Some problems in the conduct of psychotropic drug trials (a review).
"Doctors use drugs of which they know little, to treat diseases of which they know still less, in patients of whom they know nothing at all".
The above words which were written over a hundred years ago by Voltaire, the French revolutionary, aptly describe the current states of the art with reference to the conduct of drug trials with psychotropic agents. This is because meaningful communication in psychopharmacology is limited in the absence of finite knowledge.
Psychiatry is defined as a branch of medicine which deals with the causes, clinical manifestations, and treatment of diseases of the mind. However, in many instances the specific cause of a psychiatric illness remains unknown. The symptoms are often not specific, and the treatment can best be described as "hopeful".
Basically, there are three kinds of drugs-
- those in which the action is known and the assessment is direct and accurate.
- those in which the action may be fairly well known and the assessment is secondary and empirical.
- and those in which the action is poorly known and the assessment is indirect and imprecise.23
Most psychiatric drugs belong to the last category. None of them belongs to the first category.
There are certain definite differences between psychotropic and other drug trials. These differences are centered around problems of classification, conduct and computation.
[TAG:] PROBLEMS OF CLASSIFICATION[/TAG:2]
There is little agreement between clinicians in evaluating the symptoms and signs of a psychiatric disease or in establishing its prevalence. For example, it has been reported that 50-90% of clinicians agree to a diagnosis of an organic brain syndrome in a given instance, while only 8-40% will agree to a diagnosis of a personality disorder. It has also been demonstrated that psychiatrists in the United States diagnose schizophrenia much more frequently than psychiatrists in U.K., while the reverse is true for diagnosing depression. The diagnosis of schizophrenia has been reported to vary not only from country to country but also from state to state, from hospital to hospital, and from unit to unit."Anxiety" is an ubiquitous term. Anxiety may exist as a trait or as a state. Anxiety can also be a manifestation of an organic disease or of some other psychiatric disease. The response to anxiolytic agents may be different in treating trait v/s state anxiety. Unfortunately, drug trials on anxiolytic agents often do not take this into consideration.
In earlier times depression or melancholia was thought to be a single disease of varying intensity. According to Sir Aubrey Lewis, there is a disease continuum ranging from normality, which passes through the next disease state of anxiety, terminating in the end state of depressive disease. Sir Martin Roth subsequently differentiated between depression which is reactive and due to depressive circumstances, from depression which is endogenous.,  Thus, we have two kinds of depression. It was later shown that there are three kinds of depression, viz., those which can be treated by drugs which inhibit nor-adrenaline, drugs which inhibit serotonin, and drugs which inhibit dopamine., ,  To make confusion worse confounded the present trend seems to be to identify four kinds of depression, viz., psychotic depression, anxious depression, "young hostile" depression and depression with personality disorder.
Is schizophrenia a single disease or is it a group of diseases? The controversy is still not settled. In process or nuclear schizophrenia, the genetic loading seems to be important while in a schizophrenic reaction, environmental factors may be almost as important. From the biochemical point of view paranoid schizophrenia may be distinct from non-paranoid schizophrenia depending upon platelet concentration of monoamine oxidase, and schizophrenia which is acute seems to respond differently to neuroleptics as compared to chronic scbizophrenia.
Even the traditional classification of psychotropic drugs into minor tranquilisers, major tranquilisers, antidepressants, etc., seems to be lacking precision. A minor tranquiliser is often used as a hypnotic while a major tranquilizer in small doses can be used to treat anxiety. Lithium is used for the treatment of mania and depression. This begs the question-should it be classified as an antidepressant or as an antieuphoric? It is because of such problems that Lehman et al have proposed the following classification of psychotropic drugs.
Classification of psychotropic drugs:
1. Arousal : (a) Stimulants,
2. Affect : (a) Anti-depressants,
3. Integration: (a) Psychostatic,
The design of a psychotropic drug trial is often dictated by the drug itself and the clinical situation in which it has to be tried. Certain questions have to be answered by the time the trial is completed, viz., is the drug effective? is it safe? on which particular symptom is it specially effective? and at what dose?
A drug trial may go wrong if specific questions are not asked, or if the asked questions are not answered, or when a specific question is not clearly answered, or when the answers obtained are for questions which are different from those which have been asked.
1. Number of subjects
A larger number of subjects may be required to conduct psychotropic drug trials as compared to other drug trials owing to the fact that the study design often involves parallel groups rather than the patient acting as his own control. It may thus become necessary to wait for a considerable length of time before suitable patients present themselves. In contrast to other drug trials, patients often have to be matched for socio-economic variables like marital status, education, religion, etc; and thus Lasagna's law often becomes operative when psychotropic drug trials are being conducted. This, in turn, may necessitate a multicentric trial using more than one investigator and producing problems of co-ordination between the various centres, as there may be differences in diagnostic criteria, types of cases selected, clinical and therapeutic procedures, etc., at the different centres.
2. Patient co-operation
This variable becomes crucial in psychotropic drug trials as quite often the trial may be long in duration and may involve a large number of patients who may be severely psychotic and therefore noncompliant. Considerably more time is required at the time of serial evaluations, as the clinical investigator has to conduct a mental status examination and/or use elaborate rating scales. Most psychotropic drugs have a delayed onset of action as they do not readily cross the blood-brain barrier. Dosage schedules have to be taken into account as patient compliance would be obviously better when following a single daily dose schedule rather than a t.i.d. schedule. Patient compliance may again be an important variable as many psychotropic drugs have side effects like drowsiness, tremors, blurring of vision, etc., which may interfere with the subject's day to day functioning.
3. Patient variables
Many psychiatric disorders and particularly anxiety states and depressive disorders, may be strongly influenced by non-drug variables, such as age, sex, marital status, education, occupation, religion, menstruation, premorbid personality etc. For example, when comparing the response to an anxiolytic drug in two groups of patients, subjects in both index and control groups must be matched for age, because there is a suggestion that anxiety after the age of 40 years or in the middle aged is often a symptom of an underlying depression. Similarly, a demographic variable like religion may indirectly affect the symptom variable "feelings of guilt" in antidepressant drug trials. This is especially so in a country like India with diverse religions, customs and ethical beliefs and dogma. The patient's premorbid personality and the phenomenon of placebo-genesis can probably affect the results of a psychotropic drug trial more so than other drug trials, as psychiatric patients are notorious placebo responders.
4. Physician variables
Peculiarly enough, the phenomenon of placebogenesis also seems to be a function of certain variables in the treating physician for which the term "iatroplacebogenesis" has been coined. Some of the variables in the investigating physician which may influence psychotropic drug trials are his training, orientation, perseverence, and the enthusiasm with which he conducts the trial. The success of many psychiatric treatments like psychotherapy, behaviour therapy and also psychopharmacotherapy depends upon the investigator's interest, his attitude towards the patient and the treatment under investigation, and the results of the treatment.
5. Laboratory data
Unlike demographic variables, extensive laboratory investigations are generally not required to be studied in psychotropic drug trials as far as assessing the effect of the drug is concerned. This is in contrast to other drug trials, e.g., trial of an antibiotic in upper respiratory tract infections, where laboratory investigations like serial blood counts and sputum examination are very important.
Certain points must be kept in mind when identifying the duration during which a psychotropic drug trial should last. Obviously, the longer the trial the greater the chance of attrition of subjects participating in the trial, of the initial raters changing subsequently as the trial progresses, etc. The shorter the trial the lesser the chance of generating meaningful data, because of the prolonged nature of most psychiatric illnesses and because of the pharmacokinetic properties of the drug. Recurrences and remissions can also create confusion, e.g., trials where the prophylactic properties of lithium are being studied in manic depressive illness. Many psychotropic drugs have a sustained duration of action. This may necessitate that placebo washout periods are incorporated not only at the beginning but also at the end of the drug trial period. This is most important when using a cross-over design.
In spite of heroic efforts invariably a substantial number of patients may dropout from the trial before the end of the study.
A patient may discontinue because he feels better or because he does not feel better, or because his symptoms have worsened, or because lie has developed side effects, or due to social and personal reasons which have no bearing on his continuing in the trial. If the trial involves a fixed dosage drug schedule, patients may drop out at the upper dosage levels because of side effects, or, at the lower dosage levels because their symptoms did not improve.
Patients who present as psychiatric emergencies and who are entered into a drug trial may have a high drop out rate because of management problems.
Side-effects or treatment-emergent symptoms may be due to the trial drug, due to an adjuvant drug, due to drug interaction, due to the primary disease process or due to some other disease process. Because of the diverse symptomatology it is important to record symptoms which are present before active drug therapy, as subsequently it may be difficult to establish whether a particular symptom was present initially before starting the treatment or whether its later emergence identifies it as a treatment-emergent symptom or a symptom due to some other illness or reason. Treatment-emergent symptoms may be subjective and may be volunteered by the patient or they may be elicited by the investigator on enquiry. They may be objective and demonstrable on clinical examination or on investigation. They may be transient or permanent. They may be a few but serious or many and mild. It thus becomes important to establish the type of side effect, its time of emergence, its frequency and intensity, and its reversibility or otherwise.
Ayd (as quoted by Klett) has listed the following as symptoms of depression: fatigue, sleep disturbances, gastrointestinal disturbances, weight gain, menstrual disturbances, cardiovascular disturbances, headache, dry mouth, dizziness, blurred vision, auditory sensations, pain, tingling and numbness, hot flushes, unreality feelings and skin disturbances. Most of these symptoms have appeared on treatment emergent symptom check lists.
Therefore it is important to make certain whether the particular symptom is present because of the drug and/or the disease process or is independent of either. The manner of eliciting side effects can also influence their occurrence. Sometimes, the same number of side effects may he reported in a parallel group comparison between an investigational drug and a placebo. This can occur if placebo reactors have no. been eliminated, or if there is interaction between the two groups of patients, or if leading questions have been asked by the rater.
The problem; in successfully completing psychotropic drug trials do not end with problems of classification and conduct. Certain salient points have to be kept in mind in the analysis and interpretation of results.
The results of a psychotropic drug trial may be assessed subjectively., , ,  Here the training, experience and objectivity of the evaluator play an important part. The precision with which these assessments are made can be increased by using rating scales. In other instances objective measures of assessment like psychophysiological parameters may have to be used.
Most rating scales used in psychiatric assessments are unidirectional. The items have a specific loading, and the sensitivity of the scale may depend upon its most insensitive measures. (e.g. the Brief Psychiatric Rating Scale has got 18 items out of which only four are concerned with depression). Therefore this scale would not be very sensitive for evaluating depressive symptoms. Frequently the scores follow a non-normal distribution and therefore non-parametric statistics should be used. Many scales have a large number of items which have to be scored (e.g. Hamilton's Psychiatric Rating Scale for Deparession has 24 items). This can create problems of multiple inference as it is possible that if a large number of variables are involved, significances which may be detected in one or two items may possibly be due to chance occurrence.
It is important to include cases on trial where the total rating scale scores are above a certain value. Inclusion of asymptomatic cases or of cases of mild severity reduces the sensitivity, decreases the average amount of change and lowers the ratio of between group to within group differences. In turn this will decrease the likelihood of rejection of a null hypothesis and the possibility of coming to a conclusion about significant differences between two groups.
Both the total scores and the individual scores of the rating scale should be analysed as they give different information. The way the total scores decrease or increase can give a general idea about the overall drug effect on the total symptom complex. In contrast, analysis of the individual scores shows the effect of the drug on specific "target" symptoms.
It must be remembered that the drug will alter more than one variable and there is a possibility of variable interaction. In the hypothetical example given in [Table - 1] the pre- and post-treatment scores with an investigational neuroleptic show that the drug has decreased hostility and hallucination but at the same time the scores on depression and retardation have increased.
Not only the direction of change but also the magnitude of change is important.
In a hypothetical example, suppose, on drug "A" the initial score of 100 is decreased after treatment to a final score of 60, which is a 40 point difference, while on drug "B" an initial score of 50 is reduced to a final score of 10, which is again a 40 point difference. Then the magnitude of change is greater under drug, "B" conditions as there is an 80% reduction as compared to a 40% reduction under drug "A" conditions.
This example also illustrates another important feature viz., that the initial scrores in the two instances under treatment are not equal.
When using an arbitary cut-off line [Fig. 1] it is important to note the distribution of patients around it. In this example if 50% reduction in scores is the criterion for seperating patients who have improved from those who have not improved, one can state with more certainty about the number of patients who have not improved with reference to drug B rather than drug A, as both the improved and the not improved patients are clustered around the cut off line in the former instance but are well away from the cut off line in the latter instance.
Rigid adherence to certain pre-conceived notions can also create problems. In most biostatistical analysis the value of 'p' is fixed at 0.05. This insures that a type I error is not committed. However, there may be instances when a good drug may be missed with this procedure, if one does not consider the possibility of safeguarding against a type II error or taking into consideration the power function of the statistical test employed.
Careful attention to interpreting the results of certain observations during a drug trial can also yield valuable information about pharmacokinetics or pharmacodynamics. The example in [Table - 2] shows the effects on sleep and anxiety when two drugs are compared viz. clobazam v/s diazepam. It was observed that during the placebo period at the end of the drug trial period, patients on clobazam reported no change in sleep patterns or aggravation of their anxiety symptoms, while patients on diazepam showed a deterioration in sleep pattern and an increase in anxiety symptoms during this period. It was postulated that this may be due to an unknown active metabolite of clobazam which may have a long half life and which was active during the placebo trial period. Subsequent drug action studies showed that this was indeed the case.
There are many problems in successfully conducting psychotropic drug trials. Only a few which are frequently encountered in day to day work have been briefly discussed in this review. The clinical drug trial is not just a device to obtain registration of a new drug for commercial purposes. It is a highly skillful procedure for generating new knowledge which is critical to the safety of human beings and the guidance of medical profession. In conclusion, it may be correct to state that conducting a psychotropic drug trial is like observing a pretty girl in a scanty bathing suit-what one sees is important, but what one does not see can be-vital.
The authors thank Dr. C. K. Deshpande, Dean, K.E.M. Hospital, Bombay for permission to publish this paper.