醫療資源耗用、疾病併發症與死亡率在糖尿病人介入組(論質計酬)與比較組之分析--台灣健保資料庫

Medical Utilizations, Disease Outcomes and All-cause Mortality between Intervention group (Pay-for-Performance) and Comparison group in Diabetes Patients of Taiwan:

陳宇清(Yu-Ching Chen)

指導教授 ： 許弘毅教授

高雄醫學大學/健康科學院/醫務管理暨醫療資訊學系碩士班/碩士(2014年)

https://doi.org/10.6832/KMU.2014.00098

摘要

摘要
背景：
雖然醫療資源耗用的資料已經被發表，知之甚少的是論質計酬（P4P）對糖尿病患者疾病併發症和存活的影響。
目的：
我們進行這項研究調查台灣P4P對糖尿病患者醫療資源耗用，疾病併發症及存活的影響。
方法：
總共有2,573人患有主診斷為糖尿病，新被納入台灣論質計酬方案於2004年1月1日至2004年12月31日組成介入組。以1比3比例，取性別，年齡和糖尿病初診斷年三變項採完全配對，得7,719糖尿病患為比較組進行比較。因此，在本研究中糖尿病的罹病長短已被納入控制。根據台灣的死亡登錄檔，我們追蹤這10,292例患者直到2009年12月31日。變項如月薪，居住區，Charlson合併症指數（CCI），糖尿病合併症的嚴重程度指數（DSCI），重大傷病（CDD）證書，門診次數，實驗室檢驗品項和測試頻率，抗糖尿病藥物處方，降血脂藥物處方， 抗高血壓藥物處方，急診就醫次數，每年的總醫療費用，醫院層級，入院天數，與在30天內再住院率都納入分析。採Cox回歸模型，分析P4P對總體死亡率的影響。
結果：
所有個案都有完整追蹤資料; 平均追蹤時間為5.18年（SD=0.98年）。在P4P組除了口服抗糖尿病藥，其他所有變項藉由GEE皆呈現顯著趨勢分析; 在non-P4P組除了入院天數，其他所有變項藉由GEE也皆呈現顯著趨勢分析。雖然門診次數，實驗室檢驗品項、測試頻率和三高藥物在P4P組較多。然而，入院天數和每年的總醫療費用在P4P組則較少。
透過Logistic回歸模型進行分析，於控制干擾變項後，stroke，any depressive disorders，bipolar disorder和a composite of seven diseases在P4P組都顯著減少; stroke, OR=0.92，CI（0.88，0.96）;any depressive disorders，OR = 0.94，CI為（0.88，0.99）;bipolar disorder, OR =0.69，CI為（0.57，0.84）;a composite of seven diseases，OR =0.91，CI為（0.88，0.94）。 10,292例患者中，有1,171（11.38％）死亡。在充分調整後的模型，最終的獨立危險因素是non-P4P，男性，年齡較大，鄉村居住，社會經濟地位較低，較高CCI，重大傷病（CDD）證書，更高的抗糖尿病藥物處方，降血脂藥物處方， 抗高血壓藥物處方，較高的年度總醫療費用，住院時間短，和更多的醫療檢驗。
由Kaplan-Meyer存活分析log-rank檢驗，P4P組存活是顯著較長的（p <0.001），介入組平均存活68.9個月而比較組為67.3個月。
基於年齡，性別，糖尿病診斷年三項完全配對，以Cox回歸模型分析死亡率，控制了年齡，性別，城鄉差異，鬱症狀和症狀性低血糖，與non-P4P組相比，P4P組的死亡風險是顯著降低，HR=0.62（信賴區間0.53-0.72，P<0.001）。
結論與建議：
與non-P4P組比較，雖然P4P組有較多門診次數，實驗室檢驗項目及頻率與使用較多藥物控制血糖，降低血脂和血壓，P4P組花費較低的年度醫療費用。 P4P減少stroke，any depressive disorders，bipolar disorder，以及a composite of seven diseases的發生率。控制潛在的confounding factors後，P4P可以提高糖尿病患者的存活率。如果上面的數據可以在2千3百萬計的全國人口資料檔予以證實，會更具說服力。

關鍵字

糖尿病 ； 論質計酬 ； 醫療資源耗用 ； 疾病併發症 ； 死亡率

並列摘要

Abstract
Background:
While data of medical utilizations have been reported, little is known about the impact of pay-for-performance (P4P) on disease outcomes and survival of patients with diabetes.
Objective:
We conducted this study to investigate the impact of P4P on medical utilizations, disease outcomes and survival of patients with diabetes in Taiwan.
Methods:
A total of 2,573 individuals with a primary diagnosis of diabetes and who had been newly enrolled into pay-for-performance services in Taiwan between January 1, 2004 and December 31, 2004 comprised the intervention group. They were
compared with comparison group of 7,719 diabetes individuals matched by gender, age and first calendar year of diabetes diagnosis. Therefore, the duration of diabetes was controlled in this study. Based on our national mortality data base, we followed these 10,292 patients until December 31, 2009. Insurance premium, residence, Charlson comorbidity index (CCI), diabetes severity comorbidity index (DSCI), catastrophic disabling disease (CDD) certificates, outpatients visits, laboratory items and frequency of medical test, anti-diabetes drugs, lipid-lowering drugs,
anti-hypertensive drugs, emergency medical visits, annual total medical cost, accredited level of hospital, duration of hospital admission, and re-admission within 30 days were all included in the analysis. Cox regression model was utilized to re-veal the effect of P4P on all-cause mortality.
Results:
Follow-up information was available for all individuals; the mean follow-up time was 5.18 years (SD= 0.98 years). While except oral anti-diabetic drugs, all other var-iables show significant trend analysis by GEE in P4P group, all other variables show significant trend analysis by GEE in non-P4P group, except admission durations. Physician visits, laboratory, and medications are more in P4P. However, admission duration and annual total cost get less in P4P.

Analyzed by logistic regression model, the result shows that stroke, depressive disorder, bipolar disorder and composite of 7 diseases are significantly reduced in P4P after controlling for controlling confounding covariates, with OR=0.92, CI (0.88, 0.96) for stroke, OR=0.94, CI (0.88, 0.99) for depressive disorder, OR=0.69, CI (0.57, 0.84) for bipolar disorder, OR=0.91, CI (0.88, 0.94) for composite of seven diseases.
Of 10,292 patients, 1,171 (11.38%) individuals expired. In the fully adjusted model, the final independent risk factors were non-P4P, male, older age, rural residence, lower socioeconomic status, higher CCI, CDD certificate, higher anti-hypertensive use, lower anti-lipid use, higher anti-diabetes use, higher annual total medical cost, short hospital stay, and more medical test.
Conclusions and Suggestions:
While P4P is associated with more physician visits, lab examination items and fre-quency and more medications for glycemic control, lipid lowering, and blood pressure, it has less annual medical cost, compared with non-P4P. P4P reduces the incidence of stroke, depressive disorders, bipolar disorder, and a composite of 7 diseases. P4P can improve survival of patients with diabetes after controlling potential confounding factors. It would be better if above data could be confirmed in nationwide claims of 23 millions population.

Key Words: Diabetes mellitus, Pay for performance, Medical utilizations,
Disease outcomes, Mortality

並列關鍵字

Diabetes mellitus ； Pay for performance ； Medical utilizations

參考文獻

References:

1. Genuth S, Eastman R, Kahn R, et al. Implications of the United kingdom prospective diabetes study. Diabetes care 2003; 26 Suppl 1: S28-32.

2. Nathan DM, Cleary PA, Backlund JY, et al. Intensive diabetes treatment and cardiovascular disease in patients with type 1 diabetes. The New England journal of medicine 2005; 353(25): 2643-53.

3. Christianson JB, Leatherman S, Sutherland K. Lessons from evaluations of purchaser pay-for-performance programs: a review of the evidence. Medical care research and review : MCRR 2008; 65(6 Suppl): 5S-35S.

4. Giorgino F, Leonardini A, Laviola L. Cardiovascular disease and glycemic control in type 2 diabetes: now that the dust is settling from large clinical trials. Annals of the New York Academy of Sciences 2013; 1281: 36-50.