Reliance on Published Survey Data Doesn’t Necessarily Yield FMV

Given the regulatory risks associated with failure to pay fair market value (FMV) in certain physician compensation arrangements, parties contracting with physicians for services are rightly concerned about FMV and are justified in their efforts to find effective means by which FMV can be determined and documented.

One such means for the assessment of FMV is the use of widely available published survey reports on physician productivity and compensation. Several sources exist for this information, and there is a broad acceptance of published surveys in the industry as indicative of compensation for physicians in many specialties. Let’s look at what the government—specifically CMS—says about the use of surveys.

In Stark II, Phase III, after eliminating the now-defunct FMV Safe Harbor, CMS responded to commenters by saying, “Reference to multiple, objective, independently published salary surveys remains a prudent practice for evaluating fair market value.” [72 Fed. Reg. (Sep. 5, 2007), 51015.]  CMS, however, in the preamble to Phase I precluded the use of data tainted by a referral relationship: “Depending on the circumstances, the ‘volume or value’ restriction will preclude reliance on comparables that involve entities and physicians in a position to refer or generate business.” [66 Fed. Reg. (Jan. 4, 2001), 944.]

Gathering information from every physician in the country about what they produce and earn isn’t going to happen; thus, sampling would be a viable method of studying the nation’s population of physicians. Consider a survey derived from a random sample, the size of which was determined from the population of every practicing physician. Not in my lifetime. However, most surveys begin with a non-representative sample, often based on association or society membership. Not to critique existing surveys (most are well thought-out, extremely competent, and contain very useful information), but nearly all are limited in their ability to capture a statistically valid sample of the nation’s physician population. Thus to rely on published survey data as statistically valid when based on non-statistical sampling is flawed logic.

I’ve lost count of the number of folks who have argued that anything below median—insert any other quartile/decile—is FMV. Notwithstanding the concerns over statistical validity, median is simply the mid-point of respondent data; thus, half are paid below and half make more than median. While median might reflect one’s level of risk tolerance, it isn’t a broad indicator of FMV.

The point to be considered is this: sole, unschooled reliance on published survey data doesn’t necessarily yield FMV, and the result could be materially incorrect! Prudent practice in the use of the surveys means using “multiple, objective, independently published salary surveys” according to CMS (supra, 72 Fed. Reg.). Other prudent uses of survey data include measures of productivity where possible, such as methods that compare similar production and compensation levels (Percentile Matching Technique) and apply rates of compensation per unit production (Median Rate Technique).

Surveys are susceptible to trends, and the healthcare industry and its regulatory environment have created some trends that make for unique difficulties in applying survey data. For example, some surveys have seen a transition in respondent ownership over time, transitioning from physician-owned practices (POPs) to hospital-owned practices (HOPs). What was primarily POPs data a decade ago has become mostly HOPs data, a trend accompanied by the bent of many hospitals with ancillary-dependent practices to shift ancillaries into the hospital outpatient department and away from the HOP. These ancillary profits are often derived from Designated Health Services prohibited under the Stark Law, creating a conundrum for those using the data. Furthermore, practice-specific payer arrangements, revenue cycle and payer mix nuances, and local economic conditions aren’t reflected in survey data, which are by nature much broader.

Cost and income approaches represent other commercially reasonable methods for the valuation of compensation arrangements, provided their use is applicable and they are applied within the regulatory framework. While this blog won’t dissect the methods associated with these approaches, the use of multiple methods serves to mitigate the weaknesses in one or more others. The strengths and weaknesses of each are then identified in the synthesis process, and a thought-through conclusion emerges. That, in my way of thinking, outweighs the sole, unschooled reliance on survey data.

For weekly insights into healthcare, please sign up here: