LAIs versus oral antipsychotic medications: adherence

Numerous researchers have examined whether LAIs improve adherence as compared with oral medications, but there is no definitive answer. While LAIs decrease how often a patient has to decide whether to take a medication, they do not eradicate adherence issues, as patients may choose to discontinue the monthly or bimonthly injections.

Observational studies, though subject to confounding, provide suggestive evidence that LAIs improve medication adherence in routine practice.7–11 For example, Brnabic et al12 conducted a post hoc analysis of a prospective observational study that collected data at 31 international sites. The study matched 40 participants taking an LAI to an equal number taking an oral antipsychotic medication. The study found that compared to those treated with LAIs, twice as many on oral medications switched, augmented or discontinued their medications.

Randomised controlled trials (RCTs), however, have not found improved adherence with LAIs.13–19 A systematic review and meta-analysis by Leucht et al20 identified five RCTs that reported results on adherence and found no significant difference in adherence between those on LAIs and those on oral medications, although adherence was typically not measured rigorously. The methodological differences between observational studies and RCTs that may explain these conflicting findings are described below.

LAIs versus oral medications: effectiveness

The literature on LAIs and relapse prevention is also conflicting, as illustrated by two meta-analyses by the same research group. The first meta-analysis in 2011 collected outpatient RCTs that compared LAIs with oral antipsychotic medications and lasted for at least 12 months (10 studies, n=1700).20 The analysis found that patients on LAIs are 30% less likely to relapse compared with those on oral antipsychotic medications. This translates to a number needed to treat of 10. The second meta-analysis in 2014 used a broader set of inclusion criteria, including inpatient and outpatient studies of at least 6-month duration, and included two new large studies that did not find an overall advantage for LAIs over oral medications. This will be discussed below.21 Contrary to the earlier study, this second meta-analysis (21 studies, n=5176) concluded that LAIs did not significantly reduce rates of relapse compared with oral antipsychotic medications, except in studies that used ‘first-generation’ antipsychotic LAIs and in studies published in 1991 or earlier. The authors postulate a number of reasons for the superiority of LAIs in these two subgroups, including changes over the decades in definitions and thresholds for relapse, differences between older and newer medications, and recent trends towards the use of lower doses of antipsychotic medications. Since there are striking differences in results between the two meta-analyses, the authors reanalysed the newer data using the same criteria as the first study. By now including the two large additional trials that lasted for more than 12 months and found no advantage for LAIs, this updated analysis showed no significant difference in relapse rates between LAIs and oral medications.

The two recent RCTs that were added to the earlier meta-analysis used slightly different approaches to compare the effectiveness of LAIs with oral medications in preventing relapse.19 ,22 The PROACTIVE study was a double-blind RCT at eight clinical sites that randomised 305 patients with schizophrenia or schizoaffective disorder and a history of an exacerbation in the past year to LAI risperidone or to a newer oral antipsychotic medication of their physician's choice.22 The primary outcomes were rates of relapse and hospitalisation over 30 months. In this broad sample of patients, the PROACTIVE trial found no significant difference between the LAI and oral groups in the primary outcomes of time to first relapse or time to first hospitalisation. Secondary analyses found that patients in the LAI group had greater improvements in some types of psychiatric symptoms.

In their discussion of the PROACTIVE study, Buckley et al22 highlighted methodological issues that may affect the comparative effectiveness research on LAIs versus oral antipsychotics. The authors noted that frequent study contacts, access to free study medications, implicit or explicit efforts to monitor adherence, and patient selection may lead to improved adherence to all medications in these studies. These factors, which are present in RCTs but not in routine practice, could diminish differences between LAIs and oral medications with respect to adherence and relapse prevention compared with what would be expected in usual community care.

The issue of patient selection deserves special focus. Clinical trials, due to the importance of informed consent, select patients who are willing to be enrolled in a clinical trial that involves medications, to have their adherence monitored, and to attend frequent appointments. This population may have greater insight into their illness, greater engagement with their providers, and less non-adherence than the usual patients whom providers would consider for LAI medications. In short, if LAI clinical trials select patients who are relatively adherent and stable, then it is possible that LAIs might have a different effect in the real-world patients with unstable illness who are often prescribed LAIs.23

To address the issue of patient selection, Rosenheck et al19 conducted a 2-year study in the US Veterans Health Affairs system that specifically selected patients with unstable illness by including only those deemed at risk of psychiatric hospitalisation. The double-blind RCT included 369 individuals with schizophrenia who were assigned to LAI risperidone or the clinician's choice of oral antipsychotic medication. Similar to the PROACTIVE study, this study found no significant differences between LAIs and oral drugs in rates of psychiatric hospitalisations, psychotic symptoms, quality of life or social functioning.

Kirson et al24 used meta-analyses to examine the influence of study design on the results of LAI versus oral antipsychotic medication comparative effectiveness research. The authors examined 12 studies that included 19 different LAI–oral medication combinations. They found no superiority of LAIs over oral medications in RCTs, but found significant benefits of LAIs over oral formulations in observational studies. To explain these findings, the authors highlight the tension between methodological rigour and real-world generalisibility in RCTs and observational studies, respectively. RCTs include methods to standardise the conditions between the LAI and oral medication arms such as adherence monitoring and frequent appointments. Kirson et al also describe a possible ‘Hawthorne effect’—where behaviour is affected by the awareness of being observed—that would lead study subjects to have greater adherence to oral medications than they would have in real-world practice because of their knowledge that their adherence is being monitored. The authors conclude that these aspects of RCTs attenuate the advantages of LAIs over oral medications that are seen in observational studies.

Another issue is stage of illness. Some have argued that LAIs may be most beneficial when used for people experiencing a first episode of psychosis, when knowledge about the illness is low and ambivalence about medications is high. A large-scale study, using Finnish national health registries of 2588 participants after a first hospitalisation for schizophrenia, compared LAIs with oral antipsychotic medications in routine practice conditions.10 The authors found that those on LAI haloperidol, risperidone or perphenazine were significantly less likely to discontinue their medication (p<0.0001) compared to those on the oral versions of these medications. When these results were pooled, they found that those on LAIs were 59% less likely to discontinue their medication up to 7 years of follow-up and 36% less likely to be rehospitalised within 2 years. Interestingly, these benefits have also been recently reported in a single-site RCT (n=83) that compared oral with LAI risperidone among those with recent-onset schizophrenia.25 In this study, those treated with the LAI had significantly lower rates of relapse (5% vs 33%) and were four times less likely to be hospitalised compared with those on the oral formulation. These two studies, a large-scale naturalistic study and an RCT, agree in their observations of substantial benefits of LAIs over oral medications in those with recent-onset schizophrenia.

Older LAIs versus newer LAIs

Beginning with clozapine in 1989, a new wave of oral antipsychotic medications that came to be known as second-generation or atypical antipsychotics was released. The initial hope was that the newer medications would be more efficacious than the older antipsychotic medications. However, a considerable amount of research has called into question this initial assumption. For example, the CATIE (Clinical Antipsychotic Trials of Intervention Effectiveness) schizophrenia trial found that several newer antipsychotic medications were not more effective than the older antipsychotic medication perphenazine.26 Other comparative effectiveness trials and meta-analyses comparing newer with older oral antipsychotic medications have shown similar results.27–30 Research has shown that some of the newer antipsychotics are associated with a lower incidence of neurological side effects as well as high rates of metabolic adverse effects compared with older medications.31 ,32 Additionally, increasing evidence suggests that there is considerable variability among the older and newer drugs, which calls into question the validity of the ‘generation’ categories.28 ,33

Few studies have looked at how newer LAIs are compared with older LAIs. Rubio et al34 examined the effects of the drugs on substance use outcomes in an RCT of 115 individuals with schizophrenia and a substance use disorder. Participants were randomly assigned to risperidone LAI or zuclopenthixol LAI on an open-label basis for 6 months. The authors found those on LAI risperidone to have significantly fewer positive urine drug tests and significantly lower ratings on a psychotic symptom scale (positive and negative syndrome scale, PANSS) compared with those on zuclopenthixol LAI.

In contrast, a 2013 study by Lammers et al35 found no significant difference between LAI risperidone and ‘first-generation’ LAIs. The authors conducted a single-site, outpatient retrospective comparative effectiveness study comparing those with schizophrenia spectrum illness on risperidone LAI (n=70) versus ‘first-generation’ LAIs (n=102). Lammers et al found no significant differences between the groups for the primary outcomes of time to treatment discontinuation or hospitalisation, but did find that a higher incidence of extrapyramidal symptoms in patients on ‘first-generation’ LAIs.

Two recent multisite RCTs have taken on this question of newer versus older LAIs.36 ,37 A small trial by Covell et al36 enrolled patients on haloperidol decanoate (n=40) or fluphenazine decanoate (n=22), and then randomised these patients to remain on their current LAI or switch to risperidone LAI. The study period was for 12 months, six of which followed a study protocol followed by 6 months of more naturalistic observation. The primary outcome was time to treatment discontinuation, with a number of secondary outcomes. The authors found that those randomised to switch to risperidone LAI had higher rates of treatment discontinuation and significantly greater increases in weight and prolactin levels.

McEvoy et al37 conducted a large prospective comparative effectiveness trial of the newer LAI paliperidone palmitate to haloperidol decanoate. A Comparison of LAI Medications (ACLAIMS) was a double-blind RCT at 22 clinical sites. The patients (n=311) received LAI paliperidone or haloperidol monthly for as long as 24 months. Inclusion criteria explicitly required a history of medication non-adherence and/or substance use, both of which are risk factors for the primary outcome of efficacy failure. The authors defined efficacy failure as “a psychiatric hospitalisation, a need for crisis stabilisation, a substantial increase in frequency of outpatient visits, a clinician's decision that oral antipsychotic could not be discontinued within 8 weeks after starting the long-acting injectable antipsychotics, or a clinician's decision to discontinued the assigned long-acting injectable due to inadequate therapeutic benefit.” The ACLAIMS trial found no significant difference in efficacy failure between those taking paliperidone palmitate and haloperidol decanoate. However, those in the paliperidone palmitate group had higher serum prolactin levels and gained weight, while the haloperidol decanoate group experienced more akathisia, used more antiparkinsonian medications and lost weight. In summary, ACLAIMS did not find that the newer LAI was superior to the older LAI, but found significant differences in side effects.