Editorial Article
The Largest Trial Found Nothing. The Market Grew Anyway.
In April 2025 a systematic review in the Journal of Pharmacy Practice and Research pooled twenty-two randomised controlled trials and concluded that dose administration aids “can considerably improve medication adherence.” It was widely cited, and it is the kind of study an industry likes: positive, pooled, and abstracted into a slide.
Six months later the same journal published a response from Rohan A. Elliott that took the pooling apart, and one sentence in it should be read by anyone who sells, buys or invests in medication dispensing. The largest trial included in that meta-analysis — 26,635 patients — reported “no difference in the odds of optimal adherence in patients who received a DAA (odds ratio 1.03, 95 per cent confidence interval 0.95–1.13).”
Twenty-six thousand patients. No effect.
And yet the global smart pill dispensing market is USD 1,850.0 million in 2025 and reaches USD 4,912.0 million by 2033 in the Pheonix Research model. Both facts are true at once, and the reconciliation is the most interesting thing in this market: the device that gets sold is not the device that was tested.
What the review actually found, and what the critique says
Elliott’s objections are methodological and, read commercially, devastating. Many of the trials that showed a benefit bundled the device with something else — regular pharmacist follow-up, an hour of one-to-one education, telephone calls after missed doses. The adherence gain cannot be attributed to the pack when the pack arrived with a pharmacist attached. Several included trials did not test a dose administration aid as the protocol defined one: smart pill bottles, blister packs, sachet systems and compartmentalised plastic boxes were pooled together, though they intervene in completely different ways. Adherence data were combined across single-drug short courses and multi-drug chronic regimens, self-filled and pharmacy-filled packs, and devices with and without electronic reminders.
And the effect estimate itself was dominated by a null result. When the largest single trial in a pooled analysis finds nothing, the pooled average is being carried by small studies with heterogeneous interventions — a familiar pattern in adherence research, and one that earlier reviews had already read as insufficient evidence, especially in the population that uses these devices most: older people.
This is not an argument that packaging does not work. It is an argument that the evidence does not support the specific claim the category makes — that a dispenser raises adherence — and that the claim has been kept alive by pooling interventions that were never the same thing.
So what is being bought?
Something more modest and far more defensible: logistics. A filled, sealed, time-stamped pack that contains the right drug, at the right dose, at the right time, and a connected device that logs whether the compartment was opened. That is not a behavioural intervention. It is a supply-chain intervention with an audit trail.
Once you see it that way, the market’s odd shape stops being odd. A pack cannot be double-dosed by a confused patient. A sealed pack cannot be filled from the wrong bottle by a tired caregiver. A logged dose time is a record a pharmacist can reconcile and a plan can audit. None of that requires an adherence effect — and all of it is worth money to somebody carrying the risk of a medication error or a readmission.
It also explains which products are surviving. Philips’ medication dispenser line, the PMD, is out of production: no new units, no factory parts, no factory repairs, and the aging and care business that carried it was acquired by Connect America. The category’s most institutionalised hardware is a supported installed base rather than a shipping product. Meanwhile pharmacy-filled multi-dose packaging keeps scaling, because it requires no behaviour change from the patient, no charger, no Wi-Fi and no explanation. The thing that requires nothing of the patient is the thing that scales.
The marketing problem this creates
Consumer-facing material for smart dispensers routinely quotes adherence in the nineties — one widely circulated guide to the category puts it at 93 to 98 per cent “across studies.” Those numbers sit next to a twenty-six-thousand-patient trial that found no effect, and buyers eventually notice. The result is a marketing funnel with high return rates: a 30-day money-back trial, an emotional purchase by an adult child, and a product that has to keep proving itself to a family that has no way to measure whether it is working.
Contrast that with the two channels where the purchase is contractual. PACE organisations grew from 178 at the end of 2024 to 198 by December 2025, take full capitated risk for frail older adults, and buy services whose savings land on their own income statement. State Medicaid programmes that reimburse compliance packaging — Washington State pays a filling fee plus the cost of the packaging device for eligible at-risk clients — are buying a documented process, not a promise about behaviour.
What a buyer should underwrite instead
Three things, none of which is an adherence lift.
Process reliability. Fill accuracy, on-time delivery, pack integrity, and the rate at which a missed fill is detected before it reaches the patient. These are measurable from the vendor’s own system, and they are what a pharmacy contract should be priced against.
An auditable record. The dispensing event, the drug, the dose, the time and the person. That record is the only artefact in the chain that a payer can reconcile, and it is the reason packaging can be reimbursed while a reminder appliance cannot.
Avoided utilisation in a defined population. Not adherence in the abstract — admissions in a cohort, against a matched comparison. This is the only endpoint that justifies a capitated contract, and the category’s evidence problem is that almost nobody has published it at the population level in the populations that use the devices.
Why the forecast still says 12.98 per cent
Because the buyer is not buying the randomised trial. Our model rests on patient-years under managed dispensing rising from 3.20 million to 6.24 million, and revenue per patient-year rising from USD 578.13 to USD 787.18 as the bundle — pack, delivery, monitoring, pharmacist review — gets richer. Two-thirds of the growth is more covered patients, one-third is more revenue per patient. Neither input depends on a disputed adherence effect; both depend on whether a risk-bearing payer signs.
That is also why the scenario spread is so wide. Hold revenue per patient-year flat and 2033 is USD 3,607.5 million. Stall the payer shift and it is USD 2,852.0 million. If capitated and Medicaid channels buy aggressively, it is USD 6,375.0 million. The variable is not the science. It is the contracting.
There is a version of this market in which better evidence arrives — utilisation endpoints, in defined populations, showing that a sealed and audited pack reduces admissions. That is the upside case, and it is the case worth waiting for. Until then, the honest pitch is narrower and stronger: we will not promise you that the patient takes the pill. We will promise you that the pill was in the right compartment at the right time, and that we can prove it.
Figures in this article are drawn from the Pheonix Research report Global Smart Pill Dispenser Market Report, Size, Share and Forecast 2026–2033. Sources include the peer-reviewed adherence literature, CMS 2026 Star Ratings publications, CMS PACE programme data, the Washington State Administrative Code, and vendor and distributor disclosure.
Frequently Asked Questions
What does the clinical evidence say about pill dispensers and adherence?
The largest randomised trial of dose administration aids, enrolling 26,635 patients, found no difference in the odds of optimal adherence — an odds ratio of 1.03 with a 95 per cent confidence interval of 0.95 to 1.13. A 2025 critique of the most-cited meta-analysis by Rohan A. Elliott notes that many of the pooled trials bundled the device with pharmacist follow-up, education or reminder calls, and pooled different kinds of intervention together.
Why does the market grow if the evidence for adherence is weak?
Because what is sold is logistics rather than a behavioural effect: a filled, sealed, time-stamped pack containing the right drug at the right dose, plus a record of whether it was opened. That is a process intervention with an audit trail, and it is worth money to whoever carries the risk of a medication error or an avoidable admission.
Why is pharmacy-filled packaging outgrowing in-home dispensers?
A filled and sealed pharmacy pack requires no behaviour change from the patient — no charger, no connectivity, no explanation — and it produces a record that a payer or pharmacy can reconcile. Philips' PMD dispenser line is out of production, which demonstrates how exposed in-home hardware is to service discontinuity.
What should a buyer underwrite instead of an adherence claim?
Process reliability such as fill accuracy and on-time delivery, an auditable dispensing record, and avoided admissions measured in a defined population against a matched comparison. All three can be verified from a vendor's own system or a plan's claims data, unlike an adherence claim resting on a null randomised trial.
