Using logic regression to characterize extreme heat exposures and their health associations: a time-series study of emergency department visits in Atlanta

Table 3 Summary of extreme heat exposure estimated by logic regression and their short-term associations with warm-season emergency department visits in Atlanta, 1993 to 2012. Relative risk estimates and 95% confidence intervals (CI) were from time-series models adjusting for truncated continuous temperature. Within each outcome, each row of the extreme heat exposure corresponds to an “or” statement derived from the logic regression tree. “Not” statement is indicated by a superscript c

ED Visit Outcome	Heat Metric and Quantile	Extreme Heat Exposure	lag0	lag1	lag2	lag3	Frequency (days)	Relative Risk (95% CI)
HEAT	ATMN95	(lag0 and lag1)	Y	Y			237	1.400 (1.269, 1.544)
HEAT	ATMN95	OR (lag1 and lag3)		Y		Y
FLEL	ATMN95	lag1		Y			335	1.030 (1.016, 1.045)
RENAL	ATMN95	(lag0 and lag3)	Y			Y	255	1.025 (1.011, 1.039)
RENAL	ATMN95	OR (lag2^C and lag3)			N	Y
ARF	ATMN95	lag0 and lag1	Y	Y			190	1.052 (1.021, 1.085)
STK	TMX98	lag0 and lag1 and lag2^C and lag3	Y	Y	N	Y	5	1.257 (1.061, 1.477)
CIRC	ATMX95	lag1		Y			363	0.996 (0.987, 1.004)
HT	ATMX95	lag0 and lag1 and lag3^C	Y	Y		N	128	0.988 (0.977, 0.999)
		OR (lag1 and lag2 and lag3^C)		Y	Y	N
IHD	TMN99	lag0	Y				63	1.013 (0.986, 1.041)
MI	TMN99	lag0	Y				63	1.044 (0.964, 1.131)
CHF	ATMN98	lag0	Y				146	0.976 (0.947, 1.007)
DIA	TMN95	lag2			Y		238	1.009 (0.996, 1.021)
INTERN	ATMN95	lag3				Y	335	1.008 (1.002, 1.015)

HEAT heat illness, STK ischemic stroke, ARF Acute renal failure, RENAL all renal disease, FLEL Fluid and electrolyte imbalance, CIRC all circulatory system disease, CHF Congestive heart failure, IHD Ischemic heart disease, MI Myocardial infarction, DIA Diabetes, HT Hypertension, ITERN all internal causes)

ISSN: 1471-2288