For analyses used in this study, the single dotted line represents the beginning of the 33-month pre–Dependent Coverage Provision (DCP) period (January 1, 2008, to September 31, 2010), designed to be of equal length to the 33-month post-DCP/preexpansion period that ran from January 1, 2011, to September 31, 2013. The 21-month postexpansion period ran from January 1, 2014, to September 31, 2015. The first set of vertical dotted lines represents introduction of the DCP on September 23, 2010, through phase-in ending on January 1, 2011. The second set of vertical dotted lines represents full rollout of the Patient Protection and Affordable Care Act (ACA) in Maryland, including Medicaid expansion and the first open enrollment period, beginning on October 1, 2013, and extending through January 1, 2014. “Other coverage” included primary insurance payers such as charity care or international patients. “Other public sponsored coverage” included insurance coverage through programs like Medicare and Title V. Q indicates quarter.
Changes in the proportions of patients who were uninsured (A), Medicaid sponsored (B), privately insured (C), and discharged to rehabilitation (D) among survivors from January 1, 2008, to September 31, 2015. The first set of vertical dotted lines represents introduction of the Dependent Coverage Provision on September 23, 2010, through phase-in ending on January 1, 2011. The second set of vertical dotted lines represents full rollout of the Patient Protection and Affordable Care Act in Maryland, including Medicaid expansion and the first open enrollment period, beginning on October 1, 2013, and extending through January 1, 2014. Time during each of the corresponding quarters (Qs) was excluded as a model lag.
eTable 1. Average Changes in Insurance Coverage and Discharge to Rehabilitation Among DCP-Eligible (18-25y) and DCP-Ineligible (26-34y) Young Adult Trauma Patients
eTable 2. Interrupted Time-Series Analysis Results of ACA-Related Changes Among Young Adult (18-34y) Trauma Patients
eTable 3. Trauma Center Categorization in Maryland, Adapted From the Maryland Institute for Emergency Medical Services Systems (MIEMSS) 2014-2015 Annual Report
eFigure 1. Changes in Primary Insurance Coverage Among Young Adult (18-34y) Trauma Patients Managed in the Emergency Department and as Medical/Surgical Outpatients From 2012-2015
eFigure 2. Unadjusted Variations in Median Total Hospital Charge per Patient, Stratified by Patients’ Primary Insurance Payer, Among Young Adult, 18-34y, Trauma Patients From January 2005-September 2015
eFigure 3. Unadjusted Variations in the Volume of Total Inpatient Admissions, Stratified by Patients’ Primary Insurance Payer, Contributed by Young Adult, 18-34y, Trauma Patients From January 2005-September 2015
eFigure 4. Unadjusted Variations in Total Hospital Charges, Stratified by Patients’ Primary Insurance Payer, Contributed by Young Adult, 18-34y, Trauma Patients From January 2005-September 2015
Zogg CK, Payró Chew F, Scott JW, Wolf LL, Tsai TC, Najjar P, Olufajo OA, Schneider EB, Haut ER, Haider AH, Canner JK. Implications of the Patient Protection and Affordable Care Act on Insurance Coverage and Rehabilitation Use Among Young Adult Trauma Patients. JAMA Surg. 2016;151(12):e163609. doi:10.1001/jamasurg.2016.3609
To what extent were there observed changes in insurance coverage and rehabilitation use among young adult (aged 18-34 years) trauma patients in Maryland following introduction of the Patient Protection and Affordable Care Act?
Longitudinal assessment of Dependent Coverage Provision and Medicaid expansion/open enrollment implementation was conducted using risk-adjusted before-and-after, difference-in-difference, and interrupted time-series analyses among 69 507 hospitalized patients. The study found an 18.2 percentage-point reduction in the percentage of uninsured patients that was primarily driven by Medicaid and a 5.4 percentage-point increase (60% relative increase) in the use of rehabilitation.
For patients who are injured, young, and uninsured, Medicaid expansion/open enrollment has changed insurance coverage and altered patient outcomes.
Trauma is the leading cause of death and disability among young adults, who are also among the most likely to be uninsured. Efforts to increase insurance coverage, including passage of the Patient Protection and Affordable Care Act (ACA), were intended to improve access to care and promote improvements in outcomes. However, despite reported gains in coverage, the ACA’s success in promoting use of high-quality care and enacting changes in clinical end points remains unclear.
To assess for observed changes in insurance coverage and rehabilitation use among young adult trauma patients associated with the ACA, including the Dependent Coverage Provision (DCP) and Medicaid expansion/open enrollment, and to consider possible insurance and rehabilitation differences between DCP-eligible vs -ineligible patients and among stratified demographic and community subgroups.
Design, Setting, and Participants
A longitudinal assessment of DCP implementation and Medicaid expansion/open enrollment using risk-adjusted before-and-after, difference-in-difference, and interrupted time-series analyses was conducted. Eleven years (January 1, 2005, to September 31, 2015) of Maryland Health Services Cost Review Commission data, representing complete patient records from all payers within the state, were used to identify all hospitalized young adult (aged 18-34 years) trauma patients in Maryland during the study period.
Of the 69 507 hospitalized patients included, 50 548 (72.7%) were male, and the mean (SD) age was 25 (5) years. Before implementation of the DCP, 1 of 4 patients was uninsured. After ACA implementation, the number fell to less than 1 of 10, with similar patterns emerging in emergency department and outpatient settings. The change was primarily driven by Medicaid expansion/open enrollment, which corresponded to a 20.1 percentage-point increase in Medicaid (95% CI, 18.9-21.3) and an 18.2 percentage-point decrease in uninsured (95% CI, −19.3 to −17.2). No changes were detected among privately insured patients. Rehabilitation use increased by 5.4 percentage points (95% CI, 4.5-6.2)—a 60% relative increase from a baseline of 9%. Mortality (−0.5; 95% CI, −0.9 to −0.1) and failure-to-rescue rates (−4.5; 95% CI, −7.4 to −1.6) also significantly declined. Stratified changes point to significant differences in the percentage of uninsured patients and rehabilitation access across the board, mitigating or even eradicating disparities in certain cases.
Conclusions and Relevance
For patients who are injured, young, and uninsured, Medicaid expansion/open enrollment in Maryland changed insurance coverage and altered patient outcomes in ways that the DCP alone was never intended to do. Implementation of Medicaid expansion/open enrollment transformed the landscape of trauma coverage, directly affecting the health of one of the country’s most vulnerable at-risk groups.
Trauma is the leading cause of death and disability among young adults (aged 18-34 years), accounting for more than 47 500 deaths and 8.7 million nonfatal injuries within this group in the United States each year.1 The long-term consequences can be devastating. To lessen the occurrence of such consequences and promote improved functional outcomes,2- 4 access to rehabilitation after acute traumatic injury is considered an essential component of high-quality trauma care.5,6 For many young trauma patients, “rehabilitation is the final common pathway that may define the success or failure of their treatment [and] ultimate outcome.”7(p822) Patients’ ability to use rehabilitation is an important predictor of long-term outcomes of care.2- 4,7 Nevertheless, despite its known benefits, rehabilitation may not be available to all patients.8,9 Studies have demonstrated strong associations between insurance coverage and rehabilitation use8,10- 12 as well as between a lack of insurance and fatal outcomes, including in-hospital mortality13- 15 and failure to rescue (FTR).16 Defined as mortality after major complication, both FTR and mortality are insurance-sensitive markers of hospitals’ quality of care, thought to be influenced by insurance-related changes in hospitals’ financial and resource availability and other unmeasured patient-level effects (eg, family resources) needed to promote survival and timely care.13- 16
Concerns about outcomes are particularly pronounced among young adult trauma patients, who are also among the most likely to be uninsured.17,18 Efforts to increase insurance coverage, including passage of the Patient Protection and Affordable Care Act (ACA), were intended to improve access to care and, thereby, promote improvements in outcomes; however, despite reported gains in coverage,19,20 the ACA’s success in promoting use of high-quality care and enacting changes in clinical end points remains unclear.
Implementation of the ACA’s Dependent Coverage Provision (DCP) in 2010 enabled individuals whose parents had employer-sponsored private health insurance plans to remain on their parents’ insurance until age 26 years. The provision resulted in uneven insurance gains21,22 and a lack of changes in outcomes21 for young adult trauma patients despite declines in emergency department (ED) use23- 25 and increases in preventive services use.26 Predictions of how subsequent provisions, including expansion of Medicaid eligibility to 133% of the federal poverty level, introduction of state health insurance exchanges, and provision of income-based tax subsidies, could affect trauma patients and the viability of trauma hospitals are contentiously mixed.27- 32
The objective of this study was to assess for observed changes in insurance coverage and rehabilitation use among young adult trauma patients associated with the ACA, including introduction of the DCP and Medicaid expansion/open enrollment. We considered possible differences between DCP-eligible vs -ineligible patients and among stratified demographic and community subgroups. Variations in mortality, FTR, and total hospital charges (2016 US dollars) were additionally assessed.
Eleven years of data (January 1, 2005, to September 31, 2015) from the Maryland Health Services Cost Review Commission,33 representing complete patient records from all payers within the state, were queried for hospital encounters consistent with young adult trauma patients: International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis codes 800.x-959.x. Patients with late effects of injury, toxicity, or poisoning (codes 905.0-924.9), foreign body injuries (codes 930.0-930.9), or burns (codes 940.0-949.5) were excluded as were those missing information for covariates of interest (<1% of patients). The Johns Hopkins University School of Medicine Institutional Review Board approved the study and waived the need for informed consent. Data released by the Health Services Cost Review Commission are deidentified.
Patients were categorized according to primary insurance during inpatient hospitalization—private, Medicaid, uninsured, other public (eg, Medicare, Title V), other (eg, charity, international)—and by whether they were discharged to rehabilitation. Discharge to specific types of rehabilitation, including inpatient rehabilitation facilities, skilled nursing facilities, and home health agencies, was also considered. Changes in rehabilitation use and associated changes in insurance coverage served as the primary outcome measures for the study.
Patients were grouped according to age into those eligible (aged 18-25 years) vs ineligible (aged 26-34 years) for the DCP and by period of calendar time (Figure 1). Patients admitted during the 33 months before DCP implementation (January 1, 2008, to September 31, 2010) were used as a reference. The following quarter was excluded as a washout period. Patients admitted during the 33 months after DCP implementation but before Medicaid expansion in Maryland and the first open-enrollment period (January 1, 2011, to September 31, 2013) composed the post-DCP/preexpansion group. The following quarter was excluded as a washout period. Finally, patients admitted during the 21 months after Medicaid expansion/open enrollment (January 1, 2014, to September 31, 2015) represented the postexpansion group.
Differences in demographic, clinical, and community covariates were compared across periods using χ2 tests for categorical variables and Kruskal-Wallis tests for nonnormally distributed continuous length of stay. Covariates included length of inpatient stay, trauma center (TC) level (primary adult resource center [R. Adams Cowley Shock Trauma Center in Baltimore, Maryland], level I, level II, level III, and nontrauma center),34 sex, race/ethnicity (non-Hispanic white, non-Hispanic black, and other), marital status (single, married, and other), Charlson Comorbidity Index score (0 and ≥1 comorbidity), Injury Severity Score (ISS) (minor, <9; severe, 9-15; and major, >15), presence of severe head injury (maximum head Abbreviated Injury Scale score 0-3 [no severe head injury] and >3 [severe head injury]), mechanism of injury (motor vehicle collision, gunshot wound, other penetrating, struck or had a fall, and other), residential city or town population (<100 000 and ≥100 000 people), annual residential county unemployment rate (<5, 5-7, and >7%), and annual median income of residential zip code in 2016 US dollars (<$25 000, $25 000-$50 000, $50 001-$80 000, and >$80 000). For TC assignment, transfer patients were considered to have been managed at the hospital that provided their highest level of care. Charlson Comorbidity Index, ISS, Abbreviated Injury Scale, and mechanism of injury were calculated based on ICD-9-CM diagnosis and E-codes.35
Changes in the proportion of patients who were uninsured, Medicaid sponsored, privately insured, and discharged to rehabilitation were compared before and after Medicaid expansion/open enrollment using multivariable linear regression. Variations in mortality and FTR (Table 1) were also assessed. In-hospital acute care surgery-related complications used in FTR were calculated based on ICD-9-CM codes for pneumonia, pulmonary embolism, renal failure, cardiovascular accident, myocardial infarction, cardiac arrest, acute respiratory distress syndrome, and sepsis. Risk-adjusted models accounted for clustering of patients within hospitals and relied on robust SEs. Risk adjustment covariates included age, length of inpatient stay, trauma center level, sex, race/ethnicity, marital status, Charlson Comorbidity Index, ISS, severe head injury, mechanism of injury, residential city or town population, county unemployment rate, median income of residential zip code, and season. Before-and-after differences were compared among young adults and, for uninsured and rehabilitation use, among stratified subgroups of demographic and community factors. Changes in nonnormally distributed charges were compared using quantile regression (25th, 50th, and 75th).
Differences between DCP-eligible and -ineligible patients were further compared before and after DCP implementation and before and after Medicaid expansion/open enrollment using risk-adjusted difference-in-difference (DID) models to ascertain whether policy-related changes were similar between the 2 groups. DID models have been used21,22 to study DCP-related changes among young adult trauma patients. DID models are identical to before-and-after linear models except for the inclusion of a policy eligibility variable (age-group) and an interaction term (period × age group) to assess for effect modification corresponding to intervention eligibility and the association between time period and insurance or time period and rehabilitation. Differences are determined by the significance of the interaction term:
E(Y) = βperiod + βage group + βperiod × age group + βcovariates,
with E(Y) indicating the expected value of the outcome variable Y.
To account for the possibility of temporal changes within time periods, interrupted time-series analyses were also performed. Interrupted time-series analysis is a quasi-experimental regression design that uses longitudinal data to model temporal changes while accounting for preintervention trends.36- 38 Interrupted time-series analysis functions by fitting linear models to values in preintervention and postintervention periods and assessing for the magnitude of deviation in the model’s intercept from preintervention levels in the postintervention period. Phase-in lags can be incorporated. Deviations from expected values at indicated interruptions are considered a result of the intervention. Interruptions were added in January 2011 and 2014, corresponding to the DCP and Medicaid expansion/open enrollment. Data from 3 months earlier were excluded as lags.
Changes among young adult trauma patients managed within the ED and as medical or surgical outpatients from 2012 to 2015 were also examined to ascertain whether the observed changes were isolated to hospitalized patients (ie, whether the changes were likely to have occurred after presentation but before inpatient admission within the ED). Statistical analyses were conducted using Stata, version 14.1 (StataCorp LP). Models were checked for the robustness of explanatory parameters against alternative specifications. Multicolinearity and appropriateness of fit were also assayed. Two-sided P values <.05 were considered significant.
A total of 69 507 patients were included (January 1, 2008, to September 31, 2015, excluding washout-periods: 43 696), of whom 19 216 patients were admitted during the 33-month pre-DCP period (mean [SD] of 1747  patients per quarter), 16 932 during the 33-month post-DCP/preexpansion period (1539  patients per quarter), and 7548 during the 21-month postexpansion period (1078  patients per quarter). A total of 50 548 (72.7%) of the patients were male, and the mean (SD) age was 25 (5) years. Distribution of covariates remained relatively consistent across study periods (Table 1).
Unadjusted changes in insurance are presented in Figure 1. Before DCP implementation, private insurance was the most common among young adult trauma patients, remaining steady at approximately 40%. Uninsured and Medicaid patients each represented an additional 20% to 30%, with other forms of insurance and other types of public-sponsored coverage making up the remainder. Following implementation of the DCP, insurance remained largely unchanged. However, after Medicaid expansion/open enrollment, a large reduction in uninsured and an increase in Medicaid coverage occurred.
Risk-adjusted changes in insurance, discharge to rehabilitation among surviving patients, and secondary end points (in-hospital mortality, FTR, and total hospital charges) following Medicaid expansion/open enrollment are presented in Table 2. Medicaid expansion/open enrollment was associated with an 18.2 percentage-point decrease in uninsured (95% CI, −19.3 to −17.2; baseline, 24.5%), 20.1 percentage-point increase in Medicaid (95% CI, 18.9-21.3; baseline, 27.3%), and 5.4 percentage-point increase in rehabilitation (95% CI, 4.5-6.2; baseline, 9%) among young adult trauma patients aged 18 to 34 years: relative changes of −74.3%, 73.6%, and 60%, respectively. Use of inpatient rehabilitation facilitates increased by 3.3 percentage points (95% CI, 2.7-4; baseline, 6%), and use of home-health agencies increased by 1.1 percentage points (95% CI, 0.6-1.7; baseline, 3%). Fewer than 100 young adults were discharged to skilled nursing facilities.
Mortality declined by 25%, an absolute 0.5 percentage-point drop (95% CI, −0.1 to −0.9) from a baseline of 2%. Mortality declined by an absolute value of 1 percentage point among patients with ISS greater than 8 (baseline 5%). Failure to rescue dropped by 29.8% or 4.5 percentage points (95% CI, −1.6 to −7.4; baseline, 15.1%). Charges increased by a risk- and inflation-adjusted median of $1900 (95% CI, $1400-$2400) per patient. The change was significant at the 25th ($1200) and 75th ($2600) quantiles.
Stratification by DCP eligibility (eTable 1 in the Supplement) revealed that, among patients who were eligible for extended parental insurance coverage (aged 18-25 years), a significant 3.8 percentage-point drop in uninsured (95% CI, −5 to −2.7; baseline, 23.7%) corresponded to a 4.8 percentage-point increase in private coverage (95% CI, 3.5-6.1; baseline, 41.9%) without a change in Medicaid following introduction of the DCP. This was contrasted by a non–DCP-related 5.4 percentage-point drop in private coverage, 3.4 percentage-point gain in Medicaid, and insignificant change in uninsured among DCP-ineligible patients (aged 26-34 years) over the same period (DID comparing DCP-eligible vs -ineligible patients: uninsured, −5; Medicaid, −3, and private, 10.2; all P < .001). Both groups experienced a similar 1.1 percentage-point gain in rehabilitation (95% CI, 0.6-1.6) (DID P = .91). No changes in mortality, FTR, or total hospital charges were detected.
Implementation of Medicaid expansion and open enrollment was associated with a much larger drop in uninsured and gain in Medicaid coverage among both DCP-eligible and -ineligible patients (eTable 2 in the Supplement). The change was greater among ineligible patients (uninsured: −21; 95% CI, −22.5 to −19.5; Medicaid: 21.4; 95% CI, 21.4 to 24.7) vs eligible patients (uninsured: 95% CI, −15.3; 95% CI, −16.7 to −13.9; Medicaid: 16.9; 95% CI, 15.2 to 18.6) (DID, P < .001). Changes in rehabilitation did not differ significantly between groups (P = .10).
Risk-adjusted interrupted time-series analyses results are depicted graphically in Figure 2 (model results are given in eTable 2 in the Supplement). Following Medicaid expansion/open enrollment, rehabilitation use increased by 2.1 percentage points (95% CI, 0.6-3.5) beyond trends predicted during the post-DCP/preexpansion period. This change was accompanied by an instantaneous 12.8 percentage-point increase in Medicaid (95% CI, 11-14.6) and 14.6 percentage-point drop in uninsured (95% CI, −18.5 to −10.7). Each of the shifts continued to change in the same directions throughout the subsequent postexpansion period, resulting in the larger magnitude detected by the before-and-after models (Tables 1 and 2). No changes from preintervention trends were detected among privately insured patients.
Consideration of the 251 611 young adult trauma patients managed in the ED (eFigure 1A in the Supplement) and 51 830 treated as medical or surgical outpatients (eFigure 1B in the Supplement) demonstrated similar 1:1 changes in uninsured vs Medicaid (constant rates of private coverage), suggesting that the change was not isolated to hospitalized patients. Changes were similar in magnitude within both ambulatory populations: uninsured rates fell by a mean of 7.6 and 5.7 percentage points, respectively; Medicaid increased by 8.8 and 9.7 percentage points.
If insurance disparities existed before Medicaid expansion/open enrollment, such as between male and female (3519 [28.9%] vs 624 [13.1%] uninsured) and non-Hispanic black and non-Hispanic white (1898 [27.2%] vs 1497 [18.7%] uninsured) patients, the discrepancies were largely eliminated (Table 3). The proportion of uninsured dropped among men by a risk-adjusted 21.7 percentage points (−9.3 among women), bringing both groups to respective postimplementation percentages of 8.5% and 4.3% uninsured. Postimplementation uninsured rates among non-Hispanic black and non-Hispanic white patients dropped to an identical value of 5.7% (−21.3 percentage points for non-Hispanic black; −13.3 percentage points for non-Hispanic white). Similar changes were observed among variations in TC level, marital status, residential population, and income. Increases in rehabilitation were comparable among stratified subgroups, most of which started from similar preintervention levels. The notable exception was the 7.9 percentage-point increase in rehabilitation from a baseline of 10.8% among patients managed at a primary adult resource center or level I or II TCs vs the 1.5 percentage-point increase from a baseline of 6% among those managed at level III or nontrauma centers.
In a longitudinal assessment of clinical end points following ACA implementation, the results of this study demonstrated a 74.3% reduction in uninsured, 60% increase in rehabilitation, 25% reduction in mortality, 29.8% reduction in FTR, and $1900 median increase in total hospital charges among young adult patients hospitalized in Maryland for acute traumatic injury. Injured young adults in the United States are among the most likely segments of the population to be uninsured.17,18 Before the earliest ACA-related coverage expansion, 1 in 4 young adults within our study was uninsured at inpatient admission. Three years later, that number has fallen to less than 1 in 10, with similar patterns emerging among injured young adults managed as ambulatory outpatients and in the ED—groups who have historically been invisible from administrative consideration.39 The dramatic reduction in uninsured patients has been met by an equally dramatic rise in the use of Medicaid and no changes in private coverage. Whether a result of increased Medicaid enrollment of uninsured patients within the ED or, as the ubiquitousness of the effect across management settings would seem to suggest, at least a partial result of changes taking place before injury, the 20.1 percentage-point reduction in the proportion of uninsured patients coincided with significant gains in access to rehabilitation among surviving patients—a known indicator of long-term outcomes2- 4,7—and significant reductions in in-hospital mortality and FTR.
Earlier assessment of young adult trauma patients before and after implementation of the DCP revealed that changes in insurance were primarily isolated to the most privileged groups of patients22 and did not affect rehabilitation use21 on a national scale. Stratified changes related to Medicaid expansion and open enrollment, in contrast, point to significant reductions in uninsured and gains in rehabilitation access across the board, indicating that, not only did the later provisions of the ACA have an effect on young adult trauma patients but they also affected the most vulnerable groups, mitigating or even eradicating disparities in certain cases. Insurance changes related to the later part of the ACA were greater among DCP-ineligible vs -eligible patients, potentially pointing to the prior influence of the DCP and significant gains in private coverage among eligible patients demonstrated by this and other studies.21- 26 Outside of Maryland, early work by Sommers et al19,20 is indicative of similar patterns shown in self-reported survey data among general population adults aged 18 to 64 years. The investigators found greater gains in insurance coverage and reported access to care in Medicaid expansion vs nonexpansion states,19 whether using a private or traditional expansion approach,20 as well as indications of countywide reductions in all-cause mortality based on state-initiated Medicaid expansions that occurred in New York, Maine, and Arizona in 2000.40 Gains in insurance coverage and associated changes in rehabilitation use are in keeping with previous insurance-related research, which demonstrated a 9.6 percentage-point increase in rehabilitation beyond expectations among older adult trauma patients on becoming eligible for Medicare.8
The apparent dominance of Medicaid’s involvement in changes in insurance and associated clinical end points among injured young adults leads to important questions about the ACA’s effect on the provision of trauma care. The ACA was expected to “directly [address] the public-health problem of high rates of uninsured young adults,”27(p172) result in a 6.1 percentage-point increase in medically attended injuries among youth aged 26 years or younger,41 and lead to alterations in hospitals’ case mix.28 The trade-off, researchers suggested, would come at the expense of declining revenue for major TCs29 and an uncertain sufficiency in projected funding to support the added patient burden placed on smaller centers.30- 32 Among injured young adults, distributions of demographic, clinical, and community covariates remained largely unchanged. Inpatients were slightly more likely to present with 1 or more Charlson Comorbidity Index–measured comorbidities (11.8% vs 13.4%) and with increased injury severity (ISS 9-15: 18.2% vs 22.7%; ISS >15: 10.2% vs 13%) during the postexpansion period. The percentage of patients managed at a primary adult resource center or level I TC decreased (41.6% vs 34.8%), while the length of inpatient stay increased by a median of 1 day for all major payer groups. These changes corresponded to a risk- and inflation-adjusted median increase in total hospital charges of $1900 per patient and an unadjusted 461 patient median decrease in the number of admitted patients per quarter (95% CI, 317-605 patients).
Much of these later 2 changes can likely be explained by the unique structure and regulation of hospital payment in Maryland. Since 1976 (1977 for Medicaid), the Health Services Cost Review Commission’s voluntary board of commissioners has established an annual charge per admitted inpatient case based on a statewide rate-setting system that all insurers are required to pay.42,43 Rates differ among hospitals depending on their mission (eg, teaching status) and the amount of uncompensated care that they provide.42,43 The program has historically resulted in the elimination of cost-shifting among payers, more equitable distribution of the costs associated with uncompensated care and medical education, and dramatic reductions in per admission inpatient spending in the state42- 44—a subsequently steady trend that can be seen among all 3 payer types prior to the postexpansion period shown in eFigure 2 in the Supplement. The program has, however, also been met by a marked increase in the number of admissions, particularly from 2001 to 2007 when volume adjustments were temporarily removed from the calculation of annual rates.42 Volume controls were reimposed in 2008,42 leading to the gradual reduction in inpatient admissions seen between 2008 and 2013 (eFigure 3 in the Supplement).
On January 10, 2014, the Centers for Medicare & Medicaid Services and the state of Maryland jointly announced a change to Maryland’s all-payer rate-setting system based on resources that became available under the ACA.44- 46 Maryland agreed to permanently shift away from its former Centers for Medicare & Medicaid Services payment waiver, which was based on the cost per inpatient admission, in exchange for a model based on per capita total hospital cost growth. During a 5-year performance period beginning in 2014, the state promised to work toward a global payment model intended to incentivize hospitals to collaborate to prevent unnecessary hospitalizations and reduce readmission risk.45 The early result (combined with other influences of the ACA) has been a spike in hospital charges among injured young adults (eFigure 2 in the Supplement), reduction in the total number of admissions (eFigure 3 in the Supplement), and growth in the per-quarter total hospital charges contributed by all admitted injured young adults (eFigure 4 in the Supplement).
Changes in total hospital charges within the study population appear to be driven by increases in length of stay, the type of hospitals to which patients are admitted (eTable 3 in the Supplement),47 and emerging reductions in admissions for lower-risk adults. Given improvements in outcomes over the same time period, further work is warranted to consider whether and how cost and payments varied on a national scale and the extent to which potentially higher costs are justified by the provision of higher-value care.48 Changes to physicians’ fees—fixed in Maryland as a part of the Maryland Trauma Physician Services Fund49,50—in addition to hospital-specific expenses will need to be monitored and carefully weighed as projected cost-savings and value-related payments are increasingly able to be directly assayed.
The study is not without limitations, several of which come from its reliance on a retrospective administrative database in which limitations of available variables and the potential for absent or miscoded events can be a concern. Use of Maryland Health Services Cost Review Commission data,33 which requires completeness of reporting from all hospitals within the state, helps to address these limitations as does the study’s reliance on inpatient trauma diagnoses, which tend to be robustly reported. Access to these data enabled quarter-by-quarter assessment of changes through September 2015 on a statewide level. Changes in Maryland may not be nationally representative; however, in studying an early-expanding state with a well-developed trauma system,50 racially/ethnically diverse population, and uniquely regulated health care–reporting structure, the study was able to consider observed implications of the ACA on clinical end points and insurance coverage in an unprecedented way. Supplementation of the data with time-varying covariates (eg, community unemployment rates and income levels) and secondary analyses inclusive of temporal trends (eg, interrupted time-series analyses) helped to address the potential influence of changing trends outside of the ACA; nevertheless, the study lacked a temporal control, opening the door for potential confounding associated with unmeasured temporal effects. Robustness checks of measured covariates demonstrated the findings to be structurally sound against both income- and unemployment-specific effects.
In, to our knowledge, one of the first assessments to consider objective measures of the ACA’s effect, the results of this study demonstrated significant reductions in uninsured rates and improvements in clinical end points, including rehabilitation use, mortality, and FTR, among injured young adults. Medicaid expansion and open enrollment–related changes were consistent across demographic and community subgroups, attenuating the presence of preimplementation insurance disparities in many cases. Increasing Medicaid coverage in the face of stagnant proportions of patients who were privately insured primarily drove reductions in the rates of uninsured patients. The trend persisted across management settings, manifesting among inpatient, ED, and ambulatory outpatient trauma populations. For patients who were previously injured, young, and uninsured, Medicaid expansion/open enrollment in Maryland changed insurance coverage and altered patient outcomes in ways that the DCP alone was never intended to do. Medicaid expansion/open enrollment transformed the landscape of trauma coverage, directly affecting the health of one of the country’s most vulnerable at-risk groups.
Corresponding Author: Cheryl K. Zogg, MSPH, MHS, Yale School of Medicine, 367 Cedar St, Room 316 ESH, New Haven, CT 06510 (firstname.lastname@example.org).
Accepted for Publication: June 20, 2016.
Published Online: October 19, 2016. doi:10.1001/jamasurg.2016.3609
Author Contributions: Ms Zogg and Mr Canner had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and Design: Zogg, Scott, Najjar, Olufajo, Haider, Canner.
Acquisition, analysis, or interpretation of data: Zogg, Payró Chew, Scott, Wolf, Tsai, Olufajo, Schneider, Haut, Haider, Canner.
Drafting of the manuscript: Zogg.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Zogg, Payró Chew, Scott, Tsai, Olufajo.
Administrative, technical, or material support: Zogg, Tsai, Haider, Canner.
Study supervision: Zogg, Schneider, Haider.
Conflict of Interest Disclosures: The authors declare that we have no sources of funding or conflicts of interest relevant to the analysis to report. Ms Zogg is supported by National Institutes of Health Medical Scientist Training Program grant T32GM007205. Dr Haut is the primary investigator of grant 1R01HS024547-01 from the Agency for Healthcare Research and Quality and has contract CE-12-11-4489 with the Patient-Centered Outcomes Research Institute (PCORI). Dr Haut is a paid consultant and speaker for the Veterans Health Administration IMPERATIV Advantage Performance Improvement Collaborative and the Illinois Surgical Quality Improvement Collaborative. Dr Haider is the primary investigator of contract AD-1306-03980 with PCORI, Harvard Surgery Affinity Research Collaborative Program grant, and collaborative research grant from the Henry M. Jackson Foundation for the Advancement of Military Medicine. Dr Haider is a cofounder and equity holder in Patient Doctor Technologies Inc, which owns and operates the website http://www.doctella.com. No other disclosures were reported.
Previous Presentation: This work was presented as a poster at the 75th Annual Meeting of the American Association for the Surgery of Trauma and Clinical Congress of Acute Care Surgery; September 14, 2016; Waikoloa, Hawaii.