Does the AQHI reduce cardiovascular hospitalization in Hong Kong’s elderly population?

Background: Air quality alert programs have been introduced around the world to reduce the short term effects of air pollution on health. Hong Kong, a densely populated city in southern China with high levels of air pollution, introduced its first air quality health index (AQHI) on December 30th 2013. However, whether air quality alert program warnings, such as the AQHI, reduces morbidity is uncertain. Using a quasi-experimental design, we conducted the first evaluation of the AQHI in Hong Kong, focusing on cardiovascular morbidity in Hong Kong’s elderly population. Method: Interrupted time series with Poisson segmented regression from 2010 to 2016 was used to detect any sudden or gradual changes in emergency hospital admissions for cardiovascular diseases (CVD), after the AQHI policy was implemented. To account for potential confounders, models were adjusted for air pollutants (NO2, SO2, PM10, O3), temperature and humidity. The findings were validated using a negative control and three false policy periods. We also assessed effects on specific subtypes of CVD (hypertensive disease (HPD), acute myocardial infarction (AMI), heart failure, stroke and other CVD) and by sex. Results: From January 1st 2010 to December 31st, 2016, 375,672 hospital admissions for CVD occurred in Hong Kong’s elderly population. Immediately after the policy HPD and AMI dropped by16% (relative risk (RR) 0.84, 95% confidence interval (CI): 0.78–0.91) and 15% (RR 0.85, 95% CI: (0.76–0.97)) respectively. There was no significant change for all CVD or other sub-types and no differences by sex. Conclusion: Hong Kong’s AQHI helped reduced hospital admissions in the elderly for HPD and AMI but had no effect on overall emergency hospitalization for CVD. To maximize health benefits of the policy, at risk groups need to be able to follow the behavioral changes recommended by the AQHI warnings. Keywords: AQHI, Cardiovascular diseases, Elderly, Interrupted time series, Segmented regression