Magnitude of unintended pregnancy and its determinants among childbearing age women in low and middle-income countries: evidence from 61 low and middle income countries

Background Unintended pregnancy is one of the most serious health issues in low and Middle-Income Countries (LMICs), posing significant health, economic, and psychosocial costs to individuals and communities. However, there is limited evidence on the prevalence of unintended pregnancies and their determinants in LMICs. Hence, this study aimed to assess the prevalence of unintended pregnancy and its associated factors among childbearing-age women in LMICs. Method Data for the study were drawn from a recent 61 Demographic and Health Surveys (DHS) conducted in LMICs. A total sample of 187,347 mothers who gave birth in the five years preceding the survey was included. STATA version 16 was used to clean and analyze the data. Multilevel multivariable logistic regression was employed to identify individual and community-level factors of unintended pregnancy in LMICs. In the multivariable analysis, an adjusted odds ratio with a 95% confidence level was reported to indicate statistical association. Results The pooled magnitude of unintended pregnancy in LMICs was 26.46%% (95% CI: 25.30%, 27.62%), ranging from 19.25%% in Egypt to 61.71% in Bolivia. Working status (AOR =  1.03; 95% CI: 1.01, 1.06), having a husband with no education (AOR = 1.07; 95% CI: 1.00, 1.15), and primary education (AOR =  1.05; 95% CI: 1.01, 1.11), women from male-headed households (AOR = 1.04; 95% CI: 1.00, 1.08), media exposure (AOR = 1.05; 95% CI: 1.02, 1.08), unmet need for contraception (AOR = 1.05; 95% CI: 1.02, 1.08), distance from a health facility (AOR = 1.03; 95% CI: 1.00, 1.06) were significantly associated with unintended pregnancy. Conclusion Unintended pregnancy rates remain high in LMICs. Women whose husband has no education and primary education, women with media exposure, working status, women who live in a household headed by male, women with unmet need for contraception, and women with a big problem of distance to health facilities were variables that were significant predictors of unintended pregnancy. When attempting to minimize unintended pregnancy in LMICs, these factors need to be considered. Furthermore, most of these attempts should be driven by government entities in low and middle-income countries.

Conclusion: Unintended pregnancy rates remain high in LMICs. Women whose husband has no education and primary education, women with media exposure, working status, women who live in a household headed by male, women with unmet need for contraception, and women with a big problem of distance to health facilities were variables that were significant predictors of unintended pregnancy. When attempting to minimize unintended pregnancy in LMICs, these factors need to be considered. Furthermore, most of these attempts should be driven by government entities in low and middle-income countries. KEYWORDS unintended pregnancy, multilevel analysis, low and middle-income countries, child bearing age, women Background Unintended pregnancies are pregnancies that are unwanted and/ or mistimed at the time of conception (1,2). Because of the consequences associated with both the mother's and child's social and health outcomes, unintended pregnancy is a major public health concern in both high-income countries and Middle-Income countries (LMICs) (3)(4)(5). It is estimated that 208 million pregnancies occur worldwide each year, with 46% of them becoming unintended (6). Annually, more than 14 million unintended pregnancies were reported in Sub-Saharan Africa (7). Although the global trend of unintended pregnancy has decreased over time, it remains high with significant regional variations (6,(8)(9)(10)(11).
Although it is a problem in both high-income countries and LMICs, unintended pregnancy has decreased in high-income countries compared to LMICs (7). In some areas, unintended pregnancy accounts for more than half of all pregnancies (12). In low-income countries, the rate of unintended pregnancy varies between 7.2 and 59.6 per 100 person-years of follow-up (11).
Tackling unintended pregnancy is one way to reduce the high rates of maternal and neonatal mortality (20, 24-26, 49, 50). To the best of our knowledge, no studies have been conducted to investigate the magnitude of unintended pregnancy in LMICs. Because the majority of unintended pregnancies occur in lowand middle-income countries, there is a critical need to explore the underlying causes of unintended pregnancies among women in these countries. Hence, the primary goals of this study were to determine the prevalence of unintended pregnancy and to assess the effects of potential underlying factors on unintended pregnancy among women aged 15-49 in LMICs.

Study population and data sources
This study used the Demographic Health Survey (DHS) data from 61 low and middle-income countries collected between 2008 and 2020. By measuring key indicators deemed important, a DHS survey allows countries to generate data that can be used to inform policy and practice. Each country's survey includes a variety of datasets such as men, women, children, birth, and household datasets. For this study, we used individual record (IR) data. To select study participants, the DHS employs a twostage stratified sampling technique.
This study only included women between the ages of 15 and 49 who had given birth within the previous five years of the survey. As a result, the total sample size was 187,347, with respondents from each country ranging from 607 in Guyana to 9,534 in Nigeria. A detailed description of the survey year, sample size, and sample characteristics is presented in Table 1.

Study variables and measurements
The outcome variable was unintended pregnancy, which is composed of both pregnancies that are wanted no more Aragaw et al. 10.3389/frph.2023.1113926 Frontiers in Reproductive Health or wanted later (mistimed). It was a binary variable, women with mistimed pregnancies or unwanted pregnancies were recorded as "unintended pregnancies", while those who needed pregnancy then were recorded as 'intended pregnancy' (13). The study included individual-level independent variables such as the age of women (15-19, 20-24, 25-29, 30-34, 35-39, 40-44, and 45-49 years), educational status of women (no education, primary, secondary, and higher), educational status of the husband (no education, primary, secondary, and higher), media exposure (yes or no), working status (working or not working), terminated pregnancy (yes or no), household wealth status (poorest, poorer, middle, richer, or richest), household members (≤5, 6-10 or >10), and sex of household head (male or female), intention to use contraceptive (yes or no), and unmet need for contraceptive (yes or no). Community-level factors such as place of residence (rural or urban) and distance to health facilities (big problem and not a big problem) were also included. Media exposure was created by asking women about the frequency of radio, television, and newspapers. It is classified as "yes" if women had at least one type of media exposure, such as radio, newspaper, or television, and "no" otherwise.

Data processing and analysis
We appended the data from 61 LMICs after extracting the variables based on existing literature. Before any statistical analysis, the data were weighted using sampling weight to restore the survey's representativeness. The data was cleaned and statistical analyses were carried out using STATA version 16. Frequencies and percentages were used to describe the background characteristics of the study participants. We conducted a multilevel logistic regression analysis, assuming that each community has a different intercept and fixed coefficient, with a random effect applied at the cluster level. Factors with a p-value ≤0.2 in crude odds ratio (COR) were selected as candidates for the adjusted model, finally, the adjusted odds Ratio (AOR) with 95% CI was reported, and variables with p-values less than 0.05 were declared to be significant predictors of unintended pregnancy in the multivariable analysis.

Parameter estimation method and model building
The fixed effects method was utilized to estimate the relationship between unintended pregnancy and independent variables, which was expressed as an odds ratio with a 95% confidence interval and a p-value of 0.05.The random effects, which are measures of variation of unintended pregnancy across communities or clusters, were expressed in terms of the Intra-Class Correlation (ICC), the median odds ratio (MOR), and the proportional change in variance (PCV). The ICC shows the differences between clusters in unintended pregnancy among reproductive-aged women. The ICC is calculated as ICC ¼ VA VAþ3:29 Â 100, Where; VA represents the area-level variance (51)(52)(53). The MOR indicates the central value of the odd ratio between the highest and the lowest risk regions when two clusters are chosen at random. The MOR is calculated as MOR ¼ e0:95 p VA, where VA donates the area level variance (54,55).
The PCV measures the proportion of total observed individual variation that can be attributed to differences between clusters. The PCV is calculated as; VnullÀVA Vnull Â 100, whereas; V null represents the variance of the initial model, while VA represents the variance of the model with more terms (54,55).
Four models were fitted to select the best-fit model for the data using deviance: the null model (model without independent variables), model I (model with individual-level variables), model II (models including community-level variables), and model III (model with both individual and community-level variables). Deviance information criteria (DIC) (−2 × log-likelihood value) were used to assess the goodness of fit. The Variance Inflation Factor (VIF) was used to test for multicollinearity among the selected independent variables.

Ethical considerations
The data set was obtained from the DHS website after a formal request and permission from the major DHS. All methods were performed following the Demographic and Health Surveys (DHS) program's relevant guidelines and regulations. The dataset was not allowed to be shared with other organizations and has remained confidential.

Background characteristics of study participants
Majority of the participants are in the age group of 25-29 [49,946(26.66%)]. Single women had a greater proportion of unintended pregnancies (27.75%), while the least proportion was recorded among married women (25.90%). Women who are from urban areas had a greater proportion of unintended pregnancies (26.57%) than women from rural areas (25.80%). Women who are from the poorest households have a higher proportion of unintended pregnancies (26.65%) compared to women from the richest household (25.80%) ( Table 2).

Factors associated with unintended pregnancy
Based on the final model (Model III, which includes both individual and community-level variables), occupation, husband's education, unmet need for contraceptive, and media exposure, and distance to health facilities were variables statistically associated with unintended pregnancy.
Women whose husbands have no education and primary education were 1.07 (AOR = 1.07; 95% CI: 1.00, 1.15), and 1.09 (AOR = 1.09; 95% CI: 1.02, 1.17) times higher odds of experiencing unintended pregnancies respectively than women who have higher education. The odds of unintended pregnancy in those who are currently working are 1.03 (AOR = 1.03; 95% CI: 1.00, 1.06) times higher than that of women who are not working. The risk of unintended pregnancy among women who live in a household with a male head was 1.04 (AOR = 1.04; 95% CI: 1.00, 1.08) times higher as compared to women who live in a household with a female head. Women who are exposed to media have 1.05 (AOR = 1.05; 95% CI: 1.02, 1.08) times higher odds of having an unintended pregnancy than those women who have not been exposed to media. The odds of experiencing unintended pregnancy among those women who had an unmet need for contraception were 1.05 (AOR = 1.05; 95% CI: 1.02, 1.08) times higher compared to those without an unmet need for contraception. Mothers who perceive distance from the health facility as a big problem had 1.03 (AOR = 1.03; 95% CI: 1.00, 1.06) times higher odds of having an unintended pregnancy compared to their counterparts ( Table 3).

Random effects model and model fitness
The random effects results are shown in Table 3. It was found that in the empty model (Model 0), there are substantial variations in unintended pregnancies among LMICs. The ICC in the null model showed that about 2.1% (0.021) of the total variance was attributable to the community where the women live. Model III, which includes both individual and communitylevel variables, was chosen due to its low deviance (116,427). Therefore, Model III, the complete model with both the selected individual and household/community factors, was the best.

Discussion
The pooled magnitude of unintended pregnancy in LMICs was 26.46%% (95% CI: 25.30%, 27.62%). The finding is higher than a study done in Six Asian countries (19.1%) (56) and lower than a study done in SSA (29.0%) (13). It is critical to recognize that, while unintended pregnancies are common in LMICS, some variations still exist across countries. The possible reason for this variation might be the difference in the health system of each country. According to our findings, unintended pregnancy rates in low-and middleincome countries range from 19.25% in Egypt to 61.71% in Bolivia.
In a multivariable multilevel logistic regression analysis, paternal education, working status, median exposure, household wealth index, residence, and distance to health facilities were significantly associated with unintended pregnancy in low and middle-income countries.
Women whose husbands have no education and primary education have higher odds of unintended pregnancy compared to women whose husbands have higher education (57,58). Partners who have no formal education or a lower level of education are less likely to encourage their wives to use modern contraceptives and a woman's pregnancy intentions and parenting decisions are influenced by her partner's attitude (59). Improving the male partner's educational status is critical Frontiers in Reproductive Health because the male partner has a strong influence on most household decisions, including the timing of pregnancy and the number of children desired (58). Women who have media exposure have higher odds of unintended pregnancy compared to women who have no media exposure. This finding is contradictory to a study done in Ethiopia (44,60), Nepal (61), and Pakistan (62). The possible justification might be even though media exposure creates awareness, that women having exposure to media may have an increased chance of social networks that may expose them to unintended pregnancy.
Women who are currently working have higher odds of unintended pregnancy compared to women who are not currently working. This finding is consistent with a study done in Ethiopia (63,64) This finding is contradictory to a study done in Ethiopia (65) and Cambodia (66). The possible reason might be women with occupations may have a high level of social interaction and the nature of their work, which may lead to casual sex followed by unwanted pregnancy (63).
The likelihood of unintended pregnancy was higher among women who lived in a household with a male head than among women who lived in a household with a female head. The finding is similar to previous findings (67). Women living in male-headed households may not have the opportunity to actively participate in family planning decisions, resulting in an unmet need for family planning and unintended pregnancy (68).
Women with unmet family planning needs were more likely to experience unintended pregnancy than those with met needs. The finding is consistent with previous studies (69)(70)(71)(72)(73). The possible justification for this could be that the unmet need for contraception has exposed women to the risks of unplanned pregnancy. Addressing the challenges and unmet need for contraception should be a priority for reducing unintended pregnancies in low-and middle-income countries.
Women with a big problem with distance to health facilities have higher odds of having unintended pregnancies than women who do have not a big problem. The finding is consistent with a study done in Ethiopia (74)(75)(76). The possible reason might be that women facing a big problem of distance to health facilities may have a problem accessing health care such as contraception (77). The pooled prevalence of unintended pregnancy among 61 LMICS.

Strengths and limitations of the study
The use of weighted, nationally representative large datasets of low and middle-income countries, with advanced statistical analysis techniques that account for the nature of DHS data for better parameter estimation was a strength of this study. However, due to the cross-sectional type of data, uncovering the causality between dependent and independent variables is challenging. The data was obtained through self-reports from women five years before the survey, which could be a cause of recall bias. Moreover, because it depends on the factors available in the DHS data set, the most significant explanatory factors such as medical-related factors, and the quality of maternal health services may be missed.

Conclusion
Unintended pregnancy rates remain high in LMICs. Women whose husband has no education and primary education, women with media exposure, working status, women who live in a household headed by a male, women with unmet need for contraception, and women with a big problem of distance to health facilities were variables that were significant predictors of unintended pregnancy. When attempting to minimize unintended pregnancy in LMICs, these factors need to be considered. Furthermore, most of these attempts should be driven by government entities in low and middle-income countries.

Data availability statement
Publicly available datasets were analyzed in this study. This data can be found here: www.dhsprogram.com.

Ethics statement
The data set was obtained from the DHS website after a formal request and permission from the major DHS. All methods were performed following the Demographic and Health Surveys (DHS) program's relevant guidelines and regulations. The dataset was not allowed to be shared with other organizations and has remained confidential. The patients/participants provided their written informed consent to participate in this study.

Author contributions
FMA originates the research concept. FMA, AZA, RET, TA, and BAT wrote the main manuscript text. All authors contributed to the article and approved the submitted version.