In my previous post I pointed out a study that suggested that providers actually bill less, total, to uninsured patients for a doctor's visit than they bill to insured patients (and their insurers). This flies against widespread claims that the uninsured have to pay more for healthcare than insured patients because they have to pay full charges instead of the negotiated discounts insurers get. It also challenges the conclusions of all research studies based on charge data—charges being more than double, on average, than what providers actually bill for—which had been justified on the grounds that uninsured patients typically have to pay full charge prices. Today the New York Times has a column out highlighting a study in Health Affairs by Ge Bai and Gerard F. Anderson that compares charges to costs at various types of hospitals. So if charges are meaningless, what good is a charge-to-cost comparison? The Times:

"The typical response from the hospitals is that their list prices are not relevant to most patients because they don't pay the full price. The two big government insurance programs, Medicare and Medicaid, pay far less than the list prices; private insurers typically bargain for discounts. Even so, high list prices raise the ceiling from which discounts are negotiated and thus drive up premiums for many privately insured patients.
    "High prices will hit millions of people who will remain uninsured in coming years for one reason or another despite passage of the Affordable Care Act, according to the Health Affairs study."

I've not done a complete review of the literature on this question. However, I did look up each citation offered for this claim and attempted to trace these citations back to the original data. It has been a lesson in citogenesis. Bai and Anderson cited three papers in defense of their assertion that uninsured patients typically have to pay full charges:

1. Anderson (2007)
2. Thompkins, Altman, and Eilat (2006)
3. Reinhart (2006)

EIU District 1199 Care for Ohio, "Twice the Price: What Uninsured and Underinsured Patients Pay for Hospital Care," March 2005, http://s57.advocateoffice.com/vertical/Sites/{56490583-267C-4278-BC56-A7128CE248A8}/uploads/{374CBAD9-740D-48BC-8536-92ECD76D1444}.PDF (accessed 1 July 2005)

In addition, Anderson (2007) claimed to have confidential information from testimony and documents prepared for trial in lawsuits between providers and "self-pay" patients over their billing practices. While we aren't given any specifics or access to the underlying data, the existence of trials over these practices suggests, at least, a perception that self-paying patients are billed in unfair ways such as being denied discounts similar to the ones insurers get. Since the author himself is the primary source, I will refer to the endnote numbers in which he cites his own confidential communications and observations. In citation 19, Anderson notes that he had access to the testimony of the executives of healthcare providers in these lawsuits, who spoke to the reasons that their firms only offered small discounts to uninsured patients. In citation 22 Anderson notes that uninsured patients typically can get at least a 10% discount, if they ask, which is a tiny fraction of what would be offered to insurers but enough to prove that providers did not withhold discounts for regulatory reasons as some executives had claimed. Citation 23 states

"For example, Hospital Corporation of America's Web site states that HCA 'provides free care for any patient who receives non-elective treatment and whose household financial resources and/or income is at 200 percent or below the Federal Poverty Level.' Some patients above 200 percent of poverty will qualify for 'a managed care like discount.'"

That leaves us with Thompkins, Altman, and Eilat (2006). They do not merely claim that uninsured patients typically pay full charges (or close to it), but that this has been a deliberate pricing strategy by providers to offset cuts from private and public insurers (as opposed to an unintentional bureaucratic error). They have constructed a whole historical narrative of the charge-to-cost ratio premised on this claim:

"As private third-party payers consolidated in the early 1990s and their market clout grew, they moved away from negotiating with hospitals based on charges and toward contracts based on lower fee schedules or negotiated rates. Accordingly, billed charges defined prices for a shrinking proportion of patients. Hospitals responded by marking up billed charges even faster than the costs of care for such patients. This scenario resulted in an increasing gap between billed charges and the prices paid by most payers. ... This gap has grown steadily since the early 1980s and has accelerated in recent years. This acceleration is attributable to several factors. The employer-sponsored insurance market shifted in favor of managed care, giving those health plans volume and clout to obtain greater discounts from hospitals. Medicaid enrollment grew significantly, providing more stable revenues than the uninsured, but nevertheless paying rates much lower than actual costs. The enactment of the Balanced Budget Act (BBA) of 1997 lowered the growth in Medicare payments, resulting in a cut of roughly $70 billion, or 9.1 percent in hospital payments, over the five-year period 1998–2002 from the pre-BBA Medicare baseline. Thus, payments from public programs and many private third-party payers were increasingly below what hospitals believed to be appropriate for the services provided. All of these forces put pressure on hospitals to squeeze more revenues from a diminishing pool of other sources. The main technique used by many hospitals was to raise their billed charges. This raised to varying degrees the prices charged to payers not under contract, payers with contracts in which most or all payment rates were linked to charges, and services that were outside the scope of fixed- or negotiated-price contracts. Hospitals have continued to rely on charge-related sources of revenue, even from groups that generate very limited amounts of revenues such as self-paying patients."

"When I performed pooled crosssectional regression to examine the relationship between increases in charges and net revenues over the period 1984–2004, I found that hospitals received only a very small proportion of the increase in charges above the rate of increase in costs"

"In our interviews,^[2] hospital officials reported that they were reluctant to bill the uninsured for less than full charges because of insurers' common negotiating practice of insisting on being charged the same as the lowest-paying customers. They also cited Medicare guidelines and kickback regulations as reasons for not offering discounts to self-pay consumers. Related to this, they have operated with the understanding that they were subject to prosecution for fraud and abuse if they gave discounts without prior negotiation to any patients (even if they were poor and uninsured)."

The actual data in Thompkins et al is on the time-trend of the charge-cost ratio, not the proportions of charges that uninsured patients pay. For the claim that uninsured patients pay a larger share of charges they cite two other papers. The first is Pryor and Seifert (2003) which attempted interviews with administrators at 30 hospitals, getting only 4 responses. Some of the executives who responded claimed that regulations prohibited discounts for uninsured patients—again, a claim that is untrue but suggests some hospitals do charge full charges to uninsured patients. Pryor and Seifert further suggest that providers that do discount lack effective systems to handle this

"If patients are not identified as qualifying for public programs or charity care, hospitals do not have standard mechanisms for distinguishing among patients in terms of the size of their bills and their ability to pay them, or for informing patients of their ability to negotiate discounts when appropriate."

The other paper cited by Thompkins et al in support of the claim that the uninsured pay full charges was Prottas (2007), which was primarily concerned with medical debt. Prottas (2007) surveyed a similarly small number of hospitals—"a half dozen"—and found that all of the hospitals did have programs to offer discounts of up to 100 percent of billed charges for uninsured patients and others who could not pay. Interestingly, in all 6 cases, these discounts were adjusted based on income and not insurance status, so insured patients could get additional discounts on out of pocket expenses. However, for some income groups discounts were as small as 10 percent. Moreover, in all but one case, billing departments attempted to collect full charges first, and did not tell patients of their discount programs until after they refused to pay.⁴

Combined, the evidence supporting the New York Times article consists of a non-scientific sample of 10 hospitals and some non-specific, anecdotal claims from confidential trial testimony. Moreover, these samples overwhelmingly say that the providers do, in fact, routinely discount the medical bills of uninsured patients by at least 10 to 30 percent and as much as 100 percent for the poorest patients, although possibly not as much on average as they would discount for those who have insurance. Regulatory uncertainty may contribute to the complexity and opacity of hospital pricing. This evidence does not support the claim that the uninsured pay full provider charges, nor does this evidence support the claim that higher charges imply higher costs to uninsured patients. We need a larger and more representative sample to compare the actual amounts that insured and uninsured patients are billed after discounts and adjustments are made.

As a final point, I want to note that none of this is criticism of any of the papers I reviewed in pursuit of validating the New York Times and Bai and Anderson's claim about the uninsured paying full charges. Most of these cited papers were not primarily concerned with that claim as a research question, and the fact that they do not offer substantive evidence for this claim should not be taken as a suggestion that they do not offer value on the topics they actually aimed to study.

1. The source offered for the 10 percent figure is this. I did not find that claim in that source, but their data is such that it seems plausible the authors would have been able to calculate this figure. I will assume that Anderson had seen results from that study that did not make it into the final paper. In any case, Prottas (2007) provide similar numbers from—as far as I can tell—independent sources.

2. It's not clear to me what interviews the author is describing—the paper did not contain any description of interview data collection. However, one of the papers they cite, also from the Commonwealth Fund, did conduct interview surveys of hospitals—with only 4 respondents—that offered results fitting this description. They also cite this paper, which describes surveys of "a half dozen" hospitals.

3. The high prevalence of hospital billing errors, of course, are a similarly notorious and poorly sourced claim.

4. Note: all providers in each of these papers that surveyed providers did have programs that collected patient's income information and, if they were eligible, helped them enroll in public assistance programs such as medicaid and medicare. Only one of the hospitals surveyed used this information to automatically refer low-income patients who did not qualify for a public program to the hospital's discount programs. Whether the lack of automatic discounts represents a deliberate attempt to gouge patients or is merely an institutional failure, this could be a basis for the widespread perception that uninsured patients have to pay full charges for healthcare.