Table 4

Covariate estimates for the model with four groups (G2, GMA, GCum1t, Gcum1t-2)

Temperature indicator
Coefficient estimate β (SE)
Percentile P10% – P90% of the indicator
Relative risk estimate of P90%/P10%a
Pr > |Z|b

G2




     Tmint
-0.0219 (0.0027)
8.74 – 17.35
0.83
0.0145
     Tmaxt-1
-0.0124 (0.0018)
17.74 – 29.98
0.86
0.2561
     Tmint × Tmaxt-1
0.0013 (0.0001)
164.25 – 515.27
1.58
0.0000

GMA




     Moving average
-0.0158 (0.0009)
14.60 – 22.52
0.88
0.0000

GCum1t




     Tmaxt
0.0045 (0.0006)
17.75 – 29.98
1.06
0.0000
     Ctmaxt
-0.007 (0.0026)
0.00 – 10.17
0.93
0.2024
     Tmaxt × Ctmaxt
0.0003 (0.0001)
0.00 – 298.61
1.09
0.0015

     GCum1t-2




     Tmaxt-2
-0.0002 (0.0006)
17.75 – 29.98
1.00
0.9319
     Ctmaxt-2
-0.0093 (0.0027)
0.00 – 10.17
0.91
0.0181
     Tmax-2 × Ctmaxt-2
0.0004 (0.0001)
0.00 – 298.61
1.13
0.0000

a Multiplying factor for the increase in mortality when the indicator increases from the 10th percentile (P10%) to the 90th percentile (P90%),

b Pr > |Z| is the significance probability

Fouillet et al. BMC Public Health 2007 7:114   doi:10.1186/1471-2458-7-114