Scope and quality of Cochrane reviews of nutrition interventions: a cross-sectional study

Abstract

Background: All countries face significant challenges from complex manifestations of malnutrition, which affects one in three people globally. Systematic reviews provide ready-to-use syntheses of quality-appraised evidence to inform decision-making for actions. To enhance the utility and quality of future Cochrane nutrition evidence, we described the scope and quality of all nutrition systematic reviews in the Cochrane Database of Systematic Reviews (CDSR).

Methods: We screened all active CDSR records (31 July 2015) to identify reviews and protocols using pre-specified eligibility criteria and definitions. Duplicate, independent data extraction included criteria for inclusion of studies in completed reviews (PICOS). We assessed methodological quality (AMSTAR), use of GRADE, mapped reviews against 2013 Global Burden of Disease data, and categorised the paradigm (medical, lifestyle and socio-ecological) of the review question. We analysed our results using descriptive statistics.

Results: We screened 8484 records, and included 470 (8%) completed reviews (in 45 Cochrane Review Groups (CRGs)) and 169 (7%) protocols (in 41 CRGs) published by 47 of 53 CRGs with reviews. Most completed reviews were produced by the Pregnancy and Childbirth (n = 73), Neonatal (n = 64), Metabolic and Endocrine Disorders (n = 33), Developmental, Psychosocial and Learning Problems (n = 26), Kidney and Transplant (n = 18) and Heart (n = 18) CRGs. Only 27% (n = 129) of reviews had searches for new studies in 2013 or thereafter. Supplementation/supplement interventions were most common (50%; n = 235; majority with micronutrients; 73%, n = 173), followed by food interventions (20%; n = 95). All reviews included randomised controlled trials; about 5% included other designs; 25% used GRADE; the median AMSTAR score was 9 (interquartile range: 7 to 10), 51% were high (AMSTAR 9-11) and 49% moderate (AMSTAR 5-8) quality. More than 80% framed questions using a medical paradigm. For top causes of years-of-life-lost, most reviews addressed preterm birth, diabetes and ischaemic heart disease; for leading risk factors for disability-adjusted-life-years, most targeted childhood undernutrition and high body mass index.

Conclusions: Nutrition reviews comprised 8% of active CDSR records, were widely distributed across nearly all CRGs and reflected the double nutrition burden. This analysis presents a comprehensive description of the scope and quality of Cochrane nutrition reviews, and identifies gaps for future activities to support actions to address the nutrition burden, in line with the current nutrition agenda and impetus.

Keywords: Cochrane; Diet; Food; Nutrition; Systematic review.

Fig. 1

Graphic representation of the three paradigms (medical, lifestyle and socio-ecological) that conceptualize how the relationship between food and health is viewed and how the causes of nutrition problems are framed [12], along with the broad categories of nutrition interventions (nutrition-specific, nutrition-sensitive) and the enabling environment for nutrition improvement to support these interventions [2, 13]

Fig. 2

Flowchart illustrating the search results and selection process of nutrition reviews and protocols in the Cochrane Database of Systematic Reviews, and methodological quality assessment of systematic reviews [19]

Fig. 3

a) Proportions of nutrition reviews and protocols included of the total reviews and protocols screened in Cochrane Review Groups where 6% or more of all records screened per Group were included (numbers next to bars indicate the percentages of both reviews and protocols); b) Number of nutrition reviews and protocols published in the Cochrane Database per year; c) Percentage of nutrition reviews (n = 470) assessed as up-to-date by year

Fig. 4

a The percentage of completed nutrition reviews (n = 470) that examined the various categories of nutrition interventions b) The numbers of completed nutrition reviews that included the various categories of outcomes

Fig. 5

Percentages of reviews judged as meeting the recommended criterion specified for each of the 11 AMSTAR domains (n = 425 reviews methodological quality assessed using AMSTAR tool)