[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
Purchase Options:
[Skip to Content Landing]
Citations 0
December 20, 2000

Treatment of Mild Depression in Elderly Patients

Author Affiliations

Stephen J.LurieMD, PhD, Senior EditorIndividualAuthorPhil B.FontanarosaMD, Executive Deputy EditorIndividualAuthor

JAMA. 2000;284(23):2993-2994. doi:10.1001/jama.284.23.2993

To the Editor: Dr Williams and colleagues1 concluded that paroxetine showed moderate benefit for depressive symptoms and mental health functioning in elderly patients with dysthymia and more severely impaired elderly patients with minor depression. We feel that this conclusion is more optimistic than would be suggested by more precise expressions for the treatment effect, namely the absolute risk reduction (ARR) and the number needed to treat (NNT).2 It is possible to calculate the NNT from the authors' Table 3, which shows remission rates for patients attending 4 or more treatment sessions. For both dysthymia and minor depression, 52 of 106 (49.1%) patients receiving paroxetine reached remission, compared to 53 of 119 (44.5%) patients receiving placebo. The ARR is therefore 4.6%. In other words, the individual patient has a 4.6% chance of benefiting from paroxetine. However, the 95% confidence interval (CI) of the ARR is –8.5% to 17.6%.3 The interval contains 0 and so it can be concluded that the ARR is not statistically significant. Even the 90% CI does not reach significance (−6.6% to 15.5%). The NNT is 100/4.6 = 21.7. This indicates that about 22 older patients with minor depression or dysthymia need to be treated with paroxetine rather than standard or placebo treatment for 1 additional patient to benefit after 11 weeks of treatment. The 95% CI of the NNT goes to infinity because 0 is part of the 95% CI for the ARR. Thus, the 95% CI of the NNT (benefit) is 5.68 to infinity and the NNT (harm) is 11.7 to infinity.4 Given the ARR and NNT with their 95% CIs, we believe that paroxetine did not show benefit.

Finally, because the usual care in this group of patients is often less extensive than the care received by patients in the current study regardless of group, it is unfortunate that a group receiving usual care was not included in the design. In this type of more realistic experiment, it is possible that the magnitude of the effect for paroxetine would have been more substantial.

Williams Jr  JWBarrett  JOxman  T  et al.  Treatment of dysthymia and minor depression in primary care: a randomized controlled trial in older adults.  JAMA. 2000;284:1519-1526.Google Scholar
Cook  RJSackett  DL The number needed to treat: a clinically useful measure of treatment effect.  BMJ. 1995;310:452-454.Google Scholar
Gardner  SBWinter  PDGardner  MJ CIA (Confidence Interval Analysis) software program. Written for use with Gardner MJ, Altman DG, eds. Statistics With Confidence. London, England: British Medical Journal; 1989.
Altman  DG Confidence intervals for the number needed to treat.  BMJ. 1998;317:1309-1312.Google Scholar