Economic evaluations of HIV treatment and health research with people diagnosed with HIV infection and co-occurring mental health and substance use disorders.

This paper describes the research challenges involved in measuring costs in economic evaluations of patients who are coping simultaneously with HIV/AIDS and co-occurring mental health and substance abuse disorders-especially in multi-site studies. We describe the general issues that arise in measuring costs for this population and suggest some operational solutions for their resolution, drawing from our experience in a recent multi-site health services research study focused on this population. We show that while reliance on patient self-report data may be unavoidable to provide a common denominator in multi-site studies, there are also some practical ways of improving the accuracy of such data and the cost estimates that result from them. We also provide readers with a means for securing the data collection instruments developed for the cost component of this study in the hope that these may serve as templates for researchers doing similar work.

Information superhighway or billboards by the roadside? An analysis of hospital web sites.

OBJECTIVES: To determine the prevalence of hospital web sites, the types of information provided within these sites, and the relationship of information to institutional characteristics. DESIGN: Online search of hospital web sites over a 6-week period in late 1999. Web sites were abstracted for content. Bivariate comparisons were made of hospital profit status and ownership or operation by a multihospital network. PARTICIPANTS: California acute care hospitals and their web sites. MAIN OUTCOME MEASURES: Operation of web sites and web site content. RESULTS: Among 390 California hospitals, 242 (62%) had easily identifiable web sites, 59 (15%) had no web sites, and 89 (23%) had sites identified only after telephone follow-up. Hospitals without sites were more likely not-for-profit, small, rural, or unaffiliated. The presentation of information was inconsistent, although most (93%) provided basic contact information. Many hospitals provided health content information (70%) or mentioned health classes (65%), but few guaranteed the quality of this information. Patient care features (online health profiles, risk identification, e-mail) were infrequent (13%) and rudimentary. Product advertising was frequent (54%) but was often nonhealth-related and unobtrusive. Of the 36% of hospitals that reported information on quality, few of the designated measures were valid and reliable measures of quality. Overall, 21% of hospitals reported accreditation (Joint Commission on Accreditation of Healthcare Organizations) status, and for-profit hospital web sites were more likely to report this accreditation. CONCLUSION: [corrected] Consumers should be aware of current limitations in using information on hospital web sites. In the future, hospitals may better realize the potential of web sites for the delivery of health care information and patient care.

Risk adjustment alternatives in paying for behavioral health care under Medicaid.

OBJECTIVE: To compare the performance of various risk adjustment models in behavioral health applications such as setting mental health and substance abuse (MH/SA) capitation payments or overall capitation payments for populations including MH/SA users. DATA SOURCES/STUDY DESIGN: The 1991-93 administrative data from the Michigan Medicaid program were used. We compared mean absolute prediction error for several risk adjustment models and simulated the profits and losses that behavioral health care carve outs and integrated health plans would experience under risk adjustment if they enrolled beneficiaries with a history of MH/SA problems. Models included basic demographic adjustment, Adjusted Diagnostic Groups, Hierarchical Condition Categories, and specifications designed for behavioral health. PRINCIPAL FINDINGS: Differences in predictive ability among risk adjustment models were small and generally insignificant. Specifications based on relatively few MH/SA diagnostic categories did as well as or better than models controlling for additional variables such as medical diagnoses at predicting MH/SA expenditures among adults. Simulation analyses revealed that among both adults and minors considerable scope remained for behavioral health care carve outs to make profits or losses after risk adjustment based on differential enrollment of severely ill patients. Similarly, integrated health plans have strong financial incentives to avoid MH/SA users even after adjustment. CONCLUSIONS: Current risk adjustment methodologies do not eliminate the financial incentives for integrated health plans and behavioral health care carve-out plans to avoid high-utilizing patients with psychiatric disorders.

Resource utilization of patients with hypochondriacal health anxiety and somatization.

OBJECTIVES: To examine the resource utilization of patients with high levels of somatization and health-related anxiety. DESIGN: Consecutive patients on randomly chosen days completed a self-report questionnaire assessing somatization and health-related, hypochondriacal anxiety. Their medical care utilization in the year preceding and following completion of the questionnaire was obtained from an automated patient record. The utilization of patients above and below a predetermined threshold on the questionnaire was then compared. PATIENTS AND SETTING: Eight hundred seventy-six patients attending a primary care clinic in a large, urban, teaching hospital. OUTCOME MEASURES: Number of ambulatory physician visits (primary care and specialist), outpatient costs (total, physician services, and laboratory procedures), proportion of patients hospitalized, and proportion of patients receiving emergency care. RESULTS: Patients in the uppermost 14% of the clinic population on somatization and hypochondriacal health anxiety had appreciably and significantly higher utilization in the year preceding and the year following completion of the somatization questionnaire than did the rest of the patients in the clinic. After adjusting for group differences in sociodemographic characteristics and medical comorbidity, significant differences in utilization remained. In the year preceding the assessment of somatization, their adjusted total outpatient costs were $1,312 (95% CI $1154, $1481) versus $954 (95% CI $868, $1057) for the remainder of the patients and the total number of physician visits was 9.21 (95% CI 7.94, 10.40) versus 6.33 (95% CI 5.87, 6.90). In the year following the assessment of somatization, those above the threshold had adjusted total outpatient costs of $1,395 (95% CI $1243, $1586) versus $1,145 (95% CI $1038, $1282), 9.8 total physician visits (95% CI 8.66, 11.07) versus 7.2 (95% CI 6.62, 7.77), and had a 24% (95% CI 19%, 30%) versus 17% (95% CI 14%, 20%) chance of being hospitalized. CONCLUSIONS: Primary care patients who somatize and have high levels of health-related anxiety have considerably higher medical care utilization than nonsomatizers in the year before and after being assessed. This differential persists after adjusting for differences in sociodemographic characteristics and medical morbidity.

Workers' perceptions of how jobs affect health: a social ecological perspective.

A national sample of 2,048 workers was asked to rate the impact of their job on their physical and mental health. Ordered logistic regression analyses based on social ecology theory showed that the workers' responses were significantly correlated with objective and subjective features of their jobs, in addition to personality characteristics. Workers who had higher levels of perceived constraints and neuroticism, worked nights or overtime, or reported serious ongoing stress at work or higher job pressure reported more negative effects. Respondents who had a higher level of extraversion, were self-employed, or worked part time or reported greater decision latitude or use of skills on the job reported more positive effects. These findings suggest that malleable features of the work environment are associated with perceived effects of work on health, even after controlling for personality traits and other sources of reporting bias.

The setting of psychiatric care for medicare recipients in general hospitals with specialty units.

This study examined the patient and hospital characteristics associated with whether patients with psychiatric disorders were treated on the psychiatric unit or on medical wards after admission to general hospitals with psychiatric units. Medicare data for 169,798 beneficiaries who had psychiatric disorders and were admitted to general hospitals with psychiatric units were used to estimate logistic regressions of the probability of treatment on the unit. Results showed that beneficiaries who had more than one psychiatric diagnosis (except for substance use disorders), state buy-in coverage such as Medicaid, or previous psychiatric hospitalizations or who had ever been eligible for Medicare through disability were more likely to be treated on the unit. Those who were older, admitted through the emergency department, or had greater medical morbidity or primary diagnoses other than schizophrenia or bipolar or major affective disorders were less likely to be treated on the unit.

The role of profit status under imperfect information: evidence from the treatment patterns of elderly medicare beneficiaries hospitalized for psychiatric diagnoses.

Medicare claims for elderly admitted for psychiatric care were used to estimate the impact of hospital profit status on costs, length of stay (LOS), and rehospitalizations. No evidence was found that not-for-profits (NFPs) treated sicker patients or had fewer rehospitalizations. For-profits (FPs) actually treated poorer patients. Longer LOS and lower daily costs of NFPs were attributable to their other characteristics, e.g. medical school affiliation. Instrumental variables (IV) estimates suggested that NFP general hospitals actually have lower adjusted costs. These findings fail to support concerns that FP growth leads to declining access and quality or contentions that NFPs are less efficient.

Confidentiality and adolescents' use of providers for health information and for pelvic examinations.

OBJECTIVE: To examine the relationship between adolescents' perception of the confidentiality of care provided by their regular health care provider and their reported use of this provider for private health information and for pelvic examinations. DESIGN: Anonymous, self-report survey. SETTING: Thirty-two randomly selected public high schools in Massachusetts. PARTICIPANTS: Of 2224 students in systematically selected 9th and 12th grade classrooms, 1715 (50% male) had a regular provider and a checkup within the last year. RESULTS: Of teens surveyed, 76% wanted the ability to obtain confidential health care, but only 45% perceived their regular provider to provide this, and only 28% had discussed it explicitly. Logistic regression analyses revealed strong relationships between confidentiality and all outcomes studied. Among adolescents, the likelihood of having discussed sexually transmitted diseases, pregnancy prevention, and/or facts about sex with their provider was greater among teens who received a confidentiality assurance than that for teens who did not (odds ratio [OR] = 2.7; 95% confidence interval [CI], 2.2-3.4). A similar relationship for teens' likelihood of having discussed substance use with the provider was found (OR = 1.8; 95% CI, 1.4-2.3). Among sexually active females, the likelihood of a recent pelvic examination for those who received a confidentiality assurance was greater than for those who did not (OR = 3.3; 95% CI, 2.1-5.5). CONCLUSIONS: This study furthers evidence of an important link between teens' perception of confidentiality and use of health care services and information. Because teens' health risks lie largely in potential risks from health-related behaviors, confidentiality in health care may be a critical factor in disclosure and discussion of risky behaviors, and ultimately in appropriate use of health care services. Efforts should be made to increase teens' access to confidential health care sources.

Risk adjustment of capitation payments to behavioral health care carve-outs: how well do existing methodologies account for psychiatric disability?

This study used 1994-1995 administrative data from a large public employer to examine the viability of commercial risk adjustment systems for setting capitation payments to competing behavioral health care "carve-outs". The ability of Hierarchical Condition Categories and Adjusted Diagnostic Groups to predict psychiatric expenditures was improved by controlling separately for psychiatric disability. However, even the best models underpredicted expenditures of patients with psychiatric disability by 15%. Relative to full capitation, "mixed" payment systems and soft capitation reduce the ability of carve-outs to earn disproportionate profits by enrolling healthy patients and avoiding sick ones, yet also diminish incentives for cost containment.

Impact of expanding SSI on Medicaid expenditures of disabled children.

Supplemental Security Income (SSI) expansions for disabled children in the early 1990s provoked criticism that eligibility criteria were too lax and motivated the subsequent retraction of benefits for many children. However, little evidence exists on whether the clinical needs of SSI children declined during this period. The authors used Medicaid data to examine changes in average expenditures between 1989 and 1992, using an Aid to Families with Dependent Children (AFDC) comparison group to control for confounding time trends (e.g., in access). Results showed declines in average expenditures in Georgia and Tennessee but increases in California and Michigan, which are thought to have started with more liberal eligibility policies.

Diagnoses of patients treated with ECT: a comparison of evidence-based standards with reported use.

OBJECTIVE: This study assessed the extent to which patients treated with electroconvulsive therapy (ECT) had diagnoses for which ECT is an efficacious treatment according to evidence-based standards. METHODS: ECT use among all beneficiaries of a large New England insurance company in 1994 and 1995 was examined using a retrospective cohort design. Associations between provider characteristics and ECT use for diagnoses outside the standards were determined using logistic regression analysis. RESULTS: A total of 996 individuals among approximately 1.2 million beneficiaries were treated with ECT. They received a total of 1,532 ECT courses. For 86.5 percent of the courses, the diagnosis was within evidence-based indications; for 13.5 percent, the diagnosis was outside the indications. In more than half of the 13.5 percent of cases, conditions were depressive disorders for which no studies have been conducted or disorders that likely had associated depressive symptoms. Patients receiving ECT for diagnoses outside evidence-based indications were more likely to have been treated by psychiatrists who graduated from medical school between 1940 and 1960 and between 1961 and 1980 than by those who graduated between 1981 and 1990. These patients were also less likely to have been treated by psychiatrists who received their medical education outside the U.S. CONCLUSIONS: Diagnoses of patients treated with ECT were mostly within evidence-based indications. The results provide reassurance to those concerned that ECT may be used indiscriminately. If confirmed by further research, the finding that psychiatrists trained in earlier eras were more likely to use ECT for diagnoses outside evidence-based indications may offer an opportunity for targeted quality improvement.

Differences between generalists and mental health specialists in the psychiatric treatment of Medicare beneficiaries.

OBJECTIVE: To examine differences between the general medical and mental health specialty sectors in the expenditure and treatment patterns of aged and disabled Medicare beneficiaries with a physician diagnosis of psychiatric disorder. DATA SOURCES: Based on 1991-1993 Medicare Current Beneficiary Survey data, linked to the beneficiary's claims and area-level data on provider supply from the Area Resources File and the American Psychological Association. STUDY DESIGN: Outcomes examined included the number of psychiatric services received, psychiatric and total Medicare expenditures, the type of services received, whether or not the patient was hospitalized for a psychiatric disorder, the length of the psychiatric care episode, the intensity of service use, and satisfaction with care. We compared these outcomes for beneficiaries who did and did not receive mental health specialty services during the episode, using multiple regression analyses to adjust for observable population differences. We also performed sensitivity analyses using instrumental variables techniques to reduce the potential bias arising from unmeasured differences in patient case mix across sectors. PRINCIPAL FINDINGS: Relative to beneficiaries treated only in the general medical sector, those seen by a mental health specialist had longer episodes of care, were more likely to receive services specific to psychiatry, and had greater psychiatric and total expenditures. Among the elderly persons, the higher costs were due to a combination of longer episodes and greater intensity; among the persons who were disabled, they were due primarily to longer episodes. Some evidence was also found of higher satisfaction with care among the disabled individuals treated in the specialty sector. However, evidence of differences in psychiatric hospitalization rates was weaker. CONCLUSIONS: Mental health care provided to Medicare beneficiaries in the general medical sector does not appear to substitute perfectly for care provided in the specialty sector. Our study suggests that the treatment patterns in the specialty sector may be preferred by some patients; further, earlier findings indicate geographic barriers to obtaining specialty care. Thus, the matching of service use to clinical need among this vulnerable population may be inappropriate. The need for further research on outcomes is indicated.

The relationship between continuity of care and the health behaviors of patients: does having a usual physician make a difference?

BACKGROUND: Implicit in "any willing provider" and "freedom of choice" legislation is the assumption that ongoing provider relationships lead to better patient outcomes on average. Although previous studies have identified associations of usual source of care with medical utilization, its relationship to patient lifestyle has not been examined. OBJECTIVE: To determine the effect of having a usual physician on health behaviors. METHODS: Data on 3,140 adults from the 1995 Mid-Life in the US study were used to estimate logistic regressions of the effect of having a usual physician on exercise, obesity, vitamin-taking, smoking quits, substance abuse behaviors, preventive medical visits, and respondent assessments of the ability to affect one's own health and risk of heart attacks and cancer. RESULTS: Respondents with a usual physician were 3 times as likely to have had a preventive medical visit during the past year. Among lower-income respondents, those with usual physicians were one-half as likely to report substance abuse behaviors. Instrumenting reduced the magnitude of the former but not latter effect. No other significant differences were found. CONCLUSIONS: Strategies designed to foster regular patient-provider relationships may affect certain health behaviors, such as preventive care visits and substance abuse. Yet in the absence of interventions to improve the effectiveness of these relationships, they are unlikely to be a powerful policy instrument for achieving widespread improvements in patient lifestyle choices.

Changing patterns of conditions among children receiving Supplemental Security Income disability benefits.

OBJECTIVE: To determine the relative growth of types of chronic health conditions among children and adolescents receiving Supplemental Security Income (SSI) benefits before and after major SSI program changes, including changes in definitions of childhood disability and outreach to identify eligible children. DESIGN: Retrospective analysis of Medicaid claims from California, Georgia, Michigan, and Tennessee. PARTICIPANTS: All children (aged < or = 21 years) newly enrolled in SSI programs in these states from July 1989 (n=21 222) to June 1992 (n=38 789). METHODS: Medicaid data indicate eligibility status and diagnoses for services rendered. For children newly enrolled before (time 1, July 1989 to June 1990), during (time 2, July 1990 to June 1991), and after (time 3, July 1991 to June 1992) the program changes, we used claims for the first 6 months of enrollment to determine rates of chronic conditions in general and rates of asthma, attention-deficit/hyperactivity disorder (ADHD), and mental retardation specifically. We also followed up time 1 enrollees during the study period to determine the likelihood of a chronic condition claim at any time. MAIN OUTCOME MEASURE: Presence of claims for chronic conditions. RESULTS: New SSI enrollees almost doubled during the study period. Increasing numbers of new enrollees had chronic condition claims in their first 6 months (from 29% to 36%); 58% of time 1 enrollees had such claims during any study month. Rates of chronic physical conditions other than asthma increased 14% (time 1 to time 3); asthma rates increased 73%. Rates of mental health conditions other than mental retardation and ADHD increased 63%; rates of mental retardation decreased 29%, while rates of ADHD increased almost 3-fold. CONCLUSIONS: The number of children with chronic conditions receiving SSI benefits experienced rapid growth from 1989 to 1992. Growth was particularly marked for children with diagnoses of asthma and ADHD.

Trends in alternative medicine use in the United States, 1990-1997: results of a follow-up national survey.

CONTEXT: A prior national survey documented the high prevalence and costs of alternative medicine use in the United States in 1990. OBJECTIVE: To document trends in alternative medicine use in the United States between 1990 and 1997. DESIGN: Nationally representative random household telephone surveys using comparable key questions were conducted in 1991 and 1997 measuring utilization in 1990 and 1997, respectively. PARTICIPANTS: A total of 1539 adults in 1991 and 2055 in 1997. MAIN OUTCOMES MEASURES: Prevalence, estimated costs, and disclosure of alternative therapies to physicians. RESULTS: Use of at least 1 of 16 alternative therapies during the previous year increased from 33.8% in 1990 to 42.1% in 1997 (P < or = .001). The therapies increasing the most included herbal medicine, massage, megavitamins, self-help groups, folk remedies, energy healing, and homeopathy. The probability of users visiting an alternative medicine practitioner increased from 36.3% to 46.3% (P = .002). In both surveys alternative therapies were used most frequently for chronic conditions, including back problems, anxiety, depression, and headaches. There was no significant change in disclosure rates between the 2 survey years; 39.8% of alternative therapies were disclosed to physicians in 1990 vs 38.5% in 1997. The percentage of users paying entirely out-of-pocket for services provided by alternative medicine practitioners did not change significantly between 1990 (64.0%) and 1997 (58.3%) (P=.36). Extrapolations to the US population suggest a 47.3% increase in total visits to alternative medicine practitioners, from 427 million in 1990 to 629 million in 1997, thereby exceeding total visits to all US primary care physicians. An estimated 15 million adults in 1997 took prescription medications concurrently with herbal remedies and/or high-dose vitamins (18.4% of all prescription users). Estimated expenditures for alternative medicine professional services increased 45.2% between 1990 and 1997 and were conservatively estimated at $21.2 billion in 1997, with at least $12.2 billion paid out-of-pocket. This exceeds the 1997 out-of-pocket expenditures for all US hospitalizations. Total 1997 out-of-pocket expenditures relating to alternative therapies were conservatively estimated at $27.0 billion, which is comparable with the projected 1997 out-of-pocket expenditures for all US physician services. CONCLUSIONS: Alternative medicine use and expenditures increased substantially between 1990 and 1997, attributable primarily to an increase in the proportion of the population seeking alternative therapies, rather than increased visits per patient.

Inpatient psychiatric treatment of elderly Medicare beneficiaries.

OBJECTIVE: The clinical characteristics and treatment patterns of elderly Medicare beneficiaries hospitalized for psychiatric disorders were examined. METHODS: Administrative data on all elderly Medicare beneficiaries in the United States hospitalized in a nonfederal hospital for a primary psychiatric disorder in 1990-1991 were used to calculate descriptive statistics on case-mix by age group, hospital type (psychiatric hospital, general hospital psychiatric unit, or general hospital nonpsychiatric unit), and primary diagnosis. Length of stay, costs, and discharge destination by hospital type and primary diagnosis were also determined. RESULTS: A total of .6 percent of elderly Medicare beneficiaries were hospitalized for a psychiatric disorder in 1990, accounting for more than 240,000 admissions and $1 billion in Medicare payments. The most common reasons for hospitalization were major depressive disorder (28.1 percent), dementia and other organic disorders (26.8 percent), and substance-related disorders (12.6 percent). Organic disorders were particularly prevalent among the oldest old, accounting for more than half of psychiatric admissions among those 85 and older. A total of 43 percent of the psychiatric admissions were to general hospital nonpsychiatric units, 38 percent to general hospital psychiatric units, and only 19 percent to psychiatric hospitals. Within each diagnostic category, patients admitted to general hospital nonpsychiatric units had the shortest average lengths of stay and the lowest average costs. Among beneficiaries with organic, affective, and psychotic disorders other than schizophrenia, those admitted to general hospitals had shorter lengths of stay, higher rates of discharge to nursing homes, and lower rates of discharge to self-care than those treated in psychiatric hospitals. CONCLUSIONS: Case-mix-adjusted treatment patterns varied substantially across hospital types, due to differences in either illness severity or treatment styles.

Contribution of psychosocial factors to socioeconomic differences in health.

The National Survey of Mid-life Developments in the United States (MIDUS) is one of several studies that demonstrate socioeconomic gradients in mortality during midlife. When MIDUS findings on self-reported health, waist to hip ratio, and psychological well-being were analyzed for their possible roles in generating socioeconomic differences in health, they revealed clear educational gradients for women and men (i.e., higher education predicted better health). Certain potential mediating variables, like household income, parents' education, smoking behavior, and social relations contributed to an explanation of the socioeconomic gradient. In addition, two census-based measures, combined into an area poverty index, independently predicted ill health. The results suggest that a set of both early and current life circumstances cumulatively contribute toward explaining why people of lower socioeconomic status have worse health and lower psychological well-being.

High-expenditure children with Supplemental Security Income.

OBJECTIVE: To examine the clinical characteristics and health service use of children with high Medicaid expenditures. METHODOLOGY: We examined 1992 Medicaid claims and eligibility files from four states (California, Georgia, Michigan, Tennessee) for children with at least $10000 billed to Medicaid who obtained Medicaid through the Supplemental Security Income (SSI) Program and a comparison group (matched by age group and gender) of children receiving Medicaid for other reasons. We compared mean expenditures, examined expenses by category, and examined diagnoses associated with at least $10000 in expenses. RESULTS: In 1992, Medicaid paid on average approximately $1000 for children with non-SSI Medicaid enrollment. Expenditures for children with SSI were 2.9 to 9.4 times higher, but once the approximately 10% of children with high expenditures were excluded, SSI average expenditures were only 1.5 to 2.7 times higher than the non-SSI average. Children with high expenditures are likely to use hospitals and long-term care, and these services account for more than half of the average expenditures. Children with high expenditures and SSI are more likely to have chronic medical conditions than are their peers enrolled in Medicaid but not through SSI. CONCLUSIONS: A small proportion of children, even on SSI, account for very large proportions of Medicaid expenditures. Most children with SSI, despite having relatively severe mental health, physical, or developmental disabilities, have relatively modest Medicaid expenditures.

Risk adjustment of mental health and substance abuse payments.

This study used 1992 and 1993 data from private employers to compare the performance of various risk adjustment methods in predicting the mental health and substance abuse expenditures of a nonelderly insured population. The methods considered included a basic demographic model, Ambulatory Care Groups, modified Ambulatory Diagnostic Groups and Hierarchical Coexisting Conditions (a modification of Diagnostic Cost Groups), as well as a model developed in this paper to tailor risk adjustment to the unique characteristics of psychiatric disorders (the "comorbidity" model). Our primary concern was the amount of unexplained systematic risk and its relationship to the likelihood of a health plan experiencing extraordinary profits or losses stemming from enrollee selection. We used a two-part model to estimate mental health and substance abuse spending. We examined the R2 and mean absolute prediction error associated with each risk adjustment system. We also examined the profits and losses that would be incurred by the health plans serving two of the employers in our database, based on the naturally occurring selection of enrollees into these plans. The modified Ambulatory Diagnostic Groups and comorbidity model performed somewhat better than the others, but none of the models achieved R2 values above .10. Furthermore, simulations based on actual plan choices suggested that none of the risk adjustment methods reallocated payments across plans sufficiently to compensate for systematic selection.

Characteristics of psychiatrists who perform ECT.

OBJECTIVE: Use of ECT is highly variable, and previous study has linked its availability to the geographic concentration of psychiatrists. However, less than 8% of all U.S. psychiatrists provide ECT. The authors analyzed the characteristics of psychiatrists who use ECT to understand more fully the variation in its use and how changes in the psychiatric workforce may affect its availability. METHOD: Data from the 1988-1989 Professional Activities Survey were examined to investigate the influence of demographic, training, clinical practice, and geographic characteristics on whether psychiatrists use ECT. RESULTS: Psychiatrists who provided ECT were more likely to be male, to have graduated from a medical school outside the United States, and to have been trained in the 1960s or 1980s rather than the 1970s. They were more likely to provide medications than psychotherapy, to practice at private rather than state and county public hospitals, to treat patients with affective and organic disorders, and to practice in a county containing an academic medical center. CONCLUSIONS: Demographic and training characteristics significantly influence whether a psychiatrist uses ECT. Opposing trends in the U.S. psychiatric workforce could affect the availability of the procedure. Expanding training opportunities for ECT and making education, training, and testing more consistent nationwide could improve clinicians' consensus about ECT and narrow variation in its use.