Table 1.

PICOS for all articles analyzed.

Authors	Participants	Intervention	Results (Compared to Control Group)	Medical Outcomes Reported	Study Design
Pett RG et al. [26]	Pharmacists	PDMP	Frequent PDMP users were more likely to recommend naloxone.	Naloxone prescribed to shoppers	Explanatory, sequential 2-phase mixed-methods
Durand et al. [27]	Injured workers identified in worker’s compensation records	PDMP and worker’s compensation claims	No control group. Injured workers have a high prevalence of opioid use after injury, but prescribing patterns generally follow state guidelines.	Not reported	Retrospective cohort
Freeman et al. [28]	Pharmacists and providers	Interviews about PDMP	No control group. Both PCPs and pharmacists reported PDMPs are key tools to aid prescribing and dispensing.	Not reported	Qualitative
Nagarajan and Talbert [29]	Prescriber-prescriber networks	Computer model	Outliers were clearly identified in the model, which can help identify those prescribers contributing to the opioid epidemic.	Not reported	Computer model to detect prescriber outliers
Perry et al. [30]	Patients who submitted claims to commercial database in the Appalachian region of the U.S., 58.7% female	Computer model PageRank	Model clearly differentiates aberrant behavior identifying drug shoppers for both opioids and morphine milligram equivalents (MME).	Not reported	Computer model to detect prescriber outliers
Soffin et al. [31]	n/a	PDMP	n/a	Accessing real-time information about patients’ prescription opioid status using PDMP reduces opioid quantities prescribed	Opinion
Stopka et al. [32]	adults in Massachusetts prescribed opioids	Computer model	Hotspots were identified.	Not reported	Spatial epidemiological study
Wang et al. [33]	Patients in pharmacological database in Florida	Pharmacological database	Recipients of opioids, benzodiazepines, and carisoprodol in 2017 compared with 2012 were younger, more likely to be female, and geographically-localized.	Not reported	Retrospective, observational
Butler et al. [34]	n/a	PDMP, Lock-in programs	No control group. The lock-in program decreased doctor shopping.	Broader range of drug schedules and an increases frequency of updating the PDMP results in lower opioid-related mortality	Opinion
Lin et al. [35]	Patients from health insurance claim data	Medication record sharing program	The medication duplication rate was reduced 7.76 percentile, average medication overlap periods shortened 4.36 days.	Not reported	Retrospective pre-post test design
Ponte et al. [36]	Beneficiaries of the France public health system	SNIIRAM database	The strong opioid analgesics have the highest DSI (2.79%) versus 2.06% for BZD hypnotics. Flunitrazepam ranked first according to its DSI (13.2%), followed by morphine (4%), and zolpidem (2.2%).	Not reported	Retrospective, observational
Torrance et al. [37]	Beneficiaries of the Scotland National Health System (NHS)	NHS and Generation Scotland databases combined to identify trends	The number of strong opioid prescriptions more than doubled between 2003–2012. Patients in the most deprived areas were more likely to receive a strong opioid.	Not reported	Descriptive analysis
Ali et al. [38]	Respondents to the national survey of drug use and health, aged 12 and older, 48% male	Survey instrument to assess effectiveness of the PDMP	No control group. PDMP was effective at controlling for doctor shopping for opiate pain killers.	10–20 fewer days of NMPR use	Qualitative
Rutkow et al. [39]	Prescribers in four states	PDMP	Prescribers need to work with law enforcement, law-enforcement need to share data with each other, data sharing between states needs to occur.	Not reported	Qualitative
Simeone [40]	Prescriptions	PDMP, education efforts, pharmacy panels that span the country	No control group. The number of prescriptions diverted fell from 4.3 million in 2008 to 3.37 million in 2012.	Decline in overdose mortality	Retrospective, observational
Chenaf et al. [17]	Adult patients with chronic non-cancer pain (CNCP)	French national health system	No control group. Shopping very low in drugs for these conditions.	Not reported	Retrospective cohort
Delorme et al. [42]	Patients treated by opioid substitution treatment over 8 years	French national health system	No control group. Shopping behavior was only found in high dosage buprenorphine patients, but still very low.	Not reported	Retrospective cohort
Kea et al. [43]	Emergency Department (ED) discharges	PDMP	Doctor shopping was not detected in ED survey.	Not reported	Qualitative