Majority of 30-Day Readmissions After Intracerebral Hemorrhage Are Related to Infections
Background and Purpose—Infections are common after intracerebral hemorrhage, but little is known about the risk of serious infection requiring readmission after hospital discharge.
Methods—To determine if infections are prevalent in patients readmitted within 30 days of discharge, we performed a retrospective cohort study of patients discharged from nonfederal acute care hospitals in California with a primary diagnosis of intracerebral hemorrhage between 2006 and 2010. We excluded patients who died during the index admission, were discharged against medical advice, or were not California residents. Our main outcome was 30-day unplanned readmission with primary infection–related International Classification of Diseases, Ninth Revision, Clinical Modification code.
Results—There were 24 540 index intracerebral hemorrhage visits from 2006 to 2010. Unplanned readmissions occurred in 14.5% (n=3550) of index patients. Of 3550 readmissions, 777 (22%) had an infection-related primary diagnosis code. When evaluating primary and all secondary diagnosis codes, infection was associated with 1826 (51%) of readmissions. Other common diagnoses associated with readmission included stroke-related codes (n=840, 23.7%) and aspiration pneumonitis (n=154, 4.3%). The most common infection-related primary diagnosis codes were septicemia (n=420, 11.8%), pneumonia (n=124, 3.5%), urinary tract infection (n=141, 4.0%), and gastrointestinal infection (n=42, 1.2%). Patients with a primary infection–related International Classification of Diseases, Ninth Revision, Clinical Modification code on readmission had higher in-hospital mortality compared with other types of readmission (15.6% versus 8.0%, P<0.001). After controlling for other predictors of mortality, primary infection–related readmissions remained associated with in-hospital mortality (relative risk, 1.7; 95% confidence interval, 1.3–2.2).
Conclusions—Infections are associated with a majority of 30-day readmissions after intracerebral hemorrhage and increased mortality. Efforts should be made to reduce infection-related complications after hospital discharge.
Infections are common after both hemorrhagic and ischemic stroke, with ≈30% of patients admitted with intracerebral hemorrhage (ICH) developing infections during their hospital stay.1,2 There are 2 main factors that likely contribute to this high rate of infection. First, physical insults including aspiration, dysphagia, intubation, and urinary catheterization predispose patients to site-specific infections, such as ventilator-associated pneumonias and catheter-associated urinary tract infections. Second, although better studied in patients with ischemic stroke, some patients with ICH undergo a state of poststroke immunosuppression. Poststroke immunosuppression is characterized by activation of the hypothalamus–pituitary–adrenal axis and autonomic dysregulation, which leads to numerous immunologic effects such as lymphopenia and deactivation of lymphocytes and monocytes.3,4 The interaction between these 2 processes results in a high susceptibility to infection. Although the increased risk of infection during initial hospitalization is well known, to date no adequately powered multicenter studies have been performed to examine the risk of serious infection after hospital discharge. We designed this study to assess the hypothesis that infections are associated with a majority of 30-day readmissions to acute care hospitals after ICH.
We performed a retrospective observational cohort study of patients discharged from nonfederal acute care hospitals in California with a primary diagnosis of ICH between 2006 and 2010. Using administrative claims data, we sought to assess the proportion of 30-day readmissions that were either: (1) primary infection–related, defined as an infection-related International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code in the primary diagnosis position or (2) primary or secondary infection related, defined as an infection-related ICD-9-CM code in the primary diagnosis code position or in any of the subsequent 24 secondary diagnosis code positions listed in the California Statewide Inpatient Database. Details about the California Statewide Inpatient Database and Healthcare Cost and Utilization Project have been previously reported.5 Per policy of the New York University School of Medicine Institutional Review Board, this analysis of deidentified, publicly available data did not require review.
We identified all patients with first admission to an acute care hospital for ICH using primary discharge diagnosis code 431 from the ICD-9-CM. This algorithm for identification of patients with ICH from administrative data sets has high sensitivity and specificity with a positive predictive value of 89%.6 To avoid analysis of chronic cases, only the first admission with ICH was included as an index case. In addition, to ensure that the study sample was for initial ICH, we used data from 2005 as a 12-month run-in period to ensure none of the cases had previous ICH admissions. Other exclusion criteria included in-hospital death during index admission, discharge against medical advice, and non-California residency.
Study Outcomes and Measurements
The primary outcome was readmission to an acute care hospital within 30 days after index discharge with a primary discharge diagnosis code (DX1) of an infectious disease. Secondary outcome was readmission to an acute care hospital within 30 days after index discharge with any infectious disease discharge diagnosis code (the California Statewide Inpatient Database includes up to 25 discharge diagnosis codes DX1–DX25). Clinical Classification Software (CCS) categorization of ICD-9-CM codes developed by the Healthcare Cost and Utilization Project was used for cause of readmission (Table I in the online-only Data Supplement). Infections related to tuberculosis, human immunodeficiency virus/sexually transmitted diseases, and hepatitis were excluded as these are chronic conditions. Patients with likely planned readmission for diagnostic or therapeutic procedures were not included as readmissions (eg, cranioplasty and angiography/embolization; see Table II in the online-only Data Supplement for ICD-9-CM codes defined as planned readmissions). Transfers between acute care hospitals were identified using admission source data and were excluded as readmissions. Readmissions to an acute care hospital from residential, skilled nursing, and immediate care facilities were treated as readmissions. Patients with readmissions to acute care hospitals with primary rehabilitation codes were also excluded as readmissions (Table III in the online-only Data Supplement). Follow-up included data from 2011 to ensure 30-day follow-up for all cases identified in 2010.
Univariate analyses were performed using χ2 test, independent samples t test, and nonparametric tests. We used multiple logistic regression to assess risk factors for mortality during readmission. All statistical analyses were performed on commercially available software (SPSS 20, IBM).
There were 24 540 index ICH visits from 2006 to 2010. Infectious disease codes were present in 7741 (31.5%) of index admissions. Unplanned readmissions occurred in 14.5% (n=3550) of index patients. See Table 1 for the most frequent causes of 30-day readmission after ICH. Of 3550 readmission, 777 (22%) had an infectious disease primary diagnosis code. When evaluating readmission diagnosis codes at all 25 positions, infection was associated with 1826 (51%) of readmissions. Other common primary causes for readmission included stroke-related codes (n=840, 23.7%) and aspiration pneumonitis (n=154, 4.3%). The most common infectious disease primary diagnosis codes were septicemia (n=420, 11.8%), pneumonia (n=124, 3.5%), urinary tract infection (n=141, 4.0%), and gastrointestinal infection (n=42, 1.2%; Table 2).
See Table 3 for patient characteristics. Patients with primary infection–related readmissions were older, had higher prevalence of medical comorbidities, and had longer index admission length of stay compared with those without readmission as well as compared with those with readmission who was not primarily infection related.
The time to readmission was slightly longer for primary infection–related readmissions compared with other readmissions (day 11±10 versus day 10±9, P=0.004). Patients with infection-related codes during index admission were more likely to be readmitted within 30 days than those who did not (17.5% versus 13.2%, P<0.001). Patients with infection-related codes during index admission were also more likely to have primary infection–related readmissions compared with those who did not (27.1% versus 18.7%, P<0.001).
Neurosurgical procedures were performed in 3201 patients (13.0%). For patients who underwent neurosurgical procedures, 32.6% of readmissions were primary infection–related compared with 20.0% of readmissions for those who did not (P<0.001). Endotracheal intubation was performed in 4338 patients (17.1%). For patients who underwent endotracheal intubation, 31.9% of readmissions were primary infection–related compared with 19.2% for those who did not (P<0.001).
Patients with primary infection–related readmissions had higher in-hospital mortality compared with other types of readmission (15.6% versus 8.0%, P<0.001). Other predictors of readmission mortality in univariate analyses included age, do not resuscitate status, index visit length of stay, insurance status, and race. After controlling for the other predictors of mortality, primary infection–related readmissions remained associated with in-hospital mortality (odds ratio, 1.7; 95% confidence interval, 1.3–2.2).
Infections are a common inpatient complication after ICH. In this report, we demonstrate that the risk for poststroke infection continues after hospital discharge and may be a significant cause of 30-day readmissions to acute care hospitals. In addition, we demonstrate that infection-related readmissions are associated with a higher mortality than other causes for readmission, in line with our previous report from the ERICH (Ethnic/Racial Variations of Intracerebral Hemorrhage) Study that infection during initial admission is independently associated with worse outcomes, including death.2 The frequency of readmission we demonstrate is similar to those previously reported in a cohort of Medicare patients presenting with ICH (15.5%–16.0%).7 That report, interestingly, did not find a difference in readmission rates between Joint Commission certified primary stroke centers versus noncertified centers.
Our results on the association between 30-day readmission and infection are similar to those reported in single-center studies. Liotta et al8 reported a retrospective cohort of 193 ICH survivors from a large, academic medical center between 2006 and 2012. Of the 22 (11%) patients readmitted within 30 days, they found that 10 (46%) were for infections. Bjerkreim et al9 published an analysis of a cohort of 121 ICH survivors from a regional prospective stroke registry in Norway. Of the 27 (22%) patients readmitted within 90 days, infection was found to be the leading cause of readmission (26%). The fact that our results are similar to these single-center studies, which were able to identify the precise cause of readmission, lends support to the hypothesis that infections are the likely cause for readmissions associated with infection in our study. In addition, the frequency of infection during index hospitalization of ≈30% in this study was similar to our findings in the ERICH Study, which had detailed, adjudicated case, and outcome assessment. This lends validity to our use of diagnostic codes to assess for infection in the post-ICH setting.
Our findings suggest the need for improved post-discharge infection prevention in the outpatient and immediate care setting, including adoption of best practices for tracheostomy and wound care, aspiration precautions, continued mobilization, and other infection prevention measures. Patients with advanced age, medical comorbidities, and longer index admission lengths of stay should specifically be targeted given our finding that they are more likely to be readmitted with infection.
Our results should be interpreted with several limitations in mind. First, our analysis relies on ICD-9-CM codes to identify patients with ICH, and we did not have further clinical detail available. Our study is therefore subject to misclassification bias in identification of cases. However, our method of identifying cases is identical to that of Tirschwell and Longstreth who found a positive predictive value of 89% using this method.6 Second, the sensitivity and specificity of using CCS for determining the cause of readmissions is not known. However, CCS is the method by which the Centers for Medicare and Medicaid Services analyzes readmission discharge diagnoses for their 30-day readmission measures, so our results conform to regulatory practices.10 In addition, although we feel that using the primary diagnosis code is likely more specific and less sensitive than the use of all available secondary diagnosis codes, confirmation of this is out of the scope of this study and both frequencies we found are within the range of existing single-center study data. It should be noted that the sensitivity of secondary diagnosis codes is dependent on the depth of coding performed by individual hospitals and the number of diagnostic codes available in a given database. Finally, the California SID is deidentified and cannot be linked to the Social Security Death Index, so we cannot identify deaths that occur outside of inpatient settings. Therefore, we may be underestimating the burden of post discharge infectious complications if patients died of infection-related complications in the outpatient setting.
Infections are associated with a majority of 30-day readmissions to acute care hospitals in patients with ICH. Efforts should be made to reduce risk of poststroke infections even after hospital discharge.
Sources of Funding
Dr Lord was supported, in part, by the New York University-Health and Hospitals Corporation Clinical and Translational Science Institute (NYU-HHC CSTSI) grant UL1 TR000038 from the National Center for Advancing Translational Sciences, National Institutes of Health (NIH).
Dr Elkind receives compensation for serving as an expert witness for BMS-Sanofi Partnership (modest) related to Plavix and intracerebral hemorrhage and from Hi-Tech Pharmaceuticals (significant) related to 1,3 dimethlyamylamine (DMAA) and intracerebral hemorrhage.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.116.013229/-/DC1.
- Received March 1, 2016.
- Revision received May 5, 2016.
- Accepted May 19, 2016.
- © 2016 American Heart Association, Inc.
- Lord AS,
- Langefeld CD,
- Sekar P,
- Moomaw CJ,
- Badjatia N,
- Vashkevich A,
- et al
- Tirschwell DL,
- Longstreth WT Jr
- Lichtman JH,
- Jones SB,
- Leifheit-Limson EC,
- Wang Y,
- Goldstein LB
- Liotta EM,
- Singh M,
- Kosteva AR,
- Beaumont JL,
- Guth JC,
- Bauer RM,
- et al
- 10.↵Centers for Medicare and Medicaid Services. 2014 Measures updates and specifications report hospital-level 30-day risk-standardized readmission measures. 2014.