Study design and setting
This was an analytical cross-sectional household survey conducted between March and October 2018, in Wajir and Lamu counties. Two sub-counties were purposefully selected in each county, namely Wajir East and Wajir North in Wajir County, and Lamu West and Lamu East in Lamu County.
Study setting
Wajir county is ranked as one of the poorest in the country with as many as 76% of its residents having low education levels and only 4% have completed secondary level education or higher (23). In Lamu, nearly a third of the county’s residents live below the poverty line with only 13% having competed secondary education (23). Lamu County is a predominantly coastal community of the Swahili ethnic group while Wajir County is inhabited primarily by pastoralist Somalis.
Sample size and sampling procedure
Data collection involved administration of a semi structured questionnaire to female respondents aged 15–49 years in their households (Additional file 1). We calculated samples size using formulae shown below adopted from UNICEF’s Multiple Indicator Cluster Surveys (MICS) (24)
Where is n is the required sample size (number of households); 4 is a factor to achieve the 95 percent level of confidence; r is the predicted or estimated value of the indicator in target population; deff is the design effect; RR is the response rate; pb is the proportion of the target subpopulation in total population (upon which the indicator, r, is based); AveSize is the average household size (that is, average number of persons per household); RME is the relative margin of error to be tolerated at the 95% level of confidence; Currently, the target for the RME is generally is 12%. The total sample size for all domains was 350 in each county.
Respondents were selected using stratified, systematic random sampling. Each sub-county was stratified into constituent wards, sub locations and further into villages within each sub-location. From each village, we obtained a list of all households from the chief’s office. A systematic sampling (with a random start) was used to select the households for the study. All eligible household members aged 15-49 within the selected households were eligible for the study. In households where more than one woman of reproductive age 15-49 resides, a Kish grid was used to select one respondent from that household. We excluded those who did not provide consent, and women aged <18 (unless they were emancipated minors) or >49 years old.
Sampling Strategy
The study used multi-stage sampling in combination with stratified and systematic random sampling to selected the respondents. In the first stage, each sub-county was stratified into constituent wards, sub-locations and further into villages. The village comprised the primary sampling units (PSU). From each village, we obtained a list of all households from the chief’s office. In the second stage, households were selected from the household listing in each selected primary sampling units (PSU) using systematic sampling with a random start. All eligible household members aged 15-49 within the selected households were eligible for the study. However, only one woman of reproductive age was interviewed per household. In households where more than one woman of reproductive age 15-49 resides, a Kish grid[3] was used to select one respondent from that household. We excluded those who did not provide consent, and women aged <18 (unless they were emancipated minors) or >49 years old.
Data management and statistical analysis
Data were double entered and cleaned in epi infoTM 7 before statistical analyses. All analyses were conducted in STATA software version 15 (StataCorp. 2017). Descriptive statistics were used to summarise data whilst tables and figures were used for data presentation. Binary and multiple logistic regression analysis was used to identify the determinants of family planning use. Odds ratio with their corresponding 95% Confidence Intervals were used to determine magnitude and directions of the statistical associations in both bivariate and multivariable analyses. P-value was considered statistically significant at 5% level.
Outcome Variable
The study had one outcome variable which is current use of modern contraceptive method. They were coded as 1 “current use” and 0 otherwise. Current use was defined as women who were currently using any modern contraceptive method.
Explanatory variables
The explanatory variables were chosen based on expert opinion or prior knowledge. They included; the age of the participant, level of education, marital status, age at marriage, employment status, discussion with a spouse, ever given birth, family planning awareness, number of living children, acceptability in culture, FP allowed in Islam and willingness to obtain information on FP.
Data Analysis
A descriptive analysis was carried out to establish the level of current use, method preference and reasons for non-use after which frequency and 95% confidence intervals were reported. The binary logistic regression analysis was used to determine the determinants associated with family planning use. Crude Odds ratios (COR) and Adjusted Odds Ratio (AOR) with their corresponding 95% confidence intervals were used to determine the magnitude and directions of the statistical associations in both simple and multivariable analyses. A p-value of <0.05 was considered statistically significant.
Ethical considerations
Ethical approval for the study was obtained from the Research Ethics Committee of the Aga Khan University, Nairobi (2016/REC-56 (v3)). We also obtained a research permit from the National Commission for Science, Technology and Innovation (NACOSTI/P/18/14340/20946) to facilitate the conduct of research activities in the community. All participants provided written informed consent. Regarding minors below the age of 18 years only married adolescents, who are considered as emancipated, were included.
[3] The Kish grid is a method for selecting members within a household to be interviewed. It uses a pre-assigned table of random numbers to find the person to be interviewed.