Physicians as Persuaders: Evidence from Hospitals in China

This article features the work of Jia Xiang, Assistant Professor of Business Economics and Public Policy at the Kelley School of Business. 

Patients visit a physician to be treated for an illness or other medical condition.While physicians do consider their patient’s well-being when giving treatment advice, in reality, there are often more factors at play than simply the patient’s health and wellness.Physicians’ financial interests, including how much they are paid for the service, can cause their interests to be different from their patient’s. These interests become especially relevant in conditions with discretionary treatment options. 

On the other hand, the patient usually assumes — or is expected to assume — that the physician will make treatment decisions based on the best possible outcome for the patient as well as their own financial interests.  

Statement of Problem 

 In her research on physician-patient interaction in medical treatment decisions, the author explores the impact of physicians’ financial incentives, information asymmetry, and insurance rates on treatment decisions. The author’s main research objective is to determine the influence of physicians’ partially self-serving, though more informed, recommendations on patients’ treatment decisions.  

The study examines three related questions: First, to what extent are treatment decisions contrary to patients’ best interests a result of knowledge imbalance between physicians and patients? Second, how does this distortion depend on each side’s financial incentives? And finally, how does this dependence impact policies that design incentives for each side? To answer these questions, the author investigates treatment decisions in a diagnosis where both surgical and non-surgical treatment options can be effective.  

Data Sources Used 

The study utilizes anonymous health insurance claims data that was obtained from the Healthcare Security Administration in a large district in a western Chinese city. The data focuses on patients with a specific diagnosis, cervical spondylosis, which is commonly known as neck arthritis. Neck arthritis can be treated by both surgical and non-surgical treatment, and there are no clear guidelines for selecting patients who may benefit from the surgery.  

The study’s observation period ran from June 2014 to June 2018, a period of time that followed the Chinese government’s cancellation of the drug mark-up in public hospitals. To partially compensate for hospitals’ revenue loss, the government raised regulated prices for medical services and began to require that raised prices for medical services compensate for more than 80 percent of the potential loss of drug revenue. The hospital inpatient records used for the study come from all 28 hospitals in the district, some of which implemented the policy change during the observation period. Records include total cost by procedure category, out-of-pocket price to patient, insurance payment totals, patient insurance type, patient coinsurance rate, and more.  

Analytic Techniques 

Previous research frameworks examining physician-patient decision making primarily consider an objective function that weighs both patient well-being and physician financial incentives. The author builds on this by explicitly modeling how patients incorporate physician’s self-interested, though more informed, recommendations into their treatment decisions. She does so by applying a Bayesian persuasion framework to the physician-patient interaction.  Then she takes this model to data to empirically investigate the distortion caused by information asymmetry.This data set allows the author to observe independent and exogeneous variations in physicians’ compensation and patients’ coinsurance rates, which is essential to separating physicians’ roles from patients’ roles in decision making.  

Theoretically and empirically separating the role of patient choice from the physician’s role generates several new policy-relevant insights. First, there is a hump-shape relationship between the degree of physicians’ distortion and the conflict of interest between the two parties. An intense conflict of interest makes patients so skeptical, that physicians have to misdirect patients less frequently to get their treatment recommendations accepted. But when the conflict of interest is lower and patients are less skeptical, physicians can misdirect more patients toward surgery as their compensation for surgery increases.  

Second, she hypothesizes that giving patients the same amount of information as their physicians disciplines the physicians to give recommendations based solely on patients’ utility, even if the physician faces financial incentives.  

Third, if a patient with a lower coinsurance rate is easier to persuade. Because the patient becomes less price-sensitive, the physician may be able to recommend their preferred treatment more frequently.  

Results 

The author finds that for neck arthritis, which has discretionary treatment options, nearly half of the patients who received surgery would not have done so if they were fully informed. At the same time, removing physicians’ financial incentives was shown to achieve the same treatment decisions as fully informing patients does. Furthermore, the effect of physicians’ compensation on their treatment recommendations was 1.5 times larger for more insured patients.  

Policy Implications  

According to the author, the Bayesian persuasion model is valuable in this study because it allows for the measurement of a new channel of moral hazard, i.e., the greater misdirection of patients due to a lower coinsurance rate. This measurement is critical for any welfare analysis of insurance policy, and it is only feasible when patients’ decision-making process is explicitly modeled. She considers a policy that lowers patient coinsurance rates by half. A decomposition of the effect indicates that both the traditional moral hazard and the greater misdirection based on cost-sharing contribute to a rise in surgery rate of 17 percent, with the misdirection effect in particular accounting for one-fifth of this surge. While patients are better off with a welfare gain of 8 percent due to a lower out-of-pocket price, greater misdirection offsets the patient welfare gain to 2 percent, indicating that some patients are worse off due to surgeries they would not have opted for otherwise. 

 

Xiang, Jia, “Physicians as Persuaders: Evidence from Hospitals in China,” Working Paper, 2023.