Clinical and imaging data at 5 days as a surrogate for 90-day outcome in ischemic stroke.

BACKGROUND AND PURPOSE: A simple, easily measured surrogate outcome measure for use in early treatment trials for acute ischemic stroke therapies would be highly valued. We hypothesized that day-5 NIH stroke scale score (NIHSS) and day-5 diffusion weighted imaging (DWI) volume would predict clinical outcome better than either alone and could be considered as a possible surrogate outcome in early phase acute stroke trials. METHODS: The prospective Acute Stroke Accurate Prediction (ASAP) trial included a prespecified subgroup evaluated for early outcome. Logistic regression analysis was used to assess the prediction of modified Rankin (mRankin) of 0 or 1. RESULTS: A total of 204 subjects completed the substudy, and 116 (57%) had excellent outcome at 3 months. The area under the ROC curve (AUC) for day-5 NIHSS predicting 3-month excellent outcome was 0.84; for DWI volume predicting outcome was 0.76, and for the multivariable model combining both was 0.84. CONCLUSIONS: The results of the early outcome substudy of the ASAP trial suggest that early stroke severity and infarct volume measures are predictive of 3-month excellent outcome. In our data set the DWI volume does not add clinically relevant information in predicting 3-month outcome. Validation of these results is required.

Change in diffusion-weighted imaging infarct volume predicts neurologic outcome at 90 days: results of the Acute Stroke Accurate Prediction (ASAP) trial serial imaging substudy.

BACKGROUND AND PURPOSE: Predictive models of outcome after ischemic stroke have incorporated acute diffusion-weighted MRI (DWI) information with mixed results. We hypothesized that serial measurements of DWI infarct volume would be predictive of functional outcome after ischemic stroke. METHODS: The prospective Acute Stroke Accurate Prediction (ASAP) Study included a prespecified serial imaging subgroup who underwent DWI studies at baseline (<24 hours after symptom onset) and Day 5 (+/-2 days). DWI infarct volumes were calculated using the Analyze software (Rochester, Minn). Clinical outcomes were assessed at 3 months. Univariate and multivariable regression analysis was performed to assess the relationship between change in DWI lesion volume and excellent neurological outcome (modified Rankin Scale 0, 1, and Barthel Index >or=95). RESULTS: In total, 169 cases from the ASAP study had serial DWI scans with a measurable lesion at baseline, follow-up, or both. The median baseline National Institutes of Health Stroke Scale score was 6 (interquartile range, 3 to 13). For each 10 cm(3) of growth in DWI infarct volume, the OR for achieving an excellent outcome by modified Rankin Scale was 0.52 (95% CI, 0.38 to 0.71) and for the Barthel Index was 0.64 (95% CI, 0.51 to 0.79). Adjusting for clinically important covariates, the OR for an excellent modified Rankin Scale outcome was 0.57 (95% CI, 0.37 to 0.88) and excellent Barthel Index outcome was 0.75 (95% CI, 0.56 to 1.01). CONCLUSIONS: Based on these data, the likelihood of achieving an excellent neurological outcome diminishes substantially with growth in DWI infarct volume in the first 5 days after ischemic stroke of mild to moderate severity.

Which hospitals have significantly better or worse than expected mortality rates for acute myocardial infarction patients? Improved risk adjustment with present-at-admission diagnoses.

BACKGROUND: Public reports that compare hospital mortality rates for patients with acute myocardial infarction are commonly used strategies for improving the quality of care delivered to these patients. Fair comparisons of hospital mortality rates require thorough adjustments for differences among patients in baseline mortality risk. This study examines the effect on hospital mortality rate comparisons of improved risk adjustment methods using diagnoses reported as present-at-admission. METHODS AND RESULTS: Logistic regression models and related methods originally used by California to compare hospital mortality rates for patients with acute myocardial infarction are replicated. These results are contrasted with results obtained for the same hospitals by patient-level mortality risk adjustment models using present-at-admission diagnoses, using 3 statistical methods of identifying hospitals with higher or lower than expected mortality: indirect standardization, adjusted odds ratios, and hierarchical models. Models using present-at-admission diagnoses identified substantially fewer hospitals as outliers than did California model A for each of the 3 statistical methods considered. CONCLUSIONS: Large improvements in statistical performance can be achieved with the use of present-at-admission diagnoses to characterize baseline mortality risk. These improvements are important because models with better statistical performance identify different hospitals as having better or worse than expected mortality.

Validation of an acute ischemic stroke model: does diffusion-weighted imaging lesion volume offer a clinically significant improvement in prediction of outcome?

BACKGROUND AND PURPOSE: Prediction models for ischemic stroke outcome have the potential to contribute prognostic information in the clinical and/or research setting. The importance of diffusion-weighted magnetic resonance imaging (DWI) in the prediction of clinical outcome, however, is unclear. The purpose of this study was to combine acute clinical data and DWI lesion volume for ischemic stroke patients to determine whether DWI improves the prediction of clinical outcome. METHODS: Patients (N=382) with baseline DWI data from the Glycine Antagonist In Neuroprotection and citicoline (010 and 018) trials were used to develop the prediction models by multivariable logistic regression. Data from prospectively collected patients (N=266) from the Acute Stroke Accurate Prediction Study were used to externally validate the model equations. The models predicted either full recovery or nursing home-level disability/death, as defined by the National Institutes of Health Stroke Scale, Barthel Index, or modified Rankin Scale. RESULTS: The full-recovery models with DWI lesion volume had areas under the receiver operating characteristic curves (AUCs) of 0.799 to 0.821, and those without DWI lesion volume had AUCs of 0.758 to 0.798. The nursing home-level disability/death models with DWI had AUCs of 0.832 to 0.882, and those without DWI had AUCs of 0.827 to 0.867. All models had mean absolute errors < or =0.4 for calibration. CONCLUSIONS: All 12 models had excellent discrimination and calibration, with 8 of 12 meeting prespecified performance criteria (AUC > or =0.8, mean absolute error < or =0.4). Although DWI lesion volume significantly increased model explanatory power, the magnitude of increase was not large enough to be clinically important.

Present-at-admission diagnoses improved mortality risk adjustment among acute myocardial infarction patients.

OBJECTIVE: Hospital mortality outcomes for acute myocardial infarction (AMI) patients are a focus of quality improvement programs conducted by government agencies. AMI mortality risk-adjustment models using administrative data typically adjust for baseline differences in mortality risk with a limited set of common and definite comorbidities. In this study, we present an AMI mortality risk-adjustment model that adjusts for comorbid disease and for AMI severity using information from secondary diagnoses reported as present at admission for California hospital patients. STUDY DESIGN AND SETTING: AMI patients were selected from California hospital administrative data for 1996 through 1999 according to criteria used by the California Hospital Outcomes Project Report on Heart Attack Outcomes, a state-mandated public report that compares hospital mortality outcomes. We compared results for the new model to two mortality risk-adjustment models used to assess hospital AMI mortality outcomes by the state of California, and to two other models used in prior research. RESULTS: The model using present-at-admission diagnoses obtained substantially better discrimination between predicted survival and inpatient death than the other models we considered. CONCLUSION: AMI mortality risk-adjustment methods can be meaningfully improved using present-at-admission diagnoses to identify comorbid disease and conditions related closely to AMI.

Present-at-admission diagnoses improve mortality risk adjustment and allow more accurate assessment of the relationship between volume of lung cancer operations and mortality risk.

BACKGROUND: Mortality risk adjustment is a key component of studies that examine the statistical relationship between hospital lung cancer operation volume and in-hospital mortality. Previous studies of this relationship have used different methods of adjusting for factors that influence mortality risk, but none have adjusted for differences in comorbid disease using only diagnoses identified as present-at-admission. METHODS: This study uses adjustments for conditions identified as present-at-admission to examine the statistical relationship between the volume of lung cancer operations and mortality among 14,456 California hospital patients, and compares these results to other methods of risk adjustment similar to those used in previous studies. RESULTS: Mortality risk adjustment using present-at-admission diagnoses yielded better discrimination and explained more of the variability in observed deaths. Large increases in hospital procedure volume were associated with much smaller decreases in mortality risk than those estimated using comparable risk-adjustment models. CONCLUSIONS: Present-at-admission diagnoses can be used to improve mortality risk adjustment and may allow a more accurate assessment of the relationship between procedure volume and mortality risk.

Risk adjustment effect on stroke clinical trials.

BACKGROUND AND PURPOSE: The ischemic stroke population is heterogeneous. Even in balanced randomized trials, patient heterogeneity biases estimates of the treatment effect toward no effect when dichotomous end points are used. Risk adjustment statistically addresses some of the heterogeneity and can reduce bias in the treatment effect estimate. The purpose of this study was to estimate the treatment effect of tissue plasminogen activator (tPA) in the National Institute of Neurological Disorders and Stroke (NINDS) tPA data set with and without adjustment for baseline differences. METHODS: Using a prespecified predictive model, we calculated unadjusted and risk-adjusted odds ratios (ORs) for favorable outcome for the Barthel Index, National Institutes of Health Stroke Scale, and Glasgow Outcome Scale for the patients in the NINDS tPA stroke trial. To assess the importance of the difference, a new sample size was calculated through the use of the risk-adjusted analysis. RESULTS: We analyzed 615 subjects. The ORs for the Barthel Index were 1.76 (unadjusted) and 2.04 (adjusted). The National Institutes of Health Stroke Scale and Glasgow Outcome Scale analyses also demonstrated increased ORs after adjustment. The estimated sample size required for the adjusted comparison was 13% smaller than the unadjusted sample. CONCLUSIONS: Risk adjustment in this data set suggests that the true treatment effect was larger than estimated by the unadjusted analysis. Stroke clinical trials should include prospective risk adjustment methodologies.

Predicting major neurological improvement with intravenous recombinant tissue plasminogen activator treatment of stroke.

BACKGROUND AND PURPOSE: In the National Institute of Neurological Disorders and Stroke (NINDS) rt-PA Stroke Study, major neurological improvement within 24 hours (MNI) occurred significantly more frequently with recombinant tissue plasminogen activator (rtPA) treatment than with placebo. We explored the relationship between MNI and 3-month favorable outcome and sought clinical predictors of MNI. METHODS: Data from 312 rtPA-treated patients from the NINDS trial were used to assess the ability of MNI to predict favorable outcome at 3 months as defined by a modified Rankin Scale score of 0 to 1. Next, a multivariable predictive model was developed for MNI within the same data set. Clinical variables examined included age, time to treatment (TTT), diabetes, pretreatment glucose, baseline National Institutes of Health Stroke Scale score, pretreatment blood pressure, history of atrial fibrillation, weight >100 kg, and a dense artery sign. Finally, this model was used to forecast into the placebo group of the NINDS trial to assess the uniqueness of the predictors in the rtPA-treated group. RESULTS: MNI had a positive predictive value and negative predictive value of 0.70 for predicting favorable 3-month outcome. Only age [odds ratio (OR), 0.68; 95% confidence interval (CI), 0.47 to 0.99] and TTT (OR, 0.56; 95% CI, 0.34 to 0.91) appear to be independently associated with MNI. The model performed only moderately well (area under the receiver-operating characteristic curve, 0.66). Age (OR, 0.67; 95% CI, 0.45 to 0.99) but not TTT was associated with MNI in the placebo group. CONCLUSIONS: MNI may be a useful surrogate for thrombolytic activity and is predictive of favorable 3-month outcome. When rates of MNI in different populations of stroke patients treated with thrombolysis are compared, adjustments for age and TTT may be necessary.

Predicting outcome in ischemic stroke: external validation of predictive risk models.

BACKGROUND: Six multivariable models predicting 3-month outcome of acute ischemic stroke have been developed and internally validated previously. The purpose of this study was to externally validate the previous models in an independent data set. SUMMARY OF REPORT: We predicted outcomes for 299 patients with ischemic stroke who received placebo in the National Institute of Neurological Disorders and Stroke rt-PA trial. The model equations used 6 acute clinical variables and head CT infarct volume at 1 week as independent variables and 3-month National Institutes of Health Stroke Scale, Barthel Index, and Glasgow Outcome Scale as dependent variables. Previously developed model equations were used to forecast excellent and devastating outcome for subjects in the placebo tissue plasminogen activator data set. Area under the receiver operator characteristic curve was used to measure discrimination, and calibration charts were used to measure calibration. The validation data set patients were more severely ill (National Institutes of Health Stroke Scale and infarct volume) than the model development subjects. Area under the receiver operator characteristic curves demonstrated remarkably little degradation in the validation data set and ranged from 0.75 to 0.89. Calibration curves showed fair to good calibration. CONCLUSIONS: Our models have demonstrated excellent discrimination and acceptable calibration in an external data set. Development and validation of improved models using variables that are all available acutely are necessary.

Combined clinical and imaging information as an early stroke outcome measure.

BACKGROUND AND PURPOSE: Imaging information has been proposed as a potential surrogate outcome in stroke clinical trials. The purpose of this study was to determine whether an early outcome measure combining clinical and imaging information is better than either alone in predicting 3-month outcome in acute ischemic stroke patients. METHODS: Clinical information (National Institutes of Health Stroke Scale) and imaging information (CT infarct volume), measured at 1 week from 201 patients from the Randomized Trial of Tirilazad Mesylate in Acute Stroke (RANTTAS) study, were used in a multivariable logistic regression analysis to predict excellent and devastating 3-month outcome. The combined models were compared with the infarct volume models and the clinical models. Discrimination, calibration, and change in global model chi-square were assessed. RESULTS: The combined models and models using clinical information alone had areas under the receiver operating characteristic curves that did not differ significantly (probability value = 0.092 to 0.4), ranging from 0.83 to 0.95. The imaging alone models performed less well (P<0.005) and had areas under the receiver operating characteristic curves that ranged from 0.70 to 0.80. CONCLUSIONS: The National Institutes of Health Stroke Scale at 1 week is highly predictive of 3-month outcome in ischemic stroke patients. The addition of 1-week infarct volume does not improve the accuracy of the predictive model.

Hospital discharge abstract data on comorbidity improved the prediction of death among patients hospitalized with aspiration pneumonia.

OBJECTIVE: To use diagnoses reported as present at admission in California hospital discharge abstract data to identify categories of comorbid disease and conditions related to aspiration pneumonia and to assess their association with hospital mortality. STUDY DESIGN AND SETTING: The study population included all persons hospitalized in California from 1996 through 1999, with a principal diagnosis of aspiration pneumonia. Present at admission diagnoses representing comorbid diseases were separated from conditions closely related to aspiration pneumonia by a physician panel through a computer supported Delphi process. Multivariable logistic regression was used to assess the probability of hospital death after adjusting for these patient characteristics. The statistical performance of this method was compared to the performance of two independent methods for measuring comorbid disease. The practical significance of differences in statistical performance was assessed by comparing the estimated effects of age, race, and ethnicity after adjustments using each method. RESULTS: Mortality risk adjustment using present at admission diagnoses resulted in substantially better statistical performance and in different measurements of the adjusted effects of age, race, and ethnicity. CONCLUSION: Reporting present at admission diagnoses in hospital discharge data yields meaningful improvements in hospital mortality risk adjustment.

A multicenter description of intermediate-care patients: comparison with ICU low-risk monitor patients.

STUDY OBJECTIVES: To describe the characteristics and outcomes of patients admitted to intermediate-care areas (ICAs) and to compare them with those of ICU patients who receive monitoring only on day 1 and are at a low risk (i.e., < 10%) for receiving subsequent active life-supporting therapy (i.e., low-risk monitor patients). DESIGN: Nonrandomized, retrospective, cohort study. SETTING: Thirteen US teaching hospitals and 19 nonteaching hospitals. PATIENTS: A consecutive sample of 8,971 patients at 37 ICAs and 5,116 low-risk (i.e., < 10%) monitor patients at 59 ICUs in 32 US hospitals. INTERVENTIONS: None. MEASUREMENTS AND RESULTS: We recorded demographic and clinical characteristics, resource use, and outcomes for the ICA and ICU low-risk monitor patients. Patient data and outcomes for this study were collected concurrently or retrospectively. ICA and ICU low-risk monitor patients were similar in regard to gender, race, and frequency of comorbitities, but ICA patients were significantly (p < 0.001) older, had fewer physiologic abnormalities (mean acute physiology score, 16.7 vs 19.8, respectively), and were more frequently admitted due to nonoperative diagnoses. The mean length of stay for ICA patients was significantly longer (3.9 days) than for ICU low-risk monitor patients (2.6 days; p < 0.001). The hospital mortality rate was significantly higher for ICA patients (3.1%) compared to ICU low-risk monitor patients (2.3%; p = 0.002). CONCLUSIONS: The clinical features of ICA patients are similar, but not identical to, those of less severely ill ICU monitor patients. Comparisons of hospital death rates and lengths of stay for these patients should be adjusted for characteristics that previously have been shown to influence these outcomes.

Comorbid disease and the effect of race and ethnicity on in-hospital mortality from aspiration pneumonia.

BACKGROUND: Racial and ethnic disparities in mortality have been demonstrated in several diseases. African Americans are hospitalized at a significantly higher rate than whites for aspiration pneumonia; however, no studies have investigated racial and ethnic disparities in mortality in this population. OBJECTIVE: To assess the independent effect of race and ethnicity on in-hospital mortality among aspiration pneumonia discharges while comprehensively controlling for comorbid diseases, and to assess whether the prevalence and effects of comorbid illness differed across racial and ethnic categories. DESIGN, SETTING, AND PARTICIPANTS: Retrospective cohort study of 41,581 patients admitted to California hospitals for aspiration pneumonia from 1996 through 1998, using principal and secondary diagnoses present on admission. MEASUREMENT: The primary outcome measure was in-hospital mortality. RESULTS: The adjusted odds of in-hospital death for African-American compared with white discharges [odds ratio (OR)=1.01; 95% confidence interval (CI), 0.91-1.11] was not significantly different. The odds of death for Asian compared with white discharges was significantly lower (OR=0.83; 95% CI, 0.75-0.91). Hispanics had a significantly lower odds of death (OR=0.90; 95% CI, 0.82-0.988) compared to non-Hispanics. Comorbid diseases were more prevalent among African Americans and Asians than whites, and among Hispanics compared to non-Hispanics. Differences in effects of comorbid disease on mortality risk by race and ethnicity were not statistically significant. CONCLUSION: Asians have a lower risk of death, and the risk of death for African Americans is not significantly different from whites in this analysis of aspiration pneumonia discharges. Hispanics have a lower risk of death than non-Hispanics. While there are differences in prevalence of comorbid disease by racial and ethnic category, the effects of comorbid disease on mortality risk do not differ meaningfully by race or ethnicity.

Relationship between 3-month National Institutes of Health Stroke Scale score and dependence in ischemic stroke patients.

BACKGROUND: The National Institutes of Health Stroke Scale (NIHSS) provides a standardized measure of stroke severity and is frequently captured to assess 3-month outcome. Other outcome measures have been assessed for the relationship to dependence; a clinically relevant outcome. The relationship between NIHSS score and functional dependence is unknown. The purpose of this study was to assess the relationship between NIHSS score and accepted measures of dependence in surviving ischemic stroke patients. METHODS: 3-month NIHSS scores were compared to residence and Glasgow Outcome Scale (GOS) scores at 3 months in the Randomized Trial of Tirilazad Mesylate in Patients with Acute Stroke (RANTTAS). For residence, patients who were in a nursing home, chronic hospital or substantially dependent on a caregiver were characterized as 'dependent'. For GOS, a score of 3 (severely disabled) or 4 (vegetative) was characterized as 'dependent'. The sensitivity, specificity and positive (PPV) and negative predictive values (NPV) for various NIHSS score cut points compared to dependence were calculated. Logistic regression analysis was used to assess the association between the NIHSS score and dependence. RESULTS: In 385 subjects from the RANTTAS, an NIHSS score cut point of > or =15 resulted in 100% of subjects identified as being dependent by residence, sensitivity = 24%, specificity = 100%, PPV = 100% and NPV = 80%. Using GOS as the measure of dependence, the results were almost identical. NIHSS was strongly related to dependence with an area under the receiver operating characteristic curve (AUC) = 0.86 for residence and an AUC = 0.94 for GOS. CONCLUSIONS: 3-month NIHSS score is strongly associated with dependence. An NIHSS score of > or =15 at 3 months may be a reasonable estimate of subjects who are highly likely to be dependent at 3 months. These data require validation in an independent data set.

Rehabilitation of children with traumatic brain injury: descriptive analysis of a nationwide sample using the WeeFIM.

OBJECTIVE: To describe functional capability at admission and discharge of children with traumatic brain injury (TBI) in rehabilitation settings. DESIGN: Descriptive analysis. SETTING: Inpatient pediatric rehabilitation hospitals in the United States. PARTICIPANTS: Children (N=3815) in 56 pediatric inpatient rehabilitation facilities who were discharged during 1999 to 2001. INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURES: Admission and discharge WeeFIM scores. RESULTS: Admission and discharge WeeFIM scores correlated positively with age at admission, time from injury to rehabilitation admission, and length of stay (LOS). Higher admission WeeFIM scores correlated with shorter LOS, shorter time from injury to admission to rehabilitation, and higher discharge WeeFIM scores. CONCLUSIONS: Children with TBI demonstrated significant improvement in functional measures during rehabilitation. Discharge function and LOS correlated with admission severity, with children who had higher functional status and shorter time between injury and rehabilitation care having higher discharge function and shorter LOS.

Impact of different measures of comorbid disease on predicted mortality of intensive care unit patients.

BACKGROUND: Valid comparison of patient survival across ICUs requires adjustment for burden of chronic illness. The optimal measure of comorbidity in this setting remains uncertain. OBJECTIVES: To examine the impact of different measures of comorbid disease on predicted mortality for ICU patients. DESIGN: Retrospective cohort study. SUBJECTS: Seventeen thousand eight hundred ninety-three veterans from 17 geographically diverse VA Medical Centers and 43 ICUs were studied, admitted between February 1, 1996 and July 31, 1997. MEASURES: ICD-9-CM codes reflecting comorbid disease from hospital stays before and including the index hospitalization from local VA computer databases were extracted, and three measures of comorbid disease were then compared: (1) an APACHE-weighted comorbidity score using comorbid diseases used in APACHE, (2) a count of conditions described by Elixhauser, and (3) Elixhauser comorbid diseases weighted independently. Univariate analyses and multivariate logistic regression models were used to determine the contribution of each measure to in-hospital mortality predictions. RESULTS: Models using independently weighted Elixhauser comorbidities discriminated better than models using an APACHE-weighted score or a count of Elixhauser comorbidities. Twenty-three and 14 of the Elixhauser conditions were significant univariate and multivariable predictors of in-hospital mortality, respectively. In a multivariable model including all available predictors, comorbidity accounted for less (8.4%) of the model's uniquely attributable chi statistic than laboratory values (67.7%) and diagnosis (17.7%), but more than age (4.0%) and admission source (2.1%). Excluding codes from prior hospitalizations did not adversely affect model performance. CONCLUSIONS: Independently weighted comorbid conditions identified through computerized discharge abstracts can contribute significantly to ICU risk adjustment models.