Severe acute respiratory syndrome (SARS) is an emerging viral infectious disease, which has claimed 349 lives with 5,327 probable cases reported in mainland China since November 2002. 12 SARS case fatality has varied across geographical areas with higher fatality found in the northern areas of China. The SARS case fatalities of Beijing and Tianjin, two cities in the northern part of China, were 7.66% and 8%, respectively; while the case fatality rate of Guangdong, a province in the southern part of china was only 3.84%. Although sex ratio, age distribution, socioeconomic level, availability of health facilities, and clinical practices might be responsible for the geographical difference, air pollution could be an important determinant. Air pollution has been linked to acute respiratory inflammation, asthma attack, COPD exacerbation, and death from cardiorespiratory diseases by various studies. 3456 In this study, the relationship between air pollution and SARS case fatality was explored via an ecologic study design.

Publicly accessible data on SARS morbidity and mortality were utilized in the study. 2 Case fatality was estimated by dividing the number of reported deaths by the number of probable cases. Air pollution was evaluated by air pollution index (API) provided by the Chinese National Environmental Protection Agency (CNEPA). 7 CNEPA calculated individual pollution indices for five major air pollutants including particulate matter (PM10), sulfur dioxide (SO2), and nitrogen dioxide (NO2), ground-level ozone (O₃), and carbon monoxide (CO). The maximal individual pollution index was set as the comprehensive API for the monitoring area. In most of monitoring areas, PM10 was considered as a major pollutant. 7

API could be categorized into seven groups with 50, 100, 200, 300, 400, and 500 as the cut-off points. APIs less than 100 were thought healthy for general population. Since the majority of patients were diagnosed during April and May, the average of daily APIs collected within these months in each region was used for data analyses mainly. In addition, to evaluate potential influence of long-term exposure to air pollution on the case fatality rate, the average of APIs between June 2000 and October 2002 before SARS epidemic was also analyzed.

Ecologic analysis was conducted to explore the correlation and association between air pollution during April and May 2003 and SARS case fatality. A linear model was fitted to the data among five regions with 100 cases or more. Furthermore, we conducted data analysis by grouping SARS patients into three exposure levels and evaluating short-term exposure (April-May 2003) and long-term exposure (June 2000-October 2002), respectively. The cut-off points were 75 and 100 API. Case fatalities of patients from regions with high APIs (API>100) and patients from regions with moderate APIs (75–100) were compared to that of patients from regions with low APIs (API<75). Relative risks (RR) and 95% confidence intervals (CI) were calculated. Monotonic trend test with exposure scored as 1, 2, and 3 for each exposure group was also conducted to investigate whether case fatalities increased with the increase of APIs. All data analyses were performed in Statistical Analysis Software (SAS).

During the epidemic, 5,327 Chinese were diagnosed as SARS probable cases. Among them, 349 died from the disease and the rest recovered, which accounted for the case fatality of 6.55%. The average APIs ranged from 52 to 126 for different regions during April and May 2003. Ecologic analysis was conducted among 5 regions with 100 or more SARS cases (Guangdong, Shanxi, Hebei, Beijing, and Tianjin) to explore the relationship between air pollution and SARS case fatality. Inner Mongolia, which had 282 probable cases, was excluded from the analysis due to poor data quality reported by the Chinese Center for Disease Control. 8 4,870 SARS cases occurred in these five regions, which represented 91.4% of all cases in China. The APIs of Guangdong, Shanxi, Hebei, Beijing, and Tianjin during April and May were 75, 95, 98, 99, and 104, and the corresponding fatality were 3.84%, 5.36%, 5.58%, 7.66% and 8%, respectively. Data were fitted to the linear model: case fatality = -0.063+0.001*API, showing case fatality rates increased with the increment of API. The correlation coefficient between air pollution index and SARS fatality was 0.8568 (p = 0.0636). (See figure 1)

According to the air pollution data during April and May 2003, a total of 1,546 patients resided in regions with low APIs, 63 of whom died from SARS. Among 3,590 patients from regions with moderate APIs, 269 died from the disease. 17 out of 191 patients from regions with high APIs also died. The corresponding fatalities were 4.08%, 7.49% and 8.90%, respectively. SARS patients from regions with moderate APIs had an 84% increased risk of dying from SARS compared to those from regions with low APIs (RR = 1.84, 95% CI: 1.41–2.40). Similarly, SARS patients from regions with high APIs were twice as likely to die from SARS compared to those from regions with low APIs. (RR = 2.18, 95% CI: 1.31–3.65). The trend test showed that SARS case fatality increased with the increase of air pollution level (P for trend <0.001). (See Table 1)

Our study demonstrated a positive association between levels of air pollution and SARS case fatality in Chinese population by analyzing SARS data and air pollution indices. Although the interpretation of this association remains uncertain, the possibility of a detrimental effect of air pollution on the prognosis of SARS patients deserves further investigation.

SARS: Severe Acute Respiratory Syndrome

COPD: Chronic Obstructive Pulmonary Disease

API: Air Pollution Index

CNEPA: Chinese National Environmental Protection Agency

PM10: Particulate Matter

SO2: Sulfur Dioxide

NO2: Nitrogen Dioxide

CO: Carbon Monoxide

O3: Ozone

RR: Risk Ratio

CI: Confidence Interval

SAS: Statistical Analysis Software

None declared.

Yan Cui was responsible for data analysis, summarization of results, and manuscript preparation. Zuo-Feng Zhang took the overall responsibility in hypothesis generation, study design, data collection and analysis, and manuscript preparation. John Friones was involved in the generation of hypothesis, interpretation of results, and manuscript preparation. Jinkou Zhao and Hua Wang were involved in data collection and manuscript preparation. Shun-Zhang Yu was involved in generation of hypothesis and study design, and manuscript preparation. Roger Detels was involved in generation of hypothesis and study design, interpretation of results, and manuscript preparation.