Readmission Prediction for Patients with Heterogeneous Medical History: A Trajectory-Based Deep Learning Approach

Hospital readmission refers to the situation where a patient is re-hospitalized with the same primary diagnosis within a specific time interval after discharge. Hospital readmission causes $26 billion preventable expenses to the U.S. health systems annually and often indicates suboptimal patient care. To alleviate those severe financial and health consequences, it is crucial to proactively predict patients’ readmission risk. Such prediction is challenging because the evolution of patients’ medical history is dynamic and complex. The state-of-the-art studies apply statistical models which use static predictors in a period, failing to consider patients’ heterogeneous medical history. Our approach – Trajectory-BAsed DEep Learning (TADEL) – is motivated to tackle the deficiencies of the existing approaches by capturing dynamic medical history. We evaluate TADEL on a five-year national Medicare claims dataset including 3.6 million patients per year over all hospitals in the United States, reaching an F1 score of 87.3% and an AUC of 88.4%. Our approach significantly outperforms all the state-of-the-art methods. Our findings suggest that health status factors and insurance coverage are important predictors for readmission. This study contributes to IS literature and analytical methodology by formulating the trajectory-based readmission prediction problem and developing a novel deep-learning-based readmission risk prediction framework. From a health IT perspective, this research delivers implementable methods to assess patients’ readmission risk and take early interventions to avoid potential negative consequences.

Published models that predict hospital readmission: a critical appraisal

INTRODUCTION

The number of readmission risk prediction models available has increased rapidly, and these models are used extensively for health decision-making. Unfortunately, readmission models can be subject to flaws in their development and validation, as well as limitations in their clinical usefulness.

OBJECTIVE

To critically appraise readmission models in the published literature using Delphi-based recommendations for their development and validation.

METHODS

We used the modified Delphi process to create Critical Appraisal of Models that Predict Readmission (CAMPR), which lists expert recommendations focused on development and validation of readmission models. Guided by CAMPR, two researchers independently appraised published readmission models in two recent systematic reviews and concurrently extracted data to generate reference lists of eligibility criteria and risk factors.

RESULTS

We found that published models (n=81) followed 6.8 recommendations (45%) on average. Many models had weaknesses in their development, including failure to internally validate (12%), failure to account for readmission at other institutions (93%), failure to account for missing data (68%), failure to discuss data preprocessing (67%) and failure to state the model's eligibility criteria (33%).

CONCLUSIONS

The high prevalence of weaknesses in model development identified in the published literature is concerning, as these weaknesses are known to compromise predictive validity. CAMPR may support researchers, clinicians and administrators to identify and prevent future weaknesses in model development.

Frequent identical admission-readmission episodes are associated with increased mortality

Frequent emergency readmissions may associate with health consequences. We examined the association between readmissions within 28 days of hospital discharge and mortality in 32,270 alive-discharge episodes (18-107 years). Data collected between 1 April 2017 and 31 March 2019 are presented as age- and sex-adjusted hazard ratios (HR) with 95% confidence interval (CI).Compared with no readmission, mortality risk over a 2-year period was increased with one non-identical admission-readmission (AR) episode: HR = 2.4 (2.2-2.7), two or more non-identical AR episodes: HR = 3.0 (2.7-3.4), one identical AR episode: HR = 4.7 (3.6-6.1) and two or more identical AR episodes: HR = 5.0 (3.8-6.7). Eight conditions associated with AR episodes had increased risk of mortality including congestive heart failure: HR = 2.7 (2.2-3.2), chronic pulmonary obstructive disease: HR = 3.0 (2.5-3.6), pneumonia: HR = 2.0 (1.8-2.3), sepsis: HR = 2.2 (1.9-2.5), endocrine disorders: HR = 1.9 (1.6-2.3), urinary tract infection: HR = 1.5 (1.3-1.7), psychiatric disorders: HR = 1.5 (1.1-2.1) and haematological disorders: HR = 1.5 (1.2-1.9). Frequent identical AR episodes, particularly from chronic and age-related conditions, are associated with increased mortality.

Predictive risk score for unplanned 30-day rehospitalizations in the French universal health care system based on a medico-administrative database

Background

Reducing unplanned rehospitalizations is one of the priorities of health care policies in France and other Western countries. An easy-to-use algorithm for identifying patients at higher risk of rehospitalizations would help clinicians prioritize actions and care concerning discharge transitions. Our objective was to develop a predictive unplanned 30-day all-cause rehospitalization risk score based on the French hospital medico-administrative database.

Methods

This was a retrospective cohort study of all 2015 discharges from acute-care inpatient hospitalizations in a tertiary-care university center comprising four hospitals. The study endpoint was unplanned 30-day all-cause rehospitalization via emergency departments, and we collected sociodemographic, clinical, and hospital characteristics based on hospitalization database computed for reimbursement of fees. We derived a predictive rehospitalization risk score using a split-sample design and multivariate logistic regression, and we compared the discriminative properties with the LACE index risk-score.

Result

Our analysis included 118,650 hospitalizations, of which 4,127 (3.5%) led to rehospitalizations via emergency departments. Variables independently associated with rehospitalization were age, gender, state-funded medical assistance, as well as disease category and severity, Charlson comorbidity index, hospitalization via emergency departments, length of stay (LOS), and previous hospitalizations 6 months before. The predictive rehospitalization risk score yielded satisfactory discriminant properties (C statistic: 0.74) exceeding the LACE index (0.66).

Conclusion

Our findings indicate that the possibility of unplanned rehospitalization remains high for some patient characteristics, indicating that targeted interventions could be beneficial for patients at the greatest risk. We developed an easy-to-use predictive rehospitalization risk-score of unplanned 30-day all-cause rehospitalizations with satisfactory discriminant properties. Future works should, however, explore if other data from electronic medical records and other databases could improve the accuracy of our predictive rehospitalization risk score based on medico-administrative data.

Disparities in health outcomes among seniors without a family physician in the North West Local Health Integration Network: a retrospective cohort study

BACKGROUND

The relationship between having a family physician and in-hospital and postdischarge health outcomes among older adults is unclear. We ascertained the proportion of seniors who did not have a family physician and were admitted to an Ontario tertiary care centre, and we determined the association between having/not having a family physician and in-hospital mortality, 1-year mortality and readmission after live discharge.

METHODS

This was a retrospective cohort study of community-dwelling seniors who were admitted to a medical service at Thunder Bay Regional Health Sciences Centre. We conducted regression analyses adjusted for demographic factors, prior health care utilization, and factors associated with the index admission to determine the association between family physician status and the study outcomes.

RESULTS

Among the 12 033 seniors admitted to hospital between Apr. 1, 2004, and Mar. 31, 2013, 40.7% lacked a family physician. Among those without a family physician, 8.0% (390/4899) died during the index admission and 15.8% (714/4509) died in the subsequent year. Adjusted regression models showed that not having a family physician was significantly associated with in-hospital mortality (odds ratio 1.56, 95% confidence interval [CI] 1.33-1.83). Regression models of all-cause 1-year mortality and readmission also suggested that lack of a family physician was associated with detrimental health outcomes (hazard ratio 1.14, 95% CI 1.04-1.26; subdistribution hazard ratio 1.17, 95% CI 1.10-1.24, respectively).

INTERPRETATION

Elders without family physicians were disadvantaged during their hospital admission as well as in the subsequent year. Additional interventions aimed at increasing the proportion of seniors admitted to hospital who are connected with a family physician are warranted.

Predicting ICU readmission using grouped physiological and medication trends

BACKGROUND

Patients who are readmitted to an intensive care unit (ICU) usually have a high risk of mortality and an increased length of stay. ICU readmission risk prediction may help physicians to re-evaluate the patient's physical conditions before patients are discharged and avoid preventable readmissions. ICU readmission prediction models are often built based on physiological variables. Intuitively, snapshot measurements, especially the last measurements, are effective predictors that are widely used by researchers. However, methods that only use snapshot measurements neglect predictive information contained in the trends of physiological and medication variables. Mean, maximum or minimum values take multiple time points into account and capture their summary statistics, however, these statistics are not able to catch the detailed picture of temporal trends. In this study, we find strong predictors with ability of capturing detailed temporal trends of variables for 30-day readmission risk and build prediction models with high accuracy.

METHODS

We study physiological measurements and medications from the Multiparameter Intelligent Monitoring in Intensive Care II (MIMIC-II) clinical dataset. Time series of each variable are converted into trend graphs with nodes being discretized measurements of each variable. Then we extract important temporal trends by applying frequent subgraph mining on the trend graphs. The frequency of a subgraph is a good cue to find important temporal trends since similar patients often share similar trends regarding their pathophysiological evolution under medical interventions. Important temporal trends are then grouped automatically by non-negative matrix factorization. The grouped trends could be considered as an approximate representation of patients' pathophysiological states and medication profiles. We train a logistic regression model to predict 30-day ICU readmission risk based on snapshot measurements, grouped physiological trends and medication trends.

RESULTS

Our dataset consists of 1170 patients who are alive 30 days after discharge from ICU and have at least 12 h of data. In the dataset, 860 patients were not readmitted and 310 were readmitted, within 30 days after discharge. Our model outperforms all comparison models, and shows an improvement in the area under the receiver operating characteristic curve (AUC) of almost 4% from the best comparison model.

CONCLUSIONS

Grouped physiological and medication trends carry predictive information for ICU readmission risk. In order to build predictive models with higher accuracy, we should add grouped physiological and medication trends as complementary features to snapshot measurements.

Iatrogenic risk factors associated with hospital readmission of elderly patients: A matched case-control study using a clinical data warehouse

WHAT IS KNOWN

Hospital readmission within 30 days of patient discharge has become a standard to judge the quality of hospitalizations. It is estimated that 14% of the elderly, people over 75 years old or those over 65 with comorbidities, are at risk of readmission, of which 23% are avoidable. It may be possible to identify elderly patients at risk of readmission and implement steps to reduce avoidable readmissions.

OBJECTIVE

The aim of this study was to identify iatrogenic risk factors for readmission. The secondary objective was to evaluate the rate of drug-related readmissions (DRRs) among all readmissions and compare it to the rate of readmissions for other reasons.

METHODS

We conducted a retrospective, matched, case-control study to identify non-demographic risk factors for avoidable readmission, specifically DRRs. The study included patients hospitalized between 1 September 2014 and 31 October 2015 in an 800-bed university hospital. We included patients aged 75 and over. Cases consisted of patients readmitted to the emergency department within 30 days of initial discharge. Controls did not return to the emergency department within 30 days. Cases and controls were matched on sex and age because they are known as readmissions risk factors. After comparison of the mean or percentage between cases and controls for each variable, we conducted a conditional logistic regression.

RESULTS

The risk factors identified were an emergency admission at the index hospitalization, returning home after discharge, a history of unplanned readmissions and prescription of nervous system drugs. Otherwise, 11.4% of the readmissions were DRRs, of which 30% were caused by an overdose of antihypertensive. The number of drugs at readmission was higher, and potentially inappropriate medications were more widely prescribed for DRRs than for readmissions for other reasons.

WHAT IS NEW AND CONCLUSION

In this matched case-control retrospective study, after controlling for gender and age, we identified the typical profile of elderly patients at risk of readmission. These patients had an unplanned admission at the index hospitalization and prescribed nervous system drugs at discharge from the index admission; they have a history of unplanned readmission within 30 days and return home after discharge.

Impact of an integrated medicines management service on preventable medicines-related readmission to hospital: a descriptive study

Background

Medication contributes to 5-20% of hospital admissions, of which half are considered preventable. An integrated medicines management service (IMMS) was developed at a large general hospital in London to identify and manage patients at risk of a preventable medicines-related readmission (PMRR) to reduce the risk of PMRR.

Objective

To investigate the effect of the pharmacy IMMS on the rate of PMRR within 30 days of the first discharge.

Method

744 patients were identified between October 2008 and October 2014, using the PREVENT tool. Patients at risk were managed by the IMMS with medication reconciliation, review, consultation and follow-up, as required.

Results

Of 744 patients, 119 were readmitted within 30 days of discharge, with a PMRR for 2 patients (1.7%). The main reason for referral to the service was to assess the need to start a compliance aid. Most interventions involved communication: 84% included patient consultations with 50% involving discussion with the patient's community pharmacist and 32% with their general practitioner surgery.

Conclusions

An IMMS may be an effective method of reducing the rate of PMRR. Further work is needed to establish the cost-effectiveness of the service.

Hospital Readmission Through the Emergency Department

BACKGROUND

Hospital readmission places a high burden on both health care systems and patients. Most readmissions are thought to be related to the quality of the health care system.

OBJECTIVES

The aim of this study was to examine the causes and rates of early readmission in emergency department in a Tehran hospital.

PATIENTS AND METHODS

A cross-sectional investigation was performed to study readmission of inpatients at a large academic hospital in Tehran, Iran. Patients admitted to hospital from July 1, 2014 to December 30, 2014 via the emergency department were enrolled. Descriptive statistics were used to summarize the distribution demographics in the sample. Data was analyzed by chi2 test using SPSS 20 software.

RESULTS

The main cause of readmission was complications related to surgical procedures (31.0%). Discharge from hospital based on patient request at the patient's own risk was a risk factor for emergency readmission in 8.5%, a very small number were readmitted after complete treatment (0.6%). The only direct complication of treatment was infection (17%).

CONCLUSIONS

Postoperative complications increase the probability of patients returning to hospital. Physicians, nurses, etc., should focus on these specific patient populations to minimize the risk of postoperative complications. Future studies should assess the relative connections of various types of patient information (e.g., social and psychosocial factors) to readmission risk prediction by comparing the performance of models with and without this information in a specific population.

Clustering of elderly patient subgroups to identify medication-related readmission risks

INTRODUCTION

High Risk Medication Regimen (HRMR) scores are weakly predictive of hospital readmissions for elderly home health care patients. HRMR is composed of three elements related to drug risks: polypharmacy (number of medications); Potentially Inappropriate Medications (PIM) known to be harmful to the elderly; and the Medication Regimen Complexity Index (MRCI) that weighs drugs by the complexity of their dosing and instructions. In this paper, we hypothesized that HRMR scores are more predictive for demographic subgroups of elderly patients. The study used Outcome and Assessment Information Set (OASIS) variables to identify subgroups of patients for whom the HRMR measures appeared more predictive for hospital readmissions.

METHODS

OASIS and medication data were reused from a study of 911 patients (355 males, 556 females; mean age 78.9) from 15 Medicare-certified home health care agencies that established the relationship between HRMR and hospital readmissions. Hierarchical agglomerative clustering using the Jaccard distance measure and average-link method identified patient subgroups based on the OASIS data. Receiver operating curve (ROC) analyses evaluated the predictive strength of the HRMR variables for each subgroup. Additional False Discovery Rate (FDR) analyses assessed whether the clustered relationships were chance.

RESULTS

Clustering of OASIS data for 911 patients identified six subgroups: patients with Good Functional Status (n=382); Females with Moderate to Severe Pain (n=354); patients with poor prognosis needing functional status assistance (n=419); patients with Poor Functional Status (n=287); Males with Adult Children as Caregiver (n=198); adults living alone with spouses as primary caregiver (n=127). ROC results relating these subgroups to HRMR risks were strongest for Males with Adult Children as Caregivers (AUC: polypharmacy, 0.73; PIM, 0.64; MRCI, 0.77). The findings for this subgroup also met the FDR analysis threshold (<=0.20).

CONCLUSIONS

A risk of medication-related readmissions in elderly men with adult children as caregivers is consistent with research showing problems in medication adherence when seniors are supported by informal caregivers. The results from clustering analysis present a hypothesis for research on HRMR and on the relationship between adult caregivers and their fathers.

Development, Validation and Deployment of a Real Time 30 Day Hospital Readmission Risk Assessment Tool in the Maine Healthcare Information Exchange

Objectives

Identifying patients at risk of a 30-day readmission can help providers design interventions, and provide targeted care to improve clinical effectiveness. This study developed a risk model to predict a 30-day inpatient hospital readmission for patients in Maine, across all payers, all diseases and all demographic groups.

Methods

Our objective was to develop a model to determine the risk for inpatient hospital readmission within 30 days post discharge. All patients within the Maine Health Information Exchange (HIE) system were included. The model was retrospectively developed on inpatient encounters between January 1, 2012 to December 31, 2012 from 24 randomly chosen hospitals, and then prospectively validated on inpatient encounters from January 1, 2013 to December 31, 2013 using all HIE patients.

Results

A risk assessment tool partitioned the entire HIE population into subgroups that corresponded to probability of hospital readmission as determined by a corresponding positive predictive value (PPV). An overall model c-statistic of 0.72 was achieved. The total 30-day readmission rates in low (score of 0–30), intermediate (score of 30–70) and high (score of 70–100) risk groupings were 8.67%, 24.10% and 74.10%, respectively. A time to event analysis revealed the higher risk groups readmitted to a hospital earlier than the lower risk groups. Six high-risk patient subgroup patterns were revealed through unsupervised clustering. Our model was successfully integrated into the statewide HIE to identify patient readmission risk upon admission and daily during hospitalization or for 30 days subsequently, providing daily risk score updates.

Conclusions

The risk model was validated as an effective tool for predicting 30-day readmissions for patients across all payer, disease and demographic groups within the Maine HIE. Exposing the key clinical, demographic and utilization profiles driving each patient’s risk of readmission score may be useful to providers in developing individualized post discharge care plans.

Development of an automated model to predict the risk of elderly emergency medical admissions within a month following an index hospital visit: A Hong Kong experience

Objectives:

To develop an automated risk prediction model to identify elderly patients at high risk of emergency admission to medical wards within 28 days following an index hospital visit.

Methods:

A retrospective data analysis of 41 hospitals and 48 specialist outpatient clinics in Hong Kong. The study subjects were elderly patients aged 65 years or above, who had index hospital visit(s) in the year of 2005, which included hospitalizations at medical wards and attendances at the accident and emergency departments or specialist outpatient clinics for medical conditions. Multiple logistic regression was used to estimate the risk of emergency medical admission in 28 days after an index hospital visit. Model validation was performed against the complete cohort in 2006.

Results:

Over a million of episodes were included in the derivation cohort. A total of 14 predictor variables included patient socio-demographics, service utilization in the previous year, presence and number of chronic diseases and type of index episode. The model has a good discriminative ability with the area under receiver-operating characteristic curve at 0.819 and 0.824 for the derivation and validation cohorts, respectively. The model has a sensitivity of 70.3 per cent, specificity of 78.4 per cent, positive predictive value of 21.7 per cent and negative predictive value of 96.9 per cent.

Conclusion:

This simple, accurate and objective risk prediction model has been computerized into an automated screening tool to recruit high-risk elderly patients discharged from all public hospitals in Hong Kong into the Community Health Call Centre service with an aim to prevent avoidable hospitalizations.

Integrated medicines management - can routine implementation improve quality?

RATIONALE, AIMS AND OBJECTIVES

Previous service development work in the area of integrated medicines management (IMM) has demonstrated clear quality improvements in a targeted group of patients within a hospital in Northern Ireland. In order to determine whether this programme could be transferable to routine practice and thereby assess its generalizability, research has been carried out to quantify the health care benefits of incorporating the concept of IMM as routine clinical practice.

METHOD

The IMM programme of care was delivered to all eligible patients (subject to inclusion criteria) across two hospital sites in Northern Ireland during normal pharmacy opening hours. All patients were followed up for a period of 12 months from their time of hospital admission. All patient data were collected using the custom-designed Electronic Pharmacist Intervention Clinical System at each stage of their hospital journey, that is, admission, inpatient stay and discharge.

RESULTS

Patients who received the IMM service benefited from a reduced length of hospital stay on their reference admission (1.42 days; P = 0.020) as well as a reduced length of stay during the first rehospitalization (5.86 days; P = 0.013). There was also a trend of a reduced number of readmissions and a longer time to readmission during the 12-month follow-up period. Potential significant opportunity cost savings were demonstrated as well as a significant improvement in medication appropriateness (discharge vs. reference admission).

CONCLUSIONS

The IMM programme of care has proven to be transferable to routine hospital care within two hospitals in Northern Ireland. It is anticipated that this current research will further inform the development of IMM as routine clinical practice across Northern Ireland and beyond.

Physician perspective on propoxyphene as a potentially inappropriate medication in Tennessee

Medicare Part D data from the Quality Improvement Organization's 9th Statement of Work drug safety indicator project under the direction of the Centers for Medicare & Medicaid Services define the potentially inappropriate medications (PIMs) list for Tennessee. These data reveal propoxyphene as the main contributor to the state's PIM rate. In Tennessee, PIM and drug-drug interaction (DDI) rates indicate propoxyphene as the most prescribed medication among elderly patients despite decades of attention for potentially adverse effects. During this project, physicians agreed that PIM rates are too high, but disagreed in approach preference, i.e., administrative limits and bans versus a proactive educational approach. Physicians were interested in participating in quality improvement by using individual pharmacy data to influence prescribing patterns. Exploring alternatives in research and survey, a potential and reachable point of intervention was found, a prescribing paradigm proposed by researchers to improve outcomes by reducing adverse effects in minimizing PIMs and DDIs.

Risk factors for hospital readmissions in elderly patients: a systematic review

Population ageing is associated with an increase in hospital admissions. Defining the factors that affect the risk of hospital readmission could identify individuals at high risk and enable targeted interventions to be designed. This aim of this study was to identify the risk factors for hospital readmission in elderly people. A systematic review of the literature published in English or Spanish was performed by electronically searching EMBASE, MEDLINE, CINAHL, SCI and SSCI. Some keywords were aged, elder, readmission, risk, etc. Selection criteria were: prospective cohort studies with suitable statistical analysis such as logistic regression, that explored the relationship between the risk of readmission with clinical, socio-demographic or other factors in elderly patients (aged at least 75 years) admitted to hospital. Studies that fulfilled these criteria were reviewed and data were extracted by two reviewers. We assessed the methodological quality of the studies and prepared a narrative synthesis. We included 12 studies: 11 were selected from 1392 articles identified from the electronic search and one additional reference was selected by manual review. Socio-demographic factors were only explanatory in a few models, while prior admissions and duration of hospital stay were frequently relevant factors in others. Morbidity and functional disability were the most common risk factors. The results demonstrate the need for increased vigilance of elderly patients who are admitted to hospital with specific characteristics that include previous hospital admissions, duration of hospital stay, morbidity and functional disability.

[Factors associated with hospital readmissions in the elderly]

OBJECTIVE

To identify factors associated with hospital readmissions in the elderly.

DESIGN

Observational descriptive study.

SETTING

Health area 3 of Madrid.

PARTICIPANTS

Patients 74 years-old and over with a hospital admission to Hospital Universitario Príncipe de Asturias in 2006. Those with a hospital admission in the previous 6 months were excluded. A total of 1051 patients were identified.

MAIN MEASURES

Hospital Discharge Minimum Basic Data Set and primary care information system were used to develop logistical regression models. The dependent variable was the hospital readmission in a 6 month period. Independent variables were socio-demographics, health status and health care activity.

RESULTS

There were 22.6% readmissions in the first 6 months. Variables associated with higher risk of readmission were, hospital stay (hospital stay greater than 15 days had an OR: 1.73 (95% CI:1.17-2.54), the total number of medicines prescribed to the patient (OR: 1.05; 95% CI:1.01-1.09), having hypertension (OR:1.56; 95% CI:1.11-2.18), heart failure (OR: 1.56; 95% CI:1.00-2.44) or ischemic heart disease (OR: 1.51; 95% CI:1.00-2.26), and the primary care nursing mean attendance pressure (OR: 0.93; 95% CI:0.87-0.98). The model that integrates information from both systems explains a higher number of factors associated with readmission.

CONCLUSIONS

Hospital readmissions have been associated to patient medical condition and to factors related to the health care received. Integrating information from hospital and primary care administrative data bases improves the capacity to identify factors associated with a higher readmission risk.

Benzodiazepines and injury: a risk adjusted model

BACKGROUND

Benzodiazepines (BZD) are one class of medications that are generally acknowledged to be a risk factor for injuries.

OBJECTIVE

Our objective was to link outpatient prescription data with clinical data in order to develop a risk adjusted binary model that associates BZD usage with the risk for a healthcare encounter for an injury.

METHODS

In total, 3 years of outpatient BZD prescription data, totaling 133 872 outpatient BZD prescriptions for 13 745 patients for a VA medical center, were combined with data from inpatient and outpatient administrative databases. The model incorporated Elixhauser comorbidity measures with 1-year look back period, along with hospital discharges, marital status, age, mean arterial pressure and body mass index. The model also included the dose of the drug, converted to valium equivalents and its duration. The model was analyzed using generalized estimation equations (GEE).

RESULTS

Dose, duration, discharges and various comorbidities were associated with an increased risk for injury, while being married reduced the risk. Increased body mass was associated with increased injury risk. Increased mean arterial pressure was associated with decreased risk.

CONCLUSIONS

These findings offer guidance on how specific combinations of risk factors and potential protective effects may impact accidental injury risk. Clinicians prescribing or adjusting BZDs can use these results to more accurately tailor medication regimens for a patient. Our findings suggest that clinicians should also consider the nature of the social support system available to the patient in assessing total injury risk.

Effect of Concomitant Use of Benzodiazepines and Other Drugs on the Risk of Injury in a Veterans Population

BACKGROUND

Benzodiazepines comprise a class of drugs that when used as monotherapy are generally acknowledged to pose a risk for injury by increasing the likelihood of falls, fall-related injuries, adverse drug events and car accidents. Benzodiazepines may also be used concomitantly with other high risk medications that may further exacerbate the risk of injury. The aim of this study is to examine the occurrence of the concomitant use of benzodiazepines and other drugs and then quantify the indirect effect of these drug combinations on the likelihood of an injury-related health care episode.

METHODS

A multivariate model was specified that included outpatient prescription data and inpatient/outpatient medical utilisation records for 13,745 patients at a Veterans Administration hospital system over a 3-year period (1999-2001). We analysed 1,33,872 outpatient benzodiazepine prescriptions and >1.5 million non-benzodiazepine prescriptions for the study population. Micromedex software was used to identify combinations of benzodiazepines and other drugs that are likely to result in 'major' interactions. We then further restricted our focus to the use of these drug combinations within a 30-day period prior to an injury-related medical event. The adjusted odds ratio on a variable characterising concomitant use of a benzodiazepine and another drug within this period was used to quantify the relative risk of injury. The principal outcome was the estimated risk of an injury-related health care episode within a 30-day period when taking both a benzodiazepine and another drug with a 'major' severity rating as defined by Micromedex. The risk of injury was adjusted for comorbidities, hospital discharges, marital status, age, mean arterial pressure and body mass index, as well as the dose of benzodiazepine (converted to diazepam equivalents) and duration of benzodiazepine treatment.

RESULTS

Of the 1,110 unique individuals who experienced an injury, 790 (71.2%) patients had used a benzodiazepine in combination with another drug. Furthermore, only 4.3% (320/7522) of the patients taking benzodiazepines who did not have concomitant drug use experienced an injury. The occurrence of this concomitant use increased the odds of an injury >2-fold in the model. Dose and duration of benzodiazepine use, as well as certain comorbidities, were also associated with an increased risk for injury, whereas being married reduced the risk.

CONCLUSIONS

This is the first large-scale study to quantify the impact of concomitant use of benzodiazepines and other drugs on the risk of injury in a population of Veterans Administration patients. It demonstrates the utility of expanding the focus of inappropriate medication usage to include analyses that link potentially inappropriate drug use with health care utilisation for injuries.