Out-of-network billing rates represent a frequency-weighted mean across 7 procedures, chosen to represent a diverse range of specialties and both ambulatory and inpatient settings: arthroscopic meniscal repair, laparoscopic cholecystectomy, hysterectomy, total knee replacement, breast lumpectomy, colectomy, and coronary artery bypass graft surgery. Episodes were defined as beginning on the day of surgery and ending at the end of the surgical confinement. These out-of-network billing rates were conditional on the primary surgeon and facility being in-network. Bills came primarily from surgical assistants, anesthesiologists, pathologists, medical consultants, and radiologists.
eFigure. Cohort Selection Flow Diagram
eTable 1. Procedure Codes
eTable 2. Baseline Characteristics
eTable 3. Top Five Sources of Surgical Assistant Out-of-Network Bills, by Procedure
eTable 4. Top Five Sources of “Other” Out-of-Network Bills, by Procedure
eTable 5. Risk Factors for Out-of-Network Bills, Stratified by Specialty
eTable 6. Frequency of Common Sources of Clinician Charges
Customize your JAMA Network experience by selecting one or more topics from the list below.
Chhabra KR, Sheetz KH, Nuliyalu U, Dekhne MS, Ryan AM, Dimick JB. Out-of-Network Bills for Privately Insured Patients Undergoing Elective Surgery With In-Network Primary Surgeons and Facilities. JAMA. 2020;323(6):538–547. doi:10.1001/jama.2019.21463
How often are patients undergoing elective surgery with in-network primary surgeons at in-network facilities at risk for receiving out-of-network bills?
In this retrospective analysis of 347 356 surgical episodes among commercially insured patients who had undergone elective surgery with in-network primary surgeons and facilities, 20% of episodes involved out-of-network charges.
Patients undergoing elective surgery with in-network primary surgeons and facilities may be at risk of receiving out-of-network bills.
Privately insured patients who receive care from in-network physicians may receive unexpected out-of-network bills (“surprise bills”) from out-of-network clinicians they did not choose. In elective surgery, this can occur if patients choose in-network surgeons and hospitals but receive out-of-network bills from other involved clinicians.
To evaluate out-of-network billing across common elective operations performed with in-network primary surgeons and facilities.
Design, Setting, and Participants
Retrospective analysis of claims data from a large US commercial insurer, representing 347 356 patients who had undergone 1 of 7 common elective operations (arthroscopic meniscal repair [116 749]; laparoscopic cholecystectomy [82 372]; hysterectomy [67 452]; total knee replacement [42 313]; breast lumpectomy [18 018]; colectomy [14 074]; coronary artery bypass graft surgery ) by an in-network primary surgeon at an in-network facility between January 1, 2012, and September 30, 2017. Follow-up ended November 8, 2017.
Patient, clinician, and insurance factors potentially related to out-of-network bills.
Main Outcomes and Measures
The primary outcome was the proportion of episodes with out-of-network bills. The secondary outcome was the estimated potential balance bill associated with out-of-network bills from each surgical procedure, calculated as total out-of-network charges less the typical in-network price for the same service.
Among 347 356 patients (mean age, 48 [SD, 11] years; 66% women) who underwent surgery with in-network primary surgeons and facilities, 20.5% of episodes (95% CI, 19.4%-21.7%) had an out-of-network bill. In these episodes, the mean potential balance bill per episode was $2011 (95% CI, $1866-$2157) when present. Out-of-network bills were associated with surgical assistants in 37% of these episodes; when present, the mean potential balance bill was $3633 (95% CI, $3384-$3883). Out-of-network bills were associated with anesthesiologists in 37% of episodes; when present, the mean potential balance bill was $1219 (95% CI, $1049-$1388). Membership in health insurance exchange plans, compared with nonexchange plans, was associated with a significantly higher risk of out-of-network bills (27% vs 20%, respectively; risk difference, 6% [95% CI, 3.9%-8.9%]; P < .001). Surgical complications were associated with a significantly higher risk of out-of-network bills, compared with episodes with no complications (28% vs 20%, respectively; risk difference, 7% [95% CI, 5.8%-8.8%]; P < .001). Among 83 021 procedures performed at ambulatory surgery centers with in-network primary surgeons, 6.7% (95% CI, 5.8%-7.7%) included an out-of-network facility bill and 17.2% (95% CI, 15.7%-18.8%) included an out-of-network professional bill.
Conclusions and Relevance
In this retrospective analysis of commercially insured patients who had undergone elective surgery at in-network facilities with in-network primary surgeons, a substantial proportion of operations were associated with out-of-network bills.
In 2019, lawmakers in both the Senate and House of Representatives proposed multiple bills to end unexpected (ie, “surprise”) medical billing.1,2 These bills occur when patients obtain care from physicians and hospitals that participate in their insurance network (“in network”) but receive separate and often unexpected bills from a clinician who does not participate in their insurance network (“out of network”). Out-of-network clinicians charge patients what they believe they should be paid and may be reimbursed for the entire amount charged, a fraction of it, or none at all. “Balance billing” occurs when out-of-network clinicians are paid less by the insurance plan than they charged and send the patient a “balance bill” for the difference between their charges and the plan payment. Balance bills from out-of-network clinicians can amount to thousands of dollars, without the usual limits on out-of-pocket obligations (eg, deductibles, out-of-pocket maximums).3-6 Out-of-network billing can also increase health care costs for patients who do not receive balance bills, as many insurance plans require higher cost-sharing (deductibles, co-insurance, co-pays) for out-of-network care, and the option for clinicians to go out of network may increase their ability to negotiate higher in-network prices.7,8
Most prior studies on unexpected out-of-network billing have involved emergency department care or general inpatient care.4,8-12 Unlike those settings, in elective surgery, patients can usually choose in-network surgeons and facilities. However, they can rarely choose other involved clinicians (eg, assistants, anesthesiologists, pathologists). The extent to which each of these clinical specialties accounts for out-of-network bills is unclear.
To better understand the factors associated with out-of-network billing in surgical episodes, this study evaluated out-of-network billing for elective surgical episodes using claims data from a large commercial insurer. This study also examined variation in the rates of out-of-network billing associated with patient, physician, and insurance characteristics, and across geographic regions in the United States.
The data source for this study was the Clinformatics DataMart (Eden Prairie, Minnesota), which contains insurance claims from a large national insurer. The study was deemed “not regulated” by the University of Michigan Institutional Review Board because it represented a secondary analysis of deidentified data.
The study cohort consisted of commercially insured patients aged 18 to 64 years who had undergone 1 of 7 operations between 2012 and 2017. The operations were chosen to represent some of the most common surgical procedures undergone by commercially insured patients,13 as well as a diverse range of acuity levels, surgical subspecialties, and surgical settings: laparoscopic cholecystectomy, total knee replacement, arthroscopic meniscal repair, coronary artery bypass graft (CABG) surgery, colectomy, hysterectomy, and breast lumpectomy (see eAppendix eTable 1 in the Supplement for procedure codes).
We included only elective cases, as defined in the database. We restricted the analysis to cases in which the facility and primary surgeon were in-network. We identified primary surgeons as the surgeons whose Current Procedural Terminology code matched the principal procedure and who did not submit a modifier code identifying themselves as assistants or co-surgeons. Surgical assistants included those with assistant surgeon or co-surgeon modifier codes, as well as other clinicians providing a surgical type of service any time during the hospitalization. We excluded cases with missing facility identifiers or missing patient-level information. (See the eAppendix in the Supplement for a flow diagram, full description of exclusion criteria, and details on how primary surgeons and other clinician specialties were identified.)
We identified Elixhauser comorbidities based on International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) and International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Clinical Modification (ICD-10-CM) diagnosis codes documented in the 3 months before the surgical episode.14 We also used ICD-9-CM and ICD-10-CM codes to identify postoperative complications documented in the inpatient setting based on the diagnosis codes previously described to have the greatest sensitivity and specificity.15
We identified out-of-network claims using the “participating provider” variable in the data set, which indicated whether the clinician or facility associated with a claim has a contract with the payer to provide services at negotiated payment rates (ie, is “in-network”).
Because patients seeking elective surgery can generally choose in-network hospitals and surgeons, we specifically identified out-of-network bills in which the facility and primary surgeon were in-network, but other clinicians were out-of-network. The primary outcome was the proportion of elective surgical episodes involving an out-of-network bill.
The secondary outcome was the estimated potential balance bill, which was calculated for each episode as the total of out-of-network charges less the standardized payment (ie, typical in-network price) for each out-of-network charge, in line with prior research.12 The potential balance bill excludes expected sources of cost-sharing, such as deductibles and co-insurance. We also reported total out-of-network charges, total episode charges, standardized payments, and out-of-pocket cost-sharing. All charges and payments were winsorized to the 1st and 99th percentiles of charges and payments within each procedure.
We calculated the percentage of out-of-network bills for each procedure by dividing the number of episodes with any out-of-network bill by the total number of episodes of each procedure. We calculated the percentage of out-of-network bills from each specialty by dividing the number of episodes with an out-of-network bill from that specialty by the number of episodes with any out-of-network bill. Episodes with missing data were excluded (see eFigure in the Supplement).
To assess factors associated with out-of-network bills, we constructed a logistic regression at the surgical episode level, with the presence of an out-of-network bill as the outcome variable. Explanatory variables included patient factors such as age, sex, comorbidities (modeled with indicators for 0, 1, 2, or ≥3 comorbidities), and surgical complications (modeled with indicators for 0, 1, 2, or ≥3 complications); location of service (hospital inpatient, hospital outpatient, or ambulatory surgery center [ASC]); insurance plan characteristics (self-insured or fully insured; health insurance exchange vs nonexchange), procedure, geographic region, and year. Model fit was assessed using receiver operating characteristic curves. We estimated adjusted risk differences by calculating population-averaged marginal effects at each value of the explanatory variables. Because of the potential for type I error due to multiple comparisons, the findings of this secondary analysis should be interpreted as exploratory.
Since prior research has focused on out-of-network billing in the inpatient setting, we also separately calculated the proportion of episodes with in-network primary surgeons involving out-of-network bills at ASCs. To specifically measure the proportion of episodes at ASCs with out-of-network facility fees, having an in-network facility was not an inclusion criterion for this analysis.
All statistical testing was conducted at the 5% significance level with 2-sided tests, accounting for clustering at the facility level. SAS version 9.4 (SAS Institute Inc) was used for cohort preparation and descriptive analyses; Stata version 14.2 (StataCorp) was used for regression models.
Among 347 356 patients who had undergone elective surgery with in-network primary surgeons at in-network facilities, the mean age was 48 (SD, 11) years; 66% were women; 116 749 (34%) underwent arthroscopic meniscal repair; 82 372 (24%) underwent laparoscopic cholecystectomy; 67 452 (19%) underwent hysterectomy; 42 313 (12%) underwent total knee replacement; 18 018 (5%) underwent breast lumpectomy; 14 074 (4%) underwent colectomy; and 6378 (2%) underwent CABG surgery (eTable 2 in the Supplement). After applying exclusion criteria (eFigure in the Supplement), there were no missing data.
The prevalence of out-of-network bills varied by state, from 3% (Nebraska) to 46% (Alaska) (Figure). High prevalences were common in parts of the South and Northwest, whereas low prevalences were common in the Midwest.
Among 347 356 identified episodes, an out-of-network bill was present in 20.5% of episodes (95% CI, 19.4% to 21.7%) when the primary surgeon and facility were in-network (Table 1).
The presence of an out-of-network bill was associated with significantly higher total charges ($48 383 vs $34 300; difference, $14 083 [95% CI, $12 883 to $15 281]), standardized payments ($21 656 vs $16 200; difference, $5456 [95% CI, $5029 to $5886]), and out-of-pocket cost-sharing ($1768 vs $1444; difference, $324 [95% CI, $292 to $355]) relative to cases with all bills in-network (P < .001 for all). The secondary outcome of potential balance bills (difference between charges and standardized payments) ranged from $1255 (95% CI, $1078 to $1432) for laparoscopic cholecystectomy to $3449 (95% CI, $2947 to $3950) for CABG surgery.
Of 71 228 episodes with out-of-network bills, anesthesiologists and surgical assistants were both associated with out-of-network claims in 37% of episodes (Table 2). The mean potential balance bill from surgical assistants was $3633 (95% CI, $3384 to $3883) and from anesthesiologists was $1219 (95% CI, $1049 to $1388). Out-of-network surgical assistants were most commonly physician assistants (26% of all out-of-network surgical assistant claims), registered nurses (24%), certified surgical assistants (19%), and other surgeons (6%-27%, depending on the procedure) (eTable 3 in the Supplement). In procedures commonly requiring surgical pathology review (colectomy, hysterectomy, cholecystectomy, breast lumpectomy), the proportion of out-of-network bills involving pathologists ranged from 25% to 43% of episodes.
When potential balance bills were present, the mean charges for medical consultants were $708 (95% CI, $599 to $816); for pathologists, $284 (95% CI, $257 to $311); for radiologists, $321 (95% CI, $103 to $539); and for other clinicians, $754 (95% CI, $673 to $835). The most common clinician types categorized as “other” are listed in eTable 4 in the Supplement.
In a separate analysis of 83 021 operations performed at ASCs, 6.7% (95% CI, 5.8% to 7.7%) of episodes involved out-of-network facility bills, and 17.2% (95% CI, 15.7% to 18.8%) involved out-of-network professional bills (Table 3). Although fewer episodes involved out-of-network facility bills, the potential balance bills from out-of-network facility charges were significantly higher than those from out-of-network professional charges ($12 624 vs $6872, respectively; difference, $5752 [95% CI, $4849 to $6656]; P < .001).
In multivariable analysis (Table 4), self-insured plan membership was associated with a significantly lower risk of out-of-network bills, compared with fully insured plan membership (19% vs 22%, respectively; risk difference, –3% [95% CI, –3.3% to –1.9%], P < .001). Exchange plan membership was associated with a significantly higher risk when compared with nonexchange plans (27% vs 20%, respectively; risk difference, 6% [95% CI, 3.9% to 8.9%]; P < .001). The occurrence of surgical complications was also associated with a significantly higher risk of out-of-network bills, compared with episodes with no complications (28% vs 20%, respectively; risk difference, 7% [95% CI, 5.8% to 8.8%]; P < .001). Complications were significantly associated with out-of-network bills from radiologists, medical consultants, and “other” clinicians (eTable 5 in the Supplement). Surgery at an ASC or in the hospital outpatient setting was associated with a significantly lower risk of out-of-network bills (20% and 18%, respectively) relative to the hospital inpatient setting (24%, P < .001 for both). However, the difference between ASCs and the hospital outpatient setting was not statistically significant (difference, 1.1%; [95% CI, –0.9% to 3.1%]; P = .26). The prevalence of out-of-network bills was not significantly different between 2012 and 2017.
In this analysis of commercially insured patients who had undergone elective surgery with an in-network surgeon at an in-network facility, approximately 1 in 5 received an out-of-network bill, with a mean potential balance bill of $2011.
The patterns of out-of-network bills varied with the clinical scenario. Simpler ambulatory procedures that tend to involve 1 surgeon (arthroscopic meniscal repair, breast lumpectomy) had fewer out-of-network bills (13%-15% of cases), whereas inpatient procedures (hysterectomy, knee replacement, colectomy, CABG surgery) had more frequent out-of-network bills (24%-33% of cases). These more complex procedures were also associated with larger potential balance bills, in the range of $2000 to $4000.
Prior studies have estimated that the rate of out-of-network bills in general inpatient care is between 10% and 40%.10,12 By focusing on a granular set of elective surgical procedures, the current study was able to more precisely estimate the risk of an out-of-network bill as approximately 20%. Prior estimates have also suggested that anesthesiology was involved in 8% to 22% of out-of-network bills in general inpatient admissions overall,7,10,12,16 but the current study found that in surgical care, anesthesiology was involved in 37%. This study also found that episodes with out-of-network bills were associated with significantly higher overall charges, payments, and cost sharing than episodes without any out-of-network bills.
Although anesthesiologists are often cited as the most common source of out-of-network bills in surgery, the current analysis found that out-of-network bills from surgical assistants had a similar frequency (37% of episodes) and were significantly larger than those from anesthesiologists ($3633 vs $1219). Skilled assistance is often required in surgical care for a variety of reasons,17,18 and in the complex inpatient procedures analyzed in the current study, claims from assistants were present in the majority of episodes. Assistants may participate intraoperatively or perioperatively and may manage the care of admitted patients on behalf of the primary surgeon. At teaching hospitals, trainees often provide surgical assistance. However, when qualified trainees are unavailable, surgeons may rely on other surgeons or allied professionals (eg, nurses, physician assistants, certified surgical assistants) to assist, and these clinicians may bill the patient separately from the primary surgeon. As surgeons are often able to choose their assistants, choosing assistants who will be in-network would seem a logical opportunity for reducing financial risk to patients. However, there have been reports of surgeons billing as out-of-network assistants as an intentional billing tactic.19
In addition, this analysis found that the risk of out-of-network bills was not significantly different for patients who had undergone the same operation in an ASC vs in a hospital as an outpatient, if both the facility and primary surgeon were in-network. The subanalysis of cases at ASCs with in-network primary surgeons found that out-of-network facility claims were less frequent compared with out-of-network professional claims (6.7% of episodes vs 17.2% of episodes, respectively), even though ASCs are less tightly regulated than hospitals.20 However, out-of-network facility charges were significantly larger than out-of-network professional charges and thus may still be driving up costs for insurance plans and patients.
This analysis also examined out-of-network bills in several policy-relevant subgroups. First, self-insured plans, in which the employer takes on full responsibility for members’ health care expenses and the insurer provides administrative services (eg, billing, reimbursement) only, cover the majority of US residents with employer-sponsored insurance.21 The Employee Retirement Income Security Act (ERISA) preempts states from regulating these plans.22-24 As such, states are thought to be unable to protect self-insured patients from unexpected out-of-network bills. This analysis found that the risk of out-of-network bills was slightly lower in self-insured plans, perhaps because large self-insured employers may offer plans with broader clinician networks. However, self-insured plan members remain a large share of the commercially insured who are not being protected by current state policy.
Second, health insurance exchange plans are commonly purchased by individuals not offered employer-sponsored insurance. Exchange plans are known to have narrower networks than employer-sponsored plans (in exchange for lower premiums).25-27 This analysis found that members of these plans have a 6–percentage-point higher risk of out-of-network bills than members of nonexchange plans.
Third, this analysis also found that the prevalence of out-of-network bills varied widely across states. Various states (CA, CT, FL, IL, MA, MD, MS, NY) have enacted policies that protect patients from out-of-network billing at in-network facilities22,23; however, in this analysis, those states all had proportions of out-of-network bills above the national median. This suggests several hypotheses, including that these states may have been responding to endemic surprise billing problems, and that these policies may not have been completely effective. Although a formal evaluation of state policy was outside the scope of this analysis, future studies will need to account for this endogeneity and for variation in state policy designs.
Fourth, these findings suggest that, in surgical settings, the problem of out-of-network billing is not restricted to a single specialty or setting. Surgical care is inherently multidisciplinary and requires a team of clinicians with payer contracts that are rarely intentionally coordinated. This is in contrast to prior research on emergency care, which pointed toward a subset of emergency departments staffed by private equity–backed medical groups as the primary source of out-of-network billing.9
This study has several limitations. First, the Clinformatics database only contains claims from 1 insurer. However, this insurer is present in all 50 states, in the individual, small-group, and large-group markets, and offers both self- and fully-insured plans, thereby reflecting the diversity of commercial insurance plans available today. Regarding the prevalence of out-of-network bills, previous studies of out-of-network billing involving a single insurer4,8,9,12 were consistent with studies of multipayer data sets.7,10
Second, claims data do not indicate whether patients were balance-billed, nor whether they were aware of their out-of-network status. This limitation is common to most studies of out-of-network medical billing. However, prior analyses indicate that the majority of out-of-network contacts at inpatient facilities are involuntary.11
Third, the data set used in this study did not indicate how much the insurance plan paid for out-of-network care. The payments in the data set were adjusted to the typical in-network price for each service across the United States. Thus, this study’s calculations of potential balance bills reflected an estimate of the dollar amount patients could be balance-billed but not the actual amount that was balance-billed. If an insurance plan does not have out-of-network benefits, its members would be responsible for all out-of-network charges; the approach used in this study would underestimate the potential balance bill in this scenario. On the other hand, plans with out-of-network benefits often reimburse out-of-network clinicians at a “usual and customary rate,” which may exceed the typical in-network payment. In this case, the approach used in this study would overestimate the potential balance bill. Since the study data did not indicate whether patients’ insurance plans had out-of-network benefits, this analysis used methods previously published to estimate potential financial liability within the same data set.4,12 This is likely a conservative estimate because it excludes expected sources of cost-sharing such as deductibles, co-insurance, and co-pays.
Fourth, many factors influencing out-of-network billing cannot be observed using claims data. For instance, it is well-known that insurance network directories are sometimes inaccurate,27-29 but their role could not be measured in this study.
In this retrospective analysis of commercially insured patients who had undergone elective surgery at in-network facilities with in-network primary surgeons, a substantial proportion of operations were associated with out-of-network bills.
Corresponding Author: Karan R. Chhabra, MD, MSc, 2800 Plymouth Rd, Bldg 14, Room G100, Ann Arbor, MI 48109 (firstname.lastname@example.org).
Accepted for Publication: January 1, 2020.
Author Contributions: Dr Chhabra had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Chhabra, Sheetz, Ryan, Dimick.
Acquisition, analysis, or interpretation of data: Chhabra, Sheetz, Nuliyalu, Dekhne, Dimick.
Drafting of the manuscript: Chhabra, Sheetz, Dekhne, Dimick.
Critical revision of the manuscript for important intellectual content: Chhabra, Nuliyalu, Dekhne, Ryan, Dimick.
Statistical analysis: Chhabra, Sheetz, Nuliyalu, Ryan, Dimick.
Obtained funding: Chhabra, Dimick.
Administrative, technical, or material support: Ryan, Dimick.
Conflict of Interest Disclosures: Dr Dimick reported that he is a cofounder of ArborMetrix Inc, a company that makes software for profiling hospital quality and efficiency. No other authors reported disclosures.
Funding/Support: The University of Michigan Institute for Healthcare Policy and Innovation Policy Sprints Program supported this work. Dr Chhabra was supported by the Institute for Healthcare Policy and Innovation Clinician Scholars Program, Agency for Healthcare Research and Quality (AHRQ) T32HS000053, and the National Institutes of Health (NIH) Division of Loan Repayment. Dr Sheetz was supported by the Association for Academic Surgery Resident Research Fellowship (AHRQ 2T32HS000053-27). Dr Ryan was supported by NIH/National Institute on Aging (NIA) R01AG39434. Dr Dimick was supported by NIH/NIA R01AG039434 and AHRQ R01HS023597.
Role of the Funder/Sponsor: The University of Michigan Institute for Healthcare Policy and Innovation Policy Sprints Program provided input on study design but had no role in the conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; the decision to submit the manuscript for publication; or the decision as to which journal the manuscript was submitted. No other funding sources had a role in the study.
Meeting Presentation: This study was presented at the 2019 National Clinician Scholars Program Annual Meeting; November 13, 2019; Philadelphia, Pennsylvania.
Additional Contributions: We thank John Ayanian, MD, MPP, Eileen Kostanecki, MA, and Sarah Wang, MPH (University of Michigan Institute for Healthcare Policy and Innovation), for their comments on earlier versions of this article, for which they received no financial compensation. We also thank Jyothi Thumma, MPH (University of Michigan Center for Healthcare Outcomes and Policy), for assistance with data preparation and Ranganath Kathawate (University of Michigan) for research assistance, for which they received financial compensation.