INTRODUCTION

Health care “price transparency”—the ability of patients to obtain accurate price information before making health care decisions—is a growing policy focus.1 , 2 Hospitals in several states are now required to disclose prices to patients prior to providing care.3 Wide price variation and limited transparency for hospital services have been well-documented.4,5, – 6 By comparison, less attention has been focused on price transparency in outpatient settings, including primary care. A recent survey finds that most patients are interested in obtaining price information before receiving medical services, and over one-third who received primary care sought information before the visit.6

The ability to obtain prices before getting care, long important to the uninsured, is of growing importance to the privately insured, as rising deductibles have increased their out-of-pocket spending. The share of adults with private insurance that included a deductible rose from 60% in 2003 to 75% in 2014.7 Physician offices typically negotiate separate rates with different plans, making it challenging to provide specific rates for visits to insured patients who have not met their deductibles. Price information may be even more difficult to obtain for patients with plans purchased in the Marketplaces under the Affordable Care Act (ACA), because physicians’ offices are less familiar with these plans. Widespread confusion regarding Marketplace networks and rates was reported after implementation of the 2014 ACA insurance expansions.8

This study examines the availability of price information for patients seeking primary care visits in the summer of 2014, using data from an ongoing study of primary care appointment availability before and after the ACA’s major insurance expansions. We use a simulated patient (“audit”) methodology in which callers represent individuals with different insurance types. In 2012, the audit study asked about prices only for callers portraying uninsured individuals. Price information was available for 88% of uninsured patients who were offered appointments, and the average primary care visit cost $160.9 We hypothesized that similar patterns might continue for the uninsured in 2014, but that patients with private insurance would be less able to obtain price information than the uninsured, and challenges would be greatest for individuals with Marketplace coverage due to the relative newness of these plans. We further anticipated that price information might be more available in settings where patients were more cost-conscious, such as areas with more low-income individuals, and in smaller offices where office staff do both scheduling and billing. Alternatively, larger offices might be better equipped to provide detailed price information. We expected average prices to be similar across insurance groups, since the prior study showed that prices for the uninsured were similar to rates paid in those markets by private insurers.9 At the same time, differences in provider reimbursement across insurance plans are dependent on negotiated discounts and so may vary from the regular fee schedule. Finally, we expected that prices would vary somewhat by office and geographic characteristics.

To our knowledge, this is the first experimental study to compare access to price information for primary care services between privately insured and uninsured patients.

METHODS

Data Collection

Data were collected using interviewers who posed as patients seeking new patient primary care visits in ten states (Arkansas, Georgia, Illinois, Iowa, Massachusetts, Montana, New Jersey, Oregon, Pennsylvania, and Texas) between May and June 2014.10 , 11 Interviewers underwent extensive training with audit scripts. The 13 callers were female and male non-elderly adults of various ages. Primary care offices, defined as having at least one primary care physician treating adults, were selected at random within insurance type proportionate to the size of the county population with that insurance type from a representative sample frame of eligible offices. A description and validation of the sample frame, which includes federally qualified health centers (FQHCs) but not urgent care clinics, is described elsewhere.10 , 12 Only offices accepting the assigned insurance type and plan were called with that scenario.

Insurance types evaluated for the current study were employer-sponsored insurance (ESI), Marketplace insurance, and the uninsured. Calls were randomized to insurance type, and within type, to plans with a common regional carrier (e.g., Blue Cross). Plans were also assigned to a clinical scenario: checkup or evaluation for suspected hypertension (typically billed as a new evaluation and management visit). Callers did not volunteer their insurance type or plan until asked but, if not asked prior to receiving an appointment, confirmed that their insurance was accepted by the office. Callers with Marketplace plans indicated that the plan had been purchased on either the HealthCare.gov website or relevant state exchanges (e.g., the Health Connector in Massachusetts). In 2014, 85.7% of callers with the employer-sponsored insurance scenario, 82.4% of uninsured callers, and 74.3% of Marketplace callers were able to schedule appointments.13

Callers who were offered an appointment (N = 7865) asked follow-up questions related to cost. Employer-sponsored or Marketplace callers informed the scheduler they had not yet met their annual deductible and therefore wanted to know the “basic visit” price. Uninsured callers asked the same price questions.

Measures

We examined four outcomes: for all callers, (1) whether the caller was able to get any (versus no) price information; for callers unable to obtain a price, (2) whether the caller was referred to another source for this information, such as the office’s billing department or the caller’s insurance company (callers did not follow up with these sources); for callers able to obtain a price, (3) whether the visit would be free and, if not free, (4) the price. We report the exact price or the mean of the upper and lower bounds if given a range. Findings were not sensitive to excluding the top 1% of prices (>$500) and price ranges (>$360). Sample sizes for each variable differed based on the subset being analyzed (a flow diagram is provided in Online Appendix Figure 1).

For multivariable analysis, we included caller-level covariates: sex, age category (age ≤ 35 or >35), insurance type, and clinical scenario; office-level covariates: FQHC status, number of physicians practicing, and family medicine versus internal medicine specialty; county-level covariates for the office location: low, medium, or high county-level adult uninsured rate (<13%, 13–25.5%, >25.5%, respectively) and low, medium, or high county-level median income (<$45,000 per household, $45,000–$60,000, or >$60,000, respectively), and state fixed effects. The data sources for primary care office characteristics were the SK&A12 physician directory and the Uniform Data System.14 The County Health Rankings tool was used for county-level measures.15 Office-level covariates could not be linked to 6% of calls. We used regression-based imputation to fill in missing values on these covariates (results were not sensitive to excluding cases with missing values). Analyses were conducted using Stata version 13 software.16

Analysis

We calculated variable means stratified by insurance type. For each outcome, we estimated a regression model with the covariates described above. For binary outcomes, we estimated probit models; results were not sensitive to choice of logit versus probit. For price regressions, we fit the data to a gamma generalized linear model with a log link. To present estimates in either percentage points or dollars, we calculated predicted margins for each covariate and bootstrapped standard errors.

To account for differential sampling probabilities, we applied weights to all analyses in order to make calls proportional to county-level populations by insurance status. After weighting, our sample is representative of primary care offices in counties where individuals with each insurance type reside.

RESULTS

Characteristics of Callers and Offices

By design, about half the callers in each insurance type were randomized to the hypertension versus the checkup scenario (Table 1); about half were also women and <35 years of age. Family medicine physicians were contacted by 61.2% of employer plan callers. Just under half of all offices had one physician (45.9%), followed by 2–3 physicians (30.4%), and four or more physicians (23.7%). Means were similar across insurance types. Unless otherwise noted, all reported differences are statistically significant at the p < 0.05 level.

Table 1 Covariate Means
Full size table

Because separate samples were drawn for each insurance type proportional to populations in those counties with that coverage type, ESI callers were more likely to call offices in high-income counties and less likely to call offices in counties with high uninsured rates.

Ability to Obtain Price Information

The ability to obtain price information varied across insurance types. The unadjusted share of callers able to obtain a price was 89.2% for the uninsured, 61.8% for ESI, and 47.3% for Marketplace (Table 2). Among 3396 callers who did not receive a price, the unadjusted share of those referred to some other source of information was 25.8% for the uninsured, 72.1% for ESI, and 50.9% for Marketplace.

Table 2 Study Outcomes Stratified by Insurance Type
Full size table

Insurance type differences were very similar to unadjusted differences after adjusting for office- and area-level covariates (Table 3). Women were 8 percentage points more likely to obtain a price. FQHCs were 18.2 percentage points less likely to provide prices. In sub-analyses (not shown), we found that the availability of price information at FQHCs did not vary significantly between the insured and uninsured. Solo practitioners were 7.6 percentage points more likely, and offices with 2–3 physicians were 4.5 percentage points more likely, to provide a price than offices with four or more physicians. Compared to Texas, callers were significantly less likely to obtain price information in Massachusetts (−17.8 percentage points), New Jersey (−7.8 points), and Pennsylvania (−7.6 percentage points).

Table 3 Correlates of Being Able to Obtain a Price
Full size table

In a regression examining predictors of receiving a billing office or insurance company referral (Table 4) among those not offered an appointment, we found differences by insurance type similar to those identified in unadjusted analysis (Table 2). Female callers were 27.3 percentage points less likely to receive a referral. Others less likely to receive referrals were individuals ≤ 35 years of age, callers with the hypertension scenario, and callers to offices in lower-income counties.

Table 4 Correlates of Being Referred to Other Sources (Insurance Company or Billing Office) Among Individuals Unable to Obtain Price Information
Full size table

Variation in Prices

Among 4469 callers able to obtain a price, virtually no uninsured callers (0.1%) were told their visit would be free, compared to 18.3% of ESI callers and 12% of Marketplace callers. Adjusted differences by insurance were similar (Table 5). Compared to callers with the checkup scenario, women (−3.9 percentage points) and those with the hypertension scenario (−5.9 percentage points) were less likely to be told that their visit would be free. Callers to offices in low-income communities were also 6 percentage points less likely than those in moderate-income communities to be told their visit would be free, and callers in Massachusetts were 8.1 percentage points more likely.

Table 5 Correlates of Visit Prices Among Individuals Able to Obtain Price Information
Full size table

The average (non-free) price quoted was $165 for ESI, $157 for uninsured, and $161 for Marketplace, with standard deviations of $84, $85, and $92, respectively (Table 2). Regression-adjusted prices are shown in Table 5. Women were quoted prices $18 lower on average. Callers with the hypertension scenario received a price that was an average of $9 lower. Across caller groups, prices at FQHCs were $57 lower on average. In a sensitivity analysis, the lower price at FQHCs applied mainly to uninsured callers. Prices were lower from family physicians ($14) compared to internists. Solo physicians and offices with 2–3 physicians charged less ($28 and $14, respectively) than offices with ≥4 physicians. Prices were also lower in low-income counties and counties with high uninsured rates. Compared to Texas, mean prices were more than $30 higher in Massachusetts, Montana, and Oregon.

DISCUSSION

In 2014, price information at the time a visit was scheduled was able to be obtained more often by uninsured (89%) than individuals with Marketplace (47%) or ESI (62%). Providing prices to the insured may be more difficult for schedulers because offices do not typically have a single standard rate that applies to all plans, but rather negotiated rates for different plans. Lack of rate information may have contributed to lower price availability for Marketplace enrollees than those with ESI, because Marketplace plans were newer and more unfamiliar in 2014.17 Offices may have also been less willing to work with Marketplace enrollees due to perceived hassle. Future research will examine whether these trends changed as offices gained more exposure to Marketplace plans. Prices were also more difficult to obtain in calls to FQHCs than to other settings.

Compared to the uninsured, callers with ESI and Marketplace insurance were far more likely to be told to contact a billing office or insurer. While pursuing these sources could ultimately provide price information for a large majority of patients, having to call plans or billing personnel creates a barrier to comparison shopping and may ultimately discourage price-sensitive patients from seeking needed care.

When prices were provided, a small segment of insured patients—more commonly those with ESI than Marketplace insurance—were told that their visits would be free. Since 2010, the ACA has required most insurers to cover specified preventive services without cost-sharing, regardless of whether the patient has met the annual deductible. Patients seeking checkups were more commonly told the visit would be free than those seeking evaluation for hypertension (when not free, prices were lower for the hypertension scenario). Although hypertension screening is a covered preventive benefit under the ACA, the physician consultation for hypertension is not. It is unclear why more callers requesting a checkup were not told that their visit would be free. Nor is it clear why there were differences in this regard between Marketplace and employer-sponsored plans, as both are subject to ACA requirements. Schedulers may have lacked knowledge of the benefit or been uncertain of whether a visit would qualify.

We hypothesized that prices would be similar by insurance status. In fact, the uninsured almost never received free visits, but prices quoted to the uninsured were significantly lower than those offered to callers with employer plans when not zero. FQHCs also offered substantially lower prices to uninsured callers.

We found that female and male callers received very different price information. Women were more likely to obtain price information than men. Women received fewer free visits, but non-free visits were less expensive. It is possible that price information is provided differently to female callers, either because of how insurance coverage benefits apply to women (e.g., well women visits) or due to discriminatory scheduling practices. However, our study was not designed to test the impact of gender across offices. Callers may have also differed in their communication styles. Further research with a paired tester design would be required to experimentally examine gender differences.

FQHCs offered lower-cost visits, especially to the uninsured, but provided prices much less frequently than other offices. The fact that callers did not provide income information likely reduced the ability of FQHCs to provide accurate price information. FQHCs apply sliding-scale fees that require documentation to be provided before a discount can be offered; thus the audit likely represents only the upper range of FQHC prices. Yet actual patients calling FQHCs may not know how to estimate their household income, a complex calculation involving decisions regarding which family members to include, which jobs, and what time frame. According to our findings, some low-income callers seeking appointments at FQHCs may falsely assume that they would not be able to afford care. Moreover, current guidelines for notifying patients of sliding-fee discount programs at FQHCs focus on website and front desk notices, but do not address providing information to patients seeking price information by phone, as such interactions often take place in face-to-face encounters.18

We also identified important sources of geographic variation. While county-level income and uninsured rates generally had only modest associations with price information and prices, lower-income areas were less likely to provide price information but more likely to offer slightly lower average prices. The ability to obtain a price was substantially lower in Massachusetts, a state that implemented a comprehensive price transparency law in 2014.3 It is likely that this new law was not fully implemented during this study period; our results can therefore provide a baseline for tracking the law’s future impact.

Several limitations should be noted. First, the study was designed to capture only the prices of physician consultations, but patients also receive other services during visits such as laboratory tests or biopsies that may add to out-of-pocket costs and affect affordability if a patient is uninsured or if deductibles have not been met. Second, although the audit mimics a realistic scenario, actual patients may be able to provide additional information to schedulers that could enable them to obtain a price over the phone (e.g., an insurance number in order to query an electronic database, or household income to determine prices under a sliding-scale policy). Third, we cannot confirm that prices quoted to patients in fact match the negotiated rates that are paid by specific insurers in specific markets (e.g., Oregon, where prices were higher). To address this limitation, we compared our results with commercial prices at the level of larger geographic areas to confirm that prices approximated regional norms (Online Appendix Figure 2). The comparison to commercial prices indicates that the quoted prices may be somewhat low; despite asking for the full price of basic visits, it is possible that some audit callers may have been inadvertently quoted copayment amounts rather than the full rate applicable to patients who have not met their deductibles. More broadly, the audit does not mimic the diversity of benefit designs available to real patients (e.g., plans exempting primary care from the deductible). Fourth, the audit is able to capture prices offered to different groups, but may not reflect actual prices paid by patients in safety-net settings. For example, FQHCs account for a larger proportion of visits by the uninsured than their proportion in the audit study.

CONCLUSIONS

Information about health care prices is necessary for individuals to become engaged consumers. Efforts to provide greater price transparency are still evolving, but our results suggest room for improvement. Publicly available databases allowing individuals to obtain prices for procedures in their area could facilitate shopping for care,19 as could databases maintained by insurers for their members. Price transparency laws that require providers to disclose the price of care (including out-of-network fees) before a service is provided could increase pressure on providers to be more forthcoming about their prices, but these laws also create new burdens for physician offices. Current state price transparency laws are limited and generally have not applied to outpatient settings.20 Managing tradeoffs among these new tools is an important policy challenge.