A patient returns at week four. Their NPRS has dropped from 7 to 4, their Roland-Morris score has improved by six points, and you record a good response to the loading programme you prescribed.

Then you find out, months later and by accident, that they started taking 800 mg of ibuprofen before every session in week two.

The clinical data you have been collecting is not wrong, exactly. It is just measuring something other than what you think it is measuring. This piece looks at how large that gap tends to be, why patients specifically underreport to physiotherapists, and what changes when you alter the question.

The Baseline Rate of Non-Disclosure Is Higher Than Most Clinicians Assume

Levy and colleagues surveyed 4,510 US adults across two national samples and found that between 60% and 80% had withheld medically relevant information from a clinician at some point (Levy AG, et al., JAMA Network Open, 2018;1(7):e185293). The reasons clustered predictably: not wanting to be judged, not wanting to hear a lecture about a behaviour, embarrassment, and not wanting to be seen as a difficult patient.

For self-administered products the picture is worse. A meta-analysis of biologically-based complementary medicine found a pooled disclosure rate of 33% (95% CI 24–43%), with individual study rates ranging from 7% to 80% (Foley H, et al., Scientific Reports, 2019). In oncology populations, non-disclosure of complementary medicine use ran between 20% and 77% (Davis EL, et al., The Oncologist, 2012).

The leading reason across all of this literature is not defiance. It is that nobody asked.

Musculoskeletal populations are heavy users of exactly these products. The Arthritis Foundation’s 2019 survey of more than 2,600 patients found 29% currently using CBD for arthritis symptoms, 63% of those daily, and 79% either using, having used, or considering it. Patients also reach for THC products, and the newer fast-acting THC gummies built on nanoemulsion delivery are marketed on a compressed onset window – which, as the next section argues, is precisely what makes their absence from your notes a measurement problem rather than a trivia gap.

Quick tip: Ask about topicals separately. Patients frequently classify a rub-on preparation as “not really taking anything,” and topical diclofenac reaches clinically meaningful effect sizes in knee osteoarthritis (da Costa BR, et al., BMJ, 2021).

Undisclosed Co-Intervention Is Large Enough to Swallow Your MCID

This is the part that should concern anyone reporting outcomes.

The minimal clinically important difference for the NPRS in low back pain sits around 2 points. Copay’s widely cited estimate for the Oswestry Disability Index is 12.8. For the Roland-Morris, a within-patient change of 4 to 5 points, or a 30% reduction from baseline, is the usual threshold (Jordan K, et al., Journal of Clinical Epidemiology, 2006).

Now set that against analgesic effect. In a network meta-analysis of 192 trials and 102,829 participants, several oral NSAID preparations exceeded the prespecified minimal clinically relevant effect size of −0.37 with at least 99% probability (da Costa BR, et al., BMJ, 2021).

An unreported analgesic can therefore generate a change comparable to your entire treatment effect. When a patient crosses an MCID threshold, you have no way of attributing that movement between your intervention, their pharmacy, and natural history – unless you asked.

Timing compounds it. A patient who takes something 45 minutes before a session arrives with altered pain-provocation thresholds and altered end-range tolerance. Their measured range of motion is real; it just is not their untreated range of motion. Repeat that across a six-week block and the trajectory you plot is partly a dosing schedule.

Why Onset Time Changes What Patients Can Accurately Report

Oral cannabinoids have poor and variable bioavailability, roughly 4–12% for THC, with Tmax typically 1 to 2 hours. That delay drives redosing: the patient feels nothing at 40 minutes, takes more, then reports “one” when asked how much they took.

Nanoemulsion and self-emulsifying formulations were developed to shorten that window. In a crossover study of 14 healthy volunteers, a self-nanoemulsifying powder produced a Tmax for 11-hydroxy-THC of 0.86 hours versus 4.54 hours for oil-based drops, with more than double the relative bioavailability (Journal of Cannabis Research, 2025).

Treat that as directional rather than settled – fourteen participants, one formulation, industry-adjacent context, and a rat study of delta-8 nanoemulsions actually found lower bioavailability than an MCT oil solution (Pharmaceutics-indexed work, 2023). What matters clinically is the variance itself. Products with different onset profiles sit in the same drawer at home, and patients cannot reliably reconstruct dose or timing across them.

Warning worth flagging: CBD inhibits CYP2C19 and CYP2C9, THC is metabolised largely via CYP3A4 and CYP2C9. In a rheumatology caseload carrying warfarin, methotrexate and polypharmacy, that belongs in the notes even though you are not the prescriber.

The Clinical Stakes Beyond Measurement

Undisclosed NSAID use is not only a data problem.

A scoping review of 44 studies found that in vitro work consistently shows NSAIDs impairing tenocyte proliferation and collagen synthesis, while most human and animal studies of non-selective COX inhibitors showed no negative effect on soft-tissue healing – though selective COX-2 inhibitors did appear to compromise healing after surgical repair (Constantinescu DS, et al., Arthroscopy, 2019).

Muscle adaptation evidence is similarly split. Lilja and colleagues reported that high-dose anti-inflammatory use compromised strength and hypertrophic adaptation in young adults (Acta Physiologica, 2018), while other trials found no impairment at over-the-counter doses.

You cannot weigh any of that for a specific patient whose analgesic use you do not know about.

Changing the Question Changes the Answer

The disclosure literature converges on one finding: clinician inquiry is the single strongest predictor of patient disclosure. Which raises an uncomfortable point – if a patient never mentioned their supplements, whose omission was it?

Three changes that cost nothing:

Ask about products, not categories. “What do you take on a bad day?” retrieves more than “are you on any medications?”, because patients do not classify a gummy, a turmeric capsule or a topical rub as medication.

Normalise before asking. Prefacing with “most people manage flares with something at home” measurably reduces anticipated judgement, which is the dominant stated reason for withholding.

Ask about the appointment itself. “Did you take anything before coming in today?” is the highest-yield question in the set and almost nobody asks it.

Then record it as a co-intervention rather than a footnote. The adherence literature already tells us self-report is unreliable in this population – pooled home-exercise adherence estimates range from 21% to 67% depending on the review, largely because measurement is inconsistent (Peek K, et al., Physiotherapy, 2016; Argent R, et al., JMIR mHealth and uHealth, 2018). Non-disclosure is the same measurement failure pointed in the opposite direction.

Your outcome data is only as good as your intake history. Most of us are measuring a combined intervention and reporting it as a single one.

Related posts:

How Advanced Nursing Roles Are Improving Patient Care Outcomes How Healthcare Is Evolving to Meet Modern Patient Needs 5 Best 9mg Nicotine Pouches Online: The Complete Buyer’s Guide How Motion-Tracking Gadgets Improve Physical Therapy Accuracy Buy BPC-157 Peptide: Dosage, Side Effects, and How To Get BPC-157 Products in 2026 Best Peptides for Muscle Growth: A Research-Based Look at the Top Options

Stay updated, free articles. Join our Telegram channel

Join