Risk of bias and conflict of interest in clinical trials of infant formula




Oral session: Investigating bias (1)


Tuesday 22 October 2019 - 11:00 to 12:30


All authors in correct order:

Helfer B1, Dewji F1, Kunc M1, Jobson J1, Kroeger CM2, Dai Z2, Garcia-Larsen V3, Leonardi-Bee J4, Bero L2, Boyle RJ1
1 Department of Medicine, Imperial College London, UK
2 Charles Perkins Centre, University of Sydney, Australia
3 Department of International Health, Johns Hopkins Bloomberg School of Public Health, USA
4 School of Medicine, University of Nottingham, UK
Presenting author and contact person

Presenting author:

Robert Boyle

Contact person:

Abstract text
Background: infant formula is consumed by a vulnerable population and may have a substantial long-term impact on health. A strong scientific evidence-base is essential in this field, so that parents, carers and health services can make informed feeding choices. However, risk of bias and conflict of interest in clinical trials of infant formula remain under-explored, despite recent concerns regarding the quality of evidence in this field. A revised and expanded version of the Cochrane 'Risk of bias' tool (RoB 2) has recently been developed. RoB 2 seems suitable for a large-scale evaluation of clinical trials in a single field of research, but the tool has not yet been used for that purpose.

Objectives: we aimed to evaluate systematically risk of bias and conflict of interest in published infant formula trials. To identify potential sources of bias in the clinical literature in this field we applied RoB 2.

Methods: we searched MEDLINE, Embase and the Cochrane Database of Systematic Reviews to February 2019 to identify clinical trials of infant formulas in children aged less than three years published as full-text articles in peer-reviewed journals. We examined risk of bias and conflict of interest and their relationship to the trials' primary outcomes. Risk of bias was assessed using the RoB 2 tool. The review was registered in PROSPERO under CRD42014006661.

Results: we identified over 1000 eligible publications. Analysis of publications reporting trial primary outcomes showed a high level of industry involvement in this field, with the majority of trials including industry-affiliated authors or industry involvement in data analysis, or both. Risk of bias was high in the majority of trials, most commonly due to issues with the selection of reported results. Training and implementation of RoB 2 was successful, but issues were identified with potential redundancy between domains 2 and 3 when reasons for excluding participants from analysis were under-reported; with the absence of an approach to handling variations in rates of loss to follow-up in domain 3; and with making judgements about the likelihood of a prespecified analysis plan where trial registries provided insufficient details.

Conclusions: in a comprehensive analysis of a field of research characterized by high levels of conflict of interest, RoB 2 reliably identified issues with selective reporting.

Patient or healthcare consumer involvement: former participants of infant formula trials gave their feedback regarding new methodological guidance for infant formula trials developed in response to an early pilot of this systematic review, expressing their concerns at the possibility that trial results may be subject to bias due to methodological errors or commercial interests.