Weather + health: are there connections in real-world data?

Researchers are analyzing real-world data to identify associations between weather patterns and health outcomes to help life science companies improve drug development, treatment effectiveness and delivery of healthcare services. 

By | 4 minute read | September 1, 2020

Life sciences researchers have been increasingly exploring how weather acts as a variable in health outcomes through research projects. For example, weather can potentially impact disease progression or symptom severity. Or it can affect patients’ likelihood to return to their doctors and the timing of care delivery. In some cases, weather may alter treatment effectiveness or impact patient access to healthcare services.

IBM researchers are passionate about drawing insights from data. The following represents thoughts from my colleagues about some recent findings combined weather and administrative claims data sets:

Q. What inspired you to explore the connections between weather and health in your research?

Ellen Thiel, MPH, Epidemiologist, IBM Watson Health Life Sciences: Our team wanted to showcase examples of using linked weather and clinical data, so we explored the hypothesis that there is a correlation between severe migraine headaches and barometric pressure. Although there was some anecdotal evidence of weather-related triggers for migraines, there was a lack of real-world data published about this topic. Our study1 analyzed the connection between severe migraines and daily barometric pressure readings using administrative claims data linked to daily weather data. We found that healthcare encounters for severe migraine increased when there was an increase in maximum daily barometric pressure.

Elizabeth Packnett, MPH, Epidemiologist, IBM Watson Health Life Sciences: My team was motivated by some personal experiences with asthma. We also knew that there are a few ecologic studies that demonstrated an increase in asthma-related healthcare use when pollen levels are high but there has been a lack of patient-level analyses. With this new linked claims-weather data, we could add those details. Our study examined patient-level data and found the probability of an asthma-related inpatient or emergency room visit in children was significantly higher when pollen levels were high.2

Janna Manjelievskaia, PhD, MPH, Health Services Researcher, IBM Watson Health Life Sciences:

We examined the impact of multiple external and environmental factors, such as travel time associated with clinic appointments and weather, on the administration patterns of a medication given to patients undergoing chemotherapy.3 This medication must be given the day following chemotherapy receipt for optimal benefit, so timing is important. This usually requires a return visit the day after the chemotherapy administration to receive the medication, which can be difficult due to factors unrelated to the patient’s clinical condition. We found that weather conditions – such as extreme temperatures, precipitation, and high winds – were significantly associated with a lower likelihood of receiving the recommended medication overall and trended toward significant association with untimely receipt of the medication.

Q. What are some other potential connections between weather and health you might explore in the future?

Debra Irwin, PhD, MSPH, Epidemiologist, IBM Watson Health Life Sciences: Some of this work was presented at professional congresses last year, and there was a lot of interest from life science researchers about the impact of weather on the effectiveness of various drug classes and treatments. For example, which types of therapies might be more effective to help patients avoid exacerbations of chronic lung diseases during times of poor air quality? Or how might extreme weather events impact the ability of a patient with chronic disease to access healthcare?

Thiel: Our study looked at absolute daily barometric pressure, but we could further explore the connection between incidence of severe migraine events and changes in barometric pressure over an established time period. The change in pressure may have more impact than the exact barometric pressure measurement at one point in time.

Manjelievskaia: Unlike the asthma and migraine studies, weather does not necessarily impact the condition we studied. However, weather did affect a patient’s likelihood of returning to the doctor for subsequent treatments or visits. Novel data elements, such as weather, that go beyond the typical clinical measures that are standard components of outcomes research studies can be used to study the impact on how patients behave and interact with the healthcare system.

Q. What advice would you give to a researcher who is starting to use weather data?

Irwin: Utilizing weather data in observational research requires thinking through the impact that weather data may have on the study design. For example, pollen data is typically not collected during winter months in all regions of the country. Therefore, an asthma study that examined treatment effectiveness and integrated weather data may be done best by examining the impact of pollen during summer months. Researchers must consider how weather elements will affect the study design in multiple ways.

Thiel: Unlike other data sources, weather data brings in the complexity of possible regional differences that impact how conditions affect personal behavior and health. For example, people in northern states tend to be indoors more often during the colder winter months, while southern populations may spend more time indoors during summers. More time indoors may increase risk of exposure to infectious disease, such as COVID-19, so regional and seasonal weather may drive different public health outcomes depending on location.

Manjelievskaia: There are several layers of complexity when it comes to measuring the impact of weather. For example, precipitation might not be a factor by itself, but when combined with longer driving distances, it can make it more difficult for patients to get to their appointments or obtain medications. Weather data should also be interpreted in the context of the geographical regions included in the study. A temperature or snowfall that is considered extreme for one location may not be for another, so the same weather metric across multiple locations would not necessarily draw the same conclusions or have a similar impact.

Combining weather and administrative claims data presents an opportunity for researchers to think creatively and explore new connections between environment and health. The richness and diversity of real-world data that researchers can access and use will inform efforts to bring therapies to patients faster.

  1. Lew C, Min S, Thiel ER, Irwin DE. “Impact of Barometric Pressure Changes and Prophylactic Migraine Medication Use on The Occurrence of Severe Migraines: A Case-Crossover Study Using Linked Weather and Claims Data” (IBM Watson Health)
  2. Packnett E, Zimmerman N, Henriques C, Irwin D. “High Pollen Levels Increase the Likelihood Asthma-Related Inpatient Admissions and Emergency Department Visits in Children with Asthma- A Matched Case Control Study Using Patient-Level Claims and Linked Weather Data” (IBM Watson Health)
  3. Machaon Bonafede, Mark Hatfield, Tatiana Lawrence, Janna Manjelievskaia, Kristin Evans, Vicki Wing, Neel Shah.
“Incidence of febrile neutropenia and environmental factors observed with optimal receipt of pegfilgrastim via on-body injector or pre-filled syringe after chemotherapy.” ISPOR 2020: May 16-20, Orlando, FL