PT  - JOURNAL ARTICLE
AU  - Barrett, Bruce
AU  - Harahan, Brian
AU  - Brown, David
AU  - Zhang, Zhengjun
AU  - Brown, Roger
TI  - Sufficiently Important Difference for Common Cold: Severity Reduction
AID  - 10.1370/afm.698
DP  - 2007 May 01
TA  - The Annals of Family Medicine
PG  - 216--223
VI  - 5
IP  - 3
4099  - http://www.annfammed.org/content/5/3/216.short
4100  - http://www.annfammed.org/content/5/3/216.full
SO  - Ann Fam Med2007 May 01; 5
AB  - PURPOSE We undertook a study to estimate the sufficiently important difference (SID) for the common cold. The SID is the smallest benefit that an intervention would require to justify costs and risks. METHODS Benefit-harm tradeoff interviews (in-person and telephone) assessed SID in terms of overall severity reduction using evidence-based simple-language scenarios for 4 common cold treatments: vitamin C, the herbal medicine echinacea, zinc lozenges, and the unlicensed antiviral pleconaril. RESULTS Response patterns to the 4 scenarios in the telephone and in-person samples were not statistically distinguishable and were merged for most analyses. The scenario based on vitamin C led to a mean SID of 25% (95% confidence interval [CI] 0.23–0.27). For the echinacea-based scenario, mean SID was 32% (95% CI, 0.30–0.34). For the zinc-based scenario, mean SID was 47% (95% CI, 0.43–0.51). The scenario based on preliminary antiviral trials provided a mean SID of 57% (95% CI, 0.53–0.61). Multivariate analyses suggested that (1) between-scenario differences were substantive and reproducible in the 2 samples, (2) presence or severity of illness did not predict SID, and (3) SID was not influenced by age, sex, tobacco use, ethnicity, income, or education. Despite consistencies supporting the model and methods, response patterns were diverse, with wide spreads of individual SID values within and among treatment scenarios. CONCLUSIONS Depending on treatment specifics, people want an on-average 25% to 57% reduction in overall illness severity to justify costs and risks of popular cold treatments. Randomized trial evidence does not support benefits this large. This model and these methods should be further developed for use in other disease entities.