Do quality improvement interventions for person-centered family planning work? Evidence from Kenya

Introduction

While there have been significant gains in contraceptive use in the past few decades, unmet need for family planning (FP) remains a significant challenge in Kenya, with 16.8% of women reporting an unmet need in 2017¹. Consequently, the Government of Kenya has been particularly interested in improving access to quality FP services, including a new urban program to integrate FP services into existing health services and working with health officials and community groups². Researchers have hypothesized that poor quality of FP services, including provider competence, interpersonal relationships, choice of methods, information given to clients, and appropriate constellation of services³, may be a barrier to broader contraceptive use, particularly among lower socioeconomic status women^4,5.

Person-centered and woman-centered models of FP have been proposed as important strategies to improve the quality of FP services³. These approaches for FP place the client at the center of care, working with broader health systems to ensure they receive dignified, respectful care, and that they are involved in all aspects of clinical decision-making^6,7. This approach has been shown to improve client satisfaction^6,8 and method continuation⁶.

A recent review of the limited person-centered family planning (PCFP) interventions that do exist found that only two involved quality improvement (QI) approaches⁹. Overall, interventions that targeted various forms of person-centered care (PCC) were generally successful at improving client perceptions of the quality of care (usually measured with a satisfaction question). Results were mixed for outcomes such as FP knowledge, uptake and continuation. With regard to the two QI interventions identified, the first focused solely on FP in Kenya and provided training for facilities on aspects of PCC as well as the facility environment¹⁰. This intervention found impacts at the supervisory and provider levels and in observations of client-provider interactions; however, it did not find an impact on client reports of person-centered outcomes such as satisfaction, privacy, being treated well, confidentiality, and cleanliness of the facility. The second, in Malawi, focused on many aspects of maternal and reproductive health, including FP, and mostly measured more clinically related outcomes. Generally, the QI intervention did not impact outcomes for FP, however, respondents in the intervention group were more likely post intervention to say that the provider “established a cordial relationship and identified her needs”¹¹.

QI methodologies have been implemented across various global healthcare settings to improve both processes and health outcomes. In 1995, the Institute for Health Improvement developed a QI framework called the “Breakthrough Series” (BTS) to support healthcare systems in making improvements while simultaneously reducing costs¹². A key aim of the BTS is collaborative learning, which is facilitated by bringing together multi-disciplinary QI teams to work on common challenges typically over 9–12 months. Within a BTS or Improvement Collaborative, QI teams agree on collaborative and facility aims for improvement, share their performance using common measures and then work individually during quarterly “action periods”” to secure improvements. After each action period, participants in the Collaborative meet together to discuss progress from the previous period, set aims for the upcoming action period and identify potential changes they could make to improve outcomes. Improvement Collaboratives use Plan-Study-Do-Act (PDSA) cycles to test and adapt change strategies aimed at improving the selected outcome¹³. The BTS has been used to successfully secure improvements in maternal and child health in low-and middle-income countries (LMIC) settings and has been demonstrated to be cost-effective in resource-constrained environments^14,15. We conducted an intervention using QI cycles to improve PCFP in three facilities in Nairobi and Kiambu Counties, Kenya, and explored the impact on PCFP related outcomes, compared to three comparison facilities., To our knowledge, this is the first application of the BTS to secure PCFP improvements in LMICs.

Methods

Results

Table 1 shows the demographics of respondents, broken down by those in intervention and control facilities. All in all there were 227 women in the control and 292 women in the intervention facilities at baseline, and 294 women in the control and 349 women in the intervention facilities at end line (no significance difference in sample size). Intervention and control facilities were significantly different in terms of age, with control facilities having more younger women. There were no significant differences between intervention and control participants in terms of marital status, work, parity, or education. Women in the intervention facilities were more likely to be of the dominant tribe (Kikuyu) than in control facilities. Most women (77.6%) adopted a long-term FP method, with no differences between intervention and controls. About a third (33.2%) of all women said they were very satisfied with their care, and most said that the provider had no preference, a slight preference, or a moderate preference about what method they adopted, again with no difference between control and intervention facilities (80.1%).

There was a significant improvement in PCFP scores in control facilities between baseline and endline, increasing from a mean of 41.70 to 43.20 (p=0.0245) (Table 2). There was no change in PCFP scores in intervention facilities over time.

Difference-in-difference models showed that there was no impact of the intervention on the total PCFP scores or the two sub-scales of the PCFP scale (Table 3). When looking at the specific item-focuses of the QI intervention (for example, facilities that specifically worked on the PCFP topic “provider asking if the woman had any questions” did not see a change in reports of that item), we found that there was a significant negative impact of the intervention on the “providers calling respondents by their name” item for the combined effects of the intervention and time, even though this indicator significantly improved both over time and in intervention compared to control facilities. The sub-domain for Health Facility Environment and the item for “provider introducing themselves” significantly increased between the two survey rounds as well.

Discussion

This evaluation found no impact of the intervention on women’s reports of the PCFP that they experienced. This held true for the full scale, sub-scales, and also individual items that each facility focused on in their QI work. Given the substantial body of evidence pointing toward Improvement Collaboratives as an effective intervention in low- and middle-income healthcare settings, the most likely interpretation of these results is that observed challenges hindered the QI intervention. Feedback we received from study facilities corroborates this interpretation.

First, the QI process itself may have been too cumbersome to be effective. QI teams set out to improve more topics for this project than is typical in an Improvement Collaborative. Alongside the four topics for FP, another 13 were selected for maternity care. This was necessary to try and detect an impact in the overall evaluations (the PCFP and person-centered maternity care scales). However, it dramatically increased the amount of time required to organize and analyse data in weekly QI review meetings. In addition, QI teams faced an extra burden of gathering their own performance data. QI work characteristically utilizes data that is already recorded in registers or collated for performance management purposes. Person-centered care performance is rarely monitored routinely in healthcare facilities, thus QI teams were asked to conduct their own exit interviews to assess on-going progress. This required a disproportionate amount of team effort to maintain suggested sample sizes to understand changes in QI team performance.

Delays due to two national strikes by doctors and nurses extended the overall project timeline from 9 to 21 months. This exacerbated the impact of staff turnover on QI team cohesion and continuity. The overall duration may have contributed to a decline in attendance at QI team meetings in the final three-months of the program and a sense of “improvement fatigue” reported by the expert QI facilitators.

Second, system constraints within the healthcare facilities may also have hindered the QI intervention. External evaluators noted stock-outs of FP methods reported by intervention and control facilities alike during the endline data collection period. This experience is likely to have influenced overall levels of satisfaction with the FP services, as some clients may have been asked to choose an alternative FP method or return when new supplies were available. This failure is likely to have influenced overall perceptions of PCFP in both intervention and control facilities equally, however combined with challenges noted above could have had additional impact in intervention facilities.

Disentangling potential contributors can help future interventions aimed to improve quality of FP services have better success, especially for person-centered measures or that use QI approaches. Future work to improve PCC in FP should ensure a streamlined QI process with a low burden on data collection and number of topics to improve.

Ethics approval and consent to participate

This study was approved by the Institutional Review Board at The University of California, San Francisco [# 15-18008] and the Kenya Medical Research Institute’s Scientific and Ethics Review Unit {# Non-KEMRI 526}. All subjects have provided written consent to participate in study activities under these approvals.

Data availability