Table 3.
Summary of the study findings that assess temperature effects on hospital visits/admissions, infectious diseases and other neonatal outcomes.
| Main Author (Year) | Research Design | Location, Year of Study | Description of Climate | Main Temperature Exposure Variable(s) | Statistical Analysis | Outcomes Measures (Source of Data) | Effect Estimates |
|---|---|---|---|---|---|---|---|
| All-Cause/Heat-Related Hospital Visits and/or Admissions | |||||||
| Sheffield (2018) [13] | Case-crossover | New York City, USA 2005–2011 | Continental climate. | Minimum, maximum and average apparent daily temperature. Lag 0–6 days | Time-stratified case-crossover | 278,114 all-cause emergency department (ED) visits (hospital records, all-cause diagnostic codes) | Positive association with maximum temperature; 0.6% (95%CI = 0.1, 1.1). |
| Xu (2017) [44] | Time-series analysis | Brisbane, Australia2005–2015 | Humid, sub-tropical climate. Mean temperature 9.0 °C–30.9 °C | Mean daily temperature. Nine different heatwave definitions | Poisson generalised additive model and distributed lag non-linear model | 53,792 all-cause hospital admissions (hospital records) | When mean temperature was >97th percentile, the RR of hospital admissions increased by 1.05 (95%CI = 1.01, 1.10), and then further increased to 1.18 (95%CI = 1.05, 1.32) when duration of heatwave increased from 2 days to 4 days. No evidence of effect modification. |
| Kakkad (2014) [19] | Retrospective cohort | Ahmedabad, India 2009–2011 | Warm, dry conditions that often include heat waves. Average monthly maximum of 38.8 °C between March and June. Heat wave in May 2010, maximum temperature 46.8 °C. | Daily maximum temperature. No lags. | Generalised linear models and segmented regression. | Heat-related illness in neonatal admissions—defined as diagnosis of exclusion when body temperature >38 °C with any signs and symptoms such as refusal to feed, signs of dehydration, increased respiratory rate, convulsions and/or lethargy (hospital records) | Above 42 °C, each temperature increase of a degree was associated with a 43% increase in heat-related admissions (95%CI = 9.2, 88). |
| Mannan (2011) [28] | Retrospective cohort | Sylhet district, Bangladesh2004–2006 | Tropical monsoon climate. Monthly average temperatures range 19.3 °C–29.3 °C for Sylhet and 17.7 °C–29.5 °C for Mirzapur | 7 day rolling average temperature and rolling humidity index prior to diagnosis | Multivariable logistic regression. | Very severe disease in 6936 newborns in Sylhet and 5900 newborns in Mirzapur—diagnosed on history using clinical algorithms) (data from 2 cluster randomised controlled trials) | OR for temperature at time of diagnosis of very severe disease were 1.14 (95%CI = 1.08, 1.21) in Sylhet and 1.06 (95%CI = 1.04, 1.07) in Mirzapur. |
| Infectious diseases | |||||||
| Gosai (2009) [45] | Retrospective cohort | Auckland, New Zealand1994–2004 | Temperate oceanic climate. Average monthly temperatures 10.8 °C–19.8 °C | Daily minimum temperature. Lag 0–7 days | Pearson’s correlation | Hospital admissions for respiratory illness (hospital admissions data. No ICD codes given) | Correlation between minimum temperature and respiratory infections and inflammation (r = −0.42 and p < 0.001) and total whooping cough and acute bronchiolitis (r = −0.40 and p < 0.001). |
| Yan (2019) [29] | Time-series analysis | Shenzhen, China 2009–2017 | Subtropical monsoon climate. Mean temperature over period 23.3 °C. | Daily temperature, unclear which index used. Lag with a max of 30 days | Quasi-Poisson regression based on distributed lag nonlinear model | 50,657 HFMD cases (surveillance data—notifiable disease) | Cumulative RR of HFMD over 14 days in 0–1 age group was RR 0.58 (95%CI = 0.4, 0.84) for temperature in 5th percentile and RR 2.03 (95%CI = 1.77, 2.33) in the 95th percentile. 5th percentile was 12.9 °C, 95th was 30 °C and median 24.6 °C. |
| Yin (2015) [30] | Time series analysis | Chengdu, China 2013–2014 | Humid sub-tropical climate. Mean temperature for study period 16.21 °C. | Daily mean temperature. Lag 0–14 days | Poisson generalised linear regression combined with distributed lag non-linear model | 74,247 HFMD cases aged 0–5 years (surveillance data—notifiable disease) | Risk of HFMD significantly increased at temperatures 14, 17.2, 23.2 and 27 compared to 0 °C at lag 0 and lag 14 for infants <1 years old. No increase in HFMD risk for colder temperature exposure of 4.1 °C. For lag 0, RR 1.15 (95%CI = 1.03, 1.29), 1.19 (95%CI = 1.06, 1.34), 1.29 (95%CI = 1.13, 1.44), and 1.31 (95%CI 1.16, 1.48) for 14, 17.2, 23.2, and 27 °C, respectively. For lag 14 RR 1.09 (95%CI = 1.03, 1.16), 1.07 (95%CI = 1.00, 1.14), 1.06 (95%CI = 1.00, 1.13), and 1.03 (95%CI = 1.02, 1.04) for 14, 17.2, 23.2, and 27 °C, respectively. |
| Nenna (2017) [46] | Prospective cohort | Rome, Italy 2004–2014 | Mediterranean climate | Weekly average temperature. No lag | Pearson’s correlation | 723 cases of viral bronchiolitis in hospitalised infants (prospective clinical records) | The number of RSV-positive infants correlated negatively with temperature (r = −0.46, p < 0.001). |
| Kim (2017) [47] | Prospective cohort | Cheonan, South Korea 2006–2014 | Subtropical climate | Mean daily temperature. No lag days | Logistic regression | 2484 infants admitted with RSV A and RSV B (lab confirmed admissions with respiratory symptoms) | RSV A and RSV B infections were negatively correlated with average temperature; −0.056 for RSV A and −0.069 for RSV B infection, with p < 0.001 for both. |
| Hoeppner (2017) [48] | Retrospective cohort | Australia and New Zealand 2009–2011 | Perth—Mediterranean, Melbourne and Auckland-oceanic and Brisbane—humid, subtropical | Minimum temperature aggregated over a week. Lag 0–4 weeks | Linear regression. Poisson and negative binomial regression to verify results | 3876 infants admitted with bronchiolitis (data from prospective, multicentre clinical trial) | Minimum temperature and lag 0): r −0.62 (−0.75, 0.48) p < 0.001. Lag 1: r −0.58 (95%CI = −0.75, −0.43) p < 0.001. Lag 2: r −0.67 (95%CI = −0.75, −0.58) p < 0.001. Lag 3: r −0.34 (95%CI = −0.49, −0.20) p < 0.001 |
| Other neonatal outcomes | |||||||
| Iijima (2016) [49] | Prospective cohort | Hamamatsu, Japan 2012–2013 | Temperate climate. Average temperatures spring 15.3 °C, summer 27.3 °C, autumn, 18.0 °C and winter 7.0 °C. | Mean outdoor and indoor, and wind chill temperature. No lag | Simple and multivariate regression | 498 neonates. International normalised ratio on day 4 after birth (data from healthy neonates) | Significant correlation between INR and outdoor temperature (r = 0.25, p < 0.001). Weakly negative correlation between INR and room temperature (r = −0.13, p = 0.02). |
| Scrafford (2013) [31] | Retrospective cohort | Southern Nepal 2003–2006 | Humid subtropical climate | Minimum daily temperature. No lag days | Bivariate and multivariate analyses | Incidental jaundice in 18,985 neonates, defined as first report of yellow eyes/body based on visual assessment by study staff, not laboratory confirmed (part of nested pair of cluster-randomised, placebo-controlled, community based clinical trial) | OR 1.03 (95%CI = 1.02, 1.05) p < 0.001 for each 1 °C increase in minimum ambient temperature. Adjusted OR 1.04 (95%CI = 1.03, 1.06). |
RSV—Respiratory Syncytial Virus; HFMD—Hand, Foot, and Mouth Disease; CI—confidence interval; OR—odds ratio; RR—relative risk; INR—International Normalised Ratio. 
Cold exposure.