Biomedical subjects

S R Lipsitz

Publications and source records attributed to S R Lipsitz.

47 records · Page 3Linked to original sources

PubMed1994

Weighted least squares analysis of repeated categorical measurements with outcomes subject to nonresponse.

In this paper, we describe a two-step weighted least squares method for analyzing repeated categorical outcomes when some individuals are not observed at all times of follow-up. Other weighted least squares methods for analyzing repeated measures data with missing responses have previously been proposed by Koch, Imrey, and Reinfurt (1972, Biometrics 28, 663-692) and Woolson and Clarke (1984, Journal of the Royal Statistical Society, Series A 147, 87-99). These methods give consistent estimators if the responses are missing completely at random, as discussed in Rubin (1976, Biometrika 63, 581-592). We propose a two-step method that will give consistent results under the weaker condition of missing at random, and compare it with the other two methods.

Adolescent↗

PubMed1994

Performance of generalized estimating equations in practical situations.

Moment methods for analyzing repeated binary responses have been proposed by Liang and Zeger (1986, Biometrika 73, 13-22), and extended by Prentice (1988, Biometrics 44, 1033-1048). In their generalized estimating equations (GEE), both Liang and Zeger (1986) and Prentice (1988) estimate the parameters associated with the expected value of an individual's vector of binary responses as well as the correlations between pairs of binary responses. In this paper, we discuss one-step estimators, i.e., estimators obtained from one step of the generalized estimating equations, and compare their performance to that of the fully iterated estimators in small samples. In simulations, we find the performance of the one-step estimator to be qualitatively similar to that of the fully iterated estimator. When the sample size is small and the association between binary responses is high, we recommend using the one-step estimator to circumvent convergence problems associated with the fully iterated GEE algorithm. Furthermore, we find the GEE methods to be more efficient than ordinary logistic regression with variance correction for estimating the effect of a time-varying covariate.

Air Pollution↗

PubMed1993

Do the poor sue more? A case-control study of malpractice claims and socioeconomic status.

OBJECTIVE: To evaluate whether socioeconomic status is associated with risk of malpractice claims, particularly among those who have suffered medical injury. DESIGN: Case-control study. SETTING: Fifty-one hospitals in New York State. METHODS: The presence and severity of medical injury, defined as disability at the time of discharge or prolongation of the hospitalization caused by medical treatment as opposed to the disease process, were assessed through review of approximately 31,000 hospital records in New York in 1984. These sampled records were then linked to formal malpractice claims. To estimate the risk of malpractice claims by age, gender, race, insurance status, and income, we conducted a case-control study of claimant cases matched with nonclaimant controls. The cases were all those patients who filed malpractice claims referring to alleged malpractice during a sampled hospitalization. Physician reviewers had previously judged the level of disability that resulted from the medical injury. Claimants (n = 51) were each matched with five nonclaimant controls on the basis of injury. Noninjured cases were matched with noninjured controls and injured cases were matched with injured controls. RESULTS: We found that poor patients (odds ratio [OR], 0.2; 95% confidence interval [CI], 0.03 to 0.8) and uninsured patients (OR, 0.1; 95% CI, 0.005 to 0.9) were significantly less likely to file malpractice claims, after controlling for the severity of medical injury. Among patients who suffered medical injury, the elderly (OR, 0.2; 95% CI, 0.03 to 0.9) were also less likely to file claims. Gender and race were not independently associated with risk of malpractice claims. CONCLUSIONS: Poor and uninsured patients are significantly less likely to sue for malpractice, even after controlling for the presence of medical injury. Fear of malpractice risk should not be a significant factor in the decision to serve the poor. Tort reforms that would protect physicians who serve the medically indigent from malpractice suits may not be warranted.

Case-Control Studies↗

PubMed1993

The effect of hospital financial characteristics on quality of care.

OBJECTIVE: To assess the relationship among hospital financial characteristics, patient payer mix, and the incidence of negligent medical injuries. DESIGN: Retrospective medical record review linked to hospital financial reports. SETTING: Acute care hospitals in New York State in 1984. POPULATION: Stratified, random sample of 30,195 medical records from 51 acute care hospitals. MAIN OUTCOME MEASURES: Hospital rates of medical injury and substandard care were developed from reviews of 30,195 medical records at 51 acute care hospitals in New York in 1984. Hospital-level variables representing financial status, hospital staffing, and the proportion of self-pay and Medicaid hospital discharges were compiled from a variety of secondary sources. RESULTS: The likelihood of negligent medical injury was highest in those hospitals with the lowest inpatient operating costs per hospital discharge (odds ratio, 2.8; 95% confidence interval, 1.5 to 5.5). The effect of low inpatient operating costs was marked among hospitals in financial distress, many of which served indigent populations. CONCLUSIONS: Patients admitted to hospitals that are unable to expend sufficient resources on patient care may be at higher risk of substandard care. Further study of the effect of hospital financial status on quality of care appears to be warranted.

Economics, Hospital↗

PubMed1992

Socioeconomic status and risk for substandard medical care.

OBJECTIVE: To assess whether the socioeconomic status of the patient was associated with the risk of adverse events, defined as medical injuries caused by medical management, and the proportion of these events that resulted from substandard care. SETTING: 51 hospitals in New York State. METHODS: Rates of medical injury and substandard care by gender, race, income, and payer status were developed from reviews of 30,195 medical records in New York in 1984. We evaluated these socioeconomic parameters in a multivariate analysis, while controlling for hospital-level factors. RESULTS: We found that uninsured patients (odds ratio, 2.35; 95% confidence interval, 1.40 to 3.95) were at greater risk for substandard care. The characteristics of the hospitals to which patients were admitted did not affect this result. Race, gender, and income were not independently associated with risk for medical injury or substandard care in multivariate analyses. CONCLUSION: Our findings suggest that the uninsured are at greater risk for suffering medical injury due to substandard medical care.

Black or African American↗

PubMed1992

Methods for estimating the parameters of a linear model for ordered categorical data.

In many empirical analyses, the response of interest is categorical with an ordinal scale attached. Many investigators prefer to formulate a linear model, assigning scores to each category of the ordinal response and treating it as continuous. When the covariates are categorical, Haber (1985, Computational Statistics and Data Analysis 3, 1-10) has developed a method to obtain maximum likelihood (ML) estimates of the parameters of the linear model using Lagrange multipliers. However, when the covariates are continuous, the only method we found in the literature is ordinary least squares (OLS), performed under the assumption of homogeneous variance. The OLS estimates are unbiased and consistent but, since variance homogeneity is violated, the OLS estimates of variance can be biased and may not be consistent. We discuss a variance estimate (White, 1980, Econometrica 48, 817-838) that is consistent for the true variance of the OLS parameter estimates. The possible bias encountered by using the naive OLS variance estimate is discussed. An estimated generalized least squares (EGLS) estimator is proposed and its efficiency relative to OLS is discussed. Finally, an empirical comparison of OLS, EGLS, and ML estimators is made.

Abnormalities, Drug-Induced↗

PubMed1991

Hospital characteristics associated with adverse events and substandard care.

To explore the epidemiology of adverse events (AEs), which were defined as injuries due to medical treatment, and that subset of AEs caused by negligence, we studied interhospital variation in these outcomes in a sample of 31,000 medical records drawn from a random selection of 51 hospitals in New York in 1984. We found a substantial variation in both AE rates (0.2% to 7.9%; mean, 3.2%) and the percentage of AEs due to negligence (1% to 60%; mean, 24.9%) among hospitals. Univariate analyses of AEs revealed that primary teaching institutions had significantly higher rates (4.1%) and rural hospitals had significantly lower ones (1.0%). The percentage of AEs due to negligence was lower in primary teaching (10.7%) and for-profit (9.5%) hospitals and was significantly higher in hospitals with predominantly (greater than 80%) minority patients who had been discharged (37%). These findings were corroborated by multivariate analysis. Our results suggest that AEs and negligence are not randomly distributed and that certain types of hospitals have significantly higher rates of injuries due to substandard care. These observations may represent an important improvement on existing measures of quality because they take into account the fact that some hospitals' populations may be at risk of suffering a poor outcome.

Hospital Records↗

PubMed1990

Maximum likelihood regression methods for paired binary data.

We discuss maximum likelihood methods for analysing binary responses measured at two times, such as in a cross-over design. We construct a 2 x 2 table for each individual with cell probabilities corresponding to the cross-classification of the responses at the two times; the underlying likelihood for each individual is multinomial with four cells. The three dimensional parameter space of the multinomial distribution is completely specified by the two marginal probabilities of success of the 2 x 2 table and an association parameter between the binary responses at the two times. We examine a logistic model for the marginal probabilities of the 2 x 2 table for individual i; the association parameters we consider are either the correlation coefficient, the odds ratio or the relative risk. Simulations show that the parameter estimates for the logistic regression model for the marginal probabilities are not very sensitive to the parameters used to describe the association between the binary responses at the two times. Thus, we suggest choosing the measure of association for ease of interpretation.

Humans↗

PubMed1990

Analyzing correlated binary data using SAS.

We discuss methods for analyzing repeated binary measurements on the same individual. In spite of the fact that the repeated measurements on the same individual are correlated, the ordinary logistic regression maximum likelihood estimates (which assume that the repeated measures are independent) are consistent and asymptotically normal (K. Y. Liang and S. L. Zeger, Biometrika 73, 13 (1986]. However, the inverse of the estimated information matrix (assuming independence) can give inconsistent estimates of the asymptotic variance of estimated parameters. We describe how to obtain the logistic regression estimates, as well as a consistent estimate of their covariance matrix, in SAS, with minimal matrix manipulations.

Analysis of Variance↗

PubMedDate not supplied

Non-ignorable missing covariates in generalized linear models.

We propose a likelihood method for estimating parameters in generalized linear models with missing covariates and a non-ignorable missing data mechanism. In this paper, we focus on one missing covariate. We use a logistic model for the probability that the covariate is missing, and allow this probability to depend on the incomplete covariate. We allow the covariates, including the incomplete covariate, to be either categorical or continuous. We propose an EM algorithm in this case. For a missing categorical covariate, we derive a closed form expression for the E- and M-steps of the EM algorithm for obtaining the maximum likelihood estimates (MLEs). For a missing continuous covariate, we use a Monte Carlo version of the EM algorithm to obtain the MLEs via the Gibbs sampler. The methodology is illustrated using an example from a breast cancer clinical trial in which time to disease progression is the outcome, and the incomplete covariate is a quality of life physical well-being score taken after the start of therapy. This score may be missing because the patients are sicker, so this covariate could be non-ignorably missing.

Algorithms↗

PubMedDate not supplied

Missing data perspectives of the fluvoxamine data set: a review.

Fitting models to incomplete categorical data requires more care than fitting models to the complete data counterparts, not only in the setting of missing data that are non-randomly missing, but even in the familiar missing at random setting. Various aspects of this point of view have been considered in the literature. We review it using data from a multi-centre trial on the relief of psychiatric symptoms. First, it is shown how the usual expected information matrix (referred to as naive information) is biased even under a missing at random mechanism. Second, issues that arise under non-random missingness assumptions are illustrated. It is argued that at least some of these problems can be avoided using contextual information.

Data Interpretation, Statistical↗