Effects of HIV status and history of pregnancy termination on trends of family planning knowledge among women living in nine fragile countries

Background: To decrease the proportion of women in fragile or conict-affected situations (FCS) that need, but are not using modern contraception, the global health community must better understand family planning (FP) knowledge gaps; and elucidate which subgroups of women are, and are not, effectively being reached with FP information. This study investigated whether women with notable sexual and reproductive health histories that put them at risk for future pregnancy complications, namely HIV+ women and women with history of terminated pregnancy (HTP), would possess more complete FP knowledge and/or recall being given FP information more readily than HIV- women and women with no HTP (NHTP), respectively. Methods: Knowledge of several contraceptive methods, and the prevalence of women who recalled being given FP information at a healthcare visit, or via TV, radio, or periodical were estimated across HIV, HTP, and country subgroups. HIV+ and HTP status were major predictors in multivariate logistic regressions analyzing the odds of knowing a specied method of contraception or of recalling being given FP information via the modes considered. Results: HIV+ as opposed to HIV- women showed signicantly higher odds of knowing about the pill (p=0.001), IUD (p<0.001), injectables (p=0.005), male condoms (p<0.001), female condoms (p<0.001), and implants (p<0.001); but not female or male sterilization. HIV+ women demonstrated higher odds of recalling being given FP information at a health visit (p=0.001), whereas HTP women did not. HTP women showed higher odds (p<0.001) of knowing each modern method of FP with the exception of male sterilization and of recalling TV (p=0.01) and radio (p<0.001) FP messages. Conclusions: Notable ndings were the higher FP awareness among HIV+ and HTP women, and the observation that HIV+ women were seemingly provided FP information more readily at healthcare visits than were HTP women, despite their similar risk proles concerning future pregnancies. Subsequent studies may elucidate why these patterns were observed and indicate which other diagnostic groups are effectively being This study can serve as a framework for future analyses on the prevalence and predictors of FP knowledge among vulnerable groups of women living in fragile regions. Subsequently, these efforts will aid practitioners and policymakers in better targeting FP information dissemination and resources toward not only existing gaps in FP knowledge, but the causes of those gaps as well.


Statistical Analyses
The prevalence of HIV+ and HTP were determined across country, sociodemographic, reproductive health, and sexual health categories. Percentages of women who knew about each method of FP in subgroup analyses based on country, HIV status, and HTP status were calculated along with 95% con dence intervals (CI) and signi cance levels based on second order Rao-Scott design-adjusted chi-square (χ2) values. A multivariate logistic regression model was employed which included the aforementioned major predictors and covariates and each FP knowledge or messaging category as a dichotomous outcome variable. Adjusted odds ratios (AOR) were calculated and corresponding 95% CIs re ect their precision and signi cance. HIV-and NHTP women were reference groups. Therefore, any OR greater than one (1) signi ed higher odds that an HIV+ or HTP woman would possess a category of FP knowledge or recall receiving FP messaging via one of the modes investigated. Statistical analyses were completed using IBM SPSS Version 26 [63] and sample weights were applied using the IBM SPSS Complex Samples package according to the procedures outlined in the Guide to DHS Statistics, DHS-7, Version 2 [58].

Overview
The nine countries included in the study are in SSA and shown in red in Figure 1 below. The weighted percentage of the sample who were HIV+ was 5.5% (95% CI = 5. 3 -5.8). The weighted percentage of women who had HTP was 14.2% (95% CI = 13.9 -14.6). The mean age of respondents was 28.2 years, and 64.3% were married. The country that contributed the most cases was the DRC (n = 9313, 14.8%), whereas the country that contributed the least was Cote d'Ivoire (n = 4655, 7.3%).
Literate women showed a higher prevalence (p<0.001) of HIV+ (7.0%, CI = 6.6 -7.4) compared to illiterate women (3.3%, CI = 3.1 -3.6); whereas illiterate women had a higher prevalence (p<0.001) of HTP (15.2%, CI = 14.6 -15.8) than literate women (13.6%, CI = 13.1-14.1). Table 1 also includes HIV+ and HTP prevalence across wealth index, employment, electricity and communication, marriage, head of household, and recent healthcare visit variables.   Table 2 also includes HIV+ and HTP prevalence across variables indicating who has the nal say on the respondent's healthcare decisions, the respondent's ability to refuse sex, age at rst marriage/cohabitation, number of sex partners in the last 12 months, and condom use during last intercourse. Due to the large amount of information in Tables 1 and 2, all their results could not be described in the text, but the reader is encouraged to use p-values and CIs in the manner described above to observe patterns across other variables of interest.  Table 3 provides the percentage of respondents in each country with knowledge of each category of FP method and who had recalled receiving FP information at a healthcare facility (if visited in the last 12 months), via TV, radio, or in a periodical in recent months. Table 3 differs from Tables 1 and 2 because it presents column percentages.  Notes: Column percentages are presented. Percentages and 95% confidence intervals are weighted. Tests of independence were conducted across rows and based off of the second-order Rao-Scott adjusted chi-square. Statistical significance: * = p<0.001 Table 4 shows the differences between HIV+ and HIV-, and between HTP and NHTP women with respect to the FP knowledge and messaging categories.
Column percentages are presented. The second column of Table 4 shows the percentage of the total sample that reported knowledge of the corresponding FP category or who had received FP information via one of the modes considered. The   Notes: Column percentages are presented. The denominator for the bottommost row is the number of respondents in that column who visited a health facility in the last year. The denominator for all other rows is the total sample for that column. "n" values are unweighted counts. Percentages and 95% confidence intervals are weighted. Tests of independence were conducted across rows.
The χ2 value is a second-order Rao-Scott adjusted chi-square.
Logistic Regression Results  Notes: Major predicting variables in the logistic regressions are HIV status (HIV negative is the reference group) and if the respondent ever experienced a terminated pregnancy (never experiencing pregnancy termination is the reference group). Covariates controlled in the logistic regression models are: country, age (continuous), urban/rural status, education level, wealth index quintile, whether or not they are currently working, electricity in household, mobile phone ownership, current marital status, and whether the respondent considers herself to be the primary FP decider in the family.

Discussion
The primary goal of this analysis was to determine if HIV+ versus HIV-women, and HTP versus NHTP women in fragile regions would exhibit differing FP knowledge pro les and/or report receiving FP information at different rates. For every category of modern FP knowledge, HIV+ women demonstrated a higher level of awareness when compared to HIV-women in initial tests of independence. When covariates were controlled in the multivariate logistic regression models, HIV+ women did not show higher odds of knowing about female or male sterilization; but still demonstrated higher odds of knowing about all other categories of modern FP. Likewise, HTP women demonstrated higher knowledge of every modern method of FP except for male sterilization in tests of independence. In the logistic regression models, HTP women showed higher odds of knowing about every category except for again, male sterilization. It would have been reasonable to expect that HIV+ women would show higher knowledge of FP methods that also have sexually transmitted infection (STI) prevention capabilities (e.g., male and female condoms). However, the magnitude of the knowledge difference between HIV+ and HIV-women was similar for several FP methods that had no STI prevention capabilities (e.g., the pill and implants). In other words, the CIs of the AORs for knowledge of FP methods with and without STI prevention capabilities overlapped. With respect to messaging, HIV+ women were more likely than HIV-women to recall being given FP information by a provider at a recent visit to a healthcare facility. However, despite the fact that like HIV+ women, HTP women are at an increased risk of future pregnancy complications [64][65][66][67][68][69][70][71], they were not more likely than NHTP women to be given FP information at a recent healthcare visit. HTP women did however exhibit higher odds than NHTP women of recalling encountering FP messaging via TV and radio. Since TV and radio messaging could not have been targeted in a way that would have resulted in HTP women having higher environmental exposure to the messages than NHTP women, it is possible that HTP women's past pregnancy experiences sensitized them to subsequent information related to that experience, making them more likely to notice the messaging and/or remember having been exposed to it. But, as establishing that was not the goal of this research, it would need to be explored in further specialized studies.
Although the primary focus of this analysis was not to compare countries, it is important to note that Burundi, Malawi, and Zimbabwe had committed to the United Nations Foundation's Family Planning 2020 (FP2020) initiative several years prior to the DHS being carried out in those countries; whereas Cameroon, DRC, Guinea, Cote d'Ivoire, and Niger either committed to FP2020 in the same year in which the DHS was carried out or years after [72]. The time prior to implementation of the DHS with which institutions in Burundi, Malawi, and Zimbabwe had to implement FP initiatives as a result of their commitment to FP2020 could have impacted the results for these countries. As of August 2021, Angola has not committed to FP2020.
When country of residence, HIV status, and HTP status were used to categorize respondents, the percentage in each category reporting knowledge of 'at least one modern method' of FP was relatively high. However, results indicated that relying on this metric alone was misleading with respect to holistically assessing FP knowledge, since the percentage of respondents with knowledge of each individual FP method was highly variable. For example, over 85% of respondents from Niger reported knowledge of at least one modern method, but only 18.3% and 50.8% had knowledge of IUDs and male condoms, respectively. Over 95% of HIV+ women knew at least one modern method, but only 41.7% knew of male sterilization. Likewise, the percentage of respondents knowing at least one modern method was above 90% in HIV-, HTP, and NHTP women; but only about half of HIV-women knew about IUDs, 60.8% of HTP women knew about implants, and 61.1% of NHTP women knew about female condoms. Therefore, researchers assessing FP awareness are encouraged to not only focus on the percentage of women knowing at least one modern method, but to also take heed of the levels of awareness of speci c FP subcategories, which will provide a more precise understanding of where gaps in FP knowledge in their populations of interest exist.
Knowledge and awareness, along with the provision, availability, and positive attitudes about FP are prerequisites to an individual ultimately using contraception. In one of the few studies that looked at FP in FCS, low use of contraceptives by women in a union was associated with low prevalence of FP knowledge in six local FCS regions in Sudan, Uganda, and the DRC [47]. Beyond the connection between FP knowledge and subsequent contraception use, one should also take into account the trends of demand for contraception not satis ed in countries within our sample when considering this study's implications.
Slaymaker et al. [5] showed that between 1995 and 2018, Burundi, Malawi, Niger, and Zimbabwe demonstrated a decrease in the proportion of women who were sexually active and needing, but not using modern contraception. In Cameroon and the DRC there was no notable change in this proportion, while in Guinea and Cote d'Ivoire the proportion of women who were sexually active and needing, but not using modern contraception increased [5]. Angola was not included in their analysis [5]. While there is a degree of good news in the fact that this proportion decreased over time in Burundi, Malawi, Niger, and Zimbabwe, the most recently measured proportion of women who were sexually active and needing, but not using contraception in the countries within our sample, with the exception of Zimbabwe, were still extremely high. Moreira et al. [33] found that in Guinea (2012) Malawi (2015), and 14.0% in Zimbabwe (2015) [33]. For comparison purposes, this value was recently about 10% in the USA [5] and 8.6% in Columbia [33]. Decreasing these percentages is crucial for improving equity and gender issues, and health in general of women, families, and communities. When creating FP educational interventions meant to contribute to that cause, it is important that women with certain reproductive and/or sexual health histories be targeted due to their increased risk of a variety of negative outcomes. With that in mind, it is encouraging that our study indicates that HIV+ and HTP women in fragile regions seem to have been reached as evidenced by their higher levels of FP knowledge across most categories. However, the other side of this could be an indication that women that do not have easily recognizable reproductive or sexual health histories may not be proactively reached as much as necessary. In other words, women that do not fall into identi able "at risk" categories may be falling through the cracks, thereby perpetuating the cycle that results in women experiencing outcomes like HIV diagnoses or terminated pregnancies in the rst place.
There are several limitations to this study. Although the target population was fragile regions, there were many fragile and/or con ict-affected countries not able to be included because DHS and/or AIS datasets were not available for them, or they did not meet other inclusion criteria outlined in the methods section. In fact, it is debatable how to even classify regions as FCS. We chose the WBG classi cation system since it is standardized and used as a guideline by a variety of international development stakeholders. Another limitation is that the classi cation of pregnancy termination by DHS included women who had miscarried, had a stillbirth, and/or undergone an abortion. These three events, and the circumstances leading up to them can be quite different. It would have been useful if these subcategories could have been analyzed separately. Lastly, this study was based on a cross-sectional survey with self-reported measures and therefore recall bias may have been an issue, temporality of observations and causation cannot be ascertained, and intercountry analyses are of limited value since the DHS surveys for each country were done in different years. Also, social desirability bias could have played a part in women misrepresenting their knowledge of FP. Despite these limitations, this study has several strengths. First, the DHS IPUMS data management tool allowed for a high degree of standardization, harmonization, and comparability between variables. The breadth of variables available in DHS enabled the inclusion of appropriate covariates in the regression models and allowed for robust descriptive analyses of HIV+ and HTP prevalence across a variety of sociodemographic, reproductive, and sexual health subgroups, which alone provides useful information. This study is the only one to our knowledge that investigated trends and determinants of subcategories of FP knowledge along with FP messaging outcomes in a multi-country FCS target population using nationally representative surveys. Lastly, since DHS surveys are repeated approximately every ve years, longitudinal follow-up analyses will be possible.

Conclusion
With this project we aimed to contribute to the reproductive health research base in fragile regions. The descriptive portion revealed that while the percentages of respondents reporting knowledge of at least one modern method of FP may be high, knowledge of speci c subcategories of FP were extremely variable and in some cases very low. Increasing the awareness of a variety of modern methods of FP may enable women to access contraceptive methods that are better aligned with their personal desires. The inferential portion investigated predictors of FP knowledge and demonstrated the potential importance of breaking down analyses of FP knowledge by relevant medical histories of respondents. Further research is needed to clarify the mechanisms behind why we observed higher FP knowledge among HIV+ and HTP women and why HIV+ women were seemingly targeted with FP information more readily at health facility visits than were HTP women; whereas HTP women were more likely to recall TV and radio FP messaging. Subsequent studies should also seek to determine which other population subgroups at risk for pregnancy complications are effectively being reached with FP education, and which are not. Better