Practice What You Preach Is Not a Cliché. It Is a Governance Mechanism.

Based on the research ofWang, Luo, Geng, Chen and Teo, "Living Up to Online Advice: How Health Platforms Influence Physicians' Offline Practice," Information Systems Research, 2026

The advice was never the point. The public commitment was.

Qiu-Hong Wang, Kai Luo, Ruibin Geng, Xi Chen, and Hock-Hai Teo built a physician-level data set that links a leading Chinese online health Q&A platform to hospital electronic health records, then tracked what happened to the same doctors' inpatient decisions. Physicians with higher levels of online advisory activity showed reduced overall medication costs, lower uncovered medication cost ratios, and lengths of stay that hewed more closely to clinical guideline benchmarks. Mediation and stratified analyses rule out the easy explanations: this is not simply healthier patients selecting more frugal doctors, and it is not just that patients who read a doctor's online answers arrive more compliant. The moderating pattern instead lines up with cognitive consistency theory. The effect is strongest exactly where professional commitment and topical relevance run highest, and the domains most affected are the discretionary ones, where a physician's economic and institutional pressures normally have the most room to bend a decision.

The authors call the mechanism identity-based digital commitment. A publicly articulated medical answer functions as a cognitive anchor. Once a physician has told a stranger, in writing, under a real professional name, what the appropriate course is, walking into a hospital room and doing something visibly less careful creates a private, uncomfortable gap between the identity they just performed and the identity they are about to enact. Nobody has to enforce that gap. The physician closes it themselves. And the paper's cleanest reassurance is that this self-correction does not come at the expense of outcomes: readmission and mortality rates stay flat even as costs and length of stay move toward the guideline. The platform is not making doctors do less. It is making them do the same amount of medicine for less waste.

China's prescribing problem is exactly the kind of discretion this targets

The discretionary domains the paper flags, medication spend and length of stay, sit on top of a well-documented distortion. Yongbin Li and coauthors, writing in Health Affairs, examined 230,800 prescriptions written between 2007 and 2009 across 784 community health institutions in 28 Chinese cities and found patients were prescribed antibiotics at twice the rate recommended by the World Health Organization and given injections at more than three times the rate of comparable countries, a pattern the authors trace directly to the fact that drug sales constitute a major share of Chinese providers' incomes. That is the backdrop the anchor paper is working against. Overprescribing in China is not an edge case; it is a structural incentive baked into how physicians are paid, which is precisely why a lever that pushes the other direction, without adding a single line of formal oversight, is worth taking seriously.

It is also why this literature keeps returning to the same kind of platform. A closely related study by Wenjuan Fan, Qiqi Zhou, Liangfei Qiu, and Subodha Kumar, also in Information Systems Research, used data from Haodf.com's online consultations paired with Guahao.com's offline appointment bookings to show that doctors who open online consultation channels see their offline appointment volume rise, evidence that patients read online engagement as a signal of a physician's willingness to communicate well. A separate paper by Sandeep Khurana, Liangfei Qiu, and Subodha Kumar found that thoughtful, substantive answers in a health Q&A forum function as a costly signal of medical expertise, which is part of why physicians treat the writing seriously rather than phoning it in. This is not a fringe activity happening on the margins of Chinese healthcare. Ant Group's January 2025 acquisition of Haodf.com put roughly 280,000 verified doctors on a platform now wired into Alipay's 800 million healthcare users, the kind of scale that turns an individual habit into an ambient feature of how an entire profession is governed.

This is not the Yelp effect. It runs the other way.

The obvious skeptical read is that this is just reputation management with extra steps: doctors perform frugality online because they are afraid of how it will look if they don't. That story exists in American healthcare, and it is called the Yelp effect, the hypothesis that urgent-care physicians overprescribe antibiotics because they fear a bad review from a patient who wanted a prescription and didn't get one. It is worth taking seriously precisely because it predicts the opposite of what the anchor paper finds. When Dian Hu and colleagues actually tested it, publishing in the Journal of Medical Internet Research, they combed 481,825 web-based urgent-care reviews and found the Yelp effect pattern in only about 1,696 of them, and a close hand-coded read of 200 sampled reviews found it as the primary complaint in just 2.5 percent. The fear-driven story that patient ratings scare doctors into looser prescribing turns out to be mostly folklore.

The Yelp effect governs by fear and barely shows up in the data. Cognitive consistency governs by identity, and nobody built it on purpose.

That contrast is the real finding underneath the anchor paper. Reputation fear and identity consistency are not the same lever pointed in the same direction. One depends on a patient noticing, complaining, and being believed, a chain that mostly does not fire. The other depends only on the physician's own memory of what they told a stranger last week, and it fires reliably enough to move measurable costs. A governance channel that does not need an audience to work is a strange, useful thing for a hospital administrator to have, and it is not one anybody currently budgets for.

The limits of governing by conscience

None of this is a substitute for formal oversight, and the paper does not claim it is. It calls the effect "soft governance," a complement to formal review rather than a replacement for it, and the framing matters because the mechanism has an obvious ceiling. It only reaches the physicians who choose to answer questions online, it only works in domains where the physician has actually stated a position, and it depends on the discomfort of self-contradiction being stronger than the pull of a fee-for-service incentive, which the paper's own moderation results suggest is true only where professional commitment is already high. A physician who never posts, or who posts vaguely enough to avoid ever pinning themselves to a specific standard, gets none of the effect and faces none of the pressure.

The reason to pay attention anyway is that the underlying behavior is no longer optional at scale. Telehealth utilization, tracked by McKinsey in the aftermath of the pandemic surge, settled at roughly 38 times its pre-COVID baseline rather than reverting, a stabilized new floor for how much clinical interaction now happens through a screen instead of a waiting room. Physicians are going to keep answering questions online whether or not any hospital administrator decided that was a governance strategy. The choice left on the table is narrower than it looks: notice that a free, self-enforcing consistency effect is riding along inside that activity and design lightly around it, or keep treating physician Q&A time as a distraction from the real work, and leave a governance mechanism running unmanaged and unmeasured in the background of every shift.