. 2015 May 18;12(5):5256–5283. doi: 10.3390/ijerph120505256

Table 2.

Studies of the relationship between heat wave and hospitalization.

Reference	Region and Time	Heat Wave Definition	Method	Outcome Variable	Key Findings	Effect Estimate	Comments
(Manser et al. 2013) [37]	Zurich, Switzerland. 1 January 2001–31 December 2005	Any period of six days with Tmax > 5 °C above Tmax (recommended by the World Meteorological Organization)	Time-series; Poisson regression	Hospital admissions for IBD, IG and NIIs	Heat wave was significant associated with increased risk of IBD and IG flares. The strongest heat wave effect was observed on IG at lag seven days.	IBD flares: percentage increase: 4.6% (1.6%, 7.4%);	Data on one hospital was used. Day of the week, public holidays as Sundays, long-term trends and yearly seasonal patterns were adjusted.
						IG flares: percentage increase: 4.7% (1.8%, 7.4%);
						IG flares: percentage increase: 7.2% (4.6%, 9.7%) at lag 7.
(Ha et al. 2014) [33]	Allegheny County, Pennsylvania. May–September 1994–2000	Greater than two consecutive days with AT > 95th percentile (26.1 °C) of all temperatures.	Time-stratified case-crossover analysis	Stroke hospitalization for ischemic and hemorrhage stroke	Heat wave at lag-2 day was significantly associated with an increased risk for stroke hospitalization. The effect estimates were more significant for ischemic stroke, men and subjects aged 80 yrs or more.	Stroke: OR: 1.173 (1.047, 1.315) at lag2;	This analysis was stratified by gender, race, age group and type of stroke.
						Ischemic stroke: OR: 1.145 (1.009, 1.299) at lag2;
						Male: OR: 1.201 (1.008, 1.430) at lag2;
						Age, 65–79 yrs: OR: 1.161 (1.001, 1.346) at lag 2;
						Age, 80+ yrs: OR: 1.191 (1.003, 1.414) at lag 2.
(Ma et al. 2011) [29]	Shanghai, China. 2005–2008	Greater than seven consecutive days with daily Tmax above 35.0 °C and daily average temperatures above 97th percentile during the study period	The difference in the numbers of hospital admission between heat wave and reference period was used to calculate excess hospital visits, RRs and 95% CI	Hospital admission	The heat wave was significant associated with increase of total, CVD and RD.	Total: RR: 1.02 (1.01, 1.04).	The data were from database of Shanghai Health Insurance Bureau, covering most of the residents in Shanghai.
						CVD: RR: 1.08 (1.05, 1.11).
						RD: RR: 1.06 (1.00, 1.11).
(Hansen et al. 2008a) [36]	Adelaide, Australia. 1995–2006	Greater than three consecutive days when daily Tmax ≥35 °C, the 95th percentile of the Tmax range for the study period	Time-series; conditional-fixed effects Poisson regression;	Hospital admission for renal disease	During heat wave, significant increases were found on renal disease and acute renal failure. The effect estimates were higher among the elderly.	Renal disease: IRR: 1.100 (1.003, 1.206);	The analyses were stratified by three age group (15–64, 65+, 85+ yrs) and gender. Seasonality, long-term trend and over dispersion were controlled.
						ARF: IRR: 1.255 (1.037, 1.519).
						Age,15–64 yrs: IRR: 1.130 (1.025, 1.247);
						Age, 85+ yrs: IRR: 1.196 (1.036, 1.380).
(Hansen et al. 2008b) [34]	Adelaide, Australia. 1 October–31 March 1993–2006	Greater than three consecutive days when daily Tmax ≥ 35 °C, the 95th percentile of the Tmax range for the study period	Time-series; conditional-fixed effects Poisson regression;	Hospital admission for MBDs	During heat wave, significant increases were found on patients with organic illnesses, including symptomatic mental disorders; dementia; mood (affective) disorders; neurotic, stress related, and somatoform disorders; disorders of psychological development; and senility. Higher effect estimates were observed on patients with senility.	MBDs: IRR: 1.073 (1.017, 1.132);	The analyses were stratified by three age group (15–64, 65–74, 75+ yrs), and gender. Seasonality, long-term trend and over dispersion were controlled.
(Hansen et al. 2008b) [34]	Adelaide, Australia. 1 October–31 March 1993–2006		Time-series; conditional-fixed effects Poisson regression;	Hospital admission for MBDs		Senility: IRR: 2.366 (1.200, 4.667).
(Semenza et al. 1999) [30]	Cook County, Chicago. 1994–1995	13–19 July 1995	Excess admission: the weekly average (the expected number of admissions) was subtracted from the number of admissions recorded during the heat wave study period. 95% CI: a standard method based on the t-distribution	Hospital admissions	Significant excess increases were observed on patients with disorders of fluid, volume depletion, nephritis, acute renal failure, heat stroke, anhydrotic heat exhaustion, heat exhaustion, hypertensive disease, ischemic heart disease, cardiac dysrythmias, diseases of arteries, cerebrovascular disease, late effects of cerebrovascular disease, diabetes mellitus, noninsulin dependent diabetes (Type II).	Significant excess rate ranged from 19% (p = 0.019) for ischemic heart disease to 78,000% (p < 0.001) for heat stroke.	Cause-specific hospital admissions were analyzed.
(Mastrangelo et al. 2007) [38]	Veneto Region, Italy. 1 June–31 August 2002–2003	Greater than three consecutive days with Humidex above 40 °C	GEE	Hospital admissions for the elderly (≥75 yrs)	Heat wave duration increased the risk of hospital admissions for heat disease and RD.	Heat diseases: IRR: 1.16 (1.12, 1.20);	Humidex was used; heat wave characteristics were analyzed, such as duration, intensity and a dummy variable for days outside or inside a heat wave.
(Mastrangelo et al. 2007) [38]	Veneto Region, Italy. 1 June–31 August 2002–2003		GEE	Hospital admissions for the elderly (≥75 yrs)	RD: IRR: 1.05 (1.03, 1.07).	No correlation was found for fractures of femur or circulatory disease admissions
(Sheridan et al. 2014) [39]	New York, USA. April–August 1991–2004	Three consecutive days of DT (Dry Moderate) or MT+ (Moist Tropical Plus) weather type	Time-series; DLM	Hospitalizations for heat-related disease, CVD, RD.	The strongest effect of heat wave was observed on heat-related illness. Heat-related hospital admissions have increased during the time, especially during the earlier days of heat events.	Heat-related disease: RR: 25.891 (20.300, 33.022) during 1991–2004;	The analysis included different time period (1991–1996 and 1997–2004) and seasons (spring and summer).
						RR: 25.46 (15.98, 40.57) during 1991–1996;
						RR: 26.69 (20.19, 35.29) during 1997–2004.
(Williams et al. 2012) [24]	Perth, Australia. 1 January 1980–1 July 2008	Greater than three consecutive days with Tmax ≥35 °C	GEE; negative binomial regression	Hospital admissions	Total hospital admissions decreased during heat wave days.	Hospital admissions: IRR: 0.905 (0.854, 0.958).	Twenty-eight years of data were used.
(Nitschke et al. 2011) [25]	Adelaide, Australia. July 1993–March 2009	Greater than three consecutive days with Tmax ≥35 °C (the 95th percentile of Tmax for the period 1993–2009)	Case-series analysis; Negative binomial regression	Hospital admissions for total, ischemic, mental, renal, RD and direct heat disease	During heat waves, highest effect estimates were found on direct heat disease during 2008 and 2009 heat wave.	2008 heat wave: RR: 2.64 (1.32, 5.20);	The analyses were stratified by age group (0–4, 5–14, 65–74, 75+ yrs).
(Nitschke et al. 2011) [25]	Adelaide, Australia. July 1993–March 2009		Case-series analysis; Negative binomial regression			2009 heat wave: RR: 13.66 (8.80, 20.98).
(Knowlton et al. 2009) [26]	Fifty-eight counties of California, USA. 8 July 2006–22 August 2006	15 July–1 August 2006	RR: the ratio of the number of cases in the heat wave and reference period. Excess cases: the difference of the number of cases in the two period	Hospitalizations	During heat wave, a significant increase was found only on (1) Electrolyte imbalance, (2) Nephritis and nephritic syndrome (3) Acute renal failure (4) Heat-related illnesses	(1) RR: 1.09(1.07, 1.11)	The analyses were stratified by gender, age, causes and race.
						(2) RR: 1.05(1.02, 1.07)
						(3) RR: 1.11(1.08, 1.15)
						(4) RR: 10.15(7.79, 13.43)
(Nitschke et al. 2007) [27]	Adelaide, Australia. 1993–2006	Greater than three consecutive days with daily Tmax ≥35 °C	Case-series study	Hospital admissions for CVD, RD, mental and renal disease	Highest effect estimate was observed for renal patients aged 15–64 yrs.	IRR: 1.16 (1.04, 1.30)	This analysis included five age groups (0–4, 5–14, 15–64, 65–74, 75+ yrs).
(Lindstrom et al. 2013) [28]	Melbourne, Australia. 2007–2009	28 January–3 February 2009	Poisson regression	Hospital admissions, general medical admissions	During heat wave, a significant increase was observed in hospital admissions and general medical admissions.	(1) IRR: 1.11 (p < 0.05)	One hospital data was used.
(Lindstrom et al. 2013) [28]	Melbourne, Australia. 2007–2009	28 January–3 February 2009	Poisson regression	Hospital admissions, general medical admissions		(2) IRR: 1.81 (p < 0.01)	One hospital data was used.

Abbreviations: Tmax: maximum temperature; AT: apparent temperature; IBD: inflammatory bowel disease; IG: infectious gastroenteritis; NIIs: noninfectious chronic intestinal inflammations; CVD: cardiovascular disease; RD: respiratory disease; MBDs: mental and behavioral disorders; GEE: Generalized Estimating Equation; DLM: distributed linear model; OR: odds ratio; RR: rate ratio; IRR: incidence rate ratio; CI: confidence interval.