What sponsors use connected dispensing for

Sponsors come to connected dispensing for different reasons. This guide covers the five we see most often.

Replacing pill counts for medication adherence

Most trials work out adherence by counting returned tablets at each visit and comparing the count against a paper or app diary. Site staff do the count, enter it into the EDC and raise a query when the count and the diary disagree. This takes time at every visit, and it only tells you how many pills left the bottle, not when. Studies comparing pill counts with electronic monitoring have repeatedly found that counts overstate adherence, partly because some patients empty tablets before a visit.

A connected dispenser records each dispense with a timestamp as it happens, and the record reaches the platform without the patient or the site doing anything. When the protocol names the dispenser record as the adherence measure, sites no longer calculate adherence from counts, transcribe diaries or chase queries when the two disagree. Many sponsors will still count returned tablets for drug accountability, which GCP requires. That count becomes a quick cross-check against an independent record rather than the source of the adherence data.

Per-dose monitoring without switching to smart blister packs

Sponsors moving from Phase 1 to Phase 2 often want more than a count of what came back at each visit. The drug could be moving to remote, home use for the first time, and sponsors need to know doses are still taken on schedule when the patient is alone. The usual answer is to move the drug into blister packs. That means new tooling, a more complicated filling and labelling process, and stability work on a new primary pack, all before the Phase 2 timeline starts.

A connected dispenser fits onto the bottle the sponsor already used in Phase 1, so manufacturing, packaging and distribution don’t change. The participant receives the bottle as normal, and the dispenser goes on once they have it at home. The dispenser releases one tablet at a time and records each dispense with a timestamp, which gives a per-dose record like a blister would. It can also be configured to control dosing, for example the number of tablets per dose and when doses can be taken. Each dispenser is set up for a specific tablet by fitting an insert sized to it, and we verify the tablet and bottle combination before the first batch ships.

In a recent enquiry, a sponsor planning a placebo-controlled Phase 2 study with a dose of one to three tablets a day had assumed per-dose monitoring meant retooling for blisters. Keeping the Phase 1 bottle meant they didn’t need to repackage at all. Blinding works the same way as it does with bottles. Each dispenser carries a unique serial number, and any label is plain text chosen by the sponsor, who keeps the mapping to treatment on their side.

Protecting study power when adherence slips

When some participants don’t take the drug, the treatment effect the trial measures gets smaller than the drug’s real effect, and the sample size needed to detect it grows with the square of that shrinkage. As a simple illustration, assume participants who don’t take the drug get no benefit and nothing else changes. If 20% of the active arm take nothing, the observed effect falls to 80% of the true effect and you need about 1.6 times as many participants for the same power. At 30% you need about twice as many. Real adherence is usually partial, so the numbers are rarely this neat, but you get the picture. Sponsors either oversize the study to allow for it, or find out at unblinding that the study was underpowered.

An adherence run-in before randomisation identifies participants who don’t take their medication during a lead-in period so they can be excluded. This is a recognised enrichment strategy and the most direct way adherence data can support a smaller study. During the study, sites can act on non-adherence early, which limits how much the effect gets diluted. In the analysis, the statistical analysis plan can use dose data to handle non-adherence as an intercurrent event under ICH E9(R1), rather than relying on pill counts. And the next study can be sized on the adherence that was actually measured instead of an assumption. The primary analysis for regulators is typically intention-to-treat, so monitoring on its own doesn’t shrink a study, but it does lower the chance of a working drug failing because participants didn’t take it.

Dosing history for PK and exposure-response analysis

Population PK models use the dose times recorded in the study to estimate how the drug is absorbed and cleared. In outpatient studies those times usually come from diaries or from an assumption that the patient followed the schedule. When a participant takes a dose late or skips one, a concentration sample can look too low or too high, which biases the model’s parameter estimates and inflates the variability it reports.

A timestamped dispense history gives modellers the time each dose left the bottle. That is a close proxy for when it was taken, though not proof of ingestion. With it, modellers can explain outlying concentrations, tell missed doses apart from real differences in metabolism, and relate exposure to response using the doses participants dispensed. In Phase 2 studies with sparse sampling this matters more, because each sample carries more weight in the model.

Real-time adherence monitoring and course correcting

Because data arrives during the study, the trial team sets alert rules at setup, for example three days of average adherence below 80%. The platform sends an alert to the people the sponsor nominates as soon as the data shows a participant has met the rule. For participants with a phone, data arrives in the background whenever the phone’s operating system allows a read. For participants without one, the dispenser stores every event and the site reads it at the visit. Every participant uses the same device either way. The site contacts the participant, finds the cause and logs the action on the platform. Clinical decisions stay with the sponsor and sites, and Pill Connect monitors the technology.

By the end of the study the sponsor has a record for every participant. If adherence held, it shows that. If it didn’t, it shows each problem and how the site dealt with it. That record is what supports exposure confidence when the results are reviewed, meaning reviewers can see that participants received the doses attributed to them.

Further reading

For the methods themselves, see our guides to adherence monitoring methods, decentralised trials, responding to non-adherence, smart pill bottles and smart packaging.