Article Text
Abstract
Background Currently the GRADE approach only considers the development of recommendations for treatment and diagnosis related clinical questions.
Objectives To develop a proposal using GRADE to formulate recommendations for aetiology questions.
Methods The exposure to paracetamol in the first year of life and asthma at the age of 6 years was chosen as an example. A systematic review was made, searching in Cochrane, CRD, Medline and Embase databases. Cohort studies or Meta-analysis of observational studies were selected, and CASPE and MOOSE checklists used for quality assessment. The evidence was considered high when there were no factors affecting our confidence on the estimates.
Results One systematic review (including 4 related cross-sectional studies) and 4 cohorts were found. To assess the evidence with GRADE two outcomes were used (increase of asthma and wheezing at 6 years). Risk of bias was assessed taking into account the cohort design and aetiology studies characteristics, but inconsistency, indirectness and imprecision were considered in the same manner. In contrast to cross-sectional studies, cohort studies (3 out of 4) did not find a significant relationship between paracetamol and asthma development. Due to possible risk of bias and imprecision, the quality of the evidence was downgraded two levels and a recommendation in favour of the rational use of paracetamol in children was made.
Discussion A proposal to address aetiology questions based on the GRADE approach has been developed. More examples are needed to validate it.
Implications for Guideline Developers To be able to use the GRADE approach independently of the question type.