In the UK, medications must be licensed for specific purposes before they can be used, ensuring that thorough safety and quality checks have been carried out. In some cases, medications can be used ‘off-license’, which means it has not been formally approved for that use.

Antipsychotics are mainly licensed for use in psychotic illnesses, although in certain cases they are recommended and can be used ‘off-license’ for other conditions. However, we know that antipsychotics are associated with a number of adverse effects, including metabolic, endocrine and cardiovascular changes (BNF, 2014). There is also evidence of increased mortality for typical and atypical antipsychotics, particularly in those with dementia (Schneider et al, 2005; Gill et al, 2005; Trifiro et al, 2007). In light of this, prescribers must include a large dose of caution with their script and ensure they are adhering to UK guidelines.

A group of researchers from University College London set out to examine the indications for prescribed antipsychotics and evaluate the prescribing patterns for these drugs in a cohort study (Marston et al, 2014).

Methods

This observational study examined antipsychotic prescriptions in primary care settings, using data from a UK research database called THIN (The Health Improvement Network). THIN data is derived from primary care practices using Vision software and includes demographic and deprivation data, diagnoses and symptoms (entered using Read codes, the general practice coding system) and records of all prescriptions issued, linked to the British National Formulary.

The researchers looked at all individuals on the database who had received at least one antipsychotic prescription in primary care in the UK during a five-year period and recorded:

• Indication for the antipsychotic
• Rates of prescribing
• Duration of treatment
• Average dose
• Age
• Level of social deprivation

The individuals were divided into three cohorts for the purpose of analysis:

1. Those with Severe Mental Illness (SMI) (psychosis or bipolar)
2. Those without SMI but with other mental health diagnosis (e.g. depression, personality disorder, dementia)
3. Those with none of the above recorded

Results

The researchers identified and focused on the three most commonly used First Generation Antipsychotics (FGAs):

1. Haloperidol
2. Chlorpromazine
3. Trifluoperazine

and the same for the three most commonly used Second Generation Antipsychotics (SGAs):

1. Risperidone
2. Olanzapine
3. Quetiapine

The ‘population at risk’ was determined as the total population of the GP practices over the five-year period (109,170 patients from 529 practices). The analysis gave results in terms of Incident Rate Ratios (think of these as similar to Relative Risk: a ratio of 1 indicates the same chance of an outcome in one group as another; a ratio of >1 indicates that the rate is higher in the group exposed to the factor they are examining).

Key results

• Median daily doses and duration of treatment were both found to be higher in the SMI cohort
• Those aged 80 and over were twice as likely to receive antipsychotics as those in their 40s (IRR 2.234; 95% CI 2.222 to 2.246)
• Those living in the most deprived areas were more than three times as likely to receive antipsychotics as those in the least deprived areas (IRR 3.587; 95% CI 3.567 to 3.606)
• Over half of those receiving FGAs had no diagnosis of SMI – most did have codes for other mental health conditions but between 12% and 17% had neither of these categories of diagnosis recorded
• 6% of those receiving Olanzapine, 7% of those receiving Quetiapine and 13% of those receiving Risperidone had no diagnosis of SMI or other mental health condition
• 26% of those receiving Quetiapine had a diagnosis of dementia
• Over the course of the five years of the study, prescribing rates of all of the three most commonly prescribed FGAs decreased, while the rate for all of the 3 most commonly prescribed SGAs increased.

The biggest deficit of justification for antipsychotics was seen in the group prescribed FGAs. So where is the missing percentage of prescriptions? (Or, to be precise, the missing 73%, 65% and 70% for Haloperidol, Chlorpromazine and Trifluoperazine respectively.) Antipsychotics were prescribed for conditions such as anxiety, depression, dementia, sleep problems, Attention Deficit Hyperactivity Disorder and personality disorders. Antipsychotics are recommended and useful in some isolated cases outside of the psychotic illnesses group, but it is surprising to find that these conditions account for well over half of antipsychotic prescriptions.

Strengths and limitations

Initially I was concerned about the scope for selection bias by using THIN, but the database has been found to be representative of the UK population in terms of both demographics (Blak et al, 2011) and prevalence of SMI (Nazareth et al, 1993) and it provided a very large study sample. The narrow confidence intervals from their analyses indicate precise results.

The study only included practices that were using their software fully and where, theoretically, all prescriptions would have been generated electronically. From my own experience I know many prescriptions in primary care are prescribed via paper scripts on home visits or when the computer/printer/logon system isn’t working, relying on clinicians entering this information when they return to the practice or their technology returns to them, leaving room for missed information.

Objective measurements were used and all subjects were classified into diagnostic groups using Read codes. Read codes are less familiar in mental health research, given the tendency to use ICD or DSM classifications, but they can be cross-mapped to ICD-10 diagnoses. Of course, this relies on the Read codes being correctly entered at all times. The researchers also compared patients’ Read codes to their presence on the practice SMI register, which all GP practices are obliged to have, ensuring they identified all eligible individuals.

The study itself spanned a five year period, but subjects were lost to follow up if they died, changed GP practice or reached the age of 100 before the end of the five years. The study also excluded those with less than 6 months of follow-up data from the date of the first antipsychotic prescription. There was no blinding in the study, although this is less likely to affect results given the retrospective and observational nature of the research.

The authors acknowledge that using the median daily dose is a crude method of quantifying the amount prescribed for each indication, but they argue that this allows for comparison between groups. The researchers excluded any doses at over twice the maximum recommended dose in the BNF (221 prescriptions excluded for this reason, thought to represent erroneous entries, but this could have skewed data). Prescriptions issued in secondary care would also not have been identified (this would apply to all prescriptions of Clozapine, for example).

The other major limitation with this type of study is that prescriptions did not necessarily match up with diagnoses in terms of timeframe. The authors took into account that clinicians would not necessarily re-enter diagnoses every time the patient is seen, but this does mean that a diagnosis of, for example anxiety or depression, could have been entered at one time, then a prescription of antipsychotics started four years later for a different and perhaps undocumented reason, but the correlation is made because of their previous diagnosis.

Conclusions

The findings support what we already know about high rates of antipsychotic prescribing in those with dementia and older people, despite the well-documented increased risk of stroke and increased mortality, although it is surprising to see so many antipsychotic prescriptions in people without any mental disorder recorded at all.

The authors make the valid point that, according to current UK policy, physical health monitoring is recommended only for people on the practice SMI register but, with these findings, there would be a good case for widening this recommendation to include all people prescribed antipsychotics.

Overall this was a good study design and provided consistent results across the cohorts and between medications. The subjects in the study are likely to be very similar to those seen in primary care across the UK and the findings should alert any clinician to raise their awareness of:

• Documenting the rationale for antipsychotic prescriptions
• Using antipsychotic medication within its license
• If it is used off-license, to consider the benefits and risks and clearly document the justification for this
• Physical health monitoring in all people prescribed antipsychotics, not just those on the SMI register

Links

Marston L et al. Prescribing of antipsychotics in UK primary care: a cohort study. BMJ Open 2014; 4: e006135. doi:10.1136/ bmjopen-2014-006135

Joint Formulary Committee. British National Formulary. 68 ed. London: BMJ Group and Pharmaceutical Press; 2014.

Schneider LS, Dagerman KS, Insel P. Risk of death with atypical antipsychotic drug treatment for dementia: meta-analysis of randomised placebo controlled trials. JAMA. 2005: 294 (15; 1934-43).

Gill SS et al. Atypical antipsychotic drugs and risk of ischaemic stroke; population based retrospective cohort study. BMJ. 2005: 330 (7489; 445).

Trifiro G et al. All cause mortality associated with atypical and typical antipsychotics in demented outpatients. Pharmacoepidemiol. Drug Saf. 2007: 16 (5; 538-44). [PubMed abstract]

Blak BT et al. Generalisability of The Health Improvement Network (THIN) database: demographics, 
chronic disease prevalence and mortality rates. Inform Prim Care 
2011; 19:251–5. [PubMed abstract]

Nazareth I, King M, Haines A, et al. Accuracy of diagnosis of psychosis on general practice computer system. BMJ 1993; 307:32–4.

MHRA (2012). Antipsychotics: initiative to reduce prescribing to older people with dementia: drug safety update. Medicines and Healthcare Products Regulatory Agency, 22 May 2012.