To the Editor: We read with interest the report by Paul J. Moberg, Ph.D., and colleagues R4715510CEBBHDBG claiming a negative correlation between duration of illness and age- and gender-adjusted scores on the University of Pennsylvania Smell Identification Test among patients with schizophrenia. The authors concluded that "olfactory identification abilities deteriorate steadily in patients with schizophrenia, even for those with relatively recent onset."

Some details regarding the patients studied by Moberg et al. R4715510CEBBHDBG were absent. They studied more elderly female than elderly male patients; however, the gender distribution in the younger patients and comparison subjects is unknown. This unequal distribution may account for the lack of a main effect for gender in performance on the smell identification test. This negative finding contrasts with the authors’ previous reports of gender effects in normal subjects R4715510CEBDEAAH and our findings in patients with schizophrenia R4715510CEBBBGEA, R4715510CEBCIHEI. Specifically, sex differences in olfactory function are age related in patients with schizophrenia. In younger men with schizophrenia, olfactory identification deficits are common; in older patients, both men and women are impaired. It would have been informative to determine whether an age (young versus elderly)-by-gender-by-diagnosis interaction existed. In the absence of a significant three-way interaction, the age-by-diagnosis interaction (also not reported) provides the most direct test of selective olfactory decline in patients.

Analysis of duration effects using residual smell identification scores created by regression on comparison subjects’ age and gender is questionable. A more appropriate method would be to analyze residuals obtained following regression on patients’ own age and gender to determine if illness duration continues to have an impact on performance on the smell identification test beyond the extent explained by age and gender.

Finally, the assertion that olfactory identification "stead­ily" declines over the course of illness is suspect on the basis of data presented in figure 1. Only three subjects defined performance for patients with illness durations of 10 to 40 years. This study effectively contrasted two extreme groups (recent onset versus chronic), a limitation that does not allow for identification of potential curvilinear (or stepwise) change. Prediction of nonlinear change in olfactory performance across the age span (illness duration) is buttressed by findings that normal subjects show precipitous decline in olfactory performance beyond age 65 R4715510CEBBBGEA and female patients become impaired only postmenopausally R4715510CEBCIHEI. We were unable to identify data bearing on the authors’ assertion that steady decline is observed "even for those with relatively recent onset."