: To evaluate measurement of physician quality performance, which is increasingly used by health plans as the basis of quality improvement, network design, and financial incentives, despite concerns about data and methodological challenges.

: Evaluation of health plan administrative claims and enrollment data.

: Using administrative data from 9 health plans, we analyzed results for 27 well-accepted quality measures and evaluated how many quality events (patients eligible for a measure) were available per primary care physician and how different approaches for attributing patients to physicians affect the number of quality events per physician.

: Fifty-seven percent of primary care physicians had at least 1 patient who was eligible for at least 1 of the selected quality measures. Most physicians had few quality events for any single measure. As an example, for a measure evaluating appropriate treatment for children with upper respiratory tract infections, physicians on average had 14 quality events when care was attributed to physicians if they saw the patient at least once in the measurement year. The mean number of quality events dropped to 9 when attribution required that the physician provide care in at least 50% of a patientâ€™s visits. Few physicians had more than 30 quality events for any given measure.

: Available administrative data for a single health plan may provide insufficient information for benchmarking performance for individual physicians. Efforts are needed to develop consensus on assigning measure accountability and to expand information available for each physician, including accessing electronic clinical data, exploring composite measures of performance, and aggregating data across public and private health plans.

Available administrative data for a typical health plan may provide insufficient information for benchmarking performance among individual physicians.

For any single quality measure, most physicians had few quality events attributed to them even using a less stringent rule for attributing patient measures to physicians. 

Efforts are needed to encourage the aggregation of databases across public and private health plans and to maximize the data available for characterizing physician perfor

Measurement of physician quality performance is increasingly used by health plans as the basis for quality improvement, network design, and financial incentives.

 Still, efforts to measure physician performance face a number of challenges, in particular the need for sufficient sample size to support reliable measurement and the lack of consensus on methods for attributing patient measures to clinicians.

Researchers have noted that measurement and comparison of physician quality can be hampered by sample size.

 A minimum threshold of 30 patients is a common guideline for supporting comparisons for an individual measure,

 and evidence suggests that at least 35 to 45 observations are needed to make valid comparisons.

 One challenge in obtaining sufficient sample size relates to the measure itself. Many quality measures describe a select group of patients and, by definition, will yield a small number of patients for any physician. Other measures apply to larger proportions of patients, but the ability to capture information on a physician&#8217;s entire panel of patients is limited (as when performance measurement relies on data from a single health plan).

A related issue in quality measurement is attribution. Which physicians should be responsible for a quality measure? Given the current focus on team-based chronic disease care and the reality that most patients receive care from multiple clinicians,

 some authors argue that the most appropriate level of accountability is not the individual physician but rather a formal or informal group of physicians.

 Healthcare organizations often attribute patient quality measures based on utilization or a specific set of services, despite the challenges in identifying which physician should be held responsible for the fulfillment (or lack of fulfillment) of a quality measure.

Efforts are needed to understand how these issues may affect the meaningfulness and soundness of physician profiling efforts. In this study, we used a data set that is typical of the information used by health plans to characterize physician performance. Using 27 well-accepted measures that can be obtained from administrative data, we evaluated (1) how many quality events were available per physician and (2) how different attribution rules affect the number of quality events. 

We focused on 27 measures describing acute, chronic, and preventive care activities performed by PCPs. Only measures that could be obtained through administrative claims data were included. 

) lists all quality measures used in this study, as well as the period used to attribute patients and quality events to physicians.

We identified physicians by the unique identifiers used by each health plan. Primary care physicians, including family physicians, general internists, and general pediatricians, were identified based on their specialty designated in health plan credentialing records.

In selecting an attribution approach, we considered the interactions between clinicians and patients in the course of delivering care, the kinds of services involved, the evidence of a physician&#8217;s involvement in the patient&#8217;s care, and the data sources available. For this study, we applied a measure-specific attribution logic based on administrative data. Measures were attributed to PCPs based on the outpatient visits they provided to patients during a prescribed time frame specific to each measure. Visits were defined using Healthcare Effectiveness Data and Information Set codes for preventive and ambulatory health services.

 To test a less stringent approach to attribution, a patient measure was attributed to a physician if the patient had 1 or more visits during the prescribed time frame. In addition to this &#8220;1-visit&#8221; rule, 2 more stringent rules were assessed: a PCP was attributed responsibility for a patient&#8217;s measure (1) if the patient completed at least 30% of his or her ambulatory visits with that physician (30% rule) and (2) if the patient completed at least 50% of his or her ambulatory visits with that physician (50% rule).

A quality event occurred each time a patient was eligible for a quality measure. Therefore, a single patient could contribute multiple quality events if he or she was eligible for multiple measures (eg, preventive screening and another measure).

Overall, 57% of 170,168 PCPs represented in the study claims data could be attributed responsibility for at least 1 quality event (ie, &#8805;1 of their patients was eligible for &#8805;1 of our selected quality measures). 

 summarizes findings based on the 1-visit rule and describes the percentage of PCPs with more than 30 quality events for a measure. Except for preventive measures, few PCPs had more than 30 observations for any given measure. However, these high-volume providers account for a larger share of quality events overall, particularly for preventive care measures. For example, only 17% of physicians had more than 30 quality events for colorectal cancer screening, but these physicians accounted for 78% of the quality events for this indicator. Only 1% of physicians had more than 30 quality events for annual glycosylated hemoglobin testing among patients with diabetes mellitus, but they accounted for 16% of the quality events for this measure.

 summarizes how moving from a less stringent rule to a more stringent rule for attribution affects the number of patients available for characterizing physician performance. For example, using the 1-visit rule for the measure assessing appropriate care for upper respiratory tract infections in children, physicians on average had 14 eligible patients for that measure in the measurement year. The mean number of quality events dropped to 11 when care was attributed using the 30% rule and to 9 when care was attributed using the 50% rule. Relative to the 1-visit rule, the 50% rule reduced by about half the number of quality events per physician for a measure. Adopting a more stringent rule for a measure also reduced the number of PCPs with at least 1 quality event for that measure (data not shown).

Limitations of this study included the number of measures studied, the reliance on administrative data only, and the lack of direct information about the physician&#8217;s relationship with the patient. These findings are based on only 27 quality measures from administrative data. However, all are well tested and nationally endorsed, and most are included in health plans&#8217; and employers&#8217; physician performance measurement programs for PCPs. Administrative data were used because most physician-level measurement efforts around the country rely on these data. However, administrative data limit the type of clinical actions that can be profiled.

 These limitations demonstrate the issues that health plans often face in developing meaningful provider profiles. Finally, the study describes findings for PCPs only. Within the context of this study, we found similar challenges in achieving sample sizes that provided more than 30 quality events for specialist physicians.

As our results demonstrate, several practical steps are needed to ensure that physician profiles based on administrative data have sufficient information for reliable estimates. First, pooling administrative data within communities across all health plans, government purchasers, and other entities is critical to construct a more complete database representing most or all care rendered by that community&#8217;s physicians. Regional and national quality initiatives promoted by the Centers for Medicaid & Medicare Services14 and by the Robert Wood Johnson Foundation

Second, composite measures should be considered. Care should be taken in selecting and weighting the individual measures for inclusion in a composite. Furthermore, the use of composites creates additional challenges in interpreting quality results and specific actions for improving care. However, composites constructed around a particular condition or patient care activity may provide insights into quality performance and increase the number of quality events available for comparing providers.

Third, efforts are needed to encourage physician practices, health plans, and other entities to make readily available more clinically detailed data for quality measurement construction. Initially, the more widespread availability of electronic data that is already present in some settings (eg, laboratory results) will allow for a larger number of potentially more meaningful quality measures to be constructed without the expense of medical record review, and efforts are critically needed to improve the capabilities of electronic medical records to report quality measures.

 Efforts to augment routinely available administrative data by including additional codes (eg, 

) that capture critical information about outcomes or results of patient care may be promising but have not yet been tested for accuracy or reliability in widespread applications, to our knowledge.

2. Landon BE, Normand SL, Blumenthal D, Daley J

. Physician clinical performance assessment: prospects and barriers. 

4. Hofer TP, Hayward RA, Greenfield S, Wagner EH, Kaplan SH, Manning WG

. The unreliability of individual physician â€œreport cardsâ€ for assessing the costs and quality of care of a chronic disease. 

. Measuring patientsâ€™ experiences with individual primary care physicians: results of a statewide demonstration project. 

8. Pham HH, Schrag D, Oâ€™Malley AS, Wu B, Bach PB

. Care patterns in Medicare and their implications for pay for performance. 

http://www.ingenix.com/Products/Employers/HealthandProductivity/EvidenceBased

. Reliability of individual and composite measures for profiling physician performance. 

. A comparison of administrative-only versus administrative plus chart review data. 

. The Robert Wood Johnson Foundation: health and health care improvement. 

17. Casalino LP, Elster A, Eisenberg A, Lewis E, Montgomery J, Ramos D

. Will pay-for-performance and quality reporting affect health care disparities? 

. Pay for performance at the tipping point. 

Articles

TRENDS FROM THE FIELD

Typical health plan data provide limited information for benchmarking physician performance using even a less stringent rule for attributing patient measures to physicians.

Availability of Data for Measuring Physician Quality Performance

To evaluate the association and concordance of the new 8-item self-report Morisky Medication Adherence Scale (MMAS) with pharmacy fill data in a sample of community-dwelling seniors with hypertension.

 Pharmacy records for antihypertensive medications were abstracted for 87 managed care adult patients with hypertension 65 years and older who completed a survey that included the MMAS. Continuous single-interval medication availability (CSA), medication possession ratio (MPR), and continuous multiple-interval medication gaps (CMG) were calculated using pharmacy data. The MMAS adherence was categorized as high, medium, and low (MMAS scores of 8, 6 to <8, and <6, respectively); pharmacy fill nonpersistence was defined as less than 0.8 for CSA and MPR and as greater than 0.2 for CMG.

 Overall, 58%, 33%, and 9% of participants had high, medium, and low medication adherence, respectively, by the MMAS. After adjustment for demographics and in comparison to high adherers on the MMAS, patients with low MMAS adherence were 6.89 (95% confidence interval [CI], 2.48-19.10) times more likely to have nonpersistent pharmacy fill adherence by CSA and were 5.22 (95% CI, 1.88-14.50) times more likely to have nonpersistent pharmacy fill adherence by MPR. Concordance between the MMAS and CSA, MPR, and CMG was 75% or higher.

 The MMAS is significantly associated with antihypertensive drug pharmacy refill adherence. Although further validation of the MMAS is needed, it may be useful in identifying low medication adherers in clinical settings.

Compared with pharmacy fill, the self-report Morisky Medication Adherence Scale (MMAS) performed well in identifying patients with low adherence to antihypertensive medication regimens.

Although pharmacy fill rates represent an objective assessment of medication adherence, they are impractical for real-time use in most clinical settings. 

The MMAS tool is simple and economical to use in routine outpatient settings and may provide clinicians and administrators with important information to guide treatment decisions for patients with hypertension.

Patient surveys were conducted from December 2002 to March 2003 using a standardized data collection instrument. The survey data (including sociodemographic data and medication adherence) were entered into a Microsoft Access database (Microsoft Corporation, Redmond, WA) and were transferred to SAS version 9.1.3 (SAS Institute, Inc, Cary, NC) for analysis; quality check revealed less than 1% data entry error. All patient identification information was collected and maintained according to Health Insurance Portability and Accountability Act of 1996 regulations and health plan privacy rules.

The managed care organization&#8217;s data warehouse system was the source of the pharmacy fill data for the present study. The data warehouse (an Oracle relational database [Oracle Corporation, Redwood Shores, CA]) was populated with historic claims data, patient roster data, diagnosis and procedural codes, and code descriptions. Data were extracted by informatics analysts using the Oracle Discoverer tool and were transported into SAS version 9.1.3. The pharmacy data were abstracted on 87 patients and included 42 different antihypertensive medications with 1578 fills captured in the study period.

Pharmacy fill data were extracted for the 2002 calendar year and included a listing of all antihypertensive prescriptions filled, the date filled, generic and brand names of the drugs, and number of pills dispensed. The following 3 measures of adherence were calculated: continuous single-interval medication availability (CSA), medication possession ratio (MPR), and continuous multiple-interval medication gaps (CMG).

The CSA was calculated by dividing the days&#8217; supply obtained at a pharmacy fill by the number of days before the next pharmacy fill for that same medication. The MPR was calculated as the sum of the days&#8217; supply obtained between the first pharmacy fill and the last fill (the supply obtained in the last fill was excluded), divided by the total number of days in this period. The CMG was calculated by dividing the total number of days without medications (ie, treatment gaps) between the first and last pharmacy fills by the number of days in this period. A graphical example of how CSA, MPR, and CMG were calculated is provided in 

For every participant, CSA was calculated for each pharmacy fill interval, and MPR and CMG were calculated by class of antihypertensive medication being taken. Values greater than 1 were truncated at the maximum value of 1 for CSA and MPR.

 Given that self-reported adherence reflects adherence to participants&#8217; antihypertensive medication regimen, 1 CSA was assigned to each participant based on the mean of all CSAs calculated from all of his or her antihypertensive drug pharmacy fill intervals. One MPR and 1 CMG were assigned to each participant. For participants filling more than 1 class of antihypertensive medication, MPR and CMG were calculated for each class and then averaged across all classes to assign a single MPR and CMG to each participant. Given that a cut point of 0.8 has been previously used to define adequate medication adherence using pharmacy data,

pharmacy fill nonpersistence was defined as less than 0.8 for CSA and MPR and as greater than 0.2 for CMG. The continuous single-interval gap statistic is the inverse of CSA and is not presented.

Of 87 patients included in the study, the mean age was 76 years, 31% were men, 48% were black, 47% had graduated from high school, 43% were married, 43% smoked cigarettes, and the mean number of antihypertensive medications being taken was 2.2 (range, 1-4 medications). The mean (SD) MMAS score was 7.4 (0.9). Demographic characteristics of the study participants by MMAS category are given in 

. A significantly higher percentage of black versus white patients were low adherers by the MMAS. There were no other significant differences in patient demographics across MMAS categories. By self-report using the MMAS, the distribution of participants&#8217; adherence to prescribed medication use in this study population was 58% high adherence, 33% medium adherence, and 9% low adherence. The distributions of the pharmacy fill adherence measures are shown in the Figure. The median CSA, MPR, and CMG were 0.91, 0.91, and 0.12, respectively (

). However, 23% and 25% of patients had nonpersistent CSA and MPR, respectively, and 35% had nonpersistent CMG.

Association of the MMAS With Nonpersistence as Determined by Pharmacy Fill

Hypertension, a public health challenge in the United States and worldwide, is a modifiable risk factor for cardiovascular events.

Clinicians require information on antihypertensive medication adherence to draw proper conclusions about the effectiveness of treatment.

 The goal is to have access to a quick, reasonably accurate self-report adherence measure for use in outpatient settings to facilitate clinical decision making. We evaluated the accuracy of a new self-report measure by assessing its association and concordance with pharmacy fill rates. To our knowledge, the concordance between the MMAS and pharmacy fill for antihypertensive medications has not been previously evaluated. In this study, the MMAS maintained a strong, graded, statistically significant association with pharmacy fills. Using CSA, MPR, and CMG for comparison, the MMAS correctly classified at least 75% of patients as being adherent or not. Furthermore, patients classified as low adherers by the MMAS were significantly more likely to be nonpersistent by each measure of pharmacy fill. Therefore, the MMAS may be a practical and valid approach for identifying low adherers to chronic medication regimens in outpatient settings.

Low adherence poses unique challenges for clinicians trying to determine if prescribed treatment is effective. If clinicians are able to accurately identify patients with low adherence, then appropriate and timely interventions can be implemented. Several modifiable factors have been reported to negatively affect adherence to prescribed therapies. These include forgetfulness,

 lack of knowledge regarding hypertension and its treatment,

 healthcare system perceptions by the patient,

 The MMAS provides information on behaviors associated with low adherence that may be unintentional (eg, forgetfulness) or intentional (eg, not taking medications when one feels worse). Identification of these behaviors can facilitate tailoring of interventions to specific patient issues.

 For example, if a patient is identified as a low adherer by the MMAS and the responses indicate forgetfulness as a major barrier, then the clinician may suggest that the patient use weekly pill boxes and engage a family member or friend to assist with medication reminders. If a patient is identified as a low adherer and responds that she stops taking medications when she feels better or worse, then the clinician can address knowledge barriers and medication adverse effects and educate the patient about the chronic nature of hypertension and the importance of taking medication as prescribed. On the other hand, if patients are found to be high adherers and their blood pressure remains uncontrolled, then the clinician should consider increasing medication dosage or adding a medication to their regimen.

It may be that the use of a simple tool (eg, the MMAS) in the outpatient setting may allow clinicians to eliminate low adherence as a contributing factor to poor blood pressure control.

Although the concordance between self-report and pharmacy fill was good in this study, it was not perfect. Shortcomings of self-report include reliance on recall and social desirability bias, with a tendency to overestimate adherence.

 In addition, pharmacy fill rates may not capture some nuances of medication adherence behavior

 and are not practical to capture in real time among outpatient clinical encounters. It is generally assumed that patients who fill medications also take them unless they have been instructed by their provider otherwise or have adverse effects that limit their medication intake.19 It is possible that patients&#8217; medication adherence varies by drug class. In the present study, 21 patients had 1 or more drug classes with an MPR of less than 0.8 and had an MPR of 0.8 or higher for other drug classes. Even after accounting for this using a generalized estimating equation, a strong and statistically significant association between selfreported medication adherence and MPR and CMG nonpersistence pharmacy fill rates remained present (data not shown). This suggests that averaging the pharmacy measures did not mask any drug class&#8211;specific relationships.

 reported that more than 20% oversupply (ie, MPR, &#8805;1.2) is related to a similar risk of future hospitalization as that experienced by patients with an undersupply of medication (ie, MPR, <0.8). In the present study, only 5 patients had a mean MPR of 1.2 or higher, none of whom had low adherence on the MMAS. After excluding these 5 patients, the associations between nonpersistence and self-report medication adherence were similar to those of the original analysis. When we reclassified these patients as nonpersistent, the results were also similar (data not shown).

There are several important clinical implications of our findings. Compared with pharmacy fill, the selfreport MMAS performed well in identifying patients with low adherence to antihypertensive medication use. These patients are likely at greatest risk for uncontrolled blood pressure and subsequent adverse outcomes and could benefit most from tailored interventions to overcome barriers to adherence. Although pharmacy fill data represent an objective assessment of medication adherence, this is impractical for use in clinical settings and does not provide information on reasons for low adherence. The MMAS tool is simple and economical to use in routine outpatient settings and may provide clinicians with important information (ie, barriers to adherence) to guide treatment decisions for patients with hypertension. Scores on the MMAS maintained a strong graded association with antihypertensive drug pharmacy fill adherence in community-dwelling seniors receiving healthcare through a managed care organization. This association suggests that patients&#8217; self-report of adherence behavior is consistent with the rate at which they fill their antihypertensive medications. The present study extends prior work demonstrating the internal reliability and predictive validity of the MMAS with respect to blood pressure control. This 8-item tool is simple and feasible to incorporate into clinical practice and may be useful in identifying patients at risk for medication adherence issues, including low adherence in outpatient settings. 

: From the Center for Health Research (MK-W), the Department of Cardiology-Hypertension Section (RNR), Ochsner Clinic Foundation, New Orleans, LA; the Department of Epidemiology (MK-W, TI), the Department of Family and Community Medicine (MK-W), the Department of Biostatistics (LSW), Tulane University, New Orleans, LA; the Department of Community Health Sciences (DEM), University of California, Los Angeles, Los Angeles, CA; and the Department of Community and Preventive Medicine (PM), Mount Sinai School of Medicine, New York, NY.

: The authors (MK-W, TI, LSW, RNR, DGM, PM) report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article. 

: The project described was supported in part by grant R01 AG022536 from the National Institute on Aging and by the Ochsner Clinic Foundation. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute on Aging or the National Institutes of Health.

: Concept and design (MK-W, RNR, DEM, PM); acquisition of data (MK-W); analysis and interpretation of data (MK-W, TI, LSW, RNR, PM); drafting of the manuscript (MK-W, LSW, DEM, PM); critical revision of the manuscript for important intellectual content (TI, LSW, PM); statistical analysis (TI, LSW, PM); obtaining funding (MK-W); and supervision (MK-W).

: Marie Krousel-Wood, MD, MSPH, Center for Health Research, Ochsner Clinic Foundation, 1514 Jefferson Hwy, New Orleans, LA 70121. E-mail: 

1. Ong KL, Cheung BM, Man YB, Lau CP, Lam KS

. Prevalence, awareness, treatment, and control of hypertension among United States adults 1999-2004. 

3. Krousel-Wood MA, Thomas S, Muntner P, Morisky DE

. Medication adherence: a key factor in achieving blood pressure control and good clinical outcomes in hypertensive patients. 

. Detection methods and strategies for improving medication compliance. 

7. Schiff GD, Fung S, Speroff T, McNutt RA

. Decompensated heart failure: symptoms, patterns of onset, and contributing factors. 

9. Grymonpre R, Cheang M, Fraser M, Metge C, Sitar DS

. Validity of a prescription claims database to estimate medication adherence in older persons. 

11. Morisky DE, Green W, Levine DM, et al

. Concurrent and predictive validity of a self-reported measure of medication adherence. 

13. Morisky DE, Ang A, Krousel-Wood MA, Ward H

. Predictive validity of a  medication adherence measure in an outpatient setting. 

15. Garber MC, Nau DP, Erickson SR, Aikens JE, Lawrence JB

. The concordance of self-report with other measures of medication adherence: a summary of the literature. 

. The assessment of refill adherence using pharmacy records: methods, validity, and applications. 

. Estimating medication adherence using administrative claims data. 

21. Kopjar B, Sales AE, Pi~neros SL, Sun H, Li YF, Hedeen AN

22. Simpson E, Beck C, Richard H, Eisenberg MJ, Pilote L

23. Fields LE, Burt VL, Cutler JA, Hughes J, Roccella EJ, Sorlie P

. The burden of adult hypertension in the United States 1999 to 2000: a rising tide. 

. Compliance bias and the interpretation of therapeutic trials. In: Haynes RB, Taylor DW, Sackett DL, eds. 

. Baltimore, MD: Johns Hopkins Press; 1979:309.

. Awareness, knowledge and attitudes of older Americans about high blood pressure. 

. Hypertension in adherence to long-term therapies: evidence for action. 2003. 

http://www.connectedhealth.org/programs/medication-adherence/research-materials

30. Wassertheil-Smoller S, Blaufox MD, Oberman A, et al

. Effect of antihypertensives on sexual function and quality of life: the TAIM study. 

32. Krousel-Wood MA, Thomas S, Jannu A, Muntner P, Morisky DE, Re RN

. Low adherence to prescribed antihypertension medication and poorer quality of life in elderly hypertensive patients. Poster presented at: American Heart Association 2nd Scientific Conference on Compliance in Healthcare and Research, Washington DC, May 18, 2004.

. Overcome clinical inertia to control systolic blood pressure. 

adherence and health care use among adults with hypertension in an urban health care system. 

METHODS

Scores on a new medication adherence scale maintained a strong graded association with antihypertensive drug pharmacy fill adherence among community-dwelling seniors in a managed care organization.

New Medication Adherence Scale Versus Pharmacy Fill Rates in Seniors With Hypertension

: To compare the cost of substitutive Hospital at Home care versus traditional inpatient care for older patients with community-acquired pneumonia, exacerbation of chronic obstructive pulmonary disease, exacerbation of congestive heart failure, or cellulitis.

: Prospective nonrandomized clinical trial involving 455 community-dwelling older patients in 3 Medicare managed care health systems and at a Department of Veterans Affairs medical center.

: Costs were analyzed across all patients, within each of the separate health systems, and by condition. Generalized linear models controlling for confounders and using a log link and gamma family specification were used to make inferences about the statistical significance of cost differences. 

Tests were used to make inferences regarding differences in follow-up utilization.

: The costs of the Hospital at Home intervention were significantly lower than those of usual acute hospital care (mean [SD], $5081 [$4427] vs $7480 [$8113]; 

 <.001). Laboratory and procedure expenditures were lower across all study sites and at each site individually. There were minimal significant differences in health service utilization between the study groups during the 8 weeks after the index hospitalization. As-treated analysis results were consistent with Hospital at Home costs being lower.

: Total costs seem to be lower when substitutive Hospital at Home care is available for patients with congestive heart failure or chronic obstructive pulmonary disease. This result may be related to the study-based requirement for continuous nursing input. Savings may be possible, particularly for care of conditions that typically use substantial laboratory tests and procedures in traditional acute settings.

The availability of substitutive Hospital at Home care was associated with apparently lower costs for the care of patients with exacerbation of chronic obstructive pulmonary disease or with exacerbation of congestive heart failure in 3 Medicare managed care settings and at a Department of Veterans Affairs medical center.

Laboratory and procedure costs were lower with Hospital at Home care.

Costs were lower for patients with chronic obstructive pulmonary disease or congestive heart failure but not for patients with community-acquired pneumonia.

Utilization in the 8 weeks following discharge were no different when Hospital at Home care was available.

Hospital at Home is a model of care that provides hospital-level care to a  patient in his or her home and substitutes entirely for an acute hospital admission.

Several secular trends favor the development of this model. First, the Institute of Medicine found that the emergency medicine system is in crisis in part because emergency departments are unable to find inpatient hospital beds to send acutely ill patients.

 Second, population and disease burden projections suggest that the supply of available acute hospital beds is not increasing at a rate sufficient to keep pace with national acute care needs.

 Third, Hospital at Home avoids the potentially hazardous effects of inpatient care such as functional decline,

Controversy persists regarding the clinical and economic effectiveness of Hospital at Home.

 Many models studied to date have been for early discharge of surgical patients rather than for substitutive care, and all previous studies have been in countries with single-payer healthcare systems. Cost implications of substitutive Hospital at Home from the perspective of third-party payers in the United States may be qualitatively different.

We have demonstrated that substitutive Hospital at Home care in the United States was clinically feasible, efficacious, and associated with reductions in complications, greater patient satisfaction, higher caregiver satisfaction, and lower caregiver stress.

We also reported summary data on lower third-party payer costs for Hospital at Home patients.

 The aim of this study was to make more detailed comparisons of thirdparty payer costs in providing substitutive Hospital at Home care versus traditional acute hospital care for older patients with acute medical illness. The primary hypothesis was that total costs would be lower for Hospital at Home patients. Specifically, procedure, room, and hospital (not physician) costs were hypothesized to be lower.

The Hospital at Home National Demonstration and Evaluation Study has been described in detail previously.

 The study was a prospective non-randomized clinical trial conducted in 2 consecutive 11-month phases. In the first (observation) phase, Hospital at Home—eligible patients were followed up through usual acute hospital care. In the second (intervention) phase, all Hospital at Home–eligible patients were offered Hospital at Home care. A randomized controlled trial was precluded because of regulations related to Medicare managed care.

The study was conducted at 3 sites. Univera Health and Independent Health, in Buffalo, New York (hereafter, Buffalo), are Medicare managed care plans that operate in an independent practice association model. The Fallon Health Care System, in Worcester, Massachusetts (hereafter, Fallon), operates a not-for-profit Medicare managed care plan and the Fallon Clinic, a for-profit multispecialty physician group that provides care on a capitated basis to Medicare managed care beneficiaries. The Portland Veterans Affairs Medical Center, in Oregon (hereafter, PVAMC), is a quaternary care and teaching facility.

Patients and the Hospital at Home Model of Care

Eligible patients were community-dwelling persons 65 years and older residing within a specific catchment area who required acute hospital admission for an exacerbation of chronic obstructive pulmonary disease (COPD) or chronic heart failure (CHF), for community-acquired pneumonia, or for cellulitis. Patients requiring acute hospital admission for one of the target conditions and who met previously validated Hospital at Home eligibility criteria were included.

 The most common reasons for medical ineligibility were uncorrectable hypoxemia, suspected myocardial ischemia, and the presence of an acute illness that required hospital admission other than the target illness. Patients were identified in an emergency department or at an ambulatory site at times when Hospital at Home could admit patients (usually between 6 am and 8 pm). After informed consent was obtained, patients were transported home via ambulance. Patients were evaluated by a Hospital at Home physician in the emergency department or shortly after arrival at home. A Hospital at Home nurse met the ambulance at home and provided initial direct one-on-one care for a mean of 16.9 hours. After direct nursing supervision, the patient had intermittent nursing visits at least daily. Patients were not required to have a caregiver in the home, and if a caregiver was available, there was no requirement that the caregiver provide assistance or care to the patient. A Lifeline medical alert device (Philips Electronics North America Corporation, Andover, MA) was placed in the home of any patient who did not have a caregiver present in the home. The Hospital at Home physician made at least daily home visits and was available at all times for urgent visits. Nursing and other care components such as durable medical equipment, oxygen therapy, skilled therapies, and pharmacy support were provided by a partner Medicarecertified home health agency and for some services (eg, home radiology) by independent contractors. Diagnostic studies such as electrocardiography, radiology, intravenous fluids, intravenous antimicrobial agents and other medicines, oxygen, and other respiratory therapies were provided at home.

Third-party payer cost data were obtained from each of the 3 study sites. The PVAMC provided costs based on a standard Veterans Affairs stepdown cost accounting approach. Buffalo provided charges, as well as a cost-to-charge ratio for hospital facility costs. Fallon provided allowable hospital charges at the level of the admission. Data from all sites include all resources used for patient care for inpatient settings and for Hospital at Home care. There is no reason to suspect that any single costing method would systematically favor the intervention.

Subtotals of costs were used for hypothesis testing. Hospital facility costs are separated from physician costs. Physician costs are divided into emergency department costs and non—emergency department costs. This is particularly relevant because the Hospital at Home admissions primarily originate in the emergency department. Hospital costs are also categorized based on UB-92 codes that Medicare and other third-party payers use for grouping charges.

 The groups were emergency (emergency department and clinic), room (all types of room charges), skilled nursing (only applicable to Hospital at Home patients), medications (any pharmacy or medication), laboratory tests and procedures (laboratory, pathology, imaging, cardiology, and pulmonary function), and other hospital charges.

We also considered postdischarge follow-up care and readmission. Ascertaining whether the medical reason for followup utilization was related to the index admission to Hospital at Home was a nontrivial matter. Therefore, we gathered data on all hospitalizations, emergency department visits, skilled nursing facility admissions, and home healthcare visits. We were limited to comparing quantities of postdischarge utilization because the care could be obtained at facilities other than those involved in the study.

Costs were compared for all patients in the observation and intervention groups, by site, and by condition across sites. Given the different costing methods and slight variations in implementation, comparisons by site are most valid. There were insufficient patients to make comparisons by condition by site.

Data on 286 observation patients and 169 intervention patients were collected between November 2000 and September 2002. Consumer price index data indicate that inflation during that time was more than 15% for hospital and related services and more than 6% for professional services. Costs were inflation adjusted to the end of the intervention period. Total costs in the initial results study

 were reported in 2002 dollars; they will be reported in the same year’s dollars herein for consistency.

Although we present means and standard deviations, inferences of statistically significant differences were made based on the results of a generalized linear model using a log link and gamma family specification in STATA 9.1 (Stata- Corp LP, College Station, Texas).13 When using all observations, we included age, being male, being white, living alone, the Acute Physiology and Chronic Health Evaluation (APACHE) II score, number of comorbidities, activities of daily living limitations at baseline, and Instrumental Activities of Daily Living limitations at baseline as explanatory variables. In the regressions using subsets of the whole data set (by site, by primary diagnosis, or as treated), we included only age, lives alone, and the APACHE II score. Despite the large number of comparisons, we did not adjust the 

 value denoting statistical significance but interpret the set of results together. Inferences regarding the statistical significance of differences in utilization during the 8 weeks after the index hospitalization were made based on the results of

The primary investigations were intent-to-treat analyses comparing all observation phase patients with all intervention phase patients. These results provide a conservative estimate of the potential differences because some patients at each site received usual hospital care during the intervention period. Some patients refused to receive Hospital at Home care. For others, admission occurred at times when Hospital at Home was not open for admissions; standard hours were between 6 am and 8 pm. The numbers who consented to data collection at each site during the intervention phase were 42 patients in Buffalo, 47 patients at Fallon, and 80 patients at PVAMC; the numbers who received Hospital at Home care were 30 patients, 10 patients, and 44 patients, respectively. Given the numbers who did not receive Hospital at Home and to allow for secular changes in care between the two 11-month periods, we also conducted secondary analyses using as-treated data limited to the intervention period. Given the potential for selection bias, the intent-to-treat and as-treated analyses bound the potential effect of Hospital at Home.

Demographic characteristics are given in 

. There are few statistically significant differences between the observation and intervention groups; the difference in APACHE II scores was not clinically significant. During the intervention period, patients treated at home had higher mean levels of comorbidity, as reflected in the Charlson score.

 gives the results of an intent-to-treat analysis of total hospital costs and UB-92—based groupings. Emergency department costs were statistically significantly higher when Hospital at Home care was available in Buffalo but were lower under the same conditions at Fallon. In contrast, room charges are statistically significantly lower (

 <.01 for all) when combining data across all sites and at each site. The mean (SD) costs for skilled nursing range from $279 ($712) at Fallon to $778 ($844) at Buffalo. Laboratory and procedure expenditures are statistically significantly lower across all sites and at each site (

 <.05 for all). Room costs were nonzero for those who were in the Hospital at Home group because some patients received usual hospital care and because 2 patients started their care in Hospital at Home but required transfer to the acute hospital to complete their care. Combining data from all sites, the mean (SD) length of stay during the observational period was 4.9 (9.9) days, and the length of stay during the intervention period was 3.2 (2.5) days.

 gives an intent-to-treat comparison of total costs (hospital and physician combined), separate hospital and physician costs, and separate emergency department and non— emergency department physician costs. Total costs of the Hospital at Home intervention were statistically significantly lower than those of usual hospital care during the observation period (mean [SD], $5081; 

 <.001). Only one of the individual sites had statistically significantly lower total costs during the intervention period, although all 3 sites had total costs that were lower. As hypothesized, across all sites hospital costs were more than $2000 lower among those who were treated when Hospital at Home care was available (

 <.001). The PVAMC was the only site with statistically significantly lower costs for both hospital and physician components of the total. The results for subcomponents of physician costs were as expected for comparisons using individuals across all sites. Emergency department physician costs were statistically significantly higher among patients who received care when Hospital at Home care was available (

 <.05). Non—emergency department physician costs were statistically significantly lower when combining data across sites (

 <.01). Other than for emergency department physician costs, site-specific comparisons mimicked the overall comparison or were not statistically significant. Emergency department physician costs were statistically significantly lower in Buffalo and statistically significantly higher at Fallon, while they were not statistically significantly different at PVAMC. This variability may be the result of how different emergency departments are managed.

 is similar to Table 1, but the columns represent conditions rather than sites. Room costs and laboratory test and procedure expenses during the intervention phase were statistically significantly lower than those during the observation phase for each condition.

 is analogous to Table 2 in the same way that Table 3 is analogous to Table 1. The direction of comparisons is similar across all 4 conditions for total costs and for separate hospital and physician costs. However, the differences are not always statistically significant. The hospital component of costs is statistically significantly different in 2 cases, while the physician component is never significantly different. The emergency department physician cost component is greater for patients when Hospital at Home care was available, although only one difference was statistically significant (

) summarizes the results in Table 1 across all sites using an as-treated analysis. These results suggest that in part the intervention costs were lower in Table 1 because the costs of patients who received usual hospital care were lower in the intervention period. However, the room and laboratory test and procedure costs were so much lower that the differences were statistically significant in the as-treated analysis. The group treated in the hospital had a longer mean (SD) length of stay (3.5 [2.8] days) compared with the group treated at home (2.8 [2.1] days).

) summarize health service utilization during the 8 weeks following the index hospitalization. In the 2 tables, only one comparison is statistically significant at 

 <.05. In all comparisons, differences are small.

Total costs seem to be lower when substitutive Hospital at Home care is available than when only traditional acute hospital care is available for patients with CHF or COPD but not for patients with community-acquired pneumonia or cellulitis. This finding may be a result of the research study—based requirement for prolonged continuous nursing input. Laboratory and procedure costs were lower and total costs were no higher when Hospital at Home was available. Results were robust across sites and conditions.

The study was not a randomized controlled trial, and it is possible that cost savings came at the expense of patient care. However, previous results showed no difference between Hospital at Home and acute hospital care in illness-specific quality measures, as well as lower rates of medical complications for Hospital at Home.

 An inference from the present study is that patients treated in the acute hospital may receive unnecessary laboratory tests and procedures because of their geographic proximity to them. This phenomenon has been observed for other high-technology procedures.

In this study, the cost difference for community-acquired pneumonia was small and not statistically significant; possibly because limited technology is used for diagnosis and treatment. In contrast, the diagnosis and treatment of COPD and CHF involve substantial technology, and the difference in costs was statistically significant for these conditions.

Although the in-hospital treatment costs were lower during the intervention phase, the Hospital at Home costs were even lower, suggesting that the availability of such services has the potential to reduce overall costs even when not all patients can be enrolled or choose to use the services. This was true in spite of a consistent finding not favoring Hospital at Home, namely, higher emergency department costs. Compared with observation patients, intervention patients spent an additional mean of 0.9 hours in the emergency department.

 Hospital at Home cost savings were sufficient to offset the costs of providing extra emergency department care and a substantial amount of initial one-on-one continuous nursing supervision. In the future, as Hospital at Home moves from research to a program model, we anticipate that the requirements for prolonged initial one-on-one continuous nursing will decrease and will result in a more favorable cost profile.

Several studies of costs of Hospital at Home care have been published previously. Some of these studies have examined models focusing on early discharge, often for treatment of surgical patients,

or have been a combination of early discharge and substitutive Hospital at Home investigations.

Results of these studies have been mixed, although most findings suggest lower costs for Hospital at Home.

Fewer studies have focused on substitutive Hospital at Home care. In a randomized trial of substitutive Hospital at Home for first uncomplicated acute ischemic stroke, costs were reduced by approximately 40% for Hospital at Home patients; total costs, including the costs of rehabilitation, were comparable to those of usual hospital care that included only part of the rehabilitation costs.

 An Australian randomized trial of substitutive Hospital at Home for medical patients found approximately 50% cost savings and no differences in clinical outcomes.

 Compared with these studies, our study is unique in that the length of stay for patients in our Hospital at Home more closely paralleled the length of stay for patients in acute care hospitals. In addition, ours is the first such study (to our knowledge) to be performed in multiple healthcare systems with different payment structures and to demonstrate reasonably consistent results.

This investigation was not intended to be a cost minimization study because the results of usual care and Hospital at Home were not identical. However, several previous economic evaluations were cost minimization studies.

If the only outcome criterion was that the results should not be worse than those of usual care, the present study could be interpreted as a cost minimization study indicating that Hospital at Home costs less to achieve similar results. Otherwise, the present study could be interpreted as a cost consequence study indicating that Hospital at Home care costs less and provides outcomes that are at least as positive across several dimensions.

This study has limitations. Patients were not randomly assigned to treatment, and differences between study groups may be due to selection bias. However, we used a highly conservative intent-to-treat type of analysis, which may have attenuated the reported economic advantages of Hospital at Home. In addition, the model studied was a research model that used mandatory costly initial one-on-one nursing care. As the model moves into a dissemination phase, it is likely that fewer patients will require that level of care, and this may also improve the cost profile of the model. Although patients may be more satisfied with high-quality care that is less costly to the healthcare system, patients may experience some economic costs in Hospital at Home that they would not experience in the acute care hospital. For example, the additional costs for food, laundry, heating, and lighting that the patient incurs at home are not accounted for in this study but should be explained to patients. The sample size was not large for any of the sites or for any of the conditions. Furthermore, the 3 sites used different methods to assign costs to the quantities of care that were provided. Although there are some difficulties in comparing Veterans Affairs and non—Veterans Affairs costs,

 the consistency of the direction of comparison and the multiple statistically significant results support the conclusion that the cost savings that can be achieved more than offset the increased operating costs.

In addition to economic costs of care, the costs of care in terms of potential burdens placed on family members need to be considered in a model such as Hospital at Home. Unlike other Hospital at Home models described in the international literature,

our model did not require that patients have a family member available or, if a family member was available, that he or she provide any service to Hospital at Home. Our previous work demonstrated that caregivers found Hospital at Home to provide more satisfying care from their standpoint9 and that Hospital at Home care was associated with less stress than that experienced by family members of patients treated in the acute care hospital.

Decision makers contemplating the implementation of a Hospital at Home program would ultimately need to consider not only the operating costs but also the startup costs and the potential for reimbursement. Payers would seem to benefit from finding a way to reimburse this type of care, for which no standards of reimbursement exist in the dominant fee-forservice Medicare system. Future implementation should consider the need for a substantial number of patients to make Hospital at Home economically viable and should consider targeting this type of care to patients for whom there is a large amount of discretionary care provided by physicians or by the use of laboratory tests and procedures. Even if Hospital at Home does not save money on a case-by-case basis, providing such care may still result in economic and systems improvements for a healthcare delivery system. Economic advantages can be achieved if inpatient beds opened up as a result of providing Hospital at Home care could be filled by more profitable patients, and systems improvements may be accrued by making additional inpatient beds available and by reducing the burden on emergency departments.

and in this study suggest that high-quality hospital-level care can be provided to older patients in their homes as a substitute for traditional inpatient care and that this is economically feasible. As the availability of beds in internal medicine services becomes limited and as patients continue to demand to be treated more like informed consumers, policy makers and hospital strategic planners will need to consider the viability of a Hospital at Home option for the health of patients and healthcare delivery systems.

: Johns Hopkins Bloomberg School of Public Health (KDF, LCB, RC, WBG, DMS, BL), Baltimore, MD; Portland Veterans Affairs Medical Center (SIM) and Oregon Health and Science University (SIM), Portland, OR; State University of New York (WBN), Buffalo; Fallon Clinic (JBB), Worcester, MA; and Johns Hopkins School of Medicine (WBG, BL).

: This study was supported by grant 98309-G from the John A. Hartford Foundation of New York and was supplemented at Portland Veterans Affairs Medical Center by Department of Veterans Affairs New Clinical Initiative Program grant 99-027a.

: The authors (KDF, LCB, RC, SIM, WBN, JBB, WBG, DMS, BL) report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

: Concept and design (KDF, LCB, JBB, WBN, WBG, DMS, BL); acquisition of data (KDF, LCB, RC, WBN, DMS, BL); analysis and interpretation of data (KDF, LCB, RC, WBG, DMS, BL); drafting of the manuscript (KDF, LCB, BL); critical revision of the manuscript for important intellectual content (KDF, WBN, WBG, BL); statistical analysis (KDF, LCB, RC); provision of study materials or patients (SIM, WBN, JBB); obtaining funding (LCB, WBN, BL); administrative, technical, or logistic support (LCB, WBN, BL); and supervision (WBG).

: Kevin D. Frick, PhD, Johns Hopkins Bloomberg School of Public Health, 624 N Broadway, Rm 606, Baltimore, MD 21205. E-mail: 

. Home hospital: toward a tighter definition [letter]. 

Substitutive Hospital at Home care for 4 common conditions is associated with cost savings.

January 2009

Availability of Data for Measuring Physician Quality Performance

New Medication Adherence Scale Versus Pharmacy Fill Rates in Seniors With Hypertension

Substitutive Hospital at Home for Older Persons: Effects on Costs

Medication, Diagnostic, and Cost Information as Predictors of High-Risk Patients in Need of Care Management

Physician Perception of Reimbursement for Outpatient Procedures Among Managed Care Patients With Diabetes Mellitus

Effect of Patient Comorbidities on Filling of Antihypertensive Prescriptions

Health of Medicare Advantage Plan Enrollees at 1 Year After Hurricane Katrina

Concern Still Warranted: Medication Burden and Persistence with Lipid-Lowering Drugs