Respiratory illness, hospital visits, and health costs: Is it air pollution or pollen?

Abstract

Background: Both air pollution and airborne pollen can cause respiratory health problems. Since both are often jointly present in ambient air, it is important to control for one while estimating the effect of the other when considering pollution-abating policies. To date only a limited number of studies have considered the health effects of both irritants jointly for a general population, and for a sufficiently long time period to allow for variation in seasonal concentrations of both components. The primary goal of this study is to determine the causal impact of fine particulate matter (PM_2.5) on hospital visits and related treatment costs, while controlling for potentially confounding pollen effects. Our study area is the metropolitan hub of Reno/Sparks in Northern Nevada.

Methods: Taking advantage of a rare sample of daily pollen counts over a prolonged period of time (2009-2015), we model the effects of PM_2.5 and pollen on respiratory-related hospital admissions for the population at large, plus specific age groups. Pollen data are provided by a local allergy clinic. Data on PM_2.5 and other air pollutants are obtained from the U.S. Environmental Protection Agency's air quality data web site. We collect daily meteorological data from the National Centers for Environmental Information's data repository. Data on hospital admissions are given by the Nevada Center for Surveys, Evaluations, and Statistics. Our econometric approach centers on a fully robust count data (Poisson) model, estimated via Quasi-Maximum Likelihood.

Results: We find that for our sample PM_2.5 effects are largely robust to the inclusion of both pollen counts and temporal indicators. In contrast, pollen effects vanish when time fixed effects are added, pointing at their correlation with unobserved temporal confounders. At the same time, model fit improves with the inclusion of temporal indicators. Based on our preferred specification, we find a significant PM_2.5 effect of approximately 0.5% additional hospital visits per day due to a one μg/m³ increase in PM_2.5. This translates into expected augmented treatment costs of $2700 per day for the same unit-change in PM_2.5. These figures can mount quickly when more pronounced and/or longer episodes of particulate matter pollution are considered, perhaps due to wildfire smoke. For instance, the expected increase in patients and costs due to a month-long 10-unit-jump of PM_2.5 over the long-run annual average would amount to an extra 70 patients and approximately $680,000 in additional treatment costs.

Keywords: Air pollution; Allergies; Count data models; Hospital visits; Respiratory health.