University of Southern California Dissertations and Theses

Development and validation of survey instrument designed for cervical cancer screening in Malawi, and other low resource settings

(USC Thesis Other)

Development and validation of survey instrument designed for cervical cancer screening in Malawi, and other low resource settings

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
Development and Validation of Survey Instrument Designed for Cervical
Cancer Screening in Malawi, and other Low Resource Settings.

By

Ruhi Sikri

--------------------------------------------------------------------------------------------

A Thesis Presented to the
Faculty of the Graduate School
University of Southern California
In Partial Fulfillment of the
Requirements for the Degree
Master of Science
(Applied Biostatistics and Epidemiology)

December 2014

Copyright 2014 Ruhi Sikri

2

Acknowledgements
This thesis was completed with the support and guidance of Dr. Victoria Cortessis, Dr.
Laila Muderspach, and Dr. Stanley Azen. I am grateful for their invaluable insight and
mentorship throughout this entire process. I also want to take a moment to especially
thank Dr. Cortessis, the head of my thesis committee, for inspiring me not only while
being my teacher at USC, but also as my research mentor. In addition, this project could
not have been completed without the love and understanding from my supportive family
and friends who saw me through the very end.

3

Table of Contents

Acknowledgements………………………………………………………………………..2
List of Tables……………………………………………………………………………...4
List of Figures……………………………………………………………………………..5
Abstract……………………………………………………………………………………6
Chapter 1: Introduction……………………………………………………………………7
Chapter 2: Methods………………………………………………………………………12
Chapter 3: Results……………………………………………………………..…………20
Chapter 4: Discussion & Future Suggestions…………………………………………....37
References………………………………………………………………………………..40
Appendix A: Original Survey Instrument………………………………………………..43
Appendix B: Revised Survey Instrument………………………………………………..45

4

List of Tables

Table 1.1 Layout of the structure for daily expectations for personnel……….……14
Table 2.1 Descriptive statistics comparing frequency of missing values
by question type for menstrual history.…………………….....………….26

Table 2.2 Descriptive statistics comparing frequency of missing values
by question type for obstetric and gynecologic history………………….29

Table 2.3 Descriptive statistics comparing frequency of missing values
by question type for pelvic exam section.…………………….…….…....33

Table 2.4 Descriptive statistics comparing frequency of missing values
by question type for bimanual exam…………………………………………35

5

List of Figures

Figure 1.1 Page 1 and 2 of survey instrument used in Malawi intervention.....……..16
Figure 1.2 REDCap’s visual aid demonstrating status of each survey entry..............17
Figure 2.1 The original version of the menstrual history portion of the survey
instrument...………...………………………………………………..…..24

Figure 2.2 Bar chart displaying the distribution in response types, including
missing frequency. …..........................................................................…..25

Figure 2.3 Revised menstrual history section of the survey proposed for use
in the next PAPS team intervention. ..……………………….....………..27

Figure 2.4 Obstetric & gynecologic history section of the survey taken from
the original survey used by the PAPS team at the Women’s
Center, in Malamulo ….................……...………………………….…....28

Figure 2.5 Revised obstetric & gynecologic history section of the survey
proposed for use in the next PAPS team intervention..........................….31

Figure 2.6 Pelvic exam portion of the physical exam section, taken from the
original survey used by the PAPS team at the Malamulo
Women’s Center………………………………………………………....32

Figure 2.7 Revised pelvic exam section of the survey proposed for use in
the next PAPS team intervention…………………………...………...….36

6

Abstract

Regions with the highest burden of cervical cancer are often those with the most
limited access to resources and care. This is especially true for countries in southeast
Africa, which carry the greatest burden of this disease, such as Malawi. Factors
contributing to such high incidence and mortality rates include not only limited
accessibility, misunderstanding of disease, and economic barriers, but also ineffective
data capturing instrumentation used by clinicians in the field. To identify the
characteristics of survey instrumentation that are found to yield missing or inapplicable
responses, data from the 2012 PAPS cervical cancer intervention in Malawi was studied.
In this weeklong intervention, 1,221 Malawian patients were seen and screened for
cervical cancer through use of Pap smears. Of the 1,210 with evaluable Pap smear tests,
124 were found to yield abnormal results. Yet, of equal interest, was the pattern of
missing values and inapplicable responses that were found in each section of the survey
instrumentation, administered during the patient consultations in the clinic. Qualitative
and quantitative reasoning was used then to analyze and justify what the most effective
and ineffective ways in which data was captured in this instrument so an improved
version of the same tool could be produced and is included in Appendix B. It was found
that missing data was least commonly found in questions that had a pre-set list of options,
and used a “checkbox” format. This research is imperative to improving the efficiency
with which the data can be collected and used to best treat the population at risk, in this
case underserved Malawian women who have rarely been screened for cervical cancer
prior to this PAPS site visit.

7

Chapter 1: Introduction
Researching Cervical Cancer
In 2006 the World Health Organization (WHO) presented a call to action stating
that reproductive and sexual health goals need to specifically focus on sexually
transmitted infections (STI), cervical cancer, and other gynecological morbidities [1].
This comes as no surprise as cervical cancer affects approximately 528,000 females
worldwide each year, of which over half die from this disease [2]. On a global scale, 80%
of the cervical cancer burden lays on women in developing countries where health care is
often lacking most [1]. However, cervical cancer is preventable through active and
effective use of screening to detect preneoplastic lesions followed by treatment before
progression to invasive cancer.
Etiology
What is unique about this disease compared to most other cancers is that etiology
is known; ninety-nine percent of all cervical cancer cases are a result of an existing, often
symptom-free, human papillomavirus (HPV) infection [3]. However, simply having HPV
is not a sufficient factor to developing cervical cancer. There are 40 different strains of
HPV. It is one of the most common STIs in the world, and a high proportion of sexually
active individuals will contract this infection some time after sexual debut [3]. While in
most cases the infection will resolve without treatment, some infections persist, and these
are believed to predispose to precancerous lesions, a warning sign of the formation of
cervical cancer. Currently, 70% of invasive cervical cancers are attributed to HPV 16 or
18, the two most prevalent carcinogenic strains of the virus [1].

8

Developed & Developing Nations
While cervical cancer is a worldwide issue, 83% of cervical cancer cases occur in
the developing world, compared to 3.6% in developed countries [4]. Specifically,
“Mortality varies 18-fold between the different regions of the world, with rates ranging
from less than 2 per 100,000 in Western Asia, Western Europe and Australia/New
Zealand to more than 20 per 100,000 in Melanesia (20.6), Middle (22.2) and Eastern
(27.6) Africa” [5]. The vast discrepancy between these rates can be cited to be a result of
effective versus ineffective secondary prevention. Cervical cancer screening is widely
established as the most effective method for detection and control of cervical cancer [6].
Use of Pap smears as a form of secondary prevention has significantly reduced cervical
cancer rates in countries such as the United States where adequate resources are available
to provide the necessary infrastructure to control and prevent new cases from emerging.
While recent advancements in the research field suggest adopting prevention
strategies that encompass primary intervention tools, such as the HPV vaccine in the form
of Cervarix or Gardasil, its impact will take time to surface as the primary target
demographic for this vaccine is young girls, prior to becoming sexually active, around the
ages of eleven or twelve [7]. Unfortunately, the vaccine is currently prohibitively
expensive for widespread use in many low-resource settings and has just been introduced
through pilot programs in developing countries, such as Malawi, in 2013 [8]. Moreover,
uptake of the vaccine of the full three dose regimen has to date been very incomplete,
even in some high resource areas such as the United States [9] where it was added to the
list of suggested vaccines for females in 2006, and males in 2009.

9

Focusing Research on Developing Nations
Sub-Saharan Africa is now more commonly targeted in the realm of research
because it has the highest burden and mortality rate associated with cervical cancer
worldwide [10]. And while screening has been accepted as the gold standard for
monitoring cervical cancer and more importantly detecting and remediating pre-
neoplastic lesions in these developing countries, barriers still exist that render the current
programs to be inadequate, resulting in persistence of high rates of cervical cancer. As
explained in a report from a WHO 2002 consultation, there is a “… need to develop
programmes that have a systemic approach, are well integrated into the existing health
system and take into account the social, cultural and economic context [11],” and a
system that encompasses all these qualities is what is lacking in the sparse current
establishments in place. A research priority is therefore identification of barriers to
instituting such programs.
Aside from the structural issues related to the lack of establishment of effective
health clinics in these high demand areas, there are other obstacles that limit access to
such care in Malawi. Past research has established that the most commonly cited
limitations of going to clinics for screening include low perceived risk, low knowledge of
developing cervical cancer, limited accessibility to these clinics due to distance or
navigation, and low perceived benefits from screening due to fatalistic attitudes towards
the disease [12]. These factors combined with the commonly held mentality that
preventive care is viewed only as a luxury make for sound reasons for why better
education on this cancer and its preventability need to be addressed.

10

Malawi Profile: A look into Southeast Africa
In eastern Africa, 35.8% of women in the population are estimated to have a
cervical HPV infection at a time, and 76.5% of the invasive cervical cancers are
attributed to HPV 16 or 18 [8]. In Malawi, the second most common cause of death from
cancer is cervical cancer [13]. According to Institut Català d'Oncologia (ICO)
Information Centre on HPV and Cancer, Malawi’s population of women 15 and up (those
at risk of developing cervical cancer) is 4.5 million women, and every year 3,684 women
are diagnosed (incidence) and 2,314 die (mortality) from this disease [8].
Still, Malawi is a country that epitomizes the characteristics desired for the
successful establishment of such cervical and breast cancer screening clinics. Malawi is a
politically stable country whose women commonly die from this form of cancer, often
due to lack of understanding of the severity of the disease, inability to go to the clinic, or
monetary issues. These women have a low life expectancy, high HIV prevalence, and
more than 65% live in poverty [14]. Malawian constituents reign mostly from a rural
population where the main economic driver is agriculture [14].
Methods Implemented by Malawian Government
The healthcare structure in Malawi follows a socialized system where the
Ministry of Health and Population is the largest provider of health services [10]. The
Malawi Ministry of Health (MOH) updated its National Reproductive Health Policy to
include cervical cancer and approve visual inspection with acetic acid (VIA) as an
accepted method in cervical cancer screening for early detection in 2002 [15]. HPV DNA
testing is not practiced in this region, and cervical cancer screening coverage was
estimated to be 2.6% in 2013 for women aged 18-69 years, screened every 3 years [8].

11

Using this background knowledge to design a better informed, multi-sectored
intervention will aid in creating an unique, country-specific call to action, addressing the
specific needs and thereby prescribing the most effective improvements in detection and
care. Specifically, research is needed to improve the current approaches to screening that
would identify and target those most likely to benefit. To this end, it is desirable to adopt
survey instrumentation to the specific setting so intake of this crucial data is more likely
to be accurate and reliable. This will allow for future strides in research that can better aid
in improving the targeted demographic, making more informed decisions with respect to
treatment type and education options in distinct regions.

12

Chapter 2: Methods
PAPS Team International, Inc.
Professionals Analyzing Pap Smears, also known as the PAPS Team International, serves
as one medium to this call to action. The PAPS approach consists of training an in-country team
to carry out secondary prevention (cytology screening and related treatment), and providing open
collaboration in the early stages of implementation to ensure the sustainability of the program
once the visiting members of the team depart. Their mission is to prevent the development of
cervical cancer in countries where individuals are underserved in receiving proper health care
due to limited accessibility to resources, namely in developing countries [16]. This California-
based non-profit has established four different sites in Kenya prior to this most recent 2012
Malamulo, Malawi visit to the New Women’s Center.
The PAPS method of preventing future cases of cervical cancer in such regions is two-
fold. Their primary goal is to establish clinics that are both accessible and affordable to the
underserved population. In addition, they structure their consultations with patients to educate
the visiting individuals, often female, regarding crucial prevention methods and the importance
of testing for other common infections in the given locality, such as HIV and HPV [16]. It is
during these visits that common misconceptions are cleared up and the severity of HIV and
cervical cancer are fully understood by these women.

13

The Team & Training
The Malawi-bound PAPS team offered cytologic screening for cervical cancer and
appropriate follow-up at the New Women’s Center in Malamulo from September 10
th
, 2012 to
September 18
th
, 2012. This team consisted of forty-eight individuals. There were nineteen
Malawian personnel who also received specific training for cervical cancer prevention. Of these,
five were physicians and clinical officers, twelve were nurse practitioners, and two were cyto-
technologists. Training for these volunteers varied based on their position. Malawian cyto-
technologists were trained for four months at Loma Linda University, in Southern California,
prior to the trip. During this time they received didactic training and shadowed counterpart
professionals who evaluated their skills in handling and preparing Pap slides, understanding the
scoring system, and being able to read and score sample Pap smears. In September 2012, the
PAPS Team travelled to Malamulo, where the rest of the Malawian team members were trained.
There, they collaborated on screening and follow up practices to ensure sufficient preparation
across all facets of the team. The nurses’ training consisted of practicing simulations where the
following was required: cervix and breast examinations, obtaining a Pap smear, as well as
counseling and appropriately referring a patient. The physicians and clinical officers were tested
to demonstrate satisfactory knowledge of cervical cancer etiology, screening and treatment,
evaluations that were both clinical and colposcopic, LEEP procedures and cryotherapy prior to
the beginning of the on-site visits from local females.
Once at the New Women’s Center, the team adhered to a pre-set agenda that laid out the
expectations for each role for each day of the intervention. The schedule followed is displayed
below in Table 1.1.

14

Table 1.1 Layout of the structure for daily expectations for personnel

This table provides a visual representation of the duties of each personnel role on each day and
was designed by Dr. Cortessis. Yellow rows refer to team members from Malamulo, those being
trained, and red refers to team members who were a part of the PAPS team.
[Note: SP = supporting personnel CT= cytotechnologists RN = resident nurses MD= doctors]

Survey (design and purpose)
The single paged printed survey instrument used on-site was created and structured by a
clinician with the intention of capturing the most important characteristics believed to be needed
to accurately understand a woman’s medical history and possible risk of contracting cervical
cancer. Care provided and documented on this form ranged from receiving cytology screening to
colposcopy and appropriate treatment for cervical lesions or malignancies
1
. One goal of this
survey was to learn the types of patients who would most likely benefit from cervical cancer
screening.
In total there were seventy fields on the survey instrument which were broken down into
demographics, menstrual history, obstetric history, gynecologic/sexual/contraceptive history, a
breast exam, a pelvic exam, and cytology sections consisting of 7, 9, 5, 13, 7, 14, 15 inquiries,

1
IRB proposal

15

respectively. Within these distinct portions, questions inquiring about characteristics that have
been established as high risk factors in past literature were included, covering topics such as
smoking status, parity, age of sexual debut, and number of sexual partners [10]. Formats of these
questions for each section varied from fill in the blank, often used for numerical or qualitative
responses that had the largest range in type of descriptive responses, and checkmark or circling,
often used for selecting an option out of a specified set of answers.
During the nine active days of the program, 1,221 women presented for care, of whom
1,210 were evaluable. Of the 1,210 Pap tests evaluated, 124 yielded abnormal Pap results, which
lead to further evaluation for possible premalignant lesions.
Working with the data
Data collected for clinical and research purposes were de-identified for use and research.
The proposal process was reviewed and approved by the USC IRB and is available online. The
specific sections of the consultation that I analyzed and critiqued were menstrual history,
gynecologic/sexual/contraceptive history, and the pelvic examination.
My role on the team began with becoming familiar with the data in the original paper
form, in order to most accurately translate the written version of the surveys into an electronic
version, using Research Electronic Data Capture (REDCap), a secure, web-based application
widely accepted in the research field [17]. This software was selected for its ease in allowing
collaboration, tracking system and clear methods of data management. It was also an ideal choice
because it is designed for medical research and can methodically export the raw data compiled
on its database into syntax files for statistical programs such as STATA, SPSS, and R [18]. The
original printed survey instrument used at the Malawi Women’s Center is provided in Figure 1.1
as a representation of the original form on which the data was collected. Figure 1.2 provides a

16

sample of the data-tracking tool, visible to users, used to document forms that have successfully
been transformed from the original version onto the REDCap database.
Figure 1.1 Page 1 and 2 of survey instrument used in Malawi intervention

Note: The survey used is ordered by category, from demographics, menstrual history, obstetric history,
gynecologic history, breast and pelvic exam, cytology. An enlarged version of the survey is included in
Appendix A.

17

Figure 1.2 REDCap’s visual aid demonstrating status of each survey entry.

Note: This is the interface through which the real-time capture of entered data is reflected.

Through this process I also replicated entry of survey data with the other team members
in order to ensure accuracy of entry and crosscheck potential room for errors that may have
arisen in the double entries. Each week the number of forms entered were kept track of and gone
over at the weekly meetings with our supervisor, Ugonna Ihenacho, and academic epidemiologist,
Dr. Victoria Cortessis. It was during these meetings that initial discussions regarding ambiguity
in certain questions were mentioned because a similar pattern in many forms became evident. In
addition, in fields where no specific direction was given, such as specifying units or requesting a
qualitative versus quantitative response, there were many answers that were left blank. A record
was kept to track which questions were capturing the recordings the most effectively and which
ones were more difficult to interpret.
This information was then presented while meeting with experts in the field of gynecology,
epidemiology, and survey design to reconcile the differences in responses. Differences in
responses consisted of discrepancies as large as providing a qualitative versus quantitative result
in the same field, which made coding and compilation difficult because it was not clear which

18

was the preferred method of documentation for the specified question. For example, such was the
case for portions of the bimanual exam where consistency and mobility were reported as either
“yes/no” or “normal.”
Another issue that arose while transforming the data to an electronic database was that
different clinicians used different units for the same exam, so directly comparing results from
different examiners was not possible. Over time, it became clear that questions were ambiguous
or could be interpreted in different ways; in some instances the ambiguities could not be resolved.
This was especially evident in the pelvic exam and cytology section. Such was the case for the
bimanual exam where uterine size was reported often as “normal,” “6-7 cm” or “5-6 weeks.”
Revision of Questionnaire Elements
Using information gathered from these meetings, the next step was to fine tune the
instrumentation. I went through the process of collaborating with survey design documentation
and self-guided research to redesign elements of the questionnaire that were not efficiently
capturing the data desired by the clinician and were ending up being the most problematic in the
data compilation stage of the study. Use of Microsoft Word’s Developer tab enabled use of
proper survey design tools to design a formal template that could be used and referred to at a
later time for other researchers [19]. I consulted with our team academic epidemiologist and
gynecologic oncologist at the USC Keck School of Medicine to corroborate my findings, and
made corresponding edits to update the survey instrument. A review of the specific suggestions
made by the clinically active gynecologic oncologist, Dr. Laila Muderspach, was done as she
represented those clinicians responsible for filling out the form during the intervention. Her
unique perspective on the design, content and limitations provided to be invaluable in the
revamping of the instrument in order to balance the needs of an epidemiologist and a clinician, a

19

balance that was arguably skewed in the original survey since the researchers only recently
joined the PAPS team.
The importance of fine-tuning the instrument
Such a survey is needed as the first step in establishing an effective approach to screening
and preventing this disease because no study to date has created a universally accepted survey
template to be used in low resource, high demand settings. In addition, understanding local social
and cultural practices, and general views on personal topics related to sexual history, parity, and
disease status is crucial in being able to appropriately seek out information and obtain the
information desired. Success in this first step can, in turn, lead to future structured studies testing
the relationships between certain risk factors and the onset of cervical cancer, and refinement of
criteria for screening. The challenge in developing such a survey that captures all this critical
information, that is still sensitive to the cultural norms of this particular region, was elucidated as
a sore spot post intervention, during compilation of the data from these forms, and was
incorporated in the revisions.

20

Chapter 3: Results
On a Holistic Scale
Positive Qualities
A number of aspects of the survey were already crafted to conform to properties
desirable for research described above. These were retained in the survey instrument
without suggested modification. Length of the instrument was one such aspect because
increasing overall length of the instrument would likely reduce the response rate or
accuracy of responses due to fatigue or time constraints put on each examiner. After all,
over 1,200 women were examined over the course of the eight days dedicated to
screening and treatment activities carried out in the Malamulo center. So, the length that
was set for this intervention is likely the longest that would be appropriate for a single
clinician examining a patient, given that the time frame and nature of this type of
intervention is aimed at serving a high demand population in a short amount of time.
Dividing the survey into different sections aided in making the tool more
aesthetically pleasing to the viewer, easier to follow the responses when translating the
paper forms to the online version, and more helpful for the physicians trying to keep track
of what topic they were covering. This was another aspect of the original survey that was
retained in the newer model to maintain these qualities. The order of sections in the
original tool was: demographics, menstrual history, gynecologic/sexual/contraceptive
history, breast exam, pelvic exam, and cytology. Similar response options were provided
for each section, likely to maintain relative uniformity in each portion and aid the process
of going through the questions.

21

Essential questions regarding risk factors established for cervical cancer,
including smoking status, age at sexual debut, number of partners, parity, and co-
infections were all included. Therefore the survey questions querying information on
established risk factors that would be most useful to epidemiologists and researchers and
were kept in the newer model. However, the manner in which these questions were asked
was where problems began to appear, namely order and response options for the highly
personal questions.
All of these factors were kept in mind when altering the structure of the survey to
better aid higher completion rates and reduce the frequency of missing values or unusable
responses.
Areas for Potential Improvement
Among these positive qualities, this tool had definite areas for potential
improvement. Although the survey queried crucial information, there was room to more
discretely accommodate social factors that also play a role in the type of results the
clinicians obtained during these consultations. Asking personal questions early on in the
visit may have created undue discomfort to the patient and reduced the response rate,
because human behavior tends to naturally require an established level of comfort with
the physician prior to opening up honestly about such topics. For example, questions
regarding sexually transmitted infections or sexual practices should not have been asked
before more general questions regarding smoking habits and allergies. It is for this reason
that the order within the entire survey was re-structured to reflect this socio-cultural
adaptation.

22

This re-organization was accomplished by splitting this tool into three different
sections, to be handled by three distinct personnel. As per the suggestion of Dr.
Muderspach, the different sections of the form could be identified by using a color-coded
schema, available by simply printing certain sections on a given color. The first and least
intrusive section would focus on obtaining background information on the patient, the
second would be the “hands on” physical breast and pelvic exam, and the third portion
would focus on cytology
2
. This ease in transition would also allow for fewer blank fields,
as questions in each section would be sure to be included in the portion of the tool where
they can actually be answered. For example, the pelvic exam section had questions that
could only be answered during cytology, a different section altogether, so they were
rarely filled in. In addition, questions regarding previous Pap tests and menopausal status
in the cytology section could be moved to an earlier demographics section where it would
be more appropriately placed. Splitting up the sections in this manner would also allow
supporting personnel to be able to check that all the data required was captured in real-
time, that is, ideally while the patient is still present. This change would increase
efficiency both during and after the site visit. This would aid efficiency in the field
because each examiner would have a clear focus while allowing for distribution of the
effort of collecting patient data among the PAPS personnel. This reduction in turnaround
time also would allow for a shorter interval between data collection and analysis post-
visit.
A second type of revision was aimed at creating a more uniform system for
response type in the survey instrument. The original instrument used several formats for

2

For the purpose of redesigning the tool, I focused on sections within the first two portions (background and
physical exam) of the tool because I was not given access to the third, cytology-based, section.

23

capturing data times, and we surmised that use of so many distinct formats factored into
the resulting high variability, including missing responses in several fields. For example,
having clinicians in certain sections checkmark the most appropriate answer out of a
given selection, but circle it in others, was one type of inconsistency noted in the original
survey instrument. One goal of statistical analysis was to learn which of these various
formats was associated with a lower frequency of missing data. Results indicated that
providing a specific set of response options and using checkmarks as the marker ended up
being the most useful way to provide response options. This is likely a result of it
minimizing all sources of variation through a set structure of responses. Using
checkmarks as the method of marking response type was preferred over circling because
multiple responses can accidentally be included when circling and personal preferences
of circling methods can increase discrepancies, as was found in this dataset. The
descriptive statistics provided below are based on of the initial 726 cases for which data
were entered into the REDCap electronic database and available to be used for analysis
thus far in the study.
Another simple measure used to re-design this instrument was setting alignments
within the response options and questions. We anticipate that the visual order that
resulted from more consistent spacing and alignment will aid the clinician in responding
to the questions in the desired manner, by making missing data items and inconsistent
entries more obvious, thereby reducing the ambiguity that was existent prior to these
updates. Making these options for each field more specific and consistent throughout the
entire survey instrument was necessary in order to increase the efficiency and accuracy of
responses.

24

[Updates by Section]
1. Original Version: Menstrual History
Figure 2.1 The original version of the menstrual history portion of the survey instrument

Note: This is the menstrual history section of the survey taken from the original survey used by the PAPS
team at the New Women’s Center, in Malamulo.

1. Comments/Justification:
The updated version incorporates alignment shifts in order to better specify which
questions were follow ups, as doing so would allow for a better understanding if no
response was marked because the value was “0”, the question was not applicable to the
patient, or if it was really missing. In most cases where there was a question with a
conditional follow-up, only the first question was answered. But, there were also other
cases where only the follow up question was filled in or both were completed. For
example, after asking the patient if she bleeds between periods, the follow up question
asks the individual to rate the level of bleeding. The resulting data for this follow up
questions is provided in Figure 2.2 below.

25

Figure 2.2 Bar chart displaying the distribution in response types, including missing frequency.

Note: Obtained from Data View in REDCap.

What cannot be determined from these results in Figure 2 is whether the 648 missing
fields were reflecting the fact that all those individuals did not bleed between periods,
they had stopped having periods and therefore this question was not applicable, or if in
fact it was just left blank and no information was provided. Better designing these
questions can aid in reducing the number of fields left blank and associated ambiguity
about genesis of missing values. This is crucial because lack of response could mean
certain types of women, those with similar characteristics, are not being captured in this
question and this differential misclassification would thereby bias results, reducing the
accuracy of the actual distribution. The frequency of missing data by question format

26

type is detailed below, in Table 2.1, which summarizes the frequencies of missing results
by question type.
Table 2.1 Descriptive statistics comparing frequency of missing values by question type for menstrual
history.
Question Format Type
Frequency of
Missing
Age period began Fill-in-blank 4.1% (30/726)
How many days does your period last? (fill-in-blank) Fill-in-blank 5.4% (39/726)
Date of your last menstrual period: DD, MM, YYYY Fill-in-blank
48.2% (350/726)
34.8% (253/726)
14.3% (104/726)
Do you bleed between periods? Checkbox 6.9% (50/726)
Have you stopped having periods? Checkbox 1.7% (12/726)
Has your womb been removed? Checkbox 5.0% (36/726)
Are you on hormone replacement therapy? Checkbox 22.2% (161/726)

As can be seen from Table 2.1, the frequency of missing values tends to be lower with
questions that follow the checkbox option, as the missing frequencies in this section
range from 1.7% to 22.2%, while missing values for the “fill-in-blank” fields had a larger
range, from 4.1% to 48.2%.
Additional changes were made on a more detailed level. In order to prevent
clinicians from providing ranges for “how many days does your period last?” an average
number was specified so it was known to provide simply one numerical value. Date of
last menstrual period was asked higher up in the order of the questions because the
answer to this question would set the foundation of the type of answers for the following
questions, such as “have you stopped having periods?” and the follow up “If yes, have
you had any vaginal bleeding since menopause?” The last question included a follow up
requesting information on duration for hormone replacement therapy, in order to be able
to assess any potential correlation in later stages between different time intervals and the
onset of cervical cancer.

27

1. Updated Version:
Figure 2.3 Revised menstrual history section of the survey proposed for use in the next PAPS team
intervention.

28

2. Original Version: Obstetric & Gynecologic/Sexual/Contraceptive History
Figure 2.4 Obstetric & gynecologic history section of the survey taken from the original survey used by the
PAPS team at the Women’s Center, in Malamulo.

29

2. Comments/Justification:
As in menstrual history, this section included distinct ranges in frequency of
missing values, by format type, listed in Table 2.2. The format type with the highest
completion rate was the checkbox format, which had a missing frequency ranging from
1.2% to 8.5%. The circle format type had a missing frequency range from 2.3% to 93.5%.
From this portion of the survey it is evident that the checkbox format yielded fewer
missing values. The fill-in-blank format type had a missing frequency range 2.1% to
23.8%.
Table 2.2 Descriptive statistics comparing frequency of missing values by question type for obstetric and
gynecologic history.
Question Format Type
Frequency of Missing
(N=726)
Age at first pregnancy Fill-in-blank 2.1% (15/726)
Number of full-term
pregnancies
Fill-in-blank 2.6% (19/726)
Number of preterm
pregnancies
Fill-in-blank 23.8% (173/726)
Number of miscarriages Fill-in-blank 20.2% (147/726)
Number of abortions Fill-in-blank 22.3% (162/726)
Age at first intercourse Fill-in-blank 5.0% (36/726)
Number of lifetime sex
partners
Checkbox 2.9% (21/726)
How many sex partners were
circumcised?
Checkbox 8.5% (62/726)
Have you had sexual relations
within the past 6 months?
Checkbox 1.7% (12/726)
Ever had a Pap smear? Checkbox 7.3% (53/726)
Current birth control method Circle 2.3% (17/726)
Have you had any of the
following infections?
Circle 46.0% (334/726)
Have you had any of the
following symptoms?
Circle 93.5% (679/726)
History of any of the following
cancers?
Circle 4.7% (34/726)
Do you smoke cigarettes? Checkbox 1.2% (9/726)
Do you have drug allergies? Checkbox 1.8% (13/726)

30

This was one section that especially benefitted from re-ordering of the questions
and response type. The first structural change made was combining the obstetric section
with this gynecologic and sexual history portion. This was done because the types of
questions asked in both could fall under the overarching theme of “medical and
reproductive history.” The more general questions, specifically those related to smoking
and allergies, were shifted to the demographics section of the survey because that is
where they would be most appropriate. The ordering of these questions within this newly
revised medical and reproductive section was also adjusted to reflect a transition from
least sensitive questions, such as common health issues and past testing for infections,
leading up to the more personal questions regarding parity, birth control, and sexual
behavior.
An indented follow up system was designed so if the answer to first question was
“no” or “not applicable,” the clinician would know to skip to the next set of questions.
Following this design would also aid in increasing the efficiency and organization of the
consultation, saving time where certain questions need not be asked.
The response type within this section was also adjusted to reflect a more
consistent output; clinicians were instructed to check the most appropriate box or fill in
numerical values for fields where it would not make sense to list out all the possible
options or the findings were qualitative in nature. Ranges for different questions were set
to be consistent with one another, such as the number of sexual partners and number of
partners circumcised.
Specific detail-oriented changes included changing the number circumcised to a
quantitative rather than qualitative question because that would be more informative later

31

on for future analysis that could be done. The changes made are reflected in the updated
version below, in Figure 2.5.
2. Updated Version:
Figure 2.5 Revised obstetric & gynecologic history section of the survey proposed for use in the next PAPS
team intervention.

32

3. Original Version: Physical Exam
Figure 2.6 Pelvic exam portion of the physical exam section, taken from the original survey used by the
PAPS team at the Malamulo Women’s Center.

3. Comments/Justification:
For this section of the survey, I consulted with Dr. Muderspach regarding the
variables included, specifically how and for what purpose they were collected during the
exam, in advance of reformulating this part of the instrument. This section focused on the
pelvic exam where suspicious lesions or abnormalities able to be seen with the naked eye,
but not those that require cytologic evaluation, would be noted. Prior to revision, this
section exemplified a section that encompassed fields that could not be answered by the
given examiner because it was not a measurement or method used in this particular
section. This was the case for the acetowhite lesion test that was done in the cytology
stage of screening, during colposcopy. For the acetowhite test 3-5% acetic acid, also
known as household vinegar, is applied to the cervix, which stains areas of the cervix that
are abnormal white [20]. Presence of any such potential white lesions is classified as
abnormal. This test was only conducted if the Pap smear yielded abnormal results and
further examination was necessary, but was documented in the cytology section. This is

33

often why these fields were left blank in the physical exam portion of the survey
instrument.
Differences in response rate for this section can be most likely attributed to lack
of appropriate placement of tests, when comparing the frequency of missing values
between questions in this section with circle response formats versus those with
checkboxes. This can be seen as the frequency for both checkbox and circle for the
vagina and cervix tests range from 75.6% to 84.8%. Missing values are reported for each
of these tests in Table 2.3.
Table 2.3 Descriptive statistics comparing frequency of missing values by question type for pelvic exam
section.
Question Format Type
Frequency of Missing
(N=726)
Pelvic exam results Checkbox 29.8% (216/726)
BSU Checkbox 27.7% (201/726)
Vagina: Acetowhite Circle 79.2% (575/726)
Vagina: White lesions Checkbox 83.6% (607/726)
Vagina: Condyloma Checkbox 75.6% (549/726)
Cervix: Acetowhite Circle 79.9% (580/726)
Cervix: White lesions Checkbox 84.8% (616/726)
Cervix: Condyloma Checkbox 76.4% (555/726)

Based on consultation with the on-site gynecologic oncologist, Dr. Muderspach,
queries regarding these two tests were moved to the third part of the screening exam,
which focused on cytologic evaluation. Condyloma visualization was kept in this section
because it simply required visual examination of the genital region for any lesion
resembling a wart and could be determined by the examiner on-hand, on the spot.

34

This section was further re-designed using boxes and shading to organize the
queries about different types of tests for visual clearly. A simple female external genitalia
diagram was added so clinicians could indicate where they noted any potential lesion.
Conditional questions inquiring about potential bleeding or possibility of pregnancy at
the time of visit were asked in order to determine which tests could be conducted at the
time of visit without threatening the health of the woman. Doing so would aid in
providing some explanation if certain sections of the exam were listed as “deferred,” as
was often the case in the collected data from this pelvic exam section.
Four attributes listed on the original bimanual exam were taken out post-
consultation with the Dr. Muderspach because in her clinical experience they provided no
further information on the potential risk of cervical cancer or disease development. The
only feature of the bimanual exam that was retained was tenderness. Collecting
information on this variable would be useful to study in the analysis stage for strength of
association with the outcome of interest, cervical cancer.
Within the bimanual exam, it is interesting to note that the question with the
highest response rate was the tenderness question, with only 63.1% of responses missing,
compared to the other fill-in-blank bimanual questions with higher percentages. Specific
values are provided in Table 2.4 below.

35

Bimanual Exam
Table 2.4 Descriptive statistics comparing frequency of missing values by question type for bimanual exam.
Question Format Type
Frequency of Missing
(N=726)
Uterine size Fill-in-blank 78.5% (570/726)
Shape Fill-in-blank 69.6% (505/726)
Position Fill-in-blank 53.2% (386/726)
Consistency Fill-in-blank 70.4% (511/726)
Mobility Fill-in-blank 71.9% (522/726)
Tenderness Checkbox 63.1% (458/726)

The response type for the “uterine size” field was revised to be highly specific, as
it gave clinicians only two response options—normal and abnormal—with respective
descriptive requirements specifying the size range for each category. Uterine size was
included because it provided a medium to detect possible pregnancy with a uterine size
larger than 8 centimeters or 8 weeks, two measurements used interchangeably. This is an
important variable to measure because checking the “abnormal” or “greater than 8 cm”
box indicated likely pregnancy and would be a reason to halt future examinations or
testing for preventing risk of harm to the child.

36

3. Updated Version:
Figure 2.7 Revised pelvic exam section of the survey proposed for use in the next PAPS team intervention.

37

Chapter 4: Discussion & Future Suggestions

This study found that the most effective formats for the types of questions being
asked in the survey instrument were those that followed the checkbox format and had a
pre-set list of responses from which to choose. Patterns of completion of survey questions
allowed for revision of the instrument that was originally designed in the context of
western medicine and culture, to be better suited to obtain information from patients
examined in a low resource, high demand setting.
1) Conducting a Pre-Intervention Visit
In addition to using the improvements made to this tool, additional revisions may
be desired for use of the instrument in other low resource, high demand settings, either
elsewhere in Malawi or in other countries. The PAPS team usually conducts a 6-month
pre-intervention field study to better understand the attributes and customs unique to the
intervention setting. Visiting the village in the region where the intervention is planned
can allow for understanding common practices, such as appropriate clothing attire or
greeting methods, that can aid in making the patients feel more comfortable with the
clinician examining them. For example, Dr. Muderspach noted that while there is no
specific national attire for Malawian men, Malawian women traditionally wear a
“chitenje,” an enlarged sarong that is often matched with a head cloth and blouse [21]. So,
adopting a similar clothing style in the clinic could also be helpful in acculturating with
the patients. Similarly, the visit would also allow the researchers to ask about the most
appropriate survey response options for the target region. Specifically, this future
addition would aid in better incorporating culturally competent response types that are
commonly used to describe symptoms for cervical cancer, but may not be common in the

38

western world of medicine. An early visit would also help in understanding the prevalent
infections that affect the general population. In Malawi, these are food or waterborne
diseases, such as hepatitis A and typhoid fever, as well as vector-borne diseases, such as
malaria and dengue fever [22]. Including these in future studies would be useful when
testing for significant predictors of cervical cancer in the data analysis portion of the
study.
Another suggestion would be using knowledge from this study regarding the most
common types of birth control used—sterilization, depo-provera, and withdrawal—and
researching other types of terms common used to describe these terms, as each culture
has its own set of ways in which to express such terms. For example, in Malawi, “tubal
ligation” was a term used interchangeably with “sterilization,” so it is possible that there
were individuals who used tubal ligation but responded “none” because they were not
familiar with the synonymous term “sterilization.”
Both efficiency and accuracy may be served by adapting the instrument to the
local language. For example, having the first of the three sections of this survey
instrument, the background history, written and conducted in the native language,
Chichewa, the official language used by 57% of the population [10] may also aid in
increasing response rate for the questions. Certain terms in the second section, the
physical exam, could also be included in both the native and English form. This would
allow the examiner to use both versions and increase the accuracy with which the data is
captured because the patient is more likely to understand what is being asked and what
the options are referring to.
2) Adding HIV and pregnancy tests if obtain an abnormal Pap smear test.

39

Because access to care is often limited in this region, cervical cancer is not the
only health predicament that these women must overcome. HIV in Malawi is another
infection that greatly affects these women’s day-to-day life and affects their ability to
contribute to their family and the greater community. Being HIV positive also comes
with its own set of quandaries because treatment including anti-retroviral therapy (ART)
is scarce in the region. But knowing that the individual is HIV positive influences
treatment decisions for cervical cancer, so even if the patient has been tested before,
including HIV testing as a part of the screening process could aid in prescribing more
effective treatment. On the same note, women are not always aware that they are
pregnant in the early stages, but being pregnant affects how the clinician conducts the
more intrusive exams, such as the acetowhite test, after the Pap smear yields abnormal
results. Therefore, both the HIV and pregnancy tests should be included as requirements
only for women whose Pap smears yield potential precancerous lesions.

3) Implement use of adjunct testing
Recent studies have been done to conduct adjunctive testing for cervical cancer in
such resource-poor settings in order to increase the sensitivity and specificity of these
screening results. A study done in low resource settings found that sequential testing of
both VIA and HPV DNA methods, followed by colposcopy and biopsy, yielded high
specificity and sensitivity and has been suggested for future studies aimed in these
regions [23]. Therefore, allocating more time researching the benefits of using multiple
exams, as well as the relative accuracy of specific algorithms whereby they would be
employed, should be considered in the future.

40

References

[1] Preparing for the introduction of HPV vaccines: policy and programme guidance
for countries. World Health Organization Web site.
http://whqlibdoc.who.int/hq/2006/WHO_RHR_06.11_eng.pdf?ua=1 Published
2006. Accessed June 2, 2014.

[2] Latest world cancer statistics. International Agency on Research on Cancer
(WHO) Web site. .http://www.iarc.fr/en/media-centre/pr/2013/pdfs/pr223_E.pdf
Published December 12, 2013. Accessed June 15, 2014.

[3] Human Papillomavirus (HPV): Genital HPV Infection-Fact Sheet. Centers for
Disease Control and Prevention Web site. http://www.cdc.gov/std/hpv/stdfact-
hpv.htm Published January 23, 2014. Updated March 20, 2014. Accessed June 3,
2014.

[4] Gakidou E, Nordhagen S, Obermeyer Z. Coverage of cervical cancer screening in
57 countries: low average levels and large inequalities. PLoS Med.
2008;5(6):e132.

[5] Cancer Fact Sheets. International Agency on Research on cancer (WHO) Web site.

[6] Lee J. Screening of uterine cervical cancer in low-resource settings. Journal of
gynecologic oncology. 2012;23:137-138.

[7] Human Papillomavirus (HPV): HPV Vaccines. Centers for Disease Control
and Prevention Web site. http://www.cdc.gov/hpv/vaccine.html Published January
29, 2013. Updated February 5, 2014. Accessed June 3, 2014.

[8] ICO Information Centre on HPV and Cancer. Malawi. Human Papillomavirus and
Related Cancers, Fact Sheet 2013. HPV Information Centre Web site.
http://www.hpvcentre.net/statistics/reports/MWI_FS.pdf Published January 31,
2014. Accessed June 2, 2014.

[9] Markowitz LE. HPV Vaccines Prophylactic, Not Therapeutic. JAMA: The
Journal of the American Medical Association. 2007;298:805-806.

[10] Ports KA. Barriers and facilitators to cervical cancer prevention in Malawi:
Developing a strategy for HPV vaccination. ProQuest, UMI Dissertations
Publishing; 2012.

[11] World Health Organization. Cervical Cancer Screening in Developing Countries,
Report of a WHO consultation. Geneva, Switzerland: Programme on Cancer
Control, Department of Reproductive Health and Research; 2002.

41

[12] Fort VK, Makin MS, Siegler AJ, Ault K, Rochat R. Barriers to cervical cancer
screening in Mulanje, Malawi: a qualitative study. Patient preference and
adherence. 2011;5:125-131.

[13] Taulo F, Malunga E, Ngwira A. Audit of gynaecological cancers Queen Elizabeth
Central Hospital, Blantyre. Malawi Medical Journal. 2009;20.

[14] World Bank Country Data. World Bank Web site.
http://www.worldbank.org/data/countrydata/countrydata.html.
(http://epri.org.za/wp-content/uploads/2011/03/29-Malawi.pdf ). 2011. Accessed
June 3, 2014.

[15] Road Map for Accelerating the Reduction of Maternal and Neonatal Mortality
and Morbidity in Malawi. www.wcf-
uk.org/attachments/article/251/Malawi_Road_Map.pdf. Published October 2005.
Accessed June 4, 2014.

[16] The PAPS Team International. http://www.the-paps-team.org/. 2009. Accessed
June 4, 2014.

[17] REDCap. Institute for Clinical & Translational Science at the University of Iowa
Web site. https://www.icts.uiowa.edu/confluence/display/REDCapDocs/REDCap.
Last updated May 16, 2013. Accessed June 4, 2014.

[18] REDCap Workshop. Feinberg School of Medicine, Northwestern University Web
site. http://www.feinberg.northwestern.edu/research/docs/REDCap.pdf . Accessed
June 4, 2014.

[19] Office Developer Tab. http://office.microsoft.com/en-us/word-help/create-forms-
that-users-complete-or-print-in-word-HA101848148.aspx . 2010. Accessed June 4,
2014.

[20] Li W, Venkataraman S, Gustafsson U, Oyama JC, Ferris DG, Lieberman RW.
Using acetowhite opacity index for detecting cervical intraepithelial neoplasia. J
Biomed Opt. 2009;14(1):014020.

[21] Dress. Friends of Malawi Web site.
http://www.friendsofmalawi.org/learn_about_malawi/culture/dress.html.
Published 2004. Accessed June 15, 2014.

[22] The World Factbook: Field Listing:: Major Infectious Diseases. Central
Intelligence Agency (CIA) Web site.
https://www.cia.gov/library/publications/the-world-factbook/fields/2193.html
Accessed June 20, 2014.

42

[23] Bhatla N, Puri K, Kriplani A, et al. Adjunctive testing for cervical cancer
screening in low resource settings. Aust N Z J Obstet Gynaecol. 2012;52:133-139.

43

Appendix A: Original Survey Instrument

44

45

Appendix B: Revised Survey Instrument

46

47

48

Abstract (if available)

Abstract Regions with the highest burden of cervical cancer are often those with the most limited access to resources and care. This is especially true for countries in southeast Africa, which carry the greatest burden of this disease, such as Malawi. Factors contributing to such high incidence and mortality rates include not only limited accessibility, misunderstanding of disease, and economic barriers, but also ineffective data capturing instrumentation used by clinicians in the field. To identify the characteristics of survey instrumentation that are found to yield missing or inapplicable responses, data from the 2012 PAPS cervical cancer intervention in Malawi was studied. In this weeklong intervention, 1,221 Malawian patients were seen and screened for cervical cancer through use of Pap smears. Of the 1,210 with evaluable Pap smear tests, 124 were found to yield abnormal results. Yet, of equal interest, was the pattern of missing values and inapplicable responses that were found in each section of the survey instrumentation, administered during the patient consultations in the clinic. Qualitative and quantitative reasoning was used then to analyze and justify what the most effective and ineffective ways in which data was captured in this instrument so an improved version of the same tool could be produced and is included in Appendix B. It was found that missing data was least commonly found in questions that had a pre-set list of options, and used a “checkbox” format. This research is imperative to improving the efficiency with which the data can be collected and used to best treat the population at risk, in this case underserved Malawian women who have rarely been screened for cervical cancer prior to this PAPS site visit.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Cervical cancer study in Malawi: a novel design for field investigation

PDF

A comparison of three different sources of data in assessing the adolescent and young adults cancer survivors

PDF

Association between informed decision-making and mental health-related quality of life in long term prostate cancer survivors

PDF

Origins of the gender disparity in bladder cancer risk: a SEER analysis

PDF

The study of germ cell tumors and related conditions: an analysis of self-reported data with characterization and comparison of Family History Questionnaires respondents and non-respondents

PDF

Predictive factors of breast cancer survival: a population-based study

PDF

Disparities in colorectal cancer survival among Latinos in California

PDF

CAFÉ (Common Application Framework Extensible) - a framework to support data capture for clinical trials

PDF

Air pollution and breast cancer survival in California teachers: using address histories and individual-level data

PDF

Red and processed meat consumption and colorectal cancer risk: meta-analysis of case-control studies

PDF

Incidence and survival rates of the three major histologies of renal cell carcinoma

PDF

Obesity paradox in acute heart failure decompensation

PDF

Methodological approaches to assessing diurnal cortisol rythms in epidemiological studies: how many salivary samples are necessary?

PDF

Associations between isoflavone soy protein (ISP) supplementation and breast cancer in postmenopausal women in the Women’s Isoflavone Soy Health (WISH) clinical trial

PDF

Fish consumption and risk of colorectal cancer

PDF

Racial/ethnic differences in colorectal cancer patient experiences, health care utilization and their association with mortality: findings from the SEER-CAHPS data

PDF

Carboplatin and vincristine chemotherapy for progressive low grade gliomas in pediatric patients with or without neurofibromatosis type 1 (NF1)

PDF

Gene expression and angiogenesis pathway across DNA methylation subtypes in colon adenocarcinoma

PDF

Validation of the Children’s International Mucositis Evaluation Scale (ChIMES) in pediatric cancer and SCT

PDF

Pathogenic variants in cancer predisposition genes and risk of non-breast multiple primary cancers in breast cancer patients

Asset Metadata

Creator Sikri, Ruhi (author)

Core Title Development and validation of survey instrument designed for cervical cancer screening in Malawi, and other low resource settings

School Keck School of Medicine

Degree Master of Science

Degree Program Applied Biostatistics and Epidemiology

Publication Date 10/27/2014

Defense Date 08/06/2014

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag cervical cancer,Malawi,OAI-PMH Harvest

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Cortessis, Victoria (committee chair), Azen, Stanley P. (committee member), Muderspach, Laila (committee member)

Creator Email ruhsikri@gmail.com,sikri@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-509199

Unique identifier UC11288273

Identifier etd-SikriRuhi-3034.pdf (filename),usctheses-c3-509199 (legacy record id)

Legacy Identifier etd-SikriRuhi-3034.pdf

Dmrecord 509199

Document Type Thesis

Rights Sikri, Ruhi

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