Menu
Menu
Menu
Menu
Background: Fifteen counties contribute 98.7% of the maternal and newborn morbidity and mortality in Kenya. The dismal maternal and newborn (MNH) outcomes in these settings are mostly attributable to limited access to skilled MNH services. Public health services are stretched and limited in reach, and many social programmes are not sustainably designed. We implemented a network of 16 self-sustaining community medical centres (Ubuntu-Afya Kiosks) in Homa Bay County, to facilitate access to MNH and other primary health services. We investigated the effect of these centres on MNH access indicators over a 2-year period of initial implementation. Methods: We conducted a baseline and end-line survey in June 2016 and May 2018 respectively, in 10 community health units (CHU) served by Ubuntu-Afya Kiosks. We targeted women of child bearing age, ensuring equal sample across the 10 CHUs. The surveys were powered to detect a 10% increase in the proportion of women who deliver under a skilled birth attendant from a perceived baseline of 55%. Background characteristics of the respondents were compared using Fisher’s exact test for the categorical data. STATA ‘svy’ commands were used to calculate confidence intervals for the proportions taking into account the clustering within CHU. Results: The coverage of antenatal care during previous pregnancy was 99% at end-line compared to 81% at baseline. Seventy one percent of mothers attended at least four antenatal care visits, compared to 64% at baseline. The proportion of women who delivered under a skilled birth attendant during previous pregnancy was higher at end-line (90%) compared to baseline (85%). There was an increase in the proportion of women who had their newborns examined within 2 day of delivery from 74 to 92% at end-line. A considerable proportion of the respondents visited private clinics at end-line (31%) compared to 3% at baseline. Conclusions: Ubuntu-Afya Kiosks were associated with enhanced access to MNH care, with significant improvements observed in newborn examination within 2 days after delivery. More women sought care from private clinics at end-line compared to baseline, indicating potential for private sector in supporting health service delivery gaps in under-served settings.
We evaluated whether the Ubuntu-Afya Kiosks had an effect on: the proportion of expectant mothers who attend at least four antenatal care (ANC) visits; the proportion of women who deliver under a skilled birth attendant; and the proportion of newborn children who undergo post-natal check within the first 2 days of delivery. We interrogated these questions using a non-randomized before-after study design. Sixteen Ubuntu-Afya Kiosks were set up in rural Homa Bay between March and December 2016, all operational to-date. In this evaluation, ten community health units (CHUs) served by 10 community medical centres were conveniently sampled from three sub-counties: Rachuonyo North, Kabondo Kasipul, and Mbita. Homa Bay has an approximate population of 1,177,181: 266,946 are women of child bearing age (15–49 Years) and 214,647 thousand are children under-5 [10]. The proportion of births assisted by a skilled provider, proportion of women having 4 or more ANC visits, and the proportion of births with a postnatal checkup in the first 2 days after birth were documented as 60, 59 and 41% by the 2014 KDHS survey report. The maternal mortality rate is 583/100,000 [11], and the infant mortality is 77/1000. As at 2016, 226 public and private health facilities were documented as duly registered [12], representing a population to health facility ratio of 2.1 per 10,000 [13]. For this evaluation, women of child bearing age (15–49 Years) with children aged 2 years or below or who had been expectant within the preceding 2 years were recruited. Women and children who had just moved into the catchment CHUs within the preceding 6 months were excluded from the study. Set up of the community medical centres was complete in December 2016. A baseline survey was conducted in June 2016. An end-line survey was conducted in May 2018. Recruitment of study subjects was done through sampling of consecutive households in CHU clusters. Interviewers visited households within the selected CHUs and administered a standardized questionnaire to the eligible participants (Additional file 1). The questionnaire was deployed using Open Data Kit (ODK) tool and included individual respondent characteristics and key questions on setting of last delivery; number of visits to a health facility; and examination of the newborn post-delivery in the previous pregnancy. Trained data collectors who understood the local language administered the standardized questionnaire with support from community health volunteers employed by the county under the community health strategy unit. The questionnaire was tested prior to the baseline assessment and revised to assure validity and reliability. Regular spot checks were conducted by the project team to verify authenticity of the data collected. Completed questionnaires were checked for missing information, inconsistencies, and wrong skip patterns which were rectified before submission for data entry. Key informant interviews were conducted at end-line to give context to the findings of the household surveys. The key informants were members of sub-county health management team, in particular: sub-county medical officers of health; District AIDS/STI coordinator; public health nurses; health promotion officers; and clinical officers. The study was designed to detect at least a 10% change in the proportion of women who deliver under a skilled birth attendant against a population prevalence of 55%, with a desired power of 80% and 5% level of significance. The estimated required number of respondents for each survey was 396 women. Evaluation of the effect of Ubuntu-Afya kiosks was based on three core outcome indicators of access to MNH services: (1) percent of women who attended four ANC visits during previous pregnancy; (2) percent of women who delivered under skilled birth attendant; and (3) percent of newborn children examined within 2 days after delivery. The study database excluded personal identifiers and was maintained in a central server with restricted access. We analysed the data using Intercooled STATA version 13.0 (StataCorp 4905 Lakeway Drive College Station, TX 77845 USA). Background characteristics (marital status, level of education, and source of income) were analysed using Fisher’s exact test to examine the hypothesis of independence between respondent characteristic at baseline and end-line. T-test was used to test the hypothesis of no difference in mean age of the respondents during baseline survey compared to end-line. Descriptive statistics were used to summarize proportion in antenatal care, skilled delivery, and postnatal care at baseline and at end-line. Based on the sampling design, individuals in a select CHU were considered likely to bear more similarity to each other compared to individuals in other CHUs. Clustering at the level of CHUs was therefore assumed and STATA ‘svy’ commands were used to calculate Confidence Intervals (CIs) for the proportions. Estimated proportions; p-values of the hypothesis of no difference between baseline and end-line; and respective 95% CIs where the CIs indicate the range within which there is 95% confidence where the true value for the estimate lies, are presented. The study was approved by the African Medical Research Foundation Ethical Review Committee (AMREF-P235–2016). Written informed consent was sought from study participants during both surveys, and parental consent and study participant’s assent were sought for participants below 16 years of age. Participants were made aware that their participation was of free will and that they could withdraw from the study should they change their minds about participation. Confidentiality was maintained during the survey and study data was anonymised and maintained in servers accessible only by the principal investigators.
