目標：近年降血壓用藥所費不貲，須分析各醫療區、各類型醫院門診降血壓處方品質之差異。方法：以健保局2009－2011年總額協商醫療品質之兩項負向指標－門診口服降血壓藥物給藥日數重複率以及具高鉀血症之高血壓病患使用保鉀型利尿劑或醛固酮拮抗劑比率，將指標值轉為二元變數，若指標值≦全國指標值，令為1，反之令為0。收集23個醫療區398家醫院資料，採多層次邏吉斯迴歸分析高層次變項（市場競爭、每萬人醫師、人口密度、健保分局別）和低層次變項（權屬別、評鑑等級、教學狀態）對個別醫院處方品質之影響。市場競爭測量以門診人次計算賀芬達（H）指數。結果：人口密度和H愈高，醫院給藥重覆低的可能性愈大（0.0001, p<.01），每萬人醫師愈多，低重覆的機率愈小（-0.018, p<.001），台北區、南區、高屏區醫院低重覆的機率比東區醫院小（-0.36, p<.001; -0.192, p<.05; -0.25, p<.01），教學醫院比非教學醫院小（-0.316, p<.01），醫學中心比地區醫院大（0.29, p<.01）。市場競爭、醫院特徵對高血鉀用藥品質無顯著影響，但台北區、北區、南區、高屏區醫院高血鉀適當用藥的可能性比東區醫院大（0.424, 0.343, 0.316, 0.324, all p<.001）。醫院兩項處方品質的變異分別有10.25%, 17.63%可歸因醫院層次，82.52%, 73.40%歸於市場層次。結論：本文呈現近三年不同醫療區醫院降血壓藥日數重複及高血鉀不當處方之概況，供衛生政策之參考。

Objectives: To analyze inappropriate prescriptions at the market and hospital levels by examining antihypertensive medication ordering quality in out-patient departments (OPD) in Taiwan hospitals. Methods: Our sample included 398 accredited acute hospitals located in 23 regions. Information about hospital's ownership, rank, teaching status, OPD visits and hospital number published by the Department of Health (DOH) and information from the 2009-2011 Quality Database of Medications Indicators published by National Health Insurance were collected. The population density (PD) and physician density per 10000 population (PDPP) in the 23 regions were generated from DOH data. Two inappropriate prescription indicators were defined as duplicate medications of oral blood pressure-lowering agents or hypertension patients with a history of hyperkalemia using PSD or AA. We used MLLR models to assess multilevel effects on the probability of good or bad hospital quality for medications. The market concentration in the 23 regions was calculated by Herfindahl-Hirschman Index (H) using OPD visits to measure hospital competition. Results: Our study showed that higher H and PD led to greater probability of low duplicate medications. A higher PDPP led to lower probability of low duplicate medications. When compared to hospitals located in the Eastern region, the probability of low duplicate medications in Taipei, Southern, Kaohsiung-Pingtung areas is lower. The probability of low duplicate medications in teaching hospitals is lower than non-teaching hospitals and medical centers are higher than local hospitals. The probability of appropriately treating patients with hyperkalemia in Taipei, Northern, Southern, and Kaohsiung-Pingtung areas was higher than that of Eastern areas. Overall, the variation of medication quality can be explained 10.25%-17.63% by hospital-level variables and 73.40%-82.52% by market-level variables. Conclusions: Hospital and market characteristics impacted inappropriate prescriptions of anti-hypertensive drugs in the OPD setting over the past 3 years.