The Relationships Among Age, Chronic Conditions, and Healthcare Costs

December 1, 2004
Wei Yu, PhD
Wei Yu, PhD

,
Arliene Ravelo, MPH
Arliene Ravelo, MPH

,
Todd H. Wagner, PhD
Todd H. Wagner, PhD

,
Paul G. Barnett, PhD
Paul G. Barnett, PhD

Volume 10, Issue 12

Objective: To learn how age and chronic illness affect costs inthe Veterans Affairs healthcare system.

Study Design: Veterans Affairs patients 65 years or older wereidentified from administrative data. We noted their healthcare utilization,cost, and diagnosis of any of 29 common chronic conditions(CCs). We examined how those 80 years or older differedfrom the younger patients.

Results: The Department of Veterans Affairs spent $8.5 billionto treat 1.6 million older patients in fiscal year 2000. Age was lessimportant than chronic illness in explaining cost differences. Theoldest patients incurred a mean of $1295 greater costs than theyounger patients, primarily because they were more likely to havea high-cost CC. The oldest patients incurred higher total costs thanthe younger patients in only 14 of 29 groups defined by CC. Long-termcare accounted for most of the extra cost of the oldestpatients. When this cost was excluded, the oldest patients incurredonly $266 more cost than the younger patients.

Conclusions: Growth in the population of the oldest patientswill increase the number of individuals with CCs requiring long-termcare. With its limited long-term care benefit, Medicare willavoid much of this financial consequence. In contrast, the financialrisk of acute and long-term care gives the Department of VeteransAffairs an incentive to develop strategies to prevent CCs associatedwith long-term care.

(Am J Manag Care. 2004;10:909-916)

The Department of Veterans Affairs (VA) healthcare system is one of the largest integrated healthcare systems in the United States, comprising 163 hospitals, 137 nursing homes, 43 domiciliaries, and 913 outpatient clinics.1 In fiscal year (FY) 2002, the VA spent $23 billion to provide medical care to 4.7 million veterans.2 A unique feature of the veteran population is that the age distribution closely relates to the timing of wars and can change sharply in a short time. As the veterans of World War II and the Korean War have aged, the number of veterans 85 years or older has increased at a mean rate of 11% per year, from 156 000 in 1990 to 510 000 in 2000; it is projected to reach 1.3 million by 2010.3,4

The rapid aging among veterans has increased the demand for VA healthcare services, especially long-term care among the oldest veterans. A population-based survey showed that the percentage of older persons who ever used nursing homes increased from 22% in the 65- to 74-year-old cohort to 58% in the 85-to 94-year-old cohort.5 The VA has to respond to these changes because the 1999 Veterans Millennium Healthcare and Benefits Act, Public Law 106-117, mandates that the VA must provide nursing home care to any veterans in need of such care for a service-connected disability.

Although the oldest patients use more long-term care, they are also less likely to receive aggressive treatment. Two studies6,7 found that Medicare expenditures decreased with increasing age because hospital and intensive care unit use by the oldest beneficiaries declined.8-10 This research suggests that there are 2 competing effects. Long-term care is expensive, but because treatment is less aggressive, the net cost effect of the increasing numbers of the oldest veterans on the VA healthcare costs is ambiguous.

To accurately predict demand, we must understand what care older adults are using and the independent effects of age and chronic conditions (CCs) on costs. Studies7,11-13 have analyzed the relationship between age, CCs, and healthcare costs. The commonly used Medicare claims data include information on medical treatments and diagnoses, but these data offer limited insight on long-term care because of Medicare's restrictive long-term care benefits. Medicaid data are also incomplete because people must spend down their wealth before they become eligible. Hence, studies based on Medicare or Medicaid data do not provide complete information on the relationships among age, CCs, and use of long-term care. A study of the VA system can help VA policy makers make projections and can offer insight on expected trends in total healthcare costs of the non-VA population.

Although we focus on veterans, understanding changes in costs incurred by older veterans will increase our understanding of how the aging of the population will affect the larger US healthcare system. The aging of the veteran population is a harbinger of changes in the US population as a whole. The number of individuals 85 years or older increased from 3.0 million in 1990 to 4.2 million in 2000; it will reach 6.1 million in 2010, when they will account for 1.2% of the population.4

We emphasize age effects, the effects of having more older people in the population on the cost of healthcare. These are different from the effects of aging, that is, the effects of increased life expectancy on healthcare cost.14,15 In this study, we analyzed healthcare costs incurred by 1.6 million older (≥65 years) VA patients. We identified 29 common CCs. Then, controlling for each CC, we compared healthcare costs of the younger (65-79 years) patients with those of the oldest (≥80 years) patients.

METHODS

To understand patterns of resource use for treating chronic illness, we grouped patients by their most expensive condition. With veterans assigned to mutually exclusive groups, we then estimated costs by type of medical services actually used. We analyzed the differences in resource use between the younger patients and the oldest patients, particularly in long-term care.

Sample and Data Selection

We collected electronic medical records on all veterans aged 65 years or older who received VA care in federal FY 2000, ending September 30, 2000. We excluded 82 713 people who used the VA system for only pharmacy benefits. The study population comprised 1 596 789 veterans, of whom 1 283 542 were younger patients and 313 247 were the oldest patients.

We did not adjust costs for patients who died during the study year. The higher mortality rates of the oldest patients have 2 conflicting effects on annual healthcare costs. As a consequence of a higher mortality rate, the oldest patients had fewer days of life during the study year and thus fewer days in which they could incur costs. On the other hand, there are high costs associated with illness sufficiently severe to result in death. In the preliminary study, we calculated the costs excluding patients who died during the study year. We found that mean costs were lower for all 29 CC groups; however, the cost differences between the oldest patients and the younger patients remained. Because including people who died during the study year reflects actual costs for a cohort in any period, we retained these subjects in this analysis.

The VA uses contract providers for those services it cannot render conveniently. In FY 2000, 7% of VA costs were paid for contracted services. Because contract care is not completely described by VA data sets, we excluded it from analysis.

Chronic Condition Identification

We chose the 29 CCs because they were included in related studies,12,16 because they are given high priority in VA research17 or quality enhancement programs,18 or because they are known to be especially common among VA patients. We identified which patients had these CCs using the International Classification of Diseases, Ninth Revision, Clinical Modification diagnostic codes recorded in VA healthcare utilization files in FY 2000. We included in our review all diagnostic codes in the inpatient and outpatient files. For 28 conditions, we classified patients with a CC based on 1 or more diagnoses. For depression, we required 2 or more outpatient diagnoses or a single diagnosis from a psychiatric clinic.

We reviewed the classification methods from a Kaiser Permanente study12 and the Clinical Classifications Software developed by the Agency for Healthcare Research and Quality.19 We also reviewed other published studies,20,21 noting when methods differed from the Kaiser Permanente and Clinical Classifications Software methods. We conducted a sensitivity analysis of these classification methods. In a conservative approach, we excluded those codes identified by our physician consultants that did not clearly specify a CC.22

Cost Data

We determined the cost of all VA-provided medical care in FY 2000. For inpatient stays that spanned multiple FYs, we allocated inpatient costs in proportion to the number of days in FY 2000. Inpatient and outpatient costs were obtained from the Health Economics Resource Center's average cost database. We allocated costs for medical/surgical hospitalizations to each hospital stay using a clinical cost function that we developed from Medicare data through a regression of cost against the diagnosis related group relative value weights, length of stay, and use of an intensive care unit.23 We adjusted nursing home costs for case mix using the resource utilization group score.24 For inpatient rehabilitation and specialty mental healthcare, costs were measured by a mean daily rate.25 Outpatient healthcare costs were based on the Current Procedural Terminology, Fourth Edition codes recorded in the database. The relative weights were based on the Medicare Resource-Based Relative Value Scale,26 the relative value units developed by Ingenix Health Intelligence,27 surveys of provider fees and fee schedules, 28-30 and the 1999 survey of the American Dental Association.31 We used these relative weights to allocate VA outpatient costs to each encounter.32 Costs for outpatient pharmacy services were obtained from the VA Decision Support System national extract, which includes overhead costs proportionally distributed to the outpatient pharmacy departments.33

Cost Hierarchy Classification

Many veterans have multiple CCs. Because we could not clearly separate medical treatments by CC, we faced the challenging question of how to group the patients such that their resource use would be closely related to their CC.

We developed a hierarchical classification method that grouped patients who had any of the 29 CCs into 29 mutually exclusive CC groups. Patients were classified into only 1 group according to their most expensive CC. For example, a patient who had lung cancer and hypertension was assigned to the lung cancer group. This method was preferable to defining groups solely by CC, as this would have assigned patients with several conditions to several groups. Under that approach, a group's mean cost would be difficult to interpret because it would be affected by the inclusion of patients with other, higher-cost conditions.

We first calculated the mean costs of all 29 CCs. In group 1, we placed patients who had the most expensive condition. We removed patients who were in group 1 from the analysis and then recalculated the mean costs for the remaining 28 CCs, placing those patients who had the most expensive CC in group 2. We repeated this process to rank all 29 conditions. Patients who did not have any diagnosis of the 29 CCs were in a single "other"group, group 30. Except for the most expensive CC group, the number of people in each CC group does not reflect the prevalence of that condition, because each CC group excludes members who also have a more expensive CC. For the same reason, the mean annual cost of each CC group (other than the first) is the lower bound of the mean cost incurred by patients who have that condition; excluded from the calculation are patients who have that condition and a more expensive one.

Data Analysis

We categorized inpatient care into 4 groups: (1) medical/ surgical, (2) rehabilitation, (3) long-term care, and (4) specialized mental health or substance abuse treatment. For outpatient care, we separated pharmacy services from all other care. To determine whether the oldest patients had more CCs than the younger patients, we calculated the proportion of patients who had 0, 1, 2, and 3 or more of the 29 CCs. To evaluate costs, we compared mean costs for both age groups for each of the 30 CC groups. We compared the long-term care costs incurred by each age group, the proportion of individuals who used any long-term care during the study year, and the proportion of total cost accounted for by long-term care. We tested statistical differences in mean costs between age groups using 2-tailed t tests.

RESULTS

In FY 2000, VA facilities provided $8.5 billion worth of care to the 1.6 million older veterans. Diagnoses of CCs were equally common in both age groups. Use of long-term care was the major difference in demand for healthcare between the 2 age groups. The patterns of resource use, however, depended heavily on CC group.

Age and Number of Chronic Conditions

Among the 1.6 million older veterans who received services from the VA in FY 2000, 85.5% had at least 1 of the 29 CCs; 40.5% had 3 or more CCs (Table 1). The number of CCs (1, 2, or = 3) was similar in the 2 age groups. The oldest group had a slightly higher percentage (16.5% vs 14.5%) of patients who had none of the CCs.

Age, Chronic Condition, and Mean Costs

Table 2 gives the mean costs of the hierarchically classified CC groups and is sorted by the mean cost of each CC for all ages. The mean costs of the oldest patients were higher than those of the younger patients in 14 of the 29 conditions. The mean costs of the oldest patients were 8% lower in renal failure and 11% lower in lung cancer than those of the younger patients. The differences in mean costs between the 2 age groups were not statistically different for the remaining 13 conditions. In both age groups, patients who had at least 1 of the 29 CCs incurred costs that were 3 to 4 times higher than those incurred by patients who had none of the 29 CCs.

The percentages of the total cost reported in Table 2 reflect the overall cost effect (cost per person and prevalence of CC) on the VA. Among the younger patients, cancer was the most costly condition, accounting for 16.0% of the total cost. The other 2 expensive conditions were congestive heart failure (11.4%) and renal failure (11.2%). The biggest difference between the 2 age groups was that the combined cost of treating patients who had dementia or Alzheimer's disease accounted for 15.0% of the total cost among the oldest patients, whereas it accounted for only 6.2% among the younger patients.

Age and Use of Long-Term Care

Table 2 also gives healthcare costs exclusive of long-term care. The oldest patients incurred more of these costs in only 5 CC groups, much fewer than the 14 CC groups that were different when we included long-term care. The number of CC groups in which the younger patients incurred more costs increased from 2 to 11. There was no significant difference in cost between the age groups in the remaining 13 CC groups. The ranking in Table 2 is the same as that in Table 1; it is based on total cost, not on long-term care cost. For some conditions, such as headache, long-term care is not the driving factor for cost.

Table 3 demonstrates the importance of long-term care costs in explaining the effect of age on costs of each CC group. Long-term care accounted for a substantially higher proportion of the oldest patients'total healthcare cost, compared with the younger patients, in all 29 CC groups. In 16 of the 29 CC groups, the proportion of total healthcare cost attributable to long-term care for the oldest patients was more than double that of the younger patients. The proportion of the oldest patients who used any long-term care was higher than that of the younger patients in all CC groups except AIDS/HIV.

Age and Overall Healthcare Costs

Table 4 gives the differences in overall resource use between the 2 age groups. Without specification of medical condition, compared with costs of the younger patients, the total mean cost of the oldest patients was 22% higher ($6152 vs $5096), and the inpatient cost was 43% higher ($3849 vs $2695). Most of the difference in inpatient cost was because of long-term care, which was 136% higher for the oldest patients than for the younger patients ($1617 vs $686). The reason that we observed such large differences in costs between the 2 age groups at the aggregate level is that a higher proportion of the oldest patients had expensive CCs. The number of people in the 10 most expensive CCs accounted for 18% of the younger patients and for 25% of the oldest patients. For some CCs, long-term care accounted for a large proportion of the cost for the oldest patients, including 61% for Alzheimer's disease, 53% for multiple sclerosis, and 47% for Parkinson's disease. These data show that, as a population cohort, the oldest veterans use more care and incur higher costs. When we look at groups defined by chronic illness, age is a much less important factor in predicting demand for healthcare.

DISCUSSION

The data yielded 2 robust results. First, age had a limited effect on cost. The oldest patients in our study incurred a mean of $1295 greater costs than the younger patients. This was primarily because of the higher prevalence of expensive CCs among the oldest patients. The presence of a CC had a more important effect on cost than age alone. We found that, within groups defined by CCs, the oldest patients did not consistently incur greater cost. Long-term care costs account for much of the difference between these groups. When long-term care cost was excluded, the oldest patients incurred a mean of only $266 more than the younger patients.

Second, resource use was concentrated in a few CCs. We found that the top 10 most expensive (total cost) CCs for the oldest patients account for 69% of that age group's cost. Developing more efficient technologies for these CCs may help control cost. Less expensive alternatives to long-term care, such as home-or community- based care, may also hold promise. However, a study34 found that home-based care for veterans was not a substitute for nursing home care.

The VA policy makers trying to predict the need for long-term care should focus predominantly on the distribution of CCs, keeping in mind other factors that could affect demand, such as the availability and price of alternatives and supply constraints. Although age is important, it has a minor role. The VA also must identify how to manage financial risk across its sectors of care (eg, medical/surgical, rehabilitation, and long-term care). We found a shift in healthcare utilization from other types of care to long-term care when the proportion of the oldest patients increases. Therefore, part of the increase in costs for long-term care for the oldest patients is offset by the decrease in costs for other types of services. As an integrated system that provides acute and long-term care, the VA has an incentive to provide programs that prevent long-term care stays. The VA benefits also include pharmacy, mental health, rehabilitation, and other care, so managers need to balance competing concerns.

Outside the VA, long-term care is financed separately, and providers have few incentives to prevent longterm care. The increased demand for long-term care will have to be absorbed by individuals, Medicaid, and other programs. As policy makers, researchers, and organizations struggle to manage risk in the face of growing demand for long-term care, the VA is in a unique position to provide empirical data. Toward that goal, more research is needed to understand how the VA integrated system affects costs and outcomes for long-term care.

Our study population was limited to veterans who used the VA healthcare system. It differs from the general population of older persons in the United States. Our subjects were 95% male, whereas the general population of older persons comprises more women than men. Our study population included a disproportionate number of the poor and disabled, who in turn are at particularly high risk for the CCs that we considered. These factors, however, are unlikely to affect the relationship between healthcare utilization and age. The demand for long-term care by the oldest women could be higher than that by men because, for example, women are generally less likely to obtain care from their spouse at home, as they usually outlive their husbands.

Because older veterans are also eligible for Medicare, some of them use both systems for their healthcare. Preliminary data from a VA study35 showed that about 40% of VA patients used Medicare in 2000. Our study did not include costs of healthcare services provided through Medicare. If we include Medicare services, the long-term care costs would be lower than the reported numbers in this study. Including Medicare services, however, may not significantly affect our results because, to our knowledge, there is no evidence showing that Medicare use is associated with age.36

We considered only long-term care provided by VA facilities. In FY 2000, the VA provided $2.0 billion of long-term care; it also purchased about $350 million of these services from non-VA providers in the community. We did not have person-level estimates of the costs of community-provided care and thus did not include such costs in our estimates. As a result of omitting the community- based nursing home care, our analyses understated the total cost of VA long-term care by about 15%. It is unclear how this omission affects the results. People choose the location of care based on many factors, including proximity to home, perceived quality of care delivered by the facility, and supply of VA long-term care beds. Further research on this topic is warranted.

As with other large databases, the VA administrative data could miss some diagnoses.37,38 The problem of diagnoses being missed in a given hospital stay or visit is partly attenuated by the design of our study. We would have missed a diagnosis only if it was missing from the medical record for every stay and visit that took place during the study year. Recent investigations included audits of VA outpatient and inpatient data at 6 medical centers and found coding accuracy to be similar to that observed in the community.39,40 The VA budget allocation systems have long provided facilities with incentive to completely code inpatient diagnoses. Current budget allocations are affected by outpatient diagnoses too. In the year of this study, about 30% of clinic visits, exclusive of laboratory and radiology visits, were assigned more than a single diagnosis code.

CONCLUSIONS

As the World War II and Korean War veterans reach 80 years or older, demand for long-term care increases accordingly. The VA will need to manage healthcare efficiently to meet the rapid increase in demand; such management requires an understanding of the relationships among age, CCs, and healthcare costs. Our data indicate that age and healthcare costs are not always positively correlated once we control for CCs.

The aging of the rest of the US population will increase demand for health services. This study found that the primary effect of age is an increase in demand for long-term care services associated with certain CCs. This has important implications for the US healthcare system. Because Medicare has a limited long-term care benefit, it will avoid the major effect of this trend.

In contrast to comprehensive healthcare sponsors like the VA, decision makers in Medicare plans have little financial incentive to invest in services that can prevent conditions that lead to increased demand for long-term care. In the VA system, investing in prevention might be offset by lower long-term care costs. The growth in the number of the oldest patients should prompt the VA to develop such interventions. Until the risk in the financing of acute and long-term care is integrated, society will invest too little in programs that prevent conditions that increase demand for long-term care. This problem will be exacerbated by the rapid growth in the number of the oldest patients.

Acknowledgments

We thank Mary Goldstein, MD, and Michael Gould, MD, for their help in classifying chronic conditions.

References

1. Veterans Health Administration Web site. VA health care, systemwide capacities, FY 1996-2002. Available at: http://www.va.gov/vetdata/ProgramStatics/stat_app02/Table%207%20(02).xls. Accessed March 23, 2004.

2. Veterans Health Administration Web site. VA health care, systemwide obligations, FY 1996-2002. Available at: http://www.va.gov/vetdata/ProgramStatics/stat_app02/Table%208%20(02).xls. Accessed March 23, 2004.

3. Veterans Health Administration Web site. VA health care, systemwide workload, FY 1996-2002. Available at: http://www.va.gov/vetdata/ProgramStatics/ stat_app02/Table%206%20(02).xls. Accessed March 23, 2004, 2004.

4. Administration on Aging Web site. Older population by age, 1900 to 2050. Available at: http://www.aoa.gov/prof/Statistics/online_stat_data/AgePop2050.asp. Accessed July 21, 2004.

5. Spillman BC, Lubitz J. New estimates of lifetime nursing home use: have patterns of use changed? Med Care. 2002;40:965-975.

6. Hogan C, Lunney J, Gabel J, Lynn J. Medicare beneficiaries' costs of care in the last year of life. Health Aff (Millwood). 2001;20(4):188-195.

7. Lubitz J, Greenberg LG, Gorina Y, Wartzman L, Gibson D. Three decades of health care use by the elderly, 1965-1998. Health Aff (Millwood). 2001;20(2):19-32.

8. Levinsky NG, Ash AS, Yu W, Moskowitz MA. Patterns of use of common major procedures in medical care of older adults. J Am Geriatr Soc. 1999;47:553-558.

9. Perls TT, Wood ER. Acute care costs of the oldest old: they cost less, their care intensity is less, and they go to nonteaching hospitals. Arch Intern Med. 1996;156:754-760.

10. Yu W, Ash AS, Levinsky NG, Moskowitz MA. Intensive care unit use and mortality in the elderly. J Gen Intern Med. 2000;15:97-102.

11. Mendelson DN, Schwartz WB. The effects of aging and population growth on health care costs. Health Aff (Millwood). 1993;12(1):119-125.

12. Ray GT, Collin F, Lieu T, et al. The cost of health conditions in a health maintenance organization. Med Care Res Rev. 2000;57:92-109.

13. Spillman BC, Lubitz J. The effect of longevity on spending for acute and long-term care. N Engl J Med. 2000;342:1409-1415.

14. Reinhardt UE. Does the aging of the population really drive the demand for health care? Health Aff (Millwood). 2003;22(6):27-39.

15. Burner ST, Waldo DR. National health expenditure projections, 1994-2005. Health Care Financ Rev. 1995;16:221-242.

16. Fishman P, Von Korff M, Lozano P, Hecht J. Chronic care costs in managed care. Health Aff (Millwood). 1997;16(3):239-247.

17. Veterans Health Administration Web site. Designated research areas for an integrated program of research, working paper, February 1, 1998. Available at: http://www1.va.gov/resdev/fr/dra.doc. Accessed November 18, 2004.

18. Demakis JG, McQueen L, Kizer KW, Feussner JR. Quality Enhancement Research Initiative (QUERI): a collaboration between research and clinical practice. Med Care. 2000;38(suppl 1):I17-I25.

19. Agency for Healthcare Research and Quality Web site. Clinical Classifications Software (CCS) for ICD-9-CM fact sheet. Available at: http://www.hcupus.ahrq.gov/toolssoftware/ccs/ccsfactsheet.jsp. Accessed November 18, 2004.

20. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45:613-619.

21. Peterson KA, Swindle RW, Phibbs CS, Recine B, Moos RH. Determinants of readmission following inpatient substance abuse treatment: a national study of VA programs. Med Care. 1994;32:535-550.

22. Yu W, Ravelo A, Wagner TH, et al. Prevalence and costs of chronic conditions in the VA health care system. Med Care Res Rev. 2003;60(suppl):146S-167S.

23. Wagner T, Chen S, Barnett P. Using average cost methods to estimate encounter-level costs for medical-surgical stays in the VA. Med Care Res Rev. 2003;60(suppl):15S-36S.

24. Fries BE, Schneider DP, Foley WJ, Dowling M. Case-mix classification of Medicare residents in skilled nursing facilities: resource utilization groups (RUGT18). Med Care. 1989;27:843-858.

25. Yu W, Wagner T, Chen S, Barnett P. Average cost of VA rehabilitation, mental health, and long-term hospital stays. Med Care Res Rev. 2003;60(suppl):40S-53S.

26. Hsiao W, Braun P, Dunn DL, et al. An overview of the development and refinement of the Resource-Based Relative Value Scale: the foundation for reform of U.S. physician payment. Med Care. 1992;30(suppl):NS1-NS12.

27. St Anthony's 2000 RBRVS: A Comprehensive Listing of RBRVS Values for all CPT and HCPCS Codes. Salt Lake City, Utah: Ingenix Health Intelligence; 2000.

28. State of California Workers' Compensation Official Medical Fee Schedule. San Francisco, Calif: Dept of Industrial Relations; 1999.

29. Wasserman Y. 2000 National Dental Advisory Service Comprehensive Fee Report. West Allis, Wis: Medical Publishers Ltd; 2000.

30. Wasserman Y. 2000 Physicians' Fee Reference Comprehensive Fee Report. West Allis, Wis: Medical Publishers Ltd; 2000.

31. American Dental Association 1999 Survey of Dental Fees. Chicago, Ill: American Dental Association; 2000.

32. Phibbs CS, Bhandari A, Yu W, Barnett PG. Estimating the costs of VA ambulatory care. Med Care Res Rev. 2003;60(suppl):54S-73S.

33. Yu W, Barnett PG. Research Guide to Decision Support System National Cost Extracts. Menlo Park, Calif: Health Economics Resource Center, VA Palo Alto Health Care System; March 2002.

34. Hughes SL, Weaver FM, Giobbie-Hurder A, et al. Effectiveness of team-managed home-based primary care: a randomized multicenter trial. JAMA. 2000;284:2877-2885.

35. VIReC VA Information Resource Center Web site. Research findings from the VA Medicare Data Merge Initiative: veterans' enrollment, access and use of Medicare and VA health services (XVA 6g-001): report to the Under Secretary for Health, Department of Veterans Affairs, September 2003. Available at: http://www. virec.research.med.va.gov/DataSourcesName/VA-MedicareData/USHreport.pdf. 2003. Accessed July 23, 2004.

36. Shen Y, Hendricks AM, Kazis LE, et al. Health Insurance and Use of Services by Veterans, 1999: Large Health Survey of Veteran Enrollees: Executive Report. Washington, DC: Office of Quality and Performance, Veterans Health Administration; 2001.

37. Lloyd SS, Rissing JP. Physician and coding errors in patient records. JAMA. 1985;254:1330-1336.

38. Kashner TM. Agreement between administrative files and written medical records: a case of the Department of Veterans Affairs. Med Care. 1998;36:1324-1336.

39. Nugent GN, Roselle G, Nugent LB, Render ML. Methods to determine private sector payment for VA outpatient services: institutional payments to providers. Med Care. 2003;41(suppl):II33-II42.

40. Render ML, Roselle G, Franchi E, Nugent LB. Methods for estimating private sector payments for VA acute inpatient stays. Med Care. 2003;41(suppl):II11-II22.

From the Health Economics Resource Center, VA Palo Alto Health Care System, Menlo Park (WY, THW, PGB); Centers for Health Policy and for Primary Care and Outcomes Research, Stanford University (WY, THW, PGB), and Department of Health Research and Policy, Stanford University Medical School (THW, PGB), Stanford; and Department of Global Health Outcomes, Strategy & Research, Allergan, Inc, Irvine (AR); Calif. Ms Ravelo was an employee at the Health Economics Resource Center at the time of this study.

This research was supported by grant SDR-ECN-99017 from the Health Services Research and Development Service, Department of Veterans Affairs, and by the VA Cooperative Studies Program. All conclusions are our own and do not necessarily reflect those of the affiliated centers or the Department of Veterans Affairs.

Address correspondence to: Wei Yu, PhD, Health Economics Resource Center, VA Palo Alto Health Care System, 795 Willow Road (152-MPD), Menlo Park, CA 94025. E-mail: wyu2@stanford.edu.

From the Health Economics Resource Center, VA Palo Alto Health Care System, MenloPark (WY, THW, PGB); Centers for Health Policy and for Primary Care and OutcomesResearch, Stanford University (WY, THW, PGB), and Department of Health Research andPolicy, Stanford University Medical School (THW, PGB), Stanford; and Department ofGlobal Health Outcomes, Strategy & Research, Allergan, Inc, Irvine (AR); Calif. Ms Ravelowas an employee at the Health Economics Resource Center at the time of this study.

This research was supported by grant SDR-ECN-99017 from the Health ServicesResearch and Development Service, Department of Veterans Affairs, and by the VACooperative Studies Program. All conclusions are our own and do not necessarily reflectthose of the affiliated centers or the Department of Veterans Affairs.

Address correspondence to: Wei Yu, PhD, Health Economics Resource Center, VAPalo Alto Health Care System, 795 Willow Road (152-MPD), Menlo Park, CA 94025.E-mail: wyu2@stanford.edu.

1. Veterans Health Administration Web site. VA health care, systemwide capacities,FY 1996-2002. Available at: http://www.va.gov/vetdata/ProgramStatics/stat_app02/Table%207%20(02).xls. Accessed March 23, 2004.

2. Veterans Health Administration Web site. VA health care, systemwide obligations,FY 1996-2002. Available at: http://www.va.gov/vetdata/ProgramStatics/stat_app02/Table%208%20(02).xls. Accessed March 23, 2004.

3. Veterans Health Administration Web site. VA health care, systemwide workload,FY 1996-2002. Available at: http://www.va.gov/vetdata/ProgramStatics/stat_app02/Table%206%20(02).xls. Accessed March 23, 2004, 2004.

4. Administration on Aging Web site. Older population by age, 1900 to 2050.Available at: http://www.aoa.gov/prof/Statistics/online_stat_data/AgePop2050.asp.Accessed July 21, 2004.

Med Care.

5. Spillman BC, Lubitz J. New estimates of lifetime nursing home use: have patternsof use changed? 2002;40:965-975.

Health Aff (Millwood).

6. Hogan C, Lunney J, Gabel J, Lynn J. Medicare beneficiaries' costs of care inthe last year of life. 2001;20(4):188-195.

Health Aff (Millwood).

7. Lubitz J, Greenberg LG, Gorina Y, Wartzman L, Gibson D. Three decades ofhealth care use by the elderly, 1965-1998. 2001;20(2):19-32.

J Am Geriatr Soc.

8. Levinsky NG, Ash AS, Yu W, Moskowitz MA. Patterns of use of common majorprocedures in medical care of older adults. 1999;47:553-558.

Arch Intern Med.

9. Perls TT, Wood ER. Acute care costs of the oldest old: they cost less, their careintensity is less, and they go to nonteaching hospitals. 1996;156:754-760.

J Gen Intern Med.

10. Yu W, Ash AS, Levinsky NG, Moskowitz MA. Intensive care unit use andmortality in the elderly. 2000;15:97-102.

Health Aff (Millwood).

11. Mendelson DN, Schwartz WB. The effects of aging and population growth onhealth care costs. 1993;12(1):119-125.

Med Care Res Rev.

12. Ray GT, Collin F, Lieu T, et al. The cost of health conditions in a health maintenanceorganization. 2000;57:92-109.

N Engl J Med.

13. Spillman BC, Lubitz J. The effect of longevity on spending for acute and long-termcare. 2000;342:1409-1415.

Health Aff (Millwood).

14. Reinhardt UE. Does the aging of the population really drive the demand forhealth care? 2003;22(6):27-39.

Health Care Financ Rev.

15. Burner ST, Waldo DR. National health expenditure projections, 1994-2005.1995;16:221-242.

Health Aff (Millwood).

16. Fishman P, Von Korff M, Lozano P, Hecht J. Chronic care costs in managedcare. 1997;16(3):239-247.

17. Veterans Health Administration Web site. Designated research areas for anintegrated program of research, working paper, February 1, 1998. Available at:http://www1.va.gov/resdev/fr/dra.doc. Accessed November 18, 2004.

Med Care.

18. Demakis JG, McQueen L, Kizer KW, Feussner JR. Quality EnhancementResearch Initiative (QUERI): a collaboration between research and clinical practice.2000;38(suppl 1):I17-I25.

ICD-9-CM

19. Agency for Healthcare Research and Quality Web site. Clinical ClassificationsSoftware (CCS) for fact sheet. Available at: http://www.hcupus.ahrq.gov/toolssoftware/ccs/ccsfactsheet.jsp. Accessed November 18, 2004.

ICD-9-CM

J Clin Epidemiol.

20. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for usewith administrative databases. 1992;45:613-619.

Med Care.

21. Peterson KA, Swindle RW, Phibbs CS, Recine B, Moos RH. Determinants ofreadmission following inpatient substance abuse treatment: a national study of VAprograms. 1994;32:535-550.

Med Care Res Rev.

22. Yu W, Ravelo A, Wagner TH, et al. Prevalence and costs of chronic conditionsin the VA health care system. 2003;60(suppl):146S-167S.

Med Care Res Rev.

23. Wagner T, Chen S, Barnett P. Using average cost methods to estimateencounter-level costs for medical-surgical stays in the VA. 2003;60(suppl):15S-36S.

Med Care.

24. Fries BE, Schneider DP, Foley WJ, Dowling M. Case-mix classification ofMedicare residents in skilled nursing facilities: resource utilization groups (RUGT18).1989;27:843-858.

Med Care Res Rev.

25. Yu W, Wagner T, Chen S, Barnett P. Average cost of VA rehabilitation, mentalhealth, and long-term hospital stays. 2003;60(suppl):40S-53S.

Med Care.

26. Hsiao W, Braun P, Dunn DL, et al. An overview of the development andrefinement of the Resource-Based Relative Value Scale: the foundation for reformof U.S. physician payment. 1992;30(suppl):NS1-NS12.

27. St Anthony's 2000 RBRVS: A Comprehensive Listing of RBRVS Values for allCPT and HCPCS Codes. Salt Lake City, Utah: Ingenix Health Intelligence; 2000.

28. State of California Workers' Compensation Official Medical Fee Schedule. SanFrancisco, Calif: Dept of Industrial Relations; 1999.

2000 National Dental Advisory Service Comprehensive Fee

Report.

29. Wasserman Y. West Allis, Wis: Medical Publishers Ltd; 2000.

2000 Physicians' Fee Reference Comprehensive Fee Report.

30. Wasserman Y. West Allis, Wis: Medical Publishers Ltd; 2000.

31. American Dental Association 1999 Survey of Dental Fees. Chicago, Ill:American Dental Association; 2000.

Med Care Res Rev.

32. Phibbs CS, Bhandari A, Yu W, Barnett PG. Estimating the costs of VA ambulatorycare. 2003;60(suppl):54S-73S.

Research Guide to Decision Support System National Cost

Extracts.

33. Yu W, Barnett PG. Menlo Park, Calif: Health Economics Resource Center, VA Palo AltoHealth Care System; March 2002.

JAMA.

34. Hughes SL, Weaver FM, Giobbie-Hurder A, et al. Effectiveness of team-managedhome-based primary care: a randomized multicenter trial. 2000;284:2877-2885.

35. VIReC VA Information Resource Center Web site. Research findings from theVA Medicare Data Merge Initiative: veterans' enrollment, access and use ofMedicare and VA health services (XVA 6g-001): report to the Under Secretary forHealth, Department of Veterans Affairs, September 2003. Available at: http://www.virec.research.med.va.gov/DataSourcesName/VA-MedicareData/USHreport.pdf.2003. Accessed July 23, 2004.

Health Insurance and Use of Services

by Veterans, 1999: Large Health Survey of Veteran Enrollees: Executive Report.

36. Shen Y, Hendricks AM, Kazis LE, et al. Washington, DC: Office of Quality and Performance, Veterans HealthAdministration; 2001.

JAMA.

37. Lloyd SS, Rissing JP. Physician and coding errors in patient records. 1985;254:1330-1336.

Med Care.

38. Kashner TM. Agreement between administrative files and written medicalrecords: a case of the Department of Veterans Affairs. 1998;36:1324-1336.

Med

Care.

39. Nugent GN, Roselle G, Nugent LB, Render ML. Methods to determine privatesector payment for VA outpatient services: institutional payments to providers. 2003;41(suppl):II33-II42.

Med Care.

40. Render ML, Roselle G, Franchi E, Nugent LB. Methods for estimating privatesector payments for VA acute inpatient stays. 2003;41(suppl):II11-II22.