study weeks of publication: retrospective cohort at two years using data available within three Prediction of citation counts for clinical articles

ABSTRACTObjective To determine if citation counts at two yearscould be predicted for clinical articles that pass basiccriteriaforcriticalappraisalusingdatawithinthreeweeksofpublicationfromexternalsourcesandanonlinearticlerating service.Design Retrospective cohort study.Setting Online rating service, Canada.Participants 1274 articles from 105 journals publishedfromJanuarytoJune2005,randomlydividedintoa60:40split to provide derivation and validation datasets.Main outcomemeasures 20 article and journal features,including ratings of clinical relevance andnewsworthiness, routinely collected by the McMasteronline rating of evidence system, compared with citationcounts at two years.Results The derivation analysis showed that theregression equation accountedfor 60% of the variation(R 2 =0.60, 95% confidence interval 0.538 to 0.629). Thismodel applied to the validation dataset gave a similarprediction (R 2 =0.56, 0.476 to 0.596, shrinkage 0.04;shrinkage measures how well the derived equationmatches data from the validation dataset). Cited articlesinthetophalfandtopthirdwerepredictedwith83%and61% sensitivity and 72% and 82% specificity. Highercitations were predicted by indexing in numerousdatabases; number of authors; abstraction in synopticjournals; clinical relevance scores; number of citedreferences; and original, multicentred, and therapyarticlesfrom journalswitha greater proportionof articlesabstracted.Conclusion Citation counts can be reliably predicted attwo years using data within three weeks of publication.INTRODUCTIONEvidencebasedmedicineincorporatesthebestclinicalevidencewithclinicians’expertiseandpatients’valuesinclinicaldecisionmaking.Findinganddisseminatingsuch evidence is a difficult task given the amount ofresearch being published continuously. The task ismade more difficult by high quality and clinicallyrelevant healthcare articles being diluted in the largerbodyofarticlesreportingbasicscience,opinions,news,and lesser quality studies.If the importance of an article to clinical practicecould be predicted soon after publication, then afocused push of such articles could be made toclinicians and other readers who could potentially usethem.Thearticlescouldalsobequicklysenttoauthorsand publishers of information resources, systematicreviews, clinical practice guidelines, and educationalprogrammes.Variousfactorshavebeenusedtopredictthe number of citations for an article, such asmethodological quality,