醫師服務量、遵循指引與糖尿病照護

Abstract

目的：探討醫師糖尿病服務量與遵循糖尿病治療指引的關係；醫師糖尿病服務量與遵循糖尿病預防指引的關係；醫師服務量與早期發現糖尿病的關係中，遵循糖尿病預防指引是否具中介效果。 方法：主要資料來源為全民健康保險申報資料的民眾就醫資料檔與醫師基本資料檔。自變項為醫師糖尿病服務量，採計醫師糖尿病人量，並分高中低三組。 醫師糖尿病服務量與遵循糖尿病治療指引的關係為多年期觀察研究，選擇2001、2005與2009年確診當日已接受單方口服降血糖藥的新發糖尿病患為對象，主要依變項為初次降血糖藥處方是否使用metformin—2006年指引的新建議，以兩個時間虛擬變數控制時間固定效果，並選擇隨機截距模型之羅吉斯迴歸（logistic regression model with random intercept）控制具有群聚傾向的潛在醫師特性或病人選擇醫師偏好的影響。 醫師糖尿病服務量與遵循糖尿病預防指引的關係，以2010年43歲以上新發第二型糖尿病患為研究對象，選擇隨機截距模型之羅吉斯迴歸探索個案發病前三年內經常就診醫師的糖尿病服務量與其接受指引建議的血糖檢測的可能性。 至於醫師糖尿病服務量與早期發現的關係中，遵循糖尿病預防指引是否具中介效果的驗證，同樣以2010年43歲以上新發第二型糖尿病患為研究對象，以確診90天內未固定用藥為早期發現，選擇因果中介分析檢定中介效果。 結果：較之求助「低」服務量的醫師，求助「高」或「中」糖尿病量的醫師較可能獲得metformin的處方—風險對比值（odds ratios，ORs）分別為2.48（95% CI：2.03-3.04）與1.76（95% CI：1.45-2.13），支持醫師服務量越高越可能提供指引建議的治療服務的假說。 86.5%於確診前三年曾接受至少一次血糖檢測。較之經常就診醫師屬「低」服務量者，醫師屬「高」或「中」服務量的新發糖尿病患較可能於確診前接受定期血糖檢測—風險對比值分別為1.74（95% CI：1.66-1.82）與1.18（95% CI：1.13-1.24），支持醫師服務量越高越可能提供指引建議的預防服務的假說。 2010年新發糖尿病患63.1%確診時屬早期發現。醫師服務量和醫師提供指引建議的預防性服務合計增加了5.9%的早期發現可能性（OR=1.059，p<0.001）。醫師服務量對早期發現的影響，12.87%係透過「醫師提供指引建議的預防性服務」（p<0.001），中介效果存在，屬部分中介，支持在醫師服務量與早期發現糖尿病的關係中，病人接受指引建議的糖尿病篩檢具中介效果。 結論：醫師服務量高者，提供指引建議的治療或預防性服務的可能性較高，早期發現的可能性也較高。醫師服務量對早期發現的影響 12.87%係透過「醫師提供指引建議的糖尿病篩檢」，均符合研究假說。惟高醫師服務量或提供指引建議的預防性服務對早期發現的影響力有限。

Objective: To examine the association between physician volume and clinical practice guideline adherence; to examine the association between physician volume and prevention guideline ad-herence; to examine the mediating effects of adherence to prevention guideline on the relation-ship of physician volume and early detection of diabetes. Methods: These studies employed nationwide claims data from Taiwan’s National Health In-surance scheme. The researcher introduced a multi-year observational study to examine the association be-tween physician volume and clinical practice guideline adherence. The study subjects were the newly diagnosed type 2 diabetes patients who received a single oral anti-hyperglycemic agent at the index date identified from 2001, 2005, and 2009 cohorts. The independent variable was a physician’s service volume, measured by the number of total patients with diabetes. The out-come variable was whether the newly diagnosed patients received metformin, which was the 2006 guideline-recommended medicine in Taiwan, as the anti-hyperglycemic agent at the index date. A random intercept logistic regression model used to control potentially unobserved char-acteristics of health care providers which could affect the choice of the initial oral glucose-lowering agent. The researcher selected newly diagnosed type 2 diabetes patients over the age of 43 in 2010 as the research subjects, and also used the random intercept model to explore the associa-tion between physician volume and prevention guideline adherence. The independent variable was a physician’s service volume, measured by the number of total patients with diabetes during 2007-2009. The outcome variable was whether the newly diagnosed patients received blood glucose test at least once within three years prior to diagnosis. Finally, the researcher chose causal mediation analysis to examine the mediating effects of adherence to prevention guideline on the relationship of physician volume and early detection of diabetes. Early detection of diabetes was defined as the medication possession ratio (MPR) of anti-hyperglycemic agent <80% within 90 days of diagnosis. Results: Patients visiting physicians with high or medium volumes were more likely to be pre-scribed metformin than those visiting physicians with low volumes; the odds ratios (ORs) of 2.48 (95% CI: 2.03-3.04) and 1.76 (95% CI: 1.45-2.13), respectively. These findings supported the hypothesis that physicians with higher service volumes are more likely to provide guide-lines-recommended treatment services. 86.5% of cases received blood glucose test at least once within three years prior to diagno-sis. Patients visiting physicians with high or medium volumes were more likely to receive regu-lar blood glucose test than visiting physicians with low volumes, with ORs of 1.74 (95% CI: 1.66-1.82) and 1.18 (95% CI: 1.13-1.24), respectively. These findings supported the hypothesis that physicians with higher service volumes are more likely to provide guidelines-recommended preventive services. In 2010, 63.1% of newly diagnosed diabetes were detected at the early stage. Physicians’ service volumes and provision of guideline-recommended preventive services together increased the likelihood of early detection by 5.9% (OR=1.059, p<0.001). 12.87% of total effect of physi-cians’ service volumes on early detection was attributed to the provision of guideline-recommended preventive services (p<0.001). The provision of guideline-recommended preven-tive services was a mediator of the relationship between physician volume and early detection of diabetes. Conclusions: Physicians with higher volume were more likely to provide guideline-recommended services, both for treatment or prevention. And they were more likely to detect diabetes at the early stage. The provision of guideline-recommended preventive services was a mediator of the relationship between physician volume and early detection of diabetes. However, higher physician volume or provision of guideline-recommended preventive services had limited impact on early detection of diabetes.