Background: Although previous immunotherapy studies have demonstrated that a local reaction does not predict a systemic reaction, no study has investigated whether a local reaction predicts a local reaction.
Objective: To determine whether a local reaction predicts a local reaction at the next immunotherapy injection.
Methods: A retrospective analysis of an electronic immunotherapy database over a 12-month period was performed at a single site that did not dose-adjust for local reactions. Total injections, small local reactions (less than or equal to the size of patient's palm), large local reactions (LLRs; larger than the patient's palm), systemic reactions, and whether a local reaction was followed by a local reaction were recorded.
Results: Between August 2005 and July 2006, 360 patients received a total of 9678 injections. Of all patients, 78.3% had at least 1 local reaction, and 7.5% had an LLR. The total local reaction rate was 16.3% (1574/9678), the small local reaction rate was 15.9% (1536/9678), and the LLR rate was 0.4% (38/9678). Of all local reactions followed by another injection, 27.2% were followed by a local reaction. The sensitivity and positive predictive value for a local reaction predicting a local reaction at the next injection were 26.2% and 27.2%, respectively. In contrast, the specificity for the absence of a local reaction predicting the absence of a subsequent local reaction was 85.5%. For LLRs, only 6.0% were followed by another LLR; the sensitivity, positive predictive value, and specificity were 5.2%, 6.0%, and 99.6%, respectively.
Conclusion: In a clinic that does not dose-adjust for local reactions, local reactions do not predict local reactions at the next immunotherapy injection.