Health Inequalities in Unmet Need for Family Planning Among Ugandan Women: Repeated Cross Sectional Surveys in the Years 2014 to 2018.

Background: Health inequalities in unmet need for family planning have been documented in Uganda, however, little is known about their magnitude and whether these have remained the same. Objective: This study sought to examine health inequalities in unmet need for family planning among Ugandan women between 15-49 years of age in the years 2014 to 2018. Methods: Five data sets of the Performance Monitoring Accountability 2020 family planning cross-sectional surveys were used to assess health inequalities in unmet need for family planning across four socio-economic position variables (age, education, wealth status and geographical location) at five time points (2014 to 2018).Chi- square (χ 2 ) test and a logistic regression-based measure, the Relative Index of Inequality (RII) were used to assess differences and relative health inequalities respectively. Results: The data of 19,670 women aged 15 to 49 years were analysed. Between 2014 and 2018, unmet need for family planning (limiting, spacing and total) were 2 or more times more prevalent among the uneducated (RII value range:1.94 to 2.73), and poorest women(RII value range:1.90 to 3.78), in comparison with women with post-secondary education and richest women respectively. Unmet need for family planning was more prevalent among women older than 35 years (RII between 0.41 and 0.63). No geographic-related health inequalities were observed. Education-related inequalities reduced, wealth-related health inequalities increased, while age-related inequalities remained fairly consistent. Conclusion: Age, education and wealth related health inequalities were observed in unmet need for family planning. The magnitude of health inequalities varied between 2014 and 2018, with largest inequalities based on wealth status variable.


Background
The use of modern contraceptive methods among women of reproductive age could reverse Uganda's high maternal and child deaths. [1][2][3] Although modern contraceptive use in Uganda has increased 4,5 , a sizeable proportion (33.2%) of fecund women who would either want to delay or stop having children are currently not using any contraceptive method (known as unmet need for family planning) persists. 1,6 Unmet Need For Family Planning (hereafter abbreviated as UNFP) has been associated with unintended pregnancies. 7,8 These could result in unsafe abortions [9][10][11][12] in Uganda where abortion is illegal. 13 Furthermore, UNFP has been associated with short inter-pregnancy intervals (the period between a live birth and following pregnancy)and high parity. 14-16 Both short inter-pregnancy intervals and high parity have been associated with anaemia, preeclampsia, gestational diabetes, congenital malformations, preterm delivery neonatal deaths and chronic malnutrition among infants. [17][18][19][20][21][22][23] Consequently, UNFP significantly contributes to maternal and child morbidity and mortality rates in Sub-Saharan countries like Uganda. 24,25 For primary health care programmes, UNFP indicates failures in reaching those in most need of family planning services and consequently their reproductive health needs. 26,27 UNFP is often unequally distributed with some disadvantaged socio-economic groups suffering higher burdens. 28,29 In response, global initiatives through the 2012 London Summit on Family Planning committed to identify women with a higher UNFP and develop tailored interventions addressing their needs. 30 Previous studies in Sub-Saharan countries including Uganda have suggested that young, rural, least educated and poorest women are more likely to have a higher UNFP compared to their counterparts. [31][32][33][34][35][36][37] Such differences in health status where lower socio-economic positions suffer higher burdens of negative health outcomes such as UNFP suggest health inequalities. 38 In Uganda, health inequalities in UNFP may limit health gains anticipated from the remarkable progress in increased modern contraceptive use. 39 Furthermore, they may prevent the achievement of Uganda's Family Planning 2020 targets of reducing UNFP to 10% by 2020. 40 Although health inequalities in UNFP in Uganda have been documented 37 37 However, these assessments were done over ten years ago and may not reflect recent changes if any. This presents a need to utilise more contemporary data for examining any changes in health inequalities in UNFP. Recent studies have been restricted to particular geographic areas 41,42 and are therefore limited in national representativeness. Therefore, this study conducted a secondary analysis of repeated cross-sectional survey data to examine health inequalities in UNFP among women aged between 15-49 years in Uganda over five years.

Data source
This study conducted a quantitative analysis of secondary data from the Performance Monitoring Accountability 2020 (PMA 2020) repeated cross-sectional surveys conducted annually in the years 2014 to 2018. These surveys collected data on socio-economic characteristics and family planning practices to inform progress in achieving family planning 2020 targets in Uganda. 43 These surveys targeted Ugandan women of reproductive age (between 15 and 49 years),who are the main focus of family planning interventions. 44 In each survey year, the sample size was determined using the Kish Leslie formula as detailed in Zimmerman. 45 Two-stage stratified cluster sampling was used to construct nationally representative sample at low cost. 46 A total of 110 enumeration areas (primary sampling unit) were selected based on rural/urban stratification within the ten sub-regions using probability proportion to size from the Uganda Bureau of Statistics master sampling frame created during the 2002 National Population and Housing Census.
A total of 44 households were randomly selected from an up to date list of households in each enumeration area. All consenting females between 15 to 49 years of age were included in the study in each of the survey rounds.
Data were collected using the household and female structured questionnaires socioeconomic and family planning information respectively as detailed in Zimmerman. 43 Data were collected using a mobile phone device by a female resident enumerator. UNFP-the proportion of all exposed women (currently married and sexually active unmarried women) that were classified as fecund, not using contraception despite not wanting more children or not wanting a child in the next two years and pregnant and postpartum amenorrheic women (up to two years) whose pregnancy or last birth was unwanted or mistimed.

Variables of interest
Pregnant and postpartum amenorrheic women with pregnancy or last birth resulting from contraceptive failure were not classified as having UNFP.
2) Geographical location-was categorised into either rural or urban areas.
3) Level of education-was categorised based on the highest level of education attained in the Ugandan education system to include: no education, primary level, secondary level and post-secondary level. The group 1(15 to 19 years), rural, no education and lowest wealth quintile categories were considered as the lowest socio-economic position in regards to age, geographic location, education, and wealth status respectively.

Data Analysis
Only completed questionnaires were considered for analysis. The survey responses were weighted for the data analysis to correct for the disproportionality of the sample introduced by the complex survey design. 48 Survey weights were derived by taking into account the probability of selected enumeration areas, households and non-response adjustments. 49 Final weights were normalised at the national level. Descriptive statistics were used to generate proportions of women with UNFP in the various categories in all the survey years. Chisquare (χ 2 ) test was used to test whether there were differences in the distribution of UNFP across socio-economic position categories for each indicator in the various survey years. 50 Statistically significant differences consisted of those with p-values less than 0.05.
The Relative Index of Inequality (RII) was used to assess relative health inequalities. RII is an odds ratio that can be interpreted as the odds of UNFP in the lowest socio-economic position category compared to the highest socio-economic position category. 51 Values of RII greater than 1 indicate health outcomes are concentrated in the lowest socio-economic position category while values less than one indicate health outcomes are concentrated in the highest socio-economic position category if the 95% Confidence Interval (CI) does not include one and p-value is less than 0.05. 52 This was computed as explained further below.
Individuals were ranked in each socio-economic position indicator. 53 To construct this, categories in each socio-economic position were arranged cumulatively from the lowest to the highest categories. Ridit scores were used to obtain the relative rank of each category based on the mid-point range of cumulative distribution in each category. The relative ranks were weighted to account for population size in each category. The lowest socio-economic position was assigned a rank of 0 while the highest socio-economic position was assigned a rank of 1. 53 A logistic regression was used to regress the obtained relative ranks in each of the socioeconomic position indicators on the binary outcomes (yes/no) for UNFP. Adjustment for covariates was done by a forward stepwise approach while maintaining the socio-economic position variable of interest. Changes in health inequalities in UNFP were assessed by comparing the values of the RII obtained in each socio-economic indicator across the survey years. All statistical analysis was conducted using the software STATA (version 12).

Ethical consideration
Anonymised publicly accessible data was obtained from the PMA 2020 website following approval of the PMA2020 Coordinating Office.

Age-related health inequalities
Throughout the study period, UNFP was concentrated among the 36-49 year age group with both the unadjusted and adjusted RII values (detailed in table 3 below) were less than 1.
These were statistically significant with 95% confidence intervals not including 1 and pvalues less than 0.05. Throughout the study period, women in the 15 to 19 years age group were 37% to 59% less likely to experience UNFP in comparison to women in the 36 to 49 years age group, following adjustments for education, geographic location and wealth status.

Wealth-related inequalities
UNFP was concentrated among the poorest women with both the unadjusted and adjusted RII values greater than 1. Wealth-related health inequalities in UNFP were observed throughout the survey period except in the year 2015(95% confidence intervals not including 1 and pvalues less than 0.05). The poorest women were 2 to 4 times more likely to experience UNFP compared to the richest women following adjustments for age, education and geographical location. Table 6 below shows the details.

Changes in health inequalities in UNFP
The  categories (namely least educated, poorest and rural) consistently reported a higher UNFP.
These suggest persistent health inequalities. UNFP generally increased with decreasing socioeconomic position except for the age variable, suggesting a social gradient.

Age-related health inequalities.
Interestingly, the results show persistent reverse age-related health inequalities in UNFP.
Reverse health inequalities occur when adverse health outcomes are concentrated among the highest 'socio-economic position'(oldest women). 56 Some studies in Nigeria 57 and Ethiopia 58 have shown that older women are less likely to use modern contraceptives. Because of their age, women older than 35 years perceive themselves as less likely to get pregnant due to their reducing sexual exposure and fecundity, and so do not use contraceptives. 59

Education-related health inequalities
The study showed sizeable education-related health inequalities. Those with no education were two to three times more likely to have of UNFP compared to women with postsecondary education. Low levels of education have been associated with lower health literacy. 60 Women with no education may, therefore, have reduced exposure to family planning information due to their inability to read, which can also contribute to inaccurate perceptions about their risk of pregnancy and the side-effects of modern contraceptives. This may make these women less likely to use modern contraceptives to fulfil their pregnancy intentions (wanting to space their next pregnancy or have no more pregnancies) in comparison to their more educated counterparts.
There was a reduction in education-related health inequalities in UNFP. This is similar to study conducted in Egypt 61

Geographic-related health inequalities
In this study, as demonstrated previously in Ethiopia 35,36,63 , rural women reported a higher UNFP. However, no geographic-related inequalities were observed following adjustments for age, wealth status and education. This might suggest that rural-urban differences could result from differences in the distribution of wealth, education attainment and age between the two areas. This lends support to previous findings in Ghana 64 that indicated that regional variation in UNFP was associated with the distribution socio-economic and demographic variables such as age, wealth and education within these areas.

Wealth-related health inequalities
The findings of this study indicated that substantial wealth-related inequalities, with the poorest women being two to four times more likely to have UNFP compared to the richest women. Wealth-related inequalities in UNFP have also been documented by Chauhan 65 in India. However, comparisons could not be made due to the different measures used to assess health inequalities (concentration index in Indian study and the RII in this study). Poor women may be unable to access free modern contraceptive services offered at health facilities due to transport costs. 66 The subsidised community-based services may remain unaffordable to the poorest women as their priority is meeting their basic needs first with the few available resources they may have. Based on this, the poor are less likely to access and therefore benefit from the increased availability of modern contraceptives compared to richer women. 67 This study shows increasing wealth-related inequalities in UNFP. This could be attributed to higher reductions in UNFP among the richest women in comparison to the poorest women over the period selected in this study.

Strengths and Limitations
This study benefitted from using large nationally representative samples with high response rates making the results generalisable to Ugandan women aged 15 to 49 years. 68,69 The use of multiple socio-economic position parameters which provided a comprehensive picture of the health inequalities in UNFP. 70 In absence of longitudinal data, repeated cross-sectional surveys allowed assessment of population-level changes in health inequalities over five years. 71 The RII allowed assessment and comparison of the magnitude of health inequalities while accounting for the changes in the distribution of age, education, wealth and geographical areas during the five years. 72 However, the proportion of women who participated in a previous survey year (14.73% to 28.93%), may have reduced the independent nature of the observations. 73 However, this is likely to had a minimal impact as a women's pregnancy intentions and contraceptive use patterns are likely to change over a short time. 74 The ranking of women categories introduced value judgments while considering the least and most advantaged positions. Results may therefore, differ if there are alterations are made. 75 The secondary data analysed did not contain information on occupation which is a key 'socio-economic position' variable. The exclusion of occupation status while adjusting for other covariates may have resulted in residual confounding. 76

Conclusion
The study showed substantial health inequalities related to age, wealth and education in UNFP, whose magnitude has changed over time. The reverse age health inequalities observed question the overly simplistic assumption that that those in lower socio-economic positions inevitably bear the highest burden of unfavourable health outcomes. Future research needs to use time series analysis further to establish the period effects, such as changes in policy and family planning interventions) that have influenced changes in health inequalities.

Declarations
Ethical Approval and Consent to participate Anonymised publicly accessible data was obtained from the PMA 2020 website following approval of the PMA2020 Coordinating Office.

Not Applicable
Availability of data and materials The datasets analysed in this study are publicly accessible data from the PMA 2020 website upon approval of the PMA2020 Coordinating Office.

Competing interests
The Author has no conflict of interest to disclose.

Funding
The author received no financial support for the research, authorship, and/or publication of this article.

Authors' contributions
Ms Wanyana Mercy Wendy was responsible for the design, analysis, interpretation and writing this research article.