醫院門診處方警示系統採用情形與不適當處方之相關探討

Abstract

用藥安全是醫療照護的重要環節，美國IOM (Institute of Medicine)報告提出醫療組織運用健康資訊系統有助於提升照護品質和改善病人安全，其中電腦醫令系統(Computerized Physician Oreder Entry, CPOE)結合臨床決策支持系統(Clinical decision support system, CDS)而成的處方警示系統(medication alert system；CPOE/CDS)則可協助醫事人員避免開立不適當處方。根據相關文獻，各國醫院採用CPOE/CDS的差異性很大，針對特定系統的調查研究甚少，也很少研究從宏觀面探討醫院門診採用系統對不適當處方的影響。台灣2005年調查醫療院所電子病歷使用狀況，區域級以上醫院使用CPOE的比例超過90%，地區醫院也有達75%以上；同年，新制醫院評鑑用藥安全規範增列「處方警示系統」項目，相比於國外的系統發展趨勢，台灣有相當程度的優勢可以深入探討處方警示系統的使用及影響，因此本研究的第一個目的是採用問卷調查台灣地區級以上醫院門診使用處方警示系統項目及警示內容與設定條件；第二個目的則是探討處方警示系統採用趨勢和不適當用藥(重複用藥以及藥物交互作用)變化的相關性。 研究材料與方法分為兩個部份，先是利用橫斷式問卷調查地區級以上醫院採用CPOE/CDS的情況，郵寄問卷380家，調查期間2013年7月至9月，最後以216家回覆問卷之醫院為研究樣本，重要觀察變項包含醫院門診處方警示系統的採用趨勢、各類警示系統的警示內容與條件、醫院創新特性(早期採用者或晚期)、系統採用狀態(採用類別、穩定度、設計維護)等。接著檢測重複用藥警示系統與藥物交互作用警示系統的採用趨勢和醫院門診不適當用藥(重複用藥、口服糖尿病藥物交互作用、高血壓藥物交互作用)的相關性，以醫院為分析單位，且根據不同依變項，有不同研究樣本。資料來源是全民健康保險學術研究資料庫2005年百萬歸人檔(1997-2011年)、藥品代碼與ATC對照表、以及研究者整理之藥物交互作用配對檔。重複用藥率的計算方式有兩種：1.醫院門診同張處方內含兩種ATC前五碼相同之藥品的處方箋比例(研究樣本259家醫院)；2.醫院內同一位門診病人在重疊日期於不同醫師處方含兩種以上ATC前五碼相同藥品之處方箋比例(研究樣本236家醫院)。口服糖尿病藥物與高血壓用藥之一級藥物交互作用定義係參考Mircormedex2.0 (嚴重度Major且文獻等級good)，藥物交互作用處方率計算方式是醫院門診一級交互作用處方箋佔標的藥物處方箋的比例(研究樣本208家醫院)。自變項為時間，有六個觀察時期(1997年、1998-2000、2001-2003、2004-2006、2007-2009、以及2010-2012年)；控制變項為醫院特性，包含層級別、權屬別、分局別、醫院處方平均用藥數、醫院門診病人平均年齡。統計分析採用SAS 9.3包含描述性統計(平均數和百分比)以及推論性統計(卡方檢定和廣義估計方程式)。 研究結果發現醫院門診處方警示系統的採用率隨時間增加(1997年小於10%，到2010-2012年是95.83%)，規模較大以及組織部門多元化的醫院(醫學中心、教學醫院、500床以上、財/社團法人醫院)，顯著為早期採用者，傾向應用技術性高的系統，且多數是由醫院自行設計維護。同處方重複用藥相關因素分析當中，相比於1997年，1998-2000年b值-2.3830(P<0.0001)，2010-2011年的b值-5.5333 (P<0.0001)；同院跨處方重複用藥分析結果則顯示1998-2000年b值-5.58164 (P<0.0001)，2010-2011年b 值-22.5398 (P<0.0001)。然而藥物交互作用處方相關因素分析卻發現口服糖尿病藥物交互作用處方率和高血壓藥物交互作用處方率在2004-2006年b值分別是1.0217 (P<0.0001)和0.7288 (P=0.0032)，到2010-2011年則是2.0035(P<0.0001)以及1.7863 (P<0.0001)。歸納來說，重複用藥警示系統採用率增加，重複用藥發生情形隨時間顯著減少，但藥物交互作用警示系統卻沒有顯著效果。

Background Drug safety is an important issue in health care delivery system nowadays. The Institute of Medicine (IOM) has indicated that health information technology (HIT), especially computerized physician order entry system (CPOE) embedded with clinical decision support system (CDS), can assist physicians to avoid inappropriate prescribing, reduce medication errors, and improve care quality. However, the adoption of CPOE/CDS seemed to be different among countries. Many of the previous studies which have conducted surveys on the usage of CPOE/CDS for specific alert items were limited, and only a few reports have assessed the impact or performance of CPOE/CDS. Therefore, this study aimed to understand the adoption of CPOE/CDS among hospitals, and to examine the association of CPOE/CDS adoption and inappropriate prescriptions (include medication duplication and drug-drug interaction) comprehensively in Taiwan. Material & methods In the first part of this study, a cross-sectional questionnaire was developed and mailed to 380 hospitals nationwide between July and September, 2009. The variables listed included: (1) time of adopting CPOE/CDS in the hospitals' outpatient departments (with 6 time periods: before year 1997, 1998-2000, 2001-2003, 2004-2006, 2007-2009, 2010-2012) , (2) alert conditions and items in the CPOE/CDS, and (3) the operation status of CPOE/CDS (complexity, stability, and maintenance). In the second part, a longitudinal NHI dataset was applied to examine the association between the adoption of the CPOE/CDS system and inappropriate medication at hospital level. We evaluated two kinds of medication duplication and drug-drug interaction alert functions. The first duplicated medication was within a single prescription and was identified as 'a patient received drugs with the same therapeutic effect (as defined by level 4th of Anatomical Therapeutic Chemical system) in one prescription', and the medication duplication rate (1) was calculated as (cases of medication duplication) / (total number of prescriptions in a hospital) x 100%. The second duplicated medication was across prescriptions and was defined as ' a patient received drugs with the same therapeutic effect in different prescriptions provided by different physicians at the same hospital within one treatment period', and the medication duplication rate (2) was calculated as (the cases of medication duplication) / (total number of prescriptions provided in a hospital within same treatment period) x 100%. Concerning the drug-drug interaction, we only focused on the drugs for diabetes and hypertension treatment in 208 hospitals. We defined the drug-drug interaction (DDI) based on the severity level (contraindicated/major) and evidence level (excellent/good), and the DDI rate was calculated as (the cases of target DDI / total number of target prescriptions provided in a hospital). In the GEE regression models, the adoption time was classified into 6 periods: years before 1997, 1998-2000, 2001-2003, 2004-2006, 2007-2009, and 2010-2011. Hospital's characteristics were considered in the analysis which included accreditation level, ownership, branch of NHI, the average number of drugs within one prescription, and the average age of patients in outpatient departments. Results A total of 208 hospitals completed and returned the questionnaires with a response rate of 56.8%. Adoption rate of CPOE/CDS increased from less than 10% before year 1997 to 95.83% after 2010. Most of the large-scale hospitals were early adopters and tended to implement advanced systems designed by themselves. After adjusting for related variables, medication duplication rate (1) decreased alone the time significantly (b=-2.3830 for years 1998-2000 and b=-5.5333 for years 2010-2011 with P<0.0001repectively; reference year was 1997), and medication duplication rate (2) also decreased alone the time (b=-5.58164 for 1998-2000 and b=-22.5398 for 2010-2011 with P<0.0001 respectively). However, both of diabetes and hypertension DDI rates were increased after years 2004-2006 (for DM, b=1.0271 for years 2004-2006 and b=2.0035 for 2010-2011 with P<0.0001 respectively; for hypertension, b=0.7288 for years 2004-2006 , P=0.0032 and b=1.7863 for 2010-2011, P<0.0001). Discussion and conclusion In Taiwan, hospitals introduced CPOE/CDS during 1997and 2010 and have reached 95% adoption rate. We found that medication duplication in hospitals has decreased significantly along with the development of hospital's medication alert system. However, the drug-drug interaction was not decreasing which might be owing to the increased prevalence of chronic conditions and the NHI drug payment design. More detailed evaluation of the impact of medication alert system is needed in the future.