Factors associated with delays in seeking post abortion care among women in Kenya

Background: Delays in seeking quality post abortion care services remain a major contributor to high levels of mortality and morbidity among women who experience unsafe abortion. However, little is known about the causes of and factors associated with delays in seeking care among women who suffer complications of unsafe abortion. This study looks at factors that are associated with delays in seeking post-abortion care among women in Kenya. Methods: Data for this study were from a nationally representative sample of 350 healthcare facilities that participated in the 2012 Incidence and Magnitude of Unsafe Abortion study in Kenya. Data included socio-demographic characteristics, reproductive health and clinical histories from all women treated with PAC during a one-month data collection period. Results: Delay in seeking care was associated with women’s age, education level, contraceptive history, fertility intentions and referral status. Conclusions: There is need to improve women’s access to quality sexual and reproductive health information and services, contraception and abortion care. Improving current PAC services at lower level facilities will also minimize delays resulting from long referral processes.

This study is descriptive, analysing cross-sectional data obtained from women seeking PAC in healthcare facilities over a 30-day period in sampled facilities. This paper uses data from the 2012 National Incidence and Magnitude of Unsafe Abortion Study led by the African Population and Health Research Center (APHRC). The study used data from a nationally representative sample of 350 level II-level VI facilities, out of all 2838 PAC-providing facilities in Kenya in 2012. The Kenyan Ministry of Health categorizes the health system into six levels which provide preventive and curative public and private health services as follows; Community health services (Level I); Primary care facilities (Level II and III) comprising of dispensaries, clinics, health centres and maternity homes; county referral health facilities (Levels IV and V) comprising of district/county hospitals, sub-district/county and provincial hospitals; and VI (national referral health facilities comprising of national hospitals) [17]. The study sampled all facilities from level II to Level VI, but excluded dispensaries, which are less likely to offer any PAC services due to staffing and equipment. The primary outcome variable, “delay in seeking care” was based on a set of responses to questions that sought to establish 1) the time it took a woman from the onset of complication (e.g. when first bleeding was spotted) to know that she was experiencing a problem, 2) The time to decide to seek care 3) the time between making the decision to seek care and arrival at health facility. We computed the delay to seek care variable used in this study as the total time of these three different durations as reported by the patient in hours. In addition, we measured the following socio-demographic characteristics: age, level of education, type of residence (rural/urban), and occupation. For all clients including referrals, delay was computed from onset of complications to arrival at health facility at which the patient was observed in this study. For patients who were referred out of the facility of observation, the duration of delay only ends at current facility of observation while patients who were referred to the current facility, the time spent in the referring facility is also captured in the total delay. As of 31st of January 2012, MoH provided a list of 2838 facilities in levels II to VI with a potential to provide PAC. All level V (17 facilities) and VI (two facilities) facilities were included in the study as well as all (thirty-seven) non-governmental facilities that provide post abortion care or known to provide comprehensive abortion care services. However, for level II-IV facilities, we drew a representative sample using varying sampling fractions at each facility level and region. Therefore, we stratified the total sample according to five geographic regions and five facility levels. We generated these five regions by merging some provinces, which are similar with respect to geographic neighborliness, proximity to shared major healthcare facilities and some level of similarity in selected health-related indicators such as maternal mortality ratios, contraceptive prevalence and total fertility rates. These regions were a) Nairobi and Central b) Nyanza and Western c) Coast and North Eastern d) Rift Valley and e) Eastern provinces. In total 350 facilities were sampled, and a national response rate of 90 % was achieved during data collection. The original survey sample was determined in order to have 80 % power to detect at regional level 10 % difference in the proportion of women with severe complications from unsafe abortion as significant, using a two-sided 5 % significance level. Trained facility-based healthcare providers who offer PAC services at the sampled facilities collected data from all patients presenting at each of the 326 facilities out of the 350 over a 30-day period. Of the remaining twenty-four facilities, twenty-two did not provide data due to low monthly caseloads while we excluded data from two facilities due to logistical challenges. This gave an average national response rate of 93 % spread according to regions as follows; Nairobi and Central (92 %), Nyanza and Western (99 %), Coast and North Eastern (94 %), Rift Valley (89 %) and Eastern provinces (99 %). The providers collected patients’ socio-demographic characteristics, reproductive and clinical histories, diagnosis, treatment and clinical procedures performed, post abortion contraception provision, and clinical management outcomes. We collected all data using paper forms, and later captured into computers using CSPro® and then exported to STATA® 12.1 for consistency checks and analysis. These analyses consisted of descriptive and inferential statistical analysis to describe some of the demand-side characteristics associated with delayed care seeking among women presenting for PAC. The analyses presented in this paper focuses on women who sought PAC. Estimates presented in this paper were weighted using sampling weights generated from the probability of a woman being interviewed in the survey. The sampling fractions were a product of the probability of a facility being selected and accepting to participate and that of a woman participating in the survey based on overall interview response rate at the regional level. To adjust our estimate’s standard errors for design effect due to the complex sampling design above, we generated all statistics presented in this article within STATA’s “svy” platform using the facility level as the primary sampling unit. We summarized delay into median time to care by woman characteristics. For this analysis, given the right-skewedness (positive) of the data (Skeweness = 7.04), we transformed the outcome variable into its natural logarithm, yielding a more symmetric outcome variable (Skeweness = −0.0152). To study factors associated with delayed care seeking, we fitted a random-effects model assuming uniform correlations and estimated the intra-cluster correlation. Past studies have categorized abortion complications into three levels based on clinical signs and symptoms. These classification categories as used in this study were adopted from two main surveys, one in South Africa [18] and another adopted in a study in Kenya [5]. The classifications are outlined in Table 1 below. Classification of severity categories of abortion complications Cases were categorized into the extreme category of abortion complications, and required only one sign or symptom to be counted in that category. (Adopted from Jewkes, Fawcus et al. [28] and Jewkes, Gumede et al. [29] The study received ethical clearance from the Ethical Review Boards of the Kenya Medical Research Institution (KEMRI), the University of Nairobi/Kenyatta National Hospital, Moi University Teaching and Referral Hospital, and Aga Khan University. The Ministries of Public Health and sanitation and the Medical Services in Kenya and the Institutional Review Board of the Guttmacher Institute also reviewed and approved the study. For ethical considerations, verbal consents were obtained from all women presenting for PAC. Deidentification of records was done before analysis to ensure that data collected on a woman, provider or facility could not be traced back to the source.

N/A

• Dima Health