The Challenges of Estimating the National Costs of Osteoarthritis: Are We Making Progress?

Address reprint requests to Dr. Maetzel.

A commonly cited source of US national cost estimates for the treatment of arthritis is a report by the National Arthritis Data Work Group published in 1995¹. In it the 1992 costs for "all forms of arthritis" were calculated to be US $64.8 billion, of which $15.2 billion were estimated to account for medical costs, and $49.6 billion for costs due to productivity losses. As only half of these patients have clinically diagnosed osteoarthritis (OA)², how high are the costs for this condition?

The study by Harrold and colleagues in this issue of The Journal makes an important contribution toward better estimating US national costs for OA³. The authors estimated OA related medical costs for members of a community health plan in Central and Eastern Massachusetts in 1996. Subjects either had an administrative diagnosis of OA or were a subset with a validated diagnosis as per the American College of Rheumatology criteria for the classification and reporting of OA of the hand, hip, and knee. These new estimates allow a "back of the envelope" approximation to the total US costs for physician diagnosed OA, an estimate of which is provided by the authors in the Discussion section of their report. By expanding these calculations it is possible to get some idea of the total costs of physician diagnosed OA in the US for the year 2000.

To obtain estimates for US national costs of OA, the following information is needed: (1) the projected prevalence of OA in 2000; (2) the increase in the medical care portion of the US consumer price index to update the 1996 costs calculated by Harrold, et al; (3) the gross domestic product of the US in 2000; and (4) the contribution of indirect costs to the total costs. Yelin and Callahan worked with a prevalence of 15.2% for "all forms of arthritis": 37.9 million of 249.5 million people living in the US in 1990¹. In a parallel article, the National Arthritis Data Working Group estimated the prevalence of physician diagnosed OA to be 8.3%: 20.7 million of 249.5 million people living in the US in 1990². By the year 2020, the prevalence of "all forms of arthritis" was expected to increase to 18.3%; therefore, assuming a parallel increase in physician diagnosed OA, the prevalence of the latter would be 9.0% in the year 2000 (24.8 million of 276.1 million people living in the US in the year 2000). The annual OA related charges identified by Harrold and colleagues would be $1064 in the year 2000, when adjusted by a 13% increase in the medical care portion of the US consumer price index between 1996 and 2000⁴. Based on these numbers, the annual economic burden of OA related medical care costs incurred by patients with physician-diagnosed OA in the US in 2000 would be US $26.4 billion (Table 1). This would represent 0.27% of the US gross domestic product for the year 2000.

Table 1. Potential range of US total costs for OA in the year 2000.

Calculation of the "total" economic burden requires estimation of other costs such as productivity losses; these costs are generally called "indirect costs." The report by Yelin and Callahan, which refers to studies by D. Rice⁵, estimates the costs of productivity losses to be $49.6 billion, 3.26 times greater than the total medical costs of 15.2 billion. Much lower ratios of indirect to direct costs were obtained by Gabriel and colleagues, who found average indirect and nonmedical costs for OA to be $824 (1992 US dollars), i.e., 31% of the $2654 (1992 US dollars) incurred for direct medical charges by Olmsted County residents with OA^6,7. A community based study of Ontario patients with OA estimated indirect costs to amount to 45.4% of the direct medical costs⁸.

These studies used the human capital method to calculate indirect costs, a method that uses the full replacement costs independent of whether the worker was replaced or not. An alternative method, the friction cost approach, only includes productivity costs during the period that is needed to restore the initial production level⁹. This differentiation is important because indirect costs shrink considerably when the friction cost approach is applied. For example, the short term indirect costs of low back pain in the Netherlands as estimated by the human capital approach were more than 3 times higher than the indirect costs estimated by the friction cost approach (US $4.6 billion vs $1.5 billion)¹⁰. Similarly, in another study among generally younger patients with ankylosing spondylitis in the Netherlands, Belgium, and France, the indirect costs as estimated by the friction cost approach amounted to only 14% of the indirect costs estimated through the human capital approach¹¹. Therefore, indirect costs measured in the studies by Gabriel, et al^6,7 and in the Ontario study⁸ may still overestimate the indirect costs incurred by society. What the yearly total costs of OA would look like with different shares of indirect costs can be seen in Table 1. In the most likely example, indirect costs would be 25% of direct costs, and yearly costs for OA in the US would be $33 billion dollars, or 0.33% of the gross domestic product.

True, these calculations are made on the "back of an envelope," but the data by Harrold and colleagues, as well as the above mentioned studies that estimated the importance of indirect costs, provide an opportunity to reconsider what the economic burden of OA might be. Clearly, several factors precluded our getting any closer to true estimates of the costs of OA, which are: (1) national representativeness of the members of the community health plan in Central and Eastern Massachusetts, (2) disease definition, (3) disease-specific attribution of costs, and (4) the method of valuing productivity losses.

National cost estimates for clinically diagnosed OA could be higher or lower depending on the national representativeness of patients enrolled in this community health plan. Representativeness is not paramount for Harrold, et al to meet the study's objectives; however, the mechanism of selecting patients into the community health plan would need to be clarified before cost estimates could be safely extrapolated. The main contribution of Harrold, et al is to illustrate that the more stringent the disease definition, the higher the costs. But a disease defined by more stringent criteria also has a lower prevalence. Thus, the total costs of "all forms of arthritis" are likely higher than the total costs of "physician diagnosed OA," even if the per-patient costs for the latter definition are higher.

Disease attribution is another important problem: not knowing which medical costs are OA related inflates the total cost estimates, as many patients with OA have comorbidities. Both Gabriel, et al and MacLean, et al¹² compared patients with arthritis to nonarthritic controls to infer "disease attributable" costs. Matching for age and sex, however, is unlikely to weed out the costs that are attributable to other comorbidities, unless these have been adjusted for. Thus, the report by Harrold and colleagues provides a unique but laborious approach to the estimation of medical costs that are truly OA related, excluding over-the-counter medicines and OA related complementary or alternative healthcare provision.

Finally, the choice of method for valuing productivity losses can double or even triple the total cost estimates; therefore what is the most appropriate method has not been without debate^13,14. From some people's perspective even the human capital approach undervalues the indirect costs of arthritis, because patients with arthritis are more likely to be female and do not participate in the workforce, either by choice or because of their disease.

With continuing improvement in cost of illness (COI) methodology, these studies may finally become useful for policymakers. However, enthusiasm is tampered by a recent review of published COI studies where the authors observed an up to 40-fold variation in COI estimates for the same disorder¹⁵. Costs for only 80 diagnoses were estimated to be 2.1 times higher than the actual 1992 expenditures for all medical care in the US and the US gross domestic product would have been reduced by "...20.5% because of indirect costs attributable to the 80 diagnoses, even though 53 diagnosis categories had no indirect cost estimates"¹⁵. The authors of this study concluded that "...little reliance can be placed on the results of most of these reviewed studies to inform policy making for any disease"¹⁵. Another review of COI studies concluded: "... the overwhelming majority of 'specific' COI studies ... have an informative function only"¹⁶.

Estimation of the national costs of illness for OA is hampered by difficulties that can only slowly be overcome as we improve our methodologies. The study by Harrold and colleagues as well as other COI studies in rheumatology provide new understanding based upon which we can revise previous estimations.

ACKNOWLEDGMENT

The author gratefully acknowledges the helpful comments and insights received from Dr. Ed Yelin in the preparation of this manuscript.

REFERENCES

Search PubMed for:

1. Yelin E, Callahan LF. The economic cost and social and psychological impact of musculoskeletal conditions. National Arthritis Data Work Groups. Arthritis Rheum 1995;38:1351-62. [MEDLINE]

2. Lawrence RC, Helmick CG, Arnett FC, et al. Estimates of the prevalence of arthritis and selected musculoskeletal disorders in the United States. Arthritis Rheum 1998;41:778-99. [MEDLINE]

3. Harrold L, Yood RA, Straus W, et al. Challenges of estimating health service utilization for osteoarthritis patients on a population level. J Rheumatol 2002;29:1931-6.

4. US Department of Labor. CPI (U.S.)(82-84=100). Monthly medical care commodities January 1967-March 2002. 30-4-2002. CPI Detailed Report (US Dept. of Labor) Monthly. CANSIM Series Label D139122.

5. Rice DP. Cost of musculoskeletal conditions. In: Praemer A, Furner S, Rice DP, editors. Musculoskeletal conditions in the United States. Chicago: American Academy of Orthopaedic Surgeons; 1999.

6. Gabriel SE, Crowson CS, Campion ME, O'Fallon WM. Direct medical costs unique to people with arthritis. J Rheumatol 1997;24:719-25. [MEDLINE]

7. Gabriel SE, Crowson CS, Campion ME, O'Fallon WM. Indirect and nonmedical costs among people with rheumatoid arthritis and osteoarthritis compared with nonarthritic controls. J Rheumatol 1997;24:43-8. [MEDLINE]

8. Maetzel A, Li LC, Pencharz J, Maguire L, Bombardier C. A prospective comparison of the costs of illness among patients with RA, OA, or hypertension [abstract]. Arthritis Rheum 2001;44 Suppl:S310.

9. Koopmanschap MA, van Ineveld BM. Towards a new approach for estimating indirect costs of disease. Soc Sci Med 1992;34:1005-10. [MEDLINE]

10. Hutubessy RC, van Tulder MW, Vondeling H, Bouter LM. Indirect costs of back pain in the Netherlands: a comparison of the human capital method with the friction cost method. Pain 1999;80:201-7. [MEDLINE]

11. Boonen A, van der Heijde D, Landewe R, et al. Work status and productivity costs due to ankylosing spondylitis: comparison of three European countries. Ann Rheum Dis 2002:61:429-37. [MEDLINE]

12. MacLean CH, Knight K, Paulus H, Brook RH, Shekelle PG. Costs attritutable to osteoarthritis. J Rheumatol 1998;25:2213-8. [MEDLINE]

13. Clarke E, Penrod J, St. Pierre Y, et al. Underestimating the value of women: assessing the indirect costs of women with systemic lupus erythematosus. Tri-Nation Study Group. J Rheumatol 2000;27:2597-604. [MEDLINE]

14. Callahan LF. The burden of rheumatoid arthritis: facts and figures. J Rheumatol 1998;25 Suppl 53:8-12. [MEDLINE]

15. Bloom BS, Bruno DJ, Maman DY, Jayadevappa R. Usefulness of US cost-of-illness studies in healthcare decision making. PharmacoEconomics 2001;19:207-13. [MEDLINE]

16. Koopmanschap MA. Cost-of-illness studies. Useful for health policy? PharmacoEconomics 1998;14:143-8.