Back

Early detection of patients with narcotic use disorder using a modified MEDD score based on the analysis of real-world prescription patterns

Kim, Y.-J.; Lee, K. H.; Kang, H.-R.; Cho, Y. S.; Kang, D. Y.; Kim, J. H.; Kim, Y. S.

2022-04-16 addiction medicine
10.1101/2022.04.12.22273679 medRxiv
Show abstract

BackgroundAddiction to prescription narcotics is a global issue, and detection of individuals with a narcotic use disorder (NUD) at an early stage can help prevent narcotics misuse and abuse. We developed a novel index to detect early NUD based on a real-world prescription pattern analysis in a large hospital. MethodsWe analyzed narcotic prescriptions of 221 887 patients prescribed by 8737 doctors from July 2000 to June 2018. For the early detection of patients who could potentially progress to developing NUD after a long history of narcotic prescription, a weighted morphine equivalent daily dose (wt-MEDD) score was developed based on the number of prescription dates on which the actual MEDD was higher than the intended MEDD. Performance of the wt-MEDD scoring system in detecting patients diagnosed with NUD by doctors was compared with that of other NUD high risk indexes such as the MEDD scoring system, number of days on prescribed narcotics, the frequency/duration of prescription, narcotics prescription across multiple doctors, and the number of early refills of narcotics. ResultsA wt-MEDD score cut-off value of 10.5 could detect all outliers, as well as patients diagnosed with NUD with 100% sensitivity and 99.6% specificity. The wt-MEDD score showed the highest sensitivity and specificity in identifying NUD among all indexes. Further, combining the wt-MEDD score with other NUD high risk indexes improved the prediction performance. ConclusionWe developed a novel index to distinguish patients with vulnerable use patterns of narcotics. The wt-MEDD score showed excellent performance in detecting early NUD.

Matching journals

The top 8 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.