Covered, With Conditions: Five State Medicaid Programmes and the Behavioural Side of Pain Care
- An economic evaluation published in JAMA Network Open on 13 August 2025 reviewed Medicaid coverage policy for co-occurring chronic pain and opioid use disorder in five states – California, Illinois, Texas, North Carolina and New York – across policy documents dated 1 January 2018 to 31 December 2023, covering roughly 35.9 million beneficiaries at peak 2023 enrolment.
- Core medication classes and physical therapy were covered in all five states (100%); cognitive behavioural therapy in four of the five (80%); chiropractic services in four of the five (80%); acupuncture in only two of the five (40%).
- Where a service was covered, the terms diverged. Texas capped cognitive behavioural therapy at 30 visits per year and peer support at 104 units per six months, about 26 hours. Illinois limited peer support to substance use disorder treatment support. New York specified 30-, 45- and 60-minute sessions with a maximum of eight people per group.
- The authors conclude that these variations "may perpetuate health disparities based on geography rather than clinical need" and that "a patient's access to evidence-based integrated care should not depend on their state of residence." For scale: 24.3% of US adults reported chronic pain in 2023 and 8.5% reported chronic pain that frequently limited life or work activities (NCHS Data Brief No. 518, National Health Interview Survey).
A colleague in Chicago and a colleague in Houston can hold the same clinical opinion about the same patient and reach different answers about what is covered for that patient. A five-state review of Medicaid coverage policy, published in JAMA Network Open in August 2025, puts numbers on that gap. Its finding is not that behavioural pain care is uncovered – mostly it is covered – but that the fine print does most of the work.
What the five-state review actually documents
The design was deliberately unglamorous: a systematic reading of state Medicaid websites, fee schedules, preferred drug lists, clinician manuals and archived policy documents from California, Illinois, Texas, North Carolina and New York, for the population with co-occurring chronic pain and opioid use disorder. At peak 2023 enrolment those five programmes covered about 35.9 million people.
The headline pattern is a hierarchy. Medications and physical therapy sat at the top: universal coverage, all five states. Cognitive behavioural therapy and chiropractic care sat one tier down at 80%. Acupuncture sat at 40% – covered in California under specific CPT codes, and in Illinois in 2023 only. Peer support was nominally universal, covered in all five states, but implemented so differently that the word "covered" carries little information on its own: Illinois tied it to substance use disorder treatment support, while Texas allowed 104 units per six-month period, roughly 26 hours of a service that is supposed to run alongside long-term care. Texas capped cognitive behavioural therapy at 30 visits per year; New York wrote session length and an eight-person group ceiling into the policy itself. Authorisation requirements varied across states for medications and interventional procedures alike, and the authors note that differing clinician credential requirements, supervision specifications and setting restrictions "may complicate coordinated care delivery."
The federal layer makes the contrast sharper rather than resolving it. In the CY 2023 Physician Fee Schedule final rule, CMS finalised HCPCS codes G3002 and G3003 for monthly chronic pain management and treatment services, and wrote into the code descriptor "facilitation and coordination of any necessary behavioral health treatment" alongside coordination with physical and occupational therapy and integrative approaches. Medicare, in other words, pays a clinician to organise multidisciplinary pain care. Which of those services are covered, for a Medicaid patient, is set one level down and varies.
What this changes at the point of referral
The practical consequence is that a referral is a coverage question before it is a clinical one, and the numbers that matter are not coverage yes or no but the caps. Thirty sessions a year is workable for a course of pain-focused cognitive behavioural therapy; it is a different proposition if the same allowance has to absorb depression, insomnia and a flare-up in the same twelve months. Peer support measured in units rather than in months invites a plan that runs out mid-course. If you work in the United States and see Medicaid patients with chronic pain, the useful habit is to read your own state's manual for the specific limit before you promise a treatment length, and to build the taper of contact into the plan from the start rather than discovering the ceiling at session 28.
The second consequence is epistemic. When two of your patients respond differently to the same referral, the variable may be the state, not the patient. That is worth naming in supervision and in the record, because it is the kind of difference that otherwise gets attributed to motivation.
For a patient with chronic pain, the question is rarely whether therapy is covered at all – it is how many sessions the state manual allows, and that is where the five states diverge.
The review documents written policy, not delivery: it does not measure clinician network adequacy, reimbursement rates, actual patient access or outcomes, and the authors say so explicitly. It covers five states through 31 December 2023, so current policy in these or other states may differ, and the 2018-2023 window includes COVID-19 emergency provisions and Medicaid continuous enrolment.