The Association of Hospital Characteristics With Medicaid Managed Care Hospital Prices

ABSTRACT

Objective

To examine the association between various hospital characteristics and Medicaid facility prices for several routine procedures.

Study Setting and Design

We conducted a cross‐sectional study of facility claims data from Elevance Health‐affiliated Medicaid plans from 22 states plus D.C. We estimated Poisson regression models where the outcome was facility allowed cost (“price”) for procedures of interest, and the exposures were hospital characteristics, including children's hospital, critical access hospital (CAH), rurality, system affiliation, teaching designation, for‐profit, religious affiliation, government ownership, and hospital size. We report incidence rate ratios (IRRs) from the regression estimates.

Data Sources and Analytic Sample

We analyzed 2021–2023 facility claims for 11 routine procedures in the outpatient or emergency department (ED) settings, including tympanostomy, colonoscopy, and ED evaluation and management (E/M) Levels 2–5. The claims were joined with data from the American Hospital Directory to identify hospital characteristics. The study included 2,250,325 children and 2,341,063 adults who received at least one of the procedures of interest.

Principal Findings

The largest price differences were at children's hospitals and CAHs. For all child procedures, prices were higher at children's hospitals relative to other acute care hospitals: the largest estimate was for ED E/M Level 2 (IRR = 2.51; 95% CI, 2.14–2.95). Children's hospital estimates represented 93%–151% higher prices for ED procedures and 23%–48% higher prices for outpatient procedures. CAHs had higher prices for almost all procedures: for children the largest estimate was for tympanostomy (IRR = 1.78; 95% CI, 1.45–2.18), representing 78% higher prices, and for adults the largest estimate was for MRI (IRR = 2.43; 95% CI, 2.13–2.76), representing 143% higher prices. Results were mixed for other hospital characteristics.

Conclusions

This study suggests much higher Medicaid managed care facility prices at children's hospitals and CAHs for several routine procedures. More research is needed to understand what is driving the observed differences in hospital prices.

Callout Box

• What is known on this topic

  • ○
    The price of care for a given procedure varies substantially across hospitals.
  • ○
    Certain hospital characteristics, like teaching status and ownership type, may be associated with market power.
  • ○
    There is little research that: (1) combines many hospital characteristics into one study when analyzing prices; (2) analyzes hospital prices for a Medicaid managed care population; or (3) studies routine procedures in the hospital outpatient or emergency department settings.
• What this study found

  • ○
    This study examined how negotiated Medicaid managed care hospital prices for routine procedures in hospital outpatient or emergency department settings differ by various hospital characteristics.
  • ○
    The results of this study suggest much higher Medicaid managed care facility prices at children's hospitals and critical access hospitals for several routine procedures.

1. Introduction

The price of care for a given procedure varies substantially across hospitals [1, 2, 3, 4, 5]. Market dynamics may play a role in this price variation. Over the last 25 years, the consolidation of hospitals has reduced competition [6], and several studies show that hospital prices can increase following mergers [7, 8, 9, 10, 11, 12, 13]. Furthermore, some hospitals, such as teaching hospitals, often provide unique services, advanced care, or brand recognition that are not easily substituted; as such, they may wield substantial negotiating power during contract discussions with insurers [14]. It follows that certain hospital characteristics, like teaching status and ownership type, may be associated with market power [6].

There remains a notable gap in empirical research examining how hospital prices vary by hospital attributes. Some studies look at one or a few hospital characteristics, including children's hospitals [15, 16, 17], government ownership [7, 18, 19, 20], or size [7, 21, 22]. For selected procedures, the studies find that children's hospitals tend to have higher prices, government hospitals tend to have lower prices, and larger hospitals tend to have higher prices. However, there is little research that (1) combines many hospital characteristics into one study, allowing for the identification of unique associations when accounting for other aspects of the hospital; (2) analyzes hospital prices for a Medicaid managed care population, for which negotiated price information is hard to obtain; or (3) studies common, routine procedures in the hospital outpatient or emergency department (ED) settings.

The objective of this study is to examine how negotiated Medicaid managed care hospital prices for routine procedures in hospital outpatient or ED settings differ by the following hospital characteristics: children's hospital; critical access hospital (CAH) [23]; rurality; system affiliation; teaching designation; for‐profit, religious affiliation, or government ownership; and size.

States operate their own Medicaid programs, and many states contract with health insurance companies (or managed care organizations (MCOs)) to provide Medicaid health benefits. As of July 2025, 41 states plus D.C. had contracted with an MCO to deliver comprehensive Medicaid coverage to at least some of their Medicaid beneficiaries [24]. The managed care contracts covered a large proportion of the overall Medicaid population: as of July 2024, approximately 66.5 million Medicaid beneficiaries were enrolled in a comprehensive plan from an MCO, which was approximately 78% of all Medicaid beneficiaries [25].

When a state contracts with a Medicaid MCO, the state Medicaid agency makes capitated payments to the MCO [24]. Similar to commercial insurance, the MCO contracts with providers and bears financial risk for beneficiaries' healthcare spending. In the Medicaid capitation arrangement, any difference between the capitated payment and the MCO's payments for healthcare services is accrued to the MCO (whether as a profit or as a loss), so the capitated payment set by the state may influence overall provider contracted rates. Other differences in market forces for MCOs compared to the commercial insurance market could also influence market prices. For example, most states use state directed payments (SDPs) that require MCOs to increase provider reimbursement rates or set minimum rates for specified services, with the intent of increasing provider access [26]. Some SDPs are directed at specific providers, such as children's hospitals and safety net hospitals [27]. Additionally, federal financing policies, such as the Affordable Care Act's enhanced funding for Medicaid expansion as well as temporary increases in federal support during the COVID‐19 pandemic (followed by decreases during the unwinding since 2023) have affected state Medicaid program enrollment and funding [28, 29]. The evolving Medicaid funding landscape (including upcoming changes [30]) and the associated potential changes in insurer bargaining power from changes in Medicaid enrollment—while not explored in this paper—may shape provider reimbursement rates. Together, these features distinguish Medicaid managed care from commercial insurance and may contribute to differences in hospital reimbursement dynamics.

2. Methods

2.1. Data Sources and Claim Identification

This study used administrative claims data from 2021 to 2023 from Elevance Health‐affiliated Medicaid managed care plans in 22 states plus D.C. [31]. We included in‐network facility claims for children (aged 0–18) and adults (aged 19+); we limited the study to in‐network claims because they reflect a Medicaid plan's negotiated prices with hospitals. In terms of representativeness relative to the entire Medicaid MCO population, Elevance Health‐affiliated Medicaid plans comprised 11% of total Medicaid MCO enrollment in July 2022 [32]. The 22 states (plus D.C.) in the data are spread across eight of the nine census divisions (the exception being New England), with at least two states in each census division.

For children, we examined the following procedures: tonsillectomy and adenoidectomy, tympanostomy, splinting, and ED evaluation and management (E/M) Levels 2–5. For adults, we examined the following procedures: knee arthroscopy, colonoscopy, obstetric ultrasound, magnetic resonance imaging (MRI) of the brain, and ED E/M Levels 2–5. For each procedure, we only included claims in either the hospital outpatient department (HOPD) setting or the ED setting. eTable 1 lists the Current Procedural Terminology (CPT) codes used for each procedure as well as the required setting. We selected common, routine procedures for the study. Focusing on routine outpatient and ED procedures reduces the likelihood that observed differences in prices are due to patient or procedure complexity; routine procedures are also common in the claims data, which allows for a greater sample size for the analysis. We also included multiple procedures, rather than focusing on a single procedure, because higher prices at a given hospital type may be heterogeneous based on procedure. Inpatient claims were not included due to the complexities in billing for inpatient services that make direct comparisons more difficult, especially for facility claims. Lastly, we used all claims from any hospital; if a hospital only had claims for some of the procedures of interest, it was still included in the analysis. For example, if a hospital did not have any tympanostomy claims during the study period but claims for all other procedures, then that hospital would be excluded from the tympanostomy analysis but included in other procedures' analyses.

The final set of claims was linked with 2024 data from the American Hospital Directory [33] to obtain hospital characteristics. The American Hospital Directory data designates each hospital as one facility type: children's hospital, CAH, other short‐term acute care hospital, long term facility, psychiatric facility, rehabilitation facility, or “Other” facility; we included children's hospitals, CAHs, and other short‐term acute care hospitals in our analysis. For teaching status, hospitals were assigned as teaching or non‐teaching. For hospital ownership type, hospitals were assigned as for‐profit, church‐affiliated nonprofit (“religiously affiliated”), other nonprofit, and government. For system affiliation, hospitals were identified as either system‐affiliated or as independent. For bed count (size), the data includes total staffed beds.

2.2. Outcome and Exposure

The outcome of interest was the facility allowed cost (i.e., negotiated rate) for the procedure of interest; the allowed cost includes both the amount paid by the Medicaid plan and the amount paid by the beneficiary (if any). Allowed costs were inflation‐adjusted to 2023 dollars using the consumer price index for medical care [34]. The allowed cost was calculated at the claim line level, meaning for the specific procedure only and not for accompanying services. Focusing on the claim line only ensures an apples‐to‐apples comparison for each procedure's allowed costs, without needing to assess differences in ancillary services provided. For some claims, payments were rolled up across all claim lines into a single claim line (e.g., for a bundled claim); 14% of child claim lines were part of a rolled‐up claim, and 15% of adult claim lines were part of a rolled‐up claim. For these claims, an allocated allowed cost that covered the procedure of interest only was calculated (see eMethods for further details). After allocation, any procedures with zero dollars in allowed costs were dropped. Last, to address outliers, we capped allowed costs at the 99th percentile for each procedure. We were concerned that administrative data, while valuable for granularity of information, may have data errors. Further, even if the data did not contain many errors, it may contain unreasonably high values that are more anomalous than indicative of some sort of pattern. We chose the 99th percentile (instead of a lower percentile) to avoid ridding the data of meaningful variation that we should capture in the analysis.

The primary exposure variables were the various hospital characteristics of interest, including: children's hospital indicator (child claims only); CAH indicator; rurality indicator (hospitals were classified as rural if located outside a metropolitan statistical area); system affiliation indicator; teaching designation indicator; for‐profit indicator; religious affiliation indicator; government ownership indicator; indicator for having 1–99 staffed beds; and indicator for having 300 or more staffed beds. We chose to include multiple staffed bed indicator variables to examine potentially nonlinear relationships between hospital size and negotiated prices. We chose 100–299 staffed beds as the comparison group to compare hospitals similar in size (i.e., not compare the largest hospitals to the smallest hospitals).

2.3. Analysis

We estimated Poisson regression models to evaluate the association between hospital characteristics and facility allowed costs (“price”). The regression models accounted for all 10 hospital characteristics of interest, in addition to a hospital wage index variable, state fixed effects, and year fixed effects; this allowed for identification of the independent association of each hospital characteristic within a state. We included state fixed effects to control for time‐invariant geographic differences, since Medicaid policy often varies at the state level. To capture more granular geographic differences in prices, we included the hospital wage index produced for the Medicare program, which varies by Core Based Statistical Area (CBSA) [35]. The following describes how to interpret each estimate from the model, based on the indicator variables included in the model. Children's hospitals and CAHs were each compared to other acute care hospitals (“other acute care”). For‐profit hospitals, government hospitals, and religiously affiliated (nonprofit) hospitals were each compared to other nonprofit hospitals (“other nonprofit”). Hospitals with 1–99 staffed beds and hospitals with 300 or more staffed beds were compared to hospitals with 100–299 staffed beds. The other characteristics—rural, system, and teaching hospitals—were compared to their counterpart hospitals: rural versus not rural; system‐affiliated versus independent; and teaching hospital versus non‐teaching hospital. Standard errors were clustered at the hospital level.

We adjusted the regression specifications in some sensitivity analyses. One sensitivity analysis calculated allowed costs at the claim level instead of the claim line level, which allowed us to capture the allowed costs of ancillary services provided in conjunction with the procedure of interest. The other sensitivity analysis used uncapped allowed costs. For the claim‐level analysis, claims were excluded if, in addition to the procedure of interest, the claim had other claim lines with a surgical procedure; surgical procedures were defined as CPT codes 10004‐69990 [36].

We performed all analyses in SAS, version 9.4 (SAS Institute). Statistical tests were 2‐tailed at significance level p < 0.05.

3. Results

3.1. Sample Characteristics

For the child population, the sample included 2,250,325 individuals. The number of claim lines ranged from 32,793 forearm splints to 2,642,747 ED E/M Level 3 visits. Average prices varied substantially by procedure, from a mean price of $90 for a forearm splint to $1869 for a tonsillectomy and adenoidectomy. The sample sizes and average facility prices of each procedure are shown in Table 1.

TABLE 1.

Summary statistics by procedure—children.

Children (ages 0–18)
	Tonsillect. and Aden.	Tympanostomy	Forearm splinting	ED E/M Level 2	ED E/M Level 3	ED E/M Level 4	ED E/M Level 5
No. of procedures	33,683	39,037	32,793	832,838	2,642,747	1,064,674	290,007
No. of unique hospitals	902	873	1843	2381	2442	2414	2217
Mean (SD) facility allowed cost a	$1869 ($1434)	$1031 ($775)	$90 ($94)	$155 ($108)	$198 ($142)	$315 ($315)	$550 ($589)
Median facility allowed cost	$1614	$856	$64	$122	$166	$226	$334
Mean (SD) patient age	7 (4)	3 (3)	11 (4)	6 (5)	7 (6)	9 (6)	10 (6)
Percentage of procedures occurring at each hospital type b
Children's hospitals	24%	21%	8%	13%	12%	14%	24%
CAHs	2%	3%	7%	8%	7%	6%	5%
Rural hospitals	14%	15%	17%	16%	15%	14%	11%
System‐affiliated hospitals	77%	76%	83%	82%	84%	84%	85%
Teaching hospitals	64%	60%	53%	59%	60%	61%	70%
For‐profit hospitals	11%	12%	19%	16%	19%	13%	10%
Religiously affiliated hospitals	13%	13%	15%	13%	14%	15%	14%
Government hospitals	17%	18%	16%	20%	16%	16%	14%
Hospitals with 1–99 staffed beds	16%	19%	20%	22%	19%	17%	13%
Hospitals with 300+ staffed beds	54%	51%	42%	48%	48%	50%	57%

Abbreviations: Aden., adenoidectomy; CAH, critical access hospital; E/M, evaluation & management; ED, emergency department; SD, standard deviation; Tonsillect., tonsillectomy.

^a Allowed costs were capped at the 99th percentile.

^b Percentages sum to more than 100% for each procedure because hospitals were characterized as more than one type.

For the adult population, the sample included 2,341,063 individuals. The number of claim lines ranged from 5963 arthroscopies to 2,566,791 ED E/M Level 4 visits. Average price varied by procedure, from a mean price of $134 for ED E/M Level 2 visits to $1394 for arthroscopies. The sample sizes and average facility prices of each procedure are shown in Table 2.

TABLE 2.

Summary statistics by procedure—adults.

Adults (ages 19+)
	Arthroscopy, knee	Colonos., with biop.	Ultrasound (Obstet.)	MRI, brain	ED E/M Level 2	ED E/M Level 3	ED E/M Level 4	ED E/M Level 5
No. of procedures	5963	61,785	88,448	65,201	559,272	2,381,958	2,566,791	1,209,522
No. of unique hospitals	1006	1658	1345	1703	2397	2460	2455	2411
Mean (SD) facility allowed cost a	$1394 ($1059)	$702 ($691)	$157 ($91)	$587 ($394)	$134 ($90)	$166 ($104)	$231 ($173)	$327 ($275)
Median facility allowed cost	$1264	$468	$119	$488	$110	$142	$191	$259
Mean (SD) patient age	43 (12)	49 (13)	27 (5)	45 (14)	36 (13)	37 (13)	38 (14)	42 (15)
Percentage of procedures occurring at each hospital type b
CAHs	5%	4%	5%	4%	11%	8%	7%	6%
Rural hospitals	19%	14%	17%	14%	19%	17%	15%	14%
System‐affiliated hospitals	81%	83%	71%	84%	78%	84%	85%	86%
Teaching hospitals	50%	60%	65%	66%	55%	55%	58%	61%
For‐profit hospitals	17%	11%	7%	7%	12%	18%	17%	11%
Religiously affiliated hospitals	17%	17%	14%	14%	17%	17%	16%	17%
Government hospitals	14%	15%	31%	22%	20%	16%	15%	15%
Hospitals with 1–99 staffed beds	22%	14%	15%	12%	22%	20%	17%	16%
Hospitals with 300+ staffed beds	35%	48%	51%	55%	43%	44%	46%	49%

Abbreviations: biop., biopsy; CAH, critical access hospital; Colonos., colonoscopy; E/M, evaluation & management; ED, emergency department; MRI, magnetic resonance imaging; Obstet., obstetric; SD, standard deviation.

^a Allowed costs were capped at the 99th percentile.

^b Percentages sum to more than 100% for each procedure because hospitals were characterized as more than one type.

Figure 1 shows the percentage of procedures (by setting) occurring within each hospital type. Most care for the procedures occurred in system‐affiliated hospitals, and approximately 60% of procedures occurred in teaching hospitals. Children's hospitals, teaching hospitals, government hospitals, and larger hospitals were more commonly used for outpatient procedures than for ED care. CAHs were less commonly used, especially for the outpatient procedures.

FIGURE 1.

Percentage of select procedures occurring in different hospital types. CAH, critical access hospital; ED, emergency department. Percentages sum to more than 100% for each setting (i.e., ED or outpatient) because hospitals were characterized as more than one type. For the adult population, claims from children's hospitals were excluded.

To understand how representative our sample was to the broader Medicaid MCO population, we compared hospitals in our sample to all hospitals in states with at least some Medicaid MCO enrollment in 2022 (eTables 2 and 3) [25]. We used chi‐square tests to measure statistically significant differences between the populations (p < 0.05). Hospitals in our sample were less likely to be CAHs (23% versus 27%), less likely to be rural (33% versus 36%), more likely to be system‐affiliated (71% versus 69%), and more likely to be larger hospitals (300 or more staffed beds; 23% versus 19%); hospitals in our sample had a similar likelihood of being teaching hospitals (35% versus 33%) and of being other nonprofit (47% versus 50%).

3.2. Regression Models

We report incidence rate ratios (IRR) from the Poisson regression estimates, which show the percentage difference in cost at indicator hospitals compared with control hospitals. IRR estimates for the child and adult populations are shown in Figures 2 and 3, respectively, as coefficient plots, and eTables 4 and 5 show the estimates in table form.

FIGURE 2.

Coefficient plots for child population. CAH, critical access hospital; ED, emergency department; E/M, evaluation & management. The outcome is allowed cost for each procedure (capped at the 99th percentile). Regressions include all explanatory variables in the figure as well as a hospital wage index variable, state fixed effects, and year fixed effects. Standard errors were clustered at the hospital level. The figure includes incidence rate ratio estimates and 95% confidence intervals from regressions. For children's hospitals and CAHs, the reference group is other acute care hospitals. For hospital ownership type, the reference group is other nonprofit hospitals. For the two hospital size variables, the reference group is 100–299 staffed beds.

FIGURE 3.

Coefficient plots for adult population. CAH, critical access hospital; ED, emergency department; E/M, evaluation & management; MRI, magnetic resonance imaging. The outcome is allowed cost for each procedure (capped at the 99th percentile). Regressions include all explanatory variables in the figure as well as a hospital wage index variable, state fixed effects, and year fixed effects. Standard errors were clustered at the hospital level. The figure includes incidence rate ratio estimates and 95% confidence intervals from regressions. For CAHs, the reference group is other acute care hospitals. For hospital ownership type, the reference group is other nonprofit hospitals. For the two hospital size variables, the reference group is 100–299 staffed beds.

For the child population, children's hospitals were substantially more expensive than other acute care hospitals across all seven procedures; the largest estimate was for ED E/M Level 2 (IRR = 2.51; 95% CI, 2.14–2.95) and the smallest estimate was for tonsillectomy and adenoidectomy (IRR = 1.23; 95% CI, 1.01–1.50). These estimates were consistently larger than estimates for any other hospital characteristic (except for CAHs for tympanostomy), and they represented an estimated 93%–151% higher prices for ED procedures and 23%–48% higher prices for outpatient procedures. Though the outpatient procedures had lower IRR estimates than the ED procedures, they had higher mean allowed costs, suggesting still large absolute cost differences between children's hospitals and other acute care hospitals. CAHs were also more expensive than other acute care hospitals, except for tonsillectomy and adenoidectomy, which did not have a significant association with price. For CAHs, the largest estimate was for tympanostomy (IRR = 1.78; 95% CI, 1.45–2.18), representing 78% higher prices; IRR estimates for ED procedures ranged from 1.25 (95% CI, 1.10–1.43) for splinting to 1.44 (95% CI, 1.27–1.63) for ED E/M Level 4. Hospitals in a system were more expensive for most procedures (splinting being the exception) as compared to independent hospitals. For‐profit hospitals were less expensive than other nonprofit hospitals for several procedures. Hospitals with 300 or more staffed beds showed higher prices for all ED E/M levels.

For the adult population, CAHs were substantially more expensive than other acute care hospitals across all eight procedures; the largest IRR estimate was for MRI (2.43; 95% CI, 2.13–2.76) and the smallest estimate was for ED E/M Level 3 (1.34; 95% CI, 1.25–1.44). These prices were consistently larger than for any other hospital characteristic, and they represented an estimated 34%–40% higher prices for ED procedures and 55%–143% higher prices for outpatient procedures. Rurality and teaching status were not significantly associated with price. System affiliation was associated with higher prices for ED E/M Levels 3–5. At for‐profit hospitals, prices for colonoscopy and ED E/M Levels 4 and 5 were lower, but prices for ED E/M Level 2 were higher. Religious affiliation was associated with lower prices for colonoscopy but was not associated with prices for the other procedures. Government ownership was associated with higher prices for ED E/M Level 5 but was not associated with prices for the other procedures. Hospitals with 1–99 staffed beds were associated with higher prices for colonoscopies and MRI, and hospitals with 300 or more staffed beds were associated with higher prices for ED E/M Levels 4 and 5.

3.3. Sensitivity Analyses

In eTables 6–9, we show the results of the sensitivity analyses, which were all qualitatively similar to the main estimates. For the claim‐level analysis, the estimates were often similar in magnitude to the main estimates (eTables 6 and 7); no hospital characteristic showed consistently higher or lower estimates for the claim‐level analysis (the exception being for‐profit hospitals for children, which had consistently higher estimates in the claim‐level analysis). In the analysis with uncapped allowed costs, some hospital characteristics had slightly higher estimates than in the main specification, especially children's hospitals (eTables 8 and 9).

4. Discussion

This study examined the association between several important hospital characteristics and a procedure's Medicaid facility allowed costs. For children, we estimated large positive associations with children's hospitals, CAHs, and system‐affiliated hospitals; for a mix of procedures, we also estimated positive associations with large hospitals; and for a mix of procedures, we estimated a negative association with for‐profit hospitals. For adults, we estimated large positive associations with CAHs; for a mix of procedures, we estimated positive associations with system‐affiliated hospitals, small hospitals, and large hospitals; and for the rest of the characteristics, associations were mixed depending on the procedure.

Existing work has generally compared procedure cost for one hospital characteristic, such as whether the hospital is a children's hospital or a teaching hospital. For example, Tian and colleagues [16] compared children's hospitals to general acute care hospitals and found higher perforated appendicitis cost at children's hospitals. The literature generally finds that children's hospitals and teaching hospitals are more expensive than other hospitals [15, 16, 17, 18, 37, 38]. However, most studies relied on hospital discharge data that included only charges, rather than negotiated prices, or studied Medicare patients. More recent work has more broadly compared hospital characteristics and the cost of care and found that nonprofit hospitals had lower commercial and Medicare costs than for‐profit hospitals [19, 21, 22].

Building on these findings, our study combined many different characteristics, allowing us to differentiate Medicaid managed care prices for each characteristic more granularly. For example, based on the claims data used in our analysis, approximately 90% of teaching hospital claims were from hospitals also in a hospital system (89% of the child claims and 91% of the adult claims). For any study comparing one characteristic, such as teaching versus non‐teaching hospitals, without accounting for other hospital characteristics, results may suffer from omitted variable bias. Some studies examined costs at teaching and non‐teaching hospitals without controlling for the system affiliation of hospitals [18, 29, 30].

Relative to the literature, we did not find strong evidence of facility price differentials between teaching and non‐teaching hospitals; however, we found that children's hospitals were associated with higher facility prices. The previous point about overlapping characteristics is also important for interpreting our study's findings. For example, based on the claims data used in our analysis, approximately 76% of children's hospital claims were from system‐affiliated hospitals. For many children's hospitals, therefore, it is appropriate to interpret the estimates for children's hospitals and system‐affiliated hospitals together, which suggests even higher prices when compared to other acute care hospitals not in a system.

The study results—and particularly, the higher prices at children's hospitals and CAHs—could be driven by many possible factors, a few of which we discuss. Children's hospitals and CAHs may be able to negotiate higher prices with Medicaid managed care plans due to state laws (like New Jersey's 2021 statute on network adequacy standards, S.3000 [39]) or regulations that, in an effort to ensure access to needed specialty care or care across geographic locations, effectively require Medicaid managed care plans to contract with certain hospitals. These laws or regulations may hamper an insurer's ability to negotiate with children's hospitals since the insurer cannot keep hospitals out of their networks.

Additionally, consumers might prefer specialty hospitals even for routine procedures. Recent research suggests that in the commercial insurance market, consumers differentiate between children's and non‐children's hospitals such that children's hospitals are not subject to competitive pressure for routine procedures [40]. It is possible that consumers' strong preferences for children's hospitals, even for routine procedures, provide leverage for children's hospitals to command higher prices for pediatric care when negotiating with insurers, including in Medicaid.

Hospitals in a system and large hospitals also tended to have higher prices for some of the procedures examined, but not all procedures had a significant association. These hospitals likely have greater market power, which may allow them to negotiate higher prices with insurers, and the trend of hospital consolidation exacerbates this dynamic. Nevertheless, we did not observe higher prices in large hospitals or system‐affiliated hospitals for all procedures we examined, potentially suggesting that hospitals may selectively focus on specific services when negotiating prices.

Another possible driver is SDPs. It is our understanding that the claims data used in the analysis could include some SDPs, but this is uncommon, as SDPs are typically paid separately (outside of claims for individual healthcare services). Quantifying the degree of SDPs in the claims data is beyond the scope of this paper. Nevertheless, if SDPs were directed at certain hospitals (particularly, children's hospitals or CAHs) in the states examined and were reflected in the claims data used for the analysis, then they could also be a potential driver of the observed results.

Finally, higher prices for children's hospitals and CAHs may also be the result of structurally different operating environments, payor mixes, staffing requirements, geographic constraints, or regulatory obligations that affect negotiated prices independent of classic market power dynamics. While the scope of a broad analysis of all factors is not possible in this study, a number of observations can be made. For example, CAHs serve a unique role in rural areas, thereby leading Medicare to reimburse them at cost rather than prospective payments. This is the result of structural challenges as CAHs provide a breadth of services at a low volume for rural communities. The lack of scale leads to higher operating costs per procedure and could require higher rates. Similarly, children's hospitals are specialized, family‐centered operations that can require additional staffing and physical space, and their payor mix often includes a larger share of low‐income children generally covered by Medicaid [41, 42]. These challenges suggest that higher Medicaid reimbursement rates may compensate for the relatively low price paid relative to the commercially insured patient population.

4.1. Limitations

This study has several limitations. First, our results may not be generalizable to other insurers or to the market as a whole. Also, we only analyzed Medicaid claims data, so the results may not be generalizable to populations with commercial insurance or with Medicare coverage. Additionally, the results may not be generalizable to procedures not examined, such as non‐routine procedures or different routine procedures.

Second, this study's analysis only examined facility prices, so it does not capture any professional fees billed separately from facility claims. However, physicians who are not directly employed by a hospital can have a negotiation process independent of the hospital, so hospital characteristics may be less relevant to the associated professional fee for these physicians.

Third, we could not control for all possible confounders. We did not observe all hospital characteristics, and we did not examine other market characteristics, such as the insurer concentration in each market. Markets with high insurer concentration can experience lower hospital prices relative to markets with low insurer concentration (all else equal) [14, 43, 44]. Studying insurer concentration alongside hospital characteristics could be an area of future research.

Finally, we did not examine the mechanisms for increased prices. While we discussed market power, it is also possible that higher operational costs—such as the cost of hospital workers or medical equipment—may result in higher prices. It is possible that these inputs are also associated with higher quality and improved health outcomes. Nevertheless, because our study focused on routine procedures, major differences in health outcomes are unlikely, particularly for the outpatient procedures.

5. Conclusions

This analysis found that children's hospitals were associated with substantially higher Medicaid managed care prices than other acute hospitals for several common outpatient and ED pediatric procedures. CAHs also tended to have higher prices for both children and adults. Results were mixed for the other hospital characteristics, depending on the procedure. More research is needed to understand what is driving the observed differences in hospital prices by various hospital characteristics.

Funding

The authors have nothing to report.

Conflicts of Interest

Dr. Alfrey and Ms. Gordon are employees of Elevance Health, and Dr. Ukert was affiliated with Elevance Health during part of the analysis. The authors conducted this study as part of employment duties and did not receive any additional funding for the work. Ms. Gordon reported owning stock in Elevance Health.

Supporting information

eTable 1. List of procedures included in analysis.

eTable 2. Characteristics of hospital sample (child population).

eTable 3. Characteristics of hospital sample (adult population).

eTable 4. Regression results for child population—claim‐line‐level analysis.

eTable 5. Regression results for adult population—claim‐line‐level analysis.

eTable 6. Sensitivity analysis for child population—claim‐level analysis.

eTable 7. Sensitivity analysis for adult population—claim‐level analysis.

eTable 8. Sensitivity analysis for child population—uncapped allowed costs.

eTable 9. Sensitivity analysis for adult population—uncapped allowed costs.

eMethods. Additional details regarding rolled up claims.

Data Availability Statement

The data are proprietary and cannot be shared.