Patient-level drivers of effectiveness: a case study in schizophrenia


The available pivotal randomised controlled trials (RCTs) on antipsychotic drugs to treat schizophrenia use highly selected patients and may not reflect the variation and heterogeneity of patients, the disease and healthcare in the real world. Consequently, this may indicate an efficacy-effectiveness gap (for a definition, see here).

Identifying the drivers of effectiveness (for a definition, see here) is important to minimise this gap in evidence and help decision makers understand the effectiveness of these medicines in a real-world population.

In practice, an efficacy-effectiveness gap is likely to occur if a key characteristic (such as a patient-, disease- or setting-related characteristic) is distributed differently in the RCT compared with real life, and if this key characteristic is an effect modifier of the drug.

A separate GetReal case study examined potential drivers of effectiveness at a healthcare setting-level in schizophrenia.

What was examined in this case study?

This case study aimed to:

  • identify effect modifiers of antipsychotic drugs among the patient characteristics that are frequently used as exclusion criteria in RCTs
  • quantify the impact of applying these exclusion criteria on the estimation of effect of the medicines.

The purpose of this work was to identify the drivers of effectiveness that cause an efficacy-effectiveness gap using:

Literature reviews
Two literature reviews were conducted using the PICOS framework:

  • The first aimed to identify patient- and disease-related characteristics that may act as effect modifiers; these were reviewed by clinical experts who identified the most clinically relevant effect modifiers.
  • A focused literature review was then conducted that aimed to list exclusion criteria used in pre-authorisation phase three trials, focusing on second-generation antipsychotic drugs in adults with schizophrenia in the US and Europe. After cross-checking and the websites of the European and American medicine regulators, only RCTs that were considered ‘pivotal’ were selected.

Analyses of real-world data
The impact of three potential drivers of effectiveness (identified from the literature reviews) was examined using the drug effect on the severity of schizophrenia symptoms after three months, when used as exclusion criteria. A subset of participants included in the multi-national European Schizophrenia Outpatient Health Outcome (SOHO) cohort study was analysed. The patients with any of the three exclusion criteria were excluded to mimic an RCT population.

The SOHO cohort study includes 10,218 people with schizophrenia treated with antipsychotic drugs in the real-world. All participants were treated as outpatients and had started or switched to an antipsychotic drug between 2001 and 2002. Participants who had started on one of the three most frequently prescribed medicines in the database (80.7% or 8250) were included in the analysis.

Exclusion criteria were considered to be effect modifiers of antipsychotic drugs when there was a significant difference in mean symptom improvement between those who did or did not have the criteria. The impact of applying the exclusion criteria on the effect of antipsychotics was estimated by measuring the outcome before and after each additional exclusion criteria was applied (sequentially).

What were the findings and conclusions?

The literature searches and expert interviews identified three possible effect modifiers of antipsychotic drugs that were also exclusion criteria in the pivotal RCTs:

  • disease chronicity
  • substance use disorder (excluding nicotine)
  • poor adherence to medicines.

When participants with any of the three characteristics were removed, 64.8% (5348/8250) remained in the ‘RCT-like’ population.

The analyses found that shorter duration of illness (3 years and under) and severe non-adherence (to antipsychotic drugs used before baseline) had better symptom improvement at 3 months when the data on all three drugs were combined; consequently, these were found to be possible effect modifiers. Excluding only patients with a short duration of illness from the cohort had the greatest impact on the antipsychotic drug effect, whereas excluding patients with only substance use disorder or only severe non-adherence on their own did not impact on the effect of the antipsychotic drug in the remaining patients. However, the absence of effect modification cannot be completely ruled out.

Overall, a greater improvement in symptom severity (the estimated effectiveness) was found in the SOHO cohort compared with the RCT-like subset of the SOHO patients without the three exclusion criteria (the estimated efficacy). This suggests that excluding patients with at least one exclusion criteria can lead to an underestimation of the effects of antipsychotic drugs.

Exclusion criteria in the pivotal RCTs for antipsychotic drugs are numerous and often include characteristics common in patients with schizophrenia in the real-world.

In this case study, patients with a shorter duration of illness appear to have a greater response to antipsychotic drugs than those with a longer duration of illness, so they should not be systematically excluded from RCTs.

More information about this work can be found here.

Key contributor

Clementine Nordon, LASER