Introduction

Anemia is a significant global health concern that impacts quality of life, economic productivity, and mental and physical development for an estimated 1.92 billion people worldwide [1]. Symptoms can include weakness, fatigue, and difficulty concentrating, and when severe, anemia can increase the risk of poor birth outcomes such as preterm labor, low birth weight, and child and infant mortality [2,3]. Iron deficiency anemia is the most common form, accounting for over 60% of the global disease burden [2]. Iron deficiency anemia can stem from multiple causes, including blood loss related to menstruation or pregnancy, as well as diet, and blood loss due to infection by parasites [2,4]. Women of reproductive age are, therefore, at high risk of anemia and often bear the majority of the health and economic costs compared to men [2]. The negative effects of heavy menstrual bleeding on women has been well documented; in addition to many other impacts, heavy menstrual bleeding can increase risk of anemia [5,6]. Preventing and treating anemia in women of reproductive age can greatly improve women’s health and quality of life while also having important population health implications. Preconception and pregnancy are times of heightened concern around anemia; when the birthing parent has low iron stores, infants may have an iron deficiency that could impact their physical and mental development [3]. Despite the global importance attributed to anemia, including the Sustainable Development Goal Indicator 2.2.3 of reducing anemia by half in women and children between 2012 and 2030, there has been little global progress toward this goal [7,8]. Therefore, it is of public health importance to research strategies and interventions to address anemia in women of reproductive age, especially in low- and middle-income countries (LMICs) where the prevalence of anemia is notably higher compared to high-income countries [2].

Family planning methods allow for delaying or avoiding pregnancy, lengthening the time between pregnancies, and—for hormonal contraceptive methods like oral contraceptive pills, injectable contraception, contraceptive implants, and hormonal intrauterine devices (IUD)—reducing or entirely pausing menstrual blood loss [9]. In general, the estrogens and progestins present in hormonal contraceptives thin the endometrial lining of the uterus which often leads to lighter bleeding or amenorrhea, which can affect anemia status [10,11]. The hormonal IUD does this primarily through local progestin release, while the others act systemically [12]. These bleeding changes caused by hormonal contraception are examples of a wider constellation of ways contraception impacts the menstrual cycle, referred to as contraceptive-induced menstrual changes (CIMCs), which can be a key factor in contraceptive decision-making [13]. These impacts are highly personal and context-specific; previous research has found a range of positive and negative responses to CIMCs, with lighter bleeding generally viewed positively, but with both positive and negative responses to completely paused bleeding [13,14].

Given hormonal contraception can reduce menstrual blood loss, there is a theoretical non-contraceptive benefit of use to prevent or treat anemia. However, different bleeding profiles across different methods of hormonal contraception mean some contraceptive methods may be more protective against anemia than others. For example, most combined oral contraceptives include placebo pills for 4 or 7 days to induce withdrawal bleeding designed to mimic menstruation, though this bleeding is often less than a typical menstrual period. Bleeding can be further reduced by continuous use of combined oral contraception, i.e., not taking the placebo pills, and several marketed formulations provide extended, 3-month coverage between withdrawal bleeding; however, it is not evident that these formulations and the continuous-use practice are common in LMICs. Additionally, oral contraceptive pills require high user adherence to daily pill taking, and most women do not achieve 100% adherence, which can result in breakthrough bleeding or pregnancy [15,16]. Still, small-scale studies in high-income countries have shown associations between oral contraceptive pill use, reduction in menstrual blood loss, and subsequent reductions in anemia and iron deficiency [17,18].

In addition to oral contraceptive pills, the contraceptive implant or injectable contraception can reduce or pause bleeding [19,20]. Compared to oral contraceptive pills, however, these longer acting methods require less user involvement, such as remembering to take a daily pill [21,22]. Another long-acting method, the hormonal IUD, has the potential to manage many menstrual and gynecologic disorders and symptoms including heavy menstrual bleeding, and there is currently a clinical trial [NCT05233956] to investigate its impact on anemia [23]. Although there are current efforts to increase access to the hormonal IUD in LMICs due to its high effectiveness and non-contraceptive benefits, its use has historically been very limited. The non-hormonal IUD, on the other hand, has been shown to increase bleeding, which could enhance the risk of anemia [24–27]. Determining whether, and to what extent, various types of hormonal contraceptives have the potential for anemia treatment or prevention may provide important information for those using or considering hormonal contraception, potentially moving the needle on a condition that has been challenging to address.

There is a growing body of research in LMICs exploring the relationship between hormonal contraception and anemia [28–30]. In 2012, Bellizzi and Ali found an association between oral contraceptive use and lower odds of having anemia (OR 0.68, 95% CI: 0.64–0.73) using Demographic and Health Surveys (DHS) data from in 12 LMICs, with the effect increasing with longer duration of pill use [29]. More recently, using DHS data from 51 LMICs, Misunas et al. found use of the contraceptive pill, injectable contraception, or contraceptive implant significantly reduced the odds of mild, moderate, and severe anemia in adolescent girls and young women aged 15–24 [28], and Aboagye et al. found similar results among all women of reproductive age in DHS from 14 sub-Saharan African countries [31]. Using DHS data from Ethiopia, Tshome et al. found users of modern contraceptive methods, which includes hormonal contraceptives, had lower odds of anemia than women using barrier methods of contraception, such as condoms [30]. Although most of the research on LMICs come from analysis of DHS data, there have been smaller community-based studies in LMICs that have also found associations between hormonal contraceptive use and anemia reduction [32,33]. Overall, this research suggests there is an association between hormonal contraceptive use and anemia reduction in LMICs, but it has largely focused on either solely oral contraceptive pill use or hormonal IUD use, or it has conflated hormonal contraceptive methods together. Additionally, there has been concern that the HemoCue testing system used to determine anemia status in DHS surveys may over- or under-estimate true anemia rates, introducing measurement error into analyses [34–39]. There is a need for further comparative studies to understand to what extent hormonal contraceptives, and specific hormonal methods, can be protective at a population level against anemia, and to what extent population level estimates of anemia may be biased.

Our study seeks to address several previous gaps in the literature. Specifically, previous studies have limited analysis to one method of contraception, or investigated multiple methods only in specific demographic populations, countries, or regions [24–26,28–32,40]. Additionally, few previous studies have conducted robust sensitivity analyses, explored moderation effects by method type, or addressed concerns about measurement error in anemia testing systems. Our analysis builds upon previous research by further exploring the relationship between hormonal contraceptive use and anemia, in aggregate and by individual method, over multiple durations of use for women of reproductive age in 46 countries across four global regions and by assessing the robustness of our results when accounting for measurement error in anemia point-of-care testing systems.

Methods

Measures

Exposure.

Data on the exposure, hormonal contraceptive use, are collected in the woman’s questionnaire. Women are asked if they or their partner are currently doing something to prevent pregnancy. If yes, they are asked the method(s) they are using. We defined hormonal contraceptive users as woman who reported using—singly or in combination with another method—oral contraceptive pills, the contraceptive implant, or injectable contraception. The DHS does not distinguish between hormonal IUDs, which can cause less menstrual bleeding, and non-hormonal IUDs, which can cause increased menstrual bleeding; therefore, we did not include IUD users (n = 23,694) in the analysis sample. Women reporting female sterilization (n = 196,506) were also not included, as sterilization can have varied effects on the menstrual cycle [43]. We excluded hormonal contraceptive users if they had not been using their method for at least 6 months (n = 20,182) to ensure the method had enough time to impact anemia and the contraceptive-induced menstrual changes had time to stabilize. We also excluded women who were over 49 years old (n = 4,801), as only some DHS include women above this age. All other women were considered non-users of hormonal contraception.

Therefore, our exposed group (users of hormonal contraception) included women using oral contraceptive pills, the contraceptive implant, or injectable contraception for more than 6 months, and our reference group (nonusers of hormonal contraception) included all women who did not use any form of contraception or women using non-hormonal forms of contraception, including male and female condoms, male sterilization, periodic abstinence, the standard days method, withdrawal, lactational amenorrhea, spermicides, or traditional/folk methods, as defined by the DHS.

Outcome.

DHS measure anemia status using 1–3 drops of capillary blood from a finger prick sample using the HemoCue 201+ or 301 + system, which provides a reading of blood hemoglobin concentration in grams per deciliter (g/dL) in one minute [44,45]. The HemoCue system is intended to be a screening, not a diagnostic, tool for anemia, and can only screen for general anemia, not specifically iron deficiency anemia, as it only measures blood hemoglobin and not and individual’s iron reserves. Because anemia is defined differently for women who are pregnant due to changes in blood volume and other factors, we did not include pregnant women (n = 56,719) in our analysis sample as part of the referent non-user group. DHS categorizes non-pregnant women as having anemia if their hemoglobin is below 12 g/dL, with mild anemia between 11.0–11.9 g/dL, moderate anemia between 8.0–10.9 g/dL, and severe anemia below 8.0 g/dL [46]. We dichotomized the outcome as either having anemia (1) or not having anemia (0), collapsing the categories of mild, moderate, and severe anemia.

Covariates.

We used a directed acyclic graph (DAG) [47,48] to identify an adjustment set of covariates, including age, wealth, smoking, education, and location (S1 Fig). DAGs are commonly used in epidemiological research to graphically depict causal relationships between variables; to create our DAG, we used the results of a thorough literature review on anemia and links between anemia and hormonal contraceptive use and our subject matter expertise [49]. Using the DAG, we identified causal and confounding pathways between our exposure of interest (hormonal contraceptive use) and our outcome (anemia), then identified which variables we need to adjust for in order to control for confounding pathways. Age was measured continuously by year, and we used a seven-level categorical variable of five-year age groups (i.e., 15–19, 20–24, etc.) for analysis, although we explored four other functional forms, including linear, quadratic, categorical, and quadratic splines. LOESS plots and precision for each functional form were similar, so we chose the categorical variable for ease of comparison with other similar studies. Wealth, used in this analysis as a proxy for social inequality, was measured as a five-level categorical variable using the DHS wealth index which uses an equation that creates wealth quintiles by considering dwelling and household characteristics, access to consumer goods and services, assets, health outcomes and the Gini coefficient, which indicates the level of wealth inequality in that country [50]. Smoking was a binary variable, measured as a self-report of tobacco cigarette consumption. Education was measured by asking women their highest level of completed education, and this was recoded as a 4-level categorical variable for completed schooling, including no school, primary school, secondary school, or more than secondary school to permit comparison across countries. Rural or urban status of residence (i.e., rurality) was determined for each country then verified during data collection and reported as a binary variable.

Models.

We used our DAG to determine which covariates we would control for in our models to study the relationship between hormonal contraceptive use and anemia. Based only on the DAG, we would control for age, wealth, rurality, smoking, and education. Based on our knowledge of the literature and expert recommendation, we tested an interaction term for rurality and found no evidence of effect modification (likelihood ratio test statistic: 2.3, p = 0.13), and therefore left this interaction term out of the final model. We weighed the benefits and drawbacks of retaining smoking and educational attainment as confounders. Removing smoking resulted in no change to the estimate or confidence intervals, likely due to the very low prevalence of smoking in our sample (0.8%) and the fact that the DHS corrects anemia levels for smoking and altitude. Therefore, we did not include smoking as a confounder and were able to increase our sample size by 10,372. There was a small change in estimates when removing educational attainment, therefore, we retained it in our model. As shown in Fig 1, our final model controlled for age, wealth, education, and rurality. A more detailed DAG, which takes into account potential confounders that are unmeasured in the DHS (including infections, dietary intake, and underlying health conditions), can be found in S1 Fig.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Simplified directed acyclic graph (DAG) illustrating relationship between hormonal contraceptive use and anemia status, with final covariate set.

For a full DAG with all variables considered, please see Supplementary S1 Fig.

https://doi.org/10.1371/journal.pone.0327083.g001

Results

Analysis sample

All 46 DHS combined included 1,375,291 women (Fig 2). Our theoretical analysis sample excluded households without anemia testing, and excluded women who were pregnant, used IUDs or female sterilization, or had been using hormonal contraceptives for less than six months. The final analysis sample included 790,065 women who were not missing exposure (n = 9), outcome (n = 10,762), or covariate data (n = 13 missing education).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Flowchart detailing the merged DHS sample to the final analysis sample.

https://doi.org/10.1371/journal.pone.0327083.g002

Anemia prevalence by country

Anemia prevalence ranged from 12.4% in Rwanda to 63.3% in Gabon, both countries in sub-Saharan Africa (Fig 3). Anemia prevalence ranged from 13.3% to 48.3% in the Latin America/Caribbean region (Guatemala and Haiti, respectively), from 13.7% to 44.8% in the North Africa/West Asia/Europe region (Armenia and Jordan, respectively), and from 19.9% to 56.6% in the South/Southeast Asia region (Timor Leste and India, respectively). Half of countries had an anemia prevalence over 40%, only six of which were outside of sub-Saharan Africa. Nine countries (i.e., Cote d’Ivoire, Congo, Mozambique, Mauritania, Benin, India, Niger, Mali, and Gabon) had an anemia prevalence over 50%.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Weighted anemia prevalence (hemoglobin concentration below 12 g/dL) among non-pregnant women of reproductive age (15-49 years), by region and country, in 46 DHS conducted between 2008 and 2022.

https://doi.org/10.1371/journal.pone.0327083.g003

Odds of anemia with hormonal contraceptive use by country

In each of the 46 countries, we compared odds of anemia in women using hormonal contraception (users) to those not using hormonal contraception (nonusers), controlling for age, wealth, education, and rurality (Fig 4). Overall, adjusted odds ratios (AOR) for anemia ranged from 0.45 in Tanzania (95% confidence interval: 0.39–0.52) and Guatemala (0.38–0.54) to 0.98 in India (0.94–1.02), indicating that users of hormonal contraception had lower odds of being anemic than non-users. In the sub-Saharan Africa region, AORs ranged from 0.45 in Tanzania to 0.96 in Guinea (0.69–1.33); in the Latin America/Caribbean region, AORs ranged from 0.45 in Guatemala to 0.82 in Haiti (0.62–1.10); in the North Africa/West Asia/Europe region, AORs ranged from 0.71 in Egypt (0.57–0.88) to 0.90 in Jordan (0.65–1.24); and in the South/Southeast Asia region from 0.49 in Cambodia (0.43–0.57) to 0.98 in India (0.94–1.02). While the difference between the odds ratios was lowest for the North Africa/West Asia/Europe region, this region also had estimates with low precision and without statistical significance. The odds of anemia in 35 countries were statistically significantly lower for users compared to nonusers at an alpha of 0.05; only 11 countries were not significant, including three countries in the sub-Saharan Africa region (i.e., Guinea, Cameroon, and the Democratic Republic of the Congo), Haiti in the Latin America/Caribbean region, five of the six countries in the North Africa/West Asia/Europe region (i.e., Albania, Armenia, Jordan, the Kyrgyz Republic, and Tajikistan), and India in the South/Southeast Asia region. While the estimates of odds of anemia ranged in precision, nearly 85% (39 out of 46) had confidence limit ratios under 2; those seven countries with a confidence limit ratio above 2 ranged from Tajikistan (2.14) to Albania (6.85) and were either in the North Africa/West Asia/Europe or Sub-Saharan Africa regions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Odds ratios and 95% confidence intervals for anemia in non-pregnant women of reproductive age (15-49 years) using hormonal contraceptives for at least six months compared to nonusers of hormonal contraception adjusted for age, wealth, education, and rural residence, by region and country, across 46 DHS conducted between 2008 and 2022.

https://doi.org/10.1371/journal.pone.0327083.g004

Sample characteristics

When we combined data from the 46 DHS, nearly half (48%) of the merged sample had anemia (Table 1). Only 9% of the merged sample were users of hormonal contraception for at least six months, meaning 91% were non-users of hormonal contraception. Among users, 36% were anemic, whereas half (49%) of non-users were anemic, and anemia was lower in women who used injectable contraception and the contraceptive implant compared to those who used the pill. In this bivariate comparison, anemia varied somewhat by age, decreased as wealth and education increased, and was higher among those living in rural areas compared to urban areas. As mentioned above, India alone had a significant impact on these results and was therefore excluded from the moderation analysis. In the moderation analysis, we included those who had used contraception for any length of time. In this sample, 18% were using hormonal contraception, and 25% of hormonal contraceptive users were anemic. For a comparison of anemia by sociodemographic characteristics among all 46 countries, India alone, and 45 countries without India see S2 Table.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Anemia by sociodemographic characteristics among non-pregnant women of reproductive age (15-49 years) across 46 or 45 DHS conducted between 2008 and 2022.

https://doi.org/10.1371/journal.pone.0327083.t001

Merged analysis

In our regression model of the merged dataset, we found users of hormonal contraception for at least six months had 0.55 times the odds of having anemia compared to non-users when controlling for age, educational attainment, wealth, and rurality (Table 2). The 95% confidence interval for this odds ratio is 0.39 to 0.79, which indicates that these data are relatively compatible with a population level adjusted odds ratio between 0.39 and 0.79, conditional on age, educational attainment, wealth, and rurality, at an alpha of 0.05.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Adjusted odds of anemia among non-pregnant women of reproductive age (15-49 years) by contraceptive and sociodemographic characteristics across 46 DHS between 2008 and 2022.

https://doi.org/10.1371/journal.pone.0327083.t002

Moderation analysis

To examine moderation effects by method type and duration of use, we only included 45 countries in the merged analysis. We decided to exclude the India DHS due to the results of our sensitivity analyses, which found a unique moderation effect by method type unseen in other countries (see Supplementary Material S1 Appendix for a detailed explanation of this sensitivity analysis and an exploration of the moderation analysis with all countries included). In the remaining 45 countries (n = 334,907), we found moderation effects by method type but not duration of use. Overall, those women using hormonal contraception for any length of time had 0.51 times the odds (0.46–0.57) of anemia compared to non-users of hormonal contraception, which decreased further to 0.45 times the odds (0.39–0.51) for those women using hormonal contraception for more than two years (Fig 5, panel A). The effect was strongest for users of the injectable contraception, with odds of anemia ranging from 0.46 (0.39–0.52) to 0.37 (0.30–0.45), depending on duration of use (Fig 5, panel B). The effect was more moderate for women using oral contraceptive pills, for whom the odds of anemia compared to non-users ranged from 0.6 (0.49–0.72) with any duration of use to 0.53 (0.43–0.66) for use of more than 2 years (Fig 5, panel C), and users of the contraceptive implant, where the effect ranged from 0.57 (0.44–0.73) with any duration of use to 0.55 (0.43–0.70) with more than two years of use (Fig 5, panel D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Odds ratios and 95% confidence intervals for anemia in non-pregnant women of reproductive age (15-49 years) using hormonal contraceptives compared to nonusers of hormonal contraception adjusted for age, wealth, education, and rural residence, stratified by method type and length of use, in 45 DHS conducted between 2008 and 2022.

https://doi.org/10.1371/journal.pone.0327083.g005

Discussion

This analysis adds to a growing body of work examining the relationship between contraceptive use and anemia in LMICs. Our study differs from previous research in several key ways. First, we examined the impact of oral contraceptive pills, contraceptive implants, and injectable contraception, unlike many previous studies that have only focused on use of oral contraceptive pills [29,40] or use of hormonal IUDs [24–26]. Second, we did not limit our analysis to specific demographic populations, countries, or regions; instead, we included women of reproductive age in multiple LMICs across four global regions in our comparative analysis. We found women using hormonal contraception had nearly half the odds of having anemia of women who were not using hormonal contraception, controlling for age, wealth, educational attainment, and area of residence. These findings are consistent with other similar studies of the relationship between anemia and contraceptive use and support the growing evidence that hormonal contraceptive use has a causative impact on anemia [28,29,31]. Especially in the Latin America/Caribbean, South/Southeast Asia, and sub-Saharan African regions, our estimates indicated hormonal contraceptive use is associated with lower odds of being anemic.

Third, although other analyses have also used DHS data to examine multiple hormonal contraceptive methods in aggregate, we conducted novel stratified analyses by both method type and length of use to examine their moderation effect and addressed concerns about the accuracy of the anemia testing system used by the DHS by assessing if our results are robust to measurement error [28,31,34,35]. Building upon the previous research that focused on oral contraceptive and IUD use, our moderation analysis found that the protective effect of hormonal contraceptive use is greater for injectable contraception compared to oral contraceptive pills at 13–24 months. This is likely due to level of user involvement and the differential impacts of these forms of contraception on the menstrual cycle. Injectable contraception is more likely than contraceptive implants to completely pause bleeding, especially with extended use, and oral contraceptive pills include combined hormonal pills, which are designed to mimic monthly menstrual bleeding with withdrawal bleeding due to the 4–7 days of placebo pills [19–21]. For oral contraceptive pills, users must take them daily, whereas injectable contraception requires less user involvement, with one visit every three months. This lack of effect by duration of use differs from what previous studies on the relationship between oral contraceptive use and anemia have found, notably a similar DHS analysis including 12 countries [29]. Additionally, there are fewer implant users in our sample, which may make the confidence interval for our estimate less precise. We did not find significant effects for duration of use within methods; while the point estimate was lower with longer duration of use for most methods, there was still overlap in confidence intervals when comparing longer and shorter durations of use.

Our use of DHS data for this analysis has both strengths and limitations. Data collected in a standardized way across countries allows us to examine broader trends across countries and regions. On the other hand, we lose the nuance of doing a more in-depth analysis of the unique context of one country. The DHS is not done in every country; this and our exclusion criteria limits our ability to make strong regional comparisons. The rigorous complex sampling design mitigates concerns about selection bias and the quality of data collection with generally low levels of missing data (e.g., we were only missing data on 10 individuals for the exposure in a sample of over 1 million women). Still, each country and regions within countries have unique characteristics, and DHS data, which are designed to be comparative across countries, limit in our ability to control for country-specific confounders. Although the DHS wealth index can account for some measures of social inequities, we are still unable to control for country-specific factors that could greatly affect anemia, including nutritional status or infectious disease. The cross-sectional nature of the DHS also limits our ability to draw causal inferences. Furthermore, because the DHS does not distinguish between the copper and hormonal IUD, we had to exclude all IUD users from our analysis given their very different impacts on the menstrual cycle, despite promising evidence of the impact of the hormonal IUD on anemia [24,26,27]. While some countries opt in to collecting data about brands of pills or condoms, the DHS does not collect data about brands or formulations of implants and injections. For injections specifically, the different menstrual bleeding patterns seen with progestin-only injectables compared to combined injectable contraception (CIC) should be taken into consideration in countries where CIC use is high [53]. Additionally, the DHS does not collect more granular data on variables like menstruation that would have been helpful to further explore possible mediating pathways between hormonal contraceptive use and anemia status.

Despite these limitations, our findings strongly suggest hormonal contraceptives may be a useful way to address anemia in women of reproductive age, and this impact may be stronger for longer acting methods that require low user involvement and/or completely pause bleeding. Our moderation analysis provides evidence that contraceptive implant and injectable contraceptive use may result in a protective effect on anemia. Our analysis for misclassification of the outcome, which is a significant concern with the HemoCue+ systems, strengthen this evidence. Even if we assumed that the sensitivity and specificity were as low as 75%, we still found hormonal contraceptive users had lower odds of having anemia compared to nonusers.

Continued research is necessary to establish a causal link between hormonal contraceptive use and anemia reduction. For example, the LISA randomized controlled trial [23] is currently exploring associations between use of hormonal IUDs and anemia status. Still, this paper adds to the growing evidence that anemia prevention is a notable non-contraceptive benefit of certain types of hormonal contraception, including contraceptive implants and injectable contraception [28–31]. This link between anemia and contraceptive use is a promising point for intervention as we expand options for treatment of and prevention of anemia, but research and translation of that research into practice in these areas remain largely siloed between those studying nutrition and family planning. Interdisciplinary work between these related fields is greatly needed in order to begin to understand the nutrition, maternal and child health, and family planning policy implications of these findings and ensure that the impact of menstrual blood loss on anemia is prioritized and addressed in detail in international anemia guidelines that currently include very little information about this potential treatment or prevention strategy [54]. Hormonal contraception should be considered one important part of a multipronged approach to preventing and treating anemia in women of reproductive age who are using or considering contraception, especially in contexts where significant improvements are needed to address other causes of anemia, such as food quality and scarcity, poverty, and infectious diseases. However, the potential for anemia treatment or prevention should not be a reason that women with anemia are disproportionately steered towards hormonal contraception. Instead, the anemia reducing benefits of contraception are a promising piece of wider efforts to address anemia, and this information provides women with a more comprehensive picture of non-contraceptive benefits to inform their contraceptive decision making as they exercise their reproductive autonomy.

Supporting information

Acknowledgments

We acknowledge the DHS and individual countries for making the DHS available, and we thank the women who participated in these surveys. Thank you to Shiza Farid for providing examples of Stata loop code, and to Olivia Chen, Qichen Wang, Jordyn Brown, Clement Achaempong, Caroline Thompson, and Brian Pence for providing feedback on the analysis plan. We appreciate the technical advice from Tina Sanghvi on anemia and iron deficiency anemia. We thank Laneta Dorflinger, Michelle Hindin, and Mario Chen for providing feedback on the manuscript.