應用關聯規則在門診處方行為與潛藏藥物交互作用分析的研究

Apply Association Rules on the Analysis of Out-patient Prescription Patterns and Potential Drug-drug Interactions

醫師開立處方時不僅要考慮病人的疾病型態、嚴重度、年紀、可供應的藥品，還會受到本身經驗的限制與病人個別要求等因素的影響。在目前醫療資源不斷緊縮、各種新藥、替代藥物的種類不斷的增加，以及醫療院因應健保制度所進行的管控措施的交互影響之下，目前醫師的開藥行為模式是一個值得深入探討的問題。此外，目前國內門診處方箋每張平均包括約4.1種藥品，是先進國家的兩倍，因此藥物間產生交互作用的機率也很高。然而目前醫療院所對於藥物交互作用的監測大著重於中重度的交互作用上，對於輕度或隱藏而不常發生的交互作用則無法顧及。近年來減肥瘦身藥、中藥與保健藥的盛行，更增加這類藥物交互作用監控與研究的困難，若能全面的重新檢視處方箋所開立的藥品，研究是否存有潛藏的藥物交互作用，將可提高用藥安全與醫療品質。本研究主要藉由資料探勘技術中的關聯法則分析，針對南部某醫學中心某月份的門診處方資料，進行探索性研究以瞭解：1.專科醫師對某些診斷是否存在某些固定的開藥行為模式2.這些行為模式的藥物間是否有存在潛在的交互作用3.這些行為模式所使用的藥物是否有不必要的用藥等議題。研究結果顯示，大部份醫師的開藥行為模式都是適當的，但是也有少數值得進一步研究的地方，例如兩類不同屬性的藥物存在關聯規則（血管擴張劑與軟便劑；支氣管擴張劑與胃藥等）。此外，透過藥物交互作用資料庫的查詢，開藥行為模式中的藥品間並沒有交互作用的記錄，此驗證了該醫院對交互作用的品管系統的適當性。但是，有些藥品卻不在交互作用資料庫的藥品檔中，這代表了現有資料庫的資料有所不足之處。本分析結果可以指引進行進一步病歷研究的方向及做為醫生開藥時的參考。未來的研究方向包括：比較不同地區或不同層級的醫師對於同一診斷的開藥行為是否有差異；對某特定醫院做縱軸分析，來了解醫師開藥行為隨時間變化的趨勢。而如果希望了解對於老人或孩童的處方模式是否有所差異時，則也可以在科別、診斷的分類下再細分不同年齡層來進行用藥行為分析。

The types and severity of disease, patients' age and the supply of drugs are factors that physicians need to take into consideration while writing prescriptions. Physicians' personal experience and patients' individual demand are also factors influencing on physician prescribing behavior. Under the effects of present restriction of medical resources, the increase of new drugs and its substitutes and medical institutions' management strategies for the impact of reimbursement policy of the Bureau of National Health Insurance, the prescribing pattern of physicians is an issue worthy of being explored. Additionally, the average number of drugs prescribed in one prescription in our country is about 4.1 kinds of category of drugs which is twice the prescription of developing countries. It increases the possibility of drug-drug interactions. The majority of the present drug-drug interactions alert system of many hospitals was focus on the moderate and major drug interactions and ignored the minor or potentially hazardous interactions. The development of new drugs, anti-obese drugs, traditional Chinese medicine and health care supplements increase the difficulties of drug interaction monitoring and research. The more comprehensive evaluation of drug items in a prescription and the potential drug-drug interactions will increase patients' drug safety and improve medical quality. The aim of this study is using the association rules to analyze data of outpatient prescription of a medical center in south Taiwan in order to understand 1) the particular prescription pattern of specialist physician with respect to a particular diagnosis. 2) The potential drug-drug interactions appeared in the above-mentioned prescription patterns. 3) Any unnecessary drugs used in the prescription patterns. The results of this study showed that the majority of physicians prescribing behavior were appropriate. However, a minority of them needed to have a further evaluation. Our preliminary results indicated that two different categories of drugs appeared to have an association rule, for example, vasodilator and laxative; bronchodilator and antacid. In addition, there are no drug-drug interactions in some of these drug pairs according to the search of drug-drug interaction database. Some drugs were not included in the database which indicated that the reference collected in the present database of drug-drug interaction was still insufficient. The results of this study can be used as a reference to do research of medical charts or to evaluate physicians' prescription behavior. The future purpose of this study includes the comparison of variations of prescriptions by different region and different level of hospitals; analyze the trend of variation of physicians' prescription pattern over time or analyze different prescription patterns for different age group under the classification of diagnosis and specialty.