Years of life lost due to premature deaths associated with air pollution: an ecological time-series study

ABSTRACT BACKGROUND: Exposure to air pollutants is associated with hospital admissions due to cardiovascular diseases and premature deaths. OBJECTIVE: To estimate years of life lost (YLL) due to premature deaths and their financial costs. DESIGN AND SETTING: Ecological time-series study carried out in São José dos Campos, Brazil, in 2016. METHODS: Data on deaths among residents of this city in 2016 were assessed to estimate the financial cost of premature deaths associated with air pollution. The diagnoses studied were ischemic heart disease, congestive heart failure and cerebrovascular disease, according to YLL. The fractions attributable to deaths associated with air pollutant exposure and to each potential year of life lost were calculated using negative binomial regression with lags of 0-7 days between exposure and outcome. Nitrogen dioxide, particulate matter (PM10) and ozone concentrations were included in the model and adjusted for temperature, humidity and seasonality. RESULTS: Exposure to particulate matter was significant at lag 3 days. There were 2177 hospitalizations over the study period, with 201 deaths (9.2%). Premature deaths led to 2035.69 years of life lost. A 10 μg/m3 increase in PM10 concentrations was correlated with 8.0% of the hospitalizations, which corresponded to 152.67 YLL (81.67 for males and 71.00 for females). The cost generated was approximately US$ 9.1 million in 2016. CONCLUSION: In this first study conducted in a medium-sized Brazilian city, using the YLL methodology, we identified an excess expense attributable to air pollution.

calculated that an excess of 650 hospitalizations, with a cost of US$ 600,000, was caused through an increase in the concentration of this pollutant (PM 2.5 ) by 10 μg/m³. 5 A study carried out in Canada using data from 2003 to 2007 also identified a significant association between exposure to nitrogen dioxide (NO 2 ) and carbon monoxide (CO) and occurrences of hemorrhagic and ischemic strokes, with relative risks of 1.46 and 1.36. This association showed dose-response behavior with highest risk values in the fourth and fifth quintiles of NO 2 concentration. 6 Exposure to air pollution due to particulate matter contributes to cardiovascular morbidity and mortality, such that exposure to PM 2.5 for a few hours would increase the RR of cardiovascular mortality by approximately 0.4% to 1.0% due to an increase of 10 μg/m³ at lag 1 day. Long-term exposure (some years) would increase the relative risk by between 1.06 and 1.76. 7 In an extensive review on 76 studies published between 2000 and 2018, Bazyar et al. 8 showed that exposure to air pollutants increased the relative risk or chance (odds ratio, OR) of the need for emergency care in emergency rooms and also of mortality due to cardiovascular diseases such as acute myocardial infarction and hypertension. In another extensive review on 41 studies, higher risk of death due to cardiovascular diseases through exposure to particulate matter was identified. 9 Positive associations between exposure to pollutants and hospitalizations leading to premature death were identified in a study conducted in Skopje, Republic of North Macedonia, using the Disability Adjusted Life Years (DALY) methodology. 10 Using this same approach, the economic impact of premature deaths associated with particulate matter concentrations in 29 Brazilian metropolitan regions was evaluated. A total of 20,050 deaths were found, resulting in a cost of US$ 1.7 billion annually. 11  All the studies conducted so far in Brazil have been in major metropolises. Thus, there are no studies in medium-sized cities in Brazil.

OBJECTIVE
The aim of this study was to estimate the cost of years of life lost (YLL) due to premature deaths in the city of São José dos Campos, Brazil, that were associated with exposure to PM 10 .

Place of study
São José dos Campos is a city located in the southeastern region of Brazil between the cities of São Paulo and Rio de Janeiro (23° 11' S and 45° 53' W). It occupies an area of 1,099 km 2 , has a population of approximately 700,000 inhabitants and has 12 hospitals. This city is an industrial, commercial and service center serving the eastern part of the state of São Paulo and the southern part of the state of Minas Gerais, with a total regional population of approximately two million inhabitants. Some important research centers are installed in this city, such as the National Institute for

Statistical analysis
This was an ecological time-series study on data relating to hospitalizations due to circulatory system diseases (International Hospitalizations were included in three groups. Ischemic heart disease (ICD-10 categories I-20 to I-25) (Group 1), congestive heart failure (I-50) (Group 2) and cerebrovascular disease (I-60 to I-64) (Group 3), which corresponded to 70% of all deaths, were selected from among causes in ICD-10 Chapter X. The premature deaths from these diseases were determined, and the financial cost of these premature deaths was calculated using the DALY methodology.
The DALY method is a summary measurement of health expressed by means of a standard indicator in time units (years).
It is obtained as the sum of two components: years of life lost (YLL) due to premature death (associated with a specific outcome) in relation to the estimated life expectancy; and years lived with disability (YLD), i.e. the time spent in an unhealthy condition. This method is consistent with the ideals from outcomes such as disease, injury or risk factors, as described by Murray and Lopez. 11 The YLL component of DALY is obtained through equation 1, as follows: (1) Where: r = discount rate (r = 0.03); K = weight-age modulation factor (K = 1); C = constant (0.1658); a = age at death event; L = life expectancy pattern at age a; and β = weight-age function parameter (β = 0.04).
The variable "a" refers to the age of the individual at the time of death, while the variable L refers to the pattern of life expectancy at age a. These were obtained from the Brazilian Institute for Geography and Statistics (IBGE). For each interval, the value at its onset was considered: for example, for a case within the age range 45-50 years the value was set at 32.2 years for males and 37.0 for females, regardless of age.
These two variables provided the data on the events analyzed that needed to be entered in order to estimate the number of years of life lost due to a given disease. This formula was programmed and developed in an Excel spreadsheet in which the original database was aggregated for calculation. An implementation was introduced in order to make life expectancy comparisons at the age of death considering the same spreadsheet file, thus providing a single output. This output consists of each individual YLL calculation for each event. A value of € 50,000 (euros) was assigned to each individual YLL.
The total number of YLL was obtained by summing the individual YLL results due to pollution-attributable disease mortality, i.e. the part of YLL that was due to the effects of atmospheric pollution within the outcome, using the calculation given in the formula. Thus, YLL due to pollution was obtained via the following formula: Where: YLL pol = YLL attributable to air pollution Σ YLL = sum of all individual YLL in the parsed database The significance level was taken to be alpha = 0.05. This study was conducted using data publicly available from the official source, from which it was not possible to identify the subjects.
Therefore, there was no need to submit the study protocol for approval by a research ethics committee.        Table 3).

This
The results obtained, and the respective percentages, according to the aggregated groups of diseases, are shown in Table 4.
It could be seen that the highest percentage of deaths occurred in According to sex, the average YLL was 9.81 (SD = 6.02) for males and 9.96 (SD = 6.40) for females, with no significant differences.
The YLL values according to sex and diagnosis group are shown in This study may have some limitations. Among these, it was developed using secondary data: even though these data came from an official source (DATASUS), they may have contained misdiagnoses, since the main purpose of DATASUS is to record financial information. In addition, the hospitalization data referred only to occurrences within the public system through SUS, thus excluding private hospitalizations or occurrences though health plans or health insurance operators. Furthermore, factors such as passive smoking, dietary habits and lifestyle were not considered because they are not available through DATASUS. The time period used in this study was one year only.

CONCLUSIONS
Despite these possible limitations, this study not only presented unpublished data from a medium-sized city in Brazil but also quantified the cost of premature deaths due to some cardiovascular diseases. If these deaths were avoided, the cost savings could be allocated to other healthcare needs in the city. In addition, it was possible to estimate the importance of YLL in an important economically active age group.