Digital Health and Access to Medicines

Most digital transitions are hard to evaluate because the service changes at the same time as the channel. Pharmacy is a partial exception. The product is standardised and regulated, prices are observable, and the counterfactual - the same medicine bought over a counter - exists in parallel. That makes it a useful case for studying what digital delivery does to access, cost and safety, without the confounding that makes most e-health evaluations inconclusive.

Access is the clearest effect

The population that benefits most is specific and identifiable: people with limited mobility, people in areas where local pharmacy provision has thinned out, and people managing long-term conditions with repeat prescriptions. For these groups the relevant comparison is not convenience against a short walk but availability against a journey they find difficult.

Rural provision is where this shows up in the data. Pharmacy closures follow population decline, and the resulting distances are exactly the case mail-order addresses. The effect is unevenly distributed in a way that matters: it is largest for the people whose alternative was worst, which is unusual for a digital service. Most channel shifts are captured by populations that were already well served, and this is a partial counterexample.

Price and the limits of comparison

Price differences are real but narrower than headline comparisons imply, and the reason is regulatory. In several European markets, prescription-only medicines are subject to fixed pricing, so competition happens entirely in over-the-counter products and in delivery and service terms. Comparisons that mix the two categories therefore overstate the achievable saving substantially.

Shipping thresholds are the other distortion. A price advantage on a single item frequently disappears once delivery is included, which is why comparisons of shipping-free providers, such as this German comparison of mail-order pharmacies, are more informative than headline unit prices. The interesting variation is in the total, and the total depends on order size in a way that unit prices conceal.

Verification is the safety problem

The genuine risk in online pharmacy is not the legitimate operators but the parallel market of sites that present as pharmacies and are not. The regulatory response in the EU is a common logo linking to a national register, so that a claim of registration can be checked against an authority rather than taken from the site making it.

This is a general pattern in digital service markets and worth stating in general terms. Moving a regulated service online separates the service from the physical signals people used to verify it - the premises, the visible professional, the local reputation. Trust then has to be reconstructed through infrastructure: registers, verifiable marks, and the practice of checking them. Where that infrastructure exists and is used, digital delivery is about as safe as the physical alternative. Where it exists and is not used, the mark becomes something to copy rather than something to check.

The data question

Digital pharmacy generates detailed records of what individuals purchase, which is among the most sensitive categories of personal data. Prescription histories are unusually identifying, both because combinations of medicines are close to unique and because they reveal conditions people have specific reasons not to disclose. Everything in digital traces and privacy applies here with particular force, and the re-identification results mean that pooled purchase data cannot be treated as anonymous simply because names were removed.

Modelling the transition

Whether a local pharmacy remains viable depends on how many customers shift, which depends on how many have already shifted - a threshold process rather than a smooth substitution. Below a level of switching, nothing changes; above it, the local provider closes, which converts the remaining customers to mail-order whether or not they wanted to switch. This is the structure described in how opinions spread, and it is a natural application for the agent-based methods discussed elsewhere on this site: the outcome of interest is a tipping point, and tipping points are precisely what aggregate demand models cannot represent.