An average of 20,000 prescriptions have been issued every year in Scotland to treat alcohol dependency and more than 15,000 prescriptions are issued every month to treat alcohol use disorder problems (‘AUDs’). However, high-quality evidence on these drugs' effectiveness and their comparative effectiveness is limited. Specifically, recent systematic reviews on pharmacologically controlled drinking [i]
and pharmacotherapy for AUDs in outpatient settings [ii]
conclude that no high-grade evidence exists and reveal a lack of head-to-head comparisons in the literature. We addressed this, by providing real world evidence through a nationwide study in Scotland of individuals prescribed medications, following an AUD (or alcohol dependence) hospitalisation between 2010 and 2019. We also provided an overview of the distribution of prescriptions for AUD in Scotland, aiming to assess whether there were significant inequalities in levels of prescribing. We focused our analysis on two cohorts: individuals with a first hospitalisation for alcohol dependence and individuals with a first hospitalisation for alcohol intoxication or harmful use. This evidence was then further informed by patient public involvement, specifically in partnership with the Scottish Recovery Consortium and patients who had experiences of AUD hospitalisations and AUD pharmacological treatments.
Precisely, aimed to answer to three research questions in our study:
- What are the prescribing levels and trends of pharmacological treatments for AUD in Scotland? Do any inequalities exist in prescribing levels by age, sex, and socioeconomic deprivation and health board?
- What short-term health outcomes (alcohol-related hospitalisations and mortality) occur for individuals prescribed different medication types?
- Does the resultant research data set offer a realistic opportunity for carrying out future instrumental variable analysis using physicians’ prescribing preferences?
Data were extracted from National Safe Haven providing data on hospitalisations and prescriptions at a national level in Scotland. To answer to our research questions we used two cohorts defining different AUD severity levels: alcohol dependence (ADS) and alcohol intoxication or harmful use (AIH).
The ADS cohort was identified by patients with a first hospitalisation for alcohol dependence (F10.20 in ICD-10 codes [10]) in the main diagnostic position of the hospital record from 1st January 2010. The first episode was then categorized as an ‘index date’. In order to identify the first episode of hospitalisation for a patient, every patient’s hospital records were checked for any previous alcohol dependence coding (in any diagnostic position) in the preceding 10 years. If previous records were seen, then that hospitalisation was no longer classified as a first hospitalisation.
The AIH cohort was identified by patients with the first hospitalisation for alcohol intoxication or harmful use (F10.0 or F10.1 in ICD-10 codes) in the main diagnostic position as ‘index date’. The same procedure as for ADS cohort was used to ensure only first hospitalisations were recorded.
Research question 1.
We provided descriptive statistics using summary statistics and graphical displays of prescribing levels and prescriptions trends for alcohol dependence in Scotland. We also used a logistic regression to assess whether inequalities exist in AUDs prescriptions across age, sex, socioeconomic deprivation, and health board sub-groups.
Research question 2.
After identifying the most common medications related to treatment of alcohol dependence, descriptive statistics regarding both medications and number of relapses were provided for both cohorts. We then used survival analysis to carry out comparative effectiveness analysis by the two most used drugs for dependence for specific outcomes. Propensity score was used to address potential confounders coming from non-randomised patients and real-world data.
Research question 3.
We constructed an instrument on the previous one and previous twenty prescriptions of a same prescriber to assess whether national level datasets could provide good instrumental variables for comparative effectiveness.
The total number of labelled prescriptions for alcohol dependence grew from 2010 to 2017. After the national shortage of disulfiram, the overall total number of prescriptions for alcohol dependence was stable at a level of around 4030 (2018) prescriptions per month, which is just 100 less than 2016 levels. Acamprosate and disulfiram were the two most prescribed medications for alcohol dependence. In both the cohort the most prescribed drugs were those with an indication to prevent complications related to alcohol abuse such as thiamine and vitamin B. However, prescriptions directly related to alcohol dependence or withdrawal were only 21% and 11% for ADS and AIH cohort, respectively. Excluding prescriptions to prevent complications, age, sex and living in different socio-economic deprivation areas influence odds of being prescribed medications for withdrawal or dependence.
Findings regarding the comparative effectiveness between acamprosate and disulfiram show a difference between the two drugs in preventing from any alcohol related relapse in favour of disulfiram in both the cohorts. However, results were not statistically significant for most of the outcomes.
On a technical note, this study showed how using instrumental variable can be a complementary analysis in comparative effectiveness research.
A significant obstacle in preventing relapses was the actual access to therapies and prescriptions: our data described a low percentage of prescribed medications after the first hospitalisation.
There are inequalities in prescriptions across groups that can reflect a higher propensity in attending GPs visits of certain sub-populations. However, it is more difficult to explain why people in certain subpopulation, already receiving prescriptions to prevent complications related to alcohol abuse, are on average less likely to receive prescriptions for withdrawal or dependence. Further studies on inequalities in prescriptions and access to therapies could shed light on this issue, also with comparison across different conditions.
Economic analyses evaluating both costs and effects are suggested to provide conclusive guidelines for medical therapies for patients with alcohol dependence.
This study, by showing some important features in prescriptions trends for alcohol dependence, highlighted relevant matters for future strategies to tackle AUD in Scotland. In particular, only a minority of patients hospitalised for dependence or abuse are prescribed medications to assist in reducing alcohol dependence or withdrawal. Ultimately, in most commonly prescribed medications for alcohol dependence, we see better outcomes for disulfiram than acamprosate. This question the decrease in prescriptions over time of disulfiram in favour of acamprosate. Additional studies on cost-effectiveness jointly with safety and complication avoidance should be undertaken to build updated guidelines on pharmacological therapies.
A published article on this study appeared in Drug and Alcohol Review: https://onlinelibrary.wiley.com/doi/10.1111/dar.13841
[i] Palpacuer, C., et al., Pharmacologically controlled drinking in the treatment of alcohol dependence or alcohol use disorders: a systematic review with direct and network meta‐analyses on nalmefene, naltrexone, acamprosate, baclofen and topiramate. Addiction, 2018. 113(2): p. 220-237
[ii] Jonas, D.E., et al., Pharmacotherapy for adults with alcohol use disorders in outpatient settings: a systematic review and meta-analysis. Jama, 2014. 311(18): p. 1889-1900.
