Further thoughts on the importance of models in the assessment of clinical evidence
| Autor: | C. David Naylor, Antoni Basinski |
|---|---|
| Rok vydání: | 1992 |
| Předmět: |
medicine.medical_specialty
Aspirin Models Statistical Epidemiology business.industry Fibrinolysis medicine.medical_treatment Myocardial Infarction Logistic regression Placebo Models Biological Clinical trial Clinical evidence Humans Medicine Statistical analysis business Intensive care medicine Probability Demography medicine.drug |
| Zdroj: | Journal of Clinical Epidemiology. 45:729-732 |
| ISSN: | 0895-4356 |
| DOI: | 10.1016/0895-4356(92)90050-w |
| Popis: | The assessment of clinical evidence relies on a meaningful and concise summary of clinical data. Any summary is likely to be meaningful only if it is biologically and statistically appropriate. If the analysis is “explanatory” rather than “pragmatic” [l], the summary may serve both to elucidate biological mechanisms and to draw statistical conclusions. Hlatky and Whittemore [2] have commented on the importance of statistical models in the identification of synergy between two therapeutic agents, a concept explored in previous publications [3,4], They have focused attention on the assessment of synergy between aspirin and fibrinolysis in acute myocardial infarction which we reported, originally for the ISIS-2 [5,6] trial data and subsequently in a metaanalysis [7]. These analyses demonstrated that the effect of fibrinolysis vs placebo when both are administered in the presence of aspirin is greater than that of fibrinolysis vs placebo when both are administered in the absence of aspirin. Logistic regression was utilized to analyze the aggregate trial data. Although it may be argued that the definition of interaction and synergy is arbitrary in the absence of external information to guide the analysis, it is rare that the biological, clinical, and statistical milieu within which the observed |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|