Meta-Analysis: The Influence of Knowledge, Education, and Husband's Support on The Selection of Long Acting Contraception Methods

The population is expected to increase by 8,5 billion persons in 2030 and 10,9 billion persons in 2100, population is growing at a rate of around 1,1% per year. One of population growth control by birth control with contraception. The most effective contraception is long-acting contraception, however, on IDHS (Indonesian Health Demographic Survey) only 13% of currently married women use long-acting contraception. Many factors affect contraceptive use among married women, such as knowledge, education, and husband’s support. The purpose of this study was to determine the effect of knowledge, education, and husband's support on the selection of MKJP. This study used metaanalysis with the search engines by Google scholar, PUBMED, science direct, and ProQuest. The study was selected using PRISMA and it was evaluated by AMSTAR. Data synthesis was conducted by STATA 16.0. The results of this study obtained by knowledge [OR = 0,99; 95% Cl : 0,90-1,08, p = 0,000; I 2 = 74,8%], education [OR = 0,84; 95% Cl : 0,77-0,92, p = 0,000; I 2 = 86,3%], and husband’s [OR = 0,94; 95% Cl : 0,69-1,20, p = 0,000; I 2 = 81,5%]. This means that knowledge, education, and husband’s support have an impact on the use of long-acting contraception.


INTRODUCTION
The results of research by Setiasih, Widjarnako, and Istiarti (2016) state that knowledge influences the choice of MKJP, but the husband's support has no effect. 5 Meanwhile, the results of research by Chacko et. al (2016) stated that education has an effect on the use of MKJP, but the husband's support has no effect. 6 In the research of Demeke et. al (2020) stated that the husband's education and support influenced the use of MKJP with a sample of 460 respondents. 7 Meanwhile, research conducted by Bhandari et.al (2019) states that knowledge and education do not effect on the use of MKJP among modern women. This research was conducted on a sample of 9,875 respondents. 8 Based on some of these studies which have been reviewed, there are differences in the results of the research which lead to a research gap (gap). For this reason, a meta-analysis study was conducted to obtain strong conclusions to be used as a benchmark in making policies so that they could support the achievement of the Contraception Program.

METHOD
This study used a quantitative study design in the form of a meta-analysis that was conducted by searching on Google Scholar, PUBMED, Science Direct, and ProQuest. The research protocol uses Preferred Reporting Items for Systematic Reviews and Meta-Analyzes (PRISMA) to identify the number of literature and trace studies that meet the metaanalysis criteria. The PRISMA stages go through 4 (four) stages, including identification, screening, eligibility, and included. 9 The identification stage was carried out by searching for data using search keywords related to knowledge, education, and support of husbands for the choice of long-acting contraceptive methods with the type of study in the form of cross-sectional. Furthermore, screening was carried out to sort journals systematically and presented in tabular form based on exclusion criteria. Then, the assessment of the feasibility of the quality of the journal is carried out by using AMSTAR (Assessment of Multiple Systematic Reviews) which is based on the criteria of high quality (score 9-11), medium quality (score 5-8), and low quality (score 0-4). 10 Finally, sorting journals based on inclusion criteria, including journals related to research topics, the publication year 2015-2020, Indonesian and English, in the form of full text, having data for synthesis (p-value, OR, LCl, and UCl), and indexed by Sinta S1-S4 (national journal) and indexed by Scope through Scimago (Q1-Q4) for international journals. Data synthesis was performed using STATA 16.0.

RESULT AND DISCUSSION
Based on the search keywords, an overall study of the initial identification results of 4 (four) search engines was obtained as many as 4,018 journals. Furthermore, screening is carried out by deleting duplicate journals, journals that do not focus on the topic, journals with qualitative methods, journals in languages other than Indonesian and English, and journals that are only abstract and cannot be accessed. The screening results from 4,018 journals were obtained as many as 77 journals, then assessed the feasibility of the journals with AMSTAR so that as many as 20 journals were selected with high quality (score [9][10][11]. Of the 20 studies that fulfilled the synthesis, 11 journals discussed the influence The effect of knowledge on the choice of long-acting contraceptive methods For data quality the effect of knowledge on the choice of long-acting contraceptive methods with a sample of relevant journal search results (n = 11), there are 7 journals that state there is an effect of knowledge, and 4 journals that state there is no influence of knowledge. Then, the results of the metaanalysis synthesis in the 11 journals resulted in a chi-square heterogeneity value (p = 0,000 under p <0.05 and a value of I 2 = 74,80% greater than 50%, meaning that it has a heterogeneous distribution, so the Random Effect is used. Model. In the forest plot in Figure 2, there is a diamond shape (♦) which is the value of the effect size for each study with a certain confidence interval and a square shape that represents the weight of each journal. The narrower the diamond size indicates that the more accurate the conclusions are drawn and the wider the diamond size indicates that the conclusions drawn have a wide deviation. 31 The largest percentage of study weight in the Gayatri study (2020) was 44,03% with a sample of 4,307 respondents. The greater the percentage weight of the study indicates that the study has a large sample size and a high odds ratio with a narrow confidence interval range so that the study has good data quality.
Of the 11 studies, the value of p = 0.000 <0.05 and the combined effect size of OR on the effect of knowledge on the choice of longacting contraceptive methods were 0.99 with a wide confidence interval (95% Cl) lower limit of 0.90-1.08. This means that there is an effect of knowledge on the choice of long-acting contraceptive methods with an effect of 0.99 times compared to respondents who do not knpw about long-acting contraceptive methods. Apart from seeing the magnitude of the combined effect of the 11 journals, publication bias also needs to be seen from the results of the funnel plot. The funnel plot is plotted from the effect size on the x-axis and the sample size or variance on the y-axis. While the use of standard errors (not sample size or variance) on the y-axis has the advantage of spreading the points at the bottom of the scale and making it easier to identify asymmetries (an indicator that research is missing or unpublished). 32 In Figure 3, the distribution plot is asymmetrical. The funnel plot is depicted based on the effect size value and the standard error value. This shows that the 11 journals form a plot, most of which are at the top of the chart and are clustered around the Summary Effect (peak) because the sample size in the journal is quite large. 32 However, the distribution of plots is asymmetrical to the right of the triangle area which indicates that there is still publication bias.

The effect of education on the choice of long-acting contraceptive methods
For the quality of data on the effect of education on the choice of long-acting contraceptive methods, there are search results (n = 16) of relevant journals, there are 8 journals that state there is an effect of education, and 8 journals that state there is no effect of education. Then, the results of the synthesis of meta-analysis in the 16 journals resulted in a chi-square heterogeneity value (p = 0,000 under p <0,05 and a value of I 2 = 86,3% greater than 50%, meaning that it has a heterogeneous distribution, so the Random Effect is used. Model. Furthermore, in Figure 4 the resulting forest plot with the combined diamond size is quite narrow, which shows that the more accurate the resulting conclusions are. The largest percentage of study weight in the journal Kusumaningrum et al. (2020) 32,55% with a sample of 6,384 respondents. The greater the percentage weight of the study indicates that the study has a large sample size and a high odds ratio with a narrow confidence interval range so that the study has good data quality.
From these 16 journals, the value of p = 0,000 <0,05 and the combined effect size of OR on the influence of knowledge on the choice of long-acting contraceptive methods are 0,84 with a wide confidence interval (95% Cl), a lower limit of 0,77-0,92. This means that there is an effect of education on the choice of long-acting contraceptive methods with an effect of 0,844 times compared to respondents who have never received formal education.  Figure 5 illustrates a fairly even distribution of the funnel plot with plots that are mostly at the top of the graph and clustered around the Summary Effect (peak) because the sample size in the journal is quite large and symmetrical in the triangle area. This shows that the distribution of plots is evenly distributed at a low confidence interval (95% Cl) so that the publication bias that occurs is not significant.

The effect of husband's support on the choice of long-acting contraceptive methods
For the quality of data on the influence of husband's support on the choice of long-acting contraceptive methods, there are search results (n = 10) of relevant journals, there are 8 journals that state there is an effect of husband's support, and 2 journals that state there is no influence of husband's support. Then, the results of the synthesis of metaanalysis in the 10 journals resulted in a chisquare heterogeneity value (p = 0,000 under p <0,05) and a value of I 2 = 81,5% greater than 50%, meaning that it has a heterogeneous distribution, so random is used. Effect Model. Furthermore, in Figure 6, the forest plot with a diamond (♦) shape is quite narrow, which shows that the more accurate the conclusions are. The largest percentage of study weight in the study of Woldu et al. (2020) of 42.69% with a sample of 381 respondents. The greater the percentage weight of the study indicates that the study has a large sample size and a high odds ratio with a narrow confidence interval range so that the study has good data quality.
Of these 10 journals, the value of p = 0,000 < 0,05 and the combined effect size of OR on the influence of husband's support on the choice of long-acting contraceptive method are 0.94 with a wide confidence interval (95% Cl), a lower limit of 0,69-1,20. This means that there is an effect of husband's support on the  In Figure 7, the Funnel Plot Influence of Husband's Support on the Choice of Longacting Contraceptive Methods illustrates the distribution of funnel plots that are mostly at the top with a range between numbers 0 and 2. The plots are mostly at the top of the graph and are clustered around the Summary Effect (peak) because the sample size in the journal is quite large and symmetrical on the right and left of the triangle area. This shows that the research carried out is representative of the population so that the publication bias that occurs is quite small.
From the research conducted, it is concluded that the value of the largest combined effect size is in the knowledge of 0.99; husband's support of 0.94; and education 0.84. This shows that knowledge is the factor that has the greatest influence on the choice of long-acting contraceptive methods among the three factors. If it is related to the Family Planning Concept by Jane T. Bertrand (1995), it explains that the willingness to use the contraceptive method is influenced by sociodemographic factors which include marital status, age, number of children, parity, religion, knowledge of contraception, partner support, education level, and the amount of income. 33 Another concept based on Upadhyay (2001) explains that one of the factors that influence the decision to have contraception is a personal situation related to socio-demographics such as age, gender, education level, marital status, health status, divorce status, number of children, income, information media, life cycle, support and communication with partners, as well as perceptions and knowledge of contraception. 34 Both of these concepts are in line with Lawrence Green's theory (1980). According to Lawrence Green's theory (1980) which states that the factors that influence the formation of health behavior in society are divided into facilitating factors (knowledge, attitudes, employment, education, income, and culture), supporting factors (distance of access and costs), and driving factors (behavior or the role of health workers and husband or family support). 35 Knowledge is a form of human sensing or the result of a person's knowledge of objects obtained through their senses such as hearing (ears) and sight (eyes). Lack of knowledge about contraception is one of the main factors in not using family planning in populations with low contraceptive prevalence. A person's education level also affects his actions in finding solutions to existing problems. Someone with higher education will act more rationally so that it will be easier to accept new ideas. In addition, the decision to choose a long-acting contraceptive method is also influenced by the support of the husband. Changes in health behavior are intervened by support and communication with partners. 36 If it is related to policies regarding longacting contraceptive methods, such as the The policy contains guidelines related to efforts to equalize access to and quality of contraceptive family planning services in the long term through promotional activities, IEC, ensuring the availability of supporting facilities, increasing partnerships in services, and monitoring evaluation. This policy was made to serve as a reference in increasing participation in the use of long-acting contraceptive methods, however, the longacting outcomes of contraceptive methods are still low. For this reason, it is necessary to study further in future policy formulation by considering aspects of the theory of health behavior in society, especially those related to knowledge, education, and support of husbands.