Does Screening for Opioid Use Disorder in Primary Care Increase the Percentage of Patients with a New Diagnosis?

Background:

Integrating care for common mental health disorders into primary care through screening and treatment has proven to be highly effective and is now a widespread practice. Primary care may also be an ideal setting to offer treatment for opioid use disorder (OUD), as primary care providers make up the bulk of buprenorphine prescribers.(1) However, substance use disorders often go unrecognized in primary care. Screening for OUD may be an effective approach to increase identification, much like it is for depression, anxiety and alcohol use disorder.(2, 3) The US Preventive Services Task Force recommends screening for substance use in primary care settings if effective treatments are available.

The objective of this research was to compare the percentage of primary care patients who are newly diagnosed with OUD before and after implementing universal screening for OUD. In contrast to psychometric studies designed to determine whether OUD screening instruments have sufficient sensitivity to potentially increase identification, our study was designed to determine whether real world screening actually increases identification. While participants in psychometric studies understand that their responses are confidential, patients in our study knew their responses would be shared with their providers and potentially impact their clinical care (e.g., discontinuation of pain medication prescriptions).

Methods:

We recruited 20 diverse primary care clinics to participate in a pre-post screening trial as part of a cluster randomized OUD treatment trial (NCT04600414). Human subjects protection oversight was provided by the Institutional Review Board ADVARRA. Screening was initiated during the acute phase of the COVID-19 pandemic. The University of Washington’s Advancing Integrated Mental Health Solutions Center provided screening practice facilitation (see Supplement 1). Screening was conducted using OUD questions from the National Institute on Drug Abuse - modified Alcohol, Smoking and Substance Involvement Screening Test (NM-ASSIST) and/or the two-item Short Opioid Screen (SOS) developed for the study because of difficulties administering and scoring the NM-ASSIST.(4)

Each clinic calculated the number and percentage of existing patients (i.e., those with encounters in past two years) with new OUD diagnoses in the six months prior to and after screening initiation. Although our protocol pre-specified a statistical test to determine whether screening increased the percentage of patients with new OUD diagnoses, this was changed to a descriptive analysis during the peer review process.

Results:

Table 1 depicts the type, size and payor mix of each healthcare system, along with each clinic’s location, screening practices, and the percentage of new OUD diagnoses in the pre- and post-periods. Among 167,710 existing unique patients with visits during the six-month post-screening period, 1,656 (0.99%) had OUD diagnoses, including 177 (0.11%) with new OUD diagnoses. The median percentage point change of patients with a new OUD diagnosis was +0.03% (range −0.08% to +0.38%; Figure 1). The median pre-post increase in the number of patients with a new OUD diagnosis was 1.5 patients per clinic (range −4 to 17).

Table 1.

Characteristics of Health Systems, Clinics, Screening Methods, and Percent of Patients with a NEW OUD Diagnosis Six Months Before and Six Months After Screening

Healthcare System	Type of Healthcare System	Unique Number of Patients in Healthcare System	Healthcare System Payor Mix1	Clinic	Unique Number of Patients in Clinic	Geographic Location and Rurality2 of Clinic	Screen(s) Used	Screening Frequency and Mode	Screening During Telehealth Visits	Pre % New OUD Diagnosis	Post % New OUD Diagnosis	Difference in Number of New OUD Diagnoses
System 1	Private non-profit	127,822	1% Uninsured 20% Medicaid 52% Medicare 27% Commercial	Clinic A	4,050	Northeast Urban Core	NM-ASSIST4 & SOS5	Annually, Paper	No	0.28%	0.67%	+17
System 1				Clinic B	5,798	Northeast Large Rural	NM-ASSIST4 & SOS5	Annually, Verbal	Yes	0.17%	0.33%	+10
System 2	Community Health Center3	99,244	10% Uninsured 59% Medicaid 11% Medicare 20% Commercial	Clinic C	4,689	West Urban Core	NM-ASSIST4 & SOS5	Every visit, Electronic	Yes	0.52%	0.55%	+3
System 2				Clinic D	3,728	West Suburban	NM-ASSIST4 & SOS5	Annually, Electronic	Yes	0.52%	0.24%	+4
System 3	Academic Medical Center	864,533	7% Uninsured 4% Medicaid 31% Medicare 58% Commercial	Clinic E	8,119	South Urban Core	NM-ASSIST4	Annually, Electronic	Yes	0.00%	0.01%	+1
System 3				Clinic F	10,154	South Urban Core	NM-ASSIST4	Annually, Electronic	Yes	0.00%	0.03%	+3
System 3				Clinic G	14,951	South Urban Core	NM-ASSIST4	Annually, Electronic	No	0.01%	0.01%	−1
System 3				Clinic H	14,948	South Urban Core	NM-ASSIST4	Annually, Electronic	Yes	0.00%	0.00%	0
System 4	Private non-profit	162,001	9% Uninsured 12% Medicaid 53% Medicare 27% Commercial	Clinic I	7,532	West Urban Core	NM-ASSIST4 & SOS5	Annually, Paper	No	0.07%	0.25%	+14
System 4				Clinic J	7,384	West Urban Core	NM-ASSIST4 & SOS5	Annually, Verbal & Paper	No	0.16%	0.08%	−4
System 4				Clinic K	7,116	West Urban Core	SOS5	Every visit, Paper	No	0.01%	0.03%	+1
System 4				Clinic L	6,684	West Urban Core	SOS5	Annually, Paper	No	0.02%	0.53%	−1
System 5	Private non-profit	64,595	12% Uninsured 15% Medicaid 17% Medicare 67% Commercial	Clinic M	3,255	Midwest Urban Core	NM-ASSIST4 &SOS5	Annually, Paper	No	0.03%	0.06%	0
System 5				Clinic N	2,992	Midwest Large Rural	NM-ASSIST4	Annually, Paper	No	0.04%	0.00%	−1
System 5				Clinic O	1,089	Midwest Suburban	NM-ASSIST4	Annually, Verbal & Paper	Yes	0.00%	0.18%	+2
System 5				Clinic P	3,534	Midwest Urban Core	NM-ASSIST4 & SOS5	Annually, Paper	No	0.00%	0.00%	+1
System 6	Community Health Center3	48,348	46% Uninsured 34% Medicaid 3% Medicare 17% Commercial	Clinic Q	24,705	Midwest Urban Core	NM-ASSIST4 & SOS5	Annually, Paper	No	0.03%	0.02%	−2
System 6				Clinic R	8,218	Midwest Urban Core	NM-ASSIST4 & SOS5	Every visit, Paper	Yes	0.03%	0.11%	+5
System 7	Private non-profit	552,672	6% Uninsured 27% Medicaid 31% Medicare 45% Commercial	Clinic S	9,724	West Urban Core	NM-ASSIST4 & SOS5	Every visit, Paper	No	0.22%	0.31%	+8
System 7				Clinic T	19,040	West Urban Core	NM-ASSIST4 & SOS5	Annually, Electronic	Yes	0.28%	0.08%	+7

^1.Payor mix was calculated as % of patients for health systems 1, 2, 5, 6, and 7 and as % of revenue for health systems 3 and 4.

^2.Rural Urban Commuting Area Codes, Classification Scheme 1

^3.Federally Qualified Health Center

^4.National Institute on Drug Abuse (NIDA)- modified Alcohol, Smoking and Substance Involvement Screening Test

^5.Short Opioid Screen (answering yes to either question below is considered a positive screen)

^a)In the past three months, have you used opioid medications prescribed for you by your healthcare provider at higher dosages or more often than prescribed? (Includes: oxycodone [Oxycontin, Percocet], hydrocodone [Vicodin], methadone, buprenorphine, tramadol, fentanyl etc.)

^b)In the past three months, have you used any opioids not prescribed for you? (Includes: oxycontin, fentanyl, heroin, kratom etc.)?

Figure 1. Histogram of pre-post percentage point changes in patients with a new OUD¹ diagnosis².

1. OUD – Opioid Use Disorder

2. A negative number indicates a pre-post decrease in the percentage of patient with a new diagnosis and a positive number indicates a pre-post increase.

Discussion:

The percentage of patients with a new OUD diagnosis did not increase in a clinically meaningful way after OUD screening in routine care. Although screening coincided with the COVID-19 pandemic, post-pandemic screening was, anecdotally, no more effective. The small change in the percentage of new OUD diagnoses after implementing universal screening observed in this study is different from the documented effectiveness of screening for depression, anxiety and alcohol.(2, 3) This may result from multiple factors, including: 1) lower true prevalence of OUD, 2) greater stigma, resulting in lower screener sensitivity than observed in psychometric studies, and/or 3) greater delays in follow-up diagnostic assessments. Stigma may lower the sensitivity of OUD screening because patients’ comfort disclosing non-medical opioid use requires trust that they will not face discrimination. Poor sensitivity was reported in a study using the NM-ASSIST in an obstetrics clinic which found that, compared to gold standard biological testing, sensitivity was 25% for non-medical use of prescription opioids and 12.5% for non-prescription opioids.(5) While OUD screening is recommend in primary care if effective treatments are available, the resulting number of patients with new OUD diagnoses will depend on a clinic’s underlying prevalence of OUD and level of stigma. Addressing stigma may increase the sensitivity of the screening instrument. To address OUD in their communities clinics may also want to conduct outreach activities and publicize their commitment to accepting new patients seeking care for OUD.