Hard facts and misfits: essential ingredients of public health nutrition research

Abstract

Policy decisions and the practice of public health nutrition need to be based on solid evidence, developed through rigorous research studies where objective measures are used and results that run counter to dogma are not dismissed but investigated. In recent years, enhancements in study designs, and methodologies for systematic reviews and meta-analysis, have improved the evidence-base for nutrition policy and practice. However, these still rely on a full appreciation of the strengths and limitations of the measures on which conclusions are drawn and on the thorough investigation of outcomes that do not fit expectations or prevailing convictions. The importance of 'hard facts' and 'misfits' in research designed to advance knowledge and improve public health nutrition is illustrated in this paper through a selection of studies from different stages in my research career, focused on the nutritional requirements of resource-poor populations in Africa and Asia.

Keywords: Africa; Calcium; Evidence-based nutrition; Gambia; Vitamin D.

Figure 1

Effects on breast-milk calcium concentration of maternal calcium supplementation during (a) lactation and (b) the preceding pregnancy. The data are expressed as mean (SE) concentration (mg/l) in breast-milk collected serially from 60 and 125 rural Gambian mothers respectively during randomised placebo-controlled trials of calcium carbonate supplementation (lactation trial=1000 mgCa/d for 12 months; pregnancy trial =1500 mgCa/d from 20 weeks gestation to delivery)^{(22, 23)}. Dark bars = calcium supplemented group; light bars = placebo group. The numbers of women in the calcium and placebo groups respectively who participated in the sub-study of breast-milk calcium were: lactation trial 30,30; pregnancy trial = 61,64. There were no significant differences between the groups at any time.

Figure 2. Effects on size-adjusted BMC (SA-BMC) of the lumbar spine (L1-4) of maternal calcium supplementation during pregnancy in Gambian mothers.

The data are expressed as mean percentage difference (SE) relative to the placebo group at 0.5 months postpartum. The scans were obtained using dual-energy X-ray absorptiometry (DXA) as part of a randomised placebo-controlled trial of 1500 mgCa/d from 20 weeks gestation to delivery with follow-up. Measurements in the DXA sub-study were made serially on each individual during the index lactation at 0.5, 3 and 12 months⁽²⁸⁾ and approximately 5 years later at a time when the mother was neither pregnant nor lactating and at least 3 months since the end of a recent lactation period (NPNL)⁽³⁰⁾. The numbers of measurements at each timepoint for the calcium and placebo groups respectively were 0.5 months = 23, 27; 3 months = 29, 29; 12 months = 40, 39; NPNL = 31, 28. Dark bars = calcium supplemented group; light bars = placebo group. The changes over time were significantly different between the two groups: in the index lactationp for interaction = 0.05; in the follow-up study p for interaction = 0.002.

Figure 3. Effects on height of calcium supplementation in pre-pubertal Gambian children.

The data are expressed as the mean (SE) difference in height between the calcium and placebo groups by year of the study obtained from regression models after adjustment for height and age at Y1. The measurements were made serially during adolescence in 160 children (80 boys, 80 girls) who had participated in a randomised placebo-controlled trial of 1000 mgCa/d, 5 days week for 12 months at the age of 8-12 years⁽³⁵⁾. The average age (years) of the boys at each measurement timepoint was approximately Y1=10.5; Y2=11.5; Y3=12.5; Y4=13.5; Y6=15.5; Y8=17.5; Y10=19.4; Y12=21.5; Y14=23.5. The average age of the girls was approximately 0.5 years younger at each timepoint. The supplementation was commenced after the Y1 measurement (baseline) and ceased after the Y2 measurement. The numbers of boys in the calcium and placebo groups respectively were: Y1-6=40, 40; Y8=39,39; Y10=37,39; Y12=34,30; Y14=29,25. The numbers for the girls were Y1-4=40, 40; Y6=38,39; Y8=39,40; Y10=29,25; Y12=33,30; Y14=25,29. On the graph, XY = baseline for boys and girls respectively; significance of difference between the groups in boys *p=0.04; **p=0.01; ***p=0.002. There were no significant differences between the groups in girls at any time.