Determinants of Long-Acting Reversible Contraceptive Method Utilization among Married Women in Assosa Town in Western Ethiopia: A Case-Control Study

Introduction

Long-acting reversible contraceptive (LARC) methods are highly effective in the prevention of pregnancy; moreover, they are suitable and cost-effective for users over time. These methods facilitate substantial cost savings for the government and contribute directly to achieve a targeted high-level goal as compared to the short-acting techniques (1). Birth control implants (e.g., Implanon and Jadelle) and intrauterine contraceptive devices (IUCD) are among the contraceptive methods that are long-acting and reversible. The mentioned methods are effective up to 3-5 years and 10-12 years for the prevention of unwanted pregnancy, respectively. By far, they have been the most effective (>99%) contraceptive methods available. As a result, global and national family planning programs encourage giving due attention to LARC methods for women without compromising the right choice (1).

According to the World Health Organization (WHO) report, 287,000 maternal deaths occurred worldwide in 2014 during which developing countries accounted for 99% of such deaths. In the same vein, Sub-Saharan Africa and South Asia accounted for 87% of global maternal deaths. The high share of Sub-Saharan Africa and Asia in maternal death statistics is mainly due to the low utilization of modern contraceptives, especially the long-acting methods (2). Similarly, in a report by Bongart in 2012, it was shown that more than 200 million women in developing world wanted to avoid pregnancy but not using modern contraceptive methods, which could result in 54 million unintended pregnancies, more than 79,000 maternal deaths, and one million neonatal deaths each year (3).

Ethiopia is one of the Sub-Saharan African countries with the highest maternal mortality ratio. This country had 676 maternal deaths per 100,000 live births in 2011 and 412 maternal deaths per 100,000 live births in 2015. At least 35% of these deaths are due to unwanted pregnancies, as well as unsafely induced abortion and childbirth, which could be avoided by using appropriate contraceptive methods, particularly LARCs (4, 5). In Ethiopia, 5% of the currently married women use birth control implants as a long-acting method, 4.8% of which occurs in Benishangul Gumuz Region. Evidence also indicates that the majority of married women had information regarding long-acting reversible contraceptive methods; however, they did not use them practically (4, 5).

The majority of maternal and neonatal deaths can be prevented by means of proven interventions. Accordingly, the current policy of Ethiopia addresses transition from short-acting to long-acting methods; nonetheless, the utilization of these methods is very low. With this background in mind, the present study was conducted to identify the determinants of the low use of LARC methods. It is hoped that the findings of this study will be used by decision makers in Benishangul Gumuz Region in particular and Ethiopia in general in designing appropriate interventions to enhance the utilization of LARCs.

Materials and Methods

This study was conducted in Assosa town, Benishangul Gumuz Regional state, from April 26 to May 10, 2016, using an unmatched case-control design. Benishangul Gumuz is one of the nine regional states of the Federal Democratic Republic of Ethiopia. This region is located in the western part of Ethiopia, 662 km away from Addis Ababa. It is one of the developing regional states, where most of the health care coverage, including LARC use, is at a very low state.  

Assosa town is the administrative town of Benishangul Gumuz regional state, which has four administrative ‘kebeles’ (i.e., the smallest administrative unit) (6). The study population of the present study corresponded to a group of married women, living for a minimum of six months prior to the study period. The participants consisted of a case group entailing the married women of reproductive age using one of the LARC methods (e.g., Implanon, Jadelle, or IUCD) during the study period and a control group, including non-users (either using short-term contraceptives or not at all) of the LARC methods during the study period. The pregnant women reported as infertile were excluded from the study.

In this study, LARC methods were defined as those techniques that prevent pregnancy for a minimum of 3 years per application (e.g., Implanon, Jadelle, or IUCD). Knowledge of women was categorized as high/good (≥80%), moderate (60-79%), and low/poor (≤60%) based on 10 questions. The participants’ attitudes were measured by calculating the mean score of 14 attitude-related items rated on a 5-point Likert scale (from strongly disagree=1 to strongly agree=5) categorized as good and poor attitudes.

The wealth index was assessed by computing principal component analysis (PCA) to produce composite variables. Based on this index, the respondents were categorized into five wealth categories (quintiles), where the 1^st and 5^th quintiles represented the poorest and richest states, respectively. Based on the decision-making power questions, the women were classified as those who decided by themselves or discussed it with their partner and those who could not decide on at least one question.

The sample size for this study was determined using Epi-info software (version 7.4.1) for the population proportion formula based on the following assumptions. Among the determinants considered, the partners’ joint discussion and decision on contraceptive use were found to result in the largest sample size. Based on this, the rate of women having joint discussion and decision on contraceptive use among non-LARC users (i.e., controls) were assumed to be 53.7% with an odds ratio of 2.00, taken from a previous study (7).

With 95% confidence level, 80% power, and case to control ratio of 1:2, the minimum sample size required was estimated at 348 (i.e., 116 cases and 232 controls). Considering 10% non-response rate, the final sample size was 384 (i.e., 128 cases and 256 controls). The study population was selected using the simple random sampling technique via computer-generated randomization out of 166 cases (i.e., users) and 3,025 controls (i.e., non-users), identified by presurvey enumeration.

In this study, the major outcome variable was LARC method utilization, whereas sociodemographic characteristics, individual factors, and reproductive health-related factors were considered as predictor variables.

Data collection was performed using structured interviews via a questionnaire prepared based on the related literature (8-13). The questionnaire was prepared in English language, and then translated to the local language (i.e., Amharic). To ensure consistency, the questionnaire was back-translated to English. The data were collected by eight nurses with a diploma degree who were supervised by two bachelor’s degree holders.

Quality of data was controlled by designing appropriate data collection materials, pretest, training of data collectors and supervisors, and careful data entry. Reliability of the attitude questionnaire and wealth index was confirmed by achieving the Cronbach’s alpha coefficients of 0.74 and 0.78, respectively. The validity of the interview questionnaire was determined using the content validity.

The data were coded and entered into Epi-data software (version 3.1) and exported into SPSS software (version 20.0) for analysis. Descriptive statistics (e.g., frequency, percentage, mean, and standard deviation) and cross tabulations were computed and presented in tables and figures to show a general picture of the data. The wealth index was computed as a composite indicator of living standard based on the variables related to the ownership of the selected household’s assets, presence of livestock, and materials used in the house. The computation was made using PCA.

The variables were generated by summing up the principal components into three components. The adequacy of the model for PCA was checked by KMS and estimated at 0.78. Three components were explained in the wealth index with an overall cumulative percentage of 68%. Quintiles of the wealth index were generated using the composite score. The Chi-square test was calculated to check whether the assumption was met or not. Statistical tests were calculated at 95% confidence interval.

Variables with a p-value less than 0.25 in bivariate analysis were selected as candidate variables for the multivariate logistic regression model. Model fitness for multivariate analysis was checked using the Hosmer-Lemeshow test (X²⁼6.5, df=8, P=0.59), and individual covariates were assessed by Wald test. Multicollinearity
(at a variance inflation factor of >10) and interaction between the independent variables were checked. Multivariate binary logistic regression analyses were computed to identify the predictor variables. Adjusted odds ratios, together with 95% confidence intervals, were determined to measure the strength and level of significance of the association.

Ethical clearance was obtained from the Ethical Review Committee of the College of Health Sciences of Jimma University in Jimma, Ethiopia. Letter of cooperation and support from the university and the ethical approval letter were presented to the Benishangul Gumuz Regional Health Bureau and Assosa Town Administration. Confidentiality terms were assured using identification numbers in the questionnaire. In addition, the participants’ privacy was guaranteed by removing names and other identifiers during the interview.

Results

A total of 382 (i.e., 128 cases and 254 controls) participants were included in the study with a response rate of 99.5%. The mean age of the participants was 27.3±5.2 years (28.4±5.6 for cases and 26.8±4.9 for controls; age range: 18-45 years). Most of the cases and controls were in the age range of 25-34 years. Regarding ethnicity, most of the women were from Amhara region (n=163, 42.67%; 41.4% and 43.3% of the cases and controls, respectively), followed by Oromia (n=125, 32.7%; 37.5% and 30.3% of the cases and controls, respectively) and Shinasha (n=30, 7.85%; 5.4 % and 9.1% of the cases and controls, respectively).

In terms of religion, nearly half of the respondents were Orthodox (n=187, 48.7%), followed by Muslim (n=94, 24.6%). With regard to the educational status, 46 (35.9%) and 75 (29.5%) women among the case and control groups had secondary education (9-12^th grade), respectively. Considering the wealth index of the respondents, the majority of the participants in the case group were in the poorest (n=34; 26.56%) and rich (n=28; 21.87%) states. Furthermore, most of the subjects in the control group were in the third quintile (n=58, 22.83%), followed by the second quintile (n=54; 21.3%). The Chi-square test was calculated to check the fulfillment of the assumption and compare ratios between the two groups (Table 1).

As measured by the composite indicator of the 10 “Yes/No” knowledge questions, the majority of the participants in the case group (n=74, 57.8%) and about one-fourth of the individuals in the control group (n=63, 24.8%) had a good level of knowledge on LARC. In addition, 75% (n=96) 53.1% (n=135) of the case and control groups knew that LARCs are available as means either to limit or space pregnancy, respectively. In addition, the Urban Health Extension Workers were identified as the main source of information on LARC in
both case (n=101, 78.9%) and control (178, n=70.1%) groups.

Based on the composite indicator score of 14 Likert scale questions, the mean attitude score

of the respondents was estimated at 49.43±7.8. Furthermore, 78.9% (n=101) of the cases had a positive attitude towards LARC use, while this rate was 36.6% (n=93) in the control group. About 1 and 2 per 10 participants in the case (n=14, 10.9%) and control (n=46, 18.1%) groups had respectively a wrong perception that there is a restriction at health facilities regarding the method use. Similarly, 12 (9.4%) cases and 63 (24.8%) controls had the perception that LARC is not available at health facilities.

The mean ages of the participants upon the first marriage and first pregnancy were 19.5±3.7 and 21.1±3.1 years, respectively. The majority of the cases (n=118, 89.8%) and controls (n=198, 77.9 %) had at least one pregnancy before. The Chi-square test was performed to check the fulfillment of the assumption and compare ratios between the two groups (Table 2).

After adjusting for the potential confounders, the spousal occupation, wealth status, and women’s decision-making power, knowledge, attitude, distance from the related facility, expectation for method availability, and intention to give birth in the future were identified as the independent predictors of LARC use in the study area.

The spouses working as day laborers were more likely to utilize LARC than those working as civil servants (OR=3.37, 95% CI: 1.02-11.22). In addition, the women in the 5^th (OR=0.11, 95% CI: 0.03-0.39) and 4^th (OR=0.2, 95% CI: 0.06-0.68) quintiles, corresponding a better wealth status, were less likely to use LARC methods than those in the lowest quintile (i.e., 1^st quintile). The results also revealed that the women who were able to make decision by themselves were more likely to use LARC methods as compared to those who were not able to decide by themselves (OR=3.42, 95% CI: 1.25-9.34).

Furthermore, the women who had medium (OR=0.22, 95% CI: 0.089-0.53) or poor levels of knowledge about LARC methods (OR=0.05, 95% CI: 0.015-0.17) had a lower probability of utilizing such methods than those having good knowledge. Regarding the attitude, the subjects with a good (or positive) attitude toward LARC methods (OR=7.58, 95% CI: 3.42-16.84) were more likely to use these methods in comparison to those with a poor (or negative) attitude toward this issue.

The women intended to give birth in the future (OR=0.39, 95% CI: 0.18-0.81) had a lower probability of utilizing LARC methods than those who had no future intention for pregnancy. In addition, the subjects living at a walking distance of 30 min or more from a health facility were less likely to use these contraceptive methods than those living at a walking distance of less than 30 min (OR=0.36, 95% CI: 0.14-0.81). Furthermore, the women who perceived that LARC is available at health facilities were more probable to utilize such methods, compared to those having a perception regarding the unavailability of these methods at the related facilities (OR=4.29, 95% CI: 1.12-10.49). During the bivariate analysis of LARC utilization, 19 variables having a p-value less than 0.25 were selected for the final multivariate logistic regression (Table 3).

Discussion