Out-of-Network Laboratory Test Spending, Utilization, and Prices in the US

Abstract

In the context of concerns about the appropriateness and costs to insurers and patients of laboratory testing, this study uses data from the IBM MarketScan Commercial Claims and Encounters Database to examine out-of-network laboratory test spending, utilization, and prices in the US in 2008-2016.

More than 12 billion medical laboratory tests are analyzed in the US annually, making them the highest-volume health care service nationwide.¹ As attention toward out-of-network billing and out-of-network laboratories grows, data on out-of-network laboratory testing remain scarce.²

Such data are important because patients may unknowingly have their tests analyzed by out-of-network laboratories, leading to higher costs for insurers and patients. Moreover, the billing practices of some independent laboratories have raised concerns about the appropriateness and costs of laboratory testing, notably in the domain of addiction medicine.^3,4 We examined out-of-network laboratory test spending, quantity, and prices per test in the US.

Methods

We studied data for individuals enrolled for at least 1 year during 2008-2016 in the IBM MarketScan Commercial Claims and Encounters Database and whose employer consistently contributed data throughout this period. The overall database contains more than 43 million covered individuals from mostly large employers nationwide.⁵ Our continuously enrolled subsample of employers eliminated changes in composition due to the entry and exit of employers. We categorized outpatient laboratory tests into blood counts, chemistries, microbiology, pathology, toxicology, and urine tests (eTable in the Supplement). A unique flag indicated whether each test was in network or out of network.

We calculated mean spending (including cost sharing, adjusted for inflation), quantity, and prices of laboratory tests per enrollee in 3-year groups. Prices were actual paid amounts per test, including the insurer portion and patient cost sharing (sum of any co-payment, coinsurance, and deductible). Spending was the product of prices and quantities. We estimated mean annual changes in the shares of spending and shares of quantity that were out of network using generalized linear models with a γ distribution and log-link function, adjusted for age, sex, diagnostic cost-group risk score (mean, 1.0; range, typically 0 to about 60), and insurance plan fixed effects and weighted by an individual’s annual spending in the category. Standard errors were clustered at the plan level. The Harvard Medical School Institutional Review Board approved this study with a waiver of informed consent. Analyses were conducted using Stata version 15 (StataCorp). Statistical significance using 2-tailed tests was defined as P < .05.

Results

The population comprised 27 833 040 individuals (mean age, 33.8 years; 52% female; mean diagnostic cost-group risk score, 1.0). In aggregate, the share of laboratory test spending that was out of network increased from 5.2% during 2008-2010 to 11.5% during 2014-2016 (Figure), an adjusted annual growth of 0.98 (95% CI, 0.82-1.15) percentage points per year (P < .001). This was driven primarily by toxicology, for which the share of spending out of network increased 6.92 percentage points per year (95% CI, 6.03-7.80; P < .001), from 11.9% in 2008-2010 to 48.2% in 2014-2016 (Table).

Figure. Share of Spending on Out-of-Network Laboratory Tests by Test Category.

Related tests were considered together when possible and specified 6 distinct categories: blood counts, chemistries, microbiology, pathology, toxicology, and urine tests. The full sample comprised 59 874 449 person-years of observations.

Table. Total and Out-of-Network Spending and In-Network and Out-of-Network Utilization and Prices for Laboratory Tests^a.

Laboratory tests	2008-2010		2011-2013		2014-2016		Mean annual change in out-of-network percentage points (95% CI)	P value
Spending on tests per enrollee per year
	Annual spending, mean $	Out of network, %	Annual spending, mean $	Out of network, %	Annual spending, mean $	Out of network, %
All laboratory tests	178	5.2	208	7.7	240	11.5	0.98 (0.82 to 1.15)	<.001
Blood counts	10	4.0	11	3.2	3	2.6	–0.25 (–0.32 to –0.18)	<.001
Chemistries	78	5.9	88	7.1	91	7.2	0.16 (0.04 to 0.28)	.007
Microbiology	27	4.3	32	5.6	36	6.3	0.30 (0.19 to 0.41)	<.001
Pathology	55	4.7	64	8.2	72	11.2	1.03 (0.83 to 1.23)	<.001
Toxicology	3	11.9	7	28.9	21	48.2	6.92 (6.03 to 7.80)	<.001
Urine tests	6	4.2	6	3.7	7	3.1	–0.22 (–0.30 to –0.13)	<.001
Utilization, mean No. of tests per 1000 enrollees per year
	In network	Out of network	In network	Out of network	In network	Out of network
All laboratory tests	1794	55	1978	88	2129	139	0.49 (0.40 to 0.58)	<.001
Blood counts	188	5	195	5	212	6	0.02 (–0.05 to 0.08)	.58
Chemistries	924	30	1002	43	1048	54	0.29 (0.21 to 0.38)	<.001
Microbiology	289	8	336	11	382	16	0.22 (0.14 to 0.30)	<.001
Pathology	173	6	183	9	183	12	0.47 (0.36 to 0.58)	<.001
Toxicology	38	3	63	15	93	46	5.81 (5.19 to 6.43)	<.001
Urine tests	183	4	199	5	210	6	0.05 (–0.01 to 0.10)	.10
Price per test, mean $
	In network	Out of network	In network	Out of network	In network	Out of network
Blood counts	16	25	17	22	19	19	NA	NA
Chemistries	26	49	27	48	27	39	NA	NA
Microbiology	29	49	29	52	29	48	NA	NA
Pathology	96	147	104	190	116	230	NA	NA
Toxicology	23	46	25	49	40	88	NA	NA
Urine tests	9	17	10	15	11	13	NA	NA

Abbreviation: NA, not applicable.

^aEstimates of the mean annual change in shares of spending out of network and utilization out of network were derived using generalized linear models with a γ distribution and log-link function, adjusted for age categories, age categories interacted with sex, risk score, and plan fixed effects, with standard errors clustered by plan. Each individual’s risk score in each year was derived using the concurrent diagnostic cost-group model, which uses concurrent demographics and International Classification of Diseases diagnoses to calculate a score that reflects the disease burden or expected health care costs of an individual relative to the population, with a range between 0 and about 60 and average risk reflected by a score of 1. Spending was adjusted for inflation using data from the Bureau of Labor Statistics. The full sample comprised 59 874 449 person-years of observations.

The number of in-network laboratory tests increased 2.3% per year (from 1794 per 1000 enrollees per year in 2008-2010 to 2129 per 1000 enrollees per year in 2014-2016), whereas out-of-network laboratory tests increased by 18.9% per year (from 55 per 1000 enrollees per year in 2008-2010 to 139 per 1000 enrollees per year in 2014-2016). This was similarly driven primarily by toxicology, which increased from 37.9 per 1000 enrollees to 93.2 per 1000 enrollees in network and from 2.6 per 1000 enrollees to 45.6 per 1000 enrollees out of network. Out-of-network prices per test exceeded in-network prices (Table).

Discussion

This study found an increasing share of laboratory spending that was out of network, with both increased utilization and higher prices of out-of-network laboratory tests relative to in-network tests, especially for toxicology tests.

Toxicology tests are frequently ordered for patients with substance use disorders, with treatment programs potentially playing a larger role.⁶ Laboratory services may be increasingly contracted to large suppliers, who may have sufficient market power to set high prices outside of insurer networks. Moreover, clinician discretion in test ordering and the lack of guidelines for many tests may render utilization more susceptible to financial incentives. These factors could help explain the increasing prices of toxicology tests and their out-of-network share.

Limitations of the study include the lack of generalizability of the employer-sponsored insurance context to individuals with other forms of coverage and the lack of further information on the laboratories. Moreover, claims may not capture the full cost of tests for patients who received balance bills (including surprise bills) in excess of the claim amounts.

Section Editor: Jody W. Zylke, MD, Deputy Editor.

Supplement.

eTable. Common Procedural Terminology Codes