Quality of Family Planning Care as Enabling Factor for Modern Contraceptive Use in Indonesia: Evidence from PMA2020 Survey

The utilization of modern family planning methods is a key pointer that mirrors universal access to reproductive health cares. Fullling unmet needs for modern contraceptives may decrease the level of maternal mortality by almost a third. This research employed the 2016 PMA2020 survey to assess multilevel variables related to modern family planning method use among women of reproductive age in Indonesia. Data from interviews of female and Service Delivery Point (SDP) of the 2016 Performance Monitoring and Accountability 2020 (PMA2020) survey were linked to generate a merged dataset comprising women’s individual, cluster and SDP factors (N=10,210). Multilevel analysis was done to assess the inuence of contextual factors including a summary index of SDP quality FP care on woman’s modern family planning practice. Adjusted odds ratios and 95% CIs were analysed and interpreted.


Abstract Background
The utilization of modern family planning methods is a key pointer that mirrors universal access to reproductive health cares. Ful lling unmet needs for modern contraceptives may decrease the level of maternal mortality by almost a third. This research employed the 2016 PMA2020 survey to assess multilevel variables related to modern family planning method use among women of reproductive age in Indonesia.

Method
Data from interviews of female and Service Delivery Point (SDP) of the 2016 Performance Monitoring and Accountability 2020 (PMA2020) survey were linked to generate a merged dataset comprising women's individual, cluster and SDP factors (N=10,210). Multilevel analysis was done to assess the in uence of contextual factors including a summary index of SDP quality FP care on woman's modern family planning practice. Adjusted odds ratios and 95% CIs were analysed and interpreted.

Results
Modern contraceptive prevalence among women of childbearing age was 42.3% in 2016. There was a considerable variation in the likelihood of modern method use across the 372 clusters/EAs. Age, education, marital status, parity, residency, region, woman's FP decision autonomy, cluster's average ideal number of children, percentage of females dissatis ed with FP, and percent of females visited by CHVs were signi cant factors for modern FP use after adjusting for multilevel characteristics. Quality of FP care was revealed to be a signi cant enabling factor for modern contraceptive use in Indonesia.

Conclusions
Modern contraceptive prevalence among females aged 15-49 years was relatively low in 2016. The ndings indicate that an advancement in modern method use can be gained by improving quality of family planning care. Likewise, an enhancement in women's FP decision autonomy and their level of education, decline in women's ideal number of children and diminution of the dissatisfaction with family planning are all major concerns to be addressed. Family planning policies must tackle adverse cultural norms and inequities in quality of family planning care and women's education that would produce welfares to women, children, and communities.

Background
Modern contraceptive utilization is a key pointer that mirrors universal access to reproductive health cares (1)(2)(3). Furthermore, the contraceptive prevalence rate (CPR) among women of reproductive ages is indicator 3.7.1 in the 3rd Sustainable Development Goals (SDGs) (4, 5). Raising modern family planning (FP) use is essential for accomplishing three of the SDG indicators, i.e. decreasing maternal deaths, enhancing child health, and lessening HIV/AIDS (6-9). Indonesia has thrived major success in FP programs in last half centuries (10, 11). Nonetheless, modern contraceptive use in Indonesia has been an issue in last decade (12)(13)(14). Building and sustaining approaches of advancing the prevalent utilization of modern methods in Indonesia is a tough task for FP program planners and managers (12)(13)(14).
Furthermore, more than 10% of women of reproductive age in Indonesia have unmet needs for family planning (FP), mainly for modern contraceptives (15). Ful lling unmet needs for modern contraceptives may decrease the level of maternal mortality up to one third (16, 17). Woman's characteristics including highest attended education and decisional autonomy are related to modern method usage (18)(19)(20)(21)(22)(23)(24). It is propositioned that some social contexts, including the cultural norms at the society wherein women live, could in uence women's decision-making process in modern FP utilization as well (25,26).
Even though drivers and barriers of modern FP practice in Indonesia have been examined in prior research, to authors' knowledge, no prior research have utilised nation-wide representative data and multilevel analysis to assess the net in uence of social and family planning service factors on modern method utilization. This research employed nation-wide representative datasets from the 2016 Performance Monitoring and Accountability 2020 (PMA2020) Indonesia survey to assess the net effect of cluster and service delivery point (SDP) characteristics on women's likelihood to practice modern family planning methods. Social contexts were analysed at the cluster/community level, as were done in prior research (27,28).

Data
Data used in this study were obtained from the 2016 PMA2020 Indonesia survey: This project was performed in 2016 in paralel with other international survey project of PMA2020 in other 10 countries, i.e. Ghana,Kenya,Uganda,Democratic Republic of Kongo,Ethiopia,Burkina Faso,Niger,Nigeria,Côte D'Ivoire,and India (29). The 2016 PMA2020 Indonesia survey consists of women in reproductive age data from a national sample included 12,107 households, 10,565 females and 995 service delivery points (SDPs) in 372 clusters or enumeration areas (EAs) (30).
In this study, authors extracted family planning data from the female and SDP datasets across all 372 randomly chosen enumeration areas/clusters (N=10,210 women). Authors then generated cluster or enumeration area (EA) data by agregating women's individual data for the purpose of multilevel analysis.
Moreover, records from female and nearest SDP were merged using closest Global Positioning System points and 'EA ID' as the key variable to link both female and SDP datasets.

Variable Measurements
Outcome Modern method usage was de ned based on female records whether she or her spouse was currently using a modern contraception, such as injectables, pills, implants, intra-uterine device (IUD), male or female sterilization, or condoms (female or male) at the time of interview. This dichotomous outcome variable was coded as 1 for "currently using a modern contraception" and 0 for "currently not using a modern contraception".

Level 1 explanatory variables
Independent factors added into level 1 of multilevel analysis comprised women's age ( ve years interval age groups), educational background (no school or primary school, junior or senior high school, and college/university), marital status (never married, married or living with spouse, and widow/divorced/ separated), number of living children (0 child, 1 child, 2 children, 3 children, and 4 or more children), socioeconomic status (lowest, low, middle, high, or highest), health insurance (no health insurance, non-free national health insurance, free national or district health insurance, and other health insurance), and family planning decision-making autonomy (low and high).

Level 2 explanatory variables
Cluster/EAcharacteristics Cluster/EA characteristics inserted into level 2 of the multilevel analysis consisted residency (urban or rural) and region (Sumatera, Java and Bali, Nusa Tenggara, Kalimantan, Sulawesi, and Maluku and Papua). Other cluster/EA characteristics was generated by agregating individual data of women living in the same cluster/EA. Cluster/EA factors were developed for two aspects, i.e. 'cultural norms' and 'demand generating effort', that were commonly considered in literatures as part of the "social determinants" of modern family planning practice.
The 'cultural norms' aspect was constructed by 2 variables: 1. Percentage of cluster (EA) members who dissatis ed with FP.
A woman was classi ed as having dissatisfaction with FP if she cited her reason for not using a modern contraceptive was because of opposed by personal/spouse/religious view, or health concerns or fear of side effects.
2. Average ideal number of children of women in childbearing age who were living in the same cluster/EA. Furthermore, the 'demand generating effort' aspect was as follows: The percentage of cluster/EA members visited by village health worker or community health volunteer (CHV) who discussed FP in the past one year. SDPcharacteristics SDP characteristics included in the level 2 of multilevel analysis were SDP classi cation (clinic or health center, delivery post or health post, private midwife or doctor, hospital, and others) and SDP quality of FP care (QoC). The variable of SDP quality FP care (QoC) was a synthesized index obtained from categorical PCA (principal component analysis). This composite variable was built using SPSS 20 by synthesizing fourteen quality components of family planning care as follows: 1. Available trained staff(s): whether trained staff(s) was available when SDP served contraceptive care to clients.
2. Days SDP offering contraceptive services: number of days in a week SDP provide contraceptive care.
3. Available access to water and electricity: whether SDP staff(s) had available access to water and electricity.
4. FP service charge: whether SDP charged clients for FP service.
5. Method choice: how many modern contraceptives offered by SDP.
. FP method stock-out: Did SDP experience FP method stock-out in past 3 months.
7. FP kits and tools: number of available FP kits and tools in the family planning examination room.
. FP Examination room: were contraception kept from rodents, water, and sun, and was there any dirts in FP examination room.
9. SDP supervision: has SDP been supervised in past 6 months.
10. SDP supporting CHVs: did SDP train or supply contraceptives to community health volunteers (CHVs).
11. Visual and auditory privacy: did SDP maintain auditory and visual privacy in FP examination room.
13. Family planning service provision for adolescents: did SDP counsel//prescribe/offer contraceptives for adolescents.
14. Post-abortion care: did SDP provide post-abortion care.
Moreover, factor weights (composite quality scores) were then normalized and each SDP was attached with a quality index. This composite index of quality FP care was classi ed by its tertiles, denoting low, medium, or high quality family planning care.

Statistical analysis
This research performed multilevel analysis using STATA 14 by two-level mixed effect logistic regression model which was able to: Assessed the independent in uences of cluster/EA and SDP factors on women's odds of using modern methods, and Addressed the cluster effects due to multistage cluster sampling design of PMA2020 survey.
This research studied a binary dependent variable of modern FP practice. Authors performed these model building approaches to gain an unbias mixed effect model which predicts the odds of women of reproductive age in Indonesia to utilise modern FP supplies: First phase: A 'null model' without any independent variables was analysed to verify whether there was a signi cant variation across clusters/EAs regarding the odds of modern method usage. This empty model con rmed whether there was any signi cant random effect on multilevel model, by cheking whether the variance of multilevel model intercept was signi cant or not.
Second phase: A full or comprehensive model with all explanatory variables (level 1 and level 2) was then built to assess the independent in uence of cluster/EA and SDP factors, and to investigate whether there was any signi cant cross-level interaction between explanatory variables at both level.

Results
Modern contraceptive prevalence Figure 1 and 2 display the modern contraceptive prevalence in 2016. The percentage of women of reproductive age using modern FP supplies in Indonesia was 42.33% and around 57.7% were contraceptive non-users and traditional method users ( Figure 1). Figure 2 depicts that among modern FP users, around 53.33% were injectable method users, 20.99% were contraceptive pill users, 9.32% were using implants, 7.77% were using IUD, 6.15% were female sterilization users, 2.31% were condom users, and about 0.12% females reported that their spouse had been sterilized.
Female individual characteristics in 2016 PMA2020 Indonesia survey Table 1 exhibits the characteristics of women of reproductive age in this study. Approximately 14.47% were aged 15-19 years, and more than one-quarter were aged over 40 years. Around 57.32% had attended high school and only 14.59% had attended academy/university. The percentage of females had 2 children was 26.75% and about 28.39% had no parity. Regarding marital status, roughly 71.87% of women of reproductive age were married or living with a spouse. Nearly 59.01% of women had a higher decisional autonomy in family planning and about 25.38% had a high socio-economic status, but more than 39% of women had no health insurance (see Table 1).
Clusters/EAs characteristics in 2016 PMA2020 Indonesia survey  Table 2).   Post-abortion care (0.964), access to water and electricity (0.838), and number of FP methods offered by SDP (0.792) were also greatly loaded. Oppositely, FP service charge, number of days SDP served FP care, and FP service provision for adolescent were the lowest loading factors (-0.247, -0.170, and -0.085 in respective order). The quality scores then were normalized as shown in Figure 3. The multilevel analysis results presented in Table 4 signi es which individual, cluster/EA, and SDP factors had substantial in uences on the odds of modern method utilization. Of the fourteen level 1 and level 2 variables, eleven had indicated a statistical relationship with modern family planning practice, i.e. female's age, high school education, number of living children, marital status, FP decision-making autonomy, free health insurance, residency and region, community average ideal number of children, proportion of cluster members dissatis ed with family planning, proportion of cluster members visited by CHV, and SDP quality of FP care (see Table 4). Nevertheless, there were 2 variables which had no signi cant relation to modern contraceptive use, i.e. socio-economic status and SDP classi cation.  * α = 0.10; ** α = 0.05; *** α = 0.01.

Discussion
The research ndings revealed that the odds of women of childbearing age using modern contraception greatly varied across the 372 cluster/EAs. This study identi ed substantial relationship between quality of family planning care and modern contraceptive use. Particularly, SDP quality of care had a positive association with women's family planning practice, females lived closest to an SDP providing high quality of care had higher odds of using modern methods. This pattern are consistent with study ndings from other research (31-33).
Our ndings also found that the cluster average ideal number of children and percentage of cluster members dissatis ed with FP in uenced the likelihood of individual woman to use a modern FP method.
More or less, it indicates that social beliefs may impede the demand for family planning. It also re ects that cultural and traditional norms play an important role in modern contraceptive prevalence. Prior research suggested that prevailing community values may in uence woman's decision in family planning practice at the individual level (25-28).
It was exciting to acknowledge that women living in rural areas had a higher likelihood in modern contraceptive use. It may be rationalised by the national FP programs in Indonesia targeting rural women (34). Therefore, it can be viewed as an achievement of national government intervention initiatives implemented at rural areas.

STRENGTHS AND LIMITATIONS
This study used the latest PMA2020 Indonesia survey datasets which were gathered between October 2016 and January 2017, thus the research ndings are still relevant to be considered for current policy analysis. Moreover, the measurement and variables employed in this study are equivalent and compatible with those commonly used in global research. In addition, the multilevel analysis let us to identify the net effect for multidimensional variable of quality family planning care, and to earn better estimates by addressing cluster effects from multistage cluster sampling design.
Though, this study has several limitations. Firstly, the 2016 PMA2020 Indonesia survey data are crosssectional, hence, study results should be interpreted with caution due to inability to infer a causal-impact relation. Also, analysing secondary data constricted authors to include unavailable factors that may in uence modern method use, such as client's satisfaction which is part of Donabedian quality framework (35). It is possible that female respondents were using or not using modern contraception due to other variables not available in the datasets. Thus, it may affect the signi cance of variance for the intercept of multilevel model, adjusted for other covariates in the study.

Conclusions
In summary, the modern contraceptive prevalence in Indonesia was relatively low in 2016. The odds of women of reproductive age using modern family planning methods statistically differed across 372 Page 17/22 clusters/EAs. The ndings indicate that an advancement in modern method use can be gained by improving quality of family planning care. Likewise, an enhancement in women's FP decision autonomy and their level of education, decline in women's ideal number of children and diminution of the dissatisfaction with family planning are all major concerns to be addressed. Family planning policies must tackle adverse cultural norms and inequities in quality of FP care and women's education that would produce welfares to women, children, and communities. Declarations revision of the manuscript. All authors read and approved the nal draft.

Funding
The research was funded by the Indonesian National Population and Family Planning Board (NPFPB).
However, NPFPB played no role in this research.

Availability of data and materials
To access the data used in this study, administrative permissions were required and obtained from The Center of Reproductive Health, University of Gajah Mada: O ce: C24 Mahoni Road, Jogjakarta, Indonesia, Postcode 55281 Email: pskespro.fkkmk@ugm.ac.id Ethics approval and consent to participate As the study entailed secondary analysis of existing data containing no personal identifying information, an exemption from ethical review requirements was granted by the the Institutional Ethical Review Board, Mahidol University, Thailand. Consent to participate was not an issue for the same reason.

Consent for publiation
Not applicable.

Competing interest
The authors have no con ict of interest to conduct this research. Contraceptive use in 2016 PMA2020 Indonesia survey.

Figure 2
Modern contraceptive prevalence in 2016 PMA2020 Indonesia survey Figure 3