Objectives: Earlier monitoring of all symptoms, hospital admissions, cancer diagnoses and causes of death during gemfibrozil treatment had raised some suspicions which called for further follow-up.
Design: Close monitoring of selected, potentially adverse events amongst treated subjects after a placebo-controlled trial and comparing occurrences to those in various untreated groups.
Setting: All participants of the Helsinki Heart Study (a controlled, 5-year, multi-clinic coronary heart disease (CHD) primary prevention trial with gemfibrozil and placebo) were offered gemfibrozil treatment and twice yearly follow-up for 3.5 years. Untreated groups in the source population and national cancer statistics were utilized in comparisons.
Subjects: Of the 2046 dyslipidaemic men initially randomized to gemfibrozil, 2002 survivors entered the 3.5-year follow-up; of the 2035 initial placebo men, 1992 continued to be monitored.
Interventions: Gemfibrozil was chosen for the follow-up by 66.3% of the gemfibrozil-treated and 68.5% of the placebo-treated men.
Main outcome measures: Gastrointestinal symptoms, surgery, strokes, cancer incidence, mortality by cause.
Results: Gastrointestinal symptoms remained more common in the original gemfibrozil group. After 8.5 years strokes numbered 32 (gemfibrozil) vs. 37 (placebo), violent deaths 16 vs. 14, and cancers 51 in both groups. Total mortality was equal during the original 5 years, but higher in the gemfibrozil group post-trial, leading to an 8.5 year mortality of 101 vs. placebo 83 (P = 0.19). This was mainly a result of higher cancer mortality in the gemfibrozil (30) than the placebo group (18, P = 0.08). An additional 18-month post-study registry follow-up disclosed 13 placebo and five gemfibrozil cancer deaths, altering the cancer mortality to gemfibrozil 35 vs. placebo 31 at 10 years.
Conclusions: The most plausible explanation for the discrepancy between cancer incidence and cancer-specific mortality, based mainly on comparison with untreated groups, is delayed diagnosis. The increased cancer and total mortality is most probably due to chance, based on the later reversal of trends.