Abstract
While replicating the association between expressed emotion and relapse in schizophrenia at five Sydney hospitals [1, 1,2], we observed an intense interest in orthomolecular treatments among patients and their relatives. Despite their awareness that orthomolecular medicine was considered unorthodox by conventional psychiatry, eight patients (9.5%) had already had a course of orthomolecular treatment either alone or in addition to antipsychotic medications. Reports in the popular press keep orthomolecular ideas current and make them the major alternative medical approach for schizophrenia. In 1997, Jorm et al. [3] reported the astonishing finding from the Australian Bureau of Statistics that as a treatment for schizophrenia, 57% of respondents indicated vitamins, minerals, tonics and herbs, 45% indicated special diets, and only 23% indicated antipsychotics as helpful.
The term orthomolecular medicine was originated in 1968 by Nobel laureate Linus Pauling [4]. He defined orthomolecular medicine as:
The achievement and preservation of good health and the prevention and treatment of disease by regulating the concentration of molecules that are normally present in the human body. Important orthomolecular substances are the vitamins, especially vitamin C.
Orthomolecular medicine is highly eclectic and draws together diverse practitioners who place themselves outside the mainstream of modern medicine. Emphasis on diet and vitamins is central to these approaches which have been considered useful for many conditions in addition to schizophrenia, most notably: the common cold; cancer; drug addiction; wound repair; atherosclerosis; and hyperactivity in children.
Orthodox nutritionists [5] have been critical of the lack of evidence supporting these approaches. Nevertheless, they remain popular with psychiatric patients and their relatives despite a critical position statement by the American Psychiatric Association in 1973 concerning niacin (vitamin B3) therapy [6] and a critical position statement by the Royal Australian and New Zeland College of Psychiatrists who suggested that controlled clinical trials of orthomolecular therapy be carried out [7].
Perhaps surprisingly, there has been an active early literature with trials reporting encouraging results with niacin. Most notably, Hoffer [8] reported a significant effect in a double-blind trial with 30 schizophrenic patients, and a second double-blind trial [9] replicated an effect with niacin in 171 patients. This study indicated that patients benefited most who were outpatients and already in remission. Length of time on niacin was important, but even those who were on niacin for short periods (less then 1 month) showed improvement. Findings of Denson [10] and Whittenborn [11] were also positive. However, a number of well-constructed controlled studies reported negative results with niacin [12, 13, 14, 15, 16]]. Hoffer's studies were criticised on methodological grounds and in turn Hoffer criticised the negative studies for using mostly chronic hospitalised patients which he had already reported were the least likely to respond.
Since Hoffer's studies with niacin, orthomolecular approaches became more diversified. Herjanic [17] reported encouraging results from adding 3 g ascorbic acid daily. Anath et al. [18] reported improved efficacy by adding pyradoxine. Reading [19, 19,20], one of the foremost exponents of orthomolecular psychiatry in Australia, reported that of 402 patients treated by him in 8 years, 90% had improved with a combination of orthomolecular approaches which included measurement of serum vitamins and the correction of deficiencies using large amounts of vitamins. It was hypothesised that these deficiencies were related to various food allergies, and diets excluding these foods were also prescribed. This technique offered a replicable procedure that could bring orthomolecular practices and therapies into conformity with orthodox methods of diagnosis and treatment.
Method
Selection
At the end of the expressed emotion study, a circular explaining the vitamin study was sent to 56 patients that were still living with their families. Twenty-nine patients indicated interest. Before admission to the study, seven dropped out leaving 22 patients who volunteered for the study.
Treatment and control groups
All patients had already satisfied diagnostic criteria for schizophrenia according to the Present State Examination [21] as previously reported [1, 1,2]. At entry to the study, random allocation was achieved by sealing patient's ID numbers in opaque envelopes, which were later distributed either to a megavitamin or a control group by an independent research worker.
At the start of the study, all patients were in partial remission and receiving phenothiazines. Three were also receiving antidepressants. Altogether, seven were partially employed. There were no differences between the two groups with respect to sex; there were seven males in both groups. The mean age was 32.1 in the vitamin group and 29.9 in the control group. The mean number of previous admissions was 2.8 in the vitamin group and 3.1 in the control group. The number of years since first admission was 7.4 in the vitamin group and 7.2 in the control group.
Outcome measures
Any patients readmitted to psychiatric units received a Present State Examination [21] to confirm psychotic relapse. In a 6-month trial, however, the relapse rate using this measure was expected to be so low as to lack sensitivity. To augment this, a self-report questionnaire (the Brief Symptom Inventory (BSI) [22], a 53-item check list) was used as a more general measure of distress and discomfort. The questionnaire was administered on six occasions: at entry to the trial and at the end of each month for 5 months.
In addition, a Behaviour Disturbance Inventory (BDI), a 45-item problem checklist, was completed by a key family member on six occasions: at entry, and at the end of each subsequent month for 5 months. Initially designed for parental report of problems in children [23], the content was altered to include problems usually reported by parents of their offspring with schizophrenia [24]. When two parents were living with the patient, each parent individually completed the questionnaire and the mean score was calculated.
Orthomolecular and control treatment
The results from the serum vitamin levels and the food allergy test were both sent to an established orthomolecular psychiatrist who assessed each blood and food allergy result. He then wrote out individual dietary advice and a megavitamin prescription for each individual patient on the basis of their serum vitamin levels and the radioallergosorbent test (RAST) [25].
Experimental group subjects were each given the individual medications as prescribed monthly. At the end of each month, the medications were delivered to the patients’ homes and the completed questionnaires were collected. Each patient received on average 18 tablets per day. The mean dosages of vitamin were: vitamin A, 6000 iu, SD = 5163 iu; vitamin B, 1345 mg, SD = 51 mg; vitamin B, 3520 mg, SD = 75 mg; vitamin B, 6223 mg, SD = 75 mg; vitamin B12, 25 mg; vitamin C, 2822 mg, SD = 909 mg; vitamin E, 204 mg, SD = 150 mg; folic acid, 5.2 mg.
Control group subjects were given tablets identical in character and quantity to the mean number given to the vitamin group. Tablets were all inert except for one, which was vitamin C 25 mg. The daily dose of this vitamin recommended by World Health Organization is 30 mg. This small dose of vitamin C was added to conform to the requirements of the Prince of Wales Ethics Committee. A dietitian was made available to consult with patients or family members by phone.
Dietary advice was based on the results of the RAST [25], which in orthomolecular medicine, is one of the most commonly used in vitro tests to measure levels of antibodies of IgE and other immunoglobulin classes in the serum to 18 common foods. Patients in the experimental group were given dietary sheets instructing them to avoid substances that tested positive. Most common foods were eggs, gluten, sugar, corn, beef and bread. The control diet was essentially a dietary challenge: patients were given sheets instructing them to eat foods testing positive, at least three times per week.
Results
After assessment but prior to commencement of the study, one patient allocated to the vitamin group and two patients allocated to the control group withdrew. One further patient withdrew from the control group in the first month of treatment. Results, therefore, are presented on 10 patients who received vitamin and dietary treatment and eight patients in the comparison group.
Mean vitamin serum levels at start and at end of trial
The serum vitamin levels at pre-treatment and at the end of the 5-month treatment period for the two groups of patients indicated a high level of compliance (Table 1). In the treatment group levels of Vitamin A, B1, B6, B12 and folate significantly increased. Vitamin C increased non-significantly with only Vitamin E showing a decrease. In marked contrast, the levels in the control group remained remarkably stable except for B6 which inexplicably fell dramatically.
During the first month of treatment, one control patient suffered a major psychotic relapse confirmed by Present State Examination and was hospitalised for 3 months. For the purposes of this study he was assessed as having deteriorated at each subsequent assessment interval (see Tables 2 and 3). Apart from this patient 97% of all BSI and BDI were completed satisfactorily.
There was a large amount of variation in the outcome scores and a small sample size, therefore it was thought that group comparisons would be less informative then comparisons based on classifying patients as ‘deteriorated’ or ‘improved’. An effect size (ES) of one standard deviation (SD) or less of symptomatic change was treated as indicating no change, falls greater then one SD indicated improvement, and increases greater then one SD indicated deterioration. The use of an effect size (standard difference score) of 1.0 is recognised as corresponding to clinically useful change (if a group has improved by 1.0 ES then 85% of patients are above the pretreatment mean).
Five-month alteration in symptoms assessed by Brief Symptom Inventory
Five-month alteration in disturbed behaviour assessed by Behaviour Disturbance Inventory
Alteration in mental states as assessed by self-report and disturbed behaviour as assessed by family members are reported in Table 2 and Table 3 in terms of numbers showing improvement, no change, or deterioration. At each time point, there was no relationship between treatment group and outcome category using Chi-squared analysis. Pearson's coefficient between the BSI and the BDI at the five time points was r = 0.71 (p < 0.0001).
A checklist of side-effects completed at the end of the study indicated few side effects with no difference between groups. One woman in the vitamin group previously diagnosed with a hiatus hernia suffered an acute episode of vomiting over a period of 3 days. She was admitted to an acute medical ward for parenteral replacement of fluids. After discharge, she completed the study without further symptoms.
Discussion
The megavitamin treatment administered over 5 months succeeded in increasing subjects’ serum levels of vitamin A, B1, B6, B12 and folate significantly. Vitamin C levels increased non-significantly with only vitamin E showing a decrease. In marked contrast, the levels in the control group remained stable except for B6, which inexplicably fell dramatically. No physical symptoms were noted with the vitamin changes in both groups. Apart from the changes in B6, there was no generalised fall in levels of vitamins in the control group that could indicate a malabsorbtion syndrome.
Levels of symptoms were measured at monthly intervals to identify onset and degree of any improvement. Comparisons of levels of self-reported symptomatology or behavioural disturbance at no point within the 5 months showed a consistent trend. A slight trend for some members if the vitamin group to show more improvement on the BSI is offset by a tendency for others to also show more deterioration. This greater deviance in outcome is mirrored to a lesser extent in the relatives’ report using the BDI. All trends, however, are weak and do not approach statistical significance.
Only one patient, the control patient who relapsed in the first week had sufficient deterioration to lead to admission. Psychotic relapse in this patient was confirmed using the PSE. The omission of an outcome measure such as the Brief Psychiatric Rating Scale [26] to measure outcome in the other patients is a weakness in this study. The BSI is designed not for measuring the levels of core schizophrenic symptomatology but to measure features in the mental state such as anxiety depression and interpersonal sensitivity. Typically correlation of the BPI and the BPRS averages 0.5 [27]. The BDI is more accepted as a valid measure of schizophrenic symptomatology. In this study, it correlated 0.71 with the BSI.
In studies with a small sample size, as in the present study, comparison of an effective therapy with a placebo can result in a trend for the therapy to be effective which does not reach statistical significance. Conclusion that the therapy is ineffective is erroneous: a type 2, beta or false negative error [28]. When a non-significant trend is noted, with small subject numbers, a replication of that study is indicated. However, there is no trend in the present study for the experimental group to have a better outcome than the control group.
Orthomolecular treatment using a combination of dietary and megavitamin treatment is the most commonly used alternative treatment for schizophrenia. We could find no evidence supporting its efficacy in the first 5 months of treatment.
The widespread divergence in views between members of the public and psychiatrists on the relative effectiveness of treatments for schizophrenia described by Jorm [3] is something that we in psychiatry are confronted with on a daily basis. In this era of evidence-based medicine, results from this study could be used by psychiatrists to inform patients and their relatives allowing them to play a more effective role in managing the patients’ mental health.
Footnotes
Acknowledgements
We are grateful to Dr Duson Hadlipablovic for statistical advice.