Interpreting and applying the EUFEST results using number needed to treat: antipsychotic effectiveness in first-episode schizophrenia

A current therapeutic controversy in the treatment of schizophrenia is the relative merit of using different antipsychotic medications. Recently reported are the results of the European First-Episode Schizophrenia Trial (EUFEST), where 498 patients were randomised (1). Similar to the US Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) for schizophrenia, the primary outcome measure was all-cause treatment discontinuation (2), and the study was conducted at multiple sites (50 in 14 countries), with each patient able to remain on their assigned medication for a lengthy period of time (1 year). The major differences are that the EUFEST patients were in the first episode of their illness, the study although randomised, was open-label, and the principal first-generation antipsychotic being examined was haloperidol, not perphenazine, albeit at a low dose of 1–4 mg/day. Rather than including risperidone, the study designers included amisulpride, a second-generation antipsychotic unavailable in the USA. Also, unlike CATIE, failure on randomised drug meant the end of that patient’s study participation instead of offering the opportunity to switch medications and continue in another phase of the study. Three different manufacturers of antipsychotics funded the study. The authors found differences in all-cause discontinuation evidencing advantages for all of the second-generation antipsychotics using survival curve analysis, but they did not find significant differences in psychopathological rating scale outcomes among the groups. Their conclusion was that clinically meaningful antipsychotic treatment of first-episode schizophrenia is possible but that second-generation antipsychotics are not necessarily more efficacious than haloperidol, based on the finding that symptom reductions as measured by the Positive and Negative Syndrome Scale (PANSS) were almost the same in all the groups, at around 60%. Moreover, depression outcomes as measured by the Calgary Depression Scale for Schizophrenia and quality of life outcomes as measured by the Manchester Short Assessment of Quality of Life Scale were also similar for the antipsychotics tested. This may miss the point that there were identifiable differences in medication ‘durability’ where patients would continue taking some of the antipsychotics for longer periods of time. Patients randomised to haloperidol had a mean time to discontinuation of 0.5 months, compared with 5.3 months for amisulpride and 6.3 months for olanzapine (Figure 1). Did patients get better relatively quickly regardless of antipsychotic but then decided not to continue for reasons that are not adequately captured? Mean time to all-cause discontinuation (months) and 95% confidence intervals [data from Ref. (1), table 2, pages 1089–90]. As per the footnote in the original publication these are Kaplan–Meier estimates for months at risk for treatment discontinuation, excluding the first 14 days after randomisation. There is no actual upper limit for the confidence interval for amisulpride and olanzapine because the upper limit is above the maximum follow-up time The concept of number needed to treat (NNT) can help place the EUFEST data into clinical perspective. Defined as ‘the number of patients who must receive an intervention of therapy during a specific period of time to prevent one adverse outcome or produce one positive outcome’ (3, page 111), NNT is one of the essential tools of evidence-based medicine that helps us gauge effect size, or clinical significance (4). Number needed to treat for all-cause discontinuation for the five tested antipsychotics can be presented in a 5 × 5 table (Table 1). The absolute number who discontinued divided by the individual group total was used to calculate discontinuation rates, rather than the Kaplan–Meier estimate so that the respective 95% confidence interval for the NNT could be determined. The rank ordering of the antipsychotics by all-cause discontinuation was (from lowest to highest rates) olanzapine, amisulpride, ziprasidone, quetiapine and haloperidol. Quantifying these differences with NNT yields statistically significant pair-wise advantages for olanzapine vs. haloperidol and quetiapine, amisulpride vs. haloperidol and quetiapine, and ziprasidone vs. haloperidol. The strongest effect sizes were olanzapine or amisulpride vs. haloperidol with a NNT of four, meaning for every four patients randomised to olanzapine or amisulpride instead of haloperidol, one additional patient on olanzapine or amisulpride completed the study on their initially assigned medication. Overall, the EUFEST NNT results for all-cause discontinuation are consistent with what has been observed in a NNT analysis of the CATIE data (5). Number needed to treat for discontinuation because of insufficient efficacy, side effects and non-adherence can also be calculated. Amisulpride and olanzapine had the lowest rates for discontinuation because of insufficient efficacy (14% for each), and haloperidol and quetiapine the highest (48% and 40% respectively). Statistically significant advantages are evident for NNT for discontinuation because of insufficient efficacy for olanzapine or amisulpride vs. haloperidol or quetiapine (NNT five for all comparisons), and ziprasidone vs. quetiapine (NNT eight). Quetiapine and olanzapine had the lowest discontinuation rates because of side effects (3% and 6% respectively), and haloperidol and amisulpride the highest (20% for each). Statistically significant advantages are evident for NNT for discontinuation because of side effects for quetiapine vs. haloperidol, amisulpride or ziprasidone (NNT 11, 11 and 16 respectively). Amisulpride and ziprasidone had the lowest rates for discontinuation because of non-adherence (13% and 14% respectively) and haloperidol and quetiapine the highest (30% and 19% respectively). No statistically significant pair-wise differences were found for NNT for discontinuation because of non-adherence. Thus, it appears that the different antipsychotic medications differed from one another in terms of overall effectiveness, and had strikingly different profiles regarding tolerability and efficacy, for example quetiapine ranked low for efficacy but high for tolerability, and haloperidol ranked low for both efficacy and side effects. Although, PANSS scores were similar for all groups, severity of illness as measured by the Clinical Global Impression (CGI) and overall functioning as measured by the Global Assessment of Functioning (GAF) differed among the groups in a manner consistent with the primary outcome measure of all-cause discontinuation. This underscores the issue that important factors that compel the continued use of medication for some individuals cannot be measured using psychopathology ratings alone. The European First-Episode Schizophrenia Trial also confirms what we already knew about the vulnerability of first-episode patients to side effects. Similar to the results from a 52-week single-company sponsored study of olanzapine, quetiapine and risperidone in the treatment of early psychosis (6), weight gain was common: in EUFAST weight gain of > 7% of baseline was observed from a low of 37% of patients randomised to ziprasidone to a high of 86% of those randomised to olanzapine. Rates of akathisia ranged from a low of 10% for those receiving olanzapine to a high of 28% for those receiving ziprasidone. Parkinsonism was evident in 34% of patients randomised to haloperidol, despite the low mean dose. Hyperprolactinaemia was observed in 89% of patients randomised to amisulpride, which was approximately double the rate seen for the other groups. Dose ranges for the second-generation antipsychotics were amisulpride 200–800 mg/day, olanzapine 5–20 mg/day, quetiapine 200–750 mg/day and ziprasidone 40–160 mg/day. These are all within the range recommended in product labelling. The mean doses received before discontinuation of treatment were amisulpride 450.8 mg/day, olanzapine 12.6 mg/day, quetiapine 498.6 mg/day and ziprasidone 107.2 mg/day, similar to the doses observed in another multi-medication early psychosis study where mean modal doses for olanzapine and quetiapine were 11.7 and 506 mg/day respectively (6). The mean dose for haloperidol was 3.0 mg/day. To what patients would the EUFEST results apply? Generalisability would be limited to patients within their first psychotic episode with only very limited prior exposure (if any) to antipsychotic medication treatment. Almost all (89%) of subjects were inpatients. Mean age of participants was 26.0 years, 40% were female and 94% were white. Average years of education were 12.5. Only 13% were living alone; 47% were employed or were students. Although 53% of patients were diagnosed with schizophrenia, 40% had a diagnosis of schizophreniform disorder. Use of mood stabilisers, benzodiazepines, antidepressants and anticholinergic drugs were permitted. Limitations to EUFEST include the open-label design and the possibility of bias; however, discontinuation rates for haloperidol were not different for patients from sites that upon query expected haloperidol to perform less well than the alternatives. Other challenges include the lack of availability of all tested antipsychotics at all sites, particularly for ziprasidone, leading to the removal of ziprasidone from the randomisation scheme for approximately 10 of the 48 months or so that the study was conducted. Consequently, the number of subjects randomised to ziprasidone (n = 82) was lower than for the alternatives (n = 103–105). The EUFEST authors note that patients with first-episode schizophrenia are likely to do better than those with chronic schizophrenia, partly explaining the robust reductions in symptoms observed for all medication groups tested. Although efficacy appears similar, overall effectiveness or ‘durability’ is not. It seems that the rate of all-cause discontinuation, discontinuation because of lack of efficacy and discontinuation because of side effects, as well as the significant differences in the CGI and GAF scores, all suggesting inferior effectiveness of haloperidol in comparison with some second-generation antipsychotics, should result in a recommendation not to use haloperidol as the first-line treatment of first-episode schizophrenia. Amisulpride or olanzapine may be better choices, consistent also with a metaanalysis published when EUFEST first began enrolling subjects (7). Clinicians will continue to need to match-up individual patients to individual drugs, taking into account individual baseline risks for adverse outcomes, individual preferences for expected side effects and engage in ongoing monitoring of both efficacy and tolerability. Leslie Citrome, is a consultant for, has received honoraria from or has conducted clinical research supported by the following: Abbott Laboratories, AstraZeneca Pharmaceuticals, Avanir Pharmaceuticals, Azur Pharma Inc., Barr Laboratories, Bristol-Myers Squibb, Eli Lilly and Company, Forest Research Institute, GlaxoSmithKline, Janssen Pharmaceuticals, Jazz Pharmaceuticals, Pfizer Inc. and Vanda Pharmaceuticals. As Psychiatry Associate Editor for the journal, Leslie Citrome withdrew from the review process and deferred all decisions to Graham Jackson.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.