並列タイトル等トウヤク ミス カイヒ ノ タメ ノ イヤクヒン トリチガエ リスク ヒョウカ システム ノ コウチク
Tōyaku misu kaihi no tame no iyakuhin torichigae risuku hyōka shisutemu no kōchiku
Development of a risk assessment system to avoid drug-related errors
一般注記type:text
薬物治療の高度化の一方で、投薬ミスによる被害は大きな社会問題となっている。投薬ミスの最大の原因は医薬品の取り違えである。医療機関、製薬企業、規制当局等により医薬品の取り違え防止対策は実施されているが、医薬品名の類似性の評価は不十分である。我々はこれまで文字列断片パターンの類似を数値化した薬名類似度指標m2-vwhtfrag(modified version 2-visually weighted head and tail-weighted fragmentary pattern based measure)を構築し、既存の指標や特徴量よりも主観的類似度を良好に反映していることを示してきた。しかしながら、m2-vwhtfragを用いた指標構築の過程において、薬名以外の医薬品類似性(規格含量や薬効等)の影響は十分考慮されていなかった。そこで本研究では、m2-vwhtfragの閾値(薬名類似による医薬品の取り違えを起こしやすい医薬品を判別するための数値)を、規格含量や薬効の類似性を考慮して改良することを目的とした。収集した医薬品取り違えが起こった医薬品の組合せ(正例)と無作為に作成した医薬品の組合せ(負例 10セット)のm2-vwhtfragを算出し、両者を判別できる指標値の閾値を求めた。取り違えが起こった医薬品の全組み合わせ(395組)、規格及び薬効類似を除いた組み合わせ(120組)をそれぞれ正例として閾値を算出した結果、それぞれ0.45, 0.45 であり、薬名以外の医薬品類似性が閾値に及ぼす影響は見かけ上認められなかった。以上から、構築した指標は薬名類似が原因で起こる取り違えを判別する指標として適切であると考えられた。今後、医療施設において集積された医薬品の取り違え事例の要因分析に本手法を導入することで、各施設の医薬品取り違えリスク傾向を明らかにし、その対策立案につなげることが可能になると期待される。
While the prevalence of advanced medical treatment, damage caused by medication errors is serious social issues. Confusion of drug names is one of the most common causes of drug-related errors. Medical institutions, pharmaceutical companies and regulatory agencies have made efforts to prevent drug-related errors, but fundamental measures to prevent them are still inadequate.
In order to evaluate a similarity of drug name, we have developed the fragmentary pattern-based measure, which is a measure of similarity for character strings based on the string resemblance system, to develop novel measures of drug name similarity. The measure we first developed was visually weighted head and tail-weighted fragmentary pattern based measure (vwhtfrag). Then, we modified vwhtfrag in consideration of the difference in character length of name pairs, which was named m2-vwhtfrag. M2-vwhtfrag showed a higher correlation to subjective similarity than did vwhtfrag and other conventional similarity measures (htco, cos1, edit). However, the influence of drug similarity factors (ingredient contents, indication etc.) other than drug name was not sufficiently considered in the measure using m2-vwhtfrag. Therefore, the purpose of the present study was to improve the method to discriminate drugs that are likely to be induced drug-related errors in considering not only drug name but also other drug similarity factors. The various pairs of drug names for which confusion errors have been reported were used as a positive control. In contrast, ten sets of random pairs of drug names were generated as negative controls. By using all the pairs (395 pairs) and the pairs excluding drugs with similarities of ingredient contents and indication as positive controls, both thresholds were calculated as 0.45, showing that the influence of similarities could not been seen apparently. As a result, the controls and the index currently in use were considered to be appropriate as indicators for discriminating confusion errors caused by drug name similarity. By introducing this method to risk factor analysis of drug-related errors collected in a medical institution, it would be possible to evaluate the risk tendency of drug-related errors and to formulate appropriate measures for each institution.
一次資料へのリンクURLhttps://koara.lib.keio.ac.jp/xoonips/modules/xoonips/download.php?koara_id=2018000005-20180270
連携機関・データベース国立情報学研究所 : 学術機関リポジトリデータベース(IRDB)(機関リポジトリ)