Factors Contributing to the Development of Preventable Adverse Drug Events

Project number: 
Approval date: 
Monday, October 22, 2018
Principal Investigator: 
University of British Columbia (UBC)
Funding Agency: 
Canadian Institutes of Health Research(CIHR),Canadian Institutes of Health Research(CIHR)
Datasets requested: 
consolidation - census geocodes
Deaths (BC Vital Statistics Agency)
Hospital Separations (BC Ministry of Health)
Consolidation file (BC Ministry of Health)
Medical Services Plan (BC Ministry of Health)
Consolidation - demographic (Ministry of Health)
Research objective: 

Research Questions:

(i) What are patient-, provider-, and system-level risk factors for preventable ADEs among patients presenting to the emergency departments? What are their health services utilization and mortality outcomes?

Hypothesis (i):

We will identify several modifiable provider- and system-level risk factors for preventable ADEs that can be used to develop preventative strategies.

Based on an American study that found no difference in hospital length of stay between patients with preventable and non-preventable ADEs, we do not believe we will find differences in health outcomes between these groups.

(ii) What is the rate of unintentional re-exposures to culprit, same-class or contraindicated drugs within one year of an ADE diagnosis, and what are patient, provider, and system-level risk factors related to re-exposures?

Hypothesis (ii):

Based on a small Dutch pilot study (n=215), we hypothesize that at least 27% of patients diagnosed with an ADE will be re-exposed to the culprit or a same-class medication.

(iii) Is it feasible to develop an automated clinical decision rule with high sensitivity that identifies patients with ADEs using variables derived from administrative data?

Hypothesis (iii):

We believe it will be feasible be to develop an algorithm to classify patients into risk groups for ADE outcomes using administrative data. We believe that such an algorithm will have a target sensitivity of at least 90% while maintaining a specificity of 30% for ADE outcomes.

(iv) What are the sensitivity and specificity of administrative health data for key ADE outcomes?

Hypothesis (iv):

We hypothesize that the sensitivity of administrative data for key ADE outcomes will be highly variables. We hypothesize that administrative data will have poor sensitivity for the following outcomes: prescription adherence, delirium, hypotension, syncope, and electrolyte abnormalities. We hypothesize that the administrative data will have good sensitivity for the following outcomes will be high: bleeding events, thromboembolic events (e.g., deep venous thrombosis, pulmonary embolus, and cerebrovascular events), hypoglycemia, and arrhythmias.

Page last revised: January 18, 2019