Abstract
Background
The quality of evidence provided by systematic reviews rests on clear reporting in original papers. A STROBE checklist aids reporting of case-control studies, but does not distinguish different case-referent designs. These latter include case-base, incidence density and case-exposure designs, as well as the (traditionally understood) case-control study. The measure estimated and, if an odds ratio, how that may be interpreted, is dependent upon the design. In spite of methodology elaborated over four decades, a 2000 review identified numerous difficulties with authors' reporting and interpretation of case-referent studies, and a review a decade later indicated that such issues prevailed.
Objective
To provide guidance on reporting of case-referent studies. METHODS: Medline search in 2000 for odds ratio, relative risk, rare disease assumption and rarity assumption; review of collection of articles from R. Bhopal; snowballing.
Results
1) State the reference series sampling scheme. Cumulative incidence (traditional or exclusive), incidence density (concurrent) or case-base (inclusive), sampling. 2) State whether incident and/or prevalent cases enrolled 3) State what your case-referent study is calculating. A true odds ratio will only result from a case-controlstudy. Incidence density sampling, where the reference series is used to calculate population time, yields an incidence density ratio and case-base sampling enables direct calculation of a risk ratio. 4) State the formal assumptions made or formulae used 5) Refrain from 'labelling' effect measures relative risk (especially in the abstract).
Conclusions
For case-referent studies to be incorporated into systematic reviews authors should report and editors should require greater technical detail. Would an extension of STROBE help?
The quality of evidence provided by systematic reviews rests on clear reporting in original papers. A STROBE checklist aids reporting of case-control studies, but does not distinguish different case-referent designs. These latter include case-base, incidence density and case-exposure designs, as well as the (traditionally understood) case-control study. The measure estimated and, if an odds ratio, how that may be interpreted, is dependent upon the design. In spite of methodology elaborated over four decades, a 2000 review identified numerous difficulties with authors' reporting and interpretation of case-referent studies, and a review a decade later indicated that such issues prevailed.
Objective
To provide guidance on reporting of case-referent studies. METHODS: Medline search in 2000 for odds ratio, relative risk, rare disease assumption and rarity assumption; review of collection of articles from R. Bhopal; snowballing.
Results
1) State the reference series sampling scheme. Cumulative incidence (traditional or exclusive), incidence density (concurrent) or case-base (inclusive), sampling. 2) State whether incident and/or prevalent cases enrolled 3) State what your case-referent study is calculating. A true odds ratio will only result from a case-controlstudy. Incidence density sampling, where the reference series is used to calculate population time, yields an incidence density ratio and case-base sampling enables direct calculation of a risk ratio. 4) State the formal assumptions made or formulae used 5) Refrain from 'labelling' effect measures relative risk (especially in the abstract).
Conclusions
For case-referent studies to be incorporated into systematic reviews authors should report and editors should require greater technical detail. Would an extension of STROBE help?
| Original language | English |
|---|---|
| Pages (from-to) | A54-A54 |
| Number of pages | 1 |
| Journal | Value in Health |
| Volume | 16 |
| Issue number | 3 |
| DOIs | |
| Publication status | Published - 1 May 2013 |