以傾向分數配對法評估糖尿病論質計酬方案之成效

An Effectiveness Evaluation of a pay-for-performance Program for Diabetes Based on the Propensity Score Matching Method

10.6288/TJPH2010-29-01-06

林文德(Wen-Der Lin)；謝其政(Chi-Jeng Hsieh)；邱尚志(Shang-Jyh Chiou)；吳慧俞(Hui-Yu Wu)；黃一展(I-Chan Huang)

糖尿病 ； 論質計酬 ； 傾向分數配對法 ； 可避免住院 ； diabetes ； pay-for-performance ； propensity score matching ； avoidable hospitalization

台灣公共衛生雜誌

29卷1期（2010 / 02 / 01）

54 - 63

繁體中文

目標：健保局於2001年11月推行糖尿病論質計酬方案，過去研究雖呈正向的成效，但未考慮加入方案者選擇偏差的問題，因此本研究目的旨在以糖尿病可避免住院為品質結果指標，並以傾向分數法配對產生可比較之配對組，以釐清該方案之成效。方法：擷取2002-2003年承保抽樣歸人檔中因糖尿病至門診就診並申報檢查者前後一年之就醫資料，藉由傾向分數(PS, propensity score)配對法，先依病患各項特性建構加入方案之機率（即PS）模型，再自未加入方案者（即對照組，共6,855人）中，以1:1的方式配對產生與加入方案者（即介入組，共647人）加入機率相當之配對組樣本（647人），之後再以廣義估計方程式比較組間可避免住院勝算之比值。結果：對照組在前一年之病患特性、就醫情形、檢查申報比例及住院等變項上與介入組有顯著差異，但配對組與介入組各變項之差異均不顯著。方案實施後，介入組之各項檢查申報比例均顯著高於對照組與配對組，但其糖尿病相關可避免住院率增加幅度高於對照組（勝算比1.01, 95%信賴區間為0.98-1.04），卻低於配對組（勝算比0.98，95%信賴區間為0.94-1.02），惟均未達顯著意義。結論：參加論質計酬方案者與未參加者之特性及過去檢查申報比例並不相同。經傾向分數配對法產生與介入組條件相當之配對組，並據此評估論質計酬方案之成效，與傳統上以未參加者為對照組的比較方式有不同的結果。以傾向分數法做為政策評估的工具，或可提供健康政策成效評估的另一種觀點。

Objectives: This study aimed to evaluate the pay-for-performance (P4P) program for diabetes by using the propensity score method to construct a comparable matching group, and by using the frequency of diabetes-related avoidable hospitalizations to assess outcome. Methods: In order to construct the intervention group, we selected diabetic cases (N=647) from the NHI beneficiaries claim data from 2002 to 2003. Using the propensity score, we selected the matched group (N=647) from the control group (N=6,855) by matching their characteristics and covariates relating to the probability of their participation in the program. Subsequently, we took a generalized equation estimate (GEE) approach with logit link to compare the likelihood of avoidable hospitalization among groups. Results: The distributions of predicators such as patient characteristics, prior ambulatory care utilization, and the rates of laboratory tests and hospitalizations were significantly different between the intervention group and the control group while there were no significant differences between the intervention group and the matched group. Although the proportion of laboratory tests was higher in the intervention group, after the P4P program this group had a higher likelihood of avoidable hospitalization than did the control group with an odds ratio (OR) of 1.01 (95% C.I.: 0.98-1.04) while having a lower likelihood than the matched group with an OR of 0.98 (95% C.I.: 0.94-1.02). Neither difference was significant. Conclusions: The characteristics of the diabetics who participated in the P4P were different from those who did not. After using propensity scores to create a matched group with the same probability as those who participated in the P4P, we found that the results were different from comparisons made with traditional methods. The propensity score method may provide another insight into understanding the effectiveness of a particular health policy.

1. quality indicators-prevention quality indicators overview
2. Barnett TE,Chumbler NR,Vogel WB,Beyth RJ,Qin H,Kobb R.(2001).The effectiveness of a care coordination home telehealth program for veterans with diabetes mellitus: a 2-year follow-up.Am J Manag Care,12,467-474.
3. Brook RH,McGlynn EA,Shekelle PG.(2000).Defining and measuring quality of care: a perspective from US researchers.Int J Qual Health Care,12,281-295.
4. Chang RE,Lin SP,Swei SC.(2008).Proceedings of the First Annual Meeting on Center of Health Insurance Research.Taipei, Taiwan:Center of Health Insurance Research.
5. Cook NR.(2008).Statistical evaluation of prognostic versus diagnostic models: beyond the ROC curve.Clin Chem,54,17-23.
6. D''Agostino RB(1998).Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group.Stat Med,17,2265-2281.
7. Fu AZ,Dow WH,Liu GG.(2007).Propensity score and difference-in-difference methods: a study of second-generation antidepressant use in patients with bipolar disorder.Health Serv Outcomes Res Method,7,23-38.
8. Gonnella JS,Hornbrook MC,Louis DZ.(1984).Staging of disease. A case-mix measurement.JAMA,251,637-644.
9. Niefeld NR,Braunstein JB,Wu AW,Saudek CD,Weller WE,Anderson GF.(2003).Preventable hospitalization among elderly Medicare beneficiaries with type 2 diabetes.Diabetes Care,26,1344-1349.
10. Parsons LS.(2001).Reducing bias in a propensity score matched-pair sample using greedy matching techniques.Proceedings of the Twenty-Sixth Annual SAS® Users Group International Conference,Cary, NC:
11. Rosenbaum PR,Rubin DB.(1983).The central role of the propensity score in observational studies for causal effects.Biometrika,70,41-55.
12. Rosenthal MB,Dudley RA.(2007).Pay-for-performance: will the latest payment trend improve care?.JAMA,297,740-744.
13. Rosenthal MB,Frank RG,Li Z,Epstein AM.(2005).Early experience with pay-for-performance: from concept to practice.JAMA,294,1788-1793.
14. Saultz JW,Lochner J.(2005).Interpersonal continuity of care and care outcomes: a critical review.Ann Fam Med,3,159-166.
15. Stümer T,Joshi M,Glynn RJ,Avorn J,Rothman KJ,Schneeweiss S.(2006).A review of the application of propensity score methods yielded increasing use, advantages in specific settings, but not substantially different estimates compared with conventional multivariable methods.J Clin Epidemiol,59,437-447.
16. Tarakji KG,Brunken R,McCarthy PM(2006).Myocardial viability testing and the effect of early intervention in patients with advanced left ventricular systolic dysfunction.Circulation,113,230-237.
17. Weissman JS,Gatsonis C,Epstein AM.(1992).Rates of avoidable hospitalization by insurance status in Massachusetts and Maryland.JAMA,268,2388-2394.
18. 中央健康保險局(2001)。全民健康保險糖尿病醫療服務改善方案試辦計畫支付標準。台北:中央健康保險局。
19. 李玉春(2006)。全民健保支付與管理制度對病人就醫品質之影響：以可避免之住院為例。行政院衛生署九十五年度科技研究計畫，計畫編號DOH95-HI-1002。
20. 李玉春(2006)。建立全民健保以共同照護模式為基礎的糖尿病人疾病管理計畫先導性研究（Ⅲ）。行政院衛生署九十一委託計畫，計畫編號DOH91-NH-1002。
21. 李待弟(2006)。碩士論文(碩士論文)。台北，台灣大學衛生政策與管理研究所。
22. 侯佳雯(2004)。碩士論文(碩士論文)。台北，台灣大學醫療機構管理研究所。
23. 劉介宇、洪永泰、莊義利(2006)。台灣地區鄉鎮巿區發展類型應用於大型健康調查抽樣設計之研究。健康管理學刊，4，1-22。

1. 施培艾(2017)。家醫科與內分泌科糖尿病照護品質之比較。國立臺灣大學流行病學與預防醫學研究所學位論文。2017。1-58。 
2. 詹舒涵(2016)。不同醫師專科別及機構層級別對多重慢性病患的照護結果之探討。國立臺灣大學健康政策與管理研究所學位論文。2016。1-129。 
3. 蔡茜如(2014)。糖尿病醫療給付改善方案、照護過程品質與可避免住院之相關性探討。國立臺灣大學健康政策與管理研究所學位論文。2014。1-98。 
4. 譚家惠(2013)。醫療給付改善方案與照護過程對糖尿病病人健康結果之影響。國立臺灣大學健康政策與管理研究所學位論文。2013。1-123。