Article type
Year
Abstract
Background: Symptom relief (e.g. pain) is a major goal of therapy for many palliative cancer treatments. The estimation of absolute and relative effectiveness is dependent on how treatment response is defined. The incorporation of multiple different endpoint definitions represent a significant barrier to qualitative and quantitative systematic reviews. Outcomes of these types of studies not infrequently are dependent on the way response definitions are set. An endpoint that is less sensitive to variations in response definitions would be desirable. The current project evaluates the impact of two different endpoint definitions on the estimation of effect size on pain relief, and explores the hypothesis that symptom progression is a potentially more robust endpoint measure.
Methods: A prospective database was maintained for all patients referred to our Rapid Response Radiotherapy Program where pain and analgesic dairy is maintained pre treatment and 1,2,4, 8, and 12 weeks post treatment. Pain score was expressed on an 11 point categorical scale. Analgesic consumption was converted to morphine equivalent to facilitate comparison where necessary. Pain response was classified using 2 different criteria. 1. "Pain score only (2" (response: pain reduced by ( 2) 2. "Integrated pain and analgesic (2 " criteria (response: (response: pain reduced by ( 2, with at least no increase in analgesic requirements or at least stable pain score with a (50% reduction in analgesic requirements)
Results: Between January -August 1999 inclusive, 115 patients who received radiotherapy for bone pain were included. At the time of consultation, 58% of patients were on strong opioids. 46% of patients had a pain score (5. Using the pain and analgesic data at 2 weeks, when "pain score only" criteria vs "integrated criteria" was used, the proportion of patients classified as having achieved a response was 47% vs 35%, stable pain 14% vs 29%. The magnitude of variation in response is much smaller for patients with progressive pain, which was 10% vs 8% respectively. Using the McNemar's test, the difference was statistically significant for response and stable disease, but not for symptom progression.
Conclusions: Our data supports the hypothesis that the variability in response rates for subjective symptoms such as pain lies in shifts in the classification between "response" and "stable" categories. The proportion of patients whose symptoms do not respond to the treatments, perhaps report larger magnitude of decreases in their response scores that minimizes its sensitivity to response definitions. Upon further confirmation of this observation in other symptom measurement studies, it may prove useful to use symptom progression as a more robust endpoint to facilitate systematic reviews and metaanalysis in topic areas where subjective symptom response is the primary endpoint.
Methods: A prospective database was maintained for all patients referred to our Rapid Response Radiotherapy Program where pain and analgesic dairy is maintained pre treatment and 1,2,4, 8, and 12 weeks post treatment. Pain score was expressed on an 11 point categorical scale. Analgesic consumption was converted to morphine equivalent to facilitate comparison where necessary. Pain response was classified using 2 different criteria. 1. "Pain score only (2" (response: pain reduced by ( 2) 2. "Integrated pain and analgesic (2 " criteria (response: (response: pain reduced by ( 2, with at least no increase in analgesic requirements or at least stable pain score with a (50% reduction in analgesic requirements)
Results: Between January -August 1999 inclusive, 115 patients who received radiotherapy for bone pain were included. At the time of consultation, 58% of patients were on strong opioids. 46% of patients had a pain score (5. Using the pain and analgesic data at 2 weeks, when "pain score only" criteria vs "integrated criteria" was used, the proportion of patients classified as having achieved a response was 47% vs 35%, stable pain 14% vs 29%. The magnitude of variation in response is much smaller for patients with progressive pain, which was 10% vs 8% respectively. Using the McNemar's test, the difference was statistically significant for response and stable disease, but not for symptom progression.
Conclusions: Our data supports the hypothesis that the variability in response rates for subjective symptoms such as pain lies in shifts in the classification between "response" and "stable" categories. The proportion of patients whose symptoms do not respond to the treatments, perhaps report larger magnitude of decreases in their response scores that minimizes its sensitivity to response definitions. Upon further confirmation of this observation in other symptom measurement studies, it may prove useful to use symptom progression as a more robust endpoint to facilitate systematic reviews and metaanalysis in topic areas where subjective symptom response is the primary endpoint.