Opioid Use Disorder

Opioid use disorder is a chronic medical disease, and the medications that treat it are among the most effective interventions in medicine. Treatment with buprenorphine or methadone substantially reduces all-cause and overdose mortality, with cohort data showing mortality roughly halved or better while patients are retained in treatment. Framing this as a moral problem, or as something to be addressed after the "real" medical issue, costs lives in a way that few other omissions on a medicine service do.

Counseling alone is not equivalent to medication. Referral to a program without medication, or a "detox" admission with no medication at discharge, is not neutral: it is associated with worse outcomes than medication treatment. This is the single most common well-intentioned error in this disease.

• Patients with OUD are admitted often -for endocarditis, skin and soft tissue infection, osteomyelitis, overdose, trauma- and each admission is a point of contact with a system they may otherwise avoid.
• Starting medication in hospital increases the chance the patient is still in treatment weeks later, compared with referral alone at discharge. The initiation is the intervention; the referral by itself frequently fails.
• Untreated withdrawal drives patients out of the building. Discharge against medical advice is common in this population, and undertreated withdrawal is a leading and modifiable reason. Treating withdrawal properly is not a comfort measure, it is what keeps the patient present long enough to finish antibiotics for their endocarditis.

• Say "person with opioid use disorder", not "abuser" or "addict". Stigmatizing terms in the chart measurably change how subsequent clinicians judge and treat the same patient.
• Toxicology results are "positive" or "negative", never "clean" or "dirty". The alternative frames a laboratory value as a statement about the person's worth.
• "Medication for opioid use disorder" (MOUD), not "replacement" or "substitution" -the latter implies one drug is being swapped for another, which misrepresents what a stable, non-euphoric, life-preserving therapy is doing.

• OUD is a DSM-5 clinical diagnosis based on a pattern of problematic use causing impairment or distress, graded by how many criteria are met: 2 to 3 mild, 4 to 5 moderate, 6 or more severe. The criteria cover loss of control, craving, failure to meet obligations, use despite harm, and continued use in hazardous situations.
• ⚠ Tolerance and physical withdrawal do not count toward the diagnosis when opioids are taken as prescribed under medical supervision. This distinction matters: a patient on long-term opioid therapy for cancer pain is physically dependent without having a use disorder, and conflating the two both stigmatizes them and misdirects treatment.
• Screen rather than wait for the obvious. A single question about past-year use of illicit drugs or non-prescribed medications performs well as an opening, and normalizing it for everyone avoids the profiling that determines who gets asked.

• The Clinical Opiate Withdrawal Scale (COWS) scores eleven items -resting pulse, sweating, restlessness, pupil size, bone or joint aches, runny nose or tearing, gastrointestinal upset, tremor, yawning, anxiety or irritability, and gooseflesh- to grade severity objectively.
• Its main use here is timing. Traditional buprenorphine initiation requires the patient to be in at least moderate withdrawal, conventionally a COWS of about 8 to 12 or higher, before the first dose, because giving buprenorphine too early is what precipitates withdrawal.
• Score objective signs, and be aware of what mimics them. Tachycardia, sweating and agitation are not specific: sepsis, alcohol withdrawal, thyrotoxicosis and pain all raise the score. In a patient with endocarditis, deciding that a rising COWS is withdrawal rather than worsening infection is a clinical judgment, not a calculation.

• The mechanism: buprenorphine's affinity for the mu receptor exceeds that of full agonists, so it displaces them. Because it is only a partial agonist, replacing a full agonist with it produces an abrupt net drop in receptor activation and therefore sudden, severe withdrawal. It is iatrogenic, deeply unpleasant, and a powerful reason for a patient to refuse treatment ever again.
• The classic prevention is timing: wait until the patient is already in moderate withdrawal (COWS about 8 to 12) so that receptor occupancy by the full agonist has fallen, then give the first dose.
• ⚠ Why that fails with fentanyl: fentanyl is highly lipophilic, so with repeated use it accumulates in peripheral fat and redistributes slowly. Patients can therefore have meaningful fentanyl still coming out of tissue even after 8 to 24 hours or more of abstinence and a convincing COWS score. Waiting for withdrawal, the rule that worked reliably in the heroin era, no longer guarantees safety.
• Low-dose initiation (also called micro-dosing or rapid overlap) gives very small, escalating buprenorphine doses while the full agonist continues, building up over several days. Because each increment is small, it does not produce the abrupt displacement that causes precipitated withdrawal, and the patient does not have to be in withdrawal to begin -which removes the requirement people find hardest to meet. Protocols reach therapeutic dosing over roughly three to four days.
• High-dose initiation takes the opposite approach, giving substantially larger initial doses to move quickly through the vulnerable window. Emergency department studies in fentanyl-positive patients reported precipitated withdrawal in only about 1 to 5 percent, and it is emerging as a safe and efficient option.
• Both are active practice, not fringe technique. The right move on a general medicine service is to involve addiction medicine early where available and to choose a protocol deliberately, rather than defaulting to a heroin-era induction and being surprised.

• A patient already on methadone should have it continued during admission -stopping it causes withdrawal and jeopardizes the admission.
• ⚠ Verify the dose with the dispensing program directly before giving it, and confirm when the last dose was taken. Do not rely on the patient's report alone: methadone accumulates, and an over-reported dose administered to someone whose tolerance has lapsed can be fatal.
• The "72-hour rule" lets a clinician who is not part of an opioid treatment program administer methadone for up to three days to relieve withdrawal while arranging linkage to a program. The medication is administered, generally one day at a time, rather than prescribed. It is a bridge, and its value depends entirely on the linkage actually being made before it expires.
• For a patient admitted for another condition, methadone can be provided as part of that hospital care; the opioid treatment program requirement governs ongoing outpatient treatment, not inpatient administration.

• Naloxone for every patient, and teach someone who lives with them to use it -a person in overdose cannot administer it themselves, so the bystander is the actual target of the training.
• Counsel explicitly on lost tolerance after any period of reduced use, and that using alone is what turns an overdose into a death. Advise against mixing with alcohol or benzodiazepines.
• Fentanyl and xylazine now contaminate the supply unpredictably, so a dose that was tolerated before may not be. Drug checking supplies, where legally available, and syringe service programs reduce infectious complications and are a route back into care.
• Meet the patient where they are. Someone not ready to stop still benefits from naloxone, vaccination, HCV treatment and a door left open. Making future care conditional on abstinence today is how people disappear from care entirely.

• Why does buprenorphine precipitate withdrawal? Its receptor affinity exceeds that of full agonists so it displaces them, but it is only a partial agonist, so net receptor activation drops abruptly and withdrawal follows.
• Why is buprenorphine safer in overdose than methadone? Partial agonism gives a ceiling effect on respiratory depression. Methadone is a full agonist with no ceiling, plus a long variable half-life that causes accumulation and QT prolongation.
• Do you still need an X-waiver? No. The DATA-waiver was eliminated by the Consolidated Appropriations Act of 2023. Any DEA registration with Schedule III authority suffices, with no patient cap.
• Why has fentanyl made traditional induction unreliable? It is highly lipophilic, accumulating in fat and redistributing slowly, so meaningful drug persists well beyond 8 to 24 hours of abstinence despite a convincing COWS score.
• What are the two newer induction strategies? Low-dose initiation, giving small escalating doses while the full agonist continues so there is no abrupt displacement; and high-dose initiation, moving rapidly through the vulnerable window, with reported precipitated withdrawal of only about 1 to 5 percent in fentanyl-positive ED patients.
• Why is extended-release naltrexone hard to start? It needs roughly 7 to 10 opioid-free days first, so the patient must complete withdrawal unmedicated, and many relapse before the first injection.
• Your patient on maintenance buprenorphine needs surgery. Stop it? No. Continue the maintenance dose and treat pain on top, expecting that higher full-agonist doses may be needed. Stopping it adds withdrawal to pain and destabilizes treatment.
• Why is a withdrawal-only admission actively dangerous? Tolerance falls faster than the drive to use returns, so overdose risk after discharge is higher than before admission. Withdrawal management without ongoing medication is not an acceptable endpoint.
• Can you give methadone without being an opioid treatment program? Yes, under the 72-hour rule, administered generally one day at a time for up to three days while arranging program linkage, and for a patient admitted with another condition as part of hospital care.

"Ms. T is a 34-year-old woman who injects fentanyl daily, admitted with tricuspid valve endocarditis and Staphylococcus aureus bacteremia, now day two of a planned six-week antibiotic course. Overnight her COWS rose to 14 with objective sweating, mydriasis and gooseflesh, and she has asked twice about leaving. I think the withdrawal is the immediate threat to her antibiotic course, because undertreated withdrawal is a leading modifiable reason patients in her situation leave against advice, and leaving on day two of endocarditis treatment is a far greater risk to her than her drug use is this week. Given daily fentanyl I am not comfortable with a traditional induction despite the COWS of 14, since fentanyl redistributes from fat and she could still precipitate. I have asked addiction medicine to help choose between a low-dose overlap and a high-dose initiation, and I have started symptomatic management meanwhile. She has not been screened for HIV or hepatitis C, so I have sent both and checked hepatitis B immunity for vaccination. For discharge she will need naloxone with training for her partner, counseling on lost tolerance, and a confirmed appointment rather than a phone number, since referral without a booked follow-up usually fails. I would also flag in the chart that her buprenorphine must not be held if she needs a procedure."

• Is withdrawal actually controlled? A rising COWS is a reason to act today, not to observe -it predicts the patient leaving before treatment is complete.
• Has medication been offered this admission, and if declined, was it re-offered? Willingness changes across an admission, and a single refusal on day one is not a standing answer.
• If already on MOUD, is the dose verified with the program and being given? Check it has not been quietly held for a procedure.
• Pain plan compatible with the maintenance medication, and documented so the overnight team does not stop it.
• Infectious screening sent (HIV, HCV, HBV) and vaccination ordered.
• Discharge: naloxone in hand with a trained bystander, a booked follow-up rather than a referral number, enough medication to bridge the gap, and lost-tolerance counseling delivered.

• The medication is the treatment. Detox without it, or referral without it, is worse than not admitting the problem at all, because it removes tolerance without removing risk.
• Fentanyl broke the induction rule you were taught. A reassuring COWS no longer guarantees a safe traditional start.
• Never hold maintenance therapy for pain. Treat on top of it.