Table 2

Daily time-series studies with temperature exposures and myocardial infarction (MI) outcomes: summary/interpretation

First author and yearMIs (n)Long-term local mean temperature and annual range* (°C)Detrimental effect of cold?Detrimental effect of heat?Adjusted for season and trend?MI events: validation or specified criteriaAdjusted for humidityAdjusted for atmospheric pressureAdjusted for infectious disease levelsAdjusted for air pollutionAdjusted for day of week?Allowance for non-linear temperature effects?Investigated lag effects?
Studies of fatal and non-fatal events
Gerber 20061426767 (−10.8 to 21.2)NoNo
Wang 200615375516.7 (5.6 to 28.3)YesNo
Enquselassie 199316388917.9 (11.9 to 23.7)Yes (fatal MIs only)No✗†
Ohlson 1991173576.3 (−3.9 to 17.3)NoN/A
Barnett 20051887 410YesNo
Morabito 200519268314.1 (5.8 to 23.5)YesYes (for increased hours of discomfort)
Ebi 200420283 03114.7 (10.1 to 18.6) to 19.3 (14.8 to 24.4)Certain regions only (1/3 regions for night temperature; 2/3 regions for day temperature)Certain regions only (2/3 regions for night temperature;0/3 regions for day temperature)
Koken 200321Not reported9.7 (−1.4 to 21.9)N/AYes
Messner 2002223322−1.4 (−15.6 to 13.5) to 4.6 (−5.5 to 16.5)NoYes (change from previous day, non-fatal MIs only)✗‡
Danet 199923331410.6 (2.0 to 19.1)YesNo✗§
Fatal events only
Dilaveris 200624312618.0 (8.6 to 28.4)YesYes✗†
Sharovsky 20042512 00718.0 (13.9 to 21.8)YesYes
Rossi 199926Approx 160012.8 (3.8 to 22.9)Not mentionedYes
  • *10-Year average of the monthly mean temperatures, and of the minimum and maximum monthly mean temperature in the study area, as recorded at the nearest available monitoring station included in the Goddard Institute for Space Studies (GISS) surface temperature data,12 using data from the years 1991–2000 inclusive; †authors adjusted for calendar month, which should have approximately captured any seasonal effect; ‡authors performed a basic adjustment for season in a sensitivity analysis, which did not change the conclusions; §authors adjusted for annual population and calendar year, which should have approximately captured any long-term trend.