Clinical trials
Once a decision has been made to develop a compound further following the extensive pre-clinical pharmacological and toxicological studies, approval for the first clinical studies must be sought from the regulatory authority (Medicines Board in Europe or the Food and Drug Administration in the USA). A clinical trial of a new drug is, in the words of Bradford Hill (in his Principles of Medical Statistics): a carefully, and ethically, designed experiment with the aim of answering some precisely framed questions. In its most vigorous form it demands equivalent groups of patients concurrently treated in different ways. These groups are formed by the random allocation of patients to one or other treatment. In principle, the method is applicable to any disease or type of treatment. It may also be applied on any scale.
Thus the purpose of a clinical trial is to describe (a) whether the new treatment is of therapeutic value, (b) how it compares with a ‘‘standard’’ drug, (c) what type of patient would benefit from the new drug, (d) what is the best route of administration, how frequently and in what dose range should it be used, (e) what are the side effects and disadvantages of the new drug. For the classical randomized control trial, the following points should be stressed:
(a) The use of equivalent groups of patients by random (chance) allocation to a placebo, standard drug or novel drug group. It is self-evident that if the treatment groups differ with respect to age, gender, race, duration and severity of the disease it will not be possible to attribute differences in outcome to the novel treatment. The randomization of patients is therefore used to eliminate systematic bias and to permit the use of appropriate statistical methods to correct for any bias.
(b) Time and place of treatment. Treatment must be undertaken concurrently and concomitantly. Using control data from past clinical studies (historical controls) is almost always unacceptable.
(c) Precisely formulated questions. These must be formulated before the start of the trial. For example, ‘‘is drug A capable of treating depression more rapidly and effectively, and with fewer side effects, than drug B?’’.
Some important concepts in evaluating clinical trials The null hypothesis postulates that there is no difference in outcome between a new and a standard drug. Thus when two groups of patients have been treated separately with the drugs (between-patient comparisons) or when each patient has received both drugs (within-patient comparisons) and the result of the outcome of treatment is apparently better with one drug than the other, it is essential to determine if this difference is statistically significant.
The test of statistical significance will enable the researcher to determine whether the differences observed between the two treatments are due to chance. In practice, it is generally agreed that if the difference between the two groups occurs five times, or less, in 100 trials then the null hypothesis is unlikely to hold true and there is a real difference between the treatment groups. The level of statistical significance which is usually considered to be acceptable is at the 5% level, or less. This is represented by a probability value P50.05 (the percentage divided by 100). If the probability of a result occurs only once in 100 trials then it is highly significant at P50.001. The confidence interval is a measure of the degree of assurance, or confidence, one may have in the process or power of the result. The confidence interval is expressed as a range of values about the mean and within which it is 95% certain that the true value of the result lies. The range may be wide, indicating uncertainty, or narrow, indicating relative certainty. If a result lacks statistical significance at the 5% level or less, it can only be interpreted as meaning that there is no clinically or experimentally useful difference between the treatments. Small numbers in experimental groups inevitably mean low precision or statistical power. Often small clinical trials are published without any statement of statistical power or the inclusion of confidence intervals which reveal their inadequacy.
Types of error
Type 1 errors arise when a difference is found between treatments when, in reality, the groups do not differ, whereas Type 2 errors commonly arise when treatments do differ but no statistical differences are found. It must be emphasized that statistical tests do not prove that there is a difference between two groups, but merely that there is a probability of this being the case. However, a difference may be statistically significant and have narrow confidence limits but yet be biologically insignificant. For example, there is experimental evidence to show that beta adrenoceptors on lymphocytes change by over 50% between midday and midnight in healthy adults yet such changes would appear to have no clinical significance. Similarly, the activity of the serotonin transporter on the platelet membrane is one-third greater at noon than at 6 a.m. without having any apparent effect on physiological functions. Thus the statistical significance of a result in any area of neuroscience must always be considered in conj unction with the biological relevance of the change. This is often overlooked, particularly when considering the variables that are linked to psychiatric and neurological disorders.
Types of clinical trial: single and double-blind trials
Because both the clinician and the patient are subject to bias as a result of their expectations of the outcome of the trial of a new drug, the doubleblind technique is usually applied to evaluate the efficacy of a new drug. In such a trial the randomized groups of patients are given identical capsules or tablets (containing either a placebo, unknown drug or standard comparator); the clinician undertaking the clinical assessment is also ‘‘blind’’ to the distribution of the different treatments. Occasionally a double-blind technique cannot be applied as, for example, when the side effects of one of the drugs is greater than that of the placebo or the second drug. A non-blinded clinical trial (i.e. lacking a placebo or standard comparator) is called an ‘‘open’’ trial and is usually used for the first clinical exposure of a novel compound once it has been approved for clinical trial by the regulatory authority. ‘‘Open’’ trials are useful for obtaining the dose range for a new drug and the frequency of side effec ts; healthy volunteers are usually involved in the first exposure to a new drug. Once the relative safety of the drug has been assessed, an ‘‘open’’ trial on a small group of well-defined patients is undertaken so that an idea of the therapeutic efficacy of the drug can be obtained. These are termed Phase 1 studies and they are followed by Phase 2 studies, which are double-blind and involve a comparator as a standard, and Phase 3 studies (usually multicentred studies involving a large number of patients under double-blind conditions). The Phase 1–Phase 3 studies would normally take at least 6 years to complete. It has also been estimated that it takes at least 12 years from the initial chemical synthesis of a novel compound to its clearance for clinical use by the regulatory authorities at a cost in excess of $230 million. In addition, for every 10 000 chemical entities that are synthesized, approximately 10 million will enter Phase 1 trials but only one would be expected to obtain regulatory approval. Use of placebos in clinical trials While serious ethical objections have been raised regarding the use of placebos in trials of drugs used in the treatment of psychiatric disorders (largely based on the possibility that the patients may commit suicide if they are inadequately treated, although such patients are usually excluded from placebo controlled trials), all regulatory authorities insist on properly conducted, placebo controlled trials as a basis for registering a new drug.
The placebo is useful in (a) distinguishing the pharmacodynamic effects of a drug from the psychological impact of the medication and the environment in which it is given (the ‘‘halo’’ effect of the enthusiastic, or pessimistic, research clinician). It is well known, for example, that the placebo effect in major depression is as high as 30% while that of an effective antidepressant is approximately 60% of the optimal response. This statistic illustrates the importance of placebo-based studies in evaluating the efficacy of a new psychotropic drug. The placebo also distinguishes drug effects from fluctuations in the disease, which is particularly relevant in the case of psychiatric disorders. Lastly, the placebo allows false negatives to be excluded. For example, if a placebo is compared to a novel drug and a standard drug and the outcome of all three treatments is the same, the conclusion reached would be that the trial design (for example, number of patients in each group) is incapable of distinguishing between an active and inactive drug and therefore should be modified. If a standard drug is not included, however, it can only be concluded that the new medication is inactive at the dose used or that the end point used for the clinical assessment is inadequate. Informed consent must always be obtained from a patient participating in any clinical trial. In the trial of a psychotropic drug, the mental status examination is critical for determining the capacity of the patient to consent and the ability to communicate is an absolute prerequisite. Both the memory and the orientation of the patient must be substantially intact if the patient is to give informed consent. Such studies may be undertaken in chronic, stable diseases that cannot be cured but whose symptoms may improve following drug treatment, for example, Parkinson’s disease and the dementias. In such cases, each patient is subject to the new drug and placebo in a random order, the patients acting as their own control. In such studies, it is important to ensure that any ‘‘carry-over’’ effects of the active drug are taken into account when the placebo, or a lower dose of the active drug, follows the highest dose of the drug.
The number of patients to be included in any trial is usually decided by a statistician. In the fixed-number type of trial, the total number of patients to be recruited is agreed before the start of the study and must be adhered to even if the result of the study does not quite reach an acceptable level of statistical significance. In such circumstances it is not permissible to add additional patients to the study in the hope that the 5% level of significance will be reached as this will not allow chance and the effect of treatment to be the sole factors involved in deciding the outcome of the study. To avoid this situation, trials are now designed so that the number of patients in each treatment group is not determined in advance but the trial design allows either continuous or intermittent assessment of the response, thereby allowing the trial to be stopped as soon as statistical difference between the groups is obtained or when such a result seems unlikely. Thus the trial can be terminated when the predete rmined result is obtained (for example, a 50% reduction in the Hamilton Depression score in an antidepressant drug trial). In such a modified sequence design study, formal analysis of the data is undertaken at several predetermined intervals. Such interim analyses may reduce the power of the statistical analysis however. Large, definitive clinical trials of the type that have been used for Phase 3 studies are difficult to organize, prolonged, very expensive to perform and often yield inconclusive results. This has led to the design of smaller, controlled trials which, while of more limited statistical power, are subject to meta-analyses. Where numerous controlled trials on a drug have been undertaken and the outcome varies, the data can be collected in a systematic review and the accumulated results analysed by appropriate statistical methods. The resultant meta (‘‘overall’’) analysis must meet the criteria for a good scientific study. Meta-analyses can involve 100 000 patients, or more, but problems may arise in identifying all the suitable trials of a new drug (and in assessing their standards from the published literature), added to which is the bias due to the fact that only positive results are usually published. Cost of treatment with psychotropic drugs It is self-evident that any new psychotropic drug is likely to be expensive. This situation also applied to the drugs used to treat other chronic illnesses such as cardiovascular disease, rheumatoid arthritis and diabetes. However, the cost of treatment implies more than the price of the drug. As most psychiatric disorders are chronic, costs must take into account the acute, maintenance and prophylactic phases of treatment. Thus in assessing the true cost, it is necessary to consider the ancillary procedures needed to monitor the patients’ response to treatment (laboratory tests, etc.), managing any adverse effects of the medication, the relative costs of inpatient versus out-patient treatment, costs saved to the social and health services as a result of imp roved compliance and better response to treatment and the comparative costs of alternative treatments. It has been estimated that the cost of a new psychotropic drug for the treatment of depression, bipolar disorder or schizophrenia is approximately 10–15% of the total cost of treatment. It is also necessary to take into account the costs of not providing adequate treatment. These factors include the risks of premature death, and the increased morbidity, including the loss of productivity and the adverse effect of the illness on the social life of the patient.
Thus despite the much higher cost of a new psychotropic drug, Quality of Life assessments, which take into account the true cost of treatment, consistently show that new antidepressants and antipsychotics are superior to the older, and cheaper, medications. Most importantly, when optimal total care is provided for the patient, the quality of life is substantially improved which surely must be the aim of clinical psychopharmacology. The limitation of pre-clinical studies on the development of novel psychotropic drugs Serendipity has played a major role in the discovery of most classes of psychotropic drugs. For example, the observation that the first antidepressants, the tricyclic antidepressants and the monoamine oxidase inhibitors, impeded the reuptake of biogenic amines into brain slices, or inhibited their metabolism, following their acute administration to rats, provided the experimenter with a mechanism that could be easily investigated in vitro. Such methods led to the development of numerous antidepressants that differed in their potency, and to some extent in their side effects (for example, the selective serotonin reuptake inhibitors) but did little to further the development of novel antidepressants showing greater therapeutic efficacy. The accidental discovery of atypical antidepressants such as mianserin led to the broadening of the basis of the animal models used to detect potential antidepressants with greater emphasis being placed on the chronic effects of the drugs in vivo. However, with few exceptions, all the in vitro and in vivo models in current use are based on the acute effects of novel compounds. Recently, with the impact of high-throughput screening methods based on the activity of compounds on multiple neurotransmitter receptors in vitro, even acute in vivo methods have been reduced in an attempt to decrease the time taken to proceed to the initial clinical assessment. Only time will tell whether the extensive use of genomics, microarray screens and receptor profiles of novel compounds will lead to genuinely novel psychotropic drugs. So far, there appears to be little advance in the therapeutic potential of the psychotropic drugs in development.
Limitation of clinical trials of new psychotropic drugs
A typical placebo controlled trial consists of 50–60 patients including a reference drug. Trials of new psychotropic drugs usually last for 6–8 weeks; the recommended duration of treatment in clinical practice is usually about 6 months. Most clinical protocols do not provide justification for the sample size used, or even specify the statistical power of the data obtained. As the proportion of non-responders for many psychotropic drugs may range from 20–30%, it can be argued that such limited trials (with a statistical power that is too low to show differences between treatments, conducted over too short a period in comparison with clinical practice and often excluding noncompliant patients from the final analysis) are unlikely to show any difference between a new and standard drug. This means that few studies are able to demonstrate an improved efficacy of a new drug over a standard drug. It has been calculated that the sample size that would be required to show a difference of 5% between two psychotropic dr ugs, assuming that approximately 20% will be non-responders or partial responders, would be between 1250 and 2380 patients per arm of the trial. The expense of such trials would, of course, be prohibitive! This does show that the apparent limited efficacy of most classes of psychotropic drugs is a reflection of the limitation of the clinical trial rather than their real potential differences, which could only be determined by using much larger numbers of patients in trials that last for several months.
Once a decision has been made to develop a compound further following the extensive pre-clinical pharmacological and toxicological studies, approval for the first clinical studies must be sought from the regulatory authority (Medicines Board in Europe or the Food and Drug Administration in the USA). A clinical trial of a new drug is, in the words of Bradford Hill (in his Principles of Medical Statistics): a carefully, and ethically, designed experiment with the aim of answering some precisely framed questions. In its most vigorous form it demands equivalent groups of patients concurrently treated in different ways. These groups are formed by the random allocation of patients to one or other treatment. In principle, the method is applicable to any disease or type of treatment. It may also be applied on any scale.
Thus the purpose of a clinical trial is to describe (a) whether the new treatment is of therapeutic value, (b) how it compares with a ‘‘standard’’ drug, (c) what type of patient would benefit from the new drug, (d) what is the best route of administration, how frequently and in what dose range should it be used, (e) what are the side effects and disadvantages of the new drug. For the classical randomized control trial, the following points should be stressed:
(a) The use of equivalent groups of patients by random (chance) allocation to a placebo, standard drug or novel drug group. It is self-evident that if the treatment groups differ with respect to age, gender, race, duration and severity of the disease it will not be possible to attribute differences in outcome to the novel treatment. The randomization of patients is therefore used to eliminate systematic bias and to permit the use of appropriate statistical methods to correct for any bias.
(b) Time and place of treatment. Treatment must be undertaken concurrently and concomitantly. Using control data from past clinical studies (historical controls) is almost always unacceptable.
(c) Precisely formulated questions. These must be formulated before the start of the trial. For example, ‘‘is drug A capable of treating depression more rapidly and effectively, and with fewer side effects, than drug B?’’.
Some important concepts in evaluating clinical trials The null hypothesis postulates that there is no difference in outcome between a new and a standard drug. Thus when two groups of patients have been treated separately with the drugs (between-patient comparisons) or when each patient has received both drugs (within-patient comparisons) and the result of the outcome of treatment is apparently better with one drug than the other, it is essential to determine if this difference is statistically significant.
The test of statistical significance will enable the researcher to determine whether the differences observed between the two treatments are due to chance. In practice, it is generally agreed that if the difference between the two groups occurs five times, or less, in 100 trials then the null hypothesis is unlikely to hold true and there is a real difference between the treatment groups. The level of statistical significance which is usually considered to be acceptable is at the 5% level, or less. This is represented by a probability value P50.05 (the percentage divided by 100). If the probability of a result occurs only once in 100 trials then it is highly significant at P50.001. The confidence interval is a measure of the degree of assurance, or confidence, one may have in the process or power of the result. The confidence interval is expressed as a range of values about the mean and within which it is 95% certain that the true value of the result lies. The range may be wide, indicating uncertainty, or narrow, indicating relative certainty. If a result lacks statistical significance at the 5% level or less, it can only be interpreted as meaning that there is no clinically or experimentally useful difference between the treatments. Small numbers in experimental groups inevitably mean low precision or statistical power. Often small clinical trials are published without any statement of statistical power or the inclusion of confidence intervals which reveal their inadequacy.
Types of error
Type 1 errors arise when a difference is found between treatments when, in reality, the groups do not differ, whereas Type 2 errors commonly arise when treatments do differ but no statistical differences are found. It must be emphasized that statistical tests do not prove that there is a difference between two groups, but merely that there is a probability of this being the case. However, a difference may be statistically significant and have narrow confidence limits but yet be biologically insignificant. For example, there is experimental evidence to show that beta adrenoceptors on lymphocytes change by over 50% between midday and midnight in healthy adults yet such changes would appear to have no clinical significance. Similarly, the activity of the serotonin transporter on the platelet membrane is one-third greater at noon than at 6 a.m. without having any apparent effect on physiological functions. Thus the statistical significance of a result in any area of neuroscience must always be considered in conj unction with the biological relevance of the change. This is often overlooked, particularly when considering the variables that are linked to psychiatric and neurological disorders.
Types of clinical trial: single and double-blind trials
Because both the clinician and the patient are subject to bias as a result of their expectations of the outcome of the trial of a new drug, the doubleblind technique is usually applied to evaluate the efficacy of a new drug. In such a trial the randomized groups of patients are given identical capsules or tablets (containing either a placebo, unknown drug or standard comparator); the clinician undertaking the clinical assessment is also ‘‘blind’’ to the distribution of the different treatments. Occasionally a double-blind technique cannot be applied as, for example, when the side effects of one of the drugs is greater than that of the placebo or the second drug. A non-blinded clinical trial (i.e. lacking a placebo or standard comparator) is called an ‘‘open’’ trial and is usually used for the first clinical exposure of a novel compound once it has been approved for clinical trial by the regulatory authority. ‘‘Open’’ trials are useful for obtaining the dose range for a new drug and the frequency of side effec ts; healthy volunteers are usually involved in the first exposure to a new drug. Once the relative safety of the drug has been assessed, an ‘‘open’’ trial on a small group of well-defined patients is undertaken so that an idea of the therapeutic efficacy of the drug can be obtained. These are termed Phase 1 studies and they are followed by Phase 2 studies, which are double-blind and involve a comparator as a standard, and Phase 3 studies (usually multicentred studies involving a large number of patients under double-blind conditions). The Phase 1–Phase 3 studies would normally take at least 6 years to complete. It has also been estimated that it takes at least 12 years from the initial chemical synthesis of a novel compound to its clearance for clinical use by the regulatory authorities at a cost in excess of $230 million. In addition, for every 10 000 chemical entities that are synthesized, approximately 10 million will enter Phase 1 trials but only one would be expected to obtain regulatory approval. Use of placebos in clinical trials While serious ethical objections have been raised regarding the use of placebos in trials of drugs used in the treatment of psychiatric disorders (largely based on the possibility that the patients may commit suicide if they are inadequately treated, although such patients are usually excluded from placebo controlled trials), all regulatory authorities insist on properly conducted, placebo controlled trials as a basis for registering a new drug.
The placebo is useful in (a) distinguishing the pharmacodynamic effects of a drug from the psychological impact of the medication and the environment in which it is given (the ‘‘halo’’ effect of the enthusiastic, or pessimistic, research clinician). It is well known, for example, that the placebo effect in major depression is as high as 30% while that of an effective antidepressant is approximately 60% of the optimal response. This statistic illustrates the importance of placebo-based studies in evaluating the efficacy of a new psychotropic drug. The placebo also distinguishes drug effects from fluctuations in the disease, which is particularly relevant in the case of psychiatric disorders. Lastly, the placebo allows false negatives to be excluded. For example, if a placebo is compared to a novel drug and a standard drug and the outcome of all three treatments is the same, the conclusion reached would be that the trial design (for example, number of patients in each group) is incapable of distinguishing between an active and inactive drug and therefore should be modified. If a standard drug is not included, however, it can only be concluded that the new medication is inactive at the dose used or that the end point used for the clinical assessment is inadequate. Informed consent must always be obtained from a patient participating in any clinical trial. In the trial of a psychotropic drug, the mental status examination is critical for determining the capacity of the patient to consent and the ability to communicate is an absolute prerequisite. Both the memory and the orientation of the patient must be substantially intact if the patient is to give informed consent. Such studies may be undertaken in chronic, stable diseases that cannot be cured but whose symptoms may improve following drug treatment, for example, Parkinson’s disease and the dementias. In such cases, each patient is subject to the new drug and placebo in a random order, the patients acting as their own control. In such studies, it is important to ensure that any ‘‘carry-over’’ effects of the active drug are taken into account when the placebo, or a lower dose of the active drug, follows the highest dose of the drug.
The number of patients to be included in any trial is usually decided by a statistician. In the fixed-number type of trial, the total number of patients to be recruited is agreed before the start of the study and must be adhered to even if the result of the study does not quite reach an acceptable level of statistical significance. In such circumstances it is not permissible to add additional patients to the study in the hope that the 5% level of significance will be reached as this will not allow chance and the effect of treatment to be the sole factors involved in deciding the outcome of the study. To avoid this situation, trials are now designed so that the number of patients in each treatment group is not determined in advance but the trial design allows either continuous or intermittent assessment of the response, thereby allowing the trial to be stopped as soon as statistical difference between the groups is obtained or when such a result seems unlikely. Thus the trial can be terminated when the predete rmined result is obtained (for example, a 50% reduction in the Hamilton Depression score in an antidepressant drug trial). In such a modified sequence design study, formal analysis of the data is undertaken at several predetermined intervals. Such interim analyses may reduce the power of the statistical analysis however. Large, definitive clinical trials of the type that have been used for Phase 3 studies are difficult to organize, prolonged, very expensive to perform and often yield inconclusive results. This has led to the design of smaller, controlled trials which, while of more limited statistical power, are subject to meta-analyses. Where numerous controlled trials on a drug have been undertaken and the outcome varies, the data can be collected in a systematic review and the accumulated results analysed by appropriate statistical methods. The resultant meta (‘‘overall’’) analysis must meet the criteria for a good scientific study. Meta-analyses can involve 100 000 patients, or more, but problems may arise in identifying all the suitable trials of a new drug (and in assessing their standards from the published literature), added to which is the bias due to the fact that only positive results are usually published. Cost of treatment with psychotropic drugs It is self-evident that any new psychotropic drug is likely to be expensive. This situation also applied to the drugs used to treat other chronic illnesses such as cardiovascular disease, rheumatoid arthritis and diabetes. However, the cost of treatment implies more than the price of the drug. As most psychiatric disorders are chronic, costs must take into account the acute, maintenance and prophylactic phases of treatment. Thus in assessing the true cost, it is necessary to consider the ancillary procedures needed to monitor the patients’ response to treatment (laboratory tests, etc.), managing any adverse effects of the medication, the relative costs of inpatient versus out-patient treatment, costs saved to the social and health services as a result of imp roved compliance and better response to treatment and the comparative costs of alternative treatments. It has been estimated that the cost of a new psychotropic drug for the treatment of depression, bipolar disorder or schizophrenia is approximately 10–15% of the total cost of treatment. It is also necessary to take into account the costs of not providing adequate treatment. These factors include the risks of premature death, and the increased morbidity, including the loss of productivity and the adverse effect of the illness on the social life of the patient.
Thus despite the much higher cost of a new psychotropic drug, Quality of Life assessments, which take into account the true cost of treatment, consistently show that new antidepressants and antipsychotics are superior to the older, and cheaper, medications. Most importantly, when optimal total care is provided for the patient, the quality of life is substantially improved which surely must be the aim of clinical psychopharmacology. The limitation of pre-clinical studies on the development of novel psychotropic drugs Serendipity has played a major role in the discovery of most classes of psychotropic drugs. For example, the observation that the first antidepressants, the tricyclic antidepressants and the monoamine oxidase inhibitors, impeded the reuptake of biogenic amines into brain slices, or inhibited their metabolism, following their acute administration to rats, provided the experimenter with a mechanism that could be easily investigated in vitro. Such methods led to the development of numerous antidepressants that differed in their potency, and to some extent in their side effects (for example, the selective serotonin reuptake inhibitors) but did little to further the development of novel antidepressants showing greater therapeutic efficacy. The accidental discovery of atypical antidepressants such as mianserin led to the broadening of the basis of the animal models used to detect potential antidepressants with greater emphasis being placed on the chronic effects of the drugs in vivo. However, with few exceptions, all the in vitro and in vivo models in current use are based on the acute effects of novel compounds. Recently, with the impact of high-throughput screening methods based on the activity of compounds on multiple neurotransmitter receptors in vitro, even acute in vivo methods have been reduced in an attempt to decrease the time taken to proceed to the initial clinical assessment. Only time will tell whether the extensive use of genomics, microarray screens and receptor profiles of novel compounds will lead to genuinely novel psychotropic drugs. So far, there appears to be little advance in the therapeutic potential of the psychotropic drugs in development.
Limitation of clinical trials of new psychotropic drugs
A typical placebo controlled trial consists of 50–60 patients including a reference drug. Trials of new psychotropic drugs usually last for 6–8 weeks; the recommended duration of treatment in clinical practice is usually about 6 months. Most clinical protocols do not provide justification for the sample size used, or even specify the statistical power of the data obtained. As the proportion of non-responders for many psychotropic drugs may range from 20–30%, it can be argued that such limited trials (with a statistical power that is too low to show differences between treatments, conducted over too short a period in comparison with clinical practice and often excluding noncompliant patients from the final analysis) are unlikely to show any difference between a new and standard drug. This means that few studies are able to demonstrate an improved efficacy of a new drug over a standard drug. It has been calculated that the sample size that would be required to show a difference of 5% between two psychotropic dr ugs, assuming that approximately 20% will be non-responders or partial responders, would be between 1250 and 2380 patients per arm of the trial. The expense of such trials would, of course, be prohibitive! This does show that the apparent limited efficacy of most classes of psychotropic drugs is a reflection of the limitation of the clinical trial rather than their real potential differences, which could only be determined by using much larger numbers of patients in trials that last for several months.
No comments:
Post a Comment