Factors associated with use of long-acting reversible and permanent contraceptives among married women in rural Kenya: A community-based cross-sectional study in Kisii and Kilifi counties

Long-acting and permanent contraceptive methods (LAPM) are effective and economical methods for delaying or limiting pregnancies, however they are not widely used. The Kenya government is promoting the use of modern methods of family planning through various mechanisms. This study aimed to determine the prevalence and factors associated with the use of LAPM among married women of reproductive age in targeted rural sub-counties of Kilifi and Kisii counties, Kenya. Baseline and end line Data from a program implemented on improving Access to Quality Care and Extending and Strengthening Health Systems (AQCESS) in Kilifi and Kisii counties of Kenya were used. Multi-stage sampling was used to sample 1117 and 1873 women for the end line and baseline surveys, respectively. Descriptive analysis was used to explore the respondents’ characteristics and use of LAPM on a self-weighted samples. Univariable and multivariable binary logistic regression models using svy command were used to assess factors associated with the use of LAPM. A total of 762 and 531 women for the baseline and end line survey, respectively were included in this study. The prevalence of use of LAPM for baseline and end line survey were 21.5% (95% CI: 18.7–24.6%) and 23.2% (95% CI: 19.6%-27.0%), p-value = 0.485. The use of LAPM in Kisii and Kilifi counties was higher than the national average in both surveys. The multivariable analysis for the end line survey showed having 3–5 number of children ever born (aOR = 2.04; 95% CI: 1.24–3.36) and future fertility preference to have another child (aOR = 0.50; 95% CI: 0.26–0.96) were significantly associated with odds of LAPM use. The baseline showed that having at least secondary education (aOR = 1.93; 95%CI: 1.04–3.60), joint decision making about woman’s own health (aOR = 2.08; 95%CI: 1.36–3.17), and intention to have another child in future (aOR = 0.59; 95%CI: 0.40–0.89) were significantly associated with the use of LAPM. Future fertility preference to have another child was significantly associated with the use of LAPM in the two surveys. Continued health promotion and targeted media campaigns on the use of LAPM in rural areas with low socioeconomic status is needed in order to improve utilization of these methods. Programs involving men in decision making on partner’s health including family planning in the rural areas should be encouraged. The study used data from the baseline and end-line survey of the AQCESS project, conducted between August and September 2016 and January and February 2020, respectively in four rural targeted implementation sub-counties of Kisii and Kilifi counties. The AQCESS project aimed to contribute to the reduction of maternal and under-five mortalities in Kenya; its organization and implementation have been described in our previous papers [35, 36]. The project promoted the use of family planning through community sensitisation messages by the community health volunteers (CHVs) as one of the implementation activities, however, there was no family planning commodity provided by the project during the implementation period. Details of design and conduct of baseline survey is discussed in our previous paper [35]; the below sections described the design and conduct of the end line survey. The repeat cross-sectional survey was conducted in Kaloleni and Rabai sub-counties in Kilifi County and Bomachoge Borabu sub-county in Kisii County with a population of 304,778 and 129,617, respectively [36]. The maternal mortality rate in Kilifi was 448 deaths per 100,000 live births, and an under-five mortality rate of 87 deaths per 1,000 live births [36], with 70% of the population living below the poverty line. Under-five mortality rates in Kisii County was 36 deaths per 100,000 live births, however only 44% of its population living below the poverty line. Both counties have a high teenage pregnancy rate (Kisii 18.4% and Kilifi 21.8%) [36], which is higher than the national average of 18% [14, 36]. Kaloleni and Rabai sub-counties are served by 40 health facilities, while Bomachoge Borabu County is served by 12 health facilities [36]. About 82% and 93% of health facilities in Kilifi and Kisii offer family planning (FP) services; 49% and 93% of which offer services to adolescents, respectively. The CPR for the regions in 2014 were 34.1% and 66.1% for Kilifi and Kisii counties [20]. There are numerous structural challenges, including limited human healthcare resources, poor access to health services due to geographical and transportation barriers, and limited healthcare infrastructure, including a high physician and nurse to population ratio in these counties [36]. The survey employed a two-stage sampling procedure considering a village as a cluster: In the first stage of sampling, a selection of 30 villages in each of the two sites were selected based on probability proportional to the number of households in each village. The second stage involved a random selection of households from a master frame of the household listing in the selected villages. The household listings were provided by the village elders who were familiar with the village boundaries. No further sampling was carried at the selected household level. All assented and consented women aged 15–49 years old and permanent residents of the selected villages were included in the survey. The minimum sample size of 1788 households was calculated for the survey based on the following assumptions: an expected increase in the contraceptive prevalence rate between the baseline (2016) and end-line (2020) periods of 10%; 80% as the power; 95% level of significance; design effect of two; and a 20% non-response rate to account for absent household members during data collection. Figs ​Figs11 and ​and22 highlights the final study sample size for endline and baseline respectively after excluding women who were pregnant, unmarried, last menstruated more than six months before the survey, menopausal, never menstruated, did not know their last date of menstruation period, and those who preferred not to answer. Trained enumerators collected data using tablets pre-installed with a standardized electronic questionnaire programmed on the Open Data Kit (ODK) software in English and the local languages of the study areas (Giriama, Ekegusi, and Swahili). The questionnaire adapted questions from the Demographic and Health Surveys and Multiple Indicator Cluster Surveys to allow for comparison. The questionnaire was pretested in the nearby villages which were not part of the actual survey before the start of the actual survey. Each supervisor was assigned team of six enumerators to manage; the supervisor ensured that the sampled households were interviewed and repeat visits conducted to absent household members conducted. After three unsuccessful repeat visits, a household was considered to have absent members for the duration of the survey. The collected data were synchronized into a secure, password-protected cloud-based server, allowing for real-time data quality assurance. Other quality checks performed during data collection included random spot-checks, close supervision of the enumerators, routine data cleaning, and addressing all identified issues prior to the start of data collection on any given day. The main outcome variable was LAPM use, which was assessed based on two questions: (i) “Are you or your partner currently doing something or using any method to delay or avoid getting pregnant?” and (ii) “What are you currently doing to delay or avoid pregnancy?”. Based on responses to the question (ii), women were classified as LAPM user (intrauterine contraceptive devices, implants, and male and female sterilization) or otherwise a non-user of LAPM. The non-users of LAPM includes users of short-acting/traditional methods (emergency contraception, injectable, male condoms, oral contraceptive pills, and traditional FP methods) and non-users of any method of contraceptive. Specifically the main outcome was binary in nature defined as: Where Yi, is the response for the ith individual woman. Independent variables were identified based on a review of literature of previous studies on LAPM [19, 30, 31, 37–44] and included age measured in completed years at the time of the interview categorized into age groups (15–19, 20–29, 30–39, and 40–49 years); level of education (none, primary, and secondary and above), household wealth tercile (poor, middle, and rich), future fertility preference (no more/none and have ((an)other) child(ren), number of children ever born (CEB) as at the time of the survey (0–2, 3–5 and 6 or more), exposure to media was classified as yes for those who had listened to radio or watched television or read newspaper or accessed social media at least once a week and no for those who had not [23] and the main decision-maker on a woman’s health (woman alone, woman with a partner, and other). The other decision-makers included the husband/partner alone, father or father-in-law, mother or mother-in-law, other male family members, and other female family members. Household wealth terciles were generated based on a wealth score computed using the principal component analysis approach using household assets and materials for the dwelling floor, roof, and external wall [45]. Variables on exposure to media and number of children ever born were not collected during the baseline survey. Data were explored descriptively using the median and interquartile range (IQR) for continuous variables and frequencies and percentages for categorical variables. Chi-square test was used to compare the difference between the categorical groups. The multivariable logistic model was used to assess the factors associated with the use of LAPM while controlling for other variables. All the variable were considered to be important in explaining the outcome and were adjusted for in the multivariable logistic regression. The crude (cOR) and adjusted odds ratio (aOR) with the 95% confidence interval (CI) for the estimates is reported, and a p-value of less than 0.05 was considered statistically significant. We used “svy” set command in Stata to account for clustering due to complex sampling design of the study at village level. All analysis were conducted in Stata (15, StataCorp LLC, College Station, TX). The study was approved by the Aga Khan University Institutional Ethics Review Committee and research permit provided by the National Commission for Science, Technology, and Innovation. County approvals were obtained from the Ministries of Health in Kisii and Kilifi Counties. Written informed consent were obtained from each participant before the start of data collection, with those who could not write providing consent using their thumbprint. Women aged 15–17 years old provided assent and additional consent from their parents/guardians or whoever they were comfortable with consenting on their behalf. The survey was conducted in line with the Helsinki Declaration on Research involving Human Subjects.