Table 3.

Studies using other data sources other than EMS or hospitalization.

Reference	Region and Time	Heat Wave Definition	Method	Outcome Variable	Key Findings	Effect Estimates	Comments
(Van Zutphen et al. 2012) [31]	New York, USA. June–August 1992–2006	Greater than three consecutive days with mean UAT above the 90th percentile	Case control study	Birth defects	Congenital cataracts were significantly associated with heat waves, while significant decrease was on gastroschisis. No statistically significant relationships were found among central nervous systems, CVD, craniofacial or genitourinary birth defect groups.	Congenital cataracts: OR: 1.97 (1.17, 3.32);	A population-based case-control study was performed. The analyses were stratified by causes.
						Gastroschisis: OR: 0.48 (0.28, 0.81).
(Schifano et al. 2013) [40]	Rome, Italy. 1 January–31 December 2001–2010	Greater than two consecutive days with MAT above the monthly 90th percentile or the daily Tmin above the monthly 90th percentile and MAT above the median monthly value	Time-series; DLM; Poisson GAM	Preterm birth	A significant increase of preterm birth was associated with heat waves.	Percentage increase: 19% (7.91%, 31.69%).	The long-term trend, seasonality and holidays were adjusted.
(Empana et al. 2009) [35]	Paris, France. 1 January–21 December 2000–2005	1–14 August 2003	Poisson regression analysis (the same period in years 2000–2002 and 2004–2005 as reference)	Out-of-hospital cardiac arrest due to heart disease and of ST-segment elevation myocardial infarction (STEMI) aged >18 yrs	During heat wave, a significant relative rate was found on out-of-hospital cardiac arrests but not on myocardial infarctions comparing to reference period. This increase estimates were higher among males and those aged above 60 yrs.	Out of Hospital Cardiac Arrests: RR: 2.34 (1.60, 3.41);	The data was from a city mobile intensive care units (MICU) database. The analysis was adjusted for gender and age.
						Myocardial Infarctions: RR: 1.09 (0.58, 2.03).
(Kent et al. 2014) [41]	Alabama, USA. May–September 1990–2010	Sixteen heat wave definitions	Time-stratified case-crossover	Preterm births	Effect of heat waves (first definition) defined as having at least two consecutive days with Tmean above the 98th percentile were much higher than that as at least two consecutive days with T mean above the 90th (second definition).	The first definition: ER: 32.4% (3.7%, 69.1%);	Sixteen heat wave definitions were performed according to previous studies.
						The second definition: ER: 3.7% (1.1%, 6.3%).	Effect estimates varied by heat wave definitions.
(Bai et al. 2014) [32]	Ningbo, China. 2011–2013	Greater than seven consecutive days with the Tmax >35 °C	Time-series; DLNM	Heat-related illness	The strongest cumulative effect of heat waves was on severe forms of illness. Males and all age groups were vulnerable to heat wave.	The strongest cumulative effect (Lag 0–5): RR: 10.69 (2.10, 54.44).	The data were collected from the national heat-related illness surveillance system. The analyses included age groups and gender.
(Xiang 2014) [42]	Adelaide, Australia. 1 October–31 March 2001–2010	Greater than three consecutive days with daily Tmax ≥ 35 °C	GEE	Workers’ compensation claim	For outdoor industries, daily claims increased significantly during heat waves. And male laborers, tradespersons aged ≥55 yrs, those employed in ‘agriculture, forestry and fishing’ and ‘electricity, gas and water’, occupational burns, wounds, lacerations, and amputations as well as heat illnesses were significantly associated with heat waves.	Outdoor industries: IRR: 1.06 (1.02, 1.10)	The analyses were stratified by gender, age, occupation and industry.

Abbreviation: UAT: universal apparent temperature; MAT: daily maximum apparent temperature; Tmax: maximum temperature; Tmin: minimum temperature; Tmean: mean temperature; CVD: cardiovascular disease; DLM: distributed lag model; DLNM: distributed lag non-linear model; GAM: generalized additive model; GEE: generalized estimating equation; OR: odds ratio; RR: relative risk.