Measuring cancer-related adverse events

Quantifying the impact of adverse events in cancer care is an important gap area in quality measurement. 

Because cancer clinical trials are constructed to test novel treatments in narrowly defined patient populations, we don’t have a rigorous epidemiological evidence base to understand errors and injuries that occur across the continuum of care and among those receiving standard therapies.

Most importantly, we don’t yet have a systematic way to identify the potentially preventable harms that the health system should be striving to avoid.

Looking to address this gap, a team led by Saul Weingart, MD, MPP, PhD, chief medical officer at Tufts Medical Center, is seeking to create and test an automated, electronic “trigger” tool for identifying adverse events among breast, colorectal, lung and prostate cancer patients. 

This research was published in the Journal of Clinical Oncology.

This project builds on previous research done at Memorial Sloan Kettering Cancer Center with collaborators at Tufts Medical Center and the University of Utah that identified 42 potential triggers — clinical indicators that suggest the possibility of an adverse event (e.g., abnormal laboratory test values).

Weingart’s team used diagnosis and procedure codes from claims data in the OptumLabs Data Warehouse to narrow the list of 42 potential triggers to 16 oncology-specific triggers. 

In an analysis, they showed that the prevalence of these triggers is quite high, ranging from 6% to 49% for the four cancer types (breast, colorectal, lung, prostate), with significantly higher rates among those with metastatic disease. 

The team has been engaged in a measure validation process to understand whether triggers by cancer type and treatment are associated with excessive mortality, inpatient readmissions and overall resource utilization. 

They found that many of these triggers are associated with greatly increased mortality risk (approximately 2.3x–4.5x the risk) within one year of index cancer treatment. 

Ultimately, the project team hopes the work can be extended to integrate cancer-specific trigger tools into the electronic health record. This innovation would allow clinicians to identify and address treatment-related adverse events in real time.