The authors have declared that no competing interests exist.
Nutrition interventions may have favourable as well as unfavourable effects. The Maternal and Infant Nutrition Interventions in Matlab (MINIMat), with early prenatal food and micronutrient supplementation, reduced infant mortality and were reported to be very cost-effective. However, the multiple micronutrients (MMS) supplement was associated with an increased risk of stunted growth in infancy and early childhood. This unfavourable outcome was not included in the previous cost-effectiveness analysis. The aim of this study is to evaluate whether the MINIMat interventions remain cost-effective in view of both favourable (decreased under-five-years mortality) and unfavourable (increased stunting) outcomes.
Pregnant women in rural Bangladesh, where food insecurity still is prevalent, were randomized to early (E) or usual (U) invitation to be given food supplementation and daily doses of 30 mg, or 60 mg iron with 400 μg of folic acid, or MMS with 15 micronutrients including 30 mg iron and 400 μg of folic acid. E reduced stunting at 4.5 years compared with U, MMS increased stunting at 4.5 years compared with Fe60, while the combination EMMS reduced infant mortality compared with UFe60. The outcome measure used was disability adjusted life years (DALYs), a measure of overall disease burden that combines years of life lost due to premature mortality (under five-year mortality) and years lived with disability (stunting). Incremental cost effectiveness ratios were calculated using cost data from already published studies.
By incrementing UFe60 (standard practice) to EMMS, one DALY could be averted at a cost of US$24.
When both favourable and unfavourable outcomes were included in the analysis, early prenatal food and multiple micronutrient interventions remained highly cost effective and seem to be meaningful from a public health perspective.
Impaired antenatal and early life nutrition increases the risk of neonatal deaths and stunting. According to global estimates, maternal and child undernutrition causes 3.1 million child deaths annually [
Reports from studies aimed at studying the efficacy and safety of multiple micronutrient (MMS) supplementation during pregnancy has varied; from encouraging results such as reduced prevalence of low birth weight[
In the randomized MINIMat trial (Maternal and Infant Nutrition Interventions, Matlab,
The MINIMat interventions have been reported to be cost-effective in respect of the primary outcome of reduced infant mortality[
The data originate from the MINIMat trial (Maternal and Infant Nutrition Interventions in Matlab), a factorial randomized trial carried out in Matlab, Bangladesh, a rural sub-district 57 km south-east of the capital Dhaka where food insecurity is still prevalent. At the time when the trial was implemented, the under-five mortality rate in Bangladesh was 74 per 1000 live births [
Written informed consent was obtained from all parents of participating children (separately for the original trial and the two follow-ups). The Ethical Review Committee at the International Centre for Diarrhoeal Disease Research, Bangladesh and the Regional Ethical Review Board at Uppsala University, Sweden approved the study (separately for the original trial and the two follow-ups).
Cost data were retrieved from a recent study by Shaheen et al.[
The effect measure is presented as disability adjusted life years (DALYs) averted. DALYs are a measure of overall disease burden that combines years of life lost due to premature mortality (YLLs) and the time spent in an impaired health state, measured as years lived with disability (YLDs). One DALY lost is to be interpreted as one lost year of healthy life. Consequently, one DALY averted is equivalent to the gain of one year of healthy life. Calculating the cost per DALY averted facilitates comparison between different health interventions as well as enables the evaluation of an intervention’s effects on both premature death and long-term disability. The term disability here refers to loss of health due to stunting.
DALYs are typically attributed to a specific disease, health state or age group. As the MINIMat interventions had an effect on both under-five deaths and stunting, DALYs were equal to YLL due to all-cause premature mortality plus YLD due to stunting. DALYs were calculated based on strictly empirical data from the trial for up to five years (DALY5s) and from a lifetime perspective based on secondary data (DALYs). To calculate YLL for DALY5s, we used the exact mortality age of the children who died and subtracted the age at death from 5 years. YLDs were calculated as the sum of months the children had been stunted from birth to 5 years (i.e., they had a height-for-age Z-score below -2 from the WHO reference median) times a disability weight[
To calculate DALYs from a lifetime perspective, the number of YLL and YLDs were created using the following assumptions. As most child deaths occurred in early infancy, the life expectancy at birth (LE) was used for YLLs. The remaining life expectancy (RLE) at 5 years was used for YLDs. LE and RLE at 5 years in the Matlab area were based on the demographic surveillance data from the year 2003 and 2008, estimated to 69.3 and 69.7 for girls and 67.8 and 66.9 years for boys[
Children were assumed to grow linearly. Height data were linearly interpolated if height measurements were missing in the age interval between birth and 2 years of age. From 2 to 4.5 years, height-for-age z-scores were imputed every 3 months.
Duration of disability: Duration of stunting, for up to 5 years, was obtained from observed data. Children who were stunted at 4.5 years were assumed to be stunted at 5 years.
Disability weights used: 0.002 (stunting) and 0.024 (developmental delay). Stunted children were assumed to be equally affected by the disability caused by stunting throughout infancy and childhood.
Discount rate: No discounting.
Duration of disability: Duration of stunting, for up to 5 years, was obtained from observed data. From 5 to 10 years, the average proportion of recovery and incidence of stunting in each intervention group between 4.5 and 10 were used. Children who were stunted at 4.5 years were assumed to be stunted at 5 years, and children who were stunted at 10 years were assumed to continue being stunted throughout life.
Disability weights used: 0.002 (stunting) and 0.024 (developmental delay). Stunted children were assumed to be equally affected by the disability caused by stunting throughout infancy, childhood and adulthood.
Discount rate: 3%.
Life expectancy: LE at birth for YLL, and RLE at 5 years for YLD. Discounted LE and RLE for girls; 28.66 and 29.03 years. Discounted LE and RLE for boys; 28.16 and 28.84
To obtain incremental cost effectiveness ratios (ICERs) for DALYs averted, we calculated the costs for supplementing 1 pregnant woman following the regimes in each intervention arm. These were then sorted according to ascending costs and dominating alternatives were excluded (i.e., alternatives with both a higher cost and a higher DALY estimate). This procedure resulted in three remaining intervention arms. Lastly, we calculated the increment in costs and DALYs across these three arms and divided the cost differences by the differences between DALYs, which gave the ICERs.
There were 4,436 women enrolled into the MINIMat trial, of whom 845 were lost to follow-up before delivery, mainly due to fetal loss, outmigration or because they withdrew their consent. Of the 3,625 live born children 2,851 had anthropometry at 4.5 years (
E = Early invitation food supplementation, U = Usual invitation food supplementation; 30F = 30 mg iron and 400 μg of folic acid; 60F = 60 mg iron and 400 μg of folic acid; MMS = multiple micronutrients, 15 micronutrients including 30 mg iron and 400 μg of folic acid.
CHARACTERISTICS | n | % | |
---|---|---|---|
Mothers | |||
Age | |||
<20 | 916/3267 | 16.0 | |
20–29 | 1862/3267 | 57.0 | |
≥30 | 882/3267 | 27.0 | |
BMI at 8 weeks | |||
<18.5 | 916/3255 | 28.1 | |
≥18.5 | 2339/3255 | 71.9 | |
Education | |||
No Education | 1016/3267 | 31.1 | |
Can read and write | 2251/3267 | 68.9 | |
SES category | |||
Low | 1329/3267 | 40.7 | |
Middle | 651/3267 | 19.9 | |
High | 1287/3267 | 39.4 | |
Children | |||
Sex | |||
Girl | 1605/3267 | 49.1 | |
SGA |
1928/3267 | 59.0 | |
LBW |
997/3267 | 30.5 | |
Preterm |
258/3267 | 7.9 |
Adherence levels and costs for a hypothetical highest-cost delivery scenario are presented in
Adherence to food packets | Adherence to micronutrient capsules | Cost for supplementing one woman |
|
---|---|---|---|
91 | 109 | 82.12 | |
88 | 113 | 79.47 | |
94 | 107 | 86.19 | |
61 | 117 | 55.34 | |
60 | 113 | 54.42 | |
61 | 110 | 56.73 |
Adherence to and cost of food supplements and micronutrient capsules and for supplementing one woman in the MINIMat trial, Bangladesh.
Years lost to premature mortality (YLL), years lived with stunting (YLS), years lived with disability (YLD) and disability adjusted life years (DALYs) lost, for up to 5 years and calculated from a lifetime perspective, are presented in Tables
YLL |
YLS |
YLD |
DALY5 |
|||
---|---|---|---|---|---|---|
Mean (sd) | Mean (sd) | Mean (sd) |
Mean (sd) 0.024 | Mean (sd) |
Mean (sd) 0.024 | |
0.2272(1.00) | 1.54 (1.82) | 0.0031 (0.0036) | 0.0369(0.04) | 0.2303(1.00) | 0.2641(0.99) | |
0.2216(1.00) | 1.34(1.73) | 0.0027 (0.0035) | 0.0322(0.04) | 0.2243(1.00) | 0.2539(0.99) | |
0.0800(0.61) | 1.56(1.83) | 0.0031 (0.0037) | 0.0375(0.04) | 0.0831(0.61) | 0.1174(0.61) | |
0.1690(0.87) | 1.59(1.86) | 0.0032 (0.0037) | 0.0383(0.05) | 0.1720(0.87) | 0.2069(0.86) | |
0.2510(1.05) | 1.59(1.90) | 0.0033 (0.0038) | 0.0383(0.05) | 0.2542(1.05) | 0.2893(1.04) | |
0.2737(1.10) | 1.68(1.93) | 0.0034 (0.0039) | 0.0404(0.05) | 0.2770(1.10) | 0.3140(1.09) |
YLL |
YLS |
YLD |
DALY |
|||
---|---|---|---|---|---|---|
Mean (sd) | Mean | Mean |
Mean |
Mean |
Mean |
|
1.7162(7.37) | 11.7417 | 0.0235 | 0.2818 | 1.7397 | 1.9980 | |
1.7054(7.34) | 10.1577 | 0.0203 | 0.2438 | 1.7257 | 1.9492 | |
0.6177(4.49) | 11.4962 | 0.0230 | 0.2759 | 0.6406 | 0.8936 | |
1.2825(6.42) | 9.7237 | 0.0194 | 0.2334 | 1.3019 | 1.5159 | |
1.9173(7.77) | 11.8206 | 0.0236 | 0.2837 | 1.9409 | 2.2010 | |
2.0938(8.09) | 12.9199 | 0.02584 | 0.3101 | 2.1196 | 2.4039 |
Incremental costs, incremental DALYs and incremental cost-effectiveness ratios (ICER) are presented in
Cost/woman | DALYs/child |
Comparison | Incremental cost | Incremental DALY5s | ICER |
|
---|---|---|---|---|---|---|
54.4157 | 0.2542 | |||||
55.3364 | 0.1720 | |||||
56.7306 | 0.2770 | |||||
79.4645 | 0.2243 | |||||
82.1221 | 0.2303 | |||||
86.1935 | 0.0831 | |||||
54.4157 | 0.2542 | U60fe-U30Fe | 0.92 | 0.08 | 11.5 | |
55.3364 | 0.1720 | U30Fe-EMMS | 30.86 | 0.09 | 342.9 | |
86.1935 | 0.0831 | 60Fe-EMMS | 31.78 | 0.17 | 186.9 |
Incremental cost-effectiveness ratios for cost per DALY averted for the different MINIMat prenatal food and micronutrient supplementation arms.
1 Disability weight 0.002
2 Incremental cost/Incremental DALYs
Cost/woman | DALYs /child |
Comparison | Incremental cost | Incremental DALYs | ICER |
|
---|---|---|---|---|---|---|
54.4157 | 1.9409 | |||||
55.3364 | 1.3019 | |||||
56.7306 | 2.1196 | |||||
79.4645 | 1.7257 | |||||
82.1221 | 1.7397 | |||||
86.1935 | 0.6406 | |||||
54.4157 | 1.9409 | 60fe-U30Fe | 0.92 | 0.64 | 1.4 | |
55.3364 | 1.3019 | 30Fe-EMMS | 30.86 | 0.66 | 46.8 | |
86.1935 | 0.6406 | 60Fe-EMMS | 31.78 | 1.30 | 24.4 |
Incremental cost-effectiveness ratios for cost per DALY averted for the different MINIMat prenatal food and micronutrient supplementation arms.
1 Disability weight 0.002
2 Incremental cost/Incremental DALYs
When favourable (decreased under-five year mortality) and unfavourable (increased stunting), outcomes of the MINIMat trial were included in the analysis the implementation of EMMS supplementation, when compared to U60Fe, remained highly cost-effective. The incremental cost effectiveness ratio per 5 years and lifetime DALY averted of US$187 and US$24 both fall well below Bangladesh’s per-capita gross national income (US$1190 in 2015) and compare favourably to corresponding interventions evaluated elsewhere in South Asia[
The main aim of this study was to evaluate the cost of switching from the routine prenatal supplementation of 60 mg iron and folate plus the usual timing (week 20) of daily food supplements (U60Fe) to the potentially superior multiple micronutrients plus an early invitation (week 9) to food supplementation (EMMS) including both favourable and unfavourable outcomes, but all the intervention arms were included in the analysis. It turned out that switching from U60Fe to U30Fe had the lowest ICER, at US$1.4 per DALY averted. As the cost for the two interventions were similar, the ICER for switching from U60Fe to U30Fe was mainly driven by the slightly lower DALYs in the U30Fe group due to primarily lower YLL but also lower YLD estimates. The fact that U30Fe would have a lower YLL estimate was somewhat expected as the U30Fe group had a lower mortality hazard ratio in the survival analysis by Persson et al. [
Although switching from U60fe to U30Fe had the lowest ICER, EMMS had the lowest DALY estimate. Switching from U60Fe to EMMS would avert twice as many DALYs as switching to U30Fe. As the ICER of US$24 can be considered highly cost-effective and affordable in a low-resource setting, switching from U60Fe to EMMS is viable both from a public health and economic perspective. In addition, as mentioned above, when modelled, the estimates that generated the lower YLLs and YLDs in the U30Fe intervention group were not statistical significant[
The lower DALYs in the EMMS group was driven primarily by the lower YLLs as the difference in YLD between U60Fe and EMMS was minor. Meta-analyses of prenatal multiple micronutrient supplementation trials have reported limited increases in birth weight in comparison with iron-folic acid supplementation alone[
What makes the MINIMat interventions different to the trials included in the above-mentioned studies is that the multiple micronutrient supplementation was combined with food supplementation early in pregnancy and the reduction in mortality was only seen as a combined effect of early food supplementation and MMS. MMS given without food supplementation in
The MINIMat trial was a community-based trial, conducted in an area with a well-established health and demographic surveillance system and an excellent research infrastructure that fulfils the prerequisites for obtaining high-quality data. The randomised design minimised the risk of potential confounding and double masking of the micronutrient intervention reduced the risk of reporting or observation bias.
In the area where the trial was conducted, women are still frequently exposed to food insecurity and enter pregnancy deficient in several micronutrients. This makes the ICERs estimates valid to other rural areas in Bangladesh and to similar settings in other countries that struggle with problems of widespread maternal malnutrition. The cost data retrieved from the study by Shaheen et al. were collected within the same time-frame and in a situation similar to that of the MINIMat trial, which is why we believe they reasonably represent costs associated with the MINIMat interventions. Shaheen et al. presented three different delivery modes, resulting in three cost scenarios, NGO-run clinics, governmental-run clinics and a highest-cost delivery mode where the highest costs for different items from NGO- and government-run clinics were combined. As we used the highest cost scenario in our analysis, our generated ICERs might be somewhat overestimated.
Most nutrition interventions have multiple objectives and outcomes. In this study, we included two outcomes; premature mortality, and disability caused by stunting, aware that this do not represent all potential benefits or harms of the intervention[
When evaluating the cost-effectiveness of multiple outcomes, the issue of assigning values to the different outcomes becomes essential. The disability weight attributed to stunting by the global burden of disease study, 0.002, is small. In order to avoid underestimating the long-term effects of stunting, we chose to include the larger disability weight attributed to development delay due to malnutrition, although not all stunted children will suffer from cognitive impairment. When using the higher disability weight, the ICER only increased from US$24 to US$26 as differences in YLS between the intervention groups were small. The differences in YLS and YLL between the intervention group with the lowest and highest estimate were 3.2 YLS (UMMS-U30Fe) and 1.47 YLL (UMMS-EMMS) respectively. Thus, for the differences in YLS to change the ranking of the highest and lowest DALY estimate, the disability weight would have to be at least as high as 0.4 (1.47/3.2), equal to multiple sclerosis or bipolar disorder[
Another key assumption associated with the impact of stunting was that all children who were stunted at 10 years continued to be stunted throughout life, based on the fact that adult stature is strongly associated with pre-pubertal height [
The estimate of US$187 per DALY averted for up to 5 years was based on strictly observed data from the trial and was not subject to any assumptions except the disability weight for stunting. However, by not considering the number of life years lost to premature mortality and disability beyond the age of 5 years, the intervention is at a disadvantage when compared with alternatives, which are usually analysed from a lifetime perspective.
Including stunting as an outcome did not alter the conclusion that the MINIMat interventions of early prenatal food and MMS supplementation were highly cost-effective in a population where maternal undernutrition is still common. The ICER of US$24 for switching from invitation to food supplementation at the usual time in pregnancy and iron-folic acid supplementation to an early initiation of food supplementation combined with MMS can be considered affordable in a low-resource setting and viable from both a public health and economic perspective. These results can hopefully inspire others to consider multiple outcomes of nutritional interventions in economic evaluations and help to support decision-makers in the prioritizing of financing for nutrition interventions targeting pregnant women’s health.
(DOCX)
(SAV)
We thank the participants and their families in Matlab for their continuing involvement in the MINIMat trial, and the field-team members and data management staff for their excellent work.