Despite global advances in access to family planning services, 214 million women of reproductive age in developing regions of the world still experience an unmet need for modern contraception ¹ . Contraceptive discontinuation is understood to be a driver of unmet need for family planning; in fact, an analysis of Demographic and Health Survey (DHS) data collected in 34 countries between 2005 and 2010 demonstrated that modern method discontinuation while in need—which occurs when a woman who wishes to avoid pregnancy stops using her modern method of contraception—accounted for over one third total estimated unmet need ² . When a person is able to continue use of their contraceptive method for the duration of their time in need, it may be an indication that the health system has met their need for family planning care. Furthermore, there is evidence to suggest that the quality of family planning (FP) services can impact continued contraceptive use ^{3– 6} . Thus, contraceptive discontinuation provides a key measurable outcome of interest in quality improvement strategies in FP service provision; however, identifying which aspects of structural and process quality are correlated with contraceptive method discontinuation can provide insights that are actionable for health systems implementers now—before poor quality service delivery manifests as unmet need.

The assessment of quality of care (QoC) in family planning programs has been largely guided by the Bruce family planning QoC framework for the past several decades. The framework articulates six fundamental domains of quality: choice of methods, information given to clients, technical competence of providers, interpersonal relations, mechanisms for follow-up and having an appropriate constellation of services ⁷ . A number of measurement tools exist which include indicators that seek to capture key elements of FP service quality; yet, questions remain as to the utility of existing tools for performance benchmarking and strategic decision making ⁸ . The broadly endorsed Bruce framework is applied inconsistently among researchers and programs seeking to understand and improve quality of care at the facility level, and limited guidance exists as to how to analyze resulting data once collected. A recent review of quality assessment tools for FP programs in LMICs identified 20 comprehensive tools for the assessment of clinical quality of care ⁹ . Some are well aligned with the Bruce framework. For example, the Quick Investigation of Quality (QIQ) was developed by MEASURE Evaluation in 2000 as a tool that would balance the feasibility of data collection with the reliability of the resulting data. The QIQ indicators align with 5 out of the 6 domains of FP service quality defined in the Bruce Framework: choice of methods, information provision, technical competence, interpersonal relations, and mechanisms for follow-up ¹⁰ . Meanwhile, large scale facility surveys, such as the Service Availability and Readiness Assessment (SARA) used by WHO for health facility assessment in LMICs, typically limit data collection to facility audits which assess infrastructure and readiness for choice only ¹¹ . Indeed, these structural aspects of quality are among the easiest and least expensive to measure.

When data to measure the domains of the Bruce framework are collected inconsistently, it creates an analytical challenge for programs that wish to summarize and compare findings related to the measurement of FP QoC. Some programs have tracked changes over time in certain indicators, while other studies have used various methods to create indices to measure all or some of the elements in the Bruce framework; in either case, indicators may be chosen from existing data collection tools based on the feasibility of data collection, or custom indicators may be developed anew. Where no standardized set of indicators exist, but instruments are designed to capture certain domains of FP QoC, data reduction techniques can be useful to enhance comparability and simplify analyses. For example, the authors of a 2014 DHS Analytical Study to assess the quality of care in FP, antenatal, and sick child services in several countries used principal components analysis (PCA) to create indices corresponding to structure, process, and client satisfaction ¹² . Elsewhere, factor analysis has been used to reduce indicators into variables representing different domains of quality corresponding to the Bruce framework ¹³ . Exploratory Factor Analysis (EFA) is particularly useful in situations where multiple latent variables are likely to be the source of variation in a set of indicators, such as is the case in the measure of QoC. EFA can help to determine how many latent variables underlie a set of indicators, and—like PCA—provide a means for data reduction in order to explore the relationship between QoC and outcomes of interest ¹⁴ .

Another measure used to assess FP service quality is the Method Information Index (MII). The MII aims to capture—from the client perspective—similar information to what would be measured in the Information Provision domain of the Bruce Framework. Specifically, the MII assesses FP counseling quality through three questions, asked of women in regard to the family planning visit where they received their contraceptive method: “Were you informed of potential side effects of the method?”, “Were you informed on what to do if you experienced side effects?”, and “Were you told about other methods of family planning apart from your current method?” The MII is one of 18 core indicators tracked by the Family Planning 2020 (FP2020) global partnership, which formed in 2012 following the London Summit on Family Planning to help reduce unmet need in the world’s poorest countries ¹⁵ . It has also been used in population-based surveys, such as the Performance Monitoring and Accountability 2020 (PMA2020) surveys and the DHS Women’s questionnaire, to report on the quality of FP counseling at the national-level. In these surveys, the MII is calculated as the percentage of women who respond ‘yes’ to all three questions, whether they have received a modern contraceptive or not ^{16, 17} .

The MII captures information provision during the counseling session and a woman’s understanding of having received that information. The simplicity and versatility of the MII and the fact that it can be measured in client exit interviews as well as in household surveys makes it an appealing choice for programs and clinics seeking to routinely approximate program performance. It can be aggregated and reported at multiple levels—from the clinic to national-level estimates—and it places the client perspective at the center of quality measurement. Recognition that the patient is the expert in the patient experience, or a foci of client-centeredness, has been an essential theme of three recent high level reports from U.S. National Academy of Medicine, the WHO, and the Lancet Commission on High Quality Health Systems ^{18– 21} . Furthermore, it is collected and analyzed consistently where tools applying the full Bruce framework are not. However, it captures only one aspect of the Bruce framework – “information given to clients” – and, as with more comprehensive measures, there is limited evidence to show how well MII predicts outcomes of interest, such as contraceptive discontinuation, when measured at the individual level or when applied programmatically.

Recent (2014) Demographic and Health Survey data from Kenya—the site of this study—indicates that more than half (58%) of women are using contraception, up from 45.5% in 2008–09, and knowledge of modern contraceptive methods among women and men age 15 to 49 years is nearly universal; yet, 18% of currently married women have an unmet need for family planning services. Of family planning use episodes, 31% were discontinued within 12 months. Eleven percent of those episodes ended with a switch to a new method, and 11% were discontinued with no switching due to side effect and health concerns (i.e., discontinuation while in need without switching) ²² . With regard to MII in Kenya, 60% of current users of modern contraceptive methods in 2014 reported that they were informed about potential side effects of their method, 52% were told what to do if they experienced side effects, and 79% were given information about other methods ²² .

The present study is a secondary analysis of data collected in select urban sites of Kenya as part of the Measurement, Learning & Evaluation (MLE) Project, implemented by the Carolina Population Center at UNC Chapel Hill. The MLE Project was the evaluation component of the Urban Reproductive Health Initiative (URHI), which operated in Kenya, Nigeria, India, and Senegal from 2010 to 2015. In Kenya, the URHI project, called Tupange, operated in five urban areas in Kenya: Nairobi, Mombasa, Kisumu, Machakos, and Kakamega. The data used come from baseline health facility surveys and household surveys of women that were conducted in 2010 to 2011 (baseline) and again in 2014 to 2015 (endline) in these five urban areas. Two methods for assessing QoC in FP are incorporated into the surveys: the MII, and a modified version of the Quick Investigation of Quality (QIQ). The QIQ has historically used client exit interviews, facility audits, and provider observation to assess a short list of quality of care indicators at the health facility level which align with 5 domains of the Bruce Framework, mentioned previously. Facility readiness is assessed through the facility audit, while the exit interview collects information related to clients’ experience of care at the health facility ¹⁰ . The MLE survey instruments incorporate provider surveys in place of provider observation to assess technical competence in clinical procedures and counseling skills. The MLE facility audit contains additional items to assess the sixth domain of the Bruce framework: constellation of services ²³ .

In this study, we demonstrate how existing metrics and data sources can be used to assess and improve the measurement of family planning QoC through an examination of the relationship between FP service quality and modern contraceptive discontinuation while-in-need using two established QoC measures.

In this study of women and facilities in urban Kenya, two measures of FP service quality were developed and assessed for their association with contraceptive discontinuation while in need without switching: constructed indicators from a QIQ-based facility quality assessment tool and the Method Information Index (MII). Women who reported receiving higher quality FP counseling according to their MII score were significantly less likely to discontinue their method over the next three years compared to women with an MII score of zero. In fact, a woman’s likelihood of discontinuation while in need without switching to a new method was cut in half whether she reported being informed of just one aspect of MII or informed of all three versus receiving no information at all.

The analysis of the comprehensive facility-based QoC assessment tool was more complex. Using EFA, we identified six domains of FP service quality captured by the assessment tool. Four of these domains – client satisfaction, readiness for choice & management support, information exchange, and technical competence—closely align with the domains that the QIQ indicators are intended to capture; however, two additional domains of quality emerged in this setting: privacy & comfort, and infrastructure & equipment. Items included in the assessment tool which may have captured mechanisms for follow-up and/or constellation of services were not sufficiently correlated to be identified as domains of quality in this analysis. Of the six that were identified, three domains of quality—privacy & comfort, technical competence, and information exchange—were found to be significantly associated with a decreased risk of contraceptive discontinuation.

Here, we also demonstrate how existing metrics and data can be used to assess and improve the measurement of quality of care related to family planning health outcomes, as others have done with MLE data from Kenya. Prior studies in Kenya and elsewhere have described the quality of family planning service delivery and have assessed the relationship between quality of care and current contraceptive use, but to our knowledge, few have assessed quality of care against contraceptive discontinuation while-in-need as an outcome of interest ^{13, 30– 32} . Furthermore, while many studies have examined aspects of client-centered counseling and how they relate to contraceptive discontinuation across a variety of contexts, few studies look specifically at the MII. In a prospective study of women in Pakistan and Uganda, Chakraborty et al. similarly found that baseline MII scores were positively associated with method continuation rates among clients in social franchises ³³ . By leveraging this existing dataset, which includes clients at both private and public sector facilities, we add depth to a growing body of work on the potential use of the MII as an indicator of contraceptive counseling quality.

Programs operating in urban Kenya that wish to reduce the rate of contraceptive discontinuation while in need among their clients might refer to these preliminary findings in their decision making. When women receive, or understand to have received, the information reported in the MII, they are less likely to discontinue their method without switching to another method. Thus, programs may choose to prioritize ensuring provision of full information on other methods and side effects in all contraceptive counseling that takes place at a facility and from this, could expect some reduction in each patient’s risk of discontinuing her method as a result. Additionally, efforts to improve aspects of privacy & comfort in FP service provision and to hire and train competent providers may strengthen rates of contraceptive continuation among clients. These results echo findings from other settings where current method use or discontinuation were the outcome of interest. For example, interventions to train providers in evidenced-based medicine or to improve provider competencies in areas such as contraceptive counseling and client-management have been shown to improve contraceptive continuation in a variety of settings ^{34– 36} . Similarly, a 2002 study conducted in Egypt found that interactions between providers and family planning clients that could be characterized as ‘client-centered’ versus ‘physician-centered’ were associated with higher rates of method continuation 7 months later ³⁷ .

As such, these findings may contribute to efforts to develop standardized, actionable measures for QoC in FP. The framework established by Bruce in 1990 ushered in a client focused approach to family planning services by establishing quality services as those in which clients are free to choose a method, are empowered to continue using the method, and are provided with avenues to seek help with or change their method. The framework achieved this paradigm shift by emphasizing choice, information provision, and mechanisms for follow-up as key tenets of quality, and establishing interpersonal relationships—vital for the facilitation of those tenets of quality—as a domain of quality in itself. However, the Bruce framework does not explicitly address issues of privacy and comfort, nor are these issues explicitly addressed in many of the existing studies linking client-centered or person-centered approaches to care to FP outcomes ^{37– 39} . For example, in the 2002 study in Egypt referenced above, ‘client-centered’ models of communication were identified as those with a high proportion of solidarity statements (versus disagreement statements), but issues of dignity, respect or privacy were not explicitly measured. In the present study, the domain most closely aligned with client-centered approaches to care contains explicit indicators of client comfort and privacy, including: ‘proportion of clients who reported they were comfortable asking questions’, and ‘proportion of clients who had visual privacy’, ‘proportion of clients who had auditory privacy’. The QIQ was based on the Bruce framework, and does not contain all of the aspects now considered important to client-centered care; yet, those measures that are present came together in EFA to form a domain independent of other domains of quality, even if it does not represent a comprehensive measure of client-centered or person-centered care. Given these findings, and an understanding of the rights of patients to dignified and respectful FP care, we agree with others who have suggested that more comprehensive and explicit measures of client-centeredness should be developed and incorporated into family planning quality measures ^{40, 41} .

In this study, structural aspects of quality, such as infrastructure, equipment and facility readiness, were not associated with contraceptive discontinuation. Overall, these domains were more homogenous across facilities, which may suggest that indicators pertaining to structural quality, while generally easier and less expensive to measure, are better suited for measuring against a minimum acceptable level of quality, and process measures are more appropriate for teasing out which aspects of a facility or program’s performance will be correlated with better outcomes.

The appeal of using the MII over more comprehensive measures of FP QoC for routine quality monitoring at the facility or program level is clear – it contains only three questions, it is easily assessed and analyzed, it can be aggregated for upward reporting and measurement, and it allows for benchmarking against national standards. Therefore, to better understand the utility of using MII at the facility-level, we examined the correlation between facility-level measures of MII and other facility-level measures of quality. MII assessed at the facility level was not correlated with all of the domains of quality that were associated with a decreased risk of discontinuation among women. Expectedly, facility-level MII was found to be correlated with the information exchange domain. Additionally, it was correlated with scores in the facility infrastructure domain and in the privacy & comfort domain. The client exit interviews (CEI) included in the SDP assessment in the MLE study are a good fit in the context of this comprehensive assessment tool, but were not designed to be representative of FP clients at any given facility, which informed our decision to exclude facility-level MII from the discontinuation analyses. Nonetheless, given these findings – that for a woman, her understanding of having received any MII information is associated with her being less likely to discontinue her method, and that when measured at the facility level, MII is correlated with several other domains of facility quality, more research is needed to determine how the measurement of MII can be refined at the facility level so that it can provide information to facilities and program managers that is actionable for improving outcomes.

Limitations of this study include the size of the analytical sample used (n = 1,033), which represents less than half of the 2,189 women interviewed at baseline and endline who were identified as current users of a modern reversible contraceptive method in 2010. The women who were dropped would have been eligible for inclusion in the analysis had it been possible to link them to the type of facility where they had received their method, to determine their method start date, and to identify their baseline episode of contraceptive use within the reproductive health calendar collected at endline. The final samples are much smaller than the original representative sample of women, and the generalizability of these results may be limited; however, the demographic characteristics of current FP users included in and excluded from the analysis were found to be comparable ( Table 1). We were unable to identify the exact facility where each woman received her FP method; thus, this study linked women to a measure of average facility quality based on their location and the type of facility where they received her method, assuming that women seek services within the city where they reside, which may be inaccurate in some cases. Additionally, quality measures aggregated at the facility type and city level may not be representative of a woman’s individual experience of care.

Several of the factors identified in EFA have Cronbach’s alpha values below 0.7, suggesting weak scales; however, refinement of these items into definitive sub-scales and scales was not the main goal of the paper. Finally, for some women, data on facility quality were collected up to a year after baseline data on current contraceptive use were collected. We chose not to exclude women whose reported baseline episodes may have begun before SDP assessments occurred, because we do not expect that quality at the SDP changed much during that time. Together these limitations may attenuate any existing associations between other domains of facility quality identified in our study and contraceptive discontinuation.

The reason for the association identified between the client satisfaction domain and an increased risk of discontinuation is unclear, but we do not interpret this to mean that that higher rates of client satisfaction will lead to higher rates of discontinuation among women. There was no single variable within the domain that could be identified as the main driver of this finding. We considered the possibility that some women may continue contraception, particularly long-acting methods, due to coercion or pressure from providers. A stratified analysis to examine the possibility of effect modification by provider-controlled versus woman-controlled methods did not explain the finding (Extended Data). The domain includes only client reported measures (of wait time, being treated well by staff and providers, and of experiencing good counseling skills during the visit). Although our factor analysis process grouped these items together, other frameworks include these measures in different conceptual areas. For example, Jain et al. would not consider wait time an indication of process quality, while Hutchinson et al. defined client satisfaction very broadly, encompassing aspects of our domains of client satisfaction, privacy and comfort, and information provision ^{40, 42} . Given that other studies have found conceptual and mathematical relationships between client reported measures and observed quality, this particular finding is not well explained within the present study.

Our study found a positive association between woman-reported MII and facility-level measures of information provision, technical competence, privacy, autonomy & dignity and contraceptive method continuation over 3 years. The findings suggest that family planning facilities and programs should emphasize information provision and client-centered approaches to care alongside technical competence in the provision of FP care. More work is needed to determine how the measurement of MII can be refined at the facility level as an actionable metric for improving outcomes. Furthermore, comprehensive and explicit measures of client-centeredness which incorporate aspects of privacy, autonomy & dignity should be emphasized and operationalized as standard measures of FP QoC are advanced.

The study protocol and tools were approved by the Institutional Review Board at the University of North Carolina at Chapel Hill and the Kenya Medical Research Institute Ethical Review Committee. All household-level participants and client exit interview participants provided verbal consent while providers provided written consent.