Using the Ethiopian 2016 DHS, we investigated whether risk of injectable contraceptive discontinuation due to side-effects (DSE) or other reasons (DOR) was associated with different physiological and sociocultural factors. We found that DSE, but not DOR, was associated with physiological factors likely indicative of low iron status, such as being anaemic, not having received iron supplements and having given birth in the last two years prior to injectable initiation. By contrast, increased risk of DOR was driven not by physiological factors, but instead only by sociocultural characteristics such as lower wealth, not being in a relationship and being of Protestant or Catholic religion. The lack of physiological risk factors for DOR suggests that risk factors for DSE are indeed specific to side-effects driven discontinuation, and not just relevant to discontinuation generally, supporting a potential biological aetiology of DSE. These results have implications for how family planning counselling is carried out and for discourses around the reasons underpinning side-effects.
Anaemia doubles the risk of DSE but has no significant effect on the risk of DOR. This result cannot be explained by the impact of contraceptive use on anaemia: whilst injectable use was associated with a reduced risk of anaemia, and thus women who continue injectable use longer would likely be less anaemic, anaemia was not found to be a risk factor for DOR, despite women who DOR having average shorter episode lengths compared to women who DSE (6.28 vs 7.01 months respectively). This suggests anaemia is a significant risk factor specifically for DSE, rather than discontinuation generally. Previous studies which have found only sociocultural factors to be associated with risk of DSE, may have been confounded by their lack of inclusion of physiological factors, such as anaemia, which are associated with sociocultural factors such as wealth.
Understanding the relationship between anaemia and side-effects in this study is necessarily imprecise due to limitations associated with DHS data. Firstly, the DHS does not collect data on experience of contraceptive side-effects directly. However, it is reasonable to assume that women who DSE experienced more negative side-effects than women who continue use or DOR. Secondly, the contraceptive calendar is measured retrospectively, making it subject to recall bias and post-rationalization. We minimized this bias by restricting the sample to injectable episodes initiated within the last two years. Third, the analysis relies on the assumption that anaemia at interview is a proxy for anaemia at contraceptive initiation. Previous studies19 have indeed shown that iron trajectories are fairly stable over time, and the current analysis shows that, if anything, haemoglobin levels increase slightly with injectable use. Therefore, it is reasonable to assume that anaemia levels were either similar or more severe at the time of injectable initiation.
Our other findings also support the role of iron levels prior to contraceptive initiation in shaping the risk of DSE. Firstly, having taken iron supplementation with last pregnancy is protective from DSE. Secondly, women who initiated injectable use less than two years after their last birth had higher risk of DSE. High levels of iron are lost during pregnancy, leading to depleted iron stores after birth, particularly so for short interbirth intervals20 and for women who did not take iron supplements. Women who have more depleted iron stores after birth may thus experience a greater burden of side-effects. One exception is that women who had given birth 0–2 months before initiating the injectable also had lower risk of DSE than those who had given birth 3–24 months before. This may be due to increased motivation to avoid pregnancy very soon after giving birth. As a proportion of our sample had not given birth in the last 5 years and thus there was not data available for whether they had taken iron supplements with their last pregnancy, we also ran a sensitivity analysis only including women with births in the last 5 years. The effect of iron supplementation remained consistent when analysing these women only. It is also interesting to note that the effects of iron supplementation and time since birth before initiation remain after adjusting for haemoglobin levels, suggesting that depleted iron stores, independent of haemoglobin levels, may themselves also contribute to side-effects. Multiple measures of iron status are thus needed to understand what may be driving side-effects.
Taken together, our findings point towards iron-deficiency anaemia specifically as a driver of DSE. Anaemia may be caused by nutritional deficiencies, pregnancy depletion, or blood loss (iron deficiency anaemia (IDA)) or immune stress causing iron withholding (anaemia of infection (AI)). Our finding that iron supplementation is protective of DSE suggests that iron-deficiency anaemia is particularly important for DSE. If the relationship between anaemia and DSE was driven instead by anaemia of infection, which is characterized by iron withholding and decreased iron absorption, we would not expect to see an effect of iron supplementation on DSE. Future analyses should include biomarkers of immune stress and additional measures of iron status (e.g. ferritin) to understand if anaemia is caused by iron deficiency (low ferritin/low haemoglobin) or iron withholding (normal-high ferritin/low haemoglobin).
Why low iron increases the risk of DSE is unclear. One possibility is that side-effects may present an additional burden on top of the symptoms of anaemia. Another is that there may be a relationship between iron status and reproductive hormones levels. Women with lower natural levels of reproductive hormones have been suggested to experience the most side-effects from typically high contraceptive doses4, but whether low iron is associated with lower reproductive hormone levels remains to be further investigated20. Studying the response of anaemic and non-anaemic women to different doses of contraception, such as the lower dose injectable, Sayana Press, may present a promising avenue for understanding the interaction of dose, hormone levels, iron and side-effects.
The association between iron status and DSE has potentially important implications for both family planning programs and research. Family planning programs might benefit from providing an integrated service package addressing anaemia as well as supplying hormonal contraception, whereby counselling prior to prescription of contraceptives could add a haemoglobin test. If anaemia is detected, the cause of low haemoglobin could be investigated and treated, either with iron supplementation, in the case of IDA, or relevant antimalarials or antibiotics, in the case of AI. Additionally, the results of this study reveal two gaps in current discontinuation research. Firstly, as discontinuation is a heterogeneous phenomenon which is not currently measured in a nuanced way2, studies could benefit from collecting prospective longitudinal data, measuring multiple discontinuation reasons per event, side-effect experiences, and fertility desire measures, which encapsulate fertility planning and motivations within a woman’s context. Secondly, further research on discontinuation and contraceptive development should include additional physiological risk factors, such as iron levels, when examining risk of side-effects. Understanding the causal relationships between physiological factors and side-effects not only may help reduce discontinuation, but also help stimulate research to improve contraceptive technology, doses and service quality for women who continue use, as well as those who discontinue. This utilises a rights-based, quality of care approach21 by focusing concurrently on quality of life during use and helping women achieve their fertility goals through either continuation or timely discontinuation of their method of contraception.