The use of benzodiazepines by opioid-dependent people is widespread.1–4 The 2011 Victorian Illicit Drug Reporting System (IDRS), a sentinel survey of people who inject drugs (PWID), reported 92% lifetime and 71% recent (in the past 6 months) use among PWID.5 PWID use benzodiazepines for a number of reasons: to enhance the intoxicating effects of heroin or other opioids, manage anxiety, or manage withdrawal symptoms.3

The contribution of benzodiazepines to heroin overdose is well established.6,7 Alprazolam is a benzodiazepine registered in Australia for short-term treatment of anxiety and panic disorder. It is not recommended as first-line treatment because of concerns about risks of dependence and its potential for misuse.4,8

Alprazolam, like other commonly misused drugs, has a rapid onset and offset of action and high potency.8 Alprazolam may also be more toxic in overdose than other benzodiazepines.9 Laboratory-based studies have found that in combination with methadone, alprazolam has significant effects on respiration.10 A review of the interaction concluded that most evidence suggests the interaction is pharmacodynamic in nature.11

Victorian IDRS reports showed recent alprazolam use increased from 8% in 200512 to 69% in 2011.5 Alprazolam is now the most commonly injected benzodiazepine,5 with a reported street price of three tablets for $10.13 Alprazolam use is associated with disproportionate levels of harm, including amnesia, violent outbursts of rage in otherwise non-violent individuals, and theft.1,13–15 In Victoria, most alprazolam (81%) used by PWID in 2011 was obtained from illicit sources.5

Given the increased number of episodes of serious harm associated with alprazolam use, we aimed to examine its public health impact, to inform prescribing and to guide appropriate policy responses. We investigated trends in alprazolam prescribing and its detection in heroin-related deaths (HRDs) in Victoria. Our hypothesis was that increased mean consumption of alprazolam is likely to have significant effects on heroin users,16 a population already vulnerable to drug toxicity.

Our study over the 21 years from 1990 to 2010 showed a number of interesting trends in alprazolam prescribing and supply and its relationship to HRDs in Victoria. First, the supply of alprazolam increased despite its status as a second-line treatment for its approved indications; second, the increase in the supply of the high-dose formulation was disproportionate to the increase in other formulations; and third, the rate of detection of alprazolam in HRDs increased more rapidly after 2005, concurrently with other reports of increasing harm among PWID.13 The association between the detection of alprazolam in HRDs and alprazolam supply was strong and significant. While alprazolam may be more toxic in overdose than other benzodiazepines,9 the accelerated rate of detection in this population since 2005 could reflect an increased preference for and use of alprazolam,1 particularly the high-dose formulation, among heroin users.

This raises questions about the increased prescribing of a drug not preferred for treatment of its primary indication,8,18 and for which little evidence exists for effectiveness beyond short-term use.4 This is especially important given that it may be more toxic in overdose.9 We have shown that the proportion of HRDs in which this benzodiazepine was detected increased over time as supply increased. Understanding the reasons for the increasing average population-level consumption of alprazolam may help to decrease its supply and the harmful effects seen among PWID. This would be consistent with a previous study that showed that the average consumption of potentially harmful products such as salt and alcohol predicts the number of people affected in the statistically “deviant” tail end of a population distribution.16,19

The number of HRDs has remained fewer than during the heroin glut in the late 1990s,20 which led to the peak in deaths shown in 1999. The lower numbers are likely to reflect trends in heroin supply and should not be interpreted as evidence that alprazolam is relatively safe.

A relative strength of our study is the reporting of all Victorian HRDs spanning a 21-year period, enabling the identification of long-term trends in alprazolam used shortly before death. These data provide valuable information for the future prevention of deaths among people who use heroin. In addition, the prescription data and DDD calculations are estimates of the total number of prescriptions dispensed, based on data from the ASM and Medicare. Incorporating the ASM data improves the accuracy of total prescription volume through the inclusion of private prescriptions.

The finding of a strong and statistically significant association between detection of alprazolam in cases of HRD and its supply in the community is useful for generating a hypothesis about possible causes of increasing detection of this drug in cases of HRD. However, this does not mean a causal relationship exists between the increasing alprazolam supply and such deaths. The contribution of alprazolam to deaths involving multiple drugs is difficult to determine, and it is therefore not possible to specify the proportion of cases of drug toxicity due to combined drugs where alprazolam contributed directly to death.21 We used detection of alprazolam as an indication of use by PWID, rather than contribution to death per se.

The absolute number and rate of cases of HRD in which alprazolam was detected has increased substantially since 2005. Concern about the misuse of alprazolam in 2010 led to a request to the Australian National Drugs and Poisons Schedule Committee to reschedule it to the more restrictive Schedule 8.22 Among the committee’s stated reasons for not doing so at that time was that there was insufficient evidence of a problem.

This study provides further evidence of the increasing problem, perhaps involving high-dose formulations, of use of diverted medications among PWID.1 Given the growing concerns with alprazolam use among PWID and its increasing involvement in HRDs, supply control measures — such as better monitoring and surveillance (including real-time prescription monitoring), rescheduling to Schedule 8, and education of health professionals — are warranted. Provision of information about the risks of concurrent use of opioids and alprazolam to PWID is also essential.

2 Estimated alprazolam defined daily dose (DDD) per 1000 population per day, by dose formulation, Victoria, 1990–2010

3 Number of heroin-related deaths (HRDs) and proportion of cases in which alprazolam was detected, Victoria, 1996–2010

4 Relationship between annual proportion of heroin-related deaths (HRD) in which alprazolam was detected and the annual defined daily dose (DDD)
per 1000 population per day, Victoria, 1990–2010*

* Yearly figures are plotted and labelled with the number of alprazolam-detected HRD in that year; any years with zero alprazolam-detected HRD deaths had the number of alprazolam HRD set to
0.5 and the proportion of all HRD calculated from this. ◆

To the Editor: The impact of population ageing on health care expenditure was a hot topic in the late 1990s in Australia.1 Previous studies have adopted various modelling approaches to estimate age- and sex-specific health care costs to inform population health care expenditure projections.2–4 Medicare funds about 3800 medical services, including consultations provided by general practitioners and specialists, medical diagnostic services (such as medical imaging and pathology), dental surgery, optometry and selected allied health services through the Medicare Benefits Schedule (MBS). It also funds about 2100 pharmaceutical therapies, or about 80% of all prescription medications dispensed in Australia, through the Pharmaceutical Benefits Scheme (PBS).

This letter reports annual Medicare costs at the person level for 10-year age and sex groups, drawing on de-identified Medicare linked data. Medicare randomly selected 10 000 men and 10 000 women from each of the following 2009 age groups, irrespective of their associated Medicare expenditure: 15–24, 25–34, 35–44, 45–54, 55–64, 65–74 and 75–84 years. Health care cost data for each of the samples were sourced from the MBS and PBS databases for the period 1 July 2005 to 31 December 2008. Costs were annualised and mean costs per person were calculated for each sample in 2007 Australian dollars.5

Medicare costs increased with each 10-year age increment, with the exception of a decrease in costs observed in the 75–84-year age group, which is likely to have resulted from the death of patients who previously had heavy health care use. Medicare costs per person peaked for both sexes in the population aged 65–74 years (women, $4518; men, $4215), where costs were 3.7 times and 5.6 times higher than equivalent costs for women and men aged 15–24 years. Medicare costs per capita were consistently higher for women. The differential was most pronounced for the younger age groups, particularly the 25–34-year and 35–44-year age groups where costs for women were more than double the equivalent costs for men (Box).

Observed health care cost data for demographic groups can be used to inform future Australian health care expenditure projections.