目標：台灣過去有關安眠藥物使用的研究，大多是對處方內容作分析，較難瞭解醫師臨床處方之考量。本研究透過深度訪談，探索醫師的安眠藥物處方行為。方法：研究對象經由從事醫療工作的朋友引見，以質性深度訪談，訪問17位有開立安眠藥物處方的執業醫師；資料的分析採用編輯式分析法。結果：影響醫師是否處方安眠藥物之考量因素包括有：醫師的專業科別、病人期待藥物的效果、恢復及維持病人的社會功能、建立和鞏固醫病關係；受訪醫師雖表示不見得會順應病人對藥物的要求，然而醫病之間的長久熟識，可能使得醫師較難抗拒病人對於藥物需求的習慣性期待。在藥物選擇上，病人過去用藥史，如果曾經使用安眠藥物，受訪醫師會處方原來的藥物給病人繼續服用；如果是首次藥物使用者，醫師大多處方非苯二氮平類的安眠藥物；對於非苯二氮平類藥物的信心，使得醫師易忽略使用這類藥物的風險。醫師對於處方訊息的告知，最常透露的是藥物名稱、服藥時間等；醫師認為病人可能不想獲得太多訊息，因此對於藥物各種作用的詳細說明、持續用藥追蹤諮詢與評估等，並未確實執行。大部分受訪醫師對於非藥物治療失眠方式所知有限，又因健保給付制度的影響，治療失眠仍以藥物為主。結論：醫療專業特質、社會規範、醫病互動以及保險支付制度對於醫師處方行為皆有影響，相關政府單位如能規劃論質計酬的支付方式，提供多元的治療選項，同時在醫師的繼續教育中，安排非藥物治療方式訓練，使得醫師可以擬定適合治療計畫，減低非必要的藥物使用。

Objectives: In Taiwan, the bulk of studies about hypnotics have focused on the content of prescriptions, with relatively less emphasis on the physician's clinical considerations. The objective of this study was to explore hypnotic prescribing behavior by physicians. Methods: Referred by friends engaged in medical work, 17 physicians who prescribed hypnotics consented to in-depth interviews. Data were subject to editing style analysis. Results: Factors influencing hypnotic prescribing behavior included physician specialization, patient expectation of drug effects, maintenance of social functioning of the patient, and the doctor-patient relationship. The physician might not want to respond to a patient's request, but the long-term doctor-patient relationship might still influence prescription behavior and the physician would prescribe the same drug again. If the patient were a first-time user, the physician would prescribe benzodiazepine receptor agonists. Because of confidence in non-benzodiazepine drugs, physicians would ignore the risk of using these drugs. Some physicians believed that patients did not want to get too many messages, and told their patients only the name of the medication and the time of ingestion, but not a detailed description of the effects or the need for continuous follow-up evaluation and consultation. Other than their impact on the health care payment system, most of the physicians had limited knowledge about non-pharmaceutical treatments for insomnia. As a result, the treatment of insomnia still heavily relies on prescription medications. Conclusions: Based on the results, we suggest that the government might adopt pay for performance to cover the cost of alternative medical treatments. To reduce the unnecessary use of hypnotics, physicians are encouraged to learn about other non-pharmaceutical techniques during continuing medication education, apply these techniques in clinical settings, and create proper treatment plans for patients with insomnia.