Search outcome
Of 4598 research papers initially identified using the search criteria, 4430 articles were found to be irrelevant after reviewing the titles. The abstracts of the remaining 168 papers with relevant articles were then examined. Of these, 130 abstracts were found to be irrelevant. Thus, full papers of the remaining 38 abstracts were accessed and 6 were found to be duplicates. Finally, 32 full papers were assessed for quality and eligibility. Thus, 22 unique citations were identified which met the inclusion criteria (Figure 1). The characteristics of the individual studies are summarized in Table 1.

Those studies which were conducted from 1990 to 2015 were retrieved for this review. Studies varied with regard to criteria for defining unintended pregnancy. Some of the studies had assessed the intention through binary outcome (intended or unintended) while others had assessed the intention under three categories (mistimed, wanted and unwanted). All studies were quantitative and among the selected studies, 8 analyzed secondary data from demographic health surveys or other national health surveys, while 14 were primary studies including 12 cross-sectional studies and 2 cohort studies.

Of the 22 studies, 18 had measured the prevalence of unintended pregnancy and 22 had measured the determinants of unintended pregnancy. Of these 22 studies, 11 were from Africa, 9 from Asia, 1 each from Europe and North America. All the determinants were categorized into three themes i.e. Socio-demographic factors, Affordability and women's knowledge, attitudes, beliefs and culture.

Prevalence of Unintended pregnancy
Of the 18 studies which had measured the prevalence of unintended pregnancy, it was found that average prevalence was estimated to be 35% ranging from 13% to 82%.

Factors determining the unintended pregnancy
Socio-demographic factors
Socio-demographic factors included women's age, women's education, parity, birth order and interval, previous pregnancy intention, age at the time of marriage, socioeconomic status, marital status, religion, caste, and ethnicity. Thirteen studies found that women's age was the best predictor of pregnancy intention (Table 1). Of these, 8 studies showed that older women were more likely to experience unintended pregnancy (15-21) while five studies had shown that younger women were at risk of unintended pregnancy (17, 21-24). Seven studies found that women's education was the best predictor of pregnancy intention (Table 1). Of these, 5 studies showed that women with primary or less education were more likely to experience unintended pregnancy (17, 19, 25-27) while two studies had shown contradictory findings(20, 26).

Ten studies found strong associations between parity and pregnancy intention (Table 1). Among all the studies, higher parity was generally found to be the positive predictor of unintended pregnancy. In addition to this, four studies found strong associations between age at the time of marriage and pregnancy intention (Table 1). All the studies had shown the inverse relation between age at the time of marriage and pregnancy intention, as the age at the time of marriage increased, the likelihood of unintended pregnancy reduced (20, 28-30) . Likewise, five studies showed that religion played a significant role for the pregnancy intention. Around 60% of these studies found that non-Muslims were less likely to experience unintended pregnancy as compared to Muslims while 40% had shown the opposite findings(17, 31).

Seven studies found strong associations between marital status and pregnancy intention (Table 1). Of these, four studies found that married women were less likely to experience unintended pregnancy as compared to unmarried women. Three studies found that married women were more at risk of experiencing unintended pregnancy (Table 1).

Affordability
Eight studies found significant relationships between economic factors (socio-economic status or income of the household, working status of woman and woman's employment) and pregnancy intention (20, 28-30). The financial constraint was the most important factor determining the intention of pregnancy. Generally, women with high economic status were less likely to experience unintended pregnancy except the one study, which contradicted the above findings.

Women's knowledge, attitudes, beliefs, and culture
Different variables were included under this category like access to media, knowledge and practice of Family planning methods. Only one study from Nepal showed that exposure to mass media (radio) significantly predicted pregnancy intention (Table 1). Women with high levels of exposure were less likely to experience unintended pregnancy(28).

Four studies had found the strong association between pregnancy intention and knowledge of Family planning methods (Table 1). Of these 4 studies, 50% had shown that knowledge was positively associated with pregnancy intention and 50% of these had shown the negative association between two variables(28, 32). Additionally, eight studies had shown that contraceptive usage had also shown significant association with unintended pregnancy, 6 studies found the positive association with unintended pregnancy while 2 studies found the negative association.

Click here for Table 1: Prevalence and determinants of unintended pregnancy

Click here for Table 2: Quality assessment* of individual studies included in the systematic review

Figure 1: Flow diagram showing the process of retrieving the articles

FLOW CHART

This review has evaluated all the peer-reviewed studies, published in English language from 1990-2015. It was found that the prevalence of unintended pregnancy ranged from 13-82%. Common predictors identified were age, parity, socioeconomic status, religion and use of family planning methods. Age was found to be positively associated with unintended pregnancy in most of the studies. Studies from Bangladesh, Ethiopia, Egypt, Nepal, Spain, Vietnam and Nigeria reported the positive association of age with an unintended pregnancy (16, 21, 23, 27, 28, 33, 34). Consistent with definitions of mistimed and unwanted pregnancies, this review showed that mistimed pregnancies occurred to younger women and were seen more often in the first and second pregnancies, while unwanted pregnancies occurred to older women and in the third or later pregnancies. Furthermore, this analysis showed that many mistimed and unwanted pregnancies occurred either as a result of no contraceptive use, or due to method failure. Although we could not explore the types of methods used by the women or reasons for method failure, the increased likelihood of unintended pregnancy among users of contraceptive methods could be due to inappropriate usage of short term methods, which has been highlighted through prior research(35).

Studies from Kenya, Ethiopia, Ghana, Tanzania and Bangladesh reported that high parity was significantly associated with unintended pregnancy (22, 23, 31-33). These findings can be explained by the fact that couples from rural areas prefer to have more sons and they may end up having more children. Furthermore, women in the Asian region usually prefer to have enough sons who can become their source of income for the family(34). Moreover, these couples want to balance the sex of their children, and will continue to give birth if all the children are of the same sex.

In this review, it was found that wealth quintile and educational status was associated with unintended pregnancy in the same direction. Studies from Ethiopia, Japan, Spain, Bangladesh and Nigeria reported that illiterate women or women with primary education are at risk of unintended pregnancy (17, 23, 26, 27, 33). This finding can be explained by the fact that women may not even realize that they are pregnant until it has become too late, and they may not be able to negotiate with their spouses with regards to safe sex.

Knowledge about contraceptives was reported to be negatively associated with unintended pregnancy(17, 28). Various studies found that women who have more knowledge about contraceptives are less likely to experience an unintended pregnancy, as compared to those who do not have adequate knowledge about these methods. Moreover, the use of contraceptive methods is positively associated with unintended pregnancy. Studies have also found a strong positive association between users of modern contraceptives and unintended pregnancy, which can be explained by the fact that users of these methods might fail. Since this review did not focus on method failure, therefore, it can be assumed that method failure might have increased the chances of unintended pregnancy among these women.

With respect to women's autonomy, men are usually considered as the main decision-makers in developing countries, and they decide when and where a woman should seek healthcare. Hence, women are often given less power in the male-dominant societies to decide for themselves, and they have to depend on the male partners/relatives for their survival and other life matters. Besides, social norms limit women's freedom to make important decisions(36). In some regions of South Asia, women have substantially lower social status and autonomy than men(37). It has been found that low social status and autonomy seems to be associated with low fertility control(38). Such women are more likely to experience an unintended pregnancy, as compared to those who have some autonomy(36).

Strengths and Limitations of the Study
We believe that this is first systematic review on prevalence and predictors of unintended pregnancy, which has synthesized the data both from developing and developed countries. Despite the extensive literature review, there are some limitations to this study. Firstly, we had reviewed all the cross-sectional studies except one cohort study, therefore causal association between various determinants and unintended pregnancy cannot be determined to establish a relationship between unintended pregnancy and various studied correlates. Secondly, different countries have different predictors for unintended pregnancy, depending upon the epidemiologic and demographic variations in those particular countries, so results might not be generalized to all settings. Thirdly, this review included papers only in the English language and we might have missed the important information published in language other than English. The decision to exclude non-English language studies was made for practical reasons based on the increased time, expense and complexity of translating and synthesizing these studies. However, much research in developing countries may not be published in peer-reviewed journals, but might be available as gray literature in local languages. These limitations should have a minor impact on the scope of the study in view of the large size of the sample and that the main variables of interest are demographic factors associated with unintended pregnancy.

This review demonstrates that unintended pregnancy is common both in developing and developed countries and age, parity, educational and economic status can be considered as important determinants of the unintended pregnancy. This means undertaking outreach in poor and inaccessible settings and providing community-based distribution of family planning methods including counseling and referral for women with unmet need. In addition to this, programs can be undertaken to target youth through youth corners in existing health facilities or training providers in offering youth friendly services. Implementation of targeted programs will guarantee access to family planning for all categories of women in need. These types of targeted approaches can help women to meet their fertility desires and reduce unintended pregnancies with the overall objective of reducing maternal mortality and morbidity.