A study on the cost-effectiveness of coronary revascularization: Introducing the simultaneous mimic health status model

Health Economics

Volumn 6, Issue 6, 1997, Pages 613-623

  Lee, Jae Sang ^a   Bailey, Martin J ^b   Jeong, Jinook ^c   Mauldin, Patrick D ^d   Weintraub, William S ^e  

^a POSCO   (South Korea)

^b EMORY UNIVERSITY   (United States)

^c Ajou University   (South Korea)

^d MEDICAL UNIVERSITY OF SOUTH CAROLINA   (United States)

^e EMORY UNIVERSITY SCHOOL OF MEDICINE   (United States)

Author keywords

CABG; Econometric model; PTCA

Indexed keywords

ARTICLE; CORONARY ARTERY BYPASS SURGERY; CORONARY ARTERY DISEASE; COST EFFECTIVENESS ANALYSIS; FOLLOW UP; HEALTH STATUS; INTERMETHOD COMPARISON; MODEL; PERCUTANEOUS TRANSLUMINAL ANGIOPLASTY; PRIORITY JOURNAL; REVASCULARIZATION;

EID: 0031435792     PISSN: 10579230     EISSN: None     Source Type: Journal    
DOI: 10.1002/(sici)1099-1050(199711)6:6<613::aid-hec305>3.3.co;2-y     Document Type: Article

References (19)

1. National Center for Health Statistics (NCHS). Detailed diagnosis and procedures for patients discharged from short-stay hospitals. United States: Vital Health Statistics, 1987.
2. Mauldin, P. D., Jeong, J., Becker, E. R. and Weintraub, W. S. Provider cost and coronary revascularization: an econometric approach. Econometrics of Health Care Evaluation, 1994; 7: 87-98.
3. Berg, R. L. Health Status Indexes: Proceedings of a Conference Conducted by Health Service Research. Chicago: Chicago Hospital Research and Educational Trust, 1973.
4. 3 defines a latent variable HSI (Health Status Index) as a weighted index of the health indicators, based on one of the suggestions of the conference on Health Status Indexes (Tucson, AZ, 1972). His specification, however, has a serious limitation. Although he defines health improvement as a weighted index of various indicators, the weights in his model were arbitrarily assigned by researchers or were determined by separate surveys. See Ref. 3, pp. 254-255.
5. i and η is a parameter vector to be estimated to correct the selectivity bias. Because a patient chooses a procedure in their model, it suffers from selectivity bias without the use of IMR.
6. Van De Ven, Wynard, P. M. M. and Hooijmans, E. M. The MIMIC health status index (what it is and what it does). In: Econometrics of Health Care, pp. 19-29. Dordrecht: Kluwer, 1991.
7. Joreskog, K. G. and Goldberger, A. S. Estimation of a model with multiple indicators and multiple cause of a single latent variables. Journal of the American Statistical Association 1975; 70: 631-639; Van De Ven, Wynard, P. M. M. and Van Der Gaag, J. Health as an unobservable: A MIMIC-model of demand for health care. Journal of Health Economics 1982; 1: 157-183.
8. Joreskog, K. G. and Goldberger, A. S. Estimation of a model with multiple indicators and multiple cause of a single latent variables. Journal of the American Statistical Association 1975; 70: 631-639; Van De Ven, Wynard, P. M. M. and Van Der Gaag, J. Health as an unobservable: A MIMIC-model of demand for health care. Journal of Health Economics 1982; 1: 157-183.
9. 1 are for difference between PTCA and CABG.
10. i matrix includes many 'zero' elements.
11. Joreskog, K. G. and Sorbom, D. DOS LISREL 8.02. Scientific Software International, 1981-93.
12. 1 is zero.
13. Joreskog, K. G. and Sorbom, D. DOS LISREL. 8.02. Scientific Software International, 1981-93.
14. King, S. B., Lembo, N. J., Weintraub, W. S., et al. A randomized trial comparing coronary angioplasty with coronary bypass surgery: The Emory Angioplasty Surgery Trial. New England Journal of Medicine 1994; 331(16), 1044-1050.
15. WEST is eligible but non-randomized data and simply means the opposite of EAST.
16. In the literature, there exist a number of alternative methods for self-selection bias correction, Jeong et al. (1996, Emory University Working Paper), for example, compare alternative correction schemes for self-selection bias. The utilization of IMR, which is adopted in this study, is a simpler way to reduce self-selection bias. As one of the referees pointed out, an IV approach could offer a broader cure for selection bias problem. In this paper, however, we did not pursue a more complex bias-correction scheme because this WEST data result is not a main focus of this work. The WEST result is presented for a comparison with the EAST result, which is not contaminated by self-selection problems. The comparison with several remedies for self-selection bias of WEST data will be the subject of future research.
17. Selection of indicators is important because, like most econometric methods, our method is 'modeldependent'. If the model is wrong, say, if a relevant indicator is omitted, the result of our method will be misleading. We interviewed physicians to decide the best set of available indicators and tested the significance of the estimated weights of the indicators not to include irrelevant ones. However, for comparison, the same set of indicators was used for WEST data, regardless of the significance.
18. The t-value is not applicable for this variable because it is required to set an indicator to 1 to estimate other variables in LISREL, and the variable PCHGANG is used for this purpose.
19. Note that the selectivity bias is a phenomenon of non-zero mean of the error term.