A network meta-analysis of randomised controlled trials of antipsychotic medications to assess their comparative efficacy and tolerability in autistic people

This network meta-analysis of 22 randomized controlled trials involving 1,562 autistic individuals found that risperidone and aripiprazole, particularly when combined with parent training, demonstrated the greatest efficacy in reducing irritability and improving global clinical impression compared to placebo, though both medications were associated with significantly increased adverse effects.

The Gist

Imagine you are a parent or caregiver trying to navigate a very crowded, noisy marketplace. The goal is to find the best "tool" to help an autistic person who is struggling with difficult behaviors like extreme irritability, aggression, or agitation. In this marketplace, the "tools" are different antipsychotic medications.

For a long time, doctors have had to guess which tool works best because there hasn't been a single map comparing them all at once. This paper is that map. It's a Network Meta-Analysis, which is like a giant round-robin tournament where every medication gets compared not just to a placebo (a sugar pill), but indirectly to every other medication in the mix.

Here is the breakdown of what the researchers found, using simple analogies:

The Main Event: Taming the "Irritability" Storm

The researchers looked at 22 different studies involving over 1,500 autistic people (mostly children and teens). They focused on a specific score called the ABC-I, which measures how much a person is struggling with irritability and aggression.

Think of the ABC-I score as a "storm meter." A high score means a huge, dangerous storm; a lower score means the weather is calmer.

• The Champion: The most effective way to calm the storm was a combination of Risperidone (a medication) and Parent Training. It was like having a superhero (the drug) and a skilled coach (the parents) working together. This combo lowered the storm meter the most.
• The Runners-Up: If you just used Risperidone alone, it was the next best at calming the storm. Aripiprazole (another medication) came in third.
• The "Maybe" Tools: Other medications like Lurasidone, Valproic acid, and Memantine were tested, but the data didn't show they were significantly better than doing nothing (the placebo).

The "Global" Score: How Much Better Do They Feel?

The researchers also looked at the CGI-I score, which is like a "global health report card" given by a doctor. It asks, "Is this person noticeably better than before?"

• Risperidone and Aripiprazole got the highest grades on this report card. They were the most likely to make a doctor say, "Yes, this patient has improved significantly."

The Cost of Doing Business: Side Effects

Every tool in the marketplace has a price tag. In this case, the price is side effects. The researchers looked at how many people dropped out of the studies because the medicine made them feel terrible, or if they gained weight or felt too sleepy.

• The "Heavy" Side: Risperidone and Aripiprazole were the most effective at stopping the storm, but they also had the highest "price tag" for side effects. People taking these were more likely to experience negative reactions compared to those taking a placebo.
• The Weight Gain Problem: If you are worried about gaining weight, Olanzapine and Valproic acid were the worst offenders. Aripiprazole was the "lightest" option regarding weight gain, though it still carried some risk.
• The Sleepy Side: Valproic acid and Olanzapine were the most likely to make people feel very drowsy (sedated), followed closely by Risperidone.
• The Dropout Rate: Interestingly, even though Risperidone and Aripiprazole had more side effects, the number of people who quit the studies because of them wasn't statistically different from the placebo group. However, the data on this was a bit shaky and hard to pin down.

Important Caveats (The Fine Print)

The authors were very careful to point out the limits of their map:

1. Mostly Kids: Almost all the data came from children and teenagers. There is very little evidence to say if these tools work the same way for adults.
2. Short-Term Glimpse: The studies were like short snapshots in time. We don't know what happens if you use these tools for years.
3. The "Add-On" Confusion: Many of these studies were "add-on" trials, meaning the kids were already taking other things (like other meds or therapies). It's hard to tell if the improvement came from the new tool or the old ones they were already using.
4. Funding: Some of the biggest studies were paid for by the companies that make the drugs. While the authors say this didn't bias the results, they noted that more independent research is needed.

The Bottom Line

If you are looking for the most effective way to reduce irritability and aggression in autistic children based on this specific study:

• Best Combo: Risperidone + Parent Training.
• Best Single Drug: Risperidone or Aripiprazole.

However, these "best" options come with a higher risk of side effects. The authors remind us that guidelines suggest trying non-drug methods (like behavioral support) first, and if drugs are needed, the choice depends on weighing the benefit of calming the storm against the cost of the side effects.

Crucially, the paper does not tell doctors exactly which drug to prescribe for every patient. It simply provides the evidence map showing which tools performed best in the studies they reviewed, while highlighting that the evidence for adults is missing and long-term effects are unknown.

Technical Summary

Technical Summary: A Network Meta-Analysis of Antipsychotic Medications in Autistic People

Problem Statement
A significant proportion of autistic individuals receive off-license antipsychotic medication, frequently prescribed for "behaviours that challenge" (BtC) rather than comorbid psychiatric conditions like psychosis. While guidelines from the National Institute for Health and Care Excellence (NICE) and international bodies recommend non-pharmacological interventions (e.g., Positive Behaviour Support) as first-line treatments, antipsychotics remain prevalent in clinical practice. Despite recent pairwise meta-analyses, there is a lack of comprehensive evidence comparing the relative efficacy and tolerability of different antipsychotic agents against each other and placebo. This gap hinders clinicians' ability to make evidence-based decisions regarding which specific medication, if any, should be utilized.

Methodology
The authors conducted a network meta-analysis (NMA) of randomized controlled trials (RCTs) following the PRISMA-NMA guidelines.

• Data Sources: Seven databases (including Embase, Medline, Cochrane, PsycINFO) were searched from inception to September 1, 2024, alongside hand-searching of ten relevant journals and grey literature.
• Inclusion Criteria: RCTs involving autistic participants of any age, testing any antipsychotic medication (or combinations with non-pharmacological interventions) against placebo, other medications, or non-pharmacological interventions. Crossover trials were included only if Phase I data were available.
• Participants: The final analysis included 22 RCTs involving 1,562 autistic individuals (1,326 completed the studies). The population was predominantly children and adolescents (2–17.5 years), with 81.3% male. Only one small study included adults.
• Statistical Analysis: Analyses were performed using a frequentist framework via the MetaInsight web-based app. Both fixed and random-effects models were computed, with results reported from the more conservative model. Interventions were ranked using P-scores (0 to 1, where higher values indicate greater efficacy).
• Outcomes:
  • Primary: Psychopathology, ASD core symptoms, and associated behaviours (specifically measured via the Aberrant Behaviour Checklist-Irritability [ABC-I] and Clinical Global Impression-Improvement [CGI-I] scales).
  • Secondary: Adverse effects (AE), sedation, weight gain, and dropout rates due to AEs.
• Quality Assessment: Two authors independently assessed risk of bias using the Cochrane Risk of Bias 2.0 (RoB 2.0) tool.

Key Results

• Efficacy (ABC-I Score): The combined intervention of risperidone plus parent training demonstrated the largest mean difference compared to placebo (MD = -11.16, 95% CI: -15.13 to -7.18), followed by risperidone alone (MD = -7.59) and aripiprazole (MD = -5.59). These three interventions held the top ranks in the treatment hierarchy. Other agents (lurasidone, divalproex sodium, memantine) showed no significant effect, with confidence intervals crossing the no-effect line.
• Efficacy (CGI-I Score): Risperidone (OR = 7.65) and aripiprazole (OR = 7.02) showed the greatest odds of improvement compared to placebo. However, the authors noted that treatment rankings for CGI-I were not robust, as sensitivity analyses revealed that removing specific studies (e.g., Marcus et al., 2009) altered the heterogeneity and ranking stability.
• Adverse Effects:
  • Overall Adverse Effects: Risperidone (OR = 4.00) and aripiprazole (OR = 2.77) had significantly higher odds of adverse effects compared to placebo. Methylphenidate showed a high odds ratio but with wide confidence intervals crossing the null.
  • Sedation: Valproic acid, olanzapine, risperidone, and aripiprazole showed the highest odds of sedation, though confidence intervals for olanzapine and others were wide. Notably, the combination of risperidone and parent training did not show a statistically significant increase in sedation compared to placebo.
  • Weight Gain: Olanzapine, valproic acid, and risperidone were associated with the highest odds of weight gain. Aripiprazole demonstrated the least association with weight gain, although the difference from placebo was not statistically significant.
• Dropout Rates: No intervention showed a statistically significant difference in dropout rates due to adverse effects compared to placebo, a finding the authors attribute to unreliable data with wide confidence intervals.

Significance and Claims
The paper claims to be the first network meta-analysis to simultaneously assess the relative efficacy and tolerability of antipsychotics specifically for autistic people. Its primary contribution is the demonstration that combining pharmacological therapy (risperidone) with non-pharmacological interventions (parent training) yields superior efficacy on irritability scores compared to pharmacological monotherapy.

The authors emphasize that while risperidone and aripiprazole are effective for reducing irritability and improving global impression, they carry a significantly higher risk of adverse effects. The study underscores the necessity for clinicians to weigh efficacy against tolerability profiles. Furthermore, the authors highlight critical limitations in the current evidence base:

1. Demographic Gap: Evidence is almost exclusively derived from children and adolescents; there is insufficient data to draw conclusions for autistic adults.
2. Outcome Limitations: No RCTs assessed the impact on core psychiatric disorders (e.g., psychosis, bipolar disorder) or quality of life.
3. Study Design: Most studies were short-term, and many were industry-sponsored (particularly the large aripiprazole trials), necessitating more independent, long-term research.
4. Transitivity: The lack of reported IQ and intellectual disability (ID) severity in most studies limits the ability to fully verify transitivity assumptions regarding these covariates.

The authors conclude that while antipsychotics have a role, they should be considered only after non-pharmacological interventions, and future research must focus on properly powered RCTs that compare combined interventions against non-pharmacological treatments alone, while addressing the lack of data for adults and long-term outcomes.