Skip to main content

Table 3 Logistic regression results examining predictors of HTE or subgroup analysis

From: Dealing with heterogeneity of treatment effects: is the literature up to the challenge?

  Predict HTE analysis
Condition OR (95% CI)*
Journal of publication
   BMJ 1.00 [reference]
   Annals 4.22 (1.18–15.08)
   JAMA 4.40 (1.43–13.53)
   Lancet 1.44 (0.55–3.77)
   NEJM 2.36 (0.85–6.56)
Year of publication  
   2004 1.00 [reference]
   1999 0.88 (0.47–1.66)
   1994 0.83 (0.41–1.67)
Medical condition under study  
   Other 1.00 [reference]
   Cardiovascular 1.07 (0.50–2.32)
   Infectious disease 1.38 (0.66–2.89)
   Cancer 1.13 (0.47–2.71)
   Psychiatry/Neurology 2.85 (0.91–8.92)
First author's study region  
   Other 1.00 [reference]
   North America 1.23 (0.67–2.29)
Sample size
   Quintile 1 (median = 37) 1.00 [reference]
   Quintile 2 (median = 124) 0.72 (0.24–2.14)
   Quintile 3 (median = 263) 4.17 (1.62–10.74)
   Quintile 4 (median = 549) 3.08 (1.20–7.91)
   Quintile 5 (median = 1560) 7.55 (2.95–19.33)
  1. Abbreviations: HTE, heterogeneity of treatment effects; OR, odds ratio; CI, confidence interval; BMJ, British Medical Journal; JAMA, Journal of the American Medical Association; NEJM, New England Journal of Medicine
  2. *Adjusted for all other variables in the model
  3. Significant at p < 0.05 for a combined Wald Chi-square test
  4. Significant at p < 0.001 for a combined Wald Chi-square test