Seasonality and Day-to-Day Variability of Dietary Diversity: Longitudinal Study of Pregnant Women Enrolled in a Randomized Controlled Efficacy Trial in Rural Burkina Faso

Abstract

Background: Panel data indicate that nonpregnant women's dietary diversity fluctuates across climatic seasons in low- and middle-income countries. The natural day-to-day variability in food group consumption during gestation is unknown.

Objectives: A longitudinal study was conducted among pregnant women enrolled in the Micronutriments pour la Santé de la Mère et de l'Enfant study 3 randomized controlled efficacy trial [i.e., daily fortified balanced energy-protein supplement and an iron-folic acid (IFA) tablet compared with an IFA tablet only] to investigate the number of 24-hour recalls required to estimate usual prenatal food group (FG) diversity and the seasonality of pregnant women's dietary diversity in Houndé, Burkina Faso.

Methods: FG consumption was assessed twice weekly by qualitative, list-based, 24-hour recalls among 1757 pregnant women (892 control, 865 intervention). The number of days needed to estimate a woman's usual prenatal 10-point FG diversity score was calculated using the within-subject coefficient of variation. Regression models, including truncated Fourier series, were fitted to assess seasonal variations in the FG diversity score and the probability of reaching Minimum Dietary Diversity for Women (MDD-W; i.e., ≥5 FGs).

Results: The monthly mean FG scores (<5 FGs) and MDD-W prevalence (<45%) were low. Five list-based recalls allowed observed FG diversity to lie within 15% of the true mean in 90% of the estimations (mean ± SD, 40.4 ± 20.7 recalls per woman). Both the FG diversity score and prevalence achieving MDD-W showed responsiveness to seasonal variations, with peaks at the end of the dry season (i.e., April or May) and troughs in the rainy season (i.e., August).

Conclusions: Five list-based recalls are sufficient to estimate usual FG diversity during gestation, although intra-annual seasonal patterns did modestly affect the FG diversity score and MDD-W prevalence. Thus, timing of repeated dietary surveys is critical to ensure nonbiased inferences of change and trends in Burkina Faso. This trial was registered at clinicaltrials.gov as NCT03533712.

Keywords: Burkina Faso; balanced energy-protein supplements; dietary diversity; food groups; list-based recall; pregnant women; seasonality.

FIGURE 1

Food group diversity score among pregnant women (n = 1757), by study month and trial arm. Y-axis ranges between a minimum of 3.5 and a maximum of 4.5 food groups. October 2019 (n = 8 women), July 2021 (n = 14), and August 2021 (n = 1) were not plotted due to the limited number of data points. All 2-sided Welch's independent-sample t-tests were nonsignificant by intervention arm (P > 0.05). Abbreviation: COVID-19, coronavirus disease 2019.

FIGURE 2

Proportion of pregnant women (n = 1757) achieving MDD-W, by study month and trial arm. The y-axis values range between a minimum of 20% and maximum of 45%. October 2019 (n = 8 women), July 2021 (n = 14), and August 2021 (n = 1) were not plotted due to the limited number of data points. All 2-sided Welch's independent-sample t-tests were nonsignificant by intervention arm (P > 0.05). Abbreviation: MDD-W, Minimum Dietary Diversity for Women.

FIGURE 3

Monthly means and seasonal variation in a 10-point food group diversity score (n = 1757 women; n = 10,955 data points) in the MISAME-III trial. The y-axis ranges between a minimum of 3.7 and maximum of 4.3 food groups. The solid lines represent the local polynomial smoothing prediction of the monthly mean with 95% CI, whereas the dashed line represents the modeled seasonal variation, with Fourier series. The food group diversity score was fitted to the first-, second-, and third-order Fourier pairs. Abbreviation: MISAME-III, Micronutriments pour la Santé de la Mère et de l'Enfant study 3.

FIGURE 4

Monthly proportion of pregnant women achieving MDD-W (n = 1757 women; n = 10,955 data points) in the MISAME-III trial. The y-axis ranges between a minimum of 29% and maximum of 42%. The solid lines represent the local polynomial smoothing prediction of the monthly mean proportion with 95% CI, whereas the dashed line represents the modeled seasonal variation, with Fourier series. The MDD-W was fitted to the first-, second-, and third-order Fourier pairs. Abbreviations: MDD-W, Minimum Dietary Diversity for Women; MISAME-III, Micronutriments pour la Santé de la Mère et de l'Enfant study 3.