A Cervical Cancer Screening Trainer for Use in Low-Resource Settings

Each year, cervical cancer causes the death of over 275,000 women worldwide with eighty percent of these cases occurring in low or lower-middle income countries. Cervical cancer screening programs reduce cervical cancer occurrence by identifying and treating pre-cancerous cervical abnormalities before they develop to malignant stages. Standard methods of screening are inappropriate for use in low-resource settings and therefore screening is very limited. Based on a design ethnography study performed over a two-month period in Ghana, a need was identified for a low-fidelity model to assist in training healthcare providers to screen for cervical cancer using visual inspection with acetic acid. The design presented in this paper is a box trainer with a simulated vaginal cavity that allows a user to insert cervical images displayed on plastic tabs or cards from the Jhpiego Visual Inspection of the Cervix Flash Card Set © . A modular electronic feedback mechanism trains the user to properly complete the steps of the screening procedure and to track the successes of their diagnoses. This paper describes the need for a cervical cancer screening trainer, the process used to design a trainer to support visual inspection with acetic acid, and the results of prototyping and preliminary usability tests. Index Terms – Cervical cancer screening, visual inspection with acetic acid, medical training, design ethnography, global health design International Journal for Service Learning in Engineering, Humanitarian Engineering and Social Entrepreneurship Vol. 11, No. 1, pp. 1-18, Spring 2016 ISSN 1555-9033


INTRODUCTION
Throughout the world, the burden of cervical cancer fatality is most prominent in low and lowermiddle income countries (LMICs). 1 This is primarily due to a lack of infrastructure for traditional cervical cancer screening techniques. 2 Visual inspection of the cervix with acetic acid (VIA) is an effective low-cost method to screen for cervical cancer in women in LMICs, but it is not widely used. 3VIA requires experiential knowledge, which can be gained from performing the procedure in a clinical or simulated setting, in order to accurately diagnose precancerous cells.The lack of sustained use of the VIA in LMICs is due in large part to a lack of awareness and training of the method. 4The goal of this work was to address training shortcomings by developing and assessing a simulation based model to assist in training midwifery students, midwives, and other healthcare providers in LMICs to accurately screen for cervical cancer using VIA.

Cervical Cancer Screening
Each year, cervical cancer causes the death of 275,000 women worldwide. 2Eighty percent of these cases occur in LMICs. 3Screening for cervical cancer has been shown to significantly reduce mortality, given that most cervical cancer cases caught in early stages are treatable.However, there is a lack of awareness of screening methods as well as healthcare providers trained to perform screenings in resource-limited settings. 3,4,5he most commonly used Western method of the Pap smear is rarely performed in LMICs because of its cost and reliance on cytology. 6Alternative screening methods to cytology (Pap smear) include: colposcopy, visual inspection with acetic acid and magnification (VIAM), cervicography, HPV screening, and visual inspection with acetic acid (VIA).Colposcopy allows a healthcare provider to gain a magnified view of the patient's cervix using a colposcope, or specialized magnifying device. 7VIAM refers to the more general set of magnification screening tests under which colposcopy is classified.While some studies show no difference in outcomes for the two methods, 8 further research is needed to confirm the advantages of one method over the other.Cervicography involves taking a photo of a magnified cervix and sending the image to a remote diagnostician, 9 and has been used as both an adjunct 10 and an alternative 11 method of cervical cancer screening.HPV screening involves testing a sample from the patient's cervix for human papillomavirus (HPV) and referring the patient to further treatment upon finding a positive result.Previous studies show that HPV screening is the most effective screening method to reduce the rates of cervical cancer, 12 but it is also necessitates sophisticated screening equipment and therefore increased costs.
Visual inspection with acetic acid (VIA) is a low-cost and easy-to-perform cervical cancer screening method that immediately informs the healthcare provider of the patient's cervical abnormalities.The procedure was specifically designed for use in LMICs due to its reliance only on commonly found materials and training. 13The VIA is performed as follows: (1) a healthcare professional performs a standard pelvic examination and inserts a speculum into the patient's vaginal cavity, (2) the healthcare professional visually inspects the patient's cervix for any existing abnormalities and then applies acetic acid (or white table vinegar) to the cervix, (3) the healthcare professional visually inspects the cervix one minute after application of acetic acid for any acetowhite lesions that may appear, signifying pre-cancerous cells, and (4) upon visualization of acetowhite lesions, the healthcare professional either refers the patient to specialized medical care or performs cryotherapy to remove the pre-cancerous cells. 5,14ne major drawback of VIA is that it cannot be used on post-menopausal women because the squamocolumnar junction is not visible to the naked eye.VIA is known to have a high falsepositive rate in this population. 5However, there is some evidence that improved training may reduce this rate. 15A recent meta-analysis found VIA to have 80% sensitivity (consistent with previous measures at approximately 79% sensitivity) and 92% specificity (an increase from the previously measured 85% specificity) for early detection of cervical cancer. 14This meta-analysis also concluded that VIA is a more effective and cost-efficient alternative to cytology in LMICs.
Although designed specifically for use in low-resource settings, visual inspection with acetic acid is not widely used due to a lack of training, and therefore a lack of healthcare providers skilled in performing VIA.The VIA depends on experiential knowledge and the acetowhite lesions are often difficult to see or distinguish from other non-premalignant cervical abnormalities.Therefore, it is imperative that healthcare providers performing VIA screenings are sufficiently trained to recognize and interpret a wide variety of visual screening outcomes. 5

Effects of Clinical Simulation on Training for Cervical Cancer Screening
Clinical simulators ranging in sophistication have been used in medical training for many years. 16,22Simulators vary based on their engineering or physical fidelity, as well as by their psychological and procedural or functional fidelity.Previous analyses of multiple simulators have shown that increases in simulator engineering fidelity lead to large increases in costs of materials and manufacturing, while leading to minimal improvements in trainee performance.Conversely, functional fidelity has proven extremely important in clinical simulator design in order to optimize "skill transfer" from simulated to actual patient interaction. 17Multiple studies have shown that high functional fidelity can be achieved with low engineering fidelity devices. 18,19Additionally, there have been successful low-cost simulator based training programs in multiple areas of medical training including trauma, maternal, and infant health. 20,21ne essential requirement to successfully screen for cervical cancer in LMICs is appropriate training programs directed towards healthcare providers.These programs help ensure proper implementation and high participation in the screening method. 13The Alliance for Cervical Cancer Prevention (ACCP) asserts, "training should be done in a clinical setting that resembles the service-delivery conditions at the program site."The training should consist of a protocol that first includes simulation models followed by clinical practice with patients.Advantages of model-based clinical simulation are exposure to a range of clinical scenarios and the ability to practice an invasive procedure without putting a patient at risk. 22

Cervical Cancer Screening Training Technologies
A variety of existing simulation and educational methods are used in cervical examination and training applications.Jhpiego, a nonprofit organization focused on global health issues in LMICs, developed the Visual Inspection of the Cervix Flash Card Set © (flashcards).The flashcards (Figure 1) were created to supplement training, determine competency, and allow for selfreview. 23Jhpiego's Atlas of Unaided Visual Inspection of the Cervix © is used as a reference in VIA screening and shows a variety of VIA screening outcomes with distinct clinical images. 24revious research supports the use of training with cervical images (i.e., Jhpeigo flashcards) in place of training with live patients.One study compared the assessment of cervical images taken after application of acetic acid by three different physicians.The researchers found a 67% level of agreement between the three physicians' assessment of 144 images. 25The ACCP reviewed several cervical cancer screening training programs and identified that ideal training should be done with simulation models initially. 13By themselves, neither the atlas nor the flashcards offer human interaction, feedback, or a realistic clinical situation.

FIGURE 1 FRONT AND BACK OF A JHPIEGO FLASHCARD
Pelvic models are used to offer realistic clinical situations for students in a variety of medical training programs.The use of pelvic models to practice clinical examinations results in a decrease in the time necessary for trainees to reach performance competency by offering a humanistic experience and reducing stress. 26Despite this, no pelvic models are currently intended for VIA training.

Design Ethnography
A design ethnography study was performed during May and June of 2013, in the Upper East, Ashanti, and Greater Accra regions of Ghana to identify unmet women's health needs as part of the Global Health Design Specialization at the University of Michigan. 27Design ethnography techniques such as observations, interviews, and focus groups were used to gain a deep understanding of stakeholders' needs and cultural and social contexts in order to increase inform front-end design decisions. 28Ghana was chosen as the site of interest due to a previously established 25-year partnership between the University of Michigan and teaching hospitals in Ghana. 29,30In Ghana, where cervical cancer is the number one cause of cancer-related death for women, less than 5% of women have been screened. 31There are limited organized efforts in Ghana to ensure that women are screened often and accurately for cervical cancer. 32Following the completion of the design ethnography study, the development of a training model to support midwives and other healthcare providers as they learn to screen for cervical cancer was pursued.
More than 80 stakeholders were contacted at 10 different health facilities across Ghana.Health facilities included teaching hospitals, district hospitals, medical school simulation centers, community health clinics, and government health facilities.Upon return to the University of Michigan, the design of the trainer was completed throughout a single semester capstone design course followed by a single semester independent design course.During these semesters, user requirements and engineering specifications were developed, several iterations of prototypes were manufactured, and preliminary evaluations of prototypes were performed.

User Requirements and Engineering Specifications
A ranked list of user requirements was developed taking into account: (1) design ethnography observation and interview results in Ghana, (2) interviews with healthcare professionals in the United States, and (3) supplemental literary research including benchmarking.Ranking was performed through team discussion and information source analysis to identify the impact and relative importance of each requirement.Ghanaian gynecologists, obstetricians, oncologists, midwives, and nurses were also asked to rank the list of user requirements, which served to validate the finalized rank ordering.The engineering specifications were developed with subject matter expert input, a wide-ranging literature review, and clinical observations.The user requirements and engineering specifications are shown in Table 1, organized by the importance of each requirement.Specific justification for each engineering specification is given in the following paragraph.
Benchmarking from a standard cervical model, the Jhpiego Visual Inspection of the Cervix Flash Card Set © , and the Jhpiego Atlas of Unaided Visual Inspection of the Cervix © informed the development of several engineering specifications (1.2, 2.1, 2.3, 2.4). 23,24,33 ccording to stakeholders in Ghana, the current VIA training generally takes three months, but it was determined that the trainer should shorten the necessary training period (1.1).Clinical observations and interviews in Ghana informed the requirements ensuring the trainer's ability to match the VIA procedure and to signal the user's correct diagnosis (2.2, 2.5, 2.6, 4.1, 4.2, 5.2, 6.1).The trainer's ability to match the VIA procedure would also need to be realistic according to a group of trained users (2.7).To determine the anatomical dimensions necessary for the trainer, several human anatomy resources were referenced (3.1, 3.2, 3.3, 3.4). 34,35Durability requirements were determined jointly through a study by the Alliance for Cervical Cancer Prevention and an interview with an end-user (5.1, 5.3). 13To define ease of use, a study of the learning curve necessary to use a virtual reality medical trainer was referenced (6.2). 36 Portability, important for ensuring that Ghanaian midwives could easily transport the trainer by motorbike as they travel to rural clinics and smaller healthcare facilities, was defined by the ability of the trainer to fit in a typical backpack, while leaving room for other items (7.1, 7.2).Cost-effectiveness of the trainer was difficult to quantify, as there are limited funds currently spent in Ghana on preventative healthcare.Clinical observations in family planning clinics provided information for the costs of high-fidelity trainers, but these amounts far exceeded what stakeholders deemed as appropriate for a low-fidelity trainer.Therefore, a reasonable, yet arbitrary price point was established as the constraint (8.1).Finally, interviews with multiple healthcare providers and officials in Ghana emphasized the desirability of domestic manufacturing to increase the likelihood of maintainability and cost-effectiveness of the trainer while serving to empower the community and the local economy (9.1).

Concept Generation and Design Iteration
Over 50 concepts were generated using a variety of ideation methods, including team brainstorming and brainsketching, 37,38 analogical thinking, 39 Design Heuristics, 40 and functional decomposition. 41These concepts were synthesized, iterated on, and down-selected through several phases of sketch modeling, CAD modeling, and physical prototyping.Multiple variants of the design were developed during two semesters of design based coursework.Remote and inperson feedback from stakeholders at the University of Michigan and healthcare centers in Ghana was obtained at multiple points throughout the design process to confirm design decisions and inform redesign.Two exemplary iterations of prototyped concept solutions are described below along with the evaluation to inform subsequent redesign.
TWO SELECT PROTOTYPED DESIGN ITERATIONS FROM LEFT TO RIGHT: "IMAGE WHEEL" "PULL-TAB" An initial fully functional prototype, the "image wheel prototype" (Figure 2) featured a wheel with 16 different overlapping cervical images.The user would first see a pre-image of a cervix (before acetic acid was applied) and then would be able to pull a tab to reveal a post-image (after acetic acid was applied) of the same cervix.The image wheel also made possible the ability to randomize the cervical image a user would be tested on by spinning the image wheel.This prototype was initially assessed during a focus group held with eight midwifery students from the University of Michigan School of Nursing.Midwifery students used the trainer prototype and completed a survey on the model's physical attributes, realism of experience, ease of use, and functional fidelity.The survey responses resulted in a score of 3.38 out of 5 for realism of experience.Key themes from the focus group included the high procedural fidelity of inserting a speculum and viewing the 'cervix.'Design decisions for the vaginal canal feature were confirmed and are included as features in the refined concept solution described in detail below.This focus group also informed a redesign for the mechanism to view pre-and post-images of the cervix as well as and reselection of material to be used for the simulated labia.Another exemplary prototype, the "pull-tab" prototype (Figure 2) featured removable multiimage pull-tabs and a modular electronic component.Each pull-tab contained a pre-image and post-image of the same cervix.These tabs are a feature of the refined prototype described below.This prototype was brought to Ghana in March 2014 to gain direct feedback from Ghanaian stakeholders including doctors, midwives, midwifery instructors, and midwifery students.Multiple design changes were informed by this evaluation trip including modifications to the electronic component and incorporation of Jhpeigo flashcards into the trainer.The resulting concept solution and subsequently referred to as the Visualize trainer, is described below.

VISUALIZE: A BOX TRAINER FOR VIA TRAINING
Visualize is a box trainer constructed of materials available in Ghana, including pressboard, PVC pipe, latex foam, and assorted hardware (Figures 3 and 4).The Visualize prototype includes a simulated vaginal cavity, made with PVC pipe and latex foam, which allows the user to insert a speculum.The user can either insert a Jhpiego flashcard or a plastic tab with cervical images, both of which are visible by looking through the simulated vaginal cavity.An Arduino microcontroller and LCD screen provide an optional modular electronic feedback mechanism to guide the user through the steps of the VIA.The Visualize trainer opens from the back, revealing a storage compartment.Here, the user can store the speculum, sponge-holding forceps, tab set, and flashcard set.This storage compartment also provides the user with easy access to the battery for the Arduino and LCD screen module.Eight multi-image laser cut tabs were created to display images of a cervix both (pre-image) before and after (post-image) application of acetic acid; hardboard or cardboard could be used to display the images if a laser cutter is not accessible.An Arduino microcontroller and LCD screen provide optional modular electronic feedback to guide the user through the VIA procedure.
When using the tab method, the LCD screen prompts the user to select a specific numbered tab.The tab is then fully inserted into the opening on the left side of the box trainer.The user is then prompted via the LCD screen to insert the speculum, inspect the cervical pre-image, and use the sponge-holding forceps to apply acetic acid to 'cervix'.Next, the user starts a one-minute timer on the LCD screen.After one minute, the user is prompted to partially retract the tab in order to visualize the cervical post-image and make a diagnosis on whether or not the cervix pictured on the tab is pre-cancerous, cancerous, or neither.If the cervix is cancerous or precancerous, the correct diagnosis is VIA positive.If the cervix is neither, the correct diagnosis is VIA negative.The user presses either VIA-positive (button A) or VIA-negative (button B). (Buttons are not shown in Figure 4).After the user makes his/her selection, the LCD screen displays the correct response.The simulator also has the ability to allow a user to complete ten different tabs presented in a random order, before outputting the results, e.g., "X out of 10 cervical images diagnosed correctly".
For the flashcard method, the user chooses a random flashcard to insert into the opening in the top of the box and then inserts a speculum into the simulated vaginal opening.Next the user visualizes the pictured cervix through the simulated vaginal opening and makes a diagnosis.After doing so, the user can check the correctness of his/her diagnosis by retracting the card and reviewing the information on the back.Note, the flashcard method does not utilize the modular electronic feedback system.Midwifery students with varying levels of experience performing internal gynecological exams were asked to insert, expand, and remove a speculum into the different simulated vaginal cavities and rate how well the simulated vaginal cavity accurately represented the insertion, expansion, and removal of a speculum during an actual cervical exam.The midwifery students were asked to rate each simulated vaginal cavity on a scale of 1 (strongly disagree) to 5 (strongly agree) to determine the extent to which the materials realistically represented inserting, expanding, or removing a speculum in the vaginal cavity.As shown in Table 3, the test revealed that a low-cost sponge was most realistic.
As a result of this test, it was determined that it was not necessary to use silicone, a highfidelity yet expensive material, to simulate the vaginal cavity.A similar materials test was performed in Ghana with a number of stakeholders with the addition of a simulated vaginal cavity made with latex foam, a locally available material with demonstrated similar properties to silicone and sponge.The feedback from Ghanaian stakeholders confirmed that latex foam was optimal for use and therefore was used in all future prototypes.Prototype evaluation performed in Ghana allowed for research and feedback on the potential for the design to be manufactured locally in Ghana.Engineers at Kwame Nkrumah University of Science and Technology in Kumasi confirmed that microcontrollers are available to import and are appropriate for use in Ghana.Furthermore, interviewers informed that local artisans from nearby wood villages would be able to construct the trainer.All other materials were found to be locally available by visiting Central Market in Kumasi and speaking with local shop owners.
Several specifications were either not met or not fully evaluated (1.1, 2.7, 5.1, 6.2, 8.1).Specification 1.1 (<3 months for a user to perform VIA accurately and independently) requires longitudinal evaluation and therefore has not yet been addressed in the current work.Specification 2.7 (focus group score for realism of experience >3) has been preliminarily evaluated for the vaginal cavity as shown in Table 3, but additional work is needed to evaluate the specification for the trainer as a whole.Specification 5.1 (≥ 5950 uses) requires a more rigorous evaluation, as the trainer was only evaluated once for up to 600 uses.Specification 6.2 (< 138 seconds for new user to learn correctly operate trainer) was tested on a small sample of engineering students (N=3), and therefore requires a more rigorous evaluation with the target user in the intended implementation setting.Specification 8.1 (≤ GHc50 to manufacture and distribute (~20 USD)) was not met, as the modular electronic component significantly drove up the price of the trainer.

DISCUSSION
The Visualize prototype achieved the objective of creating a low-fidelity trainer to help teach Ghanaian midwives to screen for cervical cancer using visual inspection with acetic acid.There is currently no formal model-based cervical cancer screening training program in Ghana.Several organizations have attempted to address the issue of cervical cancer detection and early treatment, but none through the lens of trainer-based education.The use of low-fidelity trainer has the potential to improve cervical cancer screening training and subsequently, cervical cancer detection in LMICs.In Ghana for example, VIA training, including simulation based learning, could be provided to midwifery students during their formal education and to practicing midwives through continuing professional development (e.g., workshop).
The selected concept solution leverages the advantages of both the tab and flashcard methods.By using Visualize, the user is able to practice the entire VIA procedure, including insertion of the speculum, application of acetic acid, and visualization of the change in the cervix, which would not be possible with only the flashcards.However, the user also benefits from the breadth of cervical situations presented in the Jhpiego flashcard set and is able to practice with a much larger set of possible outcomes.
Visualize's positive design attributes include its modular electronic component, its ability to provide a strong experiential experience, and its portability.The trainer's electronic component is completely modular, enabling the simulator to work as either an electromechanical or a purely mechanical system.This is beneficial for potential implementation in low and lower-middle income countries because the electronic component may further engage users and provides immediate feedback, but the trainer still remains fully functional in the absence or failure of the electronics.While the electronic component does add a significant cost to each trainer, stakeholders in Ghana expressed a strong interest in having a feedback system that allows for individual training and practice.Furthermore, Visualize provides the user with an opportunity to gain valuable experiential knowledge.It supports current cervical cancer screening theoretical education and enhances a midwife's ability to correctly perform and diagnose the outcome of a cervical cancer examination.Visualize also allows midwives to visualize an increased number of unique cervical cases than they would be able to observe on live patients.Additionally, Visualize can be easily transported by motorbike, a common mode of transportation for midwives working in rural areas of Ghana.
Visualize's design shortcomings include its relatively high cost and its lack of completed validation.The LCD component significantly increases the price of the trainer.As it stands, an individual prototype costs approximately 32 USD, exceeding the somewhat arbitrary price point of 20 USD by 12 USD.This price reflects the cost of materials only, and does not take into account the labor necessary to manufacture and distribute.Future work includes durability and time to proficiency evaluations.
The overarching goals of making VIA training more accessible throughout Ghana and encouraging screening proficiency for a large number of midwives informed the design process used to create Visualize.For example, a concept solution that augmented current training practices and educational models was intentionally pursued over a "blue-sky"/disruptive innovation for cervical cancer screening.While advantageous from a potential short-term implementation standpoint for Ghanaian midwives and midwifery programs, this approach was also advantageous from a student design project standpoint; simulator-based designs typically don't require regulatory approval and are more conducive to single-year design project budgets and timelines. 42The pursuit of an educational/training tool design versus a medical device design also facilitated active engagement among stakeholders, probably due to the increased likelihood that a student-based design solution in the simulation space was achievable within the relatively short timeframe.
Future work includes performing additional validation and usability studies and iterating on the design as needed.Prior to implementation, training strategies and curricula would need to be developed and assessed in collaboration with subject matter experts in Ghana.

CONCLUSION
The need for a tool to aid in training healthcare providers to accurately screen for cervical cancer in low and lower-middle income settings was established through a design ethnography study in Ghana.The trainer design described in this paper aimed to enhance, support, and add an experiential education component to the theoretical training currently being performed in Ghana.Cultural and clinical contexts were considered throughout the design process to increase the likelihood for adoption.Preliminary testing and feedback suggests that the concept solution could improve VIA training, but additional testing is required.Future work involves exploring the marketability of the concept and delivering prototypes to interested end-users in Ghana.

FIGURE 3 COMPUTER
FIGURE 3 COMPUTER RENDERING OF VISUALIZE.DIMENSIONS ARE IN CENTIMETERS.

Figure 5 depicts
Figure 5 depicts Visualize's stages of use.The user can choose to either insert a Jhpiego flashcard or a customized tab.The Jhpiego flashcards show only an image of a cervix after application of acetic acid, but they provide more details about the diagnosis and describe 115 clinical scenarios.Eight multi-image laser cut tabs were created to display images of a cervix both (pre-image) before and after (post-image) application of acetic acid; hardboard or cardboard

FIGURE 6 SIMULATED
FIGURE 6 SIMULATED VAGINAL CAVITIES USED FOR MATERIALS TESTING TOP ROW FROM LEFT TO RIGHT: SILICONE, COTTON, RUBBER PLACEMAT, STYROFOAM BOTTOM ROW FROM LEFT TO RIGHT: RUBBER LATTICE, FOAM PAPER, SPONGE

TABLE 3 MATERIALS
FIDELITY TESTING RESULTS (N = 5, SCALE: 1 -STRONGLY DISAGREE THAT THE MATERIAL REALISTICALLY REPRESENTED INSERTING, EXPANDING, OR REMOVING A SPECULUM INTO, IN, OR FROM THE VAGINAL CAVITY, 5 -STRONGLY AGREE THAT THE MATERIAL REALISTICALLY REPRESENTED INSERTING, EXPANDING, OR REMOVING A SPECULUM INTO, IN, OR FROM THE VAGINAL CAVITY.)