Attempted Concealment of Doctor Shopping Through Multiple Pharmacy Dispensing
Kruti Doshi, Keiki Hinami, William TrickBackground: In the US, drug overdose deaths plateaued from 2018-2020, but remain above 60,000 annual deaths; the rate involving opioid analgesics remains slightly higher than heroin. Pharmacy shopping and overlapping prescriptions are associated with overdose. Recently, we analyzed discoordinated prescribing (i.e., prescription procurement from multiple providers) contributing to high-risk exposures. For this study, we evaluated pharmacy shopping among recipients of high-risk exposures for coordinated and discoordinated prescribing.
Methods: We used prescription claims data from CountyCare, providing managed-care services in Cook County, Illinois. We included community-dwelling adult members with chronic opioid analgesic use (i.e., members with ≥90 continuous days of opioids). We excluded members with malignancy (ICD10=Z85) or prescriptions from specialists in oncology or palliative medicine. We constructed two bipartite patient-pharmacy networks: 1. Discoordinated high-risk (≥120 Morphine Milligram Equivalents) exposures 2. No discoordinated high-risk exposure. The ties represent pharmacy dispensing weighted by number of distinct prescribers ordering prescriptions. We examined network centrality measures and to identify subgraphs within the networks, we examined K-cores. We geocoded member residences and pharmacies. We used Gephi to visualize networks and to identify whether geography contributed to network cores. We used Mann-Whitney-Wilcoxon unpaired two-sample test to compare the median number of pharmacies used by members in each network and the maximum distance travelled by each member to dispensing pharmacies. We used multivariable logistic regression to estimate the association between maximum distance and risk of discooordinated high-risk doses controlling for age and gender.
Results: We evaluated 2,064 patients with chronic opioid use, of which, 499 (24%) had discooordinated high-risk exposure dispensed by 467 pharmacies; 1565 (76%) obtained opioids from 648 pharmacies. Patients with discooordinated high-risk exposures utilized more pharmacies than those unexposed (weighted degree centrality 5.97 vs 2.96, p<.001). Among patients with ≥2 prescribers, 48% of members with discooordinated high-risk exposure used ≥3 pharmacies compared to 28% of their counterparts. Network diagrams further illustrate evidence of multi-pharmacy seeking behaviors among the discoordinated network; member nodes were surrounded by multiple ties to pharmacies. We observed that core connectedness in the subgraphs was a function of geography with more connected patients living closer to the city. Patients with discooordinated high-risk exposures travelled greater distance pharmacies compared to their counterparts (average maximum distance in km: 10.83 vs 8.79, p <0.001). Each km of maximum distance was associated with greater odds of being exposed to discooordinated high dose (aOR=1.01; 95% CI, 1.00–1.01; p<0.05).
Conclusion: Patients exposed to high-risk opioid dosage from discoordinated prescribing had significantly higher pharmacy degree centrality and traveled farther to get their prescriptions, likely to mask doctor-shopping activities. Encouraging pharmacies to use Prescription Monitoring Programs could help pharmacists flag multiple overlapping prescriptions, reducing the risk of exposing patients to high-dose opioids.