We obtained daily counts of all-cause deaths and daily mean temperatures for the 1987–2000 period for 15 U.S. cities (Atlanta, Boston, Chicago, Dallas, Detroit, Houston, Los Angeles, Miami, New York City, Philadelphia, Phoenix, San Francisco, Seattle, St. Louis, and Washington, DC) from the NMMAPS (31). We statistically modeled the delayed and nonlinear relationship between daily mean temperature and all-cause mortality over the whole range of observed temperatures for each city using distributed lag nonlinear models (5, 57). This type of model simultaneously describes the exposure-response relationship and an additional lag-response relationship. We considered a lag period of up to 21 days to capture any delayed responses, as was done in previous studies (5, 6).

We used a natural cubic spline with the same internal knots as that used in Gasparrini et al. (13) to model the exposure-response curve for each city. This allows for log-linear extrapolation of the temperature-mortality relationship beyond the observed temperature range. Extrapolation is essential for higher temperatures projected under climate change (see table S2 for specific details). We acknowledge that the log-linear extrapolation applied may be conservative in estimating the true exposure-response relationship at temperatures higher than the observed range (58).

Nevertheless, the temperature-mortality relationships we found were robust to the choice of observational period and spline model. Gasparrini et al. (5) used a longer (1985–2006) dataset and a quadratic B-spline model and found similar relationships for the same cities. The 21-day lag period that we used was also sufficient to capture potential delayed responses, as our relationships were comparable to those found by Anderson and Bell (6) using a longer lag period (28 days). Furthermore, Curriero et al. (59) reported higher heat-related relative mortality risks over northeastern U.S. cities versus southeastern cities between 1973 and 1994, a characteristic that we found with our dataset and statistical model.

Note: The content above has been extracted from a research article, so it may not display correctly.



Q&A
Please log in to submit your questions online.
Your question will be posted on the Bio-101 website. We will send your questions to the authors of this protocol and Bio-protocol community members who are experienced with this method. you will be informed using the email address associated with your Bio-protocol account.



We use cookies on this site to enhance your user experience. By using our website, you are agreeing to allow the storage of cookies on your computer.