The Side Effects of Releasing Public Health Insurance Data to the Public

flying cadeucii

Three of the five largest private health insurers in the US – UnitedHealthcare, Aetna, and Humana – have decided to follow the lead of the Centers for Medicare & Medicaid Services (CMS) and release their payment information to the public. According to Bloomberg News, this data will include 5 billion individual medical claims and $1 trillion in spending.

Releasing payment information by governmental and private health insurers is an important step towards transparency. Providing researchers with access to the details of health insurance payments is an unprecedented and long-awaited opportunity to gain insights into the drivers of rising healthcare costs. Although I share the enthusiasm of many other researchers for analyzing this valuable data, I am also concerned with unanticipated consequences that may arise with unrestricted release of sensitive and complicated healthcare insurance data to the public.

Reputation of Physicians

The performance of physicians, as some of the most reputable and highly specialized professionals of our society, cannot be evaluated only based on their insurance billing history. To the untrained eye, the abnormalities in insurance charges may seem unjustifiable. Deep expertise in the medical domain is required to investigate all of the underlying causes of the abnormal prescriptions, medical procedures and equipment utilizations. Accusing physicians of malpractice or misconduct based on hasty analysis of this data and without careful examination of the unique medical context in each case, would be unfair to those who deliver medical care to patients.

Privacy of Physicians

To increase the sales of their branded and expensive drugs, pharmaceutical companies have been mining prescribing practices of physicians for many years. Although some states like New Hampshire, Vermont, and Maine set laws against mining healthcare data for commercial purposes, the Supreme Court revoked these laws primarily on the basis that they amount to an unconstitutional infringement of protected commercial speech. Although mining prescription data is now legal, physicians still can protect their privacy and have the option to opt out of having their data shared with pharmaceutical industry representatives.

Access to insurance claims data is provided to the public as a means to reduce the costs and increase the quality of the healthcare system. However, the commercial entities such as pharmaceutical companies can also use this data towards their own financial benefits without requiring the consent of physicians. These data mining practices may result in unsolicited offers and advertisements to physicians by pharmaceutical and medical manufacturers. There should be procedures for physicians to opt-out of unwanted advertising.

Impact on Comparative Effectiveness

According to the editors of New England Journal of Medicine, selling of prescribing data to pharmaceutical companies results in the manipulation of physicians’ drug-prescribing practices and increases the healthcare costs by “over-prescription of newer, costlier drugs with little-known side effects that, in fact, may have no demonstrable benefit over existing brand-name or generic choices.” If mining prescribing data can be used to effectively pitch new, yet not beneficial drugs and consequently raises the overall healthcare costs, it is not unlikely that the same unwanted results will happen by mining insurance claims.


CMS and private health insurers shall carefully consider the potential negative consequences of their well-intended decision to release claims data to the public. Adequate clarification of the data elements, its limitations and implications can be useful in mitigating the risks of unfounded attacks to physicians. Releasing data on service quality measures along with the insurance claims data can be helpful in evaluating billing practices within the medical context. Procedures should be developed that reduce the risk of unwanted commercial marketing by business entities.

1 reply »

  1. Another reason for more education – both formal and informal – on data, statistics, and understanding how to parse same. Context is, most certainly, a critical part of reading anything, but how about using this as an opportunity to teach ever’body how to understand what they’re looking at?

    Granted, the real challenge, between-the-ears-wise, is educating journos on not knee-jerking a screaming headline out of the wrong read of a dataset. But, as either Sam Clemens or Ben Disraeli said (I wasn’t there, so I can’t say for sure), the general consensus, when it comes to statistical reporting, is, “Lies, damn lies, and statistics.”