Abstract
This paper empirically investigates the resource distribution dynamics across Diagnosis Related Groups (DRGs) of elective surgery patients, in a continuing Prospective Payment System (PPS). Existing econometric literature has mainly focussed on the impact of PPS on average Length of Stay (LOS) concluding that the average LOS has declined post PPS. There is little literature on the distribution of this decline across DRGs, in a PPS. The present paper helps fill this gap. It models the evolution over time of the empirical distribution of LOS across DRGs. The empirical distributions are estimated using a non parametric “stochastic kernel approach” based on Markov Chain theory. The results for inlier episodes suggest that resource redistribution will increase capacity and expected number of admissions for DRGs having increasing waiting times. In addition, adjustments in relative cost weights are perceived as price signals by hospitals leading to a change in their casemix. The results for high outlier patients reveal that improved quality of care is one of the factors causing reduction in high outlier episodes.
Similar content being viewed by others
Notes
LOS denotes average length of stay throughout the paper unless stated otherwise.
Therefore only large/metropolitan hospitals are effectively capable of reallocating the resources across DRGs, in response to changes in cost weights. Thus only such hospitals are used for empirical analysis in the present paper. These hospitals account for most of hospital separations in Victoria.
Here high cost patients refer to the patients having relatively higher cost within the distribution of costs of an individual DRG.
This argument is particularly valid for elective non-urgent episodes of patient care which are funded under PPS. Thus only such episodes are considered for empirical analysis in this paper.
It should be noted that ex-ante hospitals would like to specialize in the “lucrative” DRG treatment [28] but ex-post in a continuing PPS regime, even a specialist hospital will have an incentive to redistribute its resources in response to temporal changes in relative DRG weights.
For example, 5 year percentage change (from 2000 to 2005) in ratio of number of patients to total patients, for three main diagnostic categories are: 0.8% (Hip & Knee), 0.6% (Gynaecology) and 0.3% (procedures of digestive system).
We are thankful to an anonymous referee for this suggestion.
It should be noted that above factors are unobservables and data enables us to observe only number of patients treated in a particular DRG.
For the technical derivation of stochastic kernel interested readers can refer to Section 4 in Quah [26].
Emergency admissions are not included in the analysis because hospital care for such patients has very little discretionary capacity and thus no scope for resource redistribution. On the other hand, public hospital systems have genuine discretion for non-urgent elective surgery patients [3] which makes this subgroup of patients appropriate to test our hypotheses of resource dynamics.
The analysis for low outlier episodes is not done in our study as there are very few DRGs with low outlier episodes which makes it almost impossible to generate a balanced panel of DRGs over 8 years.
The econometric analysis is done using the tsrf shell provided by Danny Quah.
References
AIHW (2005) Australian hospital statistics. AIHW, Canberra
Arbia G, Basile R, Piras G (2005) Analyzing intra-distribution dynamics: a reappraisal. In: 46th congress of the European regional science association. Volos, 30 August–3 September 2006
Brook C (2008) Casemix funding for acute hospital care in Victoria, Australia. http://www.health.vic.gov.au/casemix/about.htm
Chung K (1967) Markov chains with stationary transition probabilities. Die Grundlehren der mathematischen Wissenschaften 104
Craig B, Sendi P (2002) Estimation of the transition matrix of a discrete-time Markov chain. Health Econ 11:33–42
DesHarnais S, Wroblewski R, Schumacher D (1990) How the medicare prospective payment system affects psychiatric patients treated in short-term general hospitals. Inquiry 27(4):382–8
Ehsani J, Jackson T, Duckett S (2006) The incidence and cost of adverse events in Victorian hospitals 2003–04. Med J Aust 184(11):551–555
Ellis R, McGuire T (1993) Supply-side and demand-side cost sharing in health care. J Econ Perspect 7(4):135–151
Ellis R, McGuire T (1996) Hospital response to prospective payment: moral hazard, selection, and practice-style effects. J Health Econ 15(3):257–277
Freiman M, Ellis R, McGuire T (1989) Provider response to Medicare’s PPS: reductions in length of stay for psychiatric patients treated in scatter beds. Inquiry 26(2):192–201
Victorian Department of Human Services V (1998–2006) Public hospitals policy and funding guidelines. http://www.health.vic.gov.au/pfg/
Iversen T (1993) A theory of hospital waiting lists. J Health Econ 12(1):55–71
Jackson T, Duckett S, Shepheard J, Baxter K (2006) Measurement of adverse events using ‘incidence flagged’ diagnosis codes. J Health Serv Res Policy 11(1):21–26
Jacobs R, Smith P, Street A (2006) Measuring efficiency in health care. Cambridge University Press, Cambridge
Ma C (1994) Health care payment systems: cost and quality incentives. J Econ Manage Strategy 3(1):93–112
Manton K, Woodbury M, Vertrees J, Stallard E (1993) Use of Medicare services before and after introduction of the prospective payment system. Health Serv Res 28(3):269–92
Moje C, Jackson T, McNair P (2006) Adverse events in Victorian admissions for elective surgery. Aust Health Rev 30(3):333–43
Newhouse J (1983) Two prospective difficulties with prospective payment of hospitals, or, it’s better to be a resident than a patient with a complex problem. J Health Econ 2(3):269–74
Newhouse J (1996) Reimbursing health plans and health providers: efficiency in production versus selection. J Econ Lit 34(3):1236–1263
Newhouse J (2002) Pricing the priceless: a health care conundrum. MIT, Cambridge
Newhouse J, Byrne D (1988) Did Medicare’s prospective payment system cause length of stay to fall? J Health Econ 7(4):413–6
Norton E, Van Houtven C, Lindrooth R, Normand S, Dickey B (2002) Does prospective payment reduce inpatient length of stay? Health Econ 11:377–387
Norton EC (1992) Incentive regulation of nursing homes. J Health Econ 11(2):105–128
Pagan A, Ullah A (1999) Nonparametric econometrics. Cambridge University Press, Cambridge
Quah D (1990) International patterns of growth: II. Persistence, path dependence, and sustained take-off in growth transition. Manuscript, Massachusetts Institute of Technology
Quah D (1997) Empirics for growth and distribution: stratification, polarization, and convergence clubs. J Econ Growth 2(1):27–59
Quah D (2004) Growth and distribution. http://econ.lse.ac.uk/staff/dquah/p/gnd.pdf
Rauner M, Zeiles A, Schaffhauser-Linzatti M, Hornik K (2003) Modelling the effects of the Austrian inpatient reimbursement system on length-of-stay distributions. OR Spectrum 25(2):183–206
Rosenblatt M (1956) Remarks on some nonparametric estimates of a density function. Ann Math Stat 27(3):832–837
Selden T (1990) A model of capitation. J Health Econ 9(4):397–409
Siciliani L (2006) Selection of treatment under prospective payment systems in the hospital sector. J Health Econ 25(3):479–499
Silverman B (1986) Density estimation for statistics and data analysis. Chapman & Hall/CRC, London
Street A, Duckett S (1996) Are waiting lists inevitable? Health Policy 36(1):1–15
Victoria S (2001) Casemix funding. Der Internist 42(4):75–79
Acknowledgements
I am thankful to three anonymous referees, Terri Jackson, Just Stoelwinder and Bruce Hollingsworth for detailed and useful comments on an earlier version. I also wish to thank Xibin Zhang, Alistair McGuire, Brett Inder, Anthony Harris and Jenny Watts for many helpful discussions. Preety Srivastava and Annette Thom kindly agreed to edit the final draft. Furthermore, seminar participants at the 16th European Workshop on Econometrics and Health Economics, Bergen, Norway provided valuable feedback. The usual disclaimer applies.
Author information
Authors and Affiliations
Corresponding author
Additional information
The research is partially funded by New Academic Grant. This paper is part of a project supported by National Health and Medical Research Council (NHMRC) Grant ID: 334114.
Appendix
Appendix
1.1 A.1 Estimating stochastic kernels
For estimation purposes, stochastic kernel can be written as [2]:
where S is the share in period t + n and S′ is the share in period t. f n (S|S′) is the conditional density which describes the probability that a DRG moves to a specific state of share, given the share in period t. Thus a stochastic kernel can be expressed as a conditional density and its estimator can be derived from the estimation of conditional density. A nonparametric estimator for the conditional density as proposed by Rosenblatt [29] is given by:
where the estimator for the joint density \(\hat{g}_n(S',S)\) is given by:
and the estimator for the marginal density \(\hat{h}_n(S')\) is given by:
where a and b are bandwidth parameters controlling the smoothness of fit, K is the chosen kernel function and ||S′ − S′ j || and ||S − S j || are the Euclidian metrics. Substituting Eqs. 9 and 10 in Eq. 8 the conditional density estimator can be rewritten as:
where
The above kernel estimator is the Nadaraya-Watson kernel regression estimator. It shows that a conditional density can be obtained by the sum of J kernel functions in S space weighted by the w i (S′) in S′ space.
1.2 A.2 Kernel choice and bandwidth selection
Throughout this appendix f is kernel estimate, K is kernel and h is bandwidth. Given that we have used general weight function estimate as the kernel estimate for any x, the expected value and variance of such an estimate is given by:
The expected value of \(\hat{f}\) is a smoothed version of the true density, obtained by convolving f with the kernel scaled by the band width. Thus the density estimate is of the form: smoothed version of true density + random error. The smoothed density depends on the choice of parameters through kernel and band width choice. Such a choice is critical for the discrepancy of density estimator \(\hat{f}\) from the true density f. Most widely used measure of such a discrepancy which reflects the global accuracy of \(\hat{f}\) as an estimator of f is the mean integrated square error (MISE) defined as:
which can be rewritten as:
the sum of the squared bias (bias h ) and the variance at x. Assuming kernel K is symmetric function satisfying: \(\int K(t) dt =1\), \(\int tK(t)dt=0\), and \(\int t^2K(t)dt=k_2\neq 0\) and replacing y = x − ht the integrated square bias is given by:
and the variance is given by:
Given the bias and variance, approximate value of MISE will be
The above equation reveals that there is a trade-off between the bias and variance terms: the bias can be reduced at the expense of increasing the variance, and vice versa, by adjusting the amount of smoothing or the bandwidth. Suppose h is chosen to minimise MISE. Choosing very low h will eliminate the bias but increase the integrated variance. On the other hand choosing high value of h will reduce the random variation (measured by variance) but will increase the bias or the systematic error. Thus choice of bandwidth implies a trade-off between random and systematic error.
Silverman [32] shows that the optimal value of bandwidth h that minimises approximate MISE is given by:
It should be noted that the optimal value of bandwidth depends on the unknown kernel density estimate. We first discuss the choice of this unknown kernel.
Choice of kernel:
Substituting the value of h opt (Eq. 15) back into formula for approximate MISE (Eq. 14) shows that if h is chosen optimally, then the approximate value of MISE will be
where the constant C(K) is given by:
Above formulae show that minimising MISE means choosing kernel K which minimises C(K) which in turn reduces to minimising \(\int K(t) dt\). The kernel which minimises this term is the Epanechnikov kernel.
Choice of Bandwidth
As discussed earlier, Least square cross-validation method, a completely automatic method based on the dataset on hand, is widely used for choosing bandwidth. It is based on minimizing integrated square error which for any estimator \(\hat{f}\) can be written as:
Since last term of above equation does not depend on \(\hat{f}\), the optimal choice of bandwidth corresponds to the choice which minimises \(\int\hat{f}^2 -2\int\hat{f}f\). The basic principle of least square cross-validation method is to construct an estimate of this term from the data themselves and then to minimise this estimate over h to give the choice of bandwidth. Thus the optimal value of bandwidth depends on the data set used and is not chosen subjectively.
Rights and permissions
About this article
Cite this article
Sharma, A. Inter-DRG resource dynamics in a prospective payment system: a stochastic kernel approach. Health Care Manag Sci 12, 38–55 (2009). https://doi.org/10.1007/s10729-008-9078-3
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10729-008-9078-3