Seventy-nine percent of the reviewed studies excluded participants with comorbid mental health diagnoses or concomitant medications prescribed for comorbid mental health diagnoses. Research suggests that transient psychotic symptoms are observed in up to 40% of MA-using populations [79] and possibly more amongst treatment seekers. Forty-two percent of individuals who had used MA in the prior 12 months also reported being diagnosed or treated for a concurrent mental illness—three times as high as the non-illicit drug-using population [80]. Among MA users, the majority report a lifetime prevalence of depression and anxiety [81]. The exclusion of relatively common comorbidities such as polydrug dependence and mental health comorbidities limits the generalisability of many of the studies. For example, the role of antipsychotic and antidepressant medications may differ in patients presenting with psychosis or depression. Similarly, dependence to other substances such as alcohol, benzodiazepines or opioids is also likely to impact upon the safety and efficacy and choice of medications. Medications such as topiramate and naltrexone may be worth further examination in patients with comorbid alcohol use, whereas the role of naltrexone will vary according to opioid status; for example, responding to stimulant use in patients enrolled in opioid agonist treatment. Other comorbidities that continue to be poorly addressed include the management of patients with stimulant dependence and comorbid attention-deficit hyperactivity disorder (ADHD). Accurate diagnoses of ADHD in the context of AMPH/MA use can be complicated [82], and there may be differential effects of medications in patients with both conditions. More research is required regarding pharmacological responses for patients with ADHD and stimulant use disorders. Another key area that has not been adequately addressed in clinical trials is the issue of comorbid sleep disturbances in patients using AMPH/MA, and the likely impact upon the role of different medications.

The studies reviewed here report on a variety of outcomes defined, measured and analysed differently across most publications. The broad selection of outcomes and measures render it difficult to meta-analyse or otherwise collectively synthesise the study results as reported. Future endeavours to standardise outcome measures across clinical trials in addiction medicine would make it easier to interpret study results collectively and better translate research results to clinical practice. Importantly, only three studies reviewed here (7%) provided information on adverse events/serious adverse events, despite the standard reporting format adopted by most publishers (CONSORT [83]) including a minimum standard of harm reporting. This limits the capacity to appropriately assess the risk versus benefit of the pharmacotherapies reviewed here. We elected to include studies in this review irrespective of safety reporting, to provide a comprehensive review of the current status of research.

Although adherence was reported in most of the studies reviewed here, the methods and definitions were discordant across studies. Adherence thresholds varied and were often arbitrary. When plasma was assessed for active study drug/metabolite, there was no measurable metabolite included in the placebo, and the control data is therefore missing. Furthermore, presence of the study drug/metabolite does not necessarily indicate adherent consumption of the study drug, and authors varied in their assessments in that regard (i.e. present or not versus present at a defined level). Studies relying on pill count or self-report lacked critical appraisal of the results. For example, in one study where no participant returned un-used study drug, 100% adherence was inferred as opposed to examining if there were other reasons (e.g. discarding drug).

Definitions of efficacy of pharmacotherapies vary extensively. While some studies define success by abstinence from AMPH/MA, others consider a reduction in use to be a measure of treatment success. Abstinence as an outcome can be determined by self-report, or by negative UDS at time points pre-determined (see Table 5). The desired goal of pharmacotherapy will likely vary depending on the patient, and must be patient-focused and clinically relevant.

Disparate criteria were also used when determining eligibility for a study; for example, the definition of ‘low-use’ (AMPH/MA) in the studies reviewed here was 5 days, 10 days and 18 days of the past 30 in various studies. This is an important limitation in synthesising data and results, and establishment of a clinically meaningful cut-off for regular and frequent use is imperative. Studies that analysed results by baseline frequency of MA/AMPH use often grouped days of use into categories defined as ‘light use’ or ‘heavy use’. However, rationale for cut-points was often missing or ill-defined. In one study, consumption of MA was classified as ‘heavy use’ among participants providing three MA-positive UDS/fortnight, while in another study, ‘heavy use’ was classified as self-reported use of 18 days of the prior 30. In a 2007 paper, Hillhouse et al. [19] found that frequency of MA use prior to treatment predicted both treatment performance and outcomes following treatment in a psychosocial intervention for MA dependence, reporting that participants with baseline use of < 15 days in the 30 days prior to intake had better outcomes. Therefore, the disparity in groupings in pharmacotherapy research across studies and drugs makes it difficult to reliably recommend a strategy for determining cut-points. Further research and debate in this area is required to determine a clinically meaningful way of grouping frequency of use.

Studies conducted in (e.g. men who have sex with men) or excluding (e.g. women) specific populations are limited in their ability to generalise to other populations. The implications of contextual influences on outcomes are unknown. For example, no study we reviewed here assessed specific populations such as indigenous peoples. Similarly, while women were often excluded by the study design, no study examined only women. It is unknown how generalisable any of the results reviewed here are outside of the context in which they were conducted, and it is unwise therefore to combine results across populations.

Finally, because of the similarities in chemical structure and behavioural, psychological and physical effects of AMPH and MA [84], we have included studies of AMPH and MA, and studies that did not distinguish between AMPH and MA. MA and AMPH may be knowingly or unknowingly consumed or co-consumed in uncertain concentrations, with variability over time and place. However, there is little data on which to assess whether there are distinct differences in use disorders due to these two substances; further assessment is required.