BACKGROUND AND OBJECTIVES

Physician management companies (PMCs) acquire physician practices and contract with hospitals to provide physician management services. We evaluated the association between PMC-NICU affiliations and prices, spending, utilization, and clinical outcomes.

METHODS

We linked commercial claims to PMC-NICU affiliations and conducted difference- in-differences analyses comparing changes in prices paid for physician services per critical or intensive care NICU day, length of the NICU stay, physician spending (total paid amount for physician services during stay), spending on hospital services (total paid amount for hospital services during stay), and clinical outcomes in PMC-affiliated versus non-PMC–affiliated NICUs. The study included 2858 infants admitted to 34 PMC-affiliated NICUs and 92 461 infants admitted to 2348 NICUs without an affiliation.

RESULTS

PMC affiliation was associated with a differential increase in the mean price of the 5 most common types of critical and intensive care days in NICU admissions by $313 per day (95% confidence interval, $207–$419) for PMC-affiliated versus non- PMC-affiliated NICUs. This represents a 70.4% increase in prices, relative to the preaffiliation period PMC and non- PMC-affiliated NICU means. PMC-NICU affiliation was also associated with a differential increase in physician spending by $5161 per NICU stay (95% confidence interval, $3062–$7260), a 56.4% increase. There was no significant association between PMC-NICU affiliation and changes in length of stay, clinical outcomes, or hospital spending.

CONCLUSIONS

PMC affiliation was associated with large increases in prices and total spending for NICU services, but not with changes in length of stay or adverse clinical outcomes.

What’s Known on This Subject:

In previous research, physician management companies (PMCs) contracting to provide physician staffing increased payments to emergency department physicians and anesthesiologists. Little is known, however, about PMCs’ impact on utilization and price of care in neonatology, where PMCs are prevalent.

What This Study Adds:

Results extend the literature on PMCs to neonatology, showing that PMC affiliations were associated with large increases in prices paid for physician services per NICU day, without changes in utilization or clinical outcomes.

Physician management companies (PMCs) have increasingly acquired physician practices and employed physicians over the past decade, particularly in hospital-based specialties such as emergency medicine, anesthesiology, and neonatology.1,2  PMC-staffed NICUs are common: 1 PMC firm claims to provide neonatology services in 385 affiliated NICU facilities to nearly a quarter of NICU infants in the United States.3 

PMCs are for-profit companies that may be supported by private equity investments.4  PMCs typically own a practice’s nonclinical assets, provide staffing and management services, and negotiate staff contracts with hospitals and payment amounts with insurers.5  They claim to provide benefits to both neonatologists and hospitals via economies of scale and expertise in management, including contract negotiations, regulatory compliance, human resources, and billing.6 

Critics argue that PMC-owned neonatology practices are under pressure to deliver short-term profits to investors and shareholders.6,7  Profits might be generated by negotiating higher prices with insurers,4,8  increasing efficiency, reducing spending on quality improvement efforts, and increasing the use of high-margin NICU services.9,10 

Despite the increasing prevalence of PMCs, little is known about their impact in NICUs. We analyzed the association between PMC staffing of NICUs and prices, costs, utilization of care, and clinical outcomes.

We constructed a novel database identifying NICU affiliations with 2 of the largest neonatology-focused PMCs (Pediatrix and Envision Physician Services) and their subsidiaries (Appendix 1 in Supplemental Information).3,11  These PMCs supply at least 25% of the market share for neonatology services in the United States.12 

We merged PMC-NICU affiliations with 2011–2017 commercial claims data for preferred provider organization and point-of-service plans from the Health Care Cost Institute (HCCI). This database contains commercial claims from 3 large national insurers covering ∼50 million individuals.13 

We placed NICUs into 3 categories:

  1. NICUs never affiliated with PMCs;

  2. NICUs that became PMC-affiliated between July 1, 2011, and July 31, 2017; and

  3. NICUs ever affiliated with a PMC before July 1, 2011, after July 31, 2017, or whose affiliation date is unknown (Supplemental Fig 1).

    This last category was not included in analyses.

The primary study sample included 2858 infants admitted to PMC-affiliated NICUs (34 NICUs) and 92 461 infants admitted to non-PMC–affiliated NICUs (2348 NICUs) between January 1, 2011, and December 31, 2017. Following Goodman, NICU admissions were defined as live births with an intermediate or intensive professional day current procedural terminology (CPT) code or an intensive facility day revenue code.14  We excluded 510 infants with birth weight <500 g, because of the small number of such admissions and risk of adverse events that would be difficult or impossible to prevent for extremely low birth weight infants.14,15  Thirteen such infants (0.45% of PMC-affiliated admissions) were admitted to PMC-affiliated NICUs; 497 (0.54%) were admitted to non-PMC–affiliated NICUs. Further exclusions are shown in the appendix (Supplemental Fig 2).

In addition to our primary study sample, we examined the association between PMC affiliation and utilization and quality of care for patients who were admitted to a NICU but were not critically ill, because PMCs may be motivated to increase discretionary care for infants with less severe conditions. These infants had a revenue code corresponding to a NICU Level II facility (0172) or a NICU level III facility (0173).16 

Prices

We measured prices as the negotiated insurer “allowed amounts” due to be paid to the PMC for NICU services, including what the insurer paid and the amount, if any, due from the patient. The prices are geographically adjusted for hospital county (Appendix 2 in Supplemental information).17 

We focused on the mean price for physician services for 3 common NICU services:

  1. subsequent critical care, per day, for evaluation and management of critically ill neonates 28 days or younger (CPT code 99469);

  2. subsequent intensive care, per day, for evaluation and management of recovering infants weighing 1500 g to 2500 g (CPT code 99479); and

  3. subsequent intensive care, per day, for evaluation and management of recovering infants weighing 2501 g to 5000 g (CPT code 99480).14,18 

These 3 services accounted for 73% of subsequent care days in NICUs in our sample. Additionally, we examined mean prices across the 5 most common types of critical and intensive care days (CPT codes 99469, 99472, 99478, 99479, 99480). We focused on subsequent care days, because they are more frequent than initial care days, and procedure-specific prices are highly correlated across subsequent and initial care days (correlation coefficient over 80%).

Utilization

Primary utilization measures included length of stay in the NICU, and the number of intensive and critical care days (Supplemental Table 1).14 

Secondary utilization measures included the number of chest x-rays, abdominal x-rays, head MRIs, and head ultrasounds.14  Additionally, because an increase in NICU admissions for normal-weight infants might suggest overutilization,9  we examined the association between PMC affiliation and the number of NICU admissions for infants weighing at least 2500 g.

Spending

Primary spending outcomes were:

  1. spending for all physician services during a NICU stay; and

  2. spending for all services paid to the hospital during a NICU stay.

Spending for physician services depends on procedure-specific prices (allowed amounts) and the volume of services provided. Spending for hospital services is the allowed amount paid to the hospital per hospital-day.

To examine whether higher physician spending resulted from higher prices or higher volume of physician services, we calculated total standardized physician spending using CPT national median prices (Appendix 2 in Supplemental Information). Increases in total standardized physician spending imply higher utilization, rather than higher prices.

Readmissions, Vaccination, and Adverse Clinical Outcomes

We measured 7, 14, and 30-day hospital readmissions among discharged infants.19  We also measured whether an infant received a hepatitis B vaccine20  while hospitalized. Clinical outcomes included whether an infant developed necrotizing enterocolitis,21  intraventricular hemorrhage, or neonatal late-onset sepsis.22 

Out-of-Network Billing

We examined the likelihood that a NICU admission at an in-network hospital contained out-of-network physicians claims, using an out-of-network indicator available in the data from 2014 to 2017.

Covariates

All analyses included variables intended to risk-adjust for infants’ conditions, as identified in the claims data (Table 1, Supplemental Table 2).14  Infant conditions included birth weight category, infant sex, whether the delivery included a cesarean delivery or multiple births, whether the infant had a congenital anomaly, whether major procedures were conducted (Supplemental Table 2), and whether the infant was transferred from another hospital.

TABLE 1

Characteristics of NICU Admissions and NICU Facilities in PMC-Affiliated and Non-PMC–Affiliated NICUs, Before and After Physician Management Company Affiliations

Preaffiliation PeriodaPostaffiliation Perioda
Non-PMC–Affiliated NICUs,b, cMeanPMC-Affiliated NICUs,d MeanPMC-Affiliated and Non-PMC–Affiliated NICUs,e 95% CIPMC-Affiliated NICUs,d MeanPMC-Affiliated and Non-PMC–Affiliated NICUs From Pre- to Postaffiliation Period,f, g 95% CI
Infant and hospital characteristics    d  
 No. of admissionsh 13 119 1132 14 251 1726 95 319 
Infant characteristics, %      
 Female 44.21 44.08 0.00 (−3.14 to 2.88) 44.90 1.17 (−1.28 to 3.62) 
 Cesarean delivery 54.08 56.71 2.63 (−0.39 to 5.66) 54.06 −0.87 (−5.35 to 3.61) 
 Multiple births delivery 18.02 16.17 −1.85 (−4.18 to 0.47) 18.02 2.35 (−0.76 to 5.46) 
 Major procedure 3.38 3.09 −0.28 (−1.38 to 0.81) 3.65 0.37 (−2.04 to 2.79) 
 Congenital anomaly 10.48 8.48 −2.00 (−3.85 to −0.15) 8.86 −1.18 (−4.55 to 2.19) 
 Transferred in 8.39 12.28 3.89 (2.18 to 5.60) 12.86 3.15 (−2.80 to 9.11) 
Birth weight, g, %      
 500–749 1.68 1.94 0.27 (−0.52 to 1.05) 1.62 0.15 (−0.90 to 1.21) 
 750–999 2.36 2.56 0.21 (−0.72 to 1.13) 3.01 0.84 (−0.45 to 2.13) 
 1000–1249 3.05 2.83 −0.22 (−1.26 to 0.82) 3.36 0.66 (−1.02 to 2.35) 
 1250–1499 3.81 2.65 −1.17 (−2.32 to −0.00) 4.29 0.71 (−0.32 to 1.73) 
 1500–1749 4.32 5.39 1.07 (−0.18 to 2.31) 5.12 0.97 (−0.68 to 2.62) 
 1750–1999 7.17 7.07 −0.10 (−1.66 to 1.47) 8.05 2.04 (−0.49 to 4.58) 
 2000–2499 14.95 14.66 −0.28 (−2.45 to 1.88) 16.80 2.54 (0.38 to 4.70) 
 2500 and above 62.66 62.90 0.23 (−2.70 to 3.17) 57.65 −7.92 (−15.03 to −0.82) 
Hospital characteristicsi      
 Number of hospital facilities,g No. 2348 34 2382 34 2382 
 Hospital Medicaid discharges 3740.14 3122.75 −617.39 (−1932.49 to 697.72) 3382.62 939.54 (−152.96 to 2032.03) 
 Live births 1952.29 1866.29 −86.00 (−640.44 to 468.44) 1953.83 687.02 (74.78 to 1299.27) 
 NICU beds 15.43 23.62 8.18 (0.65 to 15.72) 21.46 15.26 (6.52 to 24.00) 
 Neonatal intermediate care beds 4.32 7.97 3.65 (−0.13 to 7.43) 7.73 4.85 (−4.41 to 14.11) 
 Obstetric care beds 29.37 34.44 5.07 (−3.19 to 13.33) 31.40 12.31 (1.76 to 22.87) 
 General medical and surgical beds 175.38 141.90 −33.48 (−85.80 to 18.84) 144.36 10.7 (−35.10 to 56.59) 
Preaffiliation PeriodaPostaffiliation Perioda
Non-PMC–Affiliated NICUs,b, cMeanPMC-Affiliated NICUs,d MeanPMC-Affiliated and Non-PMC–Affiliated NICUs,e 95% CIPMC-Affiliated NICUs,d MeanPMC-Affiliated and Non-PMC–Affiliated NICUs From Pre- to Postaffiliation Period,f, g 95% CI
Infant and hospital characteristics    d  
 No. of admissionsh 13 119 1132 14 251 1726 95 319 
Infant characteristics, %      
 Female 44.21 44.08 0.00 (−3.14 to 2.88) 44.90 1.17 (−1.28 to 3.62) 
 Cesarean delivery 54.08 56.71 2.63 (−0.39 to 5.66) 54.06 −0.87 (−5.35 to 3.61) 
 Multiple births delivery 18.02 16.17 −1.85 (−4.18 to 0.47) 18.02 2.35 (−0.76 to 5.46) 
 Major procedure 3.38 3.09 −0.28 (−1.38 to 0.81) 3.65 0.37 (−2.04 to 2.79) 
 Congenital anomaly 10.48 8.48 −2.00 (−3.85 to −0.15) 8.86 −1.18 (−4.55 to 2.19) 
 Transferred in 8.39 12.28 3.89 (2.18 to 5.60) 12.86 3.15 (−2.80 to 9.11) 
Birth weight, g, %      
 500–749 1.68 1.94 0.27 (−0.52 to 1.05) 1.62 0.15 (−0.90 to 1.21) 
 750–999 2.36 2.56 0.21 (−0.72 to 1.13) 3.01 0.84 (−0.45 to 2.13) 
 1000–1249 3.05 2.83 −0.22 (−1.26 to 0.82) 3.36 0.66 (−1.02 to 2.35) 
 1250–1499 3.81 2.65 −1.17 (−2.32 to −0.00) 4.29 0.71 (−0.32 to 1.73) 
 1500–1749 4.32 5.39 1.07 (−0.18 to 2.31) 5.12 0.97 (−0.68 to 2.62) 
 1750–1999 7.17 7.07 −0.10 (−1.66 to 1.47) 8.05 2.04 (−0.49 to 4.58) 
 2000–2499 14.95 14.66 −0.28 (−2.45 to 1.88) 16.80 2.54 (0.38 to 4.70) 
 2500 and above 62.66 62.90 0.23 (−2.70 to 3.17) 57.65 −7.92 (−15.03 to −0.82) 
Hospital characteristicsi      
 Number of hospital facilities,g No. 2348 34 2382 34 2382 
 Hospital Medicaid discharges 3740.14 3122.75 −617.39 (−1932.49 to 697.72) 3382.62 939.54 (−152.96 to 2032.03) 
 Live births 1952.29 1866.29 −86.00 (−640.44 to 468.44) 1953.83 687.02 (74.78 to 1299.27) 
 NICU beds 15.43 23.62 8.18 (0.65 to 15.72) 21.46 15.26 (6.52 to 24.00) 
 Neonatal intermediate care beds 4.32 7.97 3.65 (−0.13 to 7.43) 7.73 4.85 (−4.41 to 14.11) 
 Obstetric care beds 29.37 34.44 5.07 (−3.19 to 13.33) 31.40 12.31 (1.76 to 22.87) 
 General medical and surgical beds 175.38 141.90 −33.48 (−85.80 to 18.84) 144.36 10.7 (−35.10 to 56.59) 

Authors’ analysis of HCCI data, 2011 to 2017. PMC-affiliated NICUs were compared with non-PMC–affiliated NICUs, before and after PMC affiliations were implemented.

a

The postaffiliation period is defined as after the start of PMC affiliation for PMC-affiliated NICUs.

b

The unweighted means for infant and hospital variables in non-PMC–affiliated NICUs were calculated at the midpoint of the study period, 2014, following the method in Beaulieu et al (2020).7 

c

Non-PMC–affiliated PMCs were all NICUs that were not found to be affiliated with a PMC at any point in time.

d

PMC-affiliated NICUs became PMC-affiliated between July 1, 2011, and June 30, 2017. The unweighted means for infant and hospital variables were calculated for PMC-affiliated NICUs.

e

Preaffiliation differences between PMC-affiliated and non-PMC–affiliated NICUs reflects the mean for PMC-affiliated NICUs across all preaffiliation years, minus the mean for non-PMC–affiliated NICUs in 2014.

f

The differential estimate indicates the change from the pre- to postaffiliation period for PMC-affiliated NICUs relative to non-PMC–affiliated NICUs that were never affiliated with a PMC. The analyses adjusted for year fixed effects.

g

The study sample includes all non-PMC–affiliated NICUs and all PMC-affiliated NICUs present in both pre- and postaffiliation years.

h

Column 1 includes a count of all admissions to non-PMC–affiliated NICUs in 2014 and the number of admissions to PMC-affiliated NICUs summed across preaffiliation years. Column 2 includes a count of all admissions to PMC-affiliated NICUs in preaffiliation years. Column 3 contains total admissions in columns 1 and 2. Column 4 includes all admissions to PMC-affiliated NICUs in postaffiliation years. Column 5 includes total admissions for all non-PMC–affiliated and PMC-affiliated NICUs across all study years. Given that affiliation occurred in different years for different NICUs, the number of pre- and postaffiliation years and pre- and postaffiliation admissions differs across PMCs.

i

Hospital characteristics were taken from 2011–2017 American Hospital Association data.

Statistical Analysis

The primary analysis used a pooled difference-in-differences design to estimate the relationship between the start of PMC affiliation and changes in patient outcomes between infants treated by newly affiliated PMC-NICUs (PMC-affiliated NICUs) and those treated by non-PMC–affiliated NICUs.23 

The main variable of interest was an indicator for whether a NICU began an affiliation with a PMC in a given year (Appendix 3 in Supplemental Information). The pooled difference-in-differences analysis identifies the aggregated pre- to postperiod difference between the PMC-affiliated and non-PMC–affiliated groups. We also conducted event study analyses, an extension of the difference-in-differences approach, to estimate the differential change in outcomes for PMC-affiliated versus non-PMC–affiliated NICUs for each pre- and postaffiliation year, relative to the year before affiliation initiation. New PMC affiliations occurred in different years (2 PMC-NICU affiliations in 2011, 1 in 2012, 16 in 2013, 2 in 2014, 5 in 2015, 1 in 2016, and 7 in 2017). All event study analyses included interaction terms between the PMC-NICU affiliation indicator and each year, within the 3 years before and after the affiliation started.

Linear regressions were used for continuous outcomes, and logistic regressions for binary outcomes. All models included hospital fixed effects to control for hospital characteristics that are not likely to change over the study period, such as rurality, and calendar year fixed effects to control for factors changing in a year that are common to all hospitals, such as federal policies. Standard errors were clustered at the NICU level.

We conducted 6 sensitivity analyses. First, to address concerns that estimates reflect differences in markets across hospital referral regions (HRRs), we examined whether results are robust to including HRR fixed effects. Second, we repeated analyses using a smaller, linked sample of mothers and infants (45% of infants were linked). This enabled us to include 33 additional risk-adjustment factors using both mother and infant claims (Supplemental Table 3).14  Third, because infants who are transferred may be sicker, we included only infants who were not transferred. Fourth, we used the difference-in-differences estimator developed by Chaisemartin and D’haultfoeuille, which is robust to heterogenous treatment effects over time or by hospital.24,25  Fifth, we conducted several tests for robustness of findings to alternative assumptions about missing data (Appendix 4 in Supplemental Information). Finally, to account for skewed cost and utilization outcomes, we used quantile regression to analyze price and spending outcomes at different quantiles (50th, 75th, and 95th percentiles) of costs, and Poisson pseudo maximum- likelihood regressions to estimate price, spending, and utilization outcomes.

NICUs affiliated with PMCs at any point in time were largely situated in the South and West, possibly because of interstate differences in insurer payment rates and corporate practice of medicine laws (Supplemental Fig 3).26  Thirty-four NICUs in 10 states became affiliated with a PMC during the study period. The final study sample consisted of 2858 infants admitted to 34 PMC-affiliated NICUs and 92 461 infants to 2348 non-PMC–affiliated NICUs from 2011 to 2017 (Supplemental Fig 2). Before PMC affiliation, PMC-affiliated NICUs had more transfers-in, slightly more cesarean deliveries, and fewer infants with a congenital anomaly (Table 1). Among PMC-affiliated NICUs, 29% were investor-owned, 47% reported a medical school affiliation, and 94% were located in a metropolitan area compared to 12%, 56%, and 91% of non-PMC-affiliated NICUs, respectively. Additional summary statistics are in Supplemental Table 4.

The adjusted event study analyses showed no evidence of pretrends for any outcomes, because changes during the preperiod did not differ significantly between non-PMC–affiliated and non-PMC– affiliated NICUs (Figs 13, Supplemental Figs 4–8). This supports the difference-in-differences key parallel trends assumption, which requires that outcomes in PMC-affiliated and non-PMC–affiliated NICUs follow the same trajectory before PMC affiliation.

FIGURE 1

Adjusted differential changes in prices for treatment and comparison NICUs in pre- and post-PMC–affiliation years. Caption: Adjusted event study estimates between treatment and comparison NICUs. The study sample contains 2011–2017 NICU admissions from authors’ analysis of HCCI data. Comparison NICUs encompass the group of NICUs that were never affiliated with PMCs during the study period. Treatment NICUs encompass the group of NICUs that became affiliated with a PMC between July 1, 2011, and July 31, 2017. CPT code 99469 corresponds to “subsequent critical care” for infants 28 days or younger. CPT Code 99479 corresponds to “subsequent intensive care” for infants not yet requiring critical care, weighing 1500 g to 2500 g. CPT code 99472 corresponds to subsequent critical care for infants aged 29 days through 24 months. Code 99478 corresponds to subsequent intensive care for infants weighing <1500 g. Treatment NICUs must be present in all pre- and postaffiliation years. The x-axis represents the number of years relative to the year of initiation of the management company (PMC) contract with a NICU. The midpoint of the year in which the contract started between the PMC and the NICU is denoted as the transition year. Year-to-event coefficient estimates represent changes in the outcome and 95% CI (error bars); coefficients were estimated relative to the reference period; that is, the year before affiliation. The y-axis represents allowed amounts, which include the amount the insurer paid and the amount, if any, to be paid by the patient. All models controlled for year fixed effects, hospital fixed effects, birth weight category, whether the delivery included multiple births, the sex of the infant, whether the infant was delivered via cesarean delivery, whether the infant was transferred into the NICU from another hospital, the presence of a congenital anomaly, and whether major procedures were conducted during the infant’s hospital stay.

FIGURE 1

Adjusted differential changes in prices for treatment and comparison NICUs in pre- and post-PMC–affiliation years. Caption: Adjusted event study estimates between treatment and comparison NICUs. The study sample contains 2011–2017 NICU admissions from authors’ analysis of HCCI data. Comparison NICUs encompass the group of NICUs that were never affiliated with PMCs during the study period. Treatment NICUs encompass the group of NICUs that became affiliated with a PMC between July 1, 2011, and July 31, 2017. CPT code 99469 corresponds to “subsequent critical care” for infants 28 days or younger. CPT Code 99479 corresponds to “subsequent intensive care” for infants not yet requiring critical care, weighing 1500 g to 2500 g. CPT code 99472 corresponds to subsequent critical care for infants aged 29 days through 24 months. Code 99478 corresponds to subsequent intensive care for infants weighing <1500 g. Treatment NICUs must be present in all pre- and postaffiliation years. The x-axis represents the number of years relative to the year of initiation of the management company (PMC) contract with a NICU. The midpoint of the year in which the contract started between the PMC and the NICU is denoted as the transition year. Year-to-event coefficient estimates represent changes in the outcome and 95% CI (error bars); coefficients were estimated relative to the reference period; that is, the year before affiliation. The y-axis represents allowed amounts, which include the amount the insurer paid and the amount, if any, to be paid by the patient. All models controlled for year fixed effects, hospital fixed effects, birth weight category, whether the delivery included multiple births, the sex of the infant, whether the infant was delivered via cesarean delivery, whether the infant was transferred into the NICU from another hospital, the presence of a congenital anomaly, and whether major procedures were conducted during the infant’s hospital stay.

Close modal
FIGURE 2

Adjusted differential changes in primary utilization outcomes for treatment and comparison NICUs in pre- and post-PMC–affiliation years. Caption: Adjusted event study estimates between treatment and comparison NICUs. The study sample contains 2011–2017 NICU admissions from authors’ analysis of HCCI data. Comparison NICUs encompass the group of NICUs that were never affiliated with PMCs during the study period. Treatment NICUs encompass the group of NICUs that became affiliated with a PMC between July 1, 2011, and July 31, 2017. Treatment NICUs must be present in all pre- and postaffiliation years. The x-axis represents the number of years relative to the year of initiation of the management company (PMC) contract with a NICU. The midpoint of the year in which the contract started between the PMC and the NICU is denoted as the transition year. Year-to-event coefficient estimates represent changes in the outcome and 95% CI (error bars); coefficients were estimated relative to the reference period; that is, the year before affiliation. All models controlled for year fixed effects, hospital fixed effects, birth weight category, whether the delivery included multiple births, the sex of the infant, whether the infant was delivered via cesarean delivery, whether the infant was transferred into the NICU from another hospital, the presence of a congenital anomaly, and whether major procedures were conducted during the infant’s hospital stay.

FIGURE 2

Adjusted differential changes in primary utilization outcomes for treatment and comparison NICUs in pre- and post-PMC–affiliation years. Caption: Adjusted event study estimates between treatment and comparison NICUs. The study sample contains 2011–2017 NICU admissions from authors’ analysis of HCCI data. Comparison NICUs encompass the group of NICUs that were never affiliated with PMCs during the study period. Treatment NICUs encompass the group of NICUs that became affiliated with a PMC between July 1, 2011, and July 31, 2017. Treatment NICUs must be present in all pre- and postaffiliation years. The x-axis represents the number of years relative to the year of initiation of the management company (PMC) contract with a NICU. The midpoint of the year in which the contract started between the PMC and the NICU is denoted as the transition year. Year-to-event coefficient estimates represent changes in the outcome and 95% CI (error bars); coefficients were estimated relative to the reference period; that is, the year before affiliation. All models controlled for year fixed effects, hospital fixed effects, birth weight category, whether the delivery included multiple births, the sex of the infant, whether the infant was delivered via cesarean delivery, whether the infant was transferred into the NICU from another hospital, the presence of a congenital anomaly, and whether major procedures were conducted during the infant’s hospital stay.

Close modal
FIGURE 3

Adjusted differential changes in primary spending outcomes for treatment and comparison NICUs in pre- and post-PMC–affiliation years. Caption: Adjusted event study estimates between treatment and comparison NICUs. The study sample contains 2011–2017 NICU admissions from authors’ analysis of HCCI data. Comparison NICUs encompass the group of NICUs that were never affiliated with PMCs during the study period. Treatment NICUs encompass the group of NICUs that became affiliated with a PMC between July 1, 2011, and July 31, 2017. Treatment NICUs must be present in all pre- and postaffiliation years. The x-axis represents the number of years relative to the year of initiation of the management company (PMC) contract with a NICU. The midpoint of the year in which the contract started between the PMC and the NICU is denoted as the transition year. Year-to-event coefficient estimates represent changes in the outcome and 95% CI (error bars); coefficients were estimated relative to the reference period; that is, the year before affiliation. The y-axis represents allowed amounts, which include the amount the insurer paid and the amount, if any, to be paid by the patient. All models controlled for year fixed effects, hospital fixed effects, birth weight category, whether the delivery included multiple births, the sex of the infant, whether the infant was delivered via cesarean delivery, whether the infant was transferred into the NICU from another hospital, the presence of a congenital anomaly, and whether major procedures were conducted during the infant’s hospital stay.

FIGURE 3

Adjusted differential changes in primary spending outcomes for treatment and comparison NICUs in pre- and post-PMC–affiliation years. Caption: Adjusted event study estimates between treatment and comparison NICUs. The study sample contains 2011–2017 NICU admissions from authors’ analysis of HCCI data. Comparison NICUs encompass the group of NICUs that were never affiliated with PMCs during the study period. Treatment NICUs encompass the group of NICUs that became affiliated with a PMC between July 1, 2011, and July 31, 2017. Treatment NICUs must be present in all pre- and postaffiliation years. The x-axis represents the number of years relative to the year of initiation of the management company (PMC) contract with a NICU. The midpoint of the year in which the contract started between the PMC and the NICU is denoted as the transition year. Year-to-event coefficient estimates represent changes in the outcome and 95% CI (error bars); coefficients were estimated relative to the reference period; that is, the year before affiliation. The y-axis represents allowed amounts, which include the amount the insurer paid and the amount, if any, to be paid by the patient. All models controlled for year fixed effects, hospital fixed effects, birth weight category, whether the delivery included multiple births, the sex of the infant, whether the infant was delivered via cesarean delivery, whether the infant was transferred into the NICU from another hospital, the presence of a congenital anomaly, and whether major procedures were conducted during the infant’s hospital stay.

Close modal

In unadjusted analyses, preaffiliation physician prices were ∼20% lower in NICUs that later became PMC-affiliated compared with non-PMC–affiliated NICUs, but rose to above-market levels after affiliation, with no changes in other outcomes (Table 2, Supplemental Table 5).

TABLE 2

Outcomes in Non-PMC–Affiliated and Non-PMC–Affiliated NICUs, Before and After Physician Management Company Affiliations

Preaffiliation PeriodaPostaffiliationa Period
Non-PMC–Affiliated NICUs,e, fMeanPMC-Affiliated NICUs,g MeanUnadjusted Difference Between PMC-Affiliated and Non-PMC–Affiliated NICUshAdjusted Difference Between PMC-Affiliated and Non-PMC–Affiliated NICUshPMC-Affiliated NICUs, MeaneUnadjusted Differential Change From Pre- to Postaffiliation Periodb, cAdjustedd Differential Change From Pre- to Postaffiliation Periodb, c
No. of admissionsi 14 251 13 119 1132 14 251 14 251 1726 95 319 95 319 
Prices,j 95% CI         
 Price, subsequent critical care d, CPT code 99469, infants 28 d or younger 977.49 990.21 851.40 −138.81
(−196.72 to −80.90) 
−224.31
(−376.94 to −71.68) 
1440.47 481.35
(249.64 to 713.05) 
569.04
(283.10 to 854.97) 
 Price, subsequent intensive care d, CPT code 99479, infants weighing 1500 g–2500 g 350.87 357.22 283.52 −73.69
(−95.75 to −51.64) 
−126.75
(−184.33 to −69.16) 
508.81 164.07
(93.98 to 234.16) 
288.29
(194.47 to 382.11) 
 Price, subsequent intensive care d code, CPT code 99480, infants weighing 2501 g–5000 g 329.39 335.53 264.17 −71.36
(−90.41 to −52.31) 
−87.37
(−152.56 to −22.19) 
485.68 166.03
(103.82 to 228.25) 
277.62
(194.23 to 361.01) 
 Price, most common subsequent critical and intensive care, CPT codes 99469, 99472, 99478, 99479, 99480 445.04 450.99 382.86 −68.13
(−92.88 to −43.38) 
−135.61
(−216.21 to −55.02) 
659.56 215.88
(8833 to 343.43) 
313.24
(207.00 to 419.48) 
Spending, 95% CI         
 Total physician spending 9150.97 9302.16 7398.77 −1903.38
(−2857.49 to −949.28) 
−1209.37
(−2577.98 to 159.24) 
13 102.06 3931.08 (1300.86 to 6561.31) 5160.67
(3061.73 to 7259.61) 
 Total hospital spending 56 770.05 57 334.28 50 231.08 −7103.20
(−13 066.72 to −1139.68) 
4065.23
(−7186.45 to 15 316.90) 
63247.48 4658.36
(−8181.50 to 17 498.21) 
−5114.58
(−12 032.23 to 1803.06) 
 Standardized physician spendingk 9149.42 9158.33 9046.13 −112.20
(−1186.27 to 961.87) 
246.06
(−641.74 to 1133.87) 
9467.97 432.78
(−1223.86 to 2089.42) 
727.67
(−893.91 to 2349.25) 
Utilization, 95% CI         
 Total length of stay 14.90 14.89 15.01 0.12
(−1.01 to 1.24) 
2.21
(0.07 to 4.35) 
15.31 0.54
(−1.20 to 2.28) 
−0.53
(−1.78 to 0.73) 
 No. of intensive care d 6.59 6.47 6.34 −1.19
(−2.08 to −0.29) 
−0.56
(−2.70 to 1.59) 
8.48 2.00
(0.73 to 3.27) 
−0.15
(−1.86 to 1.55) 
 No. of critical care d 7.43 7.52 8.04 1.58
(1.00 to 2.15) 
2.93
(1.50 to 4.36) 
6.14 −1.23
(−3.00 to 0.52) 
−0.59
(−2.50 to 1.31) 
 No. of imaging services, chest x-rays 1.62 1.60 1.84 0.23
(−0.06 to 0.53) 
0.84
(0.06 to 1.61) 
1.33 −0.32
(−0.64 to −0.00) 
−0.13
(−0.53 to 0.26) 
 No. of imaging services, abdominal films 0.60 0.59 0.62 0.03
(−0.11 to 0.17) 
0.05
(−0.13 to 0.22) 
0.45 −0.18
(−0.34 to −0.02) 
−0.17
(−0.42 to 0.07) 
 No. of imaging services, head MRIs 0.04 0.04 0.03 −0.02
(−0.03 to −0.00) 
−0.02
(−0.03 to −0.01) 
0.04 −0.01
(−0.03 to 0.01) 
0.02
(−0.02 to 0.05) 
 No. of imaging services, head ultrasound 0.37 0.36 0.38 0.02
(−0.05 to 0.09) 
0.12
(−0.02 to 0.26) 
0.33 −0.02
(−0.11 to 0.07) 
−0.08
(−0.14 to −0.01) 
Readmissions and adverse clinical outcomes, 95% CI         
 No. of 7-d readmissions 0.02 0.02 0.01 −0.01
(−0.02 to −0.00) 
0.00
(−0.02 to 0.02) 
0.01 0.00
(−0.00 to 0.01) 
0.01
(−0.00 to 0.02) 
 No. of 14-d readmissions 0.03 0.03 0.02 −0.01
(−0.02 to 0.00) 
0.01
(−0.03 to 0.06) 
0.02 0.01
(−0.01 to 0.02) 
−0.00
(−0.01 to 0.01) 
 No. of 30-d readmissions 0.04 0.05 0.03 −0.01
(−0.03 to 0.00) 
0.03
(−0.03 to 0.09) 
0.04 0.01
(−0.01 to 0.03) 
−0.00
(−0.01 to 0.01) 
 Any necrotizing enterocolitis,l1.77 1.77 1.77 −0.00
(−0.01 to 0.08) 
−0.63
(−1.09 to −0.18) 
1.45 0.74
(0.39 to 1.39) 
0.71
(0.35 to 1.42) 
 Any sepsis,l9.82 9.22 16.91 7.69
(2.49 to 12.89) 
3.37
(−1.98 to 2.66) 
7.24 0.82
(0.38 to 1.76) 
1.84
(0.23 to 14.38) 
 Any intraventricular hemorrhage,l3.94 3.88 4.68 0.81
(−0.37 to 1.99) 
0.76
(−2.02 to 3.54) 
3.30 0.93
(0.71 to 1.21) 
0.75
(0.42 to 1.34) 
 Any hepatitis B vaccination,l6.34 6.39 5.67 −0.72
(−2.70 to 0.76) 
−0.43
(−3.82 to 4.67) 
5.75 1.13
(0.51 to 1.24) 
1.18
(0.70 to 2.00) 
 Any out-of-network physician services,l, m12.80 12.81 12.59 −0.21
(−4.24 to 0.38) 
−5.33
(−0.84 to 11.50) 
12.47 1.07
(0.60 to 1.92) 
0.48
(0.13 to 1.73) 
Preaffiliation PeriodaPostaffiliationa Period
Non-PMC–Affiliated NICUs,e, fMeanPMC-Affiliated NICUs,g MeanUnadjusted Difference Between PMC-Affiliated and Non-PMC–Affiliated NICUshAdjusted Difference Between PMC-Affiliated and Non-PMC–Affiliated NICUshPMC-Affiliated NICUs, MeaneUnadjusted Differential Change From Pre- to Postaffiliation Periodb, cAdjustedd Differential Change From Pre- to Postaffiliation Periodb, c
No. of admissionsi 14 251 13 119 1132 14 251 14 251 1726 95 319 95 319 
Prices,j 95% CI         
 Price, subsequent critical care d, CPT code 99469, infants 28 d or younger 977.49 990.21 851.40 −138.81
(−196.72 to −80.90) 
−224.31
(−376.94 to −71.68) 
1440.47 481.35
(249.64 to 713.05) 
569.04
(283.10 to 854.97) 
 Price, subsequent intensive care d, CPT code 99479, infants weighing 1500 g–2500 g 350.87 357.22 283.52 −73.69
(−95.75 to −51.64) 
−126.75
(−184.33 to −69.16) 
508.81 164.07
(93.98 to 234.16) 
288.29
(194.47 to 382.11) 
 Price, subsequent intensive care d code, CPT code 99480, infants weighing 2501 g–5000 g 329.39 335.53 264.17 −71.36
(−90.41 to −52.31) 
−87.37
(−152.56 to −22.19) 
485.68 166.03
(103.82 to 228.25) 
277.62
(194.23 to 361.01) 
 Price, most common subsequent critical and intensive care, CPT codes 99469, 99472, 99478, 99479, 99480 445.04 450.99 382.86 −68.13
(−92.88 to −43.38) 
−135.61
(−216.21 to −55.02) 
659.56 215.88
(8833 to 343.43) 
313.24
(207.00 to 419.48) 
Spending, 95% CI         
 Total physician spending 9150.97 9302.16 7398.77 −1903.38
(−2857.49 to −949.28) 
−1209.37
(−2577.98 to 159.24) 
13 102.06 3931.08 (1300.86 to 6561.31) 5160.67
(3061.73 to 7259.61) 
 Total hospital spending 56 770.05 57 334.28 50 231.08 −7103.20
(−13 066.72 to −1139.68) 
4065.23
(−7186.45 to 15 316.90) 
63247.48 4658.36
(−8181.50 to 17 498.21) 
−5114.58
(−12 032.23 to 1803.06) 
 Standardized physician spendingk 9149.42 9158.33 9046.13 −112.20
(−1186.27 to 961.87) 
246.06
(−641.74 to 1133.87) 
9467.97 432.78
(−1223.86 to 2089.42) 
727.67
(−893.91 to 2349.25) 
Utilization, 95% CI         
 Total length of stay 14.90 14.89 15.01 0.12
(−1.01 to 1.24) 
2.21
(0.07 to 4.35) 
15.31 0.54
(−1.20 to 2.28) 
−0.53
(−1.78 to 0.73) 
 No. of intensive care d 6.59 6.47 6.34 −1.19
(−2.08 to −0.29) 
−0.56
(−2.70 to 1.59) 
8.48 2.00
(0.73 to 3.27) 
−0.15
(−1.86 to 1.55) 
 No. of critical care d 7.43 7.52 8.04 1.58
(1.00 to 2.15) 
2.93
(1.50 to 4.36) 
6.14 −1.23
(−3.00 to 0.52) 
−0.59
(−2.50 to 1.31) 
 No. of imaging services, chest x-rays 1.62 1.60 1.84 0.23
(−0.06 to 0.53) 
0.84
(0.06 to 1.61) 
1.33 −0.32
(−0.64 to −0.00) 
−0.13
(−0.53 to 0.26) 
 No. of imaging services, abdominal films 0.60 0.59 0.62 0.03
(−0.11 to 0.17) 
0.05
(−0.13 to 0.22) 
0.45 −0.18
(−0.34 to −0.02) 
−0.17
(−0.42 to 0.07) 
 No. of imaging services, head MRIs 0.04 0.04 0.03 −0.02
(−0.03 to −0.00) 
−0.02
(−0.03 to −0.01) 
0.04 −0.01
(−0.03 to 0.01) 
0.02
(−0.02 to 0.05) 
 No. of imaging services, head ultrasound 0.37 0.36 0.38 0.02
(−0.05 to 0.09) 
0.12
(−0.02 to 0.26) 
0.33 −0.02
(−0.11 to 0.07) 
−0.08
(−0.14 to −0.01) 
Readmissions and adverse clinical outcomes, 95% CI         
 No. of 7-d readmissions 0.02 0.02 0.01 −0.01
(−0.02 to −0.00) 
0.00
(−0.02 to 0.02) 
0.01 0.00
(−0.00 to 0.01) 
0.01
(−0.00 to 0.02) 
 No. of 14-d readmissions 0.03 0.03 0.02 −0.01
(−0.02 to 0.00) 
0.01
(−0.03 to 0.06) 
0.02 0.01
(−0.01 to 0.02) 
−0.00
(−0.01 to 0.01) 
 No. of 30-d readmissions 0.04 0.05 0.03 −0.01
(−0.03 to 0.00) 
0.03
(−0.03 to 0.09) 
0.04 0.01
(−0.01 to 0.03) 
−0.00
(−0.01 to 0.01) 
 Any necrotizing enterocolitis,l1.77 1.77 1.77 −0.00
(−0.01 to 0.08) 
−0.63
(−1.09 to −0.18) 
1.45 0.74
(0.39 to 1.39) 
0.71
(0.35 to 1.42) 
 Any sepsis,l9.82 9.22 16.91 7.69
(2.49 to 12.89) 
3.37
(−1.98 to 2.66) 
7.24 0.82
(0.38 to 1.76) 
1.84
(0.23 to 14.38) 
 Any intraventricular hemorrhage,l3.94 3.88 4.68 0.81
(−0.37 to 1.99) 
0.76
(−2.02 to 3.54) 
3.30 0.93
(0.71 to 1.21) 
0.75
(0.42 to 1.34) 
 Any hepatitis B vaccination,l6.34 6.39 5.67 −0.72
(−2.70 to 0.76) 
−0.43
(−3.82 to 4.67) 
5.75 1.13
(0.51 to 1.24) 
1.18
(0.70 to 2.00) 
 Any out-of-network physician services,l, m12.80 12.81 12.59 −0.21
(−4.24 to 0.38) 
−5.33
(−0.84 to 11.50) 
12.47 1.07
(0.60 to 1.92) 
0.48
(0.13 to 1.73) 

Authors’ analysis of HCCI data, 2011 to 2017. PMC-affiliated NICUs were compared with non-PMC–affiliated NICUs, before and after PMC affiliations. The sample includes infants admitted to PMC-affiliated and non-PMC–affiliated NICUs between January 1, 2011, and December 31, 2017, excluding admissions where the infant weighed <500 g.

a

The postaffiliation period in this table is defined as after the start of PMC affiliation for PMC-affiliated NICUs, and the preaffiliation is before the start of PMC affiliation.

b

The differential estimate indicates the change from the pre- to postaffiliation period for PMC-affiliated NICUs relative to non-PMC–affiliated NICUs that were never affiliated with a PMC. Year fixed effects are included.

c

The study sample includes all non-PMC–affiliated NICUs and all PMC-affiliated NICUs present in both pre- and postaffiliation years.

d

The analyses adjusted for hospital fixed effects, birth weight category, the sex of the infant, whether the delivery included multiple births, whether the infant was delivered via cesarean delivery, whether major procedures were conducted during the admission, whether the infant had a congenital anomaly, and whether the infant was transferred.

e

The unweighted means for outcome measures were calculated for both non-PMC–affiliated NICU means at the midpoint of the study period, 2014, following the method in Beaulieu et al (2020),7  and PMC-affiliated NICU means during the preaffiliation period.

f

Non-PMC–affiliated PMCs were all NICUs that were not found to be affiliated with a PMC at any point in time.

g

PMC-affiliated NICUs became PMC-affiliated between July 1, 2011, and June 30, 2017.

h

Preaffiliation differences between PMC-affiliated and non-PMC–affiliated NICUs reflect the mean for PMC-affiliated NICUs minus the mean for non-PMC–affiliated NICUs.

i

Column 1 includes a count of all admissions of non-PMC–affiliated NICUs in 2014 and total admissions to PMC-affiliated NICUs in preaffiliation years. Column 2 includes a count of all admissions to non-PMC–affiliated NICUs in 2014. Column 3 includes a count of all admissions to PMC-affiliated NICUs in preaffiliation years. Columns 4 and 5 contain total admissions in columns 2 and 3. Column 6 includes all admissions to PMC-affiliated NICUs in postaffiliation years. Columns 7 and 8 contain total admissions for all non-PMC–affiliated and PMC-affiliated NICUs across all study years. Given that affiliation occurred in different years for different NICUs, the number of pre- and postaffiliation years and pre- and postaffiliation admissions differs across PMCs.

j

The most frequently billed CPT codes were used to construct price outcomes. Because these CPT codes were not billed in all admissions, the number of admissions included in the analyses varies by price outcome.

k

Admission-level standardized physician spending outcomes were calculated by replacing actual prices paid to neonatologists with the national mean price for each of the CPT codes from 2011 to 2017. Averaged prices were then aggregated over each admission to construct the standardized spending outcomes.

l

Odds ratios for these binary outcome variables are shown in the unadjusted and adjusted differential change from pre- to postaffiliation period.

m

An out-of-network physician service is provided by an out-of-network physician for a NICU service delivered at an in-network hospital. This indicator was available in the HCCI data from 2014 to 2017.

Prices

Prices for subsequent critical and intensive care days were higher in NICUs in all years after PMC affiliation. Across all postaffiliation years, prices increased by $569 per day (95% confidence interval [CI], $283–$855) for subsequent critical care days after PMC affiliation in PMC-affiliated NICUs relative to non-PMC–affiliated NICUs (Table 2), and by $288 per day (95% CI, $194 to $382) and $278 per day (95% CI, $194 to $361) for subsequent intensive care days (Table 2). This represents 58.2%, 82.2%, and 84.3% increases, respectively, relative to the unadjusted mean prices for PMC-affiliated and non-PMC–affiliated NICUs in the preaffiliation period ($977, $351, and $329). The largest increase occurred in the first and second years after PMC affiliation (Fig 1). Average prices across the 5 most common types of subsequent critical and intensive care days in PMC-affiliated NICUs increased by $313 per day (95% CI, $207 to $419) relative to price increases in non-PMC–affiliated NICUs across all postaffiliation years, a 70.4% increase.

Utilization

There were no differential changes in any of the utilization measures after PMC affiliation (Table 2, Fig 2), including total length of stay, number of intensive days, and number of critical care days, compared with changes in non-PMC–affiliated NICUs. There were also no changes in utilization of imaging services except for a very small decrease in head ultrasounds (Table 2, Supplemental Fig 4), compared with changes in non-PMC–affiliated NICUs. PMC affiliation was not associated with the number of normal-weight or all newborns admitted to NICUs (Supplemental Fig 5).

Spending

NICU affiliations with PMCs were associated with a statistically significant differential increase of $5161 per NICU stay (95% CI, $3062 to $7260) in total spending for physician services during the postaffiliation period, representing a 56.4% increase relative to the preaffiliation period (Table 2, Fig 3). There was no change in standardized total spending for physician services, suggesting that total spending increases were because of higher prices rather than higher volume or intensity (Table 2, Supplemental Fig 4).

Postaffiliation, hospital spending decreased by $5115 per NICU stay (9% of mean spending in the preaffiliation period; not statistically significant) during the postaffiliation period relative to the preaffiliation period.

Readmissions and Clinical Outcomes

PMC affiliation was not associated with differential changes in 7-day, 14-day, or 30-day hospital readmissions, hepatitis B vaccination, or the probability of necrotizing enterocolitis, intraventricular hemorrhage, or late-onset sepsis (Table 2, Supplemental Fig 6).

Out-of-Network Billing

There were no changes in the likelihood of out-of-network physician claims for PMC-affiliated NICUs compared with non-PMC–affiliated NICUs (Table 2, Supplemental Fig 7).

Infants Admitted to Level II or III NICUs

Among infants admitted to level II or III NICUs, there was no relationship between PMC affiliation and utilization, spending, or readmissions (Supplemental Fig 8).

Sensitivity Analyses

After including HRR fixed effects to account for hospital market factors, the differential change in prices were similar in magnitude to our main findings (Supplemental Table 5). In the linked sample of infants and mothers, differential price increases associated with PMC affiliation were larger in magnitude compared with results from the full study sample, suggesting that including additional risk factors may have improved the adjustment for patient clinical risk between PMC-affiliated and non-PMC–affiliated NICUs (Supplemental Fig 9). The results were otherwise qualitatively similar across primary spending, utilization, and readmissions outcomes. In the sample that excluded transferred infants, changes in primary outcomes were similar to those in the full study sample (Supplemental Fig 10).

Although price and spending results were robust to Chaisemartin and D’haultfoeuille adjustments (Supplemental Fig 11), PMC affiliation was associated with small increases in 7-day and 14-day readmissions in the second (14-day readmissions, 0.03; 95% CI, 0.01 to 0.06, 7-day readmissions, 0.03; 95% CI, 0.02 to 0.04) and third years after the affiliation year (14-day readmissions, 0.03; 95% CI, 0.01–0.07, 7-day readmissions, 0.01; 95% CI, 0.01 to 0.02). Given that 2.6% of neonates, among a national sample, were readmitted within 8 to 30 days during the same time period as our study,27  our findings suggest that PMC affiliation is associated with moderate increases in readmissions in some postaffiliation years.

Tests incorporating missing data yielded similar estimates as our primary analyses for all outcomes (Supplemental Tables 7–10). Finally, price, spending, and utilization results remained qualitatively unchanged in quantile regression and Poisson models (Supplemental Table 11).

PMC-NICU affiliation was associated with a 70.4% mean differential increase in prices paid to PMCs by insurers across the 5 most common types of critical and intensive care days in PMC-affiliated NICUs compared with price increases in non-PMC–affiliated NICUs.

These price increases are consistent with previous findings on PMC staffing of emergency physicians and of anesthesiologists in outpatient departments, which found price increases of 122% and 17%, respectively.4,8  These hospital-based specialties may benefit from lack of competition in hospital markets and, therefore, moving physicians out of network or the threat of doing so may affect insurer–PMC negotiations and exert upward pressure on prices.4,5,22  These findings also contribute to a broader literature on institutional investors in medical care, where private equity acquisitions in dermatology are associated with higher prices for routine medical visits.28 

Advocates for PMCs argue that insurers often have sufficient negotiating leverage to underpay physicians, justifying price increases negotiated by PMCs. However, the extent, if any, to which PMCs share revenue from higher prices with employed neonatologists is unknown. In addition, PMC physician contracts may include strict noncompete agreements that limit physicians’ employment opportunities.29 

Insurers argue that higher prices affect patients in 2 ways: first, through higher insurance premiums, and second, through higher patient cost-sharing, which can be quite substantial for parents whose infant requires NICU care.30,31  We found that spending on physician services increased by $5161 per NICU stay after PMC affiliation compared with price changes in nonaffiliated NICUs, whereas spending on hospital services decreased by $5115. However, the change in hospital spending, and the implications for overall spending should be interpreted with caution because the decrease in hospital spending was not statistically significant and was relatively small, 9% of hospital spending.

We also found no overall statistically significant relationship between PMC affiliation and readmissions. However, PMC affiliation was associated with moderate increases in 7-day and 14-day readmissions in the second and third postaffiliation years, in sensitivity analyses.

First, although our paper focused on 2 major neonatology PMCs that supply at least 25% of all neonatology services nationally, our findings may not generalize to smaller PMCs.12  Second, for transferred infants, our analyses only included claims from the NICU where the infant spent the majority of their NICU days. However, results were similar when we excluded transfers. Third, we were only able to include a broader set of risk factors in the mother–infant linked sample.

Fourth, to the extent that we failed to correctly categorize NICUs as PMC-affiliated NICUs (PMC-NICU contract terminations are extremely difficult to identify), these NICUs would be in the non-PMC–affiliated group or excluded from analyses, potentially biasing our results toward the null. Fifth, our sample is limited to patients of 3 large insurers represented in HCCI data. However, HCCI data cover 50 million individuals nationally. Sixth, HCCI data do not include the ∼43% of births financed by Medicaid.32  Seventh, some PMC-NICU affiliations were not included because they became affiliated before or after the study period. Eighth, our data set does not include a mortality indicator before 2014, so we were unable to measure neonatal mortality. Finally, claims data provides only a limited assessment of the quality of care. The adverse clinical outcomes we measured are extremely serious events, and high-quality care will likely lead to lower rates of these events.33  However, these events occurred infrequently, and our relatively small study sample might fail to detect differences between PMC-affiliated and non-PMC–affiliated NICUs.

PMC affiliation of NICUs was associated with large increases in prices and spending for physician services in the NICU, but not with changes in utilization, readmissions, or clinical outcomes, though this study’s ability to detect adverse outcomes was limited. Future work should examine process of care indicators and neonatal mortality. These findings should be of interest to neonatologists, obstetricians, hospital and health insurance executives, parents, researchers, and antitrust agencies.

We thank Dr Audrone La Forgia, Dr Pam Russell, and Dr Gerald M. Loughlin for valuable discussions on neonatology. We thank Cheryn Roo and Shruti Kanna for their contributions to the literature review.

Dr Yu conceptualized and designed the study, conducted the data analyses, drafted the manuscript, and reviewed and revised the manuscript; Dr Braun conceptualized and designed the study, conducted the initial data collection, coordinated and supervised data collection, and reviewed and revised the manuscript; Dr Bond conceptualized and designed the study, critically reviewed the data analyses, and reviewed and revised the manuscript; Dr La Forgia conceptualized and designed the study, critically reviewed the data analyses, and reviewed and revised the manuscript; Dr RoyChoudhury critically reviewed the manuscript for important intellectual content, and revised the manuscript; Ms Zhang conducted the initial data analyses, and reviewed and revised the manuscript. Ms Kim completed the data collection and reviewed and revised the manuscript; Dr Casalino conceptualized and designed the study, supervised the data analyses, and reviewed and revised the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

The views presented here are those of the authors and not necessarily those of the Physicians Foundation or Arnold Ventures. Neither organization had a role in the design or interpretation of the study. The authors acknowledge the assistance of the Health Care Cost Institute and its data contributors, Aetna, Humana, and UnitedHealthcare, in providing the claims data analyzed in this study.

COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2022-060788.

FUNDING: Supported by the Physicians Foundation Center for the Study of Physician Practice and Leadership at Weill Cornell Medical College and by Arnold Ventures. The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

CI

confidence interval

CPT

Current procedural terminology

HCCI

Health Care Cost Institute

HRR

hospital referral regions

PMC

physician management company

1.
Barkholz
D
.
Herding hospital docs: staffing firms buy MD groups
.
2.
Barkholz
D
.
Physician staffing firm Envision wants to be hospitals’ ‘one throat to choke’
.
3.
Mednax National Medical Group
.
Neonatal care services
.
Available at: ttps://www.mednax.com/hospitals/solutions/neonatal/. Accessed March 10, 2022
4.
Cooper
Z
,
Nguyen
H
,
Shekita
N
,
Morton
FS
.
Out-of-network billing and negotiated payments for hospital-based physicians
.
Health Aff (Millwood)
.
2020
;
39
(
1
):
24
32
5.
Burns
LR
.
Physician practice management companies
.
Health Care Manage Rev
.
1997
;
22
(
4
):
32
46
6.
Frack
B
,
Hong
N
.
Physician practice management–a new chapter
.
7.
Beaulieu
ND
,
Dafny
LS
,
Landon
BE
, %
Dalton
JB
,
Kuye
I
,
McWilliams
JM
.
Changes in quality of care after hospital mergers and acquisitions
.
N Engl J Med
.
2020
;
382
(
1
):
51
59
8.
La Forgia
A
,
Bond
AM
,
Braun
RT
, et al
.
Association of Physician Management Companies and private equity investment with commercial health care prices paid to anesthesia practitioners
.
JAMA Intern Med
.
2022
;
182
(
4
):
396
404
9.
The Dartmouth Institute
.
The Dartmouth Atlas of Neonatal Intensive Care
.
Available at: https://www.dartmouthatlas.org/Neonatal_Atlas_090419.pdf. Accessed December 18, 2021
10.
Lantos
J
.
Cruel calculus: why saving premature babies is better business than helping them thrive
.
Health Aff (Millwood)
.
2010
;
29
(
11
):
2114
2117
11.
Bloomberg
.
KKR will buy Envision Healthcare for $5.57 billion plus debt
.
12.
Pediatrix Medical Group
.
Caring for NICU babies: it takes a village
.
13.
HCCI
.
Commercial data
.
Available at: https://healthcostinstitute.org/data. Accessed May 28, 2022
14.
Goodman
DC
,
Ganduglia-Cazaban
C
,
Franzini
L
, et al
.
Neonatal intensive care variation in Medicaid-insured newborns: a population-based study
.
J Pediatr
.
2019
;
209
:
44
51.e2
15.
Brumbaugh
JE
,
Hansen
NI
,
Bell
EF
, et al.
National Institute of Child Health and Human Development Neonatal Research Network
.
Outcomes of extremely preterm infants with birth weight less than 400 g
.
JAMA Pediatr
.
2019
;
173
(
5
):
434
445
16.
American Academy of Pediatrics Committee on Fetus And Newborn
.
Levels of neonatal care
.
Pediatrics
.
2012
;
130
(
3
):
587
597
17.
Unruh
MA
,
Zhang
Y
,
Jung
H-Y
, et al
.
Physician prices and the cost and quality of care for commercially insured patients
.
Health Aff (Millwood)
.
2020
;
39
(
5
):
800
808
18.
Liechty
EA
,
Hughes
C
,
Dolan
B
.
Coding for Pediatrics: A Manual for Pediatric Documentation and Payment for Use With AMA CPT 2017
.
Elk Grove Village, IL
:
American Academy of Pediatrics
;
2017
19.
Lorch
SA
,
Passarella
M
,
Zeigler
A
.
Challenges to measuring variation in readmission rates of neonatal intensive care patients
.
Acad Pediatr
.
2014
;
14
(
5 Suppl
):
S47
S53
20.
Centers for Disease Control and Prevention
.
Hepatitis B vaccine coverage among all live newborn infants prior to hospital or birthing facility discharge (NQF 0475)
.
21.
Ganapathy
V
,
Hay
JW
,
Kim
JH
,
Lee
ML
,
Rechtman
DJ
.
Long term health care costs of infants who survived neonatal necrotizing enterocolitis: a retrospective longitudinal study among infants enrolled in Texas Medicaid
.
BMC Pediatr
.
2013
;
13
(
1
):
127
22.
Edwards
EM
,
Ehret
DEY
,
Soll
RF
,
Horbar
JD
.
Vermont Oxford Network: a worldwide learning community
.
Transl Pediatr
.
2019
;
8
(
3
):
182
192
23.
Goodman-Bacon
A
.
Difference- in-differences with variation in treatment timing
.
J Econom
.
2021
;
225
(
2
):
254
277
24.
de Chaisemartin
C
,
D’Haultfoeuille
X
.
Two-way fixed effects estimators with heterogeneous treatment effects
.
Am Econ Rev
.
2020
;
110
(
9
):
2964
2996
25.
Roth
J
,
Sant’Anna
PH
,
Bilinski
A
,
Poe
J
.
What’s trending in difference-in- differences? A synthesis of the recent econometrics literature
.
arXiv preprint. 2022;arXiv:2201.01194
26.
Michal
MH
,
Pekarske
MSL
,
McManus
MK
.
Corporate practice of medicine 50 state summary
.
27.
Bernardo
J
,
Keiser
A
,
Aucott
S
,
Yanek
LR
,
Johnson
CT
,
Donohue
P
.
Early readmission following NICU discharges among a national sample: associated factors and spending [Published online ahead of print October 11, 2021]
.
Am J Perinatol
.
10.1055/s-0041-1736286
28.
Braun
RT
,
Bond
AM
,
Qian
Y
,
Zhang
M
,
Casalino
LP
.
Private equity in dermatology: effect on price, utilization, and spending
.
Health Aff (Millwood)
.
2021
;
40
(
5
):
727
735
29.
United States Securities and Exchange Commission
.
Mednax, Inc
.
30.
Nather
D
.
Axios
.
TeamHealth letter
.
31.
Pifer
R
.
UnitedHealthcare to appeal after jury awards TeamHealth $60M in damages
.
32.
Medicaid and CHIP Payment and Access Commission
.
Medicaid’s role in financing maternity care
.
33.
Webbe
JWH
,
Duffy
JMN
,
Afonso
E
, et al
.
Core outcomes in neonatology: development of a core outcome set for neonatal research
.
Arch Dis Child Fetal Neonatal Ed
.
2020
;
105
(
4
):
425
431

Competing Interests

POTENTIAL CONFLICT OF INTEREST: Dr Casalino reported receiving research funding from the American Board of Family Medicine Foundation and received honoraria as a member of the American Medical Association Advisory Committee on Physician Professional Satisfaction and Practice Sustainability. The other authors have indicated they have no conflicts of interest relevant to this article to disclose.

Supplementary data