How representative of everyday clinical populations are schizophrenia patients enrolled in clinical trials?
Riedel M, Strassnig M, Muller N, Zwack P and Moller HJ;
Commented by , 31 Jan 2005
Objective
It is a widely held belief that clinical trials do not mirror the treatment situation in clinical routine. Especially industry sponsored studies that are undertaken for registration purposes apply a number of rather restrictive inclusion and exclusion criteria such as minimum scores of positive symptoms, exclusion of patients younger than 18 and older than 65, exclusion of suicidal patients and inclusion of only relatively cooperative patients who must consent to participate.
It is possible that such studies therefore rather represent "ideal" world outcomes than "routine care" realities. However, studies examining this issue are scarce, at least in the field of psychiatry.
Methods
The authors compared the characteristics of all inpatiens with schizophrenia according to DSM-IV who were enrolled in clinical trials at their department during the years 1995 to 1999 with patients treated in clinical routine. A retrospective case-control design was chosen and trial patients were matched with routine care patients by schizophrenia subtype, gender and age.
The following patient characteristics were compared: psychopathology was assessed with the scales Clinical Global Impressions (CGI), Global Assessment of Functioning (GAF) and the diagnostic module of the "Arbeitsgemeinschaft für Methodik und Diagnostik in der Psychiatrie" (AMDP).
Other characteristics were derived from chart reviews: antipsychotic doses transformed using chlorpromazine equivalents, concurrent non-psychiatric illnesses, and various anamnestic parameters such as
- age of onset
- duration of illness
- education
- marital status
- family history
Results
100 patients enrolled in a clinical trial were compared with 100 routine care patients. While the overall severity of illness as measured by the CGI and the GAF was not significantly different between groups, on average trial participants had more paranoid and hallucinatory symptoms, but fewer manic symptoms and less hostility.
There were fewer patients with negative symptoms among study participants (n=17) than among the controls (n=38, p < 0.05). In addition, patients with a high degree of formal thought disorder and suicidal patients (n=3) were significantly less likely to be included in a study than to be routinely treated (n=22, p < 0.05).
Patients enrolled in a clinical trial also had significantly fewer concomitant medical and neurological diseases, a shorter duration of illness, fewer hospitalisations, and they were more likely to be discharged with atypical antipsychotics than with typical antipsychotics.
Dr Leucht's comments
The main result of the study was that patients included in clinical trials differed according to a number of characteristics from patients treated in routine care. In a similar vein Hofer et al. (ref. 1) reported that in their hospital only 13.5 % of patients with schizophrenia were eligible and finally included in a clinical trial.
A strength of the current analysis was a methodologically sound case control design matching patients of both groups. Some results are tautonomous, however. E.g. since registration trials must exclude suicidal patients or extremly thought disorder patients for ethical reasons, it is a tautonomy that fewer trial patiens were suicidal and fewer were severely thought disordered.
Furthermore, although the overall severity of illness at baseline was similar in both groups, this result may be biased by the fact that neither extremly ill patients can be included in clinical trials, nor can patients with only few symptoms because the study protocols require a minimum of symptoms.
Finally, the study cannot answer the question as to whether the outcome of routine care patients would be different than that of trial patients. To resolve this issue a prospective design would be needed.
References
Hofer A, Hummer M, Huber R, Epub Kurz M, Walch T and Fleischhacker WW. Selection bias in clinical trials with antipsychotics. J Clin Psychopharmacol. 2000; 20; 699-702