Background: Mental health problems and insomnia affect up to half of all critical care survivors within a year of hospital discharge. Such conditions contribute to increased morbidity and mortality observed in survivors. Studies undertaken in Canada, the Netherlands, and Sweden have shown that psychotropic medicines such as anxiolytics and hypnotics (e.g., benzodiazepines and z-drugs) are commonly prescribed to intensive care unit (ICU) survivors after hospital discharge, but there is limited evidence on the safety of such prescriptions in survivors. My PhD work is comprised of four retrospective studies which examined community prescribing to adults following hospitalisation for critical care. The studies: 1) described psychotropic prescribing patterns in Lothian, Scotland from 2012–2019; 2) estimated the association of critical care and new psychotropic prescribing in hospitalised survivors; 3) assessed patient factors associated with new benzodiazepine or z-drug prescribing; and 4) estimated the risk of rehospitalisation and death associated with benzodiazepine and z-drug prescribing after critical care. I performed all data cleaning, analyses, and visualisations in R.
Descriptive study: The aim of my first study was to describe patterns of community psychotropic prescribing after hospitalisation for critical care in Lothian, Scotland from 2012–2019. I found that one-third (33.7%) of survivors were prescribed psychotropic medicines within 90 days of hospital discharge (25.4% antidepressants; 14.8% anxiolytics/hypnotics; and 4.3% antipsychotics/mania medicines). Prescribing patterns did not vary extensively over the study period. I also examined prescriptions by subgroup and generic name. Serotoninergic antidepressants (48%) were the most common antidepressant prescribed. Benzodiazepines (71%) were the most common anxiolytic/hypnotic prescribed (particularly diazepam (47.5%)). Lastly, second-generation antipsychotics (62%) were the most frequently prescribed subgroup of antipsychotic/mania medicines.
Cohort studies: The primary aim of my first analytic cohort study was to examine the association of critical care and community psychotropic prescribing, by comparing critical care survivors to non-critical care hospitalised survivors using multivariable Cox regression. Among patients without psychotropic prescriptions within 180 days prior to hospitalisation, the critical care group had a higher incidence of psychotropic prescribing within 90 days of hospital discharge (10.3%; 1610/15,609) compared with the non-critical care group (3.2%; 9743/307,429); unadjusted hazard ratio (HR) 3.39, 95%CI 3.22–3.57. After adjustment for potential confounders, the risk remained elevated (adjusted HR 2.03, 95% confidence interval (CI) 1.91–2.16) and persisted later in follow-up (90–365 day; adjusted HR 1.38, 95%CI 1.30–1.46). The focus of my second cohort study was restricted to benzodiazepine and z-drug prescribing. I examined patient characteristics and practice variation associated with new community benzodiazepine or z-drug prescriptions in critical care survivors without pre-admission prescribing. Patient characteristics assessed included sociodemographic information, medical history, and in-hospital factors. Using the UK Clinical Practice Research Datalink (CPRD) Aurum datasets of adults hospitalised in 2010 and 2018, I performed multilevel multivariable logistic regression for new (any prescription within 90 days) and for new-and-persistent (2+ prescriptions within 180 days) benzodiazepine or z-drug prescribing, as well as evaluated variation by primary care practice. I found that 5.2% (2769/52,846) of treatment-naïve survivors were prescribed a benzodiazepine or z-drug after hospital discharge, with almost half of those having new-and-persistent prescribing (2.5% of total, 1311/52,846). Zopiclone was the most common drug prescribed (50%) followed by diazepam (19%). A history of insomnia (adjusted odds ratio (OR) 1.96; 95%CI 1.74–2.21), anxiety or depression (adjusted OR 1.40; 95%CI 1.28–1.53), and recent opioid prescription (adjusted OR 1.47; 95%CI 1.34–1.61) were associated with new community prescription. The combination of opioid and benzodiazepine has been associated with increased risk of hospitalisation and death in other studies, making this finding concerning. After adjusting for other confounders, sex was not associated with new prescription and older patients were less likely to receive a prescription. Among new prescriptions, 2.6% of the variation was attributable to the prescribing practice. The aim for the final cohort study was to assess the risk of adverse events associated with benzodiazepine or z-drug prescribing in adult critical care survivors hospitalised in 2010 or 2018 (not limited to treatment-naïve patients as in the previous study). I used linked data from the UK CPRD to evaluate the risk of rehospitalisation or death due to falls or trauma-related events and due to any cause, comparing critical care survivors prescribed benzodiazepine or z-drugs to those not prescribed. I used risk-set matching: each exposed patient was matched to up to five unexposed patients by days from hospital discharge to prescription (or match date in unexposed patients), as well as by primary care practice and cohort year. Patients were followed up for 30 days from prescription or match date for outcome assessment. HRs were estimated using stratified Cox regression and adjusted for confounders. Additionally, I performed subgroup analyses of treatment-naïve patients. For the full cohort, I matched 4884 exposed survivors to up to five unexposed survivors (n=23,834). The median days to prescription (and match date in the unexposed) was 11 days (interquartile range (IQR) 4–25 days). Prescription of benzodiazepines or z-drugs showed no conclusive evidence of increased risk of falls or trauma-related events in the full cohort (adjusted HR 1.27; 95%CI 0.76–2.14) or in treatment-naïve individuals (adjusted HR 1.79; 95%CI 0.61–5.26), because estimates lacked precision due to low event rates. For all-cause rehospitalisation or death, benzodiazepines or z-drugs were associated with increased risk (full cohort adjusted HR 1.24, 95%CI 1.14–1.36; treatment-naïve adjusted HR 1.66, 95%CI 1.49–1.86). However, after excluding patients treated for palliative care, the association persisted only in treatment-naïve individuals (full cohort adjusted HR 1.08, 95%CI 0.98–1.19; treatment-naïve adjusted HR 1.42, 95%CI 1.25– 1.62).
Conclusions: This research demonstrated that one third of adult survivors of critical illness received a psychotropic prescription within 90 days after hospital discharge, while one in ten psychotropic-naïve critical care survivors received a new psychotropic prescription. Prospective studies and/or availability of in-hospital prescribing data are needed to address whether community prescriptions are the result of recommendations from the discharging hospital, inappropriate continuation, or initiation post-discharge. Focussing on community benzodiazepine or z-drug prescribing, I found that one in 20 treatment-naïve adult critical care survivors received a new community prescription within 90 days of hospital discharge, with almost half of those receiving more than one such prescription. Survivors of critical illness can have new or worsened physiological impairments, making them potentially vulnerable to adverse events of these medicines. Indeed, this research demonstrated that community benzodiazepine and z-drug prescribing was associated with increased risk of all-cause rehospitalisations and deaths in critical care survivors who had not been prescribed these before hospitalisation. Clinicians should balance the possible benefits with the likely harms of prescribing these drugs in this potentially vulnerable patient group.
