Search PubMed⌕ Search

Biomedical subjects

F E Harrell

Publications and source records attributed to F E Harrell.

At least 37 records · Page 2Linked to original sources
PubMed1997

Artificial neural networks improve the accuracy of cancer survival prediction.

BACKGROUND: The TNM staging system originated as a response to the need for an accurate, consistent, universal cancer outcome prediction system. Since the TNM staging system was introduced in the 1950s, new prognostic factors have been identified and new methods for integrating prognostic factors have been developed. This study compares the prediction accuracy of the TNM staging system with that of artificial neural network statistical models. METHODS: For 5-year survival of patients with breast or colorectal carcinoma, the authors compared the TNM staging system's predictive accuracy with that of artificial neural networks (ANN). The area under the receiver operating characteristic curve, as applied to an independent validation data set, was the measure of accuracy. RESULTS: For the American College of Surgeons' Patient Care Evaluation (PCE) data set, using only the TNM variables (tumor size, number of positive regional lymph nodes, and distant metastasis), the artificial neural network's predictions of the 5-year survival of patients with breast carcinoma were significantly more accurate than those of the TNM staging system (TNM, 0.720; ANN, 0.770; P < 0.001). For the National Cancer Institute's Surveillance, Epidemiology, and End Results breast carcinoma data set, using only the TNM variables, the artificial neural network's predictions of 10-year survival were significantly more accurate than those of the TNM staging system (TNM, 0.692; ANN, 0.730; P < 0.01). For the PCE colorectal data set, using only the TNM variables, the artificial neural network's predictions of the 5-year survival of patients with colorectal carcinoma were significantly more accurate than those of the TNM staging system (TNM, 0.737; ANN, 0.815; P < 0.001). Adding commonly collected demographic and anatomic variables to the TNM variables further increased the accuracy of the artificial neural network's predictions of breast carcinoma survival (0.784) and colorectal carcinoma survival (0.869). CONCLUSIONS: Artificial neural networks are significantly more accurate than the TNM staging system when both use the TNM prognostic factors alone. New prognostic factors can be added to artificial neural networks to increase prognostic accuracy further. These results are robust across different data sets and cancer sites.

Breast Neoplasms↗
PubMed1997

A randomized controlled trial of epoprostenol therapy for severe congestive heart failure: The Flolan International Randomized Survival Trial (FIRST).

This trial evaluated the effects of epoprostenol on patients with severe left ventricular failure. Patients with class IIIB/IV congestive heart failure and decreased left ventricular ejection fraction were eligible for enrollment if angiography documented severely compromised hemodynamics while the patient was receiving a regimen of digoxin, diuretics, and an angiotensin-converting enzyme inhibitor. We randomly assigned 471 patients to epoprostenol infusion or standard care. The primary end point was survival; secondary end points were clinical events, congestive heart failure symptoms, distance walked in 6 minutes, and quality-of-life measures. The median dose of epoprostenol was 4.0 ng/kg/min, resulting in a significant increase in cardiac index (1.81 to 2.61 L/min/m2), a decrease in pulmonary capillary wedge pressure (24.5 to 20.0 mm Hg), and a decrease in systemic vascular resistance (20.76 to 12.33 units). The trial was terminated early because of a strong trend toward decreased survival in the patients treated with epoprostenol. Chronic intravenous epoprostenol therapy is not associated with improvement in distance walked, quality of life, or morbid events and is associated with an increased risk of death.

Aged↗
PubMed1997

Selection of thrombolytic therapy for individual patients: development of a clinical model. GUSTO-I Investigators.

We developed a logistic regression model with data from the GUSTO-I trial to predict mortality rate differences in individual patients who received accelerated tissue plasminogen activator (TPA) versus streptokinase treatment for acute myocardial infarction. A nomogram was developed from a reduced version of this model that approximated the underlying risk of patients treated with streptokinase, and thus the benefit of TPA. The 30-day mortality rate with accelerated TPA was 0.063 versus 0.073 with streptokinase and subcutaneously administered heparin and 0.074 with streptokinase and intravenously administered heparin. No baseline patient characteristics were significantly associated with a different relative effect of TPA. Older patients and those with anterior infarction, higher Killip classification (except Killip class IV), lower blood pressure, and increased heart rate had the greatest absolute benefit with accelerated TPA. Patients with acute myocardial infarction who had more high-risk characteristics derived a greater absolute benefit from treatment with accelerated TPA versus streptokinase.

Age Factors↗
PubMed1997

Relationship of body mass index to subsequent mortality among seriously ill hospitalized patients. SUPPORT Investigators. The Study to Understand Prognoses and Preferences for Outcome and Risks of Treatments.

OBJECTIVE: To determine if body mass Index (BMI = weight [kg]/height [m]2), predictive of mortality in longitudinal epidemiologic studies, was also predictive of mortality in a sample of seriously ill hospitalized subjects. DESIGN: Prospective, multicenter study. SETTING: Five tertiary care medical centers in the United States. PATIENTS: Patients > or = 18 yrs of age who had one of nine illnesses of sufficient severity to anticipate a 6-month mortality rate of 50% were enrolled at five participating sites in the Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments (SUPPORT). INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Patients were asked their current height and weight as part of the demographic data. Stratifying body mass index by percentile rank (< or = 15, 15 to 85, and > or = 85th percentiles), risk ratios for mortality were calculated by Cox Proportional Hazards using the 15th to 85th percentile of body mass index as the reference group while controlling for multiple variables such as prior weight loss, albumin, and Acute Physiology Score. A body mass index in the < or = 15th percentile was associated with an excess risk of mortality (risk ratio = 1.23; p < .001) within 6 months. High body mass index (> or = 85th percentile) was not significantly related to risk of mortality. CONCLUSIONS: Body mass index, a simple anthropometric measure of nutrition employed in community epidemiologic studies, has now been demonstrated to be a predictor of mortality in an acutely ill population of adults at five different tertiary centers. Even when controlling for multiple disease states and physiologic variables and removing from the analysis all patients with significant prior weight loss, a body mass index below the 15th percentile remained a significant and independent predictor of mortality. Examination of patient vs. proxy data did not change the results. Future studies examining variables predictive of mortality should include body mass index, even in acutely ill populations with a poor probability of survival.

APACHE↗
PubMed1997

Effect of aging on the sensitivity of growth hormone secretion to insulin-like growth factor-I negative feedback.

To determine the effect of aging on the suppression of GH secretion by insulin-like growth factor (IGF)-I, we studied 11 healthy young adults (6 men, 5 women, mean +/- SD: 25.2 +/- 4.6 yr old; body mass index 23.7 +/- 1.8 kg/m2) and 11 older adults (6 men, 5 women, 69.5 +/- 5.8 yr old; body mass index 24.2 +/- 2.5 kg/m2). Saline (control) or recombinant human IGF-I (rhIGF-I) (2 h baseline then, in sequence, 2.5 h each of 1, 3, and 10 micrograms/kg.h) was infused iv during the last 9.5 h of a 40.5-h fast; serum glucose was clamped within 15% of baseline. Baseline serum GH concentrations (mean +/- SE: 3.3 +/- 0.7 vs. 1.9 +/- 0.5 micrograms/L, P = 0.02) and total IGF-I concentrations (219 +/- 15 vs. 103 +/- 19 micrograms/L, P < 0.01) were higher in the younger subjects. In both age groups, GH concentrations were significantly decreased by 3 and 10 micrograms/kg.h, but not by 1 microgram/kg.h rhIGF-I. The absolute decrease in GH concentrations was greater in young than in older subjects during the 3 and 10 micrograms/kg.h rhIGF-I infusion periods, but both young and older subjects suppressed to a similar GH level during the last hour of the rhIGF-I infusion (0.78 +/- 0.24 microgram/L and 0.61 +/- 0.16 microgram/L, respectively). The older subjects had a greater increase above baseline in serum concentrations of both total (306 +/- 24 vs. 244 +/- 14 micrograms/L, P = 0.04) and free IGF-I (8.5 +/- 1.4 vs. 4.2 +/- 0.6 micrograms/L, P = 0.01) than the young subjects during rhIGF-I infusion, and their GH suppression expressed in relation to increases in both total and free serum IGF-I concentrations was significantly less than in the young subjects. We conclude that the ability of exogenous rhIGF-I to suppress serum GH concentrations declines with increasing age. This suggests that increased sensitivity to endogenous IGF-I negative feedback is not a cause of the decline in GH secretion that occurs with aging.

3-Hydroxybutyric Acid↗
PubMed1997

Development of an enterprise-wide clinical data repository: merging multiple legacy databases.

We describe the development of a clinical data repository whose core consists of four years of inpatient administrative and billing data from the mainframe legacy systems of the University of Virginia Health System (UVAHS). To these data we have linked a cardiac surgery clinical database and our physician billing data (inpatient and outpatient). Other databases will be merged in the future. A relational database management system (Sybase) running on a dedicated IBM RS/6000 minicomputer was employed to assemble 2.5 Gigabytes of core data describing approximately 100,000 hospital admissions over the four year period. To enable convenient data queries, the system has been equipped with a custom-built WWW user interface, which generates Structured Query Language (SQL) automatically. We illustrate the rapid reporting capabilities of the resulting system with reference to patients undergoing coronary artery bypass graft surgery (CABG). We conclude that this information system: a) constitutes a convenient and low-cost method to increase data availability across the UVAHS; b) provides clinicians with a tool for surveillance of patient care and outcomes; c) forms the core of a comprehensive database from which clinical research may proceed; d) provides a flexible interface empowering a wide variety of clinical departments to share and enrich their own clinical data.

Computer Communication Networks↗
PubMed1996

Cardiac troponin T levels for risk stratification in acute myocardial ischemia. GUSTO IIA Investigators.

BACKGROUND: The prognosis of patients hospitalized with acute myocardial ischemia is quite variable. We examined the value of serum levels of cardiac troponin T, serum creatine kinase MB (CK-MB) levels, and electrocardiographic abnormalities for risk stratification in patients with acute myocardial ischemia. METHODS: We studied 855 patients within 12 hours of the onset of symptoms. Cardiac troponin T levels, CK-MB levels, and electrocardiograms were analyzed in a blinded fashion at the core laboratory. We used logistic regression to assess the usefulness of baseline levels of cardiac troponin T and CK-MB and the electrocardiographic category assigned at admission-ST-segment elevation, ST-segment depression, T-wave inversion, or the presence of confounding factors that impair the detection of ischemia (bundle-branch block and paced rhythms)-in predicting outcome. RESULTS: On admission, 289 of 801 patients with base-line serum samples had elevated troponin T levels (> 0.1 ng per milliliter). Mortality within 30 days was significantly higher in these patients than in patients with lower levels of troponin T (11.8 percent vs. 3.9 percent, P < 0.001). The troponin T level was the variable most strongly related to 30-day mortality (chi-square = 21, P < 0.001), followed by the electrocardiographic category (chi-square = 14, P = 0.003) and the CK-MB level (chi-square = 11, P = 0.004). Troponin T levels remained significantly predictive of 30-day mortality in a model that contained the electrocardiographic categories and CK-MB levels (chi-square = 9.2, P = 0.027). CONCLUSIONS: The cardiac troponin T level is a powerful, independent risk marker in patients who present with acute myocardial ischemia. It allows further stratification of risk when combined with standard measures such as electrocardiography and the CK-MB level.

Acute Disease↗
PubMed1996

The effectiveness of right heart catheterization in the initial care of critically ill patients. SUPPORT Investigators.

OBJECTIVE: To examine the association between the use of right heart catheterization (RHC) during the first 24 hours of care in the intensive care unit (ICU) and subsequent survival, length of stay, intensity of care, and cost of care. DESIGN: Prospective cohort study. SETTING: Five US teaching hospitals between 1989 and 1994. SUBJECTS: A total of 5735 critically ill adult patients receiving care in an ICU for 1 of 9 prespecified disease categories. MAIN OUTCOME MEASURES: Survival time, cost of care, intensity of care, and length of stay in the ICU and hospital, determined from the clinical record and from the National Death Index. A propensity score for RHC was constructed using multivariable logistic regression. Case-matching and multivariable regression modeling techniques were used to estimate the association of RHC with specific outcomes after adjusting for treatment selection using the propensity score. Sensitivity analysis was used to estimate the potential effect of an unidentified or missing covariate on the results. RESULTS: By case-matching analysis, patients with RHC had an increased 30-day mortality (odds ratio, 1.24; 95% confidence interval, 1.03-1.49). The mean cost (25th, 50th, 75th percentiles) per hospital stay was $49 300 ($17 000, $30 500, $56 600) with RHC and $35 700 ($11 300, $20 600, $39 200) without RHC. Mean length of stay in the ICU was 14.8 (5, 9, 17) days with RHC and 13.0 (4, 7, 14) days without RHC. These findings were all confirmed by multivariable modeling techniques. Subgroup analysis did not reveal any patient group or site for which RHC was associated with improved outcomes. Patients with higher baseline probability of surviving 2 months had the highest relative risk of death following RHC. Sensitivity analysis suggested that a missing covariate would have to increase the risk of death 6-fold and the risk of RHC 6-fold for a true beneficial effect of RHC to be misrepresented as harmful. CONCLUSION: In this observational study of critically ill patients, after adjustment for treatment selection bias, RHC was associated with increased mortality and increased utilization of resources. The cause of this apparent lack of benefit is unclear. The results of this analysis should be confirmed in other observational studies. These findings justify reconsideration of a randomized controlled trial of RHC and may guide patient selection for such a study.

APACHE↗
PubMed1996

Factors associated with do-not-resuscitate orders: patients' preferences, prognoses, and physicians' judgments. SUPPORT Investigators. Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatment.

BACKGROUND: Medical treatment decisions should be based on the preferences of informed patients or their proxies and on the expected outcomes of treatment. Because seriously ill patients are at risk for cardiac arrest, examination of do-not-resuscitate (DNR) practices affecting them provides useful insights into the associations between various factors and medical decision making. OBJECTIVE: To examine the association between patients' preferences for resuscitation (along with other patient and physician characteristics) and the frequency and timing of DNR orders. DESIGN: Prospective cohort study. SETTING: 5 teaching hospitals. PATIENTS: 6802 seriously ill hospitalized patients enrolled in the Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatment (SUPPORT) between 1989 and 1994. MEASUREMENTS: Patients and their surrogates were interviewed about patients' cardiopulmonary resuscitation preferences, medical records were reviewed to determine disease severity, and a multivariable regression model was constructed to predict the time to the first DNR order. RESULTS: The patients' preference for cardiopulmonary resuscitation was the most important predictor of the timing of DNR orders, but only 52% of patients who preferred not to be resuscitated actually had DNR orders written. The probability of surviving for 2 months was the next most important predictor of the timing of DNR orders. Although DNR orders were not linearly related to the probability of surviving for 2 months, they were written earlier and more frequently for patients with a 50% or lower probability of surviving for 2 months. Orders were written more quickly for patients older than 75 years of age, regardless of prognosis. After adjustment for these and other influential patient characteristics, the use and timing of DNR orders varied significantly among physician specialties and among hospitals. CONCLUSIONS: Patients' preferences and short-term prognoses are associated with the timing of DNR orders. However, the substantial variation seen among hospital sites and among physician specialties suggests that there is room for improvement. In this study, DNR orders were written earlier for patients older than 75 years of age, regardless of prognosis. This finding suggests that physicians may be using age in a way that is inconsistent with the reported association between age and survival. The process for making decisions about DNR orders needs to be improved if such orders are to routinely and accurately reflect patients' preferences and probable outcomes.

Aged↗
PubMed1996

Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors.

Multivariable regression models are powerful tools that are used frequently in studies of clinical outcomes. These models can use a mixture of categorical and continuous variables and can handle partially observed (censored) responses. However, uncritical application of modelling techniques can result in models that poorly fit the dataset at hand, or, even more likely, inaccurately predict outcomes on new subjects. One must know how to measure qualities of a model's fit in order to avoid poorly fitted or overfitted models. Measurement of predictive accuracy can be difficult for survival time data in the presence of censoring. We discuss an easily interpretable index of predictive discrimination as well as methods for assessing calibration of predicted survival probabilities. Both types of predictive accuracy should be unbiasedly validated using bootstrapping or cross-validation, before using predictions in a new data series. We discuss some of the hazards of poorly fitted and overfitted regression models and present one modelling strategy that avoids many of the problems discussed. The methods described are applicable to all regression models, but are particularly needed for binary, ordinal, and time-to-event outcomes. Methods are illustrated with a survival analysis in prostate cancer using Cox regression.

Clinical Trials as Topic↗
PubMed1996

Use of predicted risk of mortality to evaluate the efficacy of anticytokine therapy in sepsis. The rhIL-1ra Phase III Sepsis Syndrome Study Group.

OBJECTIVES: To investigate a novel anticytokine therapy in patients with sepsis syndrome, and the relationship between a patient's baseline mortality risk and survival benefit. DESIGN: Data from a recent phase III, double-blind, placebo-controlled, multicenter clinical trial with patients randomized to three treatment arms: an intravenous loading dose of recombinant human interleukin-1-receptor antagonist (rhIL-1ra) or placebo, followed by a continuous infusion of rhIL-1ra (1.0 mg/kg/hr, or 2.0 mg/kg/hr), or placebo for 72 hrs. SETTING: Sixty-three investigative centers in eight countries. PATIENTS: The study population consisted of 893 patients: 302 placebo patients; 298 patients treated with 1.0 mg/kg/hr of rhIL-1ra; and 293 patients treated with 2.0 mg/kg/hr of rhIL-1ra. MEASUREMENTS AND MAIN RESULTS: An independent, sepsis-specific, log-normal regression model that predicts the risk of mortality over 28 days was applied to all patients enrolled into the rhIL-1ra sepsis study. The ability of the Predicted Risk of Mortality model to predict 28-day mortality in the placebo patients was determined and the relationship between mortality risk and efficacy of rhIL-1ra was investigated. The trial data were also analyzed using two other risk-assessment models for comparison with Predicted Risk of Mortality. A significant increase in survival time was demonstrated for all patients treated with rhIL-1ra (n = 893, p < .02 Predicted Risk of Mortality log-normal), but patients with a Predicted Risk of Mortality of < 24% derived little benefit. Retrospective examination of time-to-death data demonstrated that rhIL-1ra reduced risk of death in the first 2 days for patients with > or = 24% Predicted Risk of Mortality (n = 580, p < .005 Predicted Risk of Mortality log-normal). This same effect was not present in patients with a Predicted Risk of Mortality of < 24% on entry into the study. The Predicted Risk of Mortality model predicted a 28-day mortality rate of 35% for placebo patients compared with 34% observed and accurately stratified patients along the full range of risks. There was a wide distribution of individual patient risks for 28-day mortality for all patients, as well as within categorical subgroups, such as shock and organ system dysfunction. Two alternate risk models were assessed and the Acute Physiology Score of Acute Physiology and Chronic Health Evaluation III also demonstrated a statistically significant survival benefit for rhIL-1ra (p = .04 Predicted Risk of Mortality log-normal) for all patients treated. CONCLUSIONS: Using an appropriate analytic model, a statistically significant increase in survival time from rhIL-1ra was measured. A direct relationship was found between a patient's Predicted Risk of Mortality at study entry to efficacy of rhIL-1ra. Individual risk or severity assessment may be a useful tool for evaluating the clinical benefit of new therapeutic approaches to sepsis and for monitoring outcomes at the bedside.

APACHE↗
PubMed1996

Outcomes following acute exacerbation of severe chronic obstructive lung disease. The SUPPORT investigators (Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments)

In order to describe the outcomes of patients hospitalized with an acute exacerbation of severe chronic obstructive pulmonary disease (COPD) and determine the relationship between patient characteristics and length of survival, we studied a prospective cohort of 1,016 adult patients from five hospitals who were admitted with an exacerbation of COPD and a PaCO2 of 50 mm Hg or more. Patient characteristics and acute physiology were determined. Outcomes were evaluated over a 6 mo period. Although only 11% of the patients died during the index hospital stay, the 60-d, 180-d, 1-yr, and 2-yr mortality was high (20%, 33%, 43%, and 49%, respectively). The median cost of the index hospital stay was $7,100 ($4,100 to $16,000; interquartile range). The median length of the index hospital stay was 9 d (5 to 15 d). After discharge, 446 patients were readmitted 754 times in the next 6 mo. At 6 mo, only 26% of the cohort were both alive and able to report a good, very good, or excellent quality of life. Survival time was independently related to severity of illness, body mass index (BMI), age, prior functional status, PaO2/FI(O2), congestive heart failure, serum albumin, and the presence of cor pulmonale. Patients and caregivers should be aware of the likelihood of poor outcomes following hospitalization for exacerbation of COPD associated with hypercarbia.

APACHE↗
PubMed1995

The restricted cubic spline as baseline hazard in the proportional hazards model with step function time-dependent covariables.

We incorporate a cubic spline function where the tails are linearly constrained, as the baseline hazard, into the proportional hazards model. We show estimation of covariable coefficients and survival probabilities with this model to be as efficient statistically as with the Cox proportional hazards model when covariables are fixed. Examples show that the inclusion of time-dependent covariables defined as step functions into the restricted cubic spline proportional hazards model reduces computation time by a factor of 213 over the Cox model. Advantages of the spline model also include flexibility of the hazard, smooth survival curves, and confidence limits for the survival and hazard estimates when there are time-dependent covariables present.

Coronary Artery Bypass↗
PubMed1995

Identification of comatose patients at high risk for death or severe disability. SUPPORT Investigators. Understand Prognoses and Preferences for Outcomes and Risks of Treatments.

OBJECTIVE: To develop and validate a simple prognostic scoring system to identify patients in nontraumatic coma at high risk for poor outcomes using data available early in the hospital course. DESIGN: Prospective cohort study. SETTING: Five geographically diverse academic medical centers. PATIENTS: A total of 596 patients in nontraumatic coma enrolled in the Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments (SUPPORT), including 247 in the model derivation set and 349 in the model validation set. MAIN OUTCOME MEASURES: Death and severe disability by 2 months. MAIN RESULTS: For the 596 patients studied (median age, 67 years; 52% female), the primary cause of coma was cardiac arrest in 31% and cerebral infarction or intracerebral hemorrhage in 36%. At 2 months 69% had died, 20% had survived with known severe disability, 8% were known to have survived without severe disability, and 3% survived with unknown functional status. Five clinical variables available on day 3 after enrollment were associated independently with 2-month mortality: abnormal brain stem response (adjusted odds ratio [OR] = 3.2; 95% confidence interval [CI], 1.3 to 8.1), absent verbal response (OR = 4.6; 95% CI, 1.8 to 11.7), absent withdrawal response to pain (OR = 4.3; 95% CI, 1.7 to 10.8), creatinine level greater than or equal to 132.6 mumol/L (1.5 mg/dL) (OR = 4.5; 95% CI, 1.8 to 11.0), and age of 70 years or older (OR = 5.1; 95% CI, 2.2 to 12.2). Mortality at 2 months for patients with four or five of these risk factors was 97% (58/60; 95% CI, 88% to 100%) in the validation set. Brain stem and motor responses best predicted death or severe disability by 2 months. For patients with either an abnormal brain stem response or absent motor response to pain, the rate of death or severe disability at 2 months was 96% (185/193; 95% CI, 92% to 98%) in the validation set. CONCLUSIONS: Five readily available clinical variables identify a large subgroup of patients in nontraumatic coma at high risk for poor outcomes. This risk stratification approach offers physicians, patients, and patients' families information that may prove useful in patient care decisions and resource allocation.

APACHE↗
PubMed1995

Predicting future functional status for seriously ill hospitalized adults. The SUPPORT prognostic model.

OBJECTIVE: To develop a model estimating the probability of an adult patient having severe functional limitations 2 months after being hospitalized with one of nine serious illnesses. DESIGN: Prospective cohort study. SETTING: Five teaching hospitals in the United States. PARTICIPANTS: 1746 patients (model development) who survived 2 months and completed an interview, selected from 4301 patients in the Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments (SUPPORT); independent validation sample of 2478 patients. MEASUREMENTS AND MAIN OUTCOMES: Patient function 2 months after admission categorized as absence or presence of severe functional limitations (defined as Sickness Impact Profile scores > or = 30 or as activities of daily living scores > or = 4 [levels that require near-constant personal assistance]). A logistic regression model was constructed to predict severe functional limitation. RESULTS: One third (n = 590) of patients who were interviewed at 2 months had severe functional limitations. Changes in functional status were common: Of those with no baseline dependencies (not dependent on personal assistance), 21% were severely limited at 2 months; of those with 4 or more baseline limitations, 30% had improved. The patient's ability to do activities of daily living was the most important predictor of functional status. Physiologic abnormalities, diagnosis, days in hospital, age, quality of life, and previous exercise capacity also contributed substantially. Model performance, assessed using receiver-operating characteristic curves, was 0.79 for the development sample and 0.75 for the validation sample. The model was well calibrated for the entire risk range. CONCLUSIONS: Functional outcome varied substantially after hospitalization for a serious illness. A small amount of readily available clinical information can estimate the probability of severe functional limitations.

Activities of Daily Living↗
PubMed1995

The SUPPORT prognostic model. Objective estimates of survival for seriously ill hospitalized adults. Study to understand prognoses and preferences for outcomes and risks of treatments.

OBJECTIVE: To develop and validate a prognostic model that estimates survival over a 180-day period for seriously ill hospitalized adults (phase I of SUPPORT [Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments]) and to compare this model's predictions with those of an existing prognostic system and with physicians' independent estimates (SUPPORT phase II). DESIGN: Prospective cohort study. SETTING: 5 tertiary care academic centers in the United States. PARTICIPANTS: 4301 hospitalized adults were selected for phase I according to diagnosis and severity of illness; 4028 patients were evaluated from phase II. MEASUREMENTS: A survival model was developed using the following predictor variables: diagnosis, age, number of days in the hospital before study entry, presence of cancer, neurologic function, and 11 physiologic measures recorded on day 3 after study entry. Physicians were interviewed on day 3. Patients were followed for survival for 180 days after study entry. RESULTS: The area under the receiver-operating characteristics (ROC) curve for prediction of surviving 180 days was 0.79 in phase I, 0.78 in the phase II independent validation, and 0.78 when the acute physiology score from the APACHE (Acute Physiology, Age, Chronic Health Evaluation) III prognostic scoring system was substituted for the SUPPORT physiology score. For phase II patients, the SUPPORT model had equal discrimination and slightly improved calibration compared with physician's estimates. Combining the SUPPORT model with physician's estimates improved both predictive accuracy (ROC curve area = 0.82) and the ability to identify patients with high probabilities of survival or death. CONCLUSIONS: A limited amount of readily available clinical information can provide a foundation for long-term survival estimates that are as accurate as physicians' estimates. The best survival estimates combine an objective prognosis with a physician's clinical estimate.

APACHE↗
PubMed1995

Late withdrawal of cyclosporine in stable renal transplant recipients.

The use of cyclosporine (CsA) in renal transplantation has been associated with an improvement in 1-year graft survival, but has not changed the rate of late graft loss. We sought to determine whether the intent to withdraw CsA late after renal transplantation affects renal transplant survival and whether there is a racial difference in the effect of CsA withdrawal. This retrospective study included 384 consecutive patients receiving a renal transplant during the 1984 to 1991 period who were treated with CsA/azathioprine/prednisone and who had a functioning allograft 6 months following transplantation. Of these, 97 were electively withdrawn from CsA at a median of 22 months following transplantation. Factors significantly associated with the decision to withdraw CsA included white race, older age, and lower serum creatinine. Acute rejection within 6 months of stopping CsA occurred in 12 patients (12.4%), including nine of 78 (11.5%) white patients and three of 19 (15.8%) black patients. For the group of 287 patients who were not withdrawn from CsA, the 6-year graft survival rate was 59% (95% confidence interval, 52%, 66%). For the group of patients taken off of CsA, the 6-year graft survival rate was 84% (95% confidence interval, 76%, 92%). Cox proportional hazard survival analysis indicated that the intent to discontinue CsA was associated with better graft survival, with a hazard ratio of 0.37 (95% confidence interval, 0.20, 0.70), independent of other variables that may affect graft survival. A separate analysis controlling for waiting time bias also favored the CsA withdrawal group. There was no detectable racial difference in the effect of CsA withdrawal on graft survival.(ABSTRACT TRUNCATED AT 250 WORDS)

Adult↗