Evidence review: efficacy

This evidence summary outlines the findings of a Cochrane systematic review, which included 7 randomised controlled trials (RCTs) of omega-3 fatty acids for treating schizophrenia and 1 additional RCT, which was published after the search date for this Cochrane review (Manteghiy et al. 2008).

Cochrane review

The Cochrane systematic review on polyunsaturated fatty acid supplementation for schizophrenia was assessed as up-to-date in February 2009. It identified 8 relevant RCTs, 7 of which compared omega-3 fatty acids with placebo (Peet et al. 2001a, Peet et al. 2001b, Fenton et al. 2001, Peet and Horrobin 2002, Emsley et al. 2002, Emsley et al. 2006 and Berger et al. 2007). The eighth RCT investigated omega-6 fatty acids, which is not relevant to this evidence summary. These RCTs also provided the key evidence base for 4 additional evidence reviews, some of which were published more recently than the Cochrane review (Freeman et al. 2006, Ross et al. 2007, Fusar-Poli and Berger 2012 and Politi et al. 2013).

The Cochrane review of the 7 RCTs of omega-3 fatty acids in a total of 501 people with schizophrenia provides the foundation of this evidence summary. For full descriptions of the underlying studies, see the full Cochrane review characteristics of included studies. They are summarised here.

Study characteristics

All 7 studies were mixed sex, with the overall age range of participants 15–65 years. Overall study sample size was small. Peet and Horrobin (2002) had the largest sample size (n=122); the other 6 studies ranged from 30 (Peet et al. 2001b) to 90 (Fenton et al. 2001) participants.

Study duration ranged from 12 to 16 weeks, took place in a mix of hospital and community settings, and all recruited people with schizophrenia diagnosed using Diagnostic and Statistical Manual of Mental Disorders 4th edition (DSM-IV). Fenton et al. (2001) also included people with schizo-affective disorder.

Most participants had chronic schizophrenia and were still symptomatic after treatment with antipsychotics. Peet et al. (2001b) was unique in including people who were recently diagnosed and had no previous treatment with antipsychotics. Berger et al. (2007) randomised relatively young people (15–29 years) experiencing their first psychotic episode. All participants in Emsley et al. (2006) had established neuroleptic-induced tardive dyskinesia.

Most of the studies compared omega-3 fatty acid capsules given in addition to existing antipsychotic medication with placebo capsules. They used either eicosapentaenoic acid (EPA), its ester, ethyl-eicosapentaenoic acid (E-EPA) or docosahexaenoic acid (DHA). Peet et al. (2001a) also investigated whether there was a difference between EPA and DHA.

Peet et al. (2001b) is the only trial not to use omega-3 fatty acids in addition to existing antipsychotic treatment. In this study, participants were allocated EPA or placebo as sole treatment unless it became necessary to prescribe standard antipsychotic drugs during the trial.

Omega-3 fatty acids were given as various supplements, and the specific dose varied. Fenton et al. (2001) compared 3 g/day E-EPA with placebo but also gave all participants an additional supplement of vitamin E. Peet et al. (2001a) and Peet et al. (2001b) compared 2 g/day EPA with placebo. Peet and Horrobin (2002) compared different doses (1 g/day, 2 g/day or 4 g/day) of E-EPA with placebo. Berger et al. (2007) and Emsley et al. (2006) compared a single dose of 2 g/day E-EPA with placebo, whereas Emsley et al. (2002) compared 3 g/day E-EPA with placebo. The omega-3 supplements used in the RCTs do not all have a licence as medicine in the UK.

The Cochrane review reported 4 main outcomes: global state, mental state, adverse events and leaving the study early. It reported that none of the studies included data on mortality, direct measures of compliance, relapse, patient and carer satisfaction, social functioning or cost of treatment.

The main symptom scales reported included:

  • Clinical Global Impression (CGI) scale:

    • A 3-item clinician-rated scale that measures illness severity (CGI-S), global improvement or change (CGI-C) and therapeutic response.

    • The CGI-S is rated on a 7-point scale with severity ranging from 1 (normal) to 7 (the most severely ill patients). CGI-C ranges from 1 (very much improved) to 7 (very much worse). Therapeutic response ranges from 0 (marked improvement and no side effects) to 4 (unchanged or worse and side effects outweigh the therapeutic effects). Each component of the CGI is rated separately; there is no overall score.

  • Positive and Negative Syndrome Scale (PANSS):

    • A brief clinician-rated scale used to assess mental state and severity of psychopathology. There are 30 items; each item is scored on a scale of 1 (absent) to 7 (extreme). PANSS can be divided into 3 subscales measuring severity of general psychopathology, severity of positive symptoms (PANSS-P) and severity of negative symptoms (PANSS-N). A low score indicates low levels of psychopathology.

The Cochrane review stated that allocation concealment was reported and represented a low risk of bias in all RCTs except Emsley et al. (2002) and Fenton et al. (2001) where it was not described – leading to an unclear risk of bias. Similarly, blinding was reported and represented a low risk of bias in all RCTs except Emsley et al. (2002), where it was not described, also leading to an unclear risk of bias.

Efficacy – overall improvements

Peet et al. (2001b), Fenton et al. (2001) and Berger et al. (2007) reported data for different measures of overall mental health score, described as 'global state' in the Cochrane review.

In Peet et al. (2001b) (where people were not receiving antipsychotics before randomisation), after a maximum follow-up period of 12 weeks, less people receiving omega-3 fatty acids (11/15) needed standard antipsychotics by the end of the trial compared with those receiving placebo (15/15), but this was not statistically significant (n=30, risk ratio (RR) 0.74; 95% confidence interval [CI] 0.54 to 1.02, p=0.067). The same study reported that the mean number of days people in the study were free of standard antipsychotics was significantly less for those randomised to receive omega-3 fatty acids compared with placebo (35.1 days [standard deviation {SD} 34.7] compared with 65.3 days [SD18.9], p<0.02). The effects of treatments were unclear because of the concurrent, additional use of antipsychotic medication part way through the trial.

In Fenton et al. (2001), there was no difference between omega-3 fatty acid and placebo group average CGI end point scores up to 16 weeks (n=87, mean difference 0.00; 95% CI −0.29 to 0.29).

In Berger et al. (2007), there was no significant difference between omega-3 fatty acids and placebo in the number of people achieving a symptomatic response by 12 weeks (n=69, RR 0.90; 95% CI 0.50 to 1.63, p=0.73).

Efficacy – mental state

Peet et al. (2001a) and Peet et al. (2001b) both deemed an improvement of greater than 25% on the PANSS as clinically meaningful. This remains a subject of debate (Irving et al. 2006). Levine et al. (2008) suggest a PANSS percentage change of around 21% to 33% is equivalent to 'minimally improved' on the CGI-C scale. In Peet et al. (2001b) (where people were not receiving antipsychotics before randomisation), more people achieved this level of symptom improvement with omega-3 fatty acids compared with placebo (n=30, RR for not achieving 25% improvement 0.54; 95% CI 0.30 to 0.96, p=0.035, number needed to treat 3; 95% CI 2 to 29). In people already using antipsychotic drugs (Peet et al. 2001a), this outcome was not statistically significant (n=29, RR for not achieving 25% improvement 0.62; 95% CI 0.37 to 1.05, p=0.073).

In Peet et al. (2001b), average PANSS scores by the end of the trial were statistically significantly lower with omega-3 fatty acids compared with placebo (n=26, mean difference −12.50; 95% CI −22.38 to −2.62, p=0.013). In Peet et al. (2001a), this outcome was of borderline statistical significance (n=29, mean difference −10.40; 95% CI −20.35 to −0.45, p=0.041). An unrelated publication Leucht et al. (2006) suggested that a PANSS reduction of between 10 and 15 points is equivalent to 'minimally improved' on the CGI-C scale.

The Cochrane review also suggests that in Emsley et al. (2002), the percentage change in PANSS score from baseline to the end of the trial favoured omega-3 fatty acids over placebo (n=40, mean percentage change in PANNS score 12.6% [SD 12.6] with omega-3 fatty acids compared with 3.1% [SD 13.3] with placebo).

In contrast to these 3 studies favouring omega-3 fatty acids over placebo, in the slightly larger study by Fenton et al. (2001), there was no statistically significant difference in average PANSS scores by the end of the trial (n=87, mean difference −1.00; 95% CI −8.15 to 6.15, p=0.78).

The Cochrane review reported no efficacy outcome data from Peet and Horrobin (2002) or Emsley et al. (2006).

Additional randomised controlled trial

A double-blind, placebo-controlled RCT by Manteghiy et al. (2008) included 85 adult inpatients (average age 37–39 years) with schizophrenia diagnosed using DSM-IV (average illness duration 14–15 years). The study recruited 106 participants, but 21 people were excluded; it is not clear at what stage. Participants were randomised to risperidone (2–8 mg/day) plus placebo, or risperidone (2–8 mg/day) plus omega-3 fatty acids (3 g/day) for 6 weeks. One capsule (1 g) of omega-3 fatty acids contained 360 mg EPA and 240 mg DHA.

No statistically significant differences in PANSS (positive and negative symptoms, general psychopathology, formal thought disorder, agitation and suspiciousness) were found between the 2 groups at any time point assessed (weeks 0, 3 and 6).