Gerald A. Coffman

IDENTIFYING COMPLICATIONS OF CARE USING ADMINISTRATIVE DATA

The Complications Screening Program (CSP) is a method using standard hospital discharge abstract data to identify 27 potentially preventable in-hospital complications, such as post-operative pneumonia, hemorrhage, medication incidents, and wound infection. The CSP was applied to over 1.9 million adult medical/surgical cases using 1988 California discharge abstract data. Cases with complications were significantly older and more likely to die, and they had much higher average total charges and lengths of stay than other cases (P < 0.0001). For most case types, 13 chronic conditions, defined using diagnosis codes, increased the relative risks of having a complication after adjusting for patient age. Cases at larger hospitals and teaching facilities generally had higher complication rates. Logistic regression models to predict complications using demographic, administrative, clinical, and hospital characteristics variables, had modest power (C statistics = 0.64 to 0.70). The CSP requires further evaluation before using it for purposes other than research.

A Randomized Trial of Treatment Options for Alcohol-Abusing Workers

BACKGROUND Employee-assistance programs sponsored by companies or labor unions identify workers who abuse alcohol and refer them for care, often to inpatient rehabilitation programs. Yet the effectiveness of inpatient treatment, as compared with a variety of less intensive alternatives, has repeatedly been called into question. In this study, anchored in the work site, we compared the effectiveness of mandatory in-hospital treatment with that of required attendance at the meetings of a self-help group and a choice of treatment options. METHODS We randomly assigned a series of 227 workers newly identified as abusing alcohol to one of three rehabilitation regimens: compulsory inpatient treatment, compulsory attendance at Alcoholics Anonymous (AA) meetings, and a choice of options. Inpatient backup was provided if needed. The groups were compared in terms of 12 job-performance variables and 12 measures of drinking and drug use during a two-year follow-up period. RESULTS All three groups improved, and no significant differences were found among the groups in job-related outcome variables. On seven measures of drinking and drug use, however, we found significant differences at several follow-up assessments. The hospital group fared best and that assigned to AA the least well; those allowed to choose a program had intermediate outcomes. Additional inpatient treatment was required significantly more often (P less than 0.0001) by the AA group (63 percent) and the choice group (38 percent) than by subjects assigned to initial treatment in the hospital (23 percent). The differences among the groups were especially pronounced for workers who had used cocaine within six months before study entry. The estimated costs of inpatient treatment for the AA and choice groups averaged only 10 percent less than the costs for the hospital group because of their higher rates of additional treatment. CONCLUSIONS Even for employed problem drinkers who are not abusing drugs and who have no serious medical problems, an initial referral to AA alone or a choice of programs, although less costly than inpatient care, involves more risk than compulsory inpatient treatment and should be accompanied by close monitoring for signs of incipient relapse.

Predicting in-hospital mortality. A comparison of severity measurement approaches

Hospital mortality statistics derived from administrative data may not adjust adequately for patient risk on admission. Using clinical data collected from the medical record, this study compared the ability of six models to predict in-hospital death, including one model based on administrative data (age, sex, and principal and secondary diagnoses), one on admission MedisGroups score, and one on an approximation of the Acute Physiology Score (APS) from the revised Acute Physiology and Chronic Health Evaluation (APACHE II), as well as three empirically derived models` The database from 24 hospitals included 16,855 cases involving five medical conditions, with an overall in-hospital mortality rate of 15.6%. The administrative data model fit least well (R-squared values ranged from 1.9-5.5% across the five conditions). Admission MedisGroups score and the proxy APS score did better, with R-squared values ranging from 4.9% to 25.9%. Two empirical models based on small subsets of explanatory variables performed best (R-squared values ranged from 18.5-29.9%). The preceding models had the same relative performances after cross-validation using split samples. However, the high R-squared values produced by the full empirical models (using 40 or more explanatory variables) were not preserved when they were cross-validated. Most of the predictive clinical findings were general physiologic measures that were similar across conditions; only a fifth of predictors were condition-specific. Therefore, an efficient approach to risk-adjusting inhospital mortality figures may involve adding a small subset of condition-specific clinical variables to a core group of acute physiologic variables. The best predictive models employ condition-specific weighting of even the generic clinical findings.

Familial Aggregation of Low Birth Weight among Whites and Blacks in the United States

BACKGROUND Studies have shown that the birth weight of infants is correlated with the birth weights of their siblings and their mothers. We investigated whether the birth weights of mothers and index children were jointly associated with the risk of low birth weight in the siblings of the index children. METHODS We used data on the live-birth cohort of the 1988 National Maternal and Infant Health Survey. The analysis included 1691 white and 1461 black mothers, each of whom had two or more live-born, singleton children. Multiple logistic regression with generalized-estimation equations was used to assess the risk of low birth weight among an index childs siblings. Four groups were studied: that in which neither the mother nor the index child had low birth weight (group 1), that in which only the mother had low birth weight (group 2), that in which only the index child had low birth weight (group 3), and that in which both the mother and the index child had low birth weight (group 4). There was adjustment for other maternal and infant covariates. RESULTS Among white siblings in groups 1, 2, 3, and 4, 3.6, 8.3, 21.2, and 38.9 percent, respectively, had low birth weight, as compared with 8.0, 19.0, 31.1, and 57.1 percent of black siblings. When group 1 was used as the reference group, the adjusted odds ratios (and 95 percent confidence intervals) for low birth weight in groups 2, 3, and 4 were 2.5 (1.4 to 4.3), 6.8 (4.7 to 9.8), and 15.4 (9.2 to 25.5), respectively, among white siblings and 2.6 (1.8 to 3.8), 4.7 (3.5 to 6.4), and 13.9 (9.2 to 20.9) among black siblings. These associations were consistently found for birth weights below 1500 g and those ranging from 1500 to 2499 g in both races and after stratification for the mothers age, parity, education, cigarette-smoking status, and weight and height before pregnancy and the infants sex. CONCLUSIONS Although selection and recall biases cannot be excluded with certainty, our data suggest a strong familial aggregation of low birth weight among both whites and blacks in the United States.

Predicting in-hospital mortality. The importance of functional status information.

Monitoring risk-adjusted outcomes is the centerpiece of efforts to ensure health care quality. Because data collection is expensive, questions arise concerning what information is essential to adjust for risk. This investigation used retrospective analysis of existing, computerized clinical databases containing laboratory test results, information on chronic coexisting conditions, and nursing evaluations of functional status to predict in-hospital mortality. We studied persons admitted to one tertiary teaching hospital between 1987 and 1992 for cerebrovascular disease or pneumonia. Predictive models for each of the conditions were developed using logistic regression; the results were validated with split samples. We compared the predictive value of the nursing functional status assessments and the clinical laboratory data. For each study condition, the functional status data had as much prognostic information as the laboratory data. Specifically, a nurses report that a patient required total assistance for bathing was the best single predictor of in-hospital mortality in the models for patients with either cerebrovascular disease or pneumonia. If hospitals admit patients with different levels of functional impairment, it is important to account for these differences before comparing outcomes across facilities. Assessments of functional status are a simple, inexpensive measure that may have considerable value.

Weight, Blood Pressure, and Dietary Benefits After 12 Months of a Web-based Nutrition Education Program (DASH for Health): Longitudinal Observational Study

Background The dietary habits of Americans are creating serious health concerns, including obesity, hypertension, diabetes, cardiovascular disease, and even some types of cancer. While considerable attention has been focused on calorie reduction and weight loss, approaches are needed that will not only help the population reduce calorie intake but also consume the type of healthy, well-balanced diet that would prevent this array of medical complications. Objective To design an Internet-based nutrition education program and to explore its effect on weight, blood pressure, and eating habits after 12 months of participation. Methods We designed the DASH for Health program to provide weekly articles about healthy nutrition via the Internet. Dietary advice was based on the DASH diet (Dietary Approaches to Stop Hypertension). The program was offered as a free benefit to the employees of EMC Corporation, and 2834 employees and spouses enrolled. Enrollees voluntarily entered information about themselves on the website (food intake), and we used these self-entered data to determine if the program had any effect. Analyses were based upon the change in weight, blood pressure, and food intake between the baseline period (before the DASH program began) and the 12th month. To be included in an outcome, a subject had to have provided both a baseline and 12th-month entry. Results After 12 months, 735 of 2834 original enrollees (26%) were still actively using the program. For subjects who were overweight/obese (body mass index > 25; n = 151), weight change at 12 months was -4.2 lbs (95% CI: -2.2, -6.2; P < .001). For subjects with hypertension or prehypertension at baseline (n = 62), systolic blood pressure fell 6.8 mmHg at 12 months (CI: -2.6, -11.0; P < .001; n = 62). Diastolic pressure fell 2.1 mmHg (P = .16). Based upon self-entered food surveys, enrollees (n = 181) at 12 months were eating significantly more fruits, more vegetables, and fewer grain products. They also reduced consumption of carbonated beverages. Enrollees who had visited the website more often tended to have greater blood pressure and weight loss effect, suggesting that use of the DASH for Health program was at least partially responsible for the benefits we observed. Conclusions We have found that continued use of a nutrition education program delivered totally via the Internet, with no person-to-person contact with health professionals, is associated with significant weight loss, blood pressure lowering, and dietary improvements after 12 months. Effective programs like DASH for Health, delivered via the Internet, can provide benefit to large numbers of subjects at low cost and may help address the nutritional public health crisis.

Costs of Potential Complications of Care for Major Surgery Patients

We examined computerized hospital discharge ab stract data from 372,680 major surgery patients admit ted to 404 California acute care hospitals in 1988 to identify potential complications of care. At least one potential in-hospital complication occurred for 10.8% of patients. Patients with complications were older and more likely to die in-hospital (9.4% compared to 1.0%, P < 0.0001). On average, patients with complications had longer stays (13.5 versus 5.4 days, p < 0.0001) and higher total charges (

Neuropsychiatric Assessment of Liver Transplantation Candidates: Delirium and other Psychiatric Disorders

30,896 versus

Reliability of DSM-III vs. DSM-II in Child Psychopathology

9,239,p < 0.0001). After adjusting for demographic, clinical, and hospital factors, patients with potential complications averaged

Admission and mid-stay MedisGroups scores as predictors of death within 30 days of hospital admission.

16,023 higher total hospital charges than uncompli cated patients. Complications were associated with 96.6% (95% confidence interval = 95.2%, 98.0%) higher hospital charges after adjusting for these factors. Across all patients, complications were related to over