There are two broad classes of models used to address the econometric problems caused by skewness in data commonly encountered in health care applications: (1) transformation to deal with skewness (e.g., ordinary least square (OLS) on ln(y)); and (2) alternative weighting approaches based on exponential conditional models (ECM) and generalized linear model (GLM) approaches. In this paper, we encompass these two classes of models using the three parameter generalized Gamma (GGM) distribution, which includes several of the standard alternatives as special cases-OLS with a normal error, OLS for the log-normal, the standard Gamma and exponential with a log link, and the Weibull. Using simulation methods, we find the tests of identifying distributions to be robust. The GGM also provides a potentially more robust alternative estimator to the standard alternatives. An example using inpatient expenditures is also analyzed.

AddictionVolume 100, Issue 9 p. 1383-1384 The Price of Smoking WILLARD G. MANNING, WILLARD G. MANNING Harris School of Public Policy Studies, The University of Chicago, Chicago, Illinois, USASearch for more papers by this author WILLARD G. MANNING, WILLARD G. MANNING Harris School of Public Policy Studies, The University of Chicago, Chicago, Illinois, USASearch for more papers by this author First published: 25 August 2005 https://doi.org/10.1111/j.1360-0443.2005.01276.xRead the full textAboutPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinked InRedditWechat Volume100, Issue9September 2005Pages 1383-1384 RelatedInformation

Health economists often use log models (based on OLS or generalized linear models) to deal with skewed outcomes such as those found in health expenditures and inpatient length of stay. Some recent studies have employed Cox proportional hazard regression as a less parametric alternative to OLS and GLM models, even when there was no need to correct for censoring. This study examines how well the alternative estimators behave econometrically in terms of bias when the data are skewed to the right. Specifically we provide evidence on the performance of the Cox model under a variety of data generating mechanisms and compare it to the estimators studied recently in Manning and Mullahy (2001). No single alternative is best under all of the conditions examined here. However, the gamma regression model with a log link seems to be more robust to alternative data generating mechanisms than either OLS on ln(y) or Cox proportional hazards regression. We find that the proportional hazard assumption is an essential requirement to obtain consistent estimate of the E(y|x) using the Cox model.

Background: Hospitalists may decrease costs and improve outcomes in hospitalized patients, but existing evidence is limited and has not identified mechanisms for such effects.Objective: To study the costs and outcomes for patients on an academic general medicine service assigned to teams led by hospitalists and non hospitalists.Design: Cohort study.Setting: Academic general medicine service.Patients: 6511 patients admitted to the hospital from July 1997 through June 1999.Intervention: All patients admitted every fourth day were assigned to 1 of 2 hospitalists caring for inpatients 6 months each year or 1 of 58 nonhospitalists caring for inpatients 1 to 2 months each year.Measurements: Length of stay; inpatient costs; and 30-, 60, and 365-day mortality.Results: Patients assigned to hospitalists (24.8%) and nonhospitalists (75.2%) did not differ in age, race, sex, diagnosis mix, or Charlson index score. In year 1, average adjusted length of stay was 0.29 day shorter for patients cared for by hospitalists than by nonhospitalists (95% Cl, -0.66 to 0.06 day; P = 0.06); in year 2, average adjusted length of stay was 0.49 day shorter for patients cared for by hospitalists (Cl, -0.79 to -0.15 day; P = 0.01). Average adjusted costs were not significantly reduced for hospitalists compared with nonhospitalists in year 1 but were reduced by $782 in year 2 (Cl, -$1313 to -$187; P = 0.01). When years 1 and 2 were combined or when year 1 was analyzed alone, 30-day mortality was not significantly different for hospitalists and non hospitalists; however, 30-day mortality was 4.2% for hospitalists compared with 6.0% for nonhospitalists in year 2 (Cl for difference, 1.8 percentage points [-3.6 to -0.1 percentage points]; P = 0.04) and the adjusted relative risk was 0.65 (Cl, 0.44 to 0.96; P = 0.03). In multivariate analyses, resource use decreased with the physician's cumulative experience in caring for a patient's primary diagnosis. Mortality showed a similar pattern.Conclusions: Hospitalist care was associated with lower costs and short-term mortality in the second but not the first year of hospitalists' experience. Disease-specific physician experience may reduce resource use and improve patient outcomes; in addition, it may be an important determinant of the effectiveness of hospitalists.

There is a potential bias in cross-sectional estimates of the effects of cigarette prices on cigarette consumption. States with the strongest antismoking sentiment will likely have the highest cigarette taxes, which result in the highest prices. Some of the lower consumption of cigarettes in high-tax states will result from such sentiments, rather than from higher taxes, so the estimated effect of cigarette taxes on consumption will be overstated. This study corrects for such bias, employing panel data for U.S. states from 1960 to 1990. We find that controlling for this bias reduces the estimated consumer response to cigarette price change by 40-50 percent.

BACKGROUNDExpanding access to high-quality depression treatment will depend on the balance of incremental benefits and costs. We examine the incremental cost-effectiveness of an organized depression management program for high utilizers of medical care.METHODSComputerized records at 3 health maintenance organizations were used to identify adult patients with outpatient medical visit rates above the 85th percentile for 2 consecutive years. A 2-step screening process identified patients with current depressive disorders, who were not in active treatment. Eligible patients were randomly assigned to continued usual care (n = 189) or to an organized depression management program (n = 218). The program included patient education, antidepressant pharmacotherapy initiated in primary care (when appropriate), systematic telephone monitoring of adherence and outcomes, and psychiatric consultation as needed. Clinical outcomes (assessed using the Hamilton Depression Rating Scale on 4 occasions throughout 12 months) were converted to measures of "depression-free days." Health services utilization and costs were estimated using health plan-standardized claims.RESULTSThe intervention program led to an adjusted increase of 47.7 depression-free days throughout 12 months (95% confidence interval [CI], 28.2-67.8 days). Estimated cost increases were $1008 per year (95% CI, $534-$1383) for outpatient health services, $1974 per year for total health services costs (95% CI, $848-$3171), and $2475 for health services plus time-in-treatment costs (95% CI, $880-$4138). Including total health services and time-in-treatment costs, estimated incremental cost per depression-free day was $51.84 (95% CI, $17.37-$108.47).CONCLUSIONSAmong high utilizers of medical care, systematic identification and treatment of depression produce significant and sustained improvements in clinical outcomes as well as significant increases in health services costs.

Health economists often use log models to deal with skewed outcomes, such as health utilization or health expenditures. The literature provides a number of alternative estimation approaches for log models, including ordinary least-squares on ln(y) and generalized linear models. This study examines how well the alternative estimators behave econometrically in terms of bias and precision when the data are skewed or have other common data problems (heteroscedasticity, heavy tails, etc.). No single alternative is best under all conditions examined. The paper provides a straightforward algorithm for choosing among the alternative estimators. Even if the estimators considered are consistent, there can be major losses in precision from selecting a less appropriate estimator.

OBJECTIVE To analyze the effect of multiple-source drug entry on price competition after patent expiration in the pharmaceutical industry. DATA SOURCES Originators and their multiple-source drugs selected from the 35 chemical entities whose patents expired from 1984 through 1987. Data were obtained from various primary and secondary sources for the patents' expiration dates, sales volume and units sold, and characteristics of drugs in the sample markets. STUDY DESIGN The study was designed to determine significant factors using the study model developed under the assumption that the off-patented market is an imperfectly segmented market. PRINCIPAL FINDINGS After patent expiration, the originators' prices continued to increase, while the price of multiple-source drugs decreased significantly over time. By the fourth year after patent expiration, originators' sales had decreased 12 percent in dollars and 30 percent in quantity. Multiple-source drugs increased their sales twofold in dollars and threefold in quantity, and possessed about one-fourth (in dollars) and half (in quantity) of the total market three years after entry. CONCLUSION After patent expiration, multiple-source drugs compete largely with other multiple-source drugs in the price-sensitive sector, but indirectly with the originator in the price-insensitive sector. Originators have first-mover advantages, and therefore have a market that is less price sensitive after multiple-source drugs enter. On the other hand, multiple-source drugs target the price-sensitive sector, using their lower-priced drugs. This trend may indicate that the off-patented market is imperfectly segmented between the price-sensitive and insensitive sector. Consumers as a whole can gain from the entry of multiple-source drugs because the average price of the market continually declines after patent expiration.

The assignment of costs in a cost-effectiveness analysis is a complex and disputed issue. The Panel on Cost-Effectiveness in Health and Medicine was convened to discuss standards that could be applied across a range of areas of cost-effectiveness. Additionally, the Panel had a mandate to resolve some controversial issues about the practice of cost-effectiveness that created difficulty in making comparisons across studies. The Panel proposed these guidelines: (1) Do at least some of the analysis from a social perspective; (2) Assign values to resources that reflect their opportunity costs; (3) Avoid zero counting of resources; (4) Avoid double counting of resources; (5) Make analyses only as exacting as necessary in a study. Difficulties in data collection were discussed. Among other questions considered by the panel were how to assign a value to the patient's time and which productivity costs to include in a cost-effectiveness analysis. With tools and suggestions from the Panel on Cost-Effectiveness in Health and Medicine, the cost analyst can report costs accurately and provide accurate comparisons of cost performance across states, trial modalities, or diseases.

Context. Physician profiling is widely used by many health care systems, but little is known about the reliability of commonly used profiling systems.Objectives. To determine the reliability of a set of physician performance measures for diabetes care,one of the most common conditions in medical practice, and to examine whether physicians could substantially improve their profiles by preferential patient selection.Design and Setting. Cohort study performed from 1990 to 1993 at 3 geographically and organizationally diverse sites, including a large staff-model health maintenance organization, an urban university teaching clinic, and a group of private-practice physicians in an urban area.Participants. A total of 3642 patients with type 2 diabetes cared for by 232 different physicians.Main Outcome Measures. Physician profiles for their patients' hospitalization and clinic visit rates, total laboratory resource utilization rate and level of glycemic control by average hemoglobin A(1c) level with and without detailed case-mix adjustment.Results. For profiles based on hospitalization rates, visit rates, laboratory utilization rates, and glycemic control, 4% or less of the overall variance was attributable to differences in physician practice and the reliability of the median physician's case-mix-adjusted profile was never better than 0.40. At this low level of physician effect, a physician would need to have more than 100 patients with diabetes in a panel for profiles to have a reliability of 0.80 or better (while more than 90% of all primary care physicians at the health maintenance organization had fewer than 60 patients with diabetes). For profiles of glycemic control, high outlier physicians could dramatically improve their physician profile simply by pruning from their panel the 1 to 3 patients with the highest hemoglobin A(1c) levels during the prior year. This advantage from gaming could not be prevented by even detailed case-mix adjustment.Conclusions. Physician "report cards" for diabetes,one of the highest-prevalence conditions in medical practice, were unable to detect reliably true practice differences within the 3 sites studied. Use of individual physician profiles may foster an environment in which physicians can most easily avoid being penalized by avoiding or deselecting patients with high prior cost, poor adherence;or response to treatments.

STUDY QUESTIONSTo determine factors that distinguish effective rural hospital consortia from ineffective ones in terms of their ability to improve members' financial performance. Two questions in particular were addressed: (1) Do large consortia have a greater collective impact on their members? (2) Does a consortium's economic environment determine the degree of collective impact on members?DATA SOURCES AND STUDY SETTINGBased on the hospital survey conducted during February 1992 by the Robert Wood Johnson Hospital-Based Rural Health Care project of rural hospital consortia. The survey data were augmented with data from Medicare Cost Reports (1985-1991), AHA Annual Surveys (1985-1991), and other secondary data.STUDY DESIGNDependent variables were total operating profit, cost per adjusted admission, and revenue per adjusted admission. Control variables included degree of group formalization, degree of inequality of resources among members (group asymmetry), affiliation with other consortium group(s), individual economic environment, common hospital characteristics (bed size, ownership type, system affiliation, case mix, etc.), year (1985-1991), and census region dummies.PRINCIPAL FINDINGSAll dependent variables have a curvilinear association with group size. The optimum group size is somewhere in the neighborhood of 45. This reveals the benefits of collective action (i.e., scale economies and/or synergy effects) and the issue of complexity as group size increases. Across analyses, no strong evidence exists of group economic environment impacts, and the environmental influences come mainly from the local economy rather than from the group economy.CONCLUSIONThere may be some success stories of collaboration among hospitals in consortia, and consortium effects vary across different collaborations.RELEVANCE/IMPACTWhen studying consortia, it makes sense to develop a typology of groups based on some performance indicators. The results of this study imply that government, rural communities, and consortium staff and steering committees should forge the consortium concept by expanding membership in order to gain greater financial benefits for individual hospitals.

Little is known about how well individuals are compensated for injuries. This study uses data from a 1989 survey to estimate both the lifetime costs and compensation for injuries. Our findings show that about 55% of the cost is compensated by public or private programs. Compensation rates are lower for disabling injuries and those of long duration. The results also suggest that compensation system reforms that would place stricter limits on maximum compensation might not be a distributionally fair solution. The reasons are that costs are highly skewed, and the share of costs recovered by compensation programs is currently lowest for injuries that are long term, disabling, and the most expensive.

Capitation reduced Medicaid costs but had limited effects on most measures of process and outcome. Clients under capitation with the poorest mental health at baseline performed more poorly over time on some measures.

OBJECTIVE:To prospectively compare inpatient and outpatient utilization rates between prepaid (PPD) and fee-for-service (FFS) insurance coverage for patients with chronic disease.DATA SOURCE/STUDY SETTING:Data from the Medical Outcomes Study, a longitudinal observational study of chronic disease patients conducted in Boston, Chicago, and Los Angeles.STUDY DESIGN:A four-year prospective study of resource utilization among 1,681 patients under treatment for hypertension, diabetes, myocardial infarction, or congestive heart failure in the practices of 367 clinicians.DATA COLLECTION/EXTRACTION METHODS:Insurance payment system (PPD or FFS), hospitalizations, and office visits were obtained from patient reports. Disease and severity indicators, sociodemographics, and self-reported functional status were used to adjust for patient mix and to compute expected utilization rates.PRINCIPAL FINDINGS:Compared to FFS, PPD patients had 31 percent fewer observed hospitalizations before adjustment for patient differences (p = .005) and 15 percent fewer hospitalizations than expected after adjustment (p = .078). The observed rate of FFS hospitalizations exceeded the expected rate by 9 percent. These results are not explained by system differences in patient mix or trends in hospital use over four years. Half of the PPD/FFS difference in hospitalization rate is due to intrinsic characteristics of the payment system itself.CONCLUSIONS:PPD patients with chronic medical conditions followed prospectively over four years, after extensive patient-mix adjustment, had 15 percent fewer hospitalizations than their FFS counterparts owing to differences intrinsic to the insurance reimbursement system.

Standard methodologies that are used to identify households vulnerable to future episodes of poverty rely on distributional assumptions that may lead to classification errors. This paper shows how quantile models improve the identification of vulnerable households by relaxing these distributional assumptions. Quantile models are robust to outliers and classical measurement error, and easy to implement which allows easy adoption by policymakers. Applying this quantile strategy to data from Uganda to illustrate its superiority to standard approaches, I find that it more accurately identifies the future poor among the general population. The accuracy is highlighted in the fact that more than 2 in 3 households identified as vulnerable using the quantile strategy became/remained poor within 1–2 years, compared with less than 1 in 2 households using standard empirical strategies. Overall, this study points to gains that researchers and policy practitioners can make by relaxing distributional assumptions when identifying the vulnerable.

As part of an evaluation of the Utah Prepaid Mental Health Plan, the Process of Care Review Form was developed to assess the quality of the process of psychiatric care provided by Utah's community mental health centers (CMHCs) to clients with schizophrenia. This article briefly reviews issues in. the measurement of quality of care and describes the development and implementation of the form. The 67-item form was designed for use by a trained abstracter to gather data from CMHC medical records in six areas: general management of the patient, medication management, medical management, social support, psychiatric hospitalization, and psychiatric assessment. A 59-item version of the form that omits the section on psychiatric assessment has been used in three waves of data collection to document data spanning five years (1990-1994) in the evaluation of the process of psychiatric care in the Utah plan. It is currently being used longitudinally to examine differences between Utah CMHCs receiving capitated payments and those paid on a fee-for-service basis by Medicaid.

OBJECTIVE We tested the hypothesis that level of glycemic control is related to medical care costs in adults with diabetes. RESEARCH DESIGN AND METHODS Regression analysis was used to estimate the relationship between glycemic control and medical care charges for 3,017 adults with diabetes who were continuously enrolled in a large health maintenance organization (HMO) over a 4-year period. Diagnosis of diabetes was ascertained from diagnostic and pharmaceutical databases using a method with an estimated sensitivity of 0.91 and an estimated specificity of 0.99. Charges for care included defined outpatient and inpatient services. Patients who disenrolled or who died during the 4-year period were excluded from the main analysis. RESULTS Charges for medical care for patients with diabetes from 1993 to 1995 were closely related to HbA1c level in 1992 before and after adjustment for age, sex, coronary heart disease, and hypertension. Standardized 3-year estimates of charges ranged from $10,439 for patients without comorbid conditions to $44,417 for those with heart disease and hypertension. Medical care charges increased significantly for every 1% increase above HbA1c of 7%. For a person with an HbA1c value of 6%, successive 1% increases in HbA1c resulted in cumulative increases in charges of ∼ 4, 10, 20, and 30%. The increase in charges accelerated as the HbA1c value increased. For patients with diabetes only, or with diabetes plus other chronic conditions, the rate of increase in charges with HbA1c was consistent. CONCLUSIONS HbA1c provides useful information to providers and patients regarding both health status and future medical care charges. Economic data suggest that clinicians should assign high importance to low HbA1c results and aggressively maintain the HbA1c status of patients who have low HbA1c values. For economic as well as clinical reasons, it may be beneficial to lower HbA1c when it is > 8% and to reduce cardiovascular risk factors. The medical charge data suggest that investment in clinical systems to improve diabetes care may benefit both payers and patients.

Context.-Although experimental studies show that insulin therapy can be safe and efficacious in improving glycemic control in type 2 diabetes under optimal conditions (ie, using patient volunteers with close monitoring under strict study protocols), little is known about its effectiveness, complication rates, and associated resource utilization in actual clinical practice.Design.-Cohort study.Setting.-Large staff-model health maintenance organization.Participants.-A total of 8668 patients with type 2 diabetes cared for by generalist physicians from 1990 through 1993.Outcome Measures.-Resource use (hospitalizations, outpatient visits, laboratory testing, and home glucose monitoring) and glycemic control were evaluated using combined clinical, survey, and administrative information systems data. Detailed clinical case-mix data, including a newly validated case-mix method, the Total Illness Burden Index, were collected on a subsample of 1738 patients.Results.-Among patients starting insulin therapy, hemoglobin A(1c) (HbA(1c)) decreased by 0.9 percentage point (95% confidence interval, 0.7-1.0) at 1 year compared with those receiving stable medication regimens; however, 2 years after starting insulin therapy, 60% still had HbA(1c) levels of 8% or greater. There was no evidence that some primary care physicians achieved better results than other primary care physicians when starting insulin therapy in their patients. Patients with the poorest baseline glycemic control achieved substantially greater HbA(1c), reductions; those with a baseline HbA(1c) level of 13% had a 3-fold greater decline in HbA(1c) than those whose baseline HbA(1c) level was 9%, For a subset of all patients for whom detailed clinical case-mix data were obtained, those taking insulin had higher resource use than those taking sulfonylureas, independent of illness severity. After adjusting for age, sex, race, socioeconomic status, disease duration, and severity of diabetes and comorbidities, insulin: users had slightly more laboratory tests performed, 2.4 more outpatient visits per year, and almost 300 more fingersticks for home glucose testing per year compared with sulfonylurea users (all P<.01). Although 15% of patients receiving insulin therapy reported weekly symptoms of hypoglycemia, insulin therapy was not associated with an increase in emergency department visits (after case-mix adjustment) and resulted in only 0.5 hypoglycemia-related hospitalizations per 100 patient-years.Conclusions.-For patients with type 2 diabetes who were cared for by generalist physicians, starting insulin therapy was generally safe and effective in achieving moderate glycemic control in patients who initially had poor glycemic control. However, insulin therapy was associated with increases in resource use and was rarely effective in achieving tight glycemic control, even for those with moderate control.