One of the problems of health research is that it is usually “top down”. For example, in treatment effectiveness studies, researchers decide what is the outcome to measure. This is partly unavoidable in that to aid statistical analysis and demonstrate effectiveness, outcome measures that are common to all participants are used and the specific outcome measures need to be decided before participants are entered into the research. (This also prevents cherry-picking of one favourable result out of many less favourable results after the study data is collected). The “top down” problem comes from who decides what the important outcomes are; usually patients or participants are not consulted.
This can also occur in clinical practice where the doctor (or other health professional) may suggest or provide treatments that improve certain aspects of the condition, whilst the patient or client may prefer focussing on other areas.
Another assumption is that a health intervention that reduces symptoms will lead to improvements in the patient’s overall quality of life taking into account any negative effects of treatment such as side-effects.
The double-blind randomised trial this blog discusses (Grunder et al, 2016) focusses on improvements in “quality of life” or QOL (measured by the Short Form-36 measure) as well as improvement in clinical features such as symptoms (as measured by the Clinical Global Improvement Scale).
In this trial, the comparison is between the older “first generation antipsychotics” (FGAs) and the newer “second generation antipsychotics” (SGAs) in the treatment of schizophrenia spectrum disorders. The paper can be thought of as a “rematch” on the question of whether SGAs resulted in better quality of life and other outcomes than FGAs.
The best-known study that attempted to answer this question, the CUTLASS study (Jones et al, 2006), found that FGAs were at the very least not inferior to SGAs and actually may be superior in terms of improving QOL and reducing symptoms in participants with schizophrenia spectrum disorders.
Participants with a clinician diagnosis of ICD-10 schizophrenia from one of 14 hospitals in Germany were eligible for the study, if they needed to start antipsychotic treatment or change antipsychotic treatment because of poor response or unacceptable side-effects. There were some exclusion criteria e.g. involuntary patients or acutely suicidal.
The number of participants required was calculated initially to be very high (630) but this power calculation proved to be based on overly pessimistic assumptions and subsequent power calculation showed only about 150 participants were required.
Potential FGAs that could be used were flupentixol and haloperidol and potential SGAs were aripiprazole, olanzapine and quetiapine. The SGAs are quite different in terms of their actions, effectiveness and side-effects, more so than the FGAs (Leucht et al, 2013). This means that pooling the SGA results together creates uncertainty; e.g. aripiprazole is associated with less weight gain than the other two SGAs, but conversely aripiprazole may be less effective in reducing psychotic symptoms scores than olanzapine.
Pair combinations were created of each FGA paired to a SGA (e.g. haloperidol & olanzapine, flupentixol & quetiapine). Participants were randomised to one of two potential pairs. The treating doctor then chose one of these pairs. The participant was then further randomised to one member of this pair, the FGA or SGA. This allowed randomisation of treatment but with some clinician input to choose medication potential options that best fitted the participants.
Quality of life was measured using the Short-form 36 questionnaire (SF-36). Clinical outcomes were measured using the Clinical Global Impression Scale- Improvement (CGI-I Scale) and the PANNS rating schedule for psychotic and associated symptoms, the Subjective Well-Being Under Neuroleptic Treatment Scale, short form (SWN-K) and the total score of the Personal and Social Performance Scale (PSP). The CGI-I rates a clinician’s global assessment of improvement or worsening of a participant’s health on a 1-7 scale. The PANNS measures symptoms as a total score and also broken down into subscale scores of Positive, Negative and General symptoms.
Additional measurements included biological variables such as weight, waist circumference and blood tests such as glycosated haemoglobin and low-density lipoprotein (LDL) cholesterol.
Assessments were made at various points over the 24 week duration of the trial. Last Observation Carried Forward (LOCF) data was used. Variables were often combined into an Area Under the Curve and statistical tests performed on this. Participants who completed at least 6 weeks of treatment were considered to have completed the treatment protocol.
2,374 patients were assessed for eligibility but only 511 were eligible of whom only 149 agreed (so less than one-third of eligible participants agreed). The participants had an average CGI score of just over 5 which indicates a large proportion of the sample were markedly ill or worse. 69 participants were allocated FGAs, 80 participants allocated SGAs. Very few participants completed the treatment protocol of at least 6 weeks, the commonest reason being “other reason” (this may have been lack of effectiveness of randomised treatment).
|Class||Drug||Allocated||Completed treatment||Included in full-analysis||Adverse effects/Serious adverse effects*|
* Where adverse effects occurred, there was often >1 adverse effect per participant.
- It is immediately apparent that most participant data for outcomes comes from participants that did not complete the treatment protocol. (It may be that the numbers for completed treatment reflects those who completed all 24 weeks and that the participants who did not complete full treatment (but completed 6 weeks) were the additional participants entered into the final data analysis; the paper is not clear on this point)
- The final analysis showed that the SGA participants had a greater improvement on their SF-36 (quality of life) scores of a small effect size (0.34) with certain sub-scale scores:
- Physical component summary
- Physical role function
- Emotional role function
- There were no significant differences on the CGI-I improvement scale or the PANNS symptom scores
- One death occurred after the study drug was stopped
- The study found a significantly greater mean increase in the Body Mass Index (BMI) in the participants given SGAs mostly due to olanzapine and quetiapine, e.g. by week 24:
- 0·13 kg/m² increase in the FGA group
- 0·68 kg/m² increase in the SGA group
Strengths and limitations
- Measured impacts on QOL of patients not just symptoms
- Used a standardised measure of QOL (SF-36) used in a variety of health studies to allow more easy comparison (though some may advocate for QOL measurement specific for schizophrenia as used in CUTLASS (Jones et al (2006))
- A randomisation method was used, but the trial design allowed some choice on the part of the clinician to allow a better fit of medication to client’s individualised needs.
- High drop-out rates and use of Last Observation Carried Forward (LOCF) data. Use of unusual Area under Curve statistical technique
- Clinical diagnosis used not based on “gold standard” standardised research interview
- Very different drugs in terms of actions, effectiveness and side-effects combined in one SGA category.
This study suggests that SGAs have a small advantage for QOL outcomes compared to FGAs but equivalent effectiveness in reducing symptoms.
If SGAs associated with marked weight gain are used, then the patient will gain more weight than if FGAs are prescribed that are associated with less weight gain (Leucht et al, 2013). On the other hand, given the high number of drop-outs and use of LOCF data, one could argue that the confidence we can have in SGAs superiority over FGAs for QOL is not high.
CUTLASS (Jones et al (2006) gave the opposite answer of non-significant superiority of FGAs compared to SGAs for QOL, but used a different measure of QOL and aripiprazole was not available to be used.
I know it’s a cliché, but more studies are required to answer this question. Future research should perhaps use direct drug versus drug comparisons, rather than grouping them into broad categories like FGA and SGA.
Another welcome development would be the inclusion of patients in the decision-making process, when it comes to deciding which outcomes to measure.
Gründer G, Heinze M, Cordes J, Mühlbauer B, Juckel G, Schulz C, Rüther E, Timm J, NeSSy Study Group (2016) Effects of first-generation antipsychotics versus second-generation antipsychotics on quality of life in schizophrenia: a double-blind, randomised study. The Lancet Psychiatry 3(8) 717-729. http://dx.doi.org/10.1016/S2215-0366(16)00085-7
Jones PB, Barnes TR, Davies L, Dunn G, Lloyd H, Hayhurst KP, Murray RM, Markwick A, Lewis SW. (2006) Randomized controlled trial of the effect on Quality of Life of second-vs first-generation antipsychotic drugs in schizophrenia: Cost Utility of the Latest Antipsychotic Drugs in Schizophrenia Study (CUtLASS 1). Archives of General Psychiatry, 63(10), pp.1079-1087.
Leucht S, Cipriani A, Spineli L, Mavridis D, Örey D, Richter F, Samara M, Barbui C, Engel RR, Geddes JR, Kissling W. (2013) Comparative efficacy and tolerability of 15 antipsychotic drugs in schizophrenia: a multiple-treatments meta-analysis (PDF). The Lancet, 382(9896), pp.951-962