Ten-Year Temporal Trends in Medical Complications After Acute Intracerebral Hemorrhage in the United States
Background and Purpose—Data on medical complications after intracerebral hemorrhage (ICH) are sparse. We assessed trends in the prevalence of urinary tract infection, pneumonia, sepsis, deep venous thrombosis (DVT), pulmonary embolism, acute renal failure (ARF), and acute myocardial infarction after ICH in the United States.
Methods—A total of 575 211 adult ICH cases were identified from the 2004 to 2013 Nationwide Inpatient Sample. Weighted complication risks were computed by sex and mechanical ventilation status. Multivariate models were used to evaluate trends in complications and assess their association with in-hospital mortality, cost, and length of stay.
Results—Overall risks of urinary tract infection, pneumonia, sepsis, DVT, pulmonary embolism, ARF, and acute myocardial infarction after ICH were 14.8%, 7.8%, 4.1%, 2.7%, 0.7%, 8.2%, and 2.0%, respectively, but risk differed by sex and mechanical ventilation status. From 2004 to 2013, odds of DVT and ARF increased, whereas odds of pneumonia, sepsis, and mortality declined over time. All complications were associated with >2.5-day increase in length of stay and >$8000 increase in cost. ARF and acute myocardial infarction were associated with increased mortality in all patients; sepsis and pneumonia were associated with increased mortality only in nonmechanical ventilation patients, whereas urinary tract infection and DVT were associated with reduced mortality in all patients.
Conclusions—Despite significant mortality reduction, ARF and DVT risk after ICH have increased, whereas odds of sepsis and pneumonia have declined over the last decade. All complications were associated with increased cost and length of stay, but their associations with mortality were variable, likely due in part to survival bias. Innovative strategies are needed to prevent ICH-associated medical complications.
Intracerebral hemorrhage (ICH) is the most deadly form of stroke1,2 and accounts for 10% of all strokes in the United States.3 Up to 50% of stroke-related mortality is attributable to medical complications.4 Over the last decade, ICH-related mortality in the United States has decreased,5 but data on the current rates of medical complications after ICH outside of clinical trials are sparse.4 The aging population and an increase in persons with multiple chronic diseases may have resulted in greater medical complication during hospitalization; however, many of these are potentially preventable, and advances in medical practice and clinical management guidelines6,7 may have led to decreased frequency. Therefore, the current magnitude and direction of ICH-associated complications are unknown.
The primary aim of this study is to describe current trends in the prevalence of pneumonia, urinary tract infection (UTI), sepsis, deep vein thrombosis (DVT), pulmonary embolism (PE), acute renal failure (ARF), and acute myocardial infarction (AMI) in hospitalized acute ICH patients in the United States from 2004 to 2013. We secondarily evaluated the current association of each complication with other clinical factors, in-hospital mortality, length of stay, and cost.
Data for this study were obtained from the 2004 to 2013 Nationwide Inpatient Sample (NIS). The NIS is the largest all-payer inpatient care database in the United States and comprises a 20% stratified random sample of all US hospital discharges. Sampling weights provided in the NIS allow for calculation of national estimates. Each individual hospital discharge in the NIS is deidentified so all discharges were considered to be independent. Further details on the NIS design are available at http://www.hcup-us.ahrq.gov.
We identified all patients with a primary diagnosis of ICH (n=116 706) by querying the NIS using the International Classification of Disease-Clinical Modification, 9th Revision code 431 (Methods in the online-only Data Supplement).
Definition of Outcomes
All complications were defined using secondary International Classification of Disease, 9th Revision codes or Healthcare Cost and Utilization Project (HCUP) constellation of codes corresponding to these diagnoses (Methods in the online-only Data Supplement). In-hospital mortality, home disposition, and length of stay were studied using HCUP variables named DIED, DISPUNIFORM and LOS, respectively.
Comorbidities associated with ICH hospitalization were used to calculate the modified Charlson comorbidity index for all patients. We defined craniotomy/craniectomy, hydrocephalus requiring external ventricular drain or ventriculoperitoneal shunt, and mechanical ventilation (MV) status using International Classification of Disease, 9th Revision or HCUP procedural codes (Methods in the online-only Data Supplement).
Baseline characteristics of participants were summarized using descriptive statistics. We computed the unadjusted national weighted prevalence of each complication and in subgroups categorized by sex. We further stratified some complications by preexisting conditions or interventions expected to influence their occurrence: pneumonia, DVT, and PE risk by MV status and ARF and MI risk by chronic kidney disease (CKD) and coronary artery disease, respectively. We evaluated trends in prevalence of each complication over time by constructing logistic regression model with each complication as the dependent variable and year of discharge as the independent variable, evaluated continuously, with significance of differences in trend over time assessed using the Wald test.
We used a series of nested logistic regression models to evaluate the association of each complication with hospitalization variables and to assess factors that may account for any observed trends in complication risks and control for confounding (Methods in the online-only Data Supplement).
The 116 706 admissions identified in the NIS represent 575 211 ICH hospitalizations in the United States from 2004 to 2013 (Table II in the online-only Data Supplement). About 49.7% of patients were females, and mean age at admission was 68.9 years. The proportion of patients with comorbid hypertension, diabetes mellitus, CKD, and baseline coagulopathy increased over time (Table II in the online-only Data Supplement; P trend <0.001, not displayed). Most patients had low Charlson comorbidity index scores, but the proportion of patients with Charlson comorbidity index score ≥4 increased from 13.3% to 24.7% over time. MV was administered to 26.0% to 29.6%, of patients of which 16.8% to 25.3% occurred in the first 48 hours of admission (Table II in the online-only Data Supplement). Mean length of stay did not change over time, but the proportion of patients transferred to skilled nursing facilities increased from 42.3% in 2004 to 46.9% in 2013 (P<0.001; Table III in the online-only Data Supplement).
Total Medical Complication Risks
Over the study period, 29.3% of all patients and 37.1% of MV patients had at least 1 medical complication during hospitalization (Table 1). The most common complications were UTI (14.8%), ARF (8.2%), and pneumonia (7.8%), but significant disparity existed by sex and prespecified clinical factors: UTI was the most common complication in females (19.8% versus 9.9% in males), whereas ARF was the most common complication in males (10.6% in males versus 5.9% in females). Overall weighted risks of most other complications, including pneumonia (9.1% versus 6.4), sepsis (4.8% versus 3.2%), DVT (3.2% versus 2.1%), and PE (0.9% versus 0.5%), were ≈30% greater in males compared with females (P value for comparison <0.001), but AMI risks did not differ by sex. In analysis stratified by specific clinical factors, we identified >2-fold higher risks of pneumonia and DVT in ventilated compared with nonventilated patients. Moreover, ARF and AMI risks were greater in patients with CKD and coronary artery disease compared with patients with no coronary artery disease and no-CKD, respectively. Among those with DVT, 10.4% had a PE diagnosis, whereas 39.4% of patients with PE had coexisting DVT.
Overall in-hospital mortality was 23.8%, but death risk in MV patients was 59.5%. Notably, 61.6% of in-hospital deaths occurred in the first 3 days of admission, and 82.4% of all deaths occurred within the first week of hospitalization (Table IV in the online-only Data Supplement). Among those that died within the first week of hospitalization, only 16.4% had at least 1 complication compared with the 59.7% complication risk in those that died after 7 days of hospitalization.
In-hospital mortality decreased from 27.8% in 2004 to 21.0% in 2013 (relative decline 24.4%; Figure 1), but the proportion of patients with at least 1 complication increased from 26.0% to 31.2% over the same period (relative increase 19.7%; Table 1). The increase in total complication risk was driven mainly by a nearly 3-fold increase in ARF risk (4.4% to 12.2%) over the 10-year period (Figure 2; Table 1). DVT risks also increased by 53.4% (2.0% to 3.2%) but mainly in MV patients, and UTI risks increased over time but only in females (Figures 1 and 2; Table 1). The unadjusted weighted prevalence of pneumonia, sepsis, and PE did not change over time, although PE risks in MV patients increased marginally over time (Figures 1 and 2; Table 1).
Multivariate Association of Each Complication With Clinical Factors
After multivariate analyses including adjusting for MV, odds of pneumonia declined by 4% (odds ratio, 0.96; 95% confidence interval, 0.95–0.97) and sepsis by 3% (odds ratio, 0.97%; 95% confidence interval, 0.96–0.98) for each year (Table 2), whereas odds of ARF increased by 9% (odds ratio, 1.09; 95% confidence interval, 1.08–1.10) per year (Table 2).
Some clinical factors were associated with multiple complications. Compared with whites, blacks had higher odds of UTI, pneumonia, and ARF but lower odds of AMI. Atrial fibrillation, dysphagia, high Charlson comorbidity index score, and Medicaid as opposed to Medicare insurance increased the odds of having ≥4 complications. Surprisingly, MV was associated with decreased odds of UTI but increased odds of all other complications, whereas smoking was associated with reduced odds of all complications except AMI (Table 2). Although odds of sepsis and ARF were greater in teaching hospitals compared with rural hospitals, no consistent significant patterns in multivariate-adjusted odds of complications existed by hospital region or yearly stroke volume (Table 2).
Multivariate Association of Complications With Length of Stay, Cost, Home Disposition, and In-Hospital Mortality
All complications were associated with increased length of stay and increased cost (Table V in the online-only Data Supplement, multivariate model 1). Sepsis and DVT were associated with the highest mean increase in length of stay (12.9 and 10.5 days, respectively) and cost ($35 755 and $32 336, respectively; Table V in the online-only Data Supplement, multivariate model 1). After mutually adjusting for other medical complications, the associations of each complication with cost and length of stay were attenuated (Table V in the online-only Data Supplement, multivariate model 2). This suggests that some of the increase in cost and length of stay associated with these complications is mediated by their association with other complications.
Similarly, all complications were associated with reduced odds of home disposition, but their associations with in-hospital death were variable. Whereas ARF and AMI were associated with increased odds of mortality, concomitant UTI diagnosis during ICH admission was associated with decreased odds of death.
When evaluating the associations of pneumonia, sepsis, and PE with mortality, we found significant effect modification of their association with mortality by MV, so mortality associations of these complications were stratified by MV status. Pneumonia and sepsis were associated with increased odds of mortality in nonventilated patients but were associated with decreased odds of mortality in ventilated patients (Table V in the online-only Data Supplement).
In this contemporary analysis of the NIS, we found that 29% of all ICH admissions in the United States have at least 1 medical complication, and any complication risk increased by ≈20% from 2004 to 2013. This increase was driven mainly by a >2-fold relative increase in proportion of ARF, a >50% increase in unadjusted DVT risk, and marginal increase in UTI prevalence. Whereas AMI risk increased only in patients with coronary artery disease and PE risk increased only in MV patients, pneumonia and sepsis risks declined over time. Overall in-hospital mortality after ICH declined by ≈25% from 27% in 2004 to 21% in 2013, but death risk in MV patients remained >50%. Over 60% of all ICH-associated deaths occurred within the first 3 days of hospitalization and over 80% within 7 days.
The high frequency of MV, high mortality in MV patients, and huge percentage of all deaths that occur within the first few days after ICH are consistent with reports of previous studies.8–10 These reflect the high proportion of ICH patients who still present with severe or devastating hemorrhage and emphasize the need for innovative severe ICH prevention and management strategies. However, as more patients survive their ICH hospitalization, emphasis is slowly shifting from survival to improving ICH-related morbidity and optimization of functional recovery. Quantification of disparities and trends in complication burden is a crucial step in ameliorating ICH outcome because it highlights aspects of ICH care where marginal progress has been made and accentuates other areas where additional efforts are needed. Complications such as DVT are potential quality measures for ICH hospitalization, but national reference data have hitherto been lacking. This study provides a robust assessment of the current national patterns of important medical complications after ICH with additional information on relevant population subgroups.
A major finding in this study is the exponential increase in ARF. Whereas the proportion of patients with hypertension, diabetes mellitus, and CKD increased over time and may have contributed to some of the observed increase, risks remained significantly high after adjusting for these conditions, suggesting that the additional factors are responsible. Potential explanations include more aggressive lowering of blood pressure in the acute period of ICH to prevent hematoma expansion4,11 or nephropathy associated with increased use of contrast imaging, but no definite conclusions can be drawn using our retrospective analysis. That aggressive blood pressure control can increase ARF risk was evidenced in the ATACH II trial (Antihypertensive Treatment of Acute Cerebral Hemorrhage II), where adverse renal events occurred in 9.0% of patients in the intensive blood pressure–lowering arm versus 4.0% of patients in the standard-treatment arm.12 Other prospective studies are needed to assess these and other possible etiologic factors.
ICH patients are at particularly high risk for DVT because of their immobility and altered mental status.13 Although DVT and PE risks reported in this study are consistent with those in previous studies,14 rising risks imply that current measures to prevent DVT in ICH patients are far from adequate. Innovative DVT preventive measures are needed particularly in ventilated patients who accounted for all of the observed increase in risks. Current guidelines recommending deferral of subcutaneous anticoagulation for the initial few days may require reappraisal. The observed increase in DVT risk seen in our study may be partly because of increased DVT detection as a result of advancement in imaging technology and improvement in clinical practices, but low frequency of DVT prophylaxis in ICH patients in the United States may also be partly responsible.15 Ventilated patients represent a subgroup of ICH patients with potentially greater morbidity and disability from their ICH, and there may possibly be a growing reluctance to use DVT prophylaxis in this group of patients.16 These factors also need to be evaluated prospectively.
The inverse association of DVT and UTI with mortality may reflect early pickup of nonfatal cases of these conditions. Similarly, MV patients are usually monitored closely in an intensive care unit, thus the negative association of sepsis and pneumonia with mortality in ventilated patients may partly be secondary to early diagnosis and aggressive treatment of mild cases. There could also be some survival bias. MV patients are likely to have the most severe ICH and consequently more likely to die quickly from ICH before they develop complications such as DVT, pneumonia, or sepsis. Moreover, the sickest ICH patients on MV are more likely to have withdrawal of clinical care. In such cases where care is withdrawn, mortality is likely greater, and there could be decreased detection and diagnosis of complications.
Disparities in odds of various complications with respect to race and insurance status may be related to differences in timing of withdrawal of care. Compared with whites, black patients in the United States are more likely to have care withdrawn late and therefore more prone to higher detection of complications.17 Black patients with ICH also have lower white blood cell counts compared with whites and consequently more predisposed to increased risk for infectious complications.18 These factors also need further evaluation.
This study has limitations. Although we relied on previously validated codes for ICH and for complications, we cannot exclude potential inaccuracies because of coding errors. There is great variability in the accuracy of International Classification of Disease, 9th Revision codes for medical complications and in how clinicians actually diagnose pneumonia,19 sepsis, and other complications. Our study of trends in complications is based on the implicit assumption that coding practices remained unchanged over time. However, numerous coding guidelines may have led to the improvement in coding over time. We were unable to provide information on the clinical and radiological severity of ICH by virtue of inherent limitations in our database. However, we used ventilator status, hydrocephalus, dysphagia, and coma as surrogate measures of ICH severity and adjusted for comorbid disease status in all multivariate analyses. A significant proportion of medical complications occur subacutely after ICH admission and still contribute to morbidity and mortality, so true prevalence estimates of complications and mortality are likely higher than reported in this analysis. Thirty-day and 90-day outcome measures may provide more accurate information on complication burden. Our finding of excess cost and length of stay associated with all complications should be interpreted with caution because we were unable to show temporal correlation between complications and these outcomes. Reverse causation is also possible. For example, patients surviving and staying longer in the hospital for other reasons not related to a medical complication may be predisposed to a DVT as opposed to a DVT leading to prolonged hospital stay. We were unable to evaluate the impact of withdrawal of care on medical complication burden because this information is unavailable in the NIS.
Despite these limitations, this study represents the most comprehensive assessment to date of medical complications after ICH. Important strengths of our study include better ICH case ascertainment by use of clinically diagnosed hospitalized ICH. The large sample size of our study with national representation and generalizability to all ICH patients in the United States provides a comprehensive evaluation of the risks, associated determinants, outcomes, and importance of medical complications in ICH patients. Medical complications are frequent and pose significant threats to morbidity and mortality for ICH patients, and our findings should trigger increased efforts to prevent them and to lessen their burden.
Dr Otite had full access to all of the study data and takes responsibility for the integrity and accuracy of the data analysis. Drs Otite and Romano provided study concept and design. All authors involved in acquisition, analysis, and interpretation of data and revised the article critically for important intellectual content. Dr Otite drafted the article. Dr Otite performed statistical analysis. Drs Malik, Chaturvedi, Sacco, and Romano provided administrative, technical, or material support. Dr Romano supervised the study.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.116.015746/-/DC1.
- Received October 13, 2016.
- Revision received January 22, 2017.
- Accepted January 23, 2017.
- © 2017 American Heart Association, Inc.
- Mozaffarian D,
- Benjamin EJ,
- Go AS,
- Arnett DK,
- Blaha MJ,
- Cushman M,
- et al
- Hemphill JC 3rd,
- Greenberg SM,
- Anderson CS,
- Becker K,
- Bendok BR,
- Cushman M,
- et al
- Ovbiagele B
- Guyatt GH,
- Akl EA,
- Crowther M,
- Gutterman DD,
- Schuünemann HJ
- Staykov D,
- Wagner I,
- Volbers B,
- Hauer EM,
- Doerfler A,
- Schwab S,
- et al
- Broderick JP,
- Brott TG,
- Duldner JE,
- Tomsick T,
- Huster G
- Qureshi AI,
- Palesch YY,
- Barsan WG,
- Hanley DF,
- Hsu CY,
- Martin RL,
- et al
- Koivunen RJ,
- Haapaniemi E,
- Satopää J,
- Niemelä M,
- Tatlisumak T,
- Putaala J
- Prabhakaran S,
- Herbers P,
- Khoury J,
- Adeoye O,
- Khatri P,
- Ferioli S,
- et al
- Lord AS,
- Langefeld CD,
- Sekar P,
- Moomaw CJ,
- Badjatia N,
- Vashkevich A,
- et al
- Kishore AK,
- Vail A,
- Chamorro A,
- Garau J,
- Hopkins SJ,
- Di Napoli M,
- et al