A Hypothesis to Reconcile Conflicting Conclusions in Studies Relating Depressed Mood to Later Cancer*
Bernard H. Fox
Boston University School of Medicine
In 1981 there appeared a paper reporting that among 2,020 male Western Electric Company employees, those whose MMPI Depression scale scores were highest among all of the MMPI clinical scales (to be designated high-d in this chapter) were 2.3 times as likely to die of cancer (that is, relative odds, or RO, = 2.3) during the 17 years following the MMPI administration as those whose high point was not the depression scale score (non-high-d; Shekelle et al., 1981). Since 1981, many references to that study have appeared, almost all citing it as evidence that high-d (i.e., depressed mood) predicts later excess of cancer mortality, and in many references, excess cancer incidence. In 1987, a 20-year follow-up by Persky, Kempthorne-Rawson, and Shekelle of the same cohort confirmed the existence of excess mortality, but time analysis by four 5-year intervals showed successive ROs of 2.6, 2.5, 2.1, and 1.6. A similar analysis of cancer incidence yielded successive ROs of 1.9, 2.0, 1.1, and 1.3.
In 1988 and 1989, four further studies appeared. In the first, Kaplan and Reynolds ( 1988) showed no relationship between scores on the Human Population Laboratory (HPL) depression scale and later cancer mortality over a period of 17 years ( 1965-1982) among either sex in a population sample of 6,848 men and women. When they restricted their investigation to men of the same ages and occupational levels as those of Shekelle et al. ( 1981), they also found no excess mortality. In the second study, Hahn and Petitti ( 1988) showed no relationship between scores on the MMPI depression scale and later breast cancer over a____________________