Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “risk model”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 325 records · Page 18Linked to original sources
PubMedDate not supplied

[Development of an acute myocardial infarction risk adjustment model for Spain: comparison to the Charlson model and the ICES model. Application for gauging care results].

BACKGROUND: Acute myocardial infarction has a major bearing on morbimortality in developed countries. This study is aimed at developing a risk adjustment model for assessing the results of managing this disease and comparing how this management is carried out with other models. METHODS: A risk adjustment model is developed for acute myocardial infarction by means of logistic regression with the information from an administrative database including various Spanish hospitals, taking the intrahospital mortality rate as the response variable and variables inherent to the patient proper as predictive variables. The predictability thereof is compared to the Charlson Model and the ICES model. These models are applied for assessing the intrahospital mortality rate. RESULTS: The age (OR: 1.07), the anterolateral location (OR: 2.32) and inferoposterior location (OR: 1.91), cardiogenic shock (OR: 39.99), arrhythmia (OR: 94.43), cerebrovascular disease (OR: 2.15) and kidney failure (OR: 1.87) are shown to be related to a higher risk of death. The model developed provides a better predictability (-2LL: 2240) than the Charlson model (-2LL: 3073) and the ICES model (-2LL: 2366). There are no significant differences in the risk-adjusted death rate for the 23 hospitals studied, nor any differences between public and private hospitals nor the care volume or technological level. Coronariography (RMAR: 0.66) and percutaneous coronary interventions (RMAR: 0.69) are procedures having a protective effect. CONCLUSIONS: The model developed may be a useful tool for assessing the hospital care provided for myocardial infarction.

Female↗
PubMed2005

Managing industrial risk--having a tested and proven system to prevent and assess risk.

Some relatively easy techniques exist to improve the risk picture/profile to aid in preventing losses. Today with the advent of computer system resources, focusing on specific aspects of risk through systematic scoring and comparison, the risk analysis can be relatively easy to achieve. Techniques like these demonstrate how working experience and common sense can be combined mathematically into a flexible risk management tool or risk model for analyzing risk. The risk assessment methodology provided by companies today is no longer the ideas and practices of one group or even one company. It is reflective of the practice of many companies, as well as the ideas and expertise of academia and government regulators. The use of multi-criteria decision making (MCDM) techniques for making critical decisions has been recognized for many years for a variety of purposes. In today's computer age, the easy accessing and user-friendly nature for using these techniques, makes them a favorable choice for use in the risk assessment environment. The new user of these methodologies should find many ideas directly applicable to his or her needs when approaching risk decision making. The user should find their ideas readily adapted, with slight modification, to accurately reflect a specific situation using MCDM techniques. This makes them an attractive feature for use in assessment and risk modeling. The main advantage of decision making techniques, such as MCDM, is that in the early stages of a risk assessment, accurate data on industrial risk, and failures are lacking. In most cases, it is still insufficient to perform a thorough risk assessment using purely statistical concepts. The practical advantages towards deviating from strict data-driven protocol seem to outweigh the drawbacks. Industry failure data often comes at a high cost when a loss occurs. We can benefit from this unfortunate acquisition of data through the continuous refining of our decisions by incorporating this new information into our assessments. MCDM techniques offer flexibility in accessing comparison within broad data sets to reflect our best estimation of their importance towards contribution to the risk picture. This allows for the accurate determination of the more probable and more consequential issues. This can later be refined using more intensive risk techniques and the avoidance of less critical issues.

Decision Making, Organizational↗
PubMed2000

Challenges in comparing risk-adjusted bypass surgery mortality results: results from the Cooperative Cardiovascular Project.

OBJECTIVES: We sought to evaluate the predictive accuracy of four bypass surgery mortality clinical risk models and to examine the extent to which hospitals' risk-adjusted surgical outcomes vary depending on which risk-adjustment method is applied. BACKGROUND: Cardiovascular "report cards" often compare risk-adjusted surgical outcomes; however, it is unclear to what extent the risk-adjustment process itself may affect these metrics. METHODS: As part of the Cooperative Cardiovascular Project's Pilot Revascularization Study, we compared the predictive accuracy of four bypass clinical risk models among 3,654 Medicare patients undergoing surgery at 28 hospitals in Alabama and Iowa. We also compared the agreement in hospital-level risk-adjusted bypass outcome performance ratings depending on which of the four risk models was applied. RESULTS: Although the four risk models had similar discriminatory abilities (C-index, 0.71 to 0.74), certain models tended to overpredict mortality in higher-risk patients. There was high correlation between a hospital's risk-adjusted mortality rates regardless of which of the four models was used (correlation between risk-adjusted rating, 0.93 to 0.97). In contrast, there was limited agreement in which hospitals were identified as "performance outliers" depending on which risk-adjustment model was used and how outlier status was defined. CONCLUSIONS: A hospital's risk-adjusted bypass surgery mortality rating, relative to its peers, was consistent regardless of the risk-adjustment model applied, supporting their use as a means of provider performance feedback. Designation of performance outliers, however, can vary significantly depending on the benchmark and methods used for this determination.

Aged↗
PubMed2005

[Developing a hospital-based high risk scoring model and screening strategy for pancreatic cancer].

OBJECTIVE: To develop a high risk scoring model and screening strategy to improve the diagnosis of symptomatic pancreatic cancer. METHODS: A hospital-based case-control study was undertaken among a cohort comprising 136 pancreatic cancer patients and 191 patients with benign gastrointestinal diseases who were hospitalized between Feb, 2002 and Mar, 2004. All patients were consulted with an epidemiological questionnaire. Risk factors and symptoms described in the questionnaire were compared between these two groups. Significant and borderline risk factors and symptoms were selected to undergo multivariate logistic regression. A high risk scoring model was constructed according to the weighted numerical scores of every variable. The diagnostic values of 4 tumor markers of pancreatic cancers (serum CA19-9, CA242, stool K-ras and p53 mutation) and 2 imaging tests (abdominal spiral CT and ultrasonography) were evaluated to provide evidence for establishing the diagnostic strategy. RESULTS: The average score was significantly higher for the pancreatic cancer patients than for the control patients [mean 49.6 (95% CI: 45.6-53.7) vs 21.6 (95% CI: 19.3-23.9); P < 0.01]. With a cutoff value of 27 points, the sensitivity and specificity of the scoring model was 87.0% and 68.1% respectively. CT had the highest sensitivity (94.7%) among the 4 tumor markers and 2 imaging tests. Combination of the two tumor markers (CA19-9 and stool K-ras) with CT or ultrasonography could improve the sensitivity to 100% with a specificity of 67.5%-73.0%. It was suggested that for high risk patients with a risk score more than 27, the combination test be recommended as the primary test, endoscopic ultrasonography (EUS) and/or endoscopic retrograde cholangipancreatography (ERCP) be considered for patients with inconclusive CT studies when risk score and tumor markers nevertheless suggest pancreatic cancer. CONCLUSION: The high risk scoring model provides a simple and feasible way to screen pancreatic patients in hospitals at all levels. Once high risk patients are identified, they can be transferred to higher level hospitals to receive further examinations. This screening strategy may help detect more resectable pancreatic cancers.

Adult↗
PubMed1992

University of North Carolina Caries Risk Assessment Study: comparisons of high risk prediction, any risk prediction, and any risk etiologic models.

The purpose of this analysis is to compare three different statistical models for predicting children likely to be at risk of developing dental caries over a 3-yr period. Data are based on 4117 children who participated in the University of North Carolina Caries Risk Assessment Study, a longitudinal study conducted in the Aiken, South Carolina, and Portland, Maine areas. The three models differed with respect to either the types of variables included or the definition of disease outcome. The two "Prediction" models included both risk factor variables thought to cause dental caries and indicator variables that are associated with dental caries, but are not thought to be causal for the disease. The "Etiologic" model included only etiologic factors as variables. A dichotomous outcome measure--none or any 3-yr increment, was used in the "Any Risk Etiologic model" and the "Any Risk Prediction Model". Another outcome, based on a gradient measure of disease, was used in the "High Risk Prediction Model". The variables that are significant in these models vary across grades and sites, but are more consistent among the Etiologic model than the Predictor models. However, among the three sets of models, the Any Risk Prediction Models have the highest sensitivity and positive predictive values, whereas the High Risk Prediction Models have the highest specificity and negative predictive values. Considerations in determining model preference are discussed.

Child↗
PubMed1995

Intentions to share injection paraphernalia: an empirical test of the AIDS Risk Reduction Model among injection drug users.

The AIDS Risk Reduction Model is a theory-based representation of sequential psychosocial processes by which people may attempt to change their HIV risk behavior. These processes are said to occur in three stages: 1) labeling one's risk behavior as problematic, 2) forming an intention to change behavior, and 3) taking action to accomplish change. Cognitive and perceptual factors are said to influence progress across stages. This study tests the degree to which hypothesized relationships among factors at Stages 1 and 2 are consistent with cross-sectional data collected from a sample of HIV-negative injection drug users who reported recent sharing of drug injection paraphernalia ("works"). Findings indicate that intentions to share works less often in the future may be influenced directly by drug users' perceived risk of infection, which in turn is influenced by their level of HIV knowledge, perceived susceptibility to HIV, and perceived peer norms regarding drug-related risk reduction.

Acquired Immunodeficiency Syndrome↗
PubMed2006

Quantitative microbial risk assessment models for consumption of raw vegetables irrigated with reclaimed water.

Quantitative microbial risk assessment models for estimating the annual risk of enteric virus infection associated with consuming raw vegetables that have been overhead irrigated with nondisinfected secondary treated reclaimed water were constructed. We ran models for several different scenarios of crop type, viral concentration in effluent, and time since last irrigation event. The mean annual risk of infection was always less for cucumber than for broccoli, cabbage, or lettuce. Across the various crops, effluent qualities, and viral decay rates considered, the annual risk of infection ranged from 10(-3) to 10(-1) when reclaimed-water irrigation ceased 1 day before harvest and from 10(-9) to 10(-3) when it ceased 2 weeks before harvest. Two previously published decay coefficients were used to describe the die-off of viruses in the environment. For all combinations of crop type and effluent quality, application of the more aggressive decay coefficient led to annual risks of infection that satisfied the commonly propounded benchmark of < or =10(-4), i.e., one infection or less per 10,000 people per year, providing that 14 days had elapsed since irrigation with reclaimed water. Conversely, this benchmark was not attained for any combination of crop and water quality when this withholding period was 1 day. The lower decay rate conferred markedly less protection, with broccoli and cucumber being the only crops satisfying the 10(-4) standard for all water qualities after a 14-day withholding period. Sensitivity analyses on the models revealed that in nearly all cases, variation in the amount of produce consumed had the most significant effect on the total uncertainty surrounding the estimate of annual infection risk. The models presented cover what would generally be considered to be worst-case scenarios: overhead irrigation and consumption of vegetables raw. Practices such as subsurface, furrow, or drip irrigation and postharvest washing/disinfection and food preparation could substantially lower risks and need to be considered in future models, particularly for developed nations where these extra risk reduction measures are more common.

Agriculture↗
PubMed1990

Therapeutic risk-assessment model for identifying patients with adverse drug reactions.

The association between factors that place patients at risk for adverse drug reactions (ADRs) and the occurrence of ADRs was examined, and a therapeutic risk-assessment model was developed. Theoretical risk factors for ADRs to digoxin and theophylline were identified through the literature by researchers at a private tertiary-care hospital. Data were then collected from two groups of 67 patient charts each during a 15-month period. One group of charts represented patients who had experienced an ADR to digoxin or theophylline. The other group represented matched control patients who had not experienced an ADR to either drug. ICD-9-CM (International Classification of Diseases, 9th Revision, Clinical Modifications) ADR codes were assigned by medical records department personnel, and the ADRs were verified by using the Naranjo algorithm. Seven risk factors for each drug were found to be significantly associated with ADRs. A serum digoxin concentration greater than 2.5 ng/mL and elevated blood urea nitrogen were the two best predictors of an ADR to digoxin. The probability of experiencing an ADR to digoxin was 94.1% for a patient with both of these risk factors. A serum theophylline concentration greater than 25 micrograms/mL was the greatest predictor of an ADR to theophylline; the probability of experiencing an ADR to theophylline was 85.2% if a patient had that risk factor. The sensitivity and specificity of the therapeutic risk-assessment model were 92.9% and 61.8%, respectively, for digoxin and 95.8% and 84.0%, respectively, for theophylline. Several laboratory-based screening criteria demonstrated an ability to predict ADRs to digoxin and theophylline.

Digoxin↗
PubMed1986

Radiation carcinogenesis: cumulative empirical distribution functions and risk projection models.

The use of cumulative empirical distribution functions (CEDF) in connection with the prediction of radio-induced cancer risk is discussed. Two models are generally used for risk prediction: the relative risk and the absolute risk models. It can be easily demonstrated that if the relative projection model is valid, then the CEDFs of the cohort exposed to radiation and of the non-exposed cohort are the same. Land and Tokunaga have shown that for several sets of data the CEDF of exposed and non-exposed cohorts are indistinguishable. They used this result to recommend the use of the relative risk projection model and the rejection of the absolute projection model. This report discusses their argument and shows that when the absolute projection model is valid, the non-exposed and exposed CEDFs are also indistinguishable.

Humans↗
PubMed1989

Expected indoor 222Rn levels in counties with very high and very low lung cancer rates.

Counties in the U.S. with high lung cancer rates should have higher average 222Rn levels than counties with low lung cancer rates, assuming the average 222Rn level in a county is not correlated with other factors that cause lung cancer. The magnitude of this effect was calculated, using the absolute risk model, the relative risk model, and an intermediate model, for females who died in 1950-1969. The results were similar for all three models. We concluded that, ignoring migration, the average Rn level in the highest lung cancer counties should be about three times higher than in the lowest lung cancer counties according to the theory. Preliminary data are presented indicating that the situation is quite the opposite: The average Rn level in the highest lung cancer counties was only about one-half that in the lowest lung cancer counties.

Cohort Studies↗
PubMed2026

Risk prediction models for blood transfusion in patients undergoing total hip and knee arthroplasty: a systematic review and meta-analysis.

OBJECTIVE: To systematically review and evaluate published risk prediction models for perioperative blood transfusion in patients undergoing total hip or knee arthroplasty (THA/TKA). METHODS: We systematically searched PubMed, Web of Science, the Cochrane Library, and Embase from inception to May 31, 2025. Two researchers independently screened the literature, extracted data, and assessed the risk of bias and applicability using the Prediction model Risk Of Bias Assessment Tool (PROBAST). The area under the receiver operating characteristic curve (AUC) values were pooled via a meta-analysis using Stata 18.0. RESULTS: d Fourteen studies containing 36 prediction models were included. The incidence of blood transfusion among THA/TKA patients ranged from 3.2% to 30.8%. Preoperative hemoglobin (Hb) level, tranexamic acid (TXA) use, operative duration, intraoperative blood loss, and age were the most frequently incorporated predictors. Model sensitivity ranged from 58% to 94.5%, and specificity ranged from 71.3% to 94%. Meta-analysis showed that the pooled AUC value of the 13 validated models was 0.87 (95% CI: 0.85-0.90), suggesting good discriminatory performance. All models were rated as having a high risk of bias. The applicability of four studies was rated as unclear. CONCLUSION: Although the included studies demonstrated promising discriminative ability of prediction models for blood transfusion in THA/TKA, all were assessed as having a high risk of bias using the PROBAST tool. Therefore, future research should prioritize the development of models with larger sample sizes, rigorous study designs, and multicenter external validation.

Humans↗
PubMed2002

Recent innovations in intensive care unit risk-prediction models.

During the past 20 years, ICU risk-prediction models have undergone significant development, validation, and refinement. Among the general ICU severity of illness scoring systems, the Acute Physiology and Chronic Health Evaluation (APACHE), Mortality Prediction Model (MPM), and the Simplified Acute Physiology Score (SAPS) have become the most accepted and used. To risk-adjust patients with longer, more severe illnesses like sepsis and acute respiratory distress syndrome, several models of organ dysfunction or failure have become available, including the Multiple Organ Dysfunction Score (MODS), the Sequential Organ Failure Assessment (SOFA), and the Logistic Organ Dysfunction Score (LODS). Recent innovations in risk adjustment include automatic physiology and diagnostic variable retrieval and the use of artificial intelligence. These innovations have the potential of extending the uses of case-mix and severity-of-illness adjustment in the areas of clinical research, patient care, and administration. The challenges facing intensivists in the next few years are to further develop these models so that they can be used throughout the IUC stay to assess quality of care and to extend them to more specific patient groups such as the elderly and patients with chronic ICU courses.

Artificial Intelligence↗
PubMed2003

Estimating group mortality and paraplegia rates after thoracoabdominal aortic aneurysm repair.

BACKGROUND: Most clinical studies regarding thoracoabdominal aortic aneurysm (TAAA) surgery are retrospective comparisons involving heterogeneous groups of patients. Risk models that evaluate susceptibility bias enhance interpretation of these intergroup comparisons. The purpose of this analysis was to derive group risk models for mortality and paraplegia after TAAA repair. METHODS: Data regarding 1,220 consecutive patients undergoing TAAA repair were analyzed via multiple logistic regression with stepwise model selection. Categorical preoperative risk factors that predicted 30-day mortality and paraplegia were used to develop risk models. RESULTS: Fifty-eight patients (4.8%) died within 30 days and 56 patients (4.6%) developed paraplegia or paraparesis. Predictors of mortality were rupture, renal insufficiency, symptomatic aneurysms, and Crawford extent II repairs. Extent of repair and acute presentation were predictors of paraplegia. The derived risk models estimated mortality and paraplegia rates that correlated well with actual frequencies reported in other contemporary series (regression slopes = 0.87 and 1.06, respectively). CONCLUSIONS: The derived risk models accurately estimate paraplegia and mortality rates in groups of patients. Prospective model validation will be required to confirm their accuracy.

Adolescent↗
PubMed2002

A proposed methodology of cancer risk assessment modeling using biomarkers.

A methodology for cancer risk assessment modeling was developed using a biomarker of DNA adduct, exposure dose, and tumor response. DNA adducts in the blood and lung were measured after single or multiple administration of [3H]benzo[a]pyrene (1 x BaP) in ICR mice. Making the assumption that DNA adducts are formed in a dose-dependent manner as observed in 1 x BaP treatment, kinetics patterns of DNA adducts were predicted at two other hypothetical BaP doses (2 x BaP, 1/2 x BaP) for single and continuous BaP treatments because the difference between the simulated and the experimental kinetic responses only amounted to 5.49-5.86% in terms of the integrated area under the curve. Correlations between the formation of DNA adducts and exposure doses or between blood DNA and lung DNA adducts were determined to be linear. The dose-response relationship between biomarker and exposure dose was further incorporated into a dose-tumor response equation, obtained from 2-yr bioassay, to predict cancer risk. The interrelationships between exposure dose, biomarker, and tumor response allowed the prediction of cancer risk in animals, once the information on biomarker levels was obtained. Moreover, this methodology could be further applied to human cancer risk assessment after appropriate safety factors were employed.

Animals↗
PubMed2003

Baseline and early lymphopenia predict for the risk of febrile neutropenia after chemotherapy.

A risk model for febrile neutropenia (FN) after conventional cytotoxic chemotherapy, based on early (day 5) lymphopenia and the dose of chemotherapy, has been described. A risk index based on parameters available at day 1 would be easier in daily practice. The objectives of this work were (1) to investigate a risk model for FN using only day 1 blood cell count and (2) to compare the day 1 and day 5 risk models. Three series of patients were used for the delineation and/or validation of these two risk models: (1) the exhaustive cohort of 950 patients treated in the Department of Medicine of the CLB in 1996 (CLB-1996 series), (2) the Elypse 1 series, a prospective series of 321 patients treated in community hospitals and regional cancer centres, and (3) a previously reported Elypse 0 series of 329 patients. Day 1 blood cell count was available in all three series, while day 5 blood cell count was available only in the Elypse 0 and 1 series. In the CLB-1996 series, 92 (9.7%) patients experienced FN; only chemotherapy dose and day 1 lymphopenia < or =700 microl(-1) had an independent prognostic value for FN in multivariate analysis. In patients with both risk factors ('high-risk group'), the incidence of FN was 44, 50 and 61% in the CLB-1996. Elypse 1 and 0 series, respectively, indicating that the 'day 1' risk model enables one to identify patients at high-risk for FN. Besides, the observed incidence of FN in the high-risk group of the 'day 5' model (i.e. patients with day 5 lymphopenia < or =700 microl(-1) and receiving high-risk CT) was 45 and 69% in the Elypse 0 and 1 series, respectively. In the Elypse 1 and 0 series, 15 and 12% of all patients who experienced FN were in the high-risk group of the 'day 1' risk model as compared to 25 and 62% for the high-risk group of the 'day 5' risk model. Both day 1 and day 5 lymphopenia are associated with an increased risk of FN in patients treated with chemotherapy. The 'day 1' model identifies a small population of patients at high risk for FN, but has a lower sensitivity than the day 5 model.

Adolescent↗
PubMed2001

Geographic information systems as a tool for control program management for schistosomiasis in Egypt.

During a 4-year study a geographic information system (GIS) risk model was constructed for predicting the relative risk of schistosomiasis in Kafr El-Sheikh governorate, Egypt. A 1-year 1990-1991 time series on diurnal temperature difference (dT) prepared from the advanced very high resolution radiometer (AVHRR) sensor on the NOAA-11 satellite was used to develop a regional risk model for the Nile delta based on thermal-hydrological domains. A May 15, 1990 Landsat TM scene (path 177, Row 38) was used to develop a local 'village-scale' environmental risk model based on higher resolution satellite sensor data (30 m picture element size at earth surface). Four of ten classes derived from a tasseled cap (Tcap) transformation of the Landsat TM scene were shown to be significantly related to a 5-year Schistosoma mansoni prevalence database from the Ministry of Health. A risk model was developed based on dT and the proportional area of the four Tcap classes in 5 km(2) buffer zones centered on rural health unit (RHU) reporting units. Available historical data on S. mansoni and its snail host Biomphalaria alexandrina, as well as recent field collected data were gathered and incorporated as separate themes. Model validation was done using data collected on snail population bionomics-infection rates, water quality, underground water table and cercariometry at 13 hydrologically representative sites. The role of soil type, water table and water quality was studied at 79 of 154 rural health unit sites. The model permitted retrieval of relevant data by RHU point location. For the first time in Egypt, the Kafr El-Sheikh GIS schistosoma prediction model can support MOH efforts to make more accurate control program decisions based on environmental predilection sites of endemic Schistosomiasis mansoni.

Animals↗
PubMed2006

The effect of including C-reactive protein in cardiovascular risk prediction models for women.

BACKGROUND: While high-sensitivity C-reactive protein (hsCRP) is an independent predictor of cardiovascular risk, global risk prediction models incorporating hsCRP have not been developed for clinical use. OBJECTIVE: To develop and compare global cardiovascular risk prediction models with and without hsCRP. DESIGN: Observational cohort study. SETTING: U.S. female health professionals. PARTICIPANTS: Initially healthy nondiabetic women age 45 years and older participating in the Women's Health Study and followed an average of 10 years. MEASUREMENTS: Incident cardiovascular events (myocardial infarction, stroke, coronary revascularization, and cardiovascular death). RESULTS: High-sensitivity CRP made a relative contribution to global risk at least as large as that provided by total, high-density lipoprotein (HDL), and low-density lipoprotein (LDL) cholesterol individually, but less than that provided by age, smoking, and blood pressure. All global measures of fit improved when hsCRP was included, with likelihood-based measures demonstrating strong preference for models that include hsCRP. With use of 10-year risk categories of 0% to less than 5%, 5% to less than 10%, 10% to less than 20%, and 20% or greater, risk prediction was more accurate in models that included hsCRP, particularly for risk between 5% and 20%. Among women initially classified with risks of 5% to less than 10% and 10% to less than 20% according to the Adult Treatment Panel III covariables, 21% and 19%, respectively, were reclassified into more accurate risk categories. Although addition of hsCRP had minimal effect on the c-statistic (a measure of model discrimination) once age, smoking, and blood pressure were accounted for, the effect was nonetheless greater than that of total, LDL, or HDL cholesterol, suggesting that the c-statistic may be insensitive in evaluating risk prediction models. LIMITATIONS: Data were available only for women. CONCLUSIONS: A global risk prediction model that includes hsCRP improves cardiovascular risk classification in women, particularly among those with a 10-year risk of 5% to 20%. In models that include age, blood pressure, and smoking status, hsCRP improves prediction at least as much as do lipid measures.

Age Factors↗
PubMed2006

Administrative versus clinical data for coronary artery bypass graft surgery report cards: the view from California.

OBJECTIVE: The objective of this study was to compare the performance of a risk model for isolated coronary artery bypass graft (CABG) surgery based on administrative data with that of a clinical risk model in predicting mortality and identifying hospital performance outliers. METHODS: Clinical data records from the California CABG Mortality Reporting Program for 38,230 isolated CABG patients undergoing surgery in 2000-2001 were linked to records in the California patient discharge data (PDD) abstract. Risk factors based on administrative data that mirrored clinical risk factors were developed using the condition present at admission indicator in the PDD to separate preoperative acute conditions from complications of surgery. Using logistic regression, risk model performance across data sources was compared along with hospital risk-adjusted mortality ranks and quality ratings. RESULTS: The administrative data showed lower prevalence of risk factors when compared with the clinical data. The clinical risk model had somewhat better discrimination (C = 0.824) than the administrative model (C = 0.799). The clinical model yielded 17 outliers and the administrative model 16 with agreement on 12 hospitals' status. Performance of the administrative risk model was minimally affected by removal of information from prior admissions and removal of risk factors not confirmed in the clinical record. CONCLUSIONS: Unique properties of the California administrative data, including the ability to distinguish acute preoperative risk factors from complications of surgery, permitted construction of an administrative risk model that predicts mortality on par with most published clinical models. Despite this, the administrative model identified slightly different hospital outliers, which may indicate somewhat biased assessments of hospital patient risk.

California↗