Impact of National Cancer Institute's Common Toxicity Criteria and Common Terminology Criteria for Adverse Events on Quality of Treatment Related Harms Reporting: An Analysis of National Cancer Institute's Co-Operative Group Phase III Randomized Controlled Trials

Informed decision making in health care requires balancing of information related to benefit and harms of treatments. However, it is shown that the amount of space dedicated for presenting information on harms in randomized trials is less than that used to list authors' names and affiliations. Poor or absent reporting of treatment-related harms can mislead decision makers at all levels (public, researchers, funding agencies). Several efforts have been made to improve the quality of data collection and reporting of treatment-related harms. In oncology, after recognizing that many cancer treatments are toxic, and not infrequently result in severe morbidity and even death, the National Cancer Institute's (NCI) lead the way to standardize terminology and data collection on treatment-related harms. The first standards for collection of harms data in cancer known as Common Toxicity Criteria (CTC) were developed in 1982 followed by three revisions. Here we investigate the impact of NCI CTC and NCI common terminology criteria for adverse events (CTCAE) on quality of reporting of treatment-related harms in a cohort of NCI cooperative group (NCI-COG) trials.

We extracted data on elements addressing assessment of treatment-related harms from protocols and matching publications from all consecutive phase III NCI-COG randomized controlled trials (RCTs) for the years 1955 to 2006. RCTs reporting treatment-related mortality data as deaths per arm and treatment-related morbidity data as per patient per arm were categorized to have good quality. RCTs reporting treatment-related morbidity data in the aggregate method such as % of events per arm were categorized as intermediate quality while RCTs which did not report any treatment-related harms data were classified as poor quality.

We reviewed 429 RCTs enrolling 157,337 patients. Twenty seven percent (115/429) of RCTs had good quality of reporting of treatment-related harms while 63% (271/429) of RCTs were of intermediate quality. Only 10% (43/429) of RCTs had poor quality of reporting of treatment-related harms. Quality of treatment-related harms reporting did not improved with time. Nine percent (37/429) of RCTs were published before introduction of NCI CTC v1.0 while 66% (284/392) were published after NCI CTC v1.0 but before introduction of NCI v2.0 and 85% (92/108) of RCTs were published after NCI CTC v2.0 but before NCI CTCAE (figure). The quality of treatment-related harms reporting improved after introduction of NCI CTC v1.0 (p-value=0.002) while NCI CTC v2.0 (p-value= 0.78) and NCI CTACE (p-value=0.001) did not have positive impact on treatment-related harms reporting (figure). The NCI CTC/NCI CTCAE or similar standardized treatment-related harms assessment instrument was explicitly referred to in 55% (235/429) of protocols vs. 47% (203/429) of publications. Forty five percent (193/429) of protocols mentioned seriousness of treatment-related harms compared to 33% (142/429) of publications. Severity of treatment-related harms was mentioned in 63% (272/429) of protocols versus 71% (303/429) of publications.

Quality of treatment-related harms reporting in NCI-COG RCTs did not improved with time. The introduction of NCI CTC initially did improve quality of treatment-related harms reporting. However, NCI CTC V2.0 and NCI CTCAE did not subsequently appear to impact reporting of quality of treatment-related harms in NCI-COG RCTs.

No relevant conflicts of interest to declare.