Pernicious anemia as a risk factor in gastric cancer. The extent of the problem.

We have analyzed the prevalence of pernicious anemia (PA) in all patients with autopsyverified gastric carcinoma (GC) registered during 1958-76 in a Swedish city, with one Department of Pathology serving approximately 240 000 inhabitants and with a high autopsy frequency. The prevalence was compared to that in a reference group of 917 sex- and age-matched individuals without GC. PA was found in 19 persons, 2.1%, (7 males and 12 females) in the GC group and in 13 persons, 1.4%, (5 males and 8 females) in the reference group. This difference is not significant (p less than 0.15). After exclusion of individuals with PA known for less than 5 years before death, the corresponding figures were 16 in the GC group against 8 in the control group. This difference is significant (p less than 0.05). Regarding individuals with a diagnosis of PA for greater than 10 or greater than 15 years, the prevalence was twice as high in the GC as in the reference group. The differences were again insignificant (p less than 0.10). PA as a risk factor for the development of GC has been overestimated probably due to inadequate reference groups. Only 2% of all GCs are associated with PA. Large-scale preventive search for GC in PA patients is thus unjustified.