Sunday, January 12, 2014

The Finnish Mental Hospital Study

Studies Associated with the Trial

Dietary Prevention of Coronary Heart Disease: Long-Term Experiment: Observations on Male Subjects (1968) [1]
Effect of cholesterol-lowering diet on mortality from coronary heart-disease and other causes. A twelve-year clinical trial in men and women (1972) [2]
Blood coagulation studies in a long-term experiment on dietary prevention of ischemic heart disease (1974) [3]
Prevalence of cholelithiasis in men and women ingesting a serum-cholesterol-lowering diet (1976) [4] (no access)
Dietary prevention of coronary heart disease: the Finnish Mental Hospital Study (1979) [5]
Dietary prevention of coronary heart disease in women: the Finnish mental hospital study (1983) [6]

Participants and Diets

Middle aged male (aged 34-64) and female (aged 44-64) patients in two different mental hospitals, without evidence of prior CVD, participated in a cross-over trial.  See the table below:

Phase 1 (1959-1964)
Phase 2 (1965-1970)
Hospital N
Hospital K

The experimental diet differed in the following ways: “whole milk was replaced by an emulsion of soybean oil in skins milk (“filled milk”) with a mean fat content of 3.9%. 2) Butter and margarine were replaced by a brand of margarine especially prepared for this experiment and considerably more unsaturated than any of tile brands marketed in Finland (“soft margarine”). The mean contents of linoleic and linolenic acids in this margarine were 26.3 and 4.5%, respectively. The margarine as well as the filled milk were fortified with vitamins A and D. 3) The use of fat meats was discouraged. 4) Generous use of vegetable oils in cooking was recommended. The main effect of these dietary changes was that the milk fat present in the original diet was almost totally replaced by vegetable oils, mainly soybeans oil”.  Whereas, in the control diet “no intentional changes were made”. [1]

While there are differences in food intake between the hospitals, the differences between the diets were better controlled, except that fruit was 31% higher in the control diet for hospital N, sucrose was 16% higher in the control diet for hospital K and the common margarine used a lot more by the control group, particularly in hospital K, which may have been higher in artificial trans fats [5].  This speculation is supported by Ramsden et al as they found that TFA was higher in the control diet, particularly hospital K’s control diet

Artificial Trans Fats
Experimental Diet
Control Diet
Hospital N
Hospital K


Cholesterol levels were consistently lower and adipose tissue linoleic acid levels consistently higher during the experimental diet period.  Cholesterol levels shifted rapidly, while adipose tissue linoleic acid levels took a few years to completely change.  There was no significant change in triglycerides [2] [5]

Linoleic Acid in Adipocytes
Near the end of the 1st period
Near the end of the 2nd period
Hospital N
Hospital K

The closest to CHD events was a change in ECG pattern among a smaller group of patients* who were  also free of abnormal ECG patterns on entry.  The more appropriate measure for CHD events may be MD as "The intermediate ECG patterns seem to be only moderately specific for CHD and may bring in false positive cases unrelated to dietary conditions." [5]  This was lower in the experimental group for both men and women [5[6].

Deaths from various causes can be seen in the table below, although these aren't aged adjusted or measured in person years [2]

And after adjusting for total person years and age the pooled experimental diet periods had a lower CHD mortality in both men and women, but there was no significant difference in total mortality [2]

Besides the unintended dietary differences, the lack of adequate randomisation resulted in a number of confounding variables in the trial, though many of those were quite minor.  However, a notable confounding variable is that the use of a cardiotoxic medication (thioridazine) was lower in the experimental group in hospital N (0.82 vs. 1.79) and higher in the experimental group in hospital K (0.43 vs. 0.14), but was used less overall in the experimental group (0.63 vs. 0.97) [5].  Ramsden et al mentions that: Thioridazine is significantly associated with risk of sudden death (adjusted OR = 5·3; 95 % CI 1·7, 16·2; P = 0·004), ‘the likely mechanism being drug-induced arrythmia’”.  It could be argued that the substantially higher TFA consumption by the control group in hospital K (0.2% vs. 2.0%) and substantially higher use of cardiotoxic medication by the control group in hospital N can explain the differences in CHD events and CHD mortality, or at the very least explain how this study is an outlier.

Another problem for the trial was the turnover rate of patients in the hospitals.  The average annual loss of subjects was about 16% in hospital N and about 13% in hospital K.  The researchers tried to make up for this so while the initial number of subjects was 234 in hospital N and 172 in hospital K, the total number of subjects studied was 327 in hospital N and 254 in hospital K.  35% of the patients in hospital N and 44% in hospital K remained in the trial for the first whole 6 year period.  Those who stayed in the hospitals for less than a month were excluded from the study [1].  Also, “to obviate the difficulties which would have arisen from a six-year higher age range during the second period, the patient populations were 'rejuvenated' at the start of the second period by discarding 6 oldest annual cohorts and by admitting 6 new cohorts at the younger end of the age range.[5].

* The reported number of participants in each group is inconsistent across different papers.  In [5] it’s reported that there were a total of 676 men, with 444 in the experimental group and 478 in the control group (due to 246 being in both groups).  In [6] it’s reported that there were a total of 591 women, with 372 in the experimental group and 341 in the control group (due to 122 being in both groups).  However, in [2] it’s reported that there were 2276 men in the experimental group, 1902 men in the control group, 3598 women in the experimental group and 2836 women in the control group.  As the number of events and deaths matches the number reported in each study it’s likely that all patients in the hospital were on the diet and all patient mortality was recorded for the study, whereas smaller group consented to have their cholesterol measured and do ECGs.

** “In hospital K the triglycerides were consistently somewhat higher than in hospital N. This may find at least a partial explanation in the higher consumption of sucrose and perhaps also in the somewhat greater average corpulence in the former hospital” [5]

*** Replacing SFAs with PUFAs didn’t have an effect on blood coaguability [3]

****There wasn’t much difference in the incidence of gallstones [4]


  1. “Thioridazine is significantly associated with risk of sudden death (adjusted OR = 5·3; 95 % CI 1·7, 16·2; P = 0·004), ‘the likely mechanism being drug-induced arrythmia’”

    This is something I have to insist on with epidemiology apologists - how many CVD deaths in the past were due to cardiotoxic meds that have since been withdrawn or fallen from favour?
    No wonder deaths have fallen in the last few decades, it needn't have anything to do with diet.
    Consider also air pollution being diminished by tighter restrictions on vehicle exhausts.
    Confounders that are never entertained.

    1. I agree. You can also add smoking, artificial trans fats and other drugs (Warfarin, aspirin, ACE inhibitors, etc) to that list. If our diet got better then why has the prevalence of obesity, T2D and other diseases all increased