Association between medication-related adverse events and non-elective readmission in acute ischemic stroke

Despite improvements in population health contributing to decreasing stroke incidence, mortality, and age-adjusted hospitalization rates in the United States (US) over the last decade [1‐3], stroke remains a leading cause of death, hospitalization, and healthcare expenditure [1, 4]. Estimates suggest that approximately 795,000 individuals experience a new or recurrent stroke each year in the US [1], and that readmissions following hospitalizations for stroke are relatively common and often occur for potentially avoidable causes, including urinary tract infections, uncontrolled diabetes, and pneumonia [4‐6].

Hospital readmission is considered to be a useful indicator of quality of healthcare services [7‐9], which has prompted some governments and insurers to implement hospital reimbursement schedules that are dependent on short-term hospital readmission rates [10, 11]. In effort to better understand factors contributing to readmission after stroke and identify populations at greatest risk of re-hospitalization, a number of studies have examined whether patient and care setting factors are associated with post-stroke readmission [5, 12‐15]. Differences in study design, exposure and outcome defintions, and statisitical analyses across studies have led to inconsistencies in reported results [8]. Nonetheless, length of index admission [6, 16, 17], discharge disposition [5], and stroke severity [13, 18] have been identified as consistent predictors of post-stroke readmission and may be essential measures in the development of risk-standardized readmission models for stroke.

Previously, a large multi-center study of stroke readmission in Australia found that experiencing an adverse event or severe complication, such as recurrent stroke, penumonia, urinary tract infection, or fall, during hospitalization was significantly associated with readmission within 28 days (adjusted odds ratio (AOR), 2.81; 95% confidence interval (CI): 1.55–5.12) [19]. Based on these findings, it is important to examine whether a similar relationship exists in the US, and whether the reported association between inpatient adverse events and acute readmission extends to medication-related adverse events. To enhance our understanding of medication-related adverse events among stroke inpatients, we examined acute ischemic stroke (AIS) hospitalizations and readmissions using the US Healthcare Cost and Utilization Project’s (HCUP) 2014 Nationwide Readmissions Database (NRD). Our primary objectives were to characterize reasons for readmission within 30 days of discharge and determine if medication-related adverse events occuring during AIS hospitalization were associated with acute readmission. Our secondary objectives were to examine whether demographic, clinical, and hospital characteristics were associated with post-AIS readmission.

We analyzed nationally representative administrative claims from 439,682 individuals hospitalized with AIS between January and November 2014 in the US in order to characterize reasons for acute readmission and determine whether inpatient medication-related adverse events and select characteristics independently predicted readmission. Primary findings indicated that 10.7% of individuals were readmitted within 30 days of AIS discharge, that readmissions were primarily for reasons frequently observed in older adult populations, and that experiencing a medication-related adverse event greatly increased the likelihood of being readmitted within 30 days for reasons other than AIS. Our secondary findings showed that: (1) several factors, including, but not limited to, younger age, public health insurance, leaving the hospital against medical advice, and multimorbidity, were associated with an increased odds of all-cause acute readmission, and (2) that the eldest individuals and those receiving care at non-metropolitan hospitals were less likely to be readmitted following inpatient care for AIS.

Dramatic reductions in stroke incidence and related mortality have been observed in the US during the last decade and the age-adjusted AIS hospitalization rate decreased by 18.4% between 2000 and 2010 [2], with improvements being similar across sex and race [3]. Nevertheless, stroke remains a leading national cause of hospitalization and patient disability. Readmission following AIS hospitalization is common, with a large study of Medicare beneficiaries hospitalized for AIS in 2005–2006 reporting that 14.4% of patients were readmitted within 30 days of discharge and that nearly 12% of all readmission were for preventable causes [4]. During the year following AIS discharge, rates of readmission are estimated to be as high as 27% [14]. Our finding that 10.7% of patients with AIS were readmitted within 30 days of discharge suggests that there has been little overall improvement to the national rate of post-stroke acute readmission in recent years. Acute readmissions following AIS hospitalization in the US are almost always (~ 90%) unplanned, are associated with more than $17 billion of annual healthcare spending, and in some circumstances may serve as an indicator of the quality of healthcare services received [5]. Although not all unplanned readmissions are preventable, knowledge of modifiable and non-modifiable risk factors associated with readmission provides the opportunity to develop validated readmission risk prediction tools to identify patients at greatest risk of acute readmission. Ideally, such tools could help target the delivery of health services, including enhanced discharge planning, to those found to be at great risk of readmission. In turn, this may improve individual health outcomes and lead to more efficient healthcare spending [26].

Medication-related adverse events may include side effects to medications taken as prescribed, accidental overdosing by patients, and medication errors [22]. Such events contribute to acute readmission and up to 50% of medication-related adverse events occurring post-discharge are considered preventable [27, 28]. Recently, a study of more than 500 readmissions to a large academic hospital in the US found that approximately 13% of all 30-day preventable readmissions were attributable to medication-related adverse events [27]. Of readmissions classified as preventable, one half were the result of prescribing errors, while the other half resulted from insufficient patient monitoring or education [27]. These findings highlight potential population and health system benefits that may result from interventions aimed at mitigating the risk of experiencing medication-related adverse events.

To date, studies of risk factors for readmission following inpatient stroke care have largely focused on patient or stroke-related factors, including, but not limited to, age, sex, socioeconomic status, place of residence, medical history, and the National Institutes of Health Stroke Scale/Score [4‐6, 17]. Similar to our findings, prior investigations have consistently identified recurrent stroke, infections, and cardiac conditions as primary causes of acute readmission in AIS, and longer length of index AIS hospitalization and increasing morbidity to be independently associated with readmission [4, 17, 29]. Other reported patient and stroke-related risk factors for short- and long-term readmission after stroke include prior coronary heart disease, heart failure, and diabetes, as well as having a feeding tube or urinary catheter [17]. Although we have confirmed prior reports that patient and geographic disparities in AIS exist and contribute to observed undesirable health outcomes such as acute readmission, characteristics such as individual age, sex, socioeconomic status, and place of residence are unlikely to be modified. At best, knowledge of these particular disparities may provide baseline data for future population-based research into other patient-level risk factors for readmission after AIS.

The path from patient and geographic disparity to timely, effective, and sustainable intervention is presumed to be very long and uncertain. In consideration of these challenges, our study focused on the association between medication-related adverse events in the inpatient setting and acute readmission after AIS discharge, since medication-related adverse events are potentially modifiable events that may affect all patients. Our research adds to the growing body of literature pertaining to the role of hospital environments, both physical and social, in contributing to readmission after hospitalization for AIS [14, 17, 30, 31]. Specifically, our finding that medication-related adverse events during hospitalization for AIS are associated with acute readmission may in part explain readmission risks observed among patients with AIS treated by hospitalists (compared to non-hospitalists; hazard ratio, 1.30; 95% CI: 1.11–1.52) [14] and those receiving care at hospitals with higher use of hospice (compared to non-use; OR, 5.86; 95% CI: 1.13–30.3) [30]. Moreover, our findings provide a clinically important contribution to knowledge of potentially avoidable outcomes related to modifiable events occurring within the hospital environment, and inspire new questions pertaining to inpatient medication monitoring and drug safety, areas that are often understudied in neurology.

Our study has a number of strengths. We used nationally representative and longitudinal health claims data and a US-validated algorithm to identify inpatients with AIS [20]. As such, it is unlikely that individuals included in our study were admitted primarily for hemorrhagic stroke, transient ischemic attack, traumatic brain injury, or other cerebrovascular conditions. Inpatient medication-related adverse events were coded using a method reported by the Agency for Healthcare Research and Quality to be efficient at classifying drug-related events in HCUP data [22]. In-depth demographic, clinical, and hospital NRD data enabled us to examine the association between various potential risk factors and acute readmission, and to control for several likely sources of confounding in our analyses. Collectively, our study methods permitted an assessment of the national inpatient burden of AIS with high external validity.

Despite these strengths, our study has certain limitations. Individuals receiving AIS care at hospitals outside of their home state were excluded from our study since it is not possible to track longitudinal follow-up across state boundaries. Therefore, the reported number of individuals admitted to hospital during our study period likely underestimates the true burden of AIS. It is also possible that not all medication-related adverse events experienced by inpatients were adequately documented within the source administrative claims data for this study. Notwithstanding this potential bias, misclassification of medication-related adverse event status is presumed to be non-differential and would therefore bias our reported associations towards the null. In addition, it remains possible that extraneous variables that could not be controlled for within our study (such as stroke severity, outpatient medication use, and comorbidities diagnosed prior to index hospitalization) may confound reported associations between examined risk factors and readmission.

In conclusion, using nationally representative health claims data from the US, we found that readmission within 30 days of inpatient care for AIS is common and associated with medication-related adverse events occurring during the index hospitalization. Although study replication is warranted, our initial findings provide compelling evidence that acute readmissions in AIS may be avoidable through implementation of interventions focused on modifiable risk factors within the hospital environment. Future studies are needed to fully elucidate how potentially modifiable factors within the hospital environment (such as medication monitoring activities, pharmacist involvement in patient management, available nursing expertise, and communication methods) influence patient risk of medication-related adverse events and acute readmission. Population-based studies are also required to characterize inpatient medication-related adverse events in AIS by medication class and determine whether such adverse events and associated readmissions are avoidable. Lastly, future studies should examine whether the odds of acute readmission after AIS hospitalization vary by type of medication-related adverse event experienced during the index admission.