We estimate the effect of acute air pollution exposure on mortality, life-years lost, and health care utilization among the US elderly. We address endogeneity and measurement error using a novel instrument for air pollution that strongly predicts changes in fine particulate matter (PM 2.5) concentrations: changes in the local wind direction. Using detailed administrative data on the universe of Medicare beneficiaries, we find that an increase in daily PM 2.5 concentrations increases three-day county-level mortality, hospitalizations, and inpatient spending, and that these effects are not explained by co-transported pollutants like ozone and carbon monoxide. We then develop a new methodology to estimate the number of life-years lost due to PM 2.5. Our estimate is much smaller than one calculated using traditional methods, which do not adequately account for the relatively low life expectancy of those killed by pollution. Heterogeneity analysis reveals that life-years lost due to PM 2.5 varies inversely with individual life expectancy, indicating that unhealthy individuals are disproportionately vulnerable to air pollution. However, the largest aggregate burden is borne by those with medium life expectancy, who are both vulnerable and comprise a large share of the elderly population.
[In the first Ordinary Least Squares (OLS) estimate presented] each 1-μg/m3 increase in daily PM 2.5 exposure is associated with 0.098 additional deaths per million elderly over the following three days, or a 0.025 percent increase relative to the average 3-day mortality rate.... Those aged 70-79 experiencing lower (and insignificant) increases in death rates than those aged 64-69 despite having higher mean death rates.... The Instrumental Variables (IV) estimates are about five times larger than the corresponding [OLS] estimates in Panel A, suggesting that OLS estimation suffers from significant bias. The IV estimates imply that each 1-μg/m3 increase in daily PM 2.5 exposure corresponds to 0.605 additional deaths per million elderly over the following three days, or a 0.15 percent increase relative to the average 3-day mortality rate. The corresponding estimate for a one standard deviation increase in daily PM 2.5 is a 1.1 percent increase in 3-day mortality. Columns (2)-(6) show a largely monotonic relationship between the mortality effect of PM 2.5 and age, with each 1-μg/m3 increase in daily PM 2.5 causing 0.263 additional deaths per million among the 65-69 population but 2.050 additional deaths per million among the 85 and over population. However because the average mortality rate is also much higher for the older elderly, the relative mortality effects across age groups follow a U-shaped pattern: each 1-μg/m3 increase in daily PM 2.5 exposure increases 3-day mortality by 0.20 percent among ages 65-69, by 0.10 percent among ages 75-79, and by 0.18 percent among ages 85 and over. This pattern is somewhat unexpected, since, if sicker individuals are more vulnerable to pollution shocks, and if age is a good proxy for health, then we would expect relative mortality to increase monotonically with age. ...
...The association between PM 2.5, hospitalization, and medical spending is mixed: each 1-μg/m3 increase in daily PM 2.5 exposure is associated with significantly less inpatient spending and fewer hospital admissions, is not associated with spending on ER admissions, and is associated with significantly more ER admissions and visits. A more consistent story emerges from our IV approach (Panel B), which shows that increases in daily PM 2.5 increase both hospitalizations and inpatient spending, driven primarily by encounters that originate in the ER. The IV estimates imply that each 1-μg/m3 increase in daily PM 2.5 causes a highly significant increase in ER inpatient spending of over $15 thousand per million beneficiaries (relative to a mean of $13.7 million).... The overall admissions rate increases by 2.03 per million beneficiaries, an increase which also can be almost entirely explained by the 1.96 additional admissions originating through the ER. Finally, we estimate that PM 2.5 increases total ER visits, including visits that do not result in a hospital admission, by 2.29 per million beneficiaries.
A simple numerical exercise helps to illustrate the policy implications of our results. The average level of PM 2.5 decreased by 3.65-μg/m3 nationwide between 1999 and 2011.... The estimate reported in Column (5) of Table 4 implies that such a decrease saved 147,098 life-years annually among the 41 million Medicare beneficiaries alive in 2011.26 If we assign each life year a standard value of $100,000 each, the mortality reduction benefits of this decrease added up to about $15 billion in 2011. The EPA's calculation of the annual costs of meeting the 1990 Clean Air Act Amendment air quality standard increased from $19.9 billion to $43.9 billion between 2000 and 2010 (EPA 2011). Thus, the estimated $15 billion in annual mortality benefits represents a large fraction of the estimated annual costs of complying with air pollution standards during this period. By contrast, the reduction in hospitalization costs implied by our estimates is an order of magnitude smaller – about $0.93 billion annually.
Kris Krug/Flickr Fort McMurray, Alberta hub of Canada's oil sands industry
... Table 4 displays estimates of equation (1) when the outcome variable is the estimated 3-day life-years lost per million beneficiaries... Column (2) displays results when every decedent’s counterfactual life expectancy is set equal to the mean for the 2-year FFS population (11.6 years). This estimate implies that each 1-μg/m3 increase in 2 daily PM 2.5 increases life-years lost by 8.6 years per million beneficiaries. This same effect can also be obtained directly by multiplying the mortality effect of 0.746 in Column (1) by the mean life expectancy of 11.6.... Accounting for decedents’ age and sex reduces the estimated impact of PM 2.5 on life-years lost by 31 percent, to 5.9 life-years per million beneficiaries.... The estimate decreases by another 40 percent when the counterfactual life-years estimates account for previously diagnosed chronic conditions.... Finally, we estimate counterfactual life expectancy using the LASSO machine learning algorithm, which allows us to optimally incorporate over 1,000 additional predictors, as described earlier. This final estimate, ... is 24 percent smaller than estimates that account only for age, sex, and chronic conditions and implies that each 1-μg/m3 increase in daily PM 2.5 increases life-years lost by 2.7 years per million beneficiaries.... This final estimate may be close to the true value.
... Adding additional predictors when estimating life expectancy can substantially reduce the estimate of life-years lost due to pollution. This reduction can occur for two reasons. First, better survival models should predict lower remaining life expectancy for decedents on average....The mean life-years lost per decedent (“LYL per decedent”) decreases from 11.56 in the model with no predictors to 4.86 in the LASSO model. Second, a better survival model should also predict a more accurate distribution of predicted life expectancies among decedents. This matters if air pollution selectively kills individuals in this population who are systematically healthier (or sicker) than the average decedent. Indeed,... This second channel also plays a role in reducing the estimated life-years lost from improved survival modeling. While the average LYL per decedent decreases by only 0.43 per million when moving from LYL estimates based on age, sex, and chronic conditions to those based on the LASSO model, the estimated effect of PM 2.5 on LYL drops by nearly twice as much (0.85 per million). This indicates that the mortality effects of PM 2.5 tend to be larger among individuals with characteristics that LASSO associates with lower life expectancy, even after conditioning on age, sex, and chronic conditions.
The estimates... can also be used to describe the estimated counterfactual life-years lost among “compliers”: those individuals who died because of increases in wind-driven PM 2.5. This estimate can be compared to the average life-years lost among all decedents to shed light on whether those dying from increased pollution appear to be differentially healthy or frail compared to those who die on a typical day. The LYL per complier is calculated by dividing the estimated effect of increased PM 2.5 on life-years lost by the estimated mortality effect (the coefficient reported in Column 1).25 When life expectancy is modeled as a function of age and sex alone, those dying from pollution appear to have slightly longer life expectancies (7.9 years) compared to the average decedent (7.8 years). However, estimates that rely on chronic conditions or the LASSO model show the opposite pattern. In Column (5), those dying from pollution appear to have somewhat shorter life expectancies (3.6 years) compared to the average decedent (4.9 years).
The mortality rate effect of PM 2.5 decreases monotonically with life expectancy. A 1-μg/m3 increase in daily PM 2.5 increases deaths among those with life expectancy of less than one year by 18.9 per million. By contrast, the effect on those with life expectancies of 5-10 years is only 0.53 deaths per million, and the mortality rate effect for those with life expectancies exceeding 10 years is even smaller and not statistically different from zero.... Relative mortality also decreases monotonically with life expectancy, which is consistent with the notion that the sickest individuals are most vulnerable to pollution shocks....
Although beneficiaries with a life expectancy of less than one year are the most likely to be killed by air pollution, beneficiaries with a life expectancy of up to 10 years are also vulnerable.
Although beneficiaries in Column (1) have less than one year of life expectancy, their high mortality rate causes their number of life-years lost due to pollution to exceed that of any other group: 11.3 life years per million beneficiaries.... By contrast, among beneficiaries with a life expectancy of 5-10 years..., the life-years lost from pollution is only equal to 3.7. Although their life expectancy is high relative to those in Column (1), their mortality rate is much lower, resulting in a smaller loss of life years. Those with 1-2 or 2-5 years of life expectancy (Columns 2 and 3) fall somewhere in between, losing 8.2 and 6.7 life years per million beneficiaries, respectively, when PM 2.5 increases by 1 μg/m3.
Weighting the life-years lost coefficients from Table 5 by the respective sizes of the groups, we see
that the largest portion of the social cost of pollution is borne by those with a life expectancy of 5-10 years (30 percent of sample, 43 percent of burden), followed by those with a life expectancy of 2-5 years (12.7 percent of sample, 33 percent of burden). While the per capita burden is highest for those with the lowest life expectancy, the majority of the aggregate social burden falls on those with intermediate life expectancy (2 to 10 additional years).