	We applaud the research performed by David Mosen and colleagues, which suggests improvements in identifying persistent  asthmatics when reporting Health Plan Employer Data and Information Set (HEDIS) data.

 We were pleased that the Kaiser Permanente Health Care Program had solicited our input into the research design, so it would help resolve questions about the specificity of the HEDIS asthma measure. The analysis determined that extending the eligibility criteria to 2 years would strengthen the specificity of the current HEDIS measure by improving administrative case identification of true persistent asthmatics.

	The authors' recommendation to increase the eligibility from 1 year of identification to 2 years has already been included in the latest measure specifications, which were published last July.

 This is a good example of accelerating the transfer of research into practice, and we encourage this sort of collaboration to ensure that HEDIS remains a robust and valid set of measures for performance measurement.

 How well do the HEDIS asthma inclusion criteria identify persistent asthma? 

2. National Committee for Quality Assurance.

 Technical Specifications. Washington, DC: National Committee for Quality Assurance; 2005;460. HEDIS 2006. 

Health Plan Employer Data & Information Set;

Articles

LETTERS TO THE EDITORS

Improvements in Identifying Persistent Asthma Patients Using HEDIS Data

To compare the performance of the Elixhauser, Charlson, and RxRisk-V comorbidity indices and several simple count measurements, including counts of prescriptions, physician visits, hospital claims, unique prescription classes, and diagnosis clusters. 

Each measurement was calculated using claims data during a 1-year period before the initial filling of an antihypertensive medication among 20 378 members of a managed care organization. The primary outcome variable was the log-transformed sum of prescription, physician, and hospital expenditures in the year following the prescription encounter. 

In addition to descriptive statistics and Spearman rank correlations between measurements, the predictive performance was determined using linear regression models and corresponding adjusted 

The Charlson index and the Elixhauser index performed similarly (adjusted 

= 0.1172 and 0.1148, respectively), while the prescription claims-based RxRisk-V (adjusted 

= 0.1573) outperformed both. An age- and gender-adjusted regression model that included a count of diagnosis clusters was the best individual predictor of payments (adjusted 

= 0.1814). This outperformed age- and gender-adjusted models of the number of unique prescriptions filled (adjusted 

= 0.1669), number of prescriptions filled (

= 0.1573), number of physician visits (adjusted 

= 0.1546), log-transformed prior healthcare payments (adjusted 

= 0.1359), and number of hospital claims (adjusted 

Simple count measurements appear to be better predictors of future expenditures than the comorbidity indices, with a count of diagnosis clusters being the single best predictor of future expenditures among the measurements examined. 

Comorbidity scores are a common tool used by researchers in epidemiological and health services research. Comorbidities are defined as coexisting medical conditions that are distinct from the primary condition under investigation.

Interest in comorbidity scores can be attributed to the importance of relationships between comorbidities and the prognosis, detection, and outcomes of many illnesses.

In studies using secondary administrative data, the absence of randomization to treatment and control groups can result in health status differences across groups. This can potentially confound the relationship between a treatment and disease under investigation. 

International Classification of Diseases, Ninth Revision,

Numerous comorbidity controls exist, including 

-based measures such as the Elixhauser and Charlson indexes

and pharmacy claims-based measures such as the chronic disease score.

Simple variables used to measure comorbidity, such as counts of medications, physician visits, or medical conditions, also have been used in research. These measurements are less complex to use, and studies

have shown them to be as effective (if not more effective) as other measurements in predicting and controlling for comorbidity. Furthermore, these measurements are not subject to biases related to misclassification. Misclassification can arise in comorbidity indexes if complications (conditions arising from the treatment or progression of a condition) are coded as comorbidities (conditions existing simultaneously with and independently of other medical conditions). 

have compared the predictive validity of commonly used comorbidity scores. The focus of most of these comparisons, however, has been on the prediction of mortality and morbidity and not on healthcare expenditures. Increasingly, comorbidity scores are being used to control for comorbidity differences in studies

in which expenditures and payments are the primary dependent variables. Because most comorbidity measures were developed to predict mortality or morbidity, potential differences may arise when these measures are applied to expenditure outcomes. For example, although patients not surviving the initial onset of an acute myocardial infarction would be included as deaths in mortality estimates, they may not have significant healthcare expenditures because of their deaths. In this case, measures that are good predictors of mortality may not necessarily predict expenditures well. 

Comorbidity indexes have been used to argue for differences in capitation payment rates.

Of particular concern in capitation arrangements is that a small fraction of insurance beneficiaries often accounts for a large portion of healthcare expenditures. These high-expenditure individuals may also be of interest to insurers wishing to target disease management programs to control health plan spending. Although comorbidity indexes may be useful in predicting high-expenditure individuals, a comparison of measurements has not been undertaken to date, to our knowledge. Therefore, the use of comorbidity measures for these objectives may be in question. 

Given the disparate comorbidity measures in use and the absence of studies comparing their performance in analyses with expenditure outcomes, this study was undertaken to compare the performance of different comorbidity measures in predicting individual healthcare expenditures. Specifically, the performance of the following 3 administrative claims-based comorbidity scores was assessed: the 

-based Charlson and Elixhauser indexes and the pharmacy claims-based RxRisk-V score. Furthermore, the performance of these scores was compared with that of simpler comorbidity measurements, including counts of prescriptions, physician visits, hospital claims, and unique prescription classes, as well as the cumulative number of diagnosis clusters, serving as a proxy for the count of unique health conditions. 

For this analysis, we used hospital, physician, and pharmacy claims data from a large managed care organization. The study population included 20 378 individuals 18 years and older who had claims for a diagnosis of hypertension (

codes 401-404, 362.11, or 437.2) and who obtained an initial prescription fill for an antihypertensive medication between January 1, 2001, and December 31, 2002. The index date in this study was defined as the original prescription purchase date. The preperiod and the postperiod include the periods 1 year before and 1 year after the index date, respectively. Individuals with gaps in enrollment longer than 31 days during the preindex and the postindex periods were excluded. 

The Charlson index is the most common index used to control for comorbidity in health outcomes studies. The original Charlson index was developed for use with medical records and consisted of 19 different diseases weighted according to disease severity as 1, 2, 3, or 6. The index has since been adapted into several 17-item weighted indexes for use with administrative data.

A comprehensive comparison performed by Schneeweiss et al

examined differences in the predictive ability of several Charlson indexes of mortality, long-term care admissions, hospitalizations, physician visits, and expenditures for physician services. Results from that study showed little difference in the performance of different Charlson indexes, with the adaptation by Romano et al

performing best. We used a modified version of the Romano-adapted Charlson index to accommodate changes in 

coding. Based on previous investigations that suggest that adding physician claims to hospital claims increases the performance of the Charlson index, we ran the index first using preperiod hospital claims alone and then using preperiod hospital and physician claims.

A newer comorbidity measurement is the index by Elixhauser et al.

The Elixhauser index measures the effect of 30 different comorbid conditions. The index distinguishes comorbidities from complications by considering only secondary diagnoses unrelated to the principal diagnosis through the use of diagnosis related groups (DRGs). For example, a patient with a claim for congestive heart failure would have this condition coded as a comorbidity only if the medical record did not contain a DRG for cardiac disease. Current coding for the Elixhauser index was downloaded from the Agency for Healthcare Research and Quality.

The index was run first using preperiod hospital claims alone and then using preperiod hospital and physician claims. Although DRGs are not available within physician claims, it was thought that many comorbid conditions would be missed if these data were not included. The final Elixhauser scores were calculated as the sum of comorbid conditions present. Hypertension was excluded from the final score because of the disease population studied. 

Prescription Claims-based Comorbidity Index

Several indexes commonly referred to as chronic disease scores have been developed for use with pharmacy claims data.

The most recent modifications to the chronic disease scores are the RxRisk score for use among general populations

and the RxRisk-V score for use among Veterans Affairs populations.

Although developed for Veterans Affairs populations, the RxRisk-V score was deemed more applicable to our study population based on the population's age and disease distribution. Coding of the RxRisk-V was completed using the medication classes provided in the original article and the corresponding Medi-Span codes.

 The RxRisk-V identifies 45 distinct comorbid conditions by linking them to medications used during treatment. We used nonweighted and weighted counts of RxRisk-V conditions. Weights for the RxRisk-V were taken directly from the originally published prospective cost coefficient estimates.

Specific counts of healthcare utilization were performed in this study. These included a 12-month count of physician visits, hospital claims, prescriptions filled, and unique prescriptions (classes of prescriptions) filled during the 1-year preperiod. 

As a proxy for the number of unique medical conditions, we categorized conditions into diagnosis clusters and used a count of diagnosis clusters as a summary score.

 Coding of conditions into 1 of 119 unique diagnosis clusters was performed using 

claims not identified in diagnosis clustering were categorized as having an "other" diagnosis cluster. Hypertension was excluded as a cluster because of the population studied. Diagnosis clusters were implemented using hospital claims, as well as both hospital and physician claims. 

The final measurements used as potential predictors of postperiod payments were a sum of preperiod payments and a sum of log-transformed preperiod payments. To avoid log-transformation errors, $0.01 was added to preperiod payments before transformation. 

The primary outcome in this study was the sum of individual healthcare expenditures, defined as the sum of hospital, physician, and prescription payments in the 1-year postperiod. Payments included the amount paid out by the health plan to the provider, any amount paid by the patient (including deductibles and copayments), ancillary payments, and any amount reserved from the health plan. Log transformation of the dependent variable was performed to account for nonnormality. 

Descriptive statistics of population characteristics were performed, including means, standard deviations, and ranges for demographic, payment, healthcare utilization, and comorbidity variables. We assessed the correlations between each comorbidity measurement using Spearman rank correlations (?). Spearman rank correlations were used to account for potential nonnormality bias in the independent variables. The performance of each comorbidity measurement was assessed through ordinary least squares linear regressions. Adjusted 

values were reported as an informal comparison of the prediction performance to adjust for the number of explanatory variables in each regression model. The adjusted 

value should be thought of as an index value of variance that is corrected for 

 values correspond to improved model fit and greater predictive ability. To examine the performance of measurements in predicting high expenditures, we used area under the receiver operating characteristic (ROC) curve comparisons. Area under the ROC curve comparisons assess the ability of each measurement to accurately predict true-positive cases while not predicting true-negative cases. Area under the ROC curve outcomes are analyzed through the use of C statistics, which can range from 0.5 (representing no predictive ability) to 1.0 (representing perfect predictive ability). The area under the ROC curve outcome was dichotomized as 0 or 1, with 1 representing high-expenditure individuals who spent at or above the 90th percentile among the study population. 

Full descriptive statistics for the 20 378 individuals who comprised our study population are given in 

Population members were on average 49 years of age. There were slightly more men (53%) than women. In the year before filling a new prescription for an antihypertensive medication, individuals averaged 1.7 hospital claims, 10.4 physician visits, 13.2 prescriptions, and 4.7 unique prescription medications used. The mean count of diagnosis clusters identified through preperiod hospital and physician visit claims was 7.1, compared with a mean of 2.0 diagnosis clusters identified through hospital claims alone. The mean comorbidity score for each diagnosis-based index was greater when physician claims were combined with hospital claims. Mean scores ranged from 0.55 for the Charlson index, 0.61 for the Elixhauser index, 1.98 for the nonweighted RxRisk-V score, and 4111 for the weighted RxRisk-V score. The mean payment was $4615 in the preperiod, and the mean payment was $6301 in the postperiod. Most payments incurred in the postperiod were hospital payments ($2756), followed by physician payments ($2053) and prescription payments ($1492). The cutoff for 90th percentile spending was $12 945. The mean postperiod spending was $3320 among individuals below the 90th percentile cutoff, and the mean postperiod spending was $33 156 among individuals above the 90th percentile cutoff. 

The strength of the correlations between the different indexes varied across types of measurements, as summarized in 

claims-based indexes (the Elixhauser and Charlson indexes) was fair (? = 0.562) in this analysis. However, the correlation between the 

and pharmacy claims-based indexes was small, with the correlations between the nonweighted RxRisk-V score and Elixhauser index being slightly better (? = 0.301) than those between the nonweighted RxRisk-V score and Charlson index (? = 0.242). Among the count measurements analyzed, there was a high correlation between the hospital claims-identified diagnosis clusters and number of hospital claims (? = 0.932) and between the hospital and physician claims-identified diagnosis clusters and number of physician visits (? = 0.820). The nonweighted RxRisk-V score was strongly correlated with the number of prescriptions used (? = 0.806) and the number of unique prescriptions used (? = 0.848). The correlations were fair between the counts of hospital and physician visits (? = 0.560) and counts of prescription fills and physician visits (? = 0.555). However, the correlations were small across the counts of prescriptions filled and hospital counts (? = 0.327). 

The predictive performance of each comorbidity measurement on postperiod payments is given in 

The amount of variation explained by each measurement increased without exception when age and gender were included as predictors. Similarly, the addition of physician claims to hospital claims increased the predictive performance of the Charlson and Elixhauser indexes. The full Charlson index (which used hospital and physician claims, as well as adjustments for age and gender) and the full Elixhauser index performed similarly (adjusted 

= 0.1172 and 0.1148, respectively). Among the 2 RxRisk-V scores, the nonweighted score (adjusted 

= 0.1381) was a better predictor than the weighted score (adjusted 

= 0.1261). Compared with the age-and gender-adjusted RxRisk-V (adjusted 

-based indexes appear to be slightly inferior. The regression model that included a count of diagnosis clusters, age, and gender was the best individual predictor of future payments (adjusted 

= 0.1814). This outperformed age- and gender-adjusted models of counts of unique prescriptions filled (adjusted 

= 0.1669), prescriptions filled (adjusted 

= 0.1359), and hospital claims (adjusted 

METHODS

A Comparison of Comorbidity Measurements to Predict Healthcare Expenditures

To evaluate the Advanced Illness Coordinated Care Program (AICCP), delivered by allied health personnel to improve care for patients coping with advanced illness and in need of preparation for end-of-life (EOL) care. 

Clinical trial involving 275 patients and 143 surrogates in 6 settings who were randomly assigned to the AICCP or usual care (UC). 

The AICCP participants met with a care coordinator for assistance with provider communication, care coordination, and support. The AICCP was evaluated for effects on satisfaction with care, advance planning, consistency of care with patient preferences, and healthcare costs. 

The AICCP increased patient satisfaction with care and communication (

= .03), and AICCP surrogates reported fewer problems with provider support (

= .03). More AICCP than UC participants completed an advance directive (AD) (69.4% vs 48.4%; 

= .006), and the AICCP group completed more ADs per participant (

= .01). Median time to AD documentation was 46 days for AICCP and 238 days for UC (

= .02). There was no difference in survival (AICCP 43% vs UC 42%). Six-month costs were lower with AICCP than with UC ($12 123 vs $16 295); however, the difference did not reach statistical significance. 

The AICCP improved satisfaction with care and helped patients develop and revise more ADs, sooner, without affecting mortality. This program may be delivered in a range of managed care, fee-for-service, and group-model settings. 

The need to improve the care of patients coping with advanced illness and preparing for end-of-life (EOL) care is well documented. Shortcomings in the medical, psychological, spiritual, and practical domains of care have been described by experts

Many Americans are dying subsequent to unwanted care, in unrelieved pain, and with their preferences for treatment inadequately discussed, documented, and followed.

Care at the EOL often is fragmented because it is delivered by different providers in different settings. 

For these reasons, the quality of life of many people with serious illness is compromised. These shortcomings will eventually affect more Americans: more than 73% of deaths occur after age 65 years, and this age group is projected to increase as the baby-boom cohort enters old age.

Because of importance of delivering improved care, the Institute of Medicine recommends conducting research on the utility of new care strategies.

The demands of the aging baby-boom generation for more input and better treatment during advanced illness are mobilizing efforts to improve EOL care.

 Recent efforts to improve care include the development of Palliative Care Leadership Centers,

 and prehospice and care management programs.

In addition, greater attention is being paid to EOL care in medical training and practice,

and to development of models for facilitating structured EOL discussions, including interventions that focus on advance planning.

The Advanced Illness Coordinated Care Program (AICCP) was developed to improve the care of people with serious illness.

It is a care coordination and support program delivered by allied health providers (eg, nurses, social workers) as a part of physician-directed care plans. The AICCP is designed to promote communication and understanding between patients and providers about advanced illness and EOL issues to promote health literacy, to achieve coordination of care, to provide emotional and social support, and to reduce barriers to the use of palliative care and hospice services. AICCP can be delivered in acute-care, primary care, or long-term care settings and as a component of disease management or case management services. It is delivered concurrently while patients are receiving life-prolonging treatments and before referral to specialty palliative services. 

The development of AICCP was guided by clinical experience,

and recommendations for improving care coordination and support from the health literature and the managed care industry.

AICCP is designed to address reported barriers to quality care,

AICCP uses proactive strategies explicitly designed to promote readiness. 

To test the effectiveness of AICCP for helping people cope with advanced illness and EOL decisions and to ascertain the program utility from quality and cost perspectives, we conducted a randomized, controlled trial comparing AICCP patients and surrogates with patients and surrogates receiving usual care (UC). We describe the impact of the AICCP on patient and surrogate satisfaction with healthcare and provider communication, formulation and development of advance directives (ADs), whether the received medical care was consistent with the patient's AD, and healthcare costs. 

Participants were recruited from 3 Department of Veterans Affairs medical centers (VAMCs), a home care organization, and 2 managed care organizations. Patients had chronic obstructive pulmonary disease (COPD), chronic heart failure (CHF), or cancer diagnoses including those of the esophagus, trachea, colon, liver, pancreas, lung, or uterus; cancers of the prostate or breast with metastasis; and melanoma, leukemia, lymphosarcoma, Hodgkin's disease, or multiple myeloma. Patients with COPD or CHF were eligible if they had experienced 1 or more admissions to an intensive-care unit or 2 or more acute-care admissions in the last 6 months. 

shows demographic information on patients. 

shows demographic information on surrogates. 

The study protocol was approved by institutional review boards at each site. Recruiters blinded to group assignment enrolled 275 patients. We used Research Randomizer version 3

to randomize participants in blocks of 10 into AICCP (n = 133) or UC (n = 142). Approximately 60% of the patients identified a surrogate, of which 168 (76 in the AICCP group and 92 in the UC group) agreed to participate. 

One hundred eighty-six patients (AICCP = 86, UC = 100) and 143 surrogates (AICCP = 67, UC = 76) completed the study and a follow-up assessment at approximately 3 months postenrollment. The AICCP was gradually implemented as part of UC in some VAMC inpatient units before study closure. Therefore, as specified in the institutional review board approval, patients were free to cross over to the alternative treatment arm in spite of their randomization status because administrators wanted to make AICCP available on units where AICCP was introduced into UC and participants requested AICCP. Eighteen UC participants crossed over to AICCP, and 2 AICCP participants crossed into UC. Intention-to-treat analyses were performed with participants in their originally assigned groups.

The AICCP delivers care coordination and support through 6 functions. The first is physician support, which consists of helping patients develop well-organized questions to make economical use of provider time and ensuring that physicians have complete information about patients. The second is health literacy, which is the capacity to understand basic health information.

 The AICCP addresses literacy concerns in each session (eg, by helping patients comprehend specialized medical terminology, which both increases their understanding and reduces their embarrassment). The third function is care coordination, which is locating and arranging linkages to medical services. The fourth is prevention, which refers to a focus on those aspects of EOL planning that often are avoided and emotionally charged. In this study, prevention referred to efforts to reduce or eliminate common psychosocial concerns related to advanced illness such as (1) coping with the loss of ability to perform valued activities; (2) identifying and addressing family conflict around difficult advanced illness and EOL decisions (eg, patient relocation, financial burdens of illness); (3) avoiding caregiver burnout (eg, by dividing care among family members); (4) anticipating emotional reactions (eg, anticipatory grief, fear of death); (5) enhancing self-management skills by preparing patients and families to cope with health system delivery shortfalls (eg, fragmentation of care delivery, gaps in care); and (6) promoting advance planning, because timely planning may avert decision making in crisis situations.

Care coordinators help clarify patient preferences for care under different health scenarios, using worksheets designed for this purpose.

If patients engage in advance planning, care coordinators assist them in formulating and documenting ADs and discussing them with providers. Family misunderstandings about care issues frequently can be resolved during meetings with care coordinators, reducing physician time spent mediating between family members. Care coordinators also provide emotional and social support. Emotional support consists of attending to affective components of illness, identifying specific emotions, helping patients cope with suffering, and providing referrals for ongoing counseling. Social support includes guidance and information, as well as tangible support. In the AICCP, structured guidance support helps patients and caregivers complete tasks needed to maintain health and function.

The AICCP provides information support in the form of guiding patients through the immense amount of medical information available to sources that are (1) adjusted for health literacy, (2) endorsed by their physicians, and (3) relevant to their situations. It provides tangible support by locating and arranging social support services. These functions are performed by nurses, nurse practitioners, or social workers. 

The AICCP was implemented in a 6-session format and delivered by existing personnel who were familiar with institutional policies and who had ongoing relationships with providers. These personnel were chosen because a reported barrier to effective implementation of an EOL program was using staff without an institutional identity and credibility.

Care coordinators'salaries were contributed by study sites. Sites replaced care coordinators if their resources allowed; if not, care coordinators'duties were reconfigured to focus on patients with advanced illness. Each care coordinator attended training and reviewed assigned readings, including the AICCP training manual.

Program delivery was standardized across sites through conference calls and followed a structured-visit format. Care coordinators were taught to individualize the program to meet specific needs; for example, patients could schedule extra meetings. The mean number of visits was 4.92 (SD = 2.94). 

To assess treatment fidelity, we examined treatment implementation checklists for a randomly selected subset of patients. Checklists covered AICCP-recommended interventions. The intervention elements were completed in 72% to 95% (mean = 83%) of patient visits. The most common reason for not completing an element was that it did not apply to patients'circumstances. 

To evaluate the impact of AICCP, participants were asked at enrollment and at 3 and 6 months postenrollment about the effectiveness of patient/provider communication, their satisfaction with care, and their willingness to participate in treatment planning. Effectiveness of patient-provider communication and satisfaction with healthcare were measured on an investigator-constructed, 10-item scale. Reliability measured at enrollment using Cronbach's alpha was .80. The patient satisfaction items on the questionnaire were measured on a 5-point Likert-type scale: 1, "very dissatisfied"; 2, "dissatisfied"; 3, "neither satisfied nor dissatisfied"; 4, "satisfied"; and 5, "very satisfied." Patients'attitudes about participating in treatment planning were assessed by a single item measuring low, medium, or high levels of involvement in decision making. 

Surrogates'experiences with the healthcare system were measured at 3 months postenrollment, using a modified EOL Family Interview. This instrument measured surrogate satisfaction with patient care and the number of problems with the healthcare system experienced by family and patients. Its psychometric properties have been established.

Surrogates indicated the presence or absence of problems in 7 domains: (1) shared decision making, (2) physical comfort and emotional support, (3) advance care planning, (4) coordination of care, (5) personal care and respect, (6) family self-efficacy, and (7) family emotional and spiritual support. The number of problems in each domain was averaged to create a problem score for that domain. In addition, 6 items measuring satisfaction with care were rated on an 11-point scale (with 0 being the worst care possible, and 10 the best care possible). These items were averaged to create a single overall rating of care. 

Program contact, salary, and overhead costs were combined for care coordinators from 3 sites (the VAMCs) to estimate the mean per-case cost of providing AICCP. Care coordinators collected data on contacts for 70 patients, including the person contacted and the length and reason for the contact. Care coordinators also documented time spent on other tasks such as charting and contacting providers. 

Cost data were gathered for 169 participants (AICCP = 93, UC = 76) from VAMC cost records from 6 months preenrollment to 6 months postenrollment. Veterans' Affairs medical centers records for 180 participants (AICCP = 85, UC = 95) were reviewed to track the formulation and documentation of ADs. Department of Veterans Affairs (VA) records were examined for ADs (healthcare proxies or living wills). The dates and types of ADs were recorded, and AD status (yes/no) was determined for participants at enrollment and 3 and 6 months postenrollment. The number of ADs found was also summed at these time points. Because, like ADs, do-not-resuscitate and intubate (DNR[I]) orders imply patient awareness of unwanted health outcomes and communication and action about them with providers, the same information was collected for DNR(I) orders. 

The AICCP and UC participants were compared on demographic and other characteristics at enrollment (Table 1 and Table 2). To assess the effect of AICCP on satisfaction with healthcare and communication with providers, we used a random effects regression model.

 Patient satisfaction scores were examined for significant effects of group, time, and group-by-time interaction. To assess the strength of the intervention effects, we calculated effect sizes as the ratio of the estimated treatment effect (follow-up satisfaction scores minus enrollment satisfaction scores) to the pooled standard deviations at enrollment.

Because surrogates were not interviewed at enrollment, groups were compared only on posttest scores for each of the 7 healthcare domains identified in the EOL Family Interview and on posttest mean ratings of overall care. Comparisons were made by using 

tests, and effect sizes were calculated by using the means and standard deviations at posttest. 

Participant-level costs from 6 months before enrollment through 3 and 6 months postenrollment were abstracted from the VA database. Costs included inpatient, outpatient, nursing home, inpatient hospice, and an "other" cost component (diagnostic services, medications, durable medical equipment, prosthetics, VA care provided in non-VA settings, and administrative overhead). Because a Wald test indicated multivariate nonnormality in the data, they were square-root-transformed in order to meet assumptions for normality. Effect size for cost was calculated the same as it was for satisfaction outcomes. 

To examine the association between group membership and time to completion of ADs, an analysis of time from enrollment to completion of the first AD was performed. Kaplan-Meier curves for the AICCP and UC groups from enrollment to 6 months postenrollment were compared, using a log-rank test. A chi-square comparison of the percentages in the 2 groups who completed at least 1 AD by 6 months postenrollment was performed. Mean differences in total numbers of ADs completed during the 3 months and the 6 months after enrollment also were compared with the 

Demographics presented in Tables 1 and 2 indicate no statistically significant differences between the AICCP and UC groups at enrollment. Although AICCP was not expected to have an impact on survival, analyses were conducted using Cox's proportional model to verify this expectation. No significant differences were found. Survival rates at 18 months postenrollment were 43% for the AICCP group versus 42% for the UC group. 

Analyses yielded a significant group-by-time interaction. The AICCP patients reported significantly greater increases in satisfaction from pretest (mean = 3.70, SD = .74) to posttest (mean = 4.07, SD = .68) than UC patients, whose pretest mean was 3.83 (SD = .76) and whose posttest mean was 3.98 (SD =.67) (F[1, 184] = 4.88, 

= .03). Effect size of AICCP on patient satisfaction was 0.18. The AICCP and UC patients did not differ in their attitudes about participating in treatment decisions (F[1, 168] = 0.01, 

Posttest score analyses also showed significant positive effects of participation on surrogates. Fewer problems were reported by AICCP surrogates (mean = .41, SD = .30) than UC surrogates (mean = .53, SD = .32) with the spiritual and emotional support delivered by providers (

= .03). The effect size for this difference was 0.39, indicating a moderate impact of AICCP on surrogates. 

The averge cost of AICCP delivery, including salary and administrative costs, was $452.12 per case. Results of analyses using the random effects regression model revealed a significant time effect, with both groups incurring higher costs over time from 6 months preenrollment to 6 months postenrollment (

). On average, AICCP costs per patient were $12 123 versus $16 295 for UC–a $4172 (25%) difference, with an effect size of .18. This represents a statistically nonsignificant trend toward total lower cost from 6 months preenrollment to 6 months postenrollment. Cost analyses were based on subjects from VAMC sites because only VAMC sites had reliable data on cost outcomes. 

At baseline there were no differences between AICCP and UC in AD status (?

= .768). Three analyses were performed on postbaseline AD data: (1) time to first AD, (2) group differences in the percentage of ADs by 3 and 6 months postbaseline, and (3) group differences in the total number of ADs by 3 and 6 months postbaseline. DNR(I)s were excluded from the following analyses because no group differences were found at any measurement period. 

Results of the first AD analysis appear in the 

 The Kaplan-Meier plot shows a significant difference in trend between the groups in time to completion of ADs. Median time to completion of first AD for AICCP was 46 days, whereas for UC it was 238 days (log-rank 

By 3 months postbaseline, 64.7% of AICCP participants versus 43.2% of UC participants had completed at least 1 AD (?

= .005). By 6 months, 69.4% of AICCP participants versus 48.4% of UC participants had completed at least 1 AD (c

CLINICAL

Effects of a Program for Coordinated Care of Advanced Illness on Patients, Surrogates, and Healthcare Costs: A Randomized Trial

To examine associations between management training of physician executives and their leadership styles, as well as effectiveness in achieving disease management goals. 

Executive directors of community health centers (269 respondents; response rate = 40.9%) were surveyed regarding their perceptions of the medical director's leadership, and for quantitative information on the center's achievement of clinical (mostly disease management) goals. The dependent variables were the medical director's scores (as perceived by the executive director) on transformational, transactional, and laissez-faire leadership, effectiveness, satisfaction with the leader, and subordinate extra effort, using an adapted Multifactor Leadership Questionnaire (43 items; 5-point Likert scale). The independent variable was the medical director's management training status. 

Compared with medical directors with <30 days of in-service training, medical directors with an MHA, MPH, or MBA, or =30 days of in-service training, had 0.32, 0.35, 0.30, 0.36, and 0.37 higher scores on transformational leadership, transactional leadership, rated effectiveness, satisfaction, and subordinate extra effort, respectively, and 0.31 lower score on laissez-faire leadership (all 

< .001). Medical directors without management degrees but with =30 days of in-service training had 0.34, 0.36, 0.50, and 0.47 higher scores on transformational leadership, transactional leadership, rated effectiveness, and satisfaction with the leader (all 

 < .02). Our data previously had demonstrated that medical directors' transformational leadership significantly influences achievement of disease management goals. 

Training may enable physician executives to develop leadership styles that are effective in influencing clinical providers' adoption of disease management guidelines under managed care. 

Translating research findings into sustainable improvements in clinical and patient outcomes remains a substantial obstacle to improving the quality and safety of healthcare.

Although each physician individually may not vary much in his or her practice pattern across patients, a group of physicians practicing 

compound the institution-level variation by the number of physicians in the group.

Variations in practice patterns often manifest as cost or quality problems. Practice variations arise because medicine is not an exact science, often involving judgment-driven application of a basic knowledge of the medical sciences. Evidence-based guidelines and protocols, illustrated by disease management programs, represent a critical managed care strategy to minimize undesirable practice variations. 

With more than 75% of healthcare decisions driven by physicians,

influencing physician practice patterns is critical. Lane and Ross observed that "physician leadership is critical for attaining balance among conflicting pressures for quality of care versus cost containment, prevention versus high technology medical intervention, and application of specialized versus primary care."

Kirz noted that "the challenge for physician leaders will be to balance the needs of professionals with the business and accountability requirements of a permanently competitive and resource constrained service industry."

The challenge for managed care organizations is to facilitate high-quality, evidence-driven care that also is cost effective and satisfying to patients. 

Physician leadership can be a vital complement to evidence-based practice guidelines and computerized reminders for influencing provider practice patterns.

 However, there is no clarity on "how to best assist the future medical staff" in acquiring the needed leadership competencies.

It is also not clear whether training actually impacts leadership styles or the achievement of managed care goals that require organization-wide adoption of evidence-based medical practices. Lacking such empirical evidence, not many organizations commit resources for physician leadership development. 

A pilot study has shown that federally funded Community Health Centers (CHCs) whose medical directors use a transformational leadership style have superior disease management outcomes.

CHCs are the nation's healthcare safety net, providing care to the nation's underserved, rural, and inner-city populations. Disease management collaboratives for selected, high-volume, chronic medical conditions represent managed care initiatives by the US Health Resources and Services Administration (HRSA) to improve CHCs' clinical performance. Under each collaborative, population-based goals are established (eg, a target mean value for glycosylated hemoglobin [A1C], representing consistent blood sugar control among diabetics), along with care protocols and center-wide processes to achieve the goals. This paper examines the association between the CHC medical directors'management training and their leadership style, as well as effectiveness in achieving disease management goals. 

Data on the leadership styles of CHC medical directors were collected in fall 2002 by anonymous mail survey of executive directors of all 663 CHCs in the contiguous United States, with a response rate of 40.9% (269 respondents). Respondents rated their medical director's leadership behaviors and effectiveness, using an adapted Multifactor Leadership Questionnaire (MLQ-5X Short).

based on a pilot survey of South Carolina CHCs and comments from national experts. Respondents were asked to judge how frequently each statement fit the medical director on a 5-point Likert-like scale (0 = not at all; 4 = frequently, if not always). We also requested information on the center's clinical performance and the medical director's demographics, specialty, and management training status. Survey solicitation was supported by a letter from the National Rural Health Association. The study was approved by the University of South Carolina's institutional review board. 

Past studies have shown that self-ratings are invariably inflated, and are poorly correlated with subordinate and supervisor ratings.

Intuitively, subordinate ratings would seem best suited to measure an executive's leadership style and effectiveness. However, subordinate and supervisor ratings are highly correlated.

The logistics and political acceptability of an exploratory study favored surveying supervisors (executive directors) rather than subordinates. CHC medical directors are salaried employees, accountable to their executive directors, who in turn, are accountable to HRSA for the CHC's performance. Executive directors rely on the medical directors' clinical leadership to complement their own organizational leadership role to achieve the center's clinical service goals. They may send their medical directors for in-service training organized or recommended by the National Association of Community Health Centers (NACHC) or the National Rural Health Association. Under the circumstances, the executive director's responses should be accurate concerning the medical director's leadership style and management training exposure. 

Bass and Avolio's Full Range of Leadership model classifies all interpersonal managerial behaviors into 3 styles: transformational, transactional, and laissez-faire leadership. Transformational leadership deals with influencing subordinates to act on hitherto latent (higher) motivations for exceptional performance and ethically inspired goals, transcending self-interest. It has 4 components: idealized influence, inspirational motivation, intellectual stimulation, and individualized consideration. Its distinctive emphasis is on stimulating followers' intrinsic needs for intellectual and motivational growth toward higher goals, followed by facilitation of such growth through individualized mentoring or coaching relationships. Transformational leadership appears to be uniquely positioned to address the managed care challenge of ushering in widespread adoption of evidence-based practices by providers to achieve population-based disease management goals. Providers have to balance fiduciary accountability to the patient with the collective fiscal accountability demands of managed care organizations, while adjusting individual treatment plans for unique biological, cultural, and psychological needs. 

Transactional leadership is establishing a credible social exchange process of rewards for performance with subordinates. It has 2 components: contingent reward and management-by-exception. Contingent reward is exchanging valued (verbal or tangible) rewards for good performance. Management-by-exception deals with intervening when standards or performance goals are not met. A laissez-faire leader is perceived as indifferent to follower actions and organizational outcomes, and avoiding responsibility to address important issues. 

Survey item scores were aggregated based on findings from analysis of exploratory factors to produce transformational, transactional, and laissez-faire leadership style scores, as well as perceptions regarding the medical director's effectiveness, satisfaction with the leader, and subordinate extra effort.

We evaluated associations of the 3 leadership styles and the 3 subjectively rated measures of leader effectiveness (dependent variables) with the medical director's management training status (the key independent variable), using multiple regression analysis. 

Management training was operationalized in 2 formats in order to distinguish between formal degree programs in public health or management (MPH, MHA, or MBA) and in-service training. Most in-service training is provided or organized by NACHC or the National Rural Health Association. Two questions were used to gain information on training status: 

Does the medical director have an MBA/MPH/ MHA/PhD in management? (Response categories: yes, no) 

Total number of training days at management training workshops/seminars in the last 3 years/during your tenure.

(Response categories: 0, 1-7, 8-29, 30-89, and 90+ days). 

Physicians with formal degrees have sacrificed earning opportunities and made additional financial investment in such education, and may be differently motivated for leadership roles. Selection bias could invalidate conclusions about associations between leadership styles and management education. To address these issues, we defined a second variable as "in-service training" to examine the effect of in-service training among medical directors without a management degree. In-service training is received without significant personal investment and often is initiated by the executive director. We also accounted for training duration, which may impact the assimilation and practice of leadership concepts. 

For statistical analysis, medical directors were classified in 2 ways. The variable "management education" classified medical directors into 2 categories: those with an MHA, MPH, or MBA or =30 days of in-service training, and those without a management degree and with = 29 days of in-service training. (There was no difference in style and effectiveness scores between those with a degree and those with = 30 days of in-service training. Moreover, statistical power was lost when these 2 groups were separated.) 

The second variable, "in-service training," excluded medical directors with management degrees and classified the remaining medical directors into 3 categories: =30 days of in-service training (category 1); 8-29 days of in-service training (category 2); and 0-7 days of in-service training (category 3). The thresholds were based on a combination of reasoning and statistical power of the data. Intuitive expectancy suggests that fewer than 7 days of training may not establish leadership competencies. Initial in-service training focuses on procedural issues, with marginal exposure to management and leadership. In comparison, =30 days of in-service training implies various types of training, providing significant exposure to and reinforcement of leadership and personal management concepts. 

New medical directors' training starts with a day of orientation training, devoted to job descriptions and CHC procedures. During the next year, it is recommended that they receive a week's training on "managing ambulatory care," which provides introductory management training with some leadership content. Other training sessions are provided at state and national conferences. This training pattern favors 0-7 days of in-service training as a category for medical directors who are less likely to have developed leadership competencies (10.5% of medical directors had 0 days of training, and 37.1% had 1-7 days of training). 

shows that only 1.6% had =90 days of training, requiring that the top 2 categories be collapsed into a =30-day category for statistical viability. The 0-to 7-day category offers a statistically viable contrast to the =30-day group. The group in between serves as a buffer to separate 2 "extreme" groups for statistically robust comparisons. 

We also tested the association between training and percent achievement of the center's priority clinical goals, calculated based on targeted and actual status of the CHC's clinical service goals.

Of 269 respondents, a subset of 68 executive directors provided usable information to calculate the degree of clinical-goal achievement. The reported information related to disease management goals for diabetes (A1C levels), prenatal care, mental health, and the elderly. 

Table 1 provides information on the demographic characteristics, leadership styles, and management training of the surveyed medical directors and executive directors. Executive directors had a mean tenure as executive directors of 9.17 years; their mean total tenure at the CHC was 11.2 years. Medical directors received the highest scores for transformational leadership (2.95 Â± 0.68) and the lowest for laissez-faire leadership (1.31 Â± 0.79). 

Of 248 medical directors with training information, 31 (12.5%) had an MHA, MPH, or MBA degree. Among those with a management degree, 2 (6.7%) had 1-7 days of in-service training, 17 (31%) had 8-29 days, and 11 (8.9%) had = 30 days. Among those without a management degree, 47.6% had either zero or 1-7 days of in-service training, 30.4% had 8-29 days, and 8.8% had = 30 days. 

Management Training Versus Leadership Style

General linear modeling using Tukey's test of significance was applied to examine associations between leadership style and management education/training. Specialty, age, sex, tenure as the CHC medical director, and total tenure at the CHC were used as control variables. Age served as a control for clinical experience, and tenure as the medical director for management experience. We also controlled for the durational impact of interpersonal associations (including before becoming the medical director) by using total tenure at the CHC as a control variable. None of the control variables were significant, singly or in combination. To conserve statistical power, these insignificant variables were deleted. Management education and in-service training were used, respectively, as single independent variables to examine whether management training predicted leadership styles and effectiveness. 

shows that medical directors with an MHA, MPH, or MBA degree or = 30 days of in-service training were significantly more likely to have a transformational leadership style (0.32 higher mean score) than those without a degree and with fewer than 30 days of in-service training. The former group also were significantly more likely to have a transactional leadership style (0.35 higher mean score), and significantly less likely to have a laissez-faire leadership style (0.31 lower score). 

shows that medical directors with a management degree or = 30 days of in-service training also were perceived as more effective (0.30 higher rated effectiveness score), and as eliciting extra effort from subordinates (0.37 higher subordinate extra effort score). Their executive directors were more satisfied with their role performance (0.36 higher score on satisfaction with the leader). Because of insufficient numbers, we could not test for the effect of each type of degree (MHA, MPH, or MBA). 

Table 2 shows that among medical directors with in-service training only, those with = 30 days were significantly more likely to have a transformational leadership style (0.34 higher mean score) than those with negligible or no in-service training (0-7 days). Medical directors with = 30 days of in-service training also were significantly more likely to have a transactional leadership style (0.36 higher mean score) than those with negligible or no in-service training. Their lower score on laissez-faire leadership (0.22 lower) was however, not statistically significant. They also were perceived as significantly more effective than those with negligible or no in-service training (rated 0.50 higher on effectiveness and 0.47 higher on satisfaction with their leadership; see Table 3). For subordinate extra effort, the difference (0.36 higher score) was not statistically significant. 

Statistical analysis showed insufficient power to detect differences in achievement of disease management goals due to management education and in-service training. 

This study examined empirical associations between CHC medical directors' management education and their supervisors' perceptions of their leader-ship style and effectiveness. The survey questions on management education captured information on formal degrees in management and public health, as well as duration of in-service training. We found that those with either a management degree or significant in-service training were perceived to rank significantly higher than those without training on those behaviors collectively defined as transformational leadership. 

Earlier, we documented an empirical association between transformational leadership and clinical performance of the center with respect to chronic disease management and prenatal care outcomes.

Our current analysis had inadequate statistical power to assess the direct association between training and the center's care management performance. We have the 2 concurrent sets of findings: (1) that leadership style predicts subjective measures of effectiveness as well as the centers' clinical performance and (2) that management training predicts leadership style as well as subjective measures of effectiveness. Together, they suggest that management training of physician executives may significantly impact their ability to influence providers' clinical performance in the desired direction, an important ingredient for the success of managed care. Transformational leadership has been documented to enable superior organizational performance in almost all settings that it has been tested, examples being industrial research teams,

It could be argued that the executive director's prior knowledge of his or her medical director's training may predispose the executive director to attribute leadership effectiveness. However, not only were medical directors with more training perceived as more effective, they also were perceived to manifest behaviors that collectively constitute transformational leadership, a style that is validated to produce superior outcomes in a variety of settings, as outlined above. Importantly, transformational leadership correlates with objective measures of CHCs' clinical performance, as reported earlier.

Therefore, it appears unlikely that single-source response bias drove our finding of an association between the medical directors' training status and leadership styles as well as effectiveness. 

There is no documentation on associations between healthcare managers' training and their leadership styles as perceived by supervisors or coworkers in the organization. Past studies on training effectiveness have examined self-assessed behavioral changes, as well as self-assessed managerial effectiveness.

February 2006

Improvements in Identifying Persistent Asthma Patients Using HEDIS Data

A Comparison of Comorbidity Measurements to Predict Healthcare Expenditures

Effects of a Program for Coordinated Care of Advanced Illness on Patients, Surrogates, and Healthcare Costs: A Randomized Trial

Management Training of Physician Executives, Their Leadership Style, and Care Management Performance: An Empirical Study

Managing the Metric vs Managing the Patient: The Physician's View of Pay for Performance

Practical Design and Implementation Considerations in Pay-for-performance Programs