Article Text

Download PDFPDF

Association between type of contraceptive use and haemoglobin status among women of reproductive age in 24 sub-Saharan Africa countries
  1. Samson Gebremedhin,
  2. Anteneh Asefa
  1. School of Public Health, Hawassa University, Hawassa, Ethiopia
  1. Correspondence to Dr Samson Gebremedhin, School of Public Health, Hawassa University, Awasa, Ethiopia; samsongmgs{at}


Background Little is known whether contraceptive methods differentially modify women’s risk of anaemia or not. We compared the haemoglobin status of women currently using different fertility regulation methods in sub-Saharan Africa (SSA) where anaemia is a major concern.

Methods We conducted the study based on the secondary data of 24 nationally representative demographic and health surveys carried out recently in SSA. The data of 105 532 women were included in the analysis. In the original surveys, respondents were selected using multistage sampling techniques and haemoglobin was determined using the HemoCue analyser. The association between method of contraception and anaemia status was determined via a mixed-effects logistic regression model adjusted for potential confounders. The outputs are presented using adjusted odds ratio (AOR) with 95% CI.

Results The mean (±SD deviation) haemoglobin was 12.3 (±1.7) g/dl and 36.7% of the women had anaemia. Current use of modern contraceptives, as compared with non-use, was associated with a 25% reduction (AOR=0.75 (95% CI: 0.73 to 0.78)) in the odds of anaemia. Comparison among individual modern methods showed, as compared with current barrier methods users, use of injectables (AOR=0.62 (95% CI: 0.57 to 0.67)), oral contraceptive pills (OCP) (AOR=0.62 (95% CI: 0.57 to 0.66)) and implants (AOR=0.63 (95% CI: 0.58 to 0.70)) were significantly associated with reduced odds of anaemia. With reference to women with less than 12 months of use, the odds were significantly reduced by about a quarter among women with more than 12 months of OCP or injectables use.

Conclusion The use of hormonal contraceptives is associated with lower odds of anaemia.

  • anemia
  • contraceptives
  • Sub-saharan Africa
  • demographic and health survey

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Key messages

  • Use of hormonal contraceptives (injectables, pills and implants) is associated with reduced risk of anaemia.

  • Prevalence of anaemia also declines with increasing duration of use of injectables and OCP.

  • The use of non-hormonal contraceptives and intrauterine device is not associated with women’s haemoglobin status.


Over the past few decades the world had made considerable progress in improving the contraceptive prevalence rate (CPR) and reducing unmet need for family planning. Between 1990 and 2015, the CPR – proportion of women of reproductive age who use, or whose sexual partner use, at least one method of contraception – increased from 55% to 64%. In the same period, unmet need declined from 15% to 12% among women in a union. However, globally there are more than 200 million women who want to avoid or delay pregnancy but are not using a contraceptive. The problem is especially gloomy in sub-Saharan Africa (SSA) where CPR remains below 40% and unmet need exceeds 20%.1 2

Family planning has multidimensional implications for maternal and child health. According to a study, in 2008 contraceptive use averted nearly 272 000 maternal deaths and without contraceptives, the estimated deaths would have been increased by 1.8-fold.3 Another estimate suggested, if the existing unmet need for contraception is fully met, each year 76 000 maternal deaths could be averted.4 An analysis of several demographic and health surveys (DHS) concluded that the infant mortality rate is reduced by 50% in countries having CPR exceeding 30% than settings having CPR below 10%.5 The consequences of short birth intervals on child mortality have also been documented by several studies conducted in the developing world.6–8

Anaemia is a problem with public health significance both in developed and developing countries. It is also an important cause of ill-health in women.9 10 According to the estimate of the WHO, in 2011 2.7 billion people, equivalent to 43% of the world’s population, were affected by anaemia. Preschool children, pregnant women and women of reproductive age are disproportionally affected. Globally, 38% of pregnant women and 29% of women generally, and in the SSA region 46% of pregnant women and 39% of women generally are anaemic.10

Contraceptive methods can affect women’s haemoglobin status in various ways.11 12 Generally, limiting or spacing births through any fertility regulation methods is likely to prevent maternal depletion syndrome and anaemia.12–14 However, individual contraceptives may differentially modify iron status and thus the risk of anaemia.12 Intuitively, the use of oral contraceptive pills (OCP) that include iron supplements prevent iron deficiency anaemia.15 16 Changes in menstrual flow – amenorrhea, irregular, lighter or heavy bleeding – associated with hormonal and intrauterine device (IUD) methods may also alter the risk of anaemia.11 12 Further, the increase in menstrual flow inducted by the copper bearing intrauterine device (Cu-IUD) can negatively affect women’s haemoglobin status.12 17 Yet, very few studies so far investigated the contraception/anaemia relationship and empirical evidence is limited. In this study we compared the prevalence of anaemia (haemoglobin level below 12 g/dL) among women using different contraception methods in SSA where anaemia is a major public health problem.


Study design and eligibility criteria

This study was conducted based on the secondary data of 24 nationally representative DHS carried out in SSA since 2010. All DHS conducted in the subcontinent that measured haemoglobin in women were eligible for the analysis. When multiple national surveys were available, the most recent survey was used. All women 15 to 49 years of age enrolled in the surveys and those selected for haemoglobin measurement were eligible for the analysis. However, pregnant women, women in the postpartum amenorrhea period, menopausal women and those with missing information about contraceptive use were excluded (figure 1).

Figure 1

Flow chart of the study.

Sample size

The data of 105 532 women were included in the analysis. As the study was conducted based on large-scale secondary data priory sample size determination has not been made. The unweighted sample size ranged from 2524 in Lesotho to 9010 in Ethiopia, wherea, the weighted sample size varied from 384 in Gambia to 720 399 in Ethiopia (online supplementary data 1).

Supplemental material

Sample procedure of demographic and health surveys

In general, DHS employs a two-stage cluster sampling procedure designed to provide a representative sample at national, place of residence (urban-rural) and sub-national region/states levels. Initially, predetermined numbers of enumeration areas (EA) were drawn using a probability proportional to size sampling approach from a complete list of EAs defined in the recent population census. Then in every selected EA an exhaustive listing of households was made and 25 to 30 households were selected using a systematic random sampling approach (ie, probability sampling approach based on a random starting point and a fixed periodic interval). In the selected households, enumeration of the entire members was made and all women of reproductive age were enrolled in the survey. However, haemoglobin was only determined for a random sub-sample of women.18

Data collection in demographic and health surveys

In the original surveys data were collected using standardised and pretested questionnaires prepared in the major local languages of the countries. Trained interviewers, supervisors and field editors collected the data. The surveys collected information about contraceptive use and type and duration of use of current fertility regulation methods in a consistent manner. Haemoglobin levels were determined from capillary blood using a portable HemoCue analyser and later adjustment was made for altitude and cigarette smoking status following a standard approach.19 Ultimately, based on level of haemoglobin concentration, anaemia status was classified as non-anaemic (above 12 g/dL) or anaemic (below 12 g/dL). Height and weight were measured using calibrated scales and body mass index (BMI) was computed using a standard formula.

Patient and public involvement

As the study was conducted based on secondary data, contraceptive users and the public were not involved in the design of the study. Clients were not engaged in the development of the research question.

Data analysis

The datasets of all the 24 surveys were separately downloaded from the DHS program database20 and merged into one mega file. Data analysis was made using STATA version 11 software (StataCorp, United States). In order to accommodate for the complex sampling design of the surveys, weighted data analysis was used. Data weights were estimated based on sampling weights designed to ensure representativeness at country level and post-stratification weight developed based on the 2015 population size of the countries. The post-stratification weight was intended to avoid over- or underrepresentation of the countries in the dataset. Ultimately, linearisation of weights was also made to balance the total weighted and unweighted sample sizes.

Frequency distribution, measures of central tendency and dispersion were employed to describe the data. The mean haemoglobin was compared across users of different contraceptive methods using one-way analysis of variance (ANOVA) with the Tukey post hoc test.

The association between contraceptive method and anaemia status was assessed using mixed-effects bivariable and multivariable logistic regression analyses with random intercept for each county. A mixed effects model was preferred over a fixed effect model because it had better fit as measured by the log-likelihood test.

The multivariable model was adjusted for nine potential confounders: household wealth index, age, total children ever born, educational status, marital status, place of residence, BMI, household size and bed net use. The confounders were selected based on their theoretical relationship with the exposure and outcome variables and their availability in the DHS dataset. Sub-sample analyses among IUD, OCP, implants and injectables' users were made to assess the association between duration of use and anaemia. The outputs of the analyses are presented using crude and adjusted (AOR) OR.

The household wealth index was computed as an indicator of living standards using principal component analysis. The index was developed based on variables related to ownership of valuable household assets, size of agricultural land, type of drinking water and toilet facilities, and materials used for housing construction. Ultimately the distribution was divided into five groups.

Ethical considerations

In all of the original surveys ethical clearances were obtained from the institutional review boards (IRB) of ORC Macro and national level IRBs of all host countries. Data were collected after taking written informed consent from the respondents.


Socio-demographic characteristics

Of the 105 532 eligible participants, capillary blood specimens for haemoglobin testing were collected from 101 341 (96%) women. Reasons for the exclusion were: refusal to consent to the study (2.8%); not present at home during the survey (0.8%); and other unspecified reasons (0.5%). The mean (±SD) age of the respondents was 26.7 (±9.0) years and three-quarters (77.4%) were less than the age of 35 years. More than half (61.0%) were selected from rural areas and 30.2% were either from the poorer or poorest wealth categories. A quarter of the respondents (26.4%) had no formal education and 53.4% were married or in a union (online supplementary data 2).

Supplemental material

Fertility pattern, contraceptive use and unmet need

The median number of children ever born to the women was 1 and ranged from 0 to 18. About one-fifth (21.1%) had five or more births and 48.2% gave at least one birth in the past 5 years.

During the study, 31.8% of all women and 35.3% of women married or in a union were currently using contraceptives to space or limit births. Among women married or in a union, 20.1% had unmet need for spacing (those who desire to postpone their next birth by at least 2 years but do not currently use a contraceptive method) and 12.7% had unmet need for limiting births (those who desire no additional child but do not currently use a contraceptive method). Of all women, the most frequently used modern contraceptives were injectables (38.5%), followed by barrier methods (condoms, diaphragms, foams/jelly) (19.4%), OCP (17.8%) and implants (15.6%). Other modern methods were less frequently used (table 1).

Table 1

Fertility status and contraceptive use pattern among women of reproductive age in 24 sub-Saharan Africa countries, 2010–2017

Contraceptive use and haemoglobin status

The mean (±SD) haemoglobin was 12.3 (±1.7) g/dl and 36.7% of the women had anaemia. Classification based on severity indicated 29.1% women had mild (10.0–11.9 g/dL), 7.1% moderate (7.0–9.9 g/dL) and 0.5% severe (less than 7.0 g/dL) anaemia.

The haemoglobin level and prevalence of anaemia did not show significant variation between women who were not using any contraceptive at the time of the survey and current users of traditional methods. Nevertheless, ANOVA suggested the haemoglobin level among current users of modern contraceptive methods (12.64±1.65 g/dL) was significantly higher than the corresponding values for nonusers (12.27±1.69 g/dL) and traditional methods' users (12.20±1.55/dl) (P<0.001). In the multivariable logistic regression model, the odds of anaemia were reduced by 25% (AOR=0.75 (95% CI: 0.73 to 0.78)) among current users of modern contraceptives compared with nonusers.

Comparison of haemoglobin among current users of various modern fertility regulation methods showed that the mean (±SD) haemoglobin concentration among injectables (12.88±1.60 g/dL), OCP (12.63±1.63 g/dL) and implants (12.82±1.65 g/dL) users were significantly higher than women using barrier methods (12.19±1.68 g/dL), female sterilisation (12.20±1.71 g/dL), IUD (12.58±1.72 g/dL) and other modern methods (12.03±1.50 g/dL). The logistic regression analysis showed that the odds of anaemia were significantly reduced by almost 40% among injectables (AOR=0.62 (95% CI: 0.57 to 0.67)), OCP (AOR=0.62 (95% CI: 0.57 to 0.66)) and implants (AOR=0.63 (95% CI: 0.58 to 0.70)) users than barrier method users. Conversely, no significant reduction in anaemia was observed among users of IUD, female sterilisation and other modern methods (table 2).

Table 2

Contraceptive use and haemoglobin status among women of reproductive age in 24 sub-Saharan Africa countries, 2010–2017

Haemoglobin status and duration of use of selected modern contraceptives

Based on sub-sample analysis, the association between haemoglobin status and duration of use of four modern contraceptives – implants, OCPs, injectables and IUD – was assessed.

It was found that, as the duration of OCP use increases, anaemia prevalence declines. As compared with women with less than 12 months of OCP use, the odds of anaemia were significantly reduced by 23% (AOR=0.77 (95% CI: 0.66 to 0.90)) and 28% (AOR=0.72 (95% CI: 0.62 to 0.83)) among women with 12 to 23 months and more than 23 months of use, respectively. Similarly, with reference to women who have used injecteables for less than a year, the odds were reduced by 26% (AOR=0.74 (95% CI: 0.65 to 0.85)) and 23% (AOR=0.77 (95% CI: 0.68 to 0.88)) among women who had used the method for 12 to 23 and 24 or more months. Conversely, we found no significant association between anaemia and durations of use of implants and IUD (table 3).

Table 3

Haemoglobin status and duration of contraceptive use among women of reproductive age in 24 sub-Saharan Africa countries, 2010–17


This study based on large-scale data collected from multiple SSA countries, compared the haemoglobin status of women who were using different fertility regulation methods. The findings suggested that current users of hormonal contraceptives generally have reduced risk of anaemia. However, similar benefit has not been observed among users of other methods. Prolonged use of injectables and OCPs is also associated with a further decline in the prevalence of anaemia.

The study demonstrated that, with reference to barrier method use, OCP use is associated with a significant 38% reduction in the odds of anaemia. Further, longer duration of use was linked with a steady decline in anaemia. A study based on DHS data of 12 low- and middle-income countries concluded that the odds of anaemia among OCP users were reduced from 32% for use of at least 6 months to 44% for use of 2 or more years.15 Likewise, a longitudinal study among women on OCP detected a significant haemoglobin rise at 12 months of use.11 The observed gain in haemoglobin could be due to the considerable reduction in cyclic blood loss frequently documented among OCP uses.21 Further, the placebo iron pills provided with many OCP brands can also contribute to the anaemia reduction.

We observed that the odds of anaemia were reduced by 38% among current users of injectables rather than users of barrier methods. Continuing use of injectables for 12 or more months was also linked with a further reduction in anaemia. Diminished menstrual bleeding commonly observed among Depot Medroxyprogesterone Acetate (DMPA) – the most commonly used injectable contraceptive – users may explain this finding. Further, episodes of amenorrhea, which tends to increase with prolonged use of DMPA, may improve women’s haemoglobin status. As earlier witnessed by a study, DMPA-induced amenorrhea increases from 50% in the sixth month of use to 80% after 5 years of use.22

The study suggested that neither current use of IUD nor its duration of use is associated with women’s haemoglobin status. The existing studies witnessed a complex interrelationship between IUDs and anaemia. Hormonal IUDs may reduce risk of anaemia through inducing amenorrhea and light bleeding. For example, the Levonorgestrel-releasing Intrauterine System (LNG-IUS) induces amenorrhea in 20% of the users after 1 year of use.23 24 A Cochrane systematic review also concluded LNG-IUS can be used for the treatment of heavy menstrual bleeding.25 Conversely, Cu-IUDs may cause heavy bleeding, thus increase the risk of anaemia.26 In this study, we did not mange to differentiate hormonal IUDs and Cu-IUDs, as brand-specific information is not provided in the DHS dataset. Accordingly, what we have observed might be the average of the two different effects. However, parallel to our findings, a prospective study concluded that after 12 months of use, haemoglobin level significantly increased among the users of OCPs and DMPA, but not among users of IUD.11

We observed that the odds of anaemia were reduced by 37% among implant users rather than barrier methods’ users. Menstrual cycle changes frequently associated with implants, including lighter and fewer days of bleeding and infrequent or missing cycles, can convincingly explain the observed association.24 However, we did not observe further reduction in the odds of anaemia with continued use of implants beyond 2 years. This can possibly be due to the fact that missed periods and amenorrhoea become less and less common among implant users after the first or second years of use.24 27

Though the study included large-scale and nationally representative data from 24 countries that contribute to 55% of the total population of the SSA, the findings may not be generalisable to the sub-continent or to other settings. As the beneficial effect of contraceptive use for reducing the risk of anaemia is likely to be modified by the baseline prevalence of anaemia in the population, results cannot be extrapolated to settings with a different magnitude of anaemia. A study suggested that women with lowest baseline haemoglobin are likely to benefit more from contraceptive use.11

The typical strength of the study is the fact that large-scale and nationally representative data from several countries in SSA where anaemia is of public health significance is analysed. However, there were some limitations and possible biases to this study. As the study followed observational design, bias from residual or unobserved confounders cannot be excluded. The findings can also be affected by selection bias as there is no guarantee that the characteristics of the contraceptives’ users had been the same at baseline. In addition, the DHS dataset does not provide information about the specific brands of contraceptives used by the individuals. Hence differentiating Cu-IUDs from hormonal IUDs and iron supplement containing OCPs from others was not possible. This may result in under- or overestimation of the underlying association. Further, the number of IUD users was relatively smaller (n=856) than other contraceptive users, hence the statistical test of association between the duration of IUD use and anaemia might have been relatively underpowered. Further, the reported prevalence of anaemia in the study might have been underestimated because capillary haemoglobin tends to be higher than venous haemoglobin level.28 29


Hormonal contraceptives’ users (injectables, pills and implants) generally have reduced risk of anaemia. However, the same was not observed among users of other methods. Prevalence of anaemia also declines with the increasing duration of use of injectables and OCP.


The author acknowledges the DHS Program for granting access to the data.



  • Contributors Both of the authors were equally involved in the conception, development and analysis of the study, and the writing of the manuscript. Both authors read and approved the manuscript.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient consent Not required.

  • Ethics approval In all of the original surveys, ethical clearances were obtained from the institutional review boards (IRB) of ORC Macro and national level IRBs of all host countries.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement The DHS data of each country is freely available from the DHS Program webpage (

Linked Articles

  • Highlights from this issue
    British Medical Journal Publishing Group