Why stop at antidepressants?
http://www.100md.com
《英国医生杂志》
1 Department of Psychological Medicine, University of Auckland, PB92019, Auckland 1, New Zealand s.hatcher@auckland.ac.nz
Two problems arise when assessing treatments for depression. The first is specific to depression and the second is inherent in using randomised controlled trials as the sole evidence for deciding questions about treatment. Moncrieff and Kirsch focus on drug treatments, yet most of these issues also apply to non-drug interventions.1
Parker has described the problems of a "one size fits all" approach to trials of treatment for depression in which people with different severities of illness and symptoms are all included under the same heading of depression.2 Ensuring participants remain blind to treatment is also a problem. Outcomes in depression trials are usually assessed with a rating scale. However, all rating scales are ordinal—someone who scores 20 on the Hamilton rating scale for depression is more depressed than someone who scores 10, but we can't say they are twice as depressed. Nevertheless, most researchers have assumed that you can, making the results (such as changes in mean scores) hard to interpret. Lastly, most trials of depression are for short periods. In the National Institute for Health and Clinical Excellence guidelines eight weeks is the cut-off for dividing trials into short term and long term studies.
Relying on randomised controlled trials as the sole evidence for making decisions about treatment also has problems. It is difficult to recruit people to randomised controlled trials, so the trials are often small and the people who participate are not representative of the wider population of depressed people. Decisions about treatment are based on adverse effects as well as benefits, yet randomised controlled trials are poor at detecting rare but important adverse effects. In the recent debate about selective serotonin reuptake inhibitors in children, it wasn't effectiveness that was the issue but suicide.
One problem that psychological treatments don't share with drug treatments is the commercial considerations driving pharmacological research. This may affect the study design3 and result in important publication bias because only positive findings are reported.4
Better evidence
So what is the way forward? Consumers need to participate in the design of treatment trials. They may be particularly helpful in devising better ways of engaging potential participants and in assessing outcome. Researchers need to think beyond conventional randomised controlled trials and consider patient preference trials or modified Zelen methods that combine the advantages of randomisation and observational studies.5 Studies on treatment in depression need to be longer than a few weeks and they should report recruitment rates. Lastly registration of all new treatment trials and full disclosure of results needs to happen.
In the meantime, however, patients, clinicians, and funders still have to make decisions about treatment. These decisions are based on the evidence but also are influenced by values and resources. Moncrieff and Kirsch make clear where their values lie—antidepressants bad, psychological therapies good. But when criticising the evidence, why stop at antidepressants?
Competing interests: None declared.
References
Moncrieff J, Kirsch I. Efficiency of antidepressant activity in adults. BMJ 2005;331: 155-7.
Parker G. Evaluating treatments for the mood disorders: time for the evidence to get real. Aust N Z J Psychiatry 2004;38: 408-14.
Safer DJ. Design and reporting modifications in industry-sponsored comparative psychopharmacology trials. J Nerv Ment Dis 2002;190: 583-92.
Gilbody S, Song F. Publication bias and the integrity of psychiatric research. Psychol Med 2000;30: 253-8.
Black N. Why we need observational studies to evaluate the effectiveness of health care. BMJ 1996;312: 1215-8.(Simon Hatcher, senior lecturer in psychi)
Two problems arise when assessing treatments for depression. The first is specific to depression and the second is inherent in using randomised controlled trials as the sole evidence for deciding questions about treatment. Moncrieff and Kirsch focus on drug treatments, yet most of these issues also apply to non-drug interventions.1
Parker has described the problems of a "one size fits all" approach to trials of treatment for depression in which people with different severities of illness and symptoms are all included under the same heading of depression.2 Ensuring participants remain blind to treatment is also a problem. Outcomes in depression trials are usually assessed with a rating scale. However, all rating scales are ordinal—someone who scores 20 on the Hamilton rating scale for depression is more depressed than someone who scores 10, but we can't say they are twice as depressed. Nevertheless, most researchers have assumed that you can, making the results (such as changes in mean scores) hard to interpret. Lastly, most trials of depression are for short periods. In the National Institute for Health and Clinical Excellence guidelines eight weeks is the cut-off for dividing trials into short term and long term studies.
Relying on randomised controlled trials as the sole evidence for making decisions about treatment also has problems. It is difficult to recruit people to randomised controlled trials, so the trials are often small and the people who participate are not representative of the wider population of depressed people. Decisions about treatment are based on adverse effects as well as benefits, yet randomised controlled trials are poor at detecting rare but important adverse effects. In the recent debate about selective serotonin reuptake inhibitors in children, it wasn't effectiveness that was the issue but suicide.
One problem that psychological treatments don't share with drug treatments is the commercial considerations driving pharmacological research. This may affect the study design3 and result in important publication bias because only positive findings are reported.4
Better evidence
So what is the way forward? Consumers need to participate in the design of treatment trials. They may be particularly helpful in devising better ways of engaging potential participants and in assessing outcome. Researchers need to think beyond conventional randomised controlled trials and consider patient preference trials or modified Zelen methods that combine the advantages of randomisation and observational studies.5 Studies on treatment in depression need to be longer than a few weeks and they should report recruitment rates. Lastly registration of all new treatment trials and full disclosure of results needs to happen.
In the meantime, however, patients, clinicians, and funders still have to make decisions about treatment. These decisions are based on the evidence but also are influenced by values and resources. Moncrieff and Kirsch make clear where their values lie—antidepressants bad, psychological therapies good. But when criticising the evidence, why stop at antidepressants?
Competing interests: None declared.
References
Moncrieff J, Kirsch I. Efficiency of antidepressant activity in adults. BMJ 2005;331: 155-7.
Parker G. Evaluating treatments for the mood disorders: time for the evidence to get real. Aust N Z J Psychiatry 2004;38: 408-14.
Safer DJ. Design and reporting modifications in industry-sponsored comparative psychopharmacology trials. J Nerv Ment Dis 2002;190: 583-92.
Gilbody S, Song F. Publication bias and the integrity of psychiatric research. Psychol Med 2000;30: 253-8.
Black N. Why we need observational studies to evaluate the effectiveness of health care. BMJ 1996;312: 1215-8.(Simon Hatcher, senior lecturer in psychi)