated with lincomycin, quinolones and tetracyclines are twice as likely to have a
revisit than those prescribed penicillins. Moreover, children who were prescribed
2
nd
-5
th
line cephalosporins also had 36% higher odds of experiencing a revisit
than children who received penicillins.
81
Discussion
An analysis of the impact of GAS testing and antibiotic prescribing
patterns on the risk of a revisit provides useful information to help guide clinicians
and health systems in their efforts to improve antibiotic prescribing and clinical
practice. Our study results suggest that, overall, GAS testing is an important
factor in reducing the revisit risk. Being prescribed an antibiotic, on its own, does
not have a significant impact on the risk of a revisit.
We analyzed the combined effect of testing and antibiotic prescription via
an interaction model (Table 7). As before, antibiotic prescription doesn’t affect the
rate of revisit by any measurable degree. An antibiotic prescription informed by
precise diagnostic testing is the optimal strategy to reduce revisit risk: testing
decreases the risk of a revisit by 47% [p=0.000, CI: 0.433, 0.657] while the use of
GAS test and antibiotic combined has 70% lower odds of a revisit [p=0.06, CI:
0.088, 0.570].
Encouragingly, when used in conjunction, GAS test and antibiotic
prescription reduce the likelihood of a revisit. Prior studies have reported variable
rates of testing among children with pharyngitis prescribed an antibiotic, ranging
from 23% to 84%.
14,15
Our analysis suggests that performing well on the testing
quality measure is associated with a reduced risk of revisit among children
diagnosed with pharyngitis. Our study results also suggest that when antibiotics
are prescribed following GAS test, the revisit risk is significantly reduced.
Antibiotics can lower revisit risk when informed by diagnostic testing tools that
have the capacity to determine the infectious organism.
Our results also suggest that there are several factors affecting the
likelihood of a child receiving the GAS test. Children 1-5 years old are
significantly more likely to receive the GAS test, more likely to be prescribed an
antibiotic as well as more likely to initiate a revisit. The recommendations for
pharyngitis treatment suggest testing and treatment for children above 3 years
old. Additionally, specialty care physicians are less inclined to implement the
GAS test because the children visiting them may be sufficiently sick to warrant an
antibiotic without the test. Black and low-income children have lower odds of
82
receiving the test, possibly due to awareness about health care. Children in the
Pacific region, compared to the entire nation, are the least likely to receive a GAS
test.
In addition to receiving the diagnostic GAS test, there are multitude of
factors that affect the revisit risk. Young children less than age 5 are more likely
to initiate a revisit compared to older children as parents and physicians are more
cautious with younger children. As seen above, black children are not only less
likely to receive a GAS test but they are also less likely to initiate a revisit.
Patients visiting non-physician providers and outpatient facility providers are less
likely to have a revisit than those visiting primary care physicians. Children
seeing non-physician providers and outpatient facility providers may be there for
single visit, perhaps for a severe infection, or may lack a usual source of care,
thus reducing the likelihood of a revisit.
Adherence to HEDIS measures for the treatment of pharyngitis vary
nationally. Compared to the Pacific region, children in the Mid-Atlantic states
have higher odds of receiving a GAS test, but the rate of treatment and revisit
does not vary significantly. The New England region has significantly higher odds
of employing the GAS test, but the odds of prescribing an antibiotic and likelihood
of initiating a revisit do not vary significantly from other regions of the country.
Lastly, there is significant value in performing well on the HEDIS measure
that recommends the use of rapid diagnostic tools in determining if the infection
is bacterial, thus informing appropriate prescription decisions. Diagnostic tools in
conjunction with adequate antibiotic have the potential to reduce the revisit risk
for pharyngitis patients which is already reasonably low at xx% in untreated
patients. As much as judicious employment of antibiotics is important in
managing infection and curbing complications, use of rapid diagnostic tools is the
determining factor in reducing revisits for pediatric pharyngitis. The risk of a
revisit for a pharyngitis patient remains greater among those prescribed an
antibiotic suggesting the employment of effective antibiotic is necessary. Lastly,
the decision to return for a revisit based on ineffective antibiotic can be averted if
optimal testing and treatment were employed.
Limitations
The Optum claims dataset only contains ICD-9 codes for diagnoses. It
does not contain infection severity and clinical symptoms, thus it is not apparent
whether a treated child presented with severe symptoms to warrant treatment. It
should be noted that lack of severity data and heterogeneity in physician
treatment decisions may affect patient outcomes. Similarly, medical claims about
GAS testing were obtained from the CPT codes within the Optum claims dataset,
we are unable to verify the test results. Since GAS test results would most likely
drive treatment decisions, lack of this information is a weakness in determining
whether the treatment decision was driven by a positive/negative test as this may
affect patient outcomes.
Revisits are identified using diagnosis categories as defined by ICD-9
codes. We are unable to measure whether the revisit suggests resolution of
infection or worsening infection symptoms. We evaluated CPT codes to develop
and identify a revisit however CPT codes appear to suggest the financial
reimbursement schemes that may affect how a visit is coded. They do not clarify
whether the infection worsened/resolved or provide information about treatment
failure.
Figure 1: Four patient groups
Diagnosed with
pharyngitis
24,685
Received a group
A test
11,549 (46.78%)
Antibiotic
prescribed
5409 (46.83%)
No Antibiotic
prescribed 6140
(53.16%)
Did not receive a
group A test
13,136 (53.21%)
Antibiotic
prescribed
6490 (49.40%)
No Antibiotic
prescribed 6646
(50.59%)
Table 1: Descriptive Statistics
Group 1
(Tested and
Rx)
Group 2
(Tested but
no Rx)
Group 3
(Not tested
and Rx)
Group 4 (Not
tested and
no Rx)
Chi-
square
test
p-
value
Total patients
(24,685)
5409 6140 6490 6646
REVISIT 28 days 179 (3.3%) 151 (2.4%) 325 (5%) 326 (5%) 0.000
Male 2451 (45%) 2967 (48%) 3085 (47%) 3204 (48%) 0.004
Female 2958 (55%) 3173 (52%) 3405 (53%) 3442 (52%) 0.004
Age (in years) 0.000
<1 5 (0.09%) 14 (0.22%) 20 (0.30%) 22 (0.33%)
1- ≤5 721 (13%) 706 (11%) 848 (13%) 764 (11%)
5 - ≤12 3057 (56%) 2762 (45%) 3493 (53%) 2933 (44%)
12 - ≤18 1626 (30%) 2658 (43%) 2139 (33%) 2927 (44%)
Household
Income
0.000
Low 489 (9%) 549 (9%) 839 (13%) 854 (13%)
Middle 678 (12%) 699 (11%) 879 (13%) 959 (14%)
High 3447 (64%) 4025 (65%) 3735 (57%) 3711 (56%)
Unknown 795 (14%) 867 (14%) 1037 (16%) 1122 (17%)
Season of the
visit
0.000
Winter (Dec- Feb) 2410 (44%) 2810 (46%) 2824 (43%) 2965 (44%)
Spring (Mar-May) 1389 (26%) 1656 (27%) 1665 (26%) 1654 (25%)
Summer (Jun-
Aug)
262 (4.8%) 313 (5%) 366 (5.6%) 376 (5.6%)
Fall (Sept –Nov) 1348 (25%) 1361 (22%) 1635 (25%) 1651 (25%)
Provider 0.000
Primary Physician 3023 (56%) 3098 (50%) 4347 (67%) 4333 (65%)
Specialty
physician
124 (2%) 143 (2%) 266 (4%) 283 (4%)
Non-physician
providers
280 (5%) 251 (4%) 416 (6%) 381 (6%)
Outpatient care
provider
1982 (37%) 2647 (43%) 1459 (22%) 1648 (25%)
ARTI related
diagnoses
0.000
AOM 83 (1.5%) 62 (1%) 208 (3.2%) 168 (2.5%)
Bronchitis 19 (0.35%) 19 (3.4%) 45 (0.69%) 42 (0.63%)
Sinusitis 54 (1%) 60 (1%) 146 (2.2%) 99 (1.49%)
Pneumonia 1 (0.02%) 5 (0.08%) 10 (0.15%) 9 (0.14%)
URI 241 (4.5%) 320 (5.2%) 360 (5.5%) 458 (6.8%)
86
Co-morbid
medical
diagnoses
Mental disorders 7 (0.13%) 18 (0.29%) 8 (0.12%) 17(0.26%)
Nervous system
disorders
55 (1%) 61 (1%) 91 (1.4%) 118(1.8%)
Respiratory
symptoms
81 (1.5%) 110 (1.8%) 217 (3.3%) 264 (4%)
Fever 189 (3.5%) 212 (3.5%) 468 (7.2%) 455 (7%)
Injury and
poisoning
19 (0.35%) 17 (0.28%) 16 (0.25%) 37 (0.5%)
Other/None 4388 (86%) 5044 (85%) 4206 (74%) 4288 (74%)
Had previous
visits (in the
prior 90 days)
608 (22%) 589 (21%) 828 (30%) 759 (27%)
0.000
Antibiotic
classes
0.132
Penicillins 3059 (56%) 0 3618 (56%) 0
1
st
line
Cephalosporins
365 (6.7%) 0 405 (6.2%) 0
Sulfonamides 135 (2.5%) 0 154 (2.3%) 0
2
nd
line
Macrolides
1069 (20%) 0 1391 (21%) 0
2
nd
– 5
th
line
cephalosporins
662 (12%) 0 754 (11.6%) 0
Other 119 (2%) 0 168 (2.5%) 0
Insurance type 0.000
EPO 672 (12%) 825 (13.4%) 760 (12%) 870 (13%)
HMO 554 (10%) 609 (9.9%) 484 (7.8%) 420 (6%)
POS 4076 (75%) 4582 (75%) 5099 (78%) 5207 (78%)
PPO 107 (2.0%) 124 (1.28%) 147 (2.18%) 149 (2.14%)
Race 0.001
White 271 (5%) 322 (5%) 294 (4.5%) 290 (4.3%)
Black 4057 (75%) 4508 (73%) 4777 (74%) 4854 (73%)
Hispanic 377 (7%) 513 (8%) 539 (8.3%) 621 (9%)
Asian 492 (9%) 588 (9.5%) 648 (10%) 650 (9.8%)
Unknown 212 (4%) 209 (3.4%) 232 (3.5%) 231 (3.5%)
Region 0.000
E N Central 784 (14%) 796 (13%) 1203 (18%) 1097 (16%)
E S Central 170 (3%) 202 (3%) 272 (4%) 291 (4%)
Mid Atlantic 486 (9%) 524 (8.5%) 500 (7.7%) 479 (7%)
Mountain 591 (11%) 638 (10%) 626 (10%) 602 (9%)
New England 170 (3%) 177 (3%) 212 (3%) 222 (3%)
87
Pacific 192 (3.5%) 188 (3%) 269 (4%) 269 (4%)
S Atlantic 1448 (27%) 1927 (31%) 1461 (22%) 1759 (26%)
W N Central 787 (15%) 764 (12%) 850 (13%) 755 (11%)
W S Central 781 (14%) 924 (15%) 1097 (17%) 1172 (18%)
Setting 0.000
Office 1783 (33%) 2413 (39%) 175 (3%) 233 (3.5%)
Clinic 3550 (65%) 3635 (59%) 5328 (82%) 5304 (80%)
Outpatient
hospital
24 (0.44%) 38 (0.61%) 23 (0.37%) 24 (0.36%)
ER 18 (0.33%) 21 (0.34%) 892 (14%) 1001 (15%)
Other 34 (0.62%) 33 (0.53%) 72 (1%) 84 (1%)
Table 2: Multivariable model of factors associated with receiving a GAS test
Group A strep test (N=11,549) Odds Ratio p-value 95% CI
Secondary ARTI diagnoses
AOM 0.517 0.000 0.422 0.633
Bronchitis 0.594 0.010 0.400 0.881
Sinusitis 0.633 0.000 0.503 0.797
Pneumonia 0.368 0.040 0.142 0.955
URI 0.867 0.019 0.770 0.976
Co-morbid medical diagnoses
Mental disorders 1.369 0.296 0.759 2.467
Nervous system disorders 0.894 0.017 0.590 0.949
Respiratory symptoms 1.081 0.000 0.459 0.660
Fever 1.260 0.000 0.572 0.745
Injury and poisoning 1.015 0.692 0.583 1.429
Other/None 1 (Reference)
Had previous ARTI visit in 90
days
1.000 0.999 0.916 1.091
Season
Winter 1.018 0.783 0.894 1.160
Spring 1.044 0.528 0.912 1.195
Summer 1 (Reference)
Fall 0.920 0.236 0.802 1.055
Provider
Primary Physician 1.234 0.000 1.097 1.389
Specialty physician 0.618 0.000 0.498 0.765
Non-physician providers 1.131 0.128 0.964 1.328
Outpatient facility provider 1( Reference)
Other facility provider 0.045 0.008 0.004 0.449
Gender
Male 1 (Reference)
Female 1.011 0.686 0.955 1.071
Age (in years)
<1 0.765 0.415 0.402 1.456
1- ≤5 1.146 0.006 1.040 1.262
5 - ≤12 1.135 0.000 1.066 1.209
12 - ≤18 1 (Reference)
Insurance Type
EPO 0.610 0.110 0.333 1.117
HMO 0.811 0.496 0.445 1.479
POS 1 (Reference)
PPO 0.616 0.135 0.327 1.162
Region
E N Central 1.375 0.000 1.156 1.636
89
E S Central 1.193 0.104 0.964 1.476
Mid Atlantic 1.511 0.000 1.259 1.825
Mountain 1.334 0.002 1.113 1.600
New England 1.797 0.000 1.424 2.268
Pacific 1 (Reference)
S Atlantic 1.436 0.000 1.215 1.698
W N Central 1.660 0.000 1.388 1.986
W S Central 1.153 0.104 0.971 1.369
Race
White 1 (Reference)
Black 0.893 0.047 0.800 0.998
Hispanic 0.937 0.223 0.845 1.039
Asian 0.919 0.298 0.786 1.076
Unknown 1.098 0.179 0.957 1.259
Household Income
Low 1 (Reference)
Middle 1.129 0.045 1.002 1.272
High 1.316 0.000 1.194 1.450
Unknown 1.105 0.089 0.985 1.240
Setting
Office 1 (Reference)
Clinic 2.017 0.002 1.302 3.125
Outpatient hospital 0.038 0.000 0.027 0.053
ER 0.569 0.000 0.418 0.775
Other 17.079 0.000 14.68 19.862
90
Table 3: Multivariable model of factors associated with receiving antibiotic
prescription
Received antibiotic (N=11,899) Odds Ratio p-value 95% CI
Not receiving GAS test 1.055 0.067 0.996 1.118
Secondary ARTI diagnoses
AOM 1.256 0.012 1.051 1.502
Bronchitis 1.111 0.562 0.777 1.588
Sinusitis 1.468 0.000 1.185 1.819
Pneumonia 0.877 0.748 0.395 1.952
URI 0.824 0.001 0.736 0.923
Co-morbid medical diagnoses
Mental disorders 0.467 0.015 0.253 0.863
Nervous system disorders 0.847 0.147 0.677 1.059
Respiratory symptoms 0.855 0.053 0.730 1.001
Fever 1.013 0.815 0.903 1.137
Injury and poisoning 0.692 0.095 0.449 1.066
Other/None 1 (Reference)
Had previous ARTI visit in 90
days
1.070 0.101 0.986 1.161
Season
Winter 0.991 0.882 0.882 1.113
Spring 1.012 0.844 0.897 1.141
Summer 1 (Reference)
Fall 1.055 0.385 0.934 1.192
Provider
Primary Physician 1 (Reference)
Specialty physician 0.995 0.956 0.852 1.162
Non-physician providers 1.125 0.044 1.002 1.263
Other facility provider 0.973 0.633 0.873 1.085
Gender
Male 1 (Reference)
Female 1.080 0.003 1.026 1.136
Age (in years)
<1 0.605 0.105 0.329 1.110
1- ≤5 1.573 0.000 1.447 1.711
5 - ≤12 1.688 0.000 1.597 1.784
12 - ≤18 1 (Reference)
Insurance Type
EPO 0.798 0.436 0.452 1.408
HMO 0.957 0.879 0.544 1.683
POS 1 (Reference)
PPO 0.753 0.347 0.416 1.360
Region
91
E N Central 1.012 0.864 0.874 1.173
E S Central 0.829 0.049 0.688 0.999
Mid Atlantic 0.943 0.472 0.804 1.105
Mountain 0.957 0.580 0.820 1.117
New England 0.944 0.567 0.777 1.147
Pacific 1 (Reference)
S Atlantic 0.769 0.000 0.668 0.886
W N Central 0.996 0.960 0.854 1.161
W S Central 0.861 0.046 0.744 0.997
Race
White 1 (Reference)
Black 0.903 0.038 0.821 0.994
Hispanic 0.986 0.759 0.902 1.078
Asian 1.024 0.733 0.892 1.175
Unknown 0.972 0.642 0.862 1.095
Household Income
Low 1 (Reference)
Middle 0.984 0.774 0.887 1.092
High 0.978 0.612 0.899 1.064
Unknown 0.961 0.436 0.869 1.062
Setting
Office 1 (Reference)
Clinic 0.619 0.027 0.405 0.946
Outpatient hospital 0.898 0.129 0.782 1.031
ER 0.818 0.163 0.617 1.084
Other 0.815 0.001 0.722 0.922
92
Table 4: Multivariable model of factors associated with revisits after diagnosis
with acute pharyngitis
Revisit (n=981) Odds Ratio p-value 95% CI
Group A strep test 0.612 0.000 0.527 0.710
Antibiotic prescribed 1.094 0.175 0.960 1.247
Secondary ARTI diagnoses
AOM 0.795 0.321 0.507 1.249
Bronchitis 0.515 0.260 0.162 1.633
Sinusitis 0.861 0.583 0.507 1.464
Pneumonia 0.925 0.940 0.123 6.915
URI 0.694 0.021 0.508 0.947
Co-morbid medical
diagnoses
Mental disorders 1.349 0.624 0.407 4.475
Nervous system disorders 1.193 0.474 0.734 1.940
Respiratory symptoms 0.899 0.597 0.608 1.331
Fever 1.027 0.845 0.783 1.346
Injury and poisoning 1.361 0.509 0.544 3.403
Other/None 1 (Reference)
Had previous ARTI visit in
90 days
1.000 0.999 0.916 1.091
Provider
Primary Physician 1( Reference)
Specialty physician 1.216 0.235 0.880 1.681
Non-physician providers 0.751 0.066 0.553 1.019
Outpatient facility provider 0.284 0.000 0.198 0.407
Other facility provider 4.605 0.198 0.449 47.19
Gender
Male 1 (Reference)
Female 0.984 0.811 0.864 1.120
Age (in years)
<1 1.648 0.362 0.562 4.831
1- ≤5 1.392 0.001 1.139 1.700
5 - ≤12 1.027 0.716 0.888 1.188
12 - ≤18 1 (Reference)
Season
Winter 1.025 0.865 0.764 1.375
Spring 1.015 0.924 0.748 1.377
Summer 1 (Reference)
Fall 0.762 0.090 0.557 1.043
Insurance Type
EPO 0.418 0.116 0.140 1.240
HMO 0.515 0.228 0.175 1.513
93
POS 1 (Reference)
PPO 0.691 0.523 0.222 2.146
Region
E N Central 0.715 0.074 0.495 1.032
E S Central 0.774 0.286 0.484 1.238
Mid Atlantic 1.128 0.530 0.773 1.647
Mountain 0.932 0.719 0.637 1.364
New England 1.008 0.973 0.628 1.618
Pacific 1 (Reference)
S Atlantic 0.890 0.510 0.630 1.257
W N Central 0.789 0.223 0.540 1.154
W S Central 0.952 0.787 0.666 1.359
Race
White 1 (Reference)
Black 0.754 0.037 0.578 0.983
Hispanic 0.823 0.114 0.647 1.047
Asian 0.951 0.780 0.672 1.346
Unknown 1.281 0.078 0.972 1.688
Household Income
Low 1 (Reference)
Middle 1.064 0.633 0.823 1.375
High 0.967 0.758 0.783 1.194
Unknown 0.905 0.447 0.702 1.168
Setting
Office 1 (Reference)
Clinic 0.427 0.174 0.125 1.452
Outpatient hospital 1.637 0.019 1.104 2.499
ER 1.073 0.862 0.488 2.430
Other 1.717 0.008 1.187 2.617
94
Table 5: Sensitivity analysis: Revisit model with 7 day revisits
Revisits
Adjusted OR
Revisit 7 days
Received Group A strep test 0.278 (p=0.000)
Antibiotic prescribed 1.001 (p=0.992)
Table 6: 28-day Revisit model with antibiotic class within treated patients only
Revisit (n=981) Odds Ratio p-value 95% CI
Antibiotic classes
Penicillins 1 (Reference)
1
st
line Cephalosporins 0.863 0.478 0.576 1.294
Sulfonamides 1.350 0.293 0.771 2.365
2
nd
line Macrolides 0.967 0.794 0.756 1.237
2
nd
– 5
th
line
cephalosporins
1.366 0.023 1.044 1.789
Other (lincomycin,
quinolones and
tetracyclines)
2.015 0.006 1.228 3.306
Table 7: Revisit model with GAS test + Antibiotic combined
Revisit 28 days (n=981) Adjusted Odds Ratio
GAS test 0.533
Antibiotic prescribed 1.001
GAS + Antibiotic 0.299
95
References
1. Choby, Beth A. Diagnosis and treatment of streptococcal pharyngitis.
American Family Physician. 2009;79(5): 383-390.
2. Briel, Matthias, et al. Procalcitonin-guided antibiotic use vs a standard
approach for acute respiratory tract infections in primary care. Archives of
Internal Medicine. 2008;168(18): 2000-2007.
3. Maltezou, Helen C., et al. Evaluation of a rapid antigen detection test in the
diagnosis of streptococcal pharyngitis in children and its impact on antibiotic
prescription. Journal of Antimicrobial Chemotherapy. 2008;62(6): 1407-1412.
4. Neuner, Joan M., et al. Diagnosis and management of adults with pharyngitis:
a cost-effectiveness analysis. Annals of Internal Medicine. 2003;139(2): 113-122.
5. Gerber, Michael A., et al. Prevention of rheumatic fever and diagnosis and
treatment of acute Streptococcal pharyngitis: a scientific statement from the
American Heart Association Rheumatic Fever, Endocarditis, and Kawasaki
Disease Committee of the Council on Cardiovascular Disease in the Young, the
Interdisciplinary Council on Functional Genomics and Translational Biology, and
the Interdisciplinary Council on Quality of Care and Outcomes Research:
endorsed by the American Academy of Pediatrics. Circulation. 2009;119(11):
1541-1551.
6. Gieseker, Karen E., et al. Evaluating the American Academy of Pediatrics
diagnostic standard for Streptococcus pyogenes pharyngitis: backup culture
versus repeat rapid antigen testing. Pediatrics. 2003;111(6): e666-e670.
7. Shapiro, Daniel J., et al. Viral features and testing for streptococcal
pharyngitis. Pediatrics. 2017: e20163403.
8. Shulman, S, Bisno, A., Clegg, H., Gerber, M., Kaplan, E., Lee, G. et al. Clinical
Practice Guideline for the Diagnosis and Management of Group A Streptococcal
Pharyngitis: 2012 Update by the Infectious Diseases Society of America. Clinical
Infectious Diseases. 2012; 55: e86–e102.
9. Mangione-Smith R, McGlynn EA, Elliott MN, McDonald L, Franz CE, Kravitz
RL. Parent Expectations for Antibiotics, Physician-Parent Communication, and
Satisfaction. Archives of Pediatric Adolescent Medicine. 2001;155(7):800–806
10. NCQA. HEDIS 2016.
http://www.ncqa.org/HEDISQualityMeasurement/HEDISMeasures/HEDIS2016.a
spx. Accessed October 20, 2015.
96
11. Linder, Jeffrey A., et al. Antibiotic treatment of children with sore throat. The
Journal of American Medical Association. 2005; 294(18): 2315-2322.
12. Crimmel, Beth Levin. Health Insurance Coverage and Income Levels for the
US Noninstitutionalized Population Under Age 65, 2001. Medical Expenditure
Panel Survey, Agency for Healthcare Research and Quality. 2004.
13. Alsan, Marcella, et al. Antibiotic use in cold and Flu season and prescribing
quality: a retrospective cohort study. Medical care. 2015. 53(12): 1066.
14. Mainous III, Arch G., et al. Streptococcal diagnostic testing and antibiotics
prescribed for pediatric tonsillopharyngitis. The Pediatric Infectious Disease
Journal.1996;15(9): 806-810.
15. Benin, Andrea L., et al. Improving diagnostic testing and reducing overuse of
antibiotics for children with pharyngitis: a useful role for the electronic medical
record. The Pediatric Infectious Disease Journal. 2003;22(12): 1043-1047.
16. Ayanruoh, Steve, et al. Impact of rapid streptococcal test on antibiotic use in
a pediatric emergency department. Pediatric Emergency Care. 2009;25(11): 748-
750.
17. Bird, Chris, et al. A Pragmatic Study to Evaluate the Use of a Rapid
Diagnostic Test to Detect Group A Streptococcal Pharyngitis in Children With the
Aim of Reducing Antibiotic Use in a UK Emergency Department. Pediatric
Emergency Care. 2018.
18. Alsan, Marcella, et al. Antibiotic use in cold and Flu season and prescribing
quality: a retrospective cohort study. Medical Care. 2015; 53(12): 1066.
19. AHFS/ASHP. American Hospital Formulary Service Drug Information. 2012.
AHFS drug information. www.ahfsdruginformation.com. Accessed January 20,
2018.
97
CHAPTER 6: Discussion
Overall, receiving an antibiotic within 3 days of the index visit results in a
lower likelihood of a revisit. Irrespective of the infection, the day of antibiotic
receipt plays a crucial role in reducing revisit risk for ARTI patients in outpatient
settings. Chapter 3 investigates time to treatment and risk of a revisit among both
the treated and untreated pediatric patients diagnosed with ARTIs. Treatment
timing is significant in reducing the likelihood of a revisit. The results suggest that
receipt of antibiotics within 3 days leads to a 51% reduction in likelihood of
initiating a revisit. Sensitivity analyses show that receiving an antibiotic within a
day of the index visit had 43% lower risk of initiating a revisit within 28 days than
the children who were not treated. Children receiving an antibiotic on Day 2-3 of
the index visit had 34% reduction in the likelihood of initiating a revisit compared
to children who did not receive antibiotics. This finding implies that receipt of an
antibiotic within 3 days hours when intended as necessary treatment, is a
pathway to reduce the revisit risk for a patient.
Chapter 4 investigates the effects of treatment timing on the risk of a
revisit in patients treated with AOM. As found in the case of ARTIs, filling an
antibiotic within 3 days has a reduction in revisit risk than compared to children
who filled after 3 days. Children who filled the antibiotic within 3 days have a 57%
reduced risk of a revisit within 28 days than those who received the antibiotic
within 4-28 days or were not treated. This suggests that prescribing antibiotics
with clear instructions to fill within 3 days is a favorable prescribing strategy for
AOM. Providers can utilize the watchful waiting approach with wait up to 3 days
of the antibiotic as a strategy towards judicious antibiotic use without
compromising revisit risks for pediatric ARTI patients.
Chapter 5 evaluates the HEDIS quality measure of “test and treat” for
acute pharyngitis. In assessing the quality of “test and treat” measure, the results
suggest that diagnostic tools in conjunction with adequate treatment have the
capacity to reduce revisit risk. It is imperative, then, to use diagnostic tools in
determining pharyngitis to inform proper treatment decisions. Group A Strep
98
testing is an important factor in reducing the revisit risk while antibiotic treatment,
on its own, does not have a significant impact on the risk of a revisit. The study
found that when antibiotics are prescribed following GAS test, there is a
reduction in revisit risk. Therefore, antibiotic treatment, as informed by diagnostic
testing tools that have the capacity to determine the infectious organism results
in lower revisit risk.
Lastly, accurate diagnosis and treatment for acute respiratory tract
infections (ARTIs) is a reasonable goal, yet many factors affect a clinician’s
ability to achieve it. Lack of precise diagnostic tools, cheap and relatively safe
antibiotics, and provider prescribing behaviors all contribute to overuse and
misuse of antibiotics in the face of uncertain diagnoses. Furthermore, the rise of
antibiotic-resistance demands appropriate and judicious use of antibiotics.
Antibiotic receipt within 3 days leads to a lower likelihood of a revisit. Providers
may recommend antibiotic treatment, in cases deemed necessary, with fill within
3 days of the initial visit.
Conclusion
In conclusion, this research suggests that antibiotic prescription can
markedly reduce revisit risk for acute respiratory tract infections in children when
filled within a certain time frame. Receiving an antibiotic within 3 days of the
index visit leads to a decline in risk of a revisit within 28 days of the diagnosis.
Delayed antibiotics, using the wait and see approach to be filled within 3 days, is
a viable strategy in promoting judicious use without affecting health outcomes
negatively. Additionally, diagnostic testing tools as an adjunct to antibiotic
treatment is a superior combination in reducing revisit risk, than prescribing an
antibiotic without determining the infectious organism. Development of improved
diagnostic tools, thus, needs to be promoted as a potential path for adequate
antibiotic use and to curb antibiotic-resistance.
Moreover, additional research is needed in the outcomes for antibiotic
use, especially revisits. To detail the cause for revisit, information beyond
diagnoses is needed to decipher whether the visit was necessary—i.e. a visit for
99
an unresolved infection or an output of the health system that may require
revisits as general follow-up. In the face of uncertain diagnoses and rising
antibiotic resistance, documenting revisits after antibiotic use and learning more
about cause for revisits has implications for acute respiratory tract infection
treatment with antibiotics.
Additional research is also needed to understand the impact of
interventions like antibiotic stewardship programs and HEDIS measures on
antibiotic use in outpatient settings across the U.S. Such programs and quality
measures promote judicious antibiotic use, thus the effects having such
interventions may have a positive effect on revisits following antibiotic use over
time. In general, evaluating the impact of antibiotic use is a formidable pathway
to understand and affect adequate treatment practices albeit in the face of
uncertain diagnoses and antibiotic-resistant organisms.
Abstract (if available)
Abstract
Pediatric acute respiratory tract infections (ARTIs) are commonly treated with antibiotics. Antibiotic treatment for acute respiratory tract infections typically lasts 7 to 28 days, and patients who fail treatment usually return within 28 days of their initial visit. Antibiotics can be prescribed for immediate use or as delayed prescription. Prescribing antibiotics in an outpatient setting is often a complex interaction between patients and physicians. Nonetheless, utilization of antibiotics appropriately and timely, can curb unnecessary medical costs, provide health benefits when necessary, and reduce the contribution to antibiotic-resistance. This dissertation focuses on the effects of antibiotic treatment timing on patient outcomes as measured by revisits for pediatric ARTIs in outpatient settings and examines the factors that correlate with antibiotic prescription decision and the use of strategies used to improve the appropriateness of the treatment decision. These strategies include tailoring the time to treatment, the use of narrow-spectrum antibiotics, and employing Group A strep diagnostic test when appropriate.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Health economics and outcomes research for informed decision making in rapidly evolving therapeutic areas
PDF
Understanding primary nonadherence to medications and its associated healthcare outcomes: a retrospective analysis of electronic medical records in an integrated healthcare setting
PDF
Pharmacotherapy among patients with type 2 diabetes
PDF
Antibiotic immunomodulation in Staphylococcus aureus bloodstream infection
PDF
Documenting the impact of under-treatment for chronic disorders
PDF
The price and clinical impact of specialty drug management in the US healthcare system
PDF
Value in health in the era of vertical integration
PDF
Three essays on comparative economic and clinical outcomes of antipsychotics using retrospective claims data
PDF
The disease burden among aging people living with HIV/AIDS
PDF
Randomized clinical trial generalizability and outcomes for children and adolescents with high-risk acute lymphoblastic leukemia
PDF
Evaluating cancer treatments with health economics methods
PDF
Essays in health economics
PDF
Integration of behavioral health outcomes into electric health records to improve patient care
PDF
Impact of pharmacy-based transitional care on healthcare utilization and costs
PDF
Three essays on estimating the effects of government programs and policies on health care among disadvantaged population
PDF
Addressing federal pain research priorities: drug policy, pain mechanisms, and integrative treatment
PDF
Inter professional education and practice in the health care setting: an innovative model using human simulation learning
PDF
Long-term impacts of childhood adversity on health and human capital
PDF
Sportiod and teen athletes
PDF
Essays in health economics and provider behavior
Asset Metadata
Creator
Sangha, Kinpritma
(author)
Core Title
Outcomes of antibiotic use among children with acute respiratory tract infections
School
School of Pharmacy
Degree
Doctor of Philosophy
Degree Program
Health Economics
Publication Date
07/28/2019
Defense Date
04/25/2019
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
acute infections,antibiotic prescribing,antibiotics,Children,health outcomes,OAI-PMH Harvest,Pediatrics,prescriptions,respiratory infections,revisits
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
McCombs, Jeffrey (
committee chair
), Padula, William (
committee member
), Steinberg, Irving (
committee member
)
Creator Email
kinpritma.sangha@gmail.com,ksangha@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c89-196975
Unique identifier
UC11663131
Identifier
etd-SanghaKinp-7655.pdf (filename),usctheses-c89-196975 (legacy record id)
Legacy Identifier
etd-SanghaKinp-7655.pdf
Dmrecord
196975
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Sangha, Kinpritma
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
acute infections
antibiotic prescribing
antibiotics
health outcomes
prescriptions
respiratory infections
revisits