Theses Doctoral

Modifiable Risk in a Changing Climate: Linking household-level temperature, humidity, and air pollution to population health

Quinn, Ashlinn Ko

Background: This dissertation comprises research conducted on two distinct projects. Project I focuses on the connection between household air pollution (HAP) from cooking with biomass fuels and blood pressure (BP); this research is situated in the context of a large randomized trial of a cookstove intervention in Ghana, West Africa. The setting of Project II, meanwhile, is the residential environment of New York City, where we explore temperature and humidity conditions in homes and relate these conditions to summertime heat wave risk and to the survival and transmission of respiratory viruses in the winter. Although these projects are quite distinct, each relates to the complex relationship between climate change and health. Reducing HAP to improve health (the focus of Project I) will simultaneously reduce climate change through a reduction in emissions of short-lived climate pollutants into the atmosphere. Meanwhile, furthering our understanding of heat and humidity levels inside urban residences (the focus of Project II) is crucial to our ability to protect health in light of projections for a changing climate. Domestic activities associated with heating, cooling, and cooking are thus very relevant both to human health and to climate change mitigation and adaptation.
Objectives and Methods: Our overall objective for Project I was to investigate exposure- response relationships between HAP and BP in a cohort of pregnant women taking part in the
Ghana Randomized Air Pollution and Health Study (GRAPHS). We first explored this association in a cross-sectional study (Chapter 1), in which we used 72-hour personal monitoring to ascertain levels of exposure among the GRAPHS women to carbon monoxide (CO), one of the pollutants emitted by traditional wood-fed cooking fires. These exposure data were collected at enrollment into the GRAPHS study, prior to the initiation of cooking with improved cookstoves. We investigated the association between these “baseline” CO exposure levels and the women’s blood pressure at enrollment into GRAPHS. A limitation of this study was that BP was only measured once. We followed this with a second study of 44 women drawn from the same cohort (Chapter 2), for whom we designed BP protocols using 24-hour ambulatory blood pressure monitoring (ABPM), the current gold standard for clinical diagnosis of hypertension. As we were not aware of any prior research in Africa that had employed ABPM, we also designed a parallel BP protocol using home blood pressure monitoring (HBPM) equipment for comparison with ABPM. The use of ABPM with concurrent personal CO monitoring enabled us to investigate hourly associations between CO exposure and changes in BP. We also evaluated BP in these women both before and after the cookstove intervention; this allowed us to investigate whether any changes in BP were associated with switching to an improved cookstove.
Our objectives for Project II were to understand the distribution of temperature and humidity conditions in a range of New York City homes during the summer and winter seasons, to evaluate the impact of structural and behavioral factors (e.g. building size, use of air conditioning, and use of humidifiers) on these conditions, and to build models that could help predict indoor conditions from more readily available outdoor measurements. We conducted this research in two ways. We first analyzed a set of indoor temperature and humidity measurements that were collected in 285 New York City apartments during portions of summers 2003-2011 and used these data to simulate indoor conditions during two heat wave scenarios, one of which was more moderate and the other of which was more extreme (Chapter 3). Second, we designed and conducted a new study in which temperature and humidity were monitored in a set of 40 NYC apartments between 2013 and 2015 (Chapters 4-6). This second study enabled us extend our research into the winter season, and also to explore how factors such as air conditioning and humidifier use impacted indoor temperature and humidity. We also investigated relationships between the monitored conditions, self-reported perceptions of the indoor environment, and symptoms that were experienced among household members.
Results: In the cross-sectional analysis of CO and BP in the GRAPHS cohort (Chapter 1), we found a significant positive association between CO exposure and diastolic blood pressure (DBP): on average, each 1 ppm increase in exposure to CO was associated with 0.43 mmHg higher DBP [0.01, 0.86]. A non-significant positive trend was also observed for systolic blood pressure (SBP). In our study of the acute relationship between CO exposure and BP (Chapter 2), we determined that peak CO exposure (defined as above the 90th percentile of the exposure distribution, or an average of 4.1ppm) in the two hours prior to BP measurement was associated with elevations in hourly systolic BP (4.3 mmHg [95% CI: 1.1, 7.4]) and diastolic BP (4.5 mmHg [95% CI: 1.9, 7.2]), as compared to BP following lower CO exposures. We also observed a non-significant trend toward lower BP following initiation of cooking with an improved cookstove. Lastly, we demonstrated that ABPM was a feasible and well-tolerated tool for BP assessment in a rural West African setting.
For Project II in New York City, we first determined that there was a great deal of variability in indoor summer heat index (HI) between homes in association with similar outdoor conditions, and that this variability increased with increasing outdoor heat (Chapter 3). Our simulation of a moderate heat wave led us to conclude that the hottest 5% of the homes would reach peak indoor heat index (HI) values of 39°C. In a more extreme heat wave simulation, HI in the hottest 5% of homes reached a peak of 41oC and did not drop below 34oC for the entire nine- day simulated heat wave period.
Our second indoor monitoring study yielded the following findings: in the summer season (Chapter 4), we found significant differences in indoor temperature and heat index according to the type of air conditioning (AC) in the home. Homes with central AC were the coolest, followed by homes with ductless AC, window AC, and no AC. Apartments on the top floor of a building were significantly hotter than other apartments regardless of the presence of AC. During the winter season (Chapter 5), median vapor pressure in our sample of apartments was 6.5mb. Comparing humidity levels in the apartments to a threshold of 10mb vapor pressure that has been proposed as protective against influenza virus transmission, levels of absolute humidity in the homes remained below this threshold for 86% of the winter: a total of over three months. Residential use of humidifiers was not associated with higher indoor humidity levels. Larger building size (above 100 units) was significantly associated with lower humidity, while the presence of a radiator heating system was non-significantly associated with higher humidity. Lastly, perceptions of indoor temperature and measured temperature were significantly associated in both the summer and the winter (Chapter 6), while sleep quality was inversely related to measured indoor temperature in the summer season only. Reports of heat- stress symptoms were associated with perceived, but not measured, temperature in the summer season.
Conclusions: The work presented in this dissertation adds to a growing body of evidence on the importance of exposures in the domestic environment to health and well-being. The research reported here on household air pollution in Ghana documents an exposure-response relationship between air pollution from cookstoves and elevations in blood pressure, on both a chronic and an acute basis. As elevated BP is a known risk factor for cardiovascular disease (CVD), our research provides support for a plausible factor linking HAP exposure to CVD. Meanwhile, our research on temperature and humidity in New York City residences provides concrete data to supplement the very slim literature to date documenting these conditions in the home environment, where Americans spend over half their time. We conclude, first, that AC may not be fully protective against summertime heat risk, and second, that the levels of humidity we observed in residential environments are consistent with levels that have been shown to promote enhanced survival and transmission of respiratory viruses in experimental settings. We suggest that interventions that can reduce exposure to household air pollution and excess indoor heat can also mitigate climate change, and that with thoughtful planning we can improve health at the same time as we foster resiliency in the face of a changing climate.


  • thumnail for Quinn_columbia_0054D_13578.pdf Quinn_columbia_0054D_13578.pdf application/pdf 11.3 MB Download File

More About This Work

Academic Units
Environmental Health Sciences
Thesis Advisors
Kinney, Patrick L.
Shaman, Jeffrey L.
Ph.D., Columbia University
Published Here
October 7, 2016