Table 1

Results of meta-regression models of the weekend effect on mortality

ParameterNumber of estimates in categoryEstimate (95% CrI)% difference in odds ratio (compared with baseline/reference category) (95% CrI)
 Intercept0.05 (−0.10, 0.20)(Baseline/reference category OR)
1.05 (0.90, 1.22)
Adequacy of statistical adjustment
 1 or 2a: Adjustment including measures of acute physiology5ReferenceReference
 2b: Adequate adjustment of main and contextual factors400.13 (−0.03, 0.30)14% (−3% , 35%)
 3: Partial adjustment400.13 (−0.03, 0.29)14% (−3%, 34%)
 4: Inadequate adjustment340.15 (−0.01, 0.31)16% (−1%, 37%)
Surgical admissions yes81−0.04 (−0.14, 0.06)−4% (−13%, 6%)
Elective admissions yes270.27 (0.21, 0.32)31% (24%, 38%)
Maternity admissions yes23−0.18 (−0.26,–0.10)−17% (−23%, −10%)
Time (linear trend)1190.00 (0.00, 0.00)0% (0%, 0%)
Total number of observations/estimates119
  • Time (year) was selected as mid-point of the data collection period. Categories 1 (comprehensive adjustment) and 2a (adequate adjustment including measures of acute physiology) were combined due to the low number of studies in these categories. Estimates can be interpreted as approximate percentage increase in the estimate of weekend effect OR. Meta-regressions also have country random effect (varying intercept for countries). Individual studies can contribute to multiple estimates of the weekend effect, for example, by individual years, different patient subgroups and individual weekdays/weekend days (eg, Saturday vs Wednesday and Sunday vs Wednesday).

  • CrI, credible interval.