Background
Female genital fistula results in uncontrollable leakage of urine and/or feces, and is often caused by obstructed labor without timely or adequate management. Globally, between 1 and 2 million women are living with fistula and thousands of new cases occur annually [
1]. Fistula disproportionally affects poor and marginalized women and girls who lack timely access to quality obstetric care and to follow-up services required in addressing maternal morbidities [
2‐
4]. Policies and programs in low- and middle- income countries (LMICs) inadequately address barriers faced by women living with fistula in accessing treatment services. Women living with fistula frequently experience humiliation, isolation and stigma, among many other barriers, that prevent them from seeking and receiving comprehensive care and treatment, including surgery and follow up support services. Studies indicate that barriers cut across psycho-social, cultural, awareness, social, financial, transportation, facility shortages, quality of care, and political dimensions [
5]. The complexity of these barriers and relative rarity of the condition create measurement challenges for countries seeking to monitoring access to fistula care.
Understanding the layered barriers and intersectional vulnerabilities of poverty, gender, stigma, and geography faced by women in need of fistula care is useful in better targeting fistula services and designing programs to reduce obstacles to care seeking and treatment [
6]. In sub-Saharan Africa, qualitative studies over the last decade have explored multiple barriers faced by women living with fistula [
5,
7‐
12]. Financial barriers observed across studies include actual and perceived costs of transportation to hospitals and ancillary services during surgery and the post-operative recovery period for women and accompanying family members [
8,
10,
13,
14]. Women also face deterrents and delays in care-seeking related to health systems quality (e.g. inadequate antenatal, childbirth and postnatal care and limited health worker knowledge) as well as gaps in individual and community awareness of fistula as a condition, its causes, consequences, and treatment options [
15,
16]. Additionally, psychosocial barriers intersect with gender dynamics in the household to leave many women isolated and reinforce a deep sense of shame and low self-worth. These barriers may affect women’s ability to manage self-care in the home, engage in socioeconomic and cultural activities, or decide to seek repair and reintegration services [
7,
9,
11,
17,
18]. While barriers to fistula care have been described qualitatively in several LMICs, limited effort has been made to quantitatively and comprehensively measure these factors, in a standardized and comparable way.
Quantitative fistula studies to date primarily focus on prevalence estimation (epidemiology) or profile experiences for clinical care [
1,
19‐
21], with limited attention to measuring the range of barriers to fistula care. These studies have, however, demonstrated aspects of women’s experience with fistula that are conceptually and empirically (cross-sectionally) associated with barriers to seeking, reaching, and receiving fistula care. In Nigeria and Uganda, according to a cross-country survey, the median length of time a woman lives with fistula before seeking care (~ 1 and ~ 2 years, respectively) is likely associated with the nature and magnitude of barriers she faces [
20]. A quantitative analysis of Demographic and Health Survey (DHS) data in sub-Saharan Africa, suggests that age and levels of education may have indirect associations with barriers and enablers of care; namely, those with higher education levels may be more likely to attend antenatal care, have better birth preparedness and elevated knowledge of fistula prevention and care options [
22]. The sociocultural and health systems contexts of fistula care-seeking are important to consider, including spousal, family, community social support or lack thereof on the one hand and formal and informal care alternatives (e.g. surgical/rehabilitation services and/or traditional healers) on the other [
20,
23,
24]. Given the often-hidden nature of the population living with fistula, limited data on the fistula burden in LMICs, and backlog of cases awaiting surgical treatment at hospitals, our current study relies on women who have reached care to make inferences about those who have not.
Composite measures (scales and indices) offer an opportunity to better understand barriers to fistula care and design interventions aimed at reducing obstacles and increasing equity-promoting solutions. While the cross-country DHS study showed seven discrete reasons for not seeking fistula care-ranging from unawareness of fistulas repairability and service locations, to expense and distance, to embarrassment, to limited permission to seek care, and poor quality care– these were not cumulatively analyzed at the level of the woman herself [
22]. Composite measures are of growing interest in maternal, newborn, and child health globally, e.g., related to the increased focus on experienced quality of care [
25‐
27]. Scales measure latent concepts like “barriers” and are recognized as acceptable in health program settings that focus on marginalized or stigmatized populations as seen in the HIV and mental health fields [
28,
29]. Given the layered, interrelated, yet conceptually congruent nature of the barriers experienced by women with fistula across LMICs countries [
5] and documented in Nigeria and Uganda in particular [
15,
16], composite scales related to this maternal morbidity may deepen our understanding of a complex psychometric phenomenon and offer measurement tools for wider research and programmatic use.
No composite measure of barriers has been developed to our knowledge in the context of fistula repair. Our study aimed to develop and validate composite measures of barriers to accessing fistula care currently faced by women living with fistula in Nigeria and Uganda. The resulting measures—indices and scales—can be used to inform future planning, programming and resource allocation for fistula care.
Methods
Study design and setting
This study applied an iterative approach, guided by scale development principles and item reduction [
30], including exploratory and confirmatory factor analyses to develop a quantitative composite measure assessing barriers to fistula care. The study was nested within an evaluation of a complex social behavior change (SBC) intervention to reduce barriers to fistula care in Nigeria and Uganda under the USAID-funded Fistula Care
Plus Project implemented by EngenderHealth [
23].
Implemented among fistula patients at accredited fistula repair centers in Ebonyi in southern Nigeria, Katsina in northern Nigeria, and the Central 1 sub-Region of Uganda, this study validated the composite measure in multiple settings. The fistula centers in Nigeria were located at well-functioning hospitals that provide surgical fistula repair on a routine basis, and treat women with fistula in a timely manner, with limited case backlog. Katsina experienced greater backlog than Ebonyi due to the higher estimated fistula burden in the north of Nigeria. The fistula center in Central 1 sub-Region, Uganda, one of the country’s leading fistula treatment sites, offers mainly pooled/camp-based services for fistula repairs, which generally indicates some level of case backlog at the community level. The estimated number of fistula cases in Ebonyi, Katsina and Central-1 sub-Region, Uganda was 696, 1895, and 758, respectively [
31].
Participant sample and method
The target population for use of these composite measures is any woman living with fistula; however, given the inability to sample this group at the population level, we relied on a study population of women with fistula who reached a care facility after experiencing similar barriers. Our study population comprises women aged 15 years and above who received fistula repair at the accredited fistula centers in Ebonyi and Katsina, Nigeria and Central Uganda. Women were invited to participate post-surgery during their two-week stay at the respective recovery wards in these centers. Trained data collectors approached potential participants at the fistula centers and engaged them in a comprehensive informed consent process. Given the study only interviewed “emancipated minors” –adolescents aged 15–17 years not living under the control of parents or guardians, married, or looking after their own child, no parent or guardian consent was required. After documenting informed consent by signature or thumbprint (for those with limited literacy), data collectors administered a post-repair client survey in the local language (Igbo and Hausa in Nigeria, and Luganda in Uganda). Post-repair client surveys were conducted at baseline and endline of the larger SBC intervention to develop and validate the composite measures. Surveys were paper-based in Nigeria and collected on tablets using Open Data Kit software in Uganda.
Measurement
Post-repair client surveys explored women’s sociodemographic characteristics, reproductive health and fistula history, and the barriers and enablers to fistula care they experienced (Additional File
1). The thematic categories of barriers and range of items that formed the composite measure were developed iteratively based on a literature review [
5], and qualitative exploration of local barrier constructs [
15,
16]. Similarly identified qualitative barriers reported across the study settings—through focus groups and in-depth interviews—allowed researchers to develop a harmonized set of items for Nigeria and Uganda. We assessed 43 items or barriers to fistula care that captured awareness of the condition’s causes, manifestations, and care options, restrictive cultural beliefs, gender norms reflecting women’s care-seeking ability, psychological consequences, social stigma, financial costs, transport and infrastructure, health care quality and interactions. Each item asked the degree to which a barrier item affected a woman’s access to fistula repair and was scored on a 4-point Likert scale with increasing options from ‘strongly disagree’, to ‘disagree’, to ‘agree’, to ‘strongly agree.’ In Uganda, the scale was administered as a 5-point Likert with a ‘neutral’ category.
Items were translated into Igbo, Hausa, and Luganda and back translated into English to check for meaning and comprehension. During the baseline assessment, the items were pretested with a several women in each fistula center prior to the study start to refine wording and understand the relevance of the full list of barriers. An open-ended section at the end of the post-repair survey asked about whether any additional barriers had been experienced by the participant, or whether any included barriers were deemed irrelevant. Fistula clients’ responses suggested the study team should retain all the items across the three sites at the start of the measure development process.
Analysis
Given the varied domains of barriers to fistula treatment identified in the literature review and formative research, we began with the assumption that this is a multidimensional construct. We applied an iterative process of item reduction and exploratory and confirmatory factor analysis using the baseline and endline samples, respectively. We reverse-coded positively worded items to create consistent response categories for barriers; each barrier was scored between 1 and 4 (‘strongly disagree’ to ‘strongly agree’). Neutral responses, which comprised less than 10% of the responses across items in Uganda were re-classified as “agree,” one of the mid-point categories, to ensure comparability of data across settings and retain statistical power. We explored distributions of all the items to assess uniformity in response categories for the psychometric analyses. We generated scale scores and evaluated the distribution of the full scale and sub-scales. All statistical analysis were conducted using Stata 13 software.
To determine the factor structure and assess construct validity, we conducted exploratory factor analyses (EFA) on the baseline sample to identify the potential number of factors that fit the data. The EFA specifications were selected based on an assessment of normality in the response distributions as well as of correlation among the potential factors. Following recommended criteria [
30], we conducted a principal component analysis (PCA), applying parallel analysis and scree plot review to identify the number of factors. We retained the number of factors with eigenvalues over 1.00. Within this plausible factor structure, we assigned items to factors based on face and content validity considerations through a consultative process and by exploring inter-item correlation matrices.
We did this in combination with an item reduction approach, removing items with extremely low (< 10%) and high (> 90%) frequencies that approached floor and ceiling effects, to ensure response variability. Items were removed if they had high proportions of missing data. We tested the normality of item responses to determine the type of factor analysis to implement. Items with low uniqueness (< 0.7), low factor loadings (< 0.3), and interpretability based on face and content validity were retained for the confirmatory factor analysis (CFA). We compiled the items excluded through this process but with high face and content validity into a separate index.
We conducted a CFA using endline sample data, to test the emergent factor structure from the EFA and retained items with statistically significant factor loadings (p < 0.05). Statistical tests measuring model adequacy were conducted and interpreted based on common criteria including the root mean square error of approximation (RMSEA), comparative fit index (CFI), Tucker-Lewis Index (TLI), and the standardized root mean square residual (SRMR). Modification indices were built into the CFA’s structural equation modeling to adjust for similarly worded items and assess goodness-of-fit. We also assessed scale invariance across sites to account for any sample differences between the distinct geographies of Ebonyi, Katsina, and Uganda.
Scale and index reliability and validity were assessed using standard correlation tests and related criteria. Reliability, or internal consistency of items, was assessed using Cronbach’s alpha and ordinal theta. The latter assumes polychoric correlation, useful when items have limited response categories (four in the case of assessed items). A standard reliability cutoff of 0.7 was adopted to assess adequate internal consistency, i.e., how well the items relate to one another to capture the overall construct of barriers to fistula care. We assessed convergent validity by exploring associations among scores with the theoretically correlated sub-scales and indices resulting from these measure development procedures. Predictive validity was assessed by exploring associations of the indices and sub-scales with variables that are theoretically expected to be correlated: the length of time a woman has been living with fistula and whether she has sought treatment in the past. Bivariate and multivariable predictive validity models were assessed, controlling for age, education level, and fistula repair center.
Discussion
To our knowledge, this is one of the first studies to develop a set of composite measures that quantitatively captures barriers to fistula care (Additional file
2: Supplemental Table 2). The multidimensional Barriers to Fistula Care Index comprises a series of latent sub-scales—Limited awareness, Social abandonment, and Internalized stigma—that can be used separately or together in research and program settings. The Financial/Transport Inaccessibility Index is conceptually distinct from the three sub-scales and overall Barriers to Fistula Care Index (Additional file
2: Supplemental Table 2). Composite measures demonstrate favorable psychometric properties, demonstrate functionality across settings, and are associated with years living with fistula and prior treatment. While the Barriers to Fistula Care Index and subscales primarily assess reasons why women are unable to obtain care and mostly relate to the ‘first delay’ in care-seeking, the Financial/Transport Inaccessibility Index speaks primarily to the ‘second delay’. The ‘third delay’ that relates to quality of health care was incongruent both conceptually and psychometrically given the data in our samples. This may be due to the rarity of the condition or to women’s limited exposure to and perceptions about relevant health services.
There are some limitations to the analysis presented here. We lacked power to sufficiently assess sub-scale stability across geographic areas. As this work was nested within a larger body of work and given the challenge of recruiting repaired women, let alone women still living with fistula, we were unable to draw a large enough sample in each individual site. Future studies should test for invariance across settings, including ensuring adequate sample size in each setting. Another limitation was the fact that our sample of women were post-repair; we have assumed similarity among the barriers faced by women who eventually reached fistula care and those women who have not yet done so. Ideally, barriers among women living with fistula who have not yet sought, reached, or received care would also be assessed to validate this assumption and adjust for any selection bias. However, in practice, this population is very difficult to identify and sample in a representative manner.
The sub-scales or dimensions of the Barriers to Fistula Care Index align with much of the qualitative literature regarding the experience of living with fistula and resonates with quantitative DHS analyses on treatment seeking in sub-Saharan Africa [
22]. Our study shows similar types and levels of barriers—including limited awareness, internalized stigma, social abandonment. It confirms that internalized stigma (e.g., that one is dirty) and being avoided or neglected by spouses, family and society (‘social abandonment’) as an commonly ascribed reasons for delayed care-seeking in sub-Saharan Africa [
5]. For example, a qualitative study found that obstetric fistula survivors in Addis Ababa coped with their emotions through suicidal thoughts, avoiding family and community members, and concealing their story and circumstances [
9]. As such, this sense of shame and compounded barriers often leads to women resorting to different coping mechanisms including but not limited to managing the condition at home, drawing limited social support mechanisms, seeking religious or spiritual guidance, and alternative care [
8,
10,
11,
17,
18]. While all three sub-scales measures barriers to care, the Internationalized stigma and Social abandonment sub-scales may be useful to further adapt in understanding care seeking for post-repair and reintegration services [
32].
Given the hard-to-reach nature and multiple vulnerabilities of and constraints faced by women living with fistula, the composite barrier measures can be used to inform future programming and resource allocation for fistula in a particular country context. As LMICs vary substantially in the way fistula surgeries, counselling, and reintegration services are structured, these composite measures can be applied at the population level to select optimal outreach, screening, and service delivery models. Quantifying barriers may help elucidate what is needed at the community level to promote access to care. The relative scores on sub-scales can influence targeting outreach, case identification, and referral. The indices and scales have implications beyond fistula strategies and programming; they may guide a better understanding of care-seeking for a wider range of reproductive and maternal morbidities as well as stigmatized conditions. For example, service design and delivery for pelvic organ prolapse and other health areas that require surgical intervention or continuous care (e.g. HIV, TB) could benefit by adapting the measures presented here [
33].
We recommend further research replicating these measures elsewhere as well as increasing sample sizes to assess sub-groups of women living with fistula by age and geography. As these indices and scales provide a lens into community-relevant barriers, we encourage the composite measures be tested further through implementation research and large-scale program and health systems monitoring to improve access for the most vulnerable groups. Despite some variation in the summary composite scores, in Nigeria and Uganda, we recommend targeting behavioral interventions toward communities that destigmatize fistula and promote inclusion (non-abandonment) of women living with and recovering from fistula repairs, alongside innovations to enhance awareness, finance and transportation barriers to fistula care.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.