From electronic data base searching, 350 articles related to unmet need for family planning and associated factors were retrieved. From those records, 60 studies were removed after reviewing their titles due to duplication. Then after reviewing titles and abstracts of the remaining articles, 271 articles excluded because of irrelevance. Fifteen articles were assessed for eligibility and considered for the final review after 4 articles [34-37] removed due to inaccessibility of full articles even after email request for the full document till three times (Figure-1).
Characteristics of included articles
A total of 17, 585 reproductive age women were included in this meta-analysis to estimate the pooled prevalence of unmet need for family planning and associated factors in Ethiopia. All studies included were cross sectional and community based except two articles done at institution level[15, 27]. The majority of the included studies were conducted in Amhara region [14, 16, 18, 21, 25, 27] and the largest sample size was recorded in SNNPR(5746) whereas the smallest in Amhara region(337). From those studies the lowest prevalence of unmet need for family planning was reported in Amhara region while the highest reported in SNNPR. All included articles were conducted from 2009-2018 except one study manipulated in 2005. In addition, almost all studies had a good response rate (≥97%) (Table-1).
Prevalence of unmet need for family planning in Ethiopia 2005-2018
From those included studies (15 articles), the estimated pooled prevalence of unmet need for family planning in Ethiopia was 31.45% (95%CI: 26.52, 36.39). The lowest prevalence was observed in Amhara region (17.4%, 95% CI: 11.80, 23.00) while the largest recorded in SNNPR (52.4%, 95%CI: 44.64, 60.16). The included studies showed substantial heterogeneity (I2 =88, P=0.000) indicated to compute random effect meta-analysis. So, to identify possible source of heterogeneity univariate meta regression model analysis using different factors like Region, study setting and study year was computed, but none of them were significant. However, publication bias was non-significant using Begg’s and Egger’s test (p=0.458 and p=0.258 respectively) (Figure-2).
Sub group analysis
Subgroup analysis was performed to compare the prevalence of unmet need for family planning using study year, sample size, region and study setting. The subgroup analysis by region showed the highest prevalence of unmet need for family planning in SNNPR, 39.5% (95%CI: 28.15, 50.85) and the lowest in others (include Oromo and AA), 27.40% (95%CI: 16.34, 38.47). The prevalence of unmet need for family planning before the year 2014 was 34.49% (95%CI: 24.92, 44.06) and it reduces to 29.03% (95%CI: 23.44, 34.62) after 2014. The prevalence of unmet need for family planning conducted on community based was 32.09% (95%CI: 24.50, 39.68) which was almost in line with studies conducted at institution based, 31.65% (95%CI: 19.94, 37.58). In relation to sample size the pooled prevalence of unmet need for family planning with sample size <614 was 30.5% (95%CI: 22.73, 38.28) whereas sample size 614 and above the polled prevalence was 32.6% (95%CI: 22.73, 42.47) (Table-2).
In order to identify factors associated with source of heterogeneity of the pooled prevalence of unmet need for family planning, meta regression was undertaken by considering both continuous and categorical data. Sample size, study year, study setting and region for each individual studies were considered in the meta-regression. But the meta regression showed that the pooled prevalence of unmet need for family planning was not associated with sample
Factors associated with unmet need for family planning
As shown below in figure-3, we tried to investigate factors associated with unmet need for family planning. Age of the women, age at first marriage, education level of both women and their partner, discussion with partner and occupation were factors assessed for association. But only age at first marriage, education level (both) and discussion were significantly associated with unmet need for family planning. Age at first marriage was reported in three articles [16, 18, 20]. Women with age at first marriage <18yrs were 2.3 times more likely to have unmet need for family planning than women marriage at 18yrs and above, OR=2.3 and (95% CI: 1.08, 4.87). Level of female education was another factor recorded in eight articles [14, 16-18, 20, 21, 24, 27]. The odds of unmet need for family planning was 1.9 times higher in illiterate women than literate women (read and write, primary and above educational level), OR= 0.9 and (95%CI: 1.19, 3.04). The association between male partner education and unmet need for family planning was reported in four articles [16, 18, 23]. The likely hood of unmet need for family planning in women with illiterate male partner was 1.78 times higher than women having literate male partner, OR=1.78 and (95%CI: 1.18, 2.68). The last but not the least factor, which was associated with unmet need for family planning was discussion, reported in six original articles[14, 16-18, 24, 25]. Women with no discussion with her partner about family planning were 3.52 times more likely to have unmet need for family planning when compared with women with pertinent discussion with partner, OR=3.52 and (95%CI: 2.56, 4.87). The other two factors (age less than 25yrs and occupation of women-being house wife) were not significantly associated with unmet need for family planning, OR=1.16 and (95%CI: 0.65, 2.08) and OR=2.07 and (95%CI: 0.78, 5.52) respectively (figure-3).