July 15, 2020

Hydrochlorothiazide and risk of melanoma and non-melanoma skin cancer

Query 19-05: Hydrochlorothiazide and risk of melanoma and non-melanoma skin cancer.

This research is funded by the Drug Safety and Effectiveness Network (DSEN) and conducted by the following investigators: Cristiano Moura, Sasha Bernatsky, and the CAN-AIM Team.


  • In Canada, an estimated 7,800 new melanoma cases and 1,300 melanoma-related deaths will occur in 2019[1].
  • Though statistics for non-melanoma skin cancer (NMSC) are usually not reported in cancer registries, NMSC is thought to be the most commonly diagnosed cancer among Canadians[1]. There are two main types of NMSC, including basal cell carcinoma (BCC) and squamous cell carcinoma (SCC).
  • Significant risk factors for developing melanoma and NMSC include race/ethnicity, altitude, occupation and habits, iatrogenic immunosuppression, innate immunodeficiency, ultraviolet exposure and photosensitizing drug use.
  • Hydrochlorothiazide, a widely prescribed antihypertensive agent, can increase skin sensitivity to sunlight and ultraviolet radiation.
  • Pharmacoepidemiologic analyses performed to date are limited. For example, sun exposure data, family history of cancer, race/ethnicity and smoking status are not available or are substantially missing in prior published analyses. Also, competing risks could confound apparent associations.
  • Further investigations are needed to answer the research questions with more certainty.

Study aim:

  • To determine if hydrochlorothiazide is associated with non-melanoma skin cancer (i.e. squamous and basal cell carcinomas) in the adult patient population and, if so, to determine if there is a cumulative dose-response relationship.
  • To determine if hydrochlorothiazide is associated with the risk of melanoma in the adult patient population.

Study description:

  • CAN-AIM proposes to use data from three cohorts of the Canadian Partnership for Tomorrow’s Health (CanPath), all linked with administrative data collected by provincial Ministries of Health. The following cohorts are involved:
  1. The CARTaGENE cohort (Quebec)
  2. The BC Generations Project (BCGP – British Columbia)
  3. The Ontario Health Study (OHS – Ontario)
  • In some of the analyses, we will apply a recently developed method that combines multiple data sources to control for unmeasured confounding [2]. This will be part of the collaboration between CAN-AIM and CNODES and will have two approaches.
  • One makes use of the CanPath linked administrative and clinical data.
  • The other will be developed with MarketScan Commercial Database containing millions of healthcare claims information from large employers, managed care organizations, and Medicare/Medicaid programs in the United States. Unlike Canadian administrative data, Medicare/Medicaid data and subsets of MarketScan data include information on race/ethnicity and other important clinical covariates.

Current status:

Preliminary analysis with MarketScan data for the NMSC project is being finalized. We are currently working on the initial sample selection from the first of the CanPath cohorts of interest, OHS. We have also been granted access to more recent years of data from BCGP (up to December 2019), and we are applying for the same update with CARTaGENE cohort.

For more information, contact: Autumn Neville, Research Coordinator. [email protected]

  1. Public Health Agency of, C., et al., Release notice – Canadian Cancer Statistics 2019. Health Promot Chronic Dis Prev Can, 2019. 39(8-9): p. 255.
  2. Burne, R.M. and M. Abrahamowicz, Martingale residual-based method to control for confounders measured only in a validation sample in time-to-event analysis. Stat Med, 2016. 35(25): p. 4588-4606.