Financial Implications of Short Stay Pediatric Hospitalizations

This retrospective cohort study included the 15 hospitals participating in the revenue management program (RMP; Children's Hospital Association, Lenexa, KS) during the calendar year 2017. The RMP is an adjunct of the Pediatric Health Information System (PHIS; Children's Hospital Association, Lenexa, KS), a comprehensive administrative database of clinical and financial details of children’s hospitals’ IP, OBS, emergency, and ambulatory surgery encounters. Hospitals participating in the RMP provide data on actual payments received from payers at the encounter level.

We identified IP and OBS encounters for patients 0 to 18 years old that were classified into a medical All Patient Refined Diagnostic Related Group (APR-DRG, 3M, St Paul, MN). To limit this analysis to pediatric patients with common low-severity medical conditions, we excluded stays >48 hours, those with major or extreme severity of illness based on the APR-DRG, mortalities, transfers, encounters for obstetrics or gynecology (Major Diagnostic Category 14), encounters associated with surgical procedures or intensive care, and normal newborn encounters (APR-DRG 626, 640 with minor or moderate severity of illness). To ensure all health care resource use for each encounter was available in PHIS, we excluded children that were transferred into or out of the included children’s hospital.

The primary outcome measure was the cost coverage ratio (CCR), defined as the payment received by the hospital divided by estimated cost and interpreted as the revenue received from a payer for each $1 of cost. For instance, a CCR = 1 indicates that the payment received by the hospital was equal to hospital cost, a CCR >1 indicates that the hospital recorded a positive margin for that hospitalization, and a CCR <1 indicates a net financial loss for the involved hospital. In the PHIS database, the cost for each encounter is estimated from charges using the hospital and year specific ratio of cost to charges and reflects overhead costs as well as costs related to the provision of services to the patient including laboratory, clinical, and imaging services.

Key variables consisted of both demographic and clinical characteristics, including age, gender, race or ethnicity, the number of complex chronic conditions,¹⁵  payer (public versus nonpublic), length of stay, and hospitalization resources intensity scores for kids (H-RISK). H-RISK was developed by the Children’s Hospital Association as a marker of illness severity and was used as a continuous variable to describe the relative severity of illness associated with each encounter.¹⁶ 

We compared demographic and clinical characteristics of unmatched IP and OBS encounters using χ² tests for categorical variables and Wilcoxon rank sum tests for continuous variables. To illustrate the variation in CCR across payers and hospitals, we stratified encounters based on their payer (public versus nonpublic) and compared CCR between IP and OBS for each hospital. Finally, we estimated each hospital’s additional revenue if OBS encounters were paid at the median inpatient CCR for both public and nonpublic payers [ie, for each encounter, we calculated (Cost × Median Payor Specific Inpatient CCR) – Actual OBS Payment]. All statistical analyses were performed by using SAS v.9.4 (SAS Institute, Cary, NC), and P values <.05 were considered statistically significant. Because we used deidentified data, the Children’s Mercy Kansas City Institutional Review Board deemed the study exempt from review.

Of the 124 789 encounters included from 15 hospitals, 70 981 (56.9%) were OBS status stays (Table 1). The proportion of OBS stays varied from 8% to 86% of encounters across hospitals (Fig 1). Hospital encounters most frequently involved children 1 to 4 years of age (30.9% of the OBS cohort and 29.5% of the IP cohort). The largest proportion of encounters involved children who were Non-Hispanic White (45.9% OBS, 47.1% IP), with smaller proportions of Hispanic and Non-Hispanic Black patients.

OBS stays were more commonly covered by public payer (56.2% of OBS stays vs 52.1% of IP stays). Most encounters involved patients without medical complexity (67.4% of OBS stays and 76.4% of IP stays without complex chronic conditions), and the length of stay was slightly longer for encounters billed under IP status (29.5 hours vs 22.7 hours in OBS). Severity of illness, as evaluated by the H-RISK method, was similar between the 2 groups (1.0 [SD 0.7] in IP, 0.8 [SD 0.7] in OBS). The 10 most common APR-DRGs were similar between IP and OBS status stays except for chemotherapy, diabetes, pulmonary edema, and respiratory failure, which were more commonly observed among the IP stays (Supplemental Table 3).

Nearly half of the included encounters resulted in a net financial loss with a CCR <1 (47.8%, Table 2). Overall, OBS status stays were more likely to result in financial loss than IP status stays (57.0% vs 35.7%). Stays paid by nonpublic payer were more likely to result in a net financial gain with a CCR >1 (71.9%); however, OBS status stays paid by nonpublic payer still resulted in a CCR <1 more frequently than IP status stays paid by nonpublic payer (30.6% vs 25.1%). The majority of stays paid by public payer had a CCR <1 (64.3%), and OBS status stays paid by public payer resulted in net financial loss most frequently (77.6%).

The median CCR was lower for OBS stays compared to IP stays (0.9 [IQR 0.3–1.7] vs 1.4 [IQR 0.7–2.4], Supplemental Table 3). For stays paid by a public payer, the median OBS CCR was 0.6 (IQR 0.2–0.9), and 1.1 (IQR 0.6–2.1) for IP, indicating probable financial loss for OBS hospitalizations under public payers. The median CCR was greater (reflecting a positive margin) for both OBS and IP stays paid by a nonpublic payer (1.6 [IQR 0.8–2.3] and 1.7 [IQR 1.0–2.6], respectively). Substantial intrahospital variation in CCR was present overall, and by both public and nonpublic payers.

Across included hospitals, OBS status stays paid by a public payer resulted in the lowest CCR (Fig 2). Most stays paid by a nonpublic payer resulted in a CCR >1; however, OBS stays still resulted in lower CCR than IP stays in general.

If OBS stays were paid at the median IP CCR for both public and nonpublic payers, the total revenue would increase by $167 million across all included hospitals (Fig 3). The potential additional revenue generated would be nearly 2.5 times greater for stays paid by public payers than for nonpublic payers ($119 million for public payer vs $48 million for nonpublic payer). The impact of applying the median IP CCR rate to all OBS encounters would result in variable financial impacts by hospital; however, all hospitals would experience a net increase in revenue.

We provide new insight into the potential financial ramifications associated with the use of OBS status at children’s hospitals. Paying encounters as IP instead of as OBS would have resulted in an estimated increase in revenue of around $167 million for the 15 hospitals included. Stays compensated by public payers would have resulted in the most substantial change ($119 million) if OBS stays were instead paid as IP stays. Whereas it may be unrealistic for all studied OBS stays to be billed under IP status, if even 50% of these stays were billed as IP, each hospital would average over $5.5 million in additional revenue annually. The differential in revenue and cost coverage between OBS and IP stays should prompt hospital administration, payers, and policy makers to evaluate the use of OBS status closely, because this designation may risk the financial health of children’s hospitals.

Previous work has shown that the costs to care for OBS and IP status stays in children’s hospitals are similar, because OBS patients are usually treated in the same wards and by the same personnel as IP status patients.¹  Additionally, with similar clinical characteristics, the categorization into either OBS or IP status can be vague and is likely because of a combination of factors including local payer contracts, hospital policy, and provider interpretation of third-party guidelines such as Milliman and InterQual.^6,7,13,14  Our study attempted to identify cohorts of OBS and IP status patients who appear clinically similar (demographics, preexisting complex chronic conditions, and LOS) and to evaluate the difference in financial outcomes between the 2 groups. In general, we found that stays designated as OBS resulted in worse financial outcomes (CCR 0.9 [IQR 0.3–1.7]) than those designated as IP (CCR 1.4 [IQR 0.7–2.4]). For all but two hospitals included, the median CCR for IP was higher than for OBS (Supplemental Table 4). The difference in financial outcome without meaningful differences in clinical characteristics highlights an important issue associated with OBS and should cause concern for hospital systems with high use of OBS status.

Our study also highlights the impact of payer on financial outcomes. Reduced payment from public payer sources is well documented in adult literature;^17–21  however, published information on financial outcomes in pediatrics is lacking. This differential in payment may be more troublesome for children’s hospital, because the most common public payer in the pediatric population is Medicaid, which despite being state run, is often tied to the national Medicare payment structure. In this study, we found that, in general, for IP stays paid by public payer, payment received exceeded costs (CCR 1.1 [IQR 0.6–2.1]), whereas OBS stays recovered around 60% of estimated costs (CCR 0.6 [IQR 0.2–0.9]). Hospitals must be cognizant of their payer mix and nonpublic payer contracts if they are to offset losses incurred when OBS stays are paid by public payer.

The financial losses associated with OBS status are a result of decreased payment without a proportional decrease in the costs associated with the hospitalization. Changes in payer structure and state law may ultimately provide a remedy for this issue; however, hospital systems may be able to impact the cost of OBS stays by evaluating internal processes. In particular, the cohorting of OBS patients in OBS units may improve throughput and efficiency, resulting in reduced LOS, and, as a result, reduced hospitalization costs. PHIS does not label OBS stays as occurring in a designated unit or on a ward alongside IP status patients. However, previous work has demonstrated that OBS units are infrequently used at most major children’s hospitals, and when they are present, they are heterogenous in their structure, function, and location.^4,10,22,23  Further work is needed to determine the impact of OBS units on the net financial outcome of OBS status stays and the fiscal health of children’s hospitals.

The original purpose of OBS status was to provide a prolonged period of evaluation for patients in the emergency department when a provider was uncertain of the need for hospitalization.⁸  From its practical origins in emergency medicine, OBS status in pediatrics has become a complicated, erratically applied billing policy which presents substantial challenges that reach beyond hospital financial status. The variable use of OBS status also affects interhospital comparison of patient populations, which can make quality benchmarking and population analyses difficult because many large databases do not include OBS status stays in the data sets.²⁴  Previous work has shown that variable OBS status use affects hospital performance on important benchmarking metrics including LOS and readmission rates.^6,7  With some states imposing penalties for hospitals on the basis of these metrics, OBS status may impact hospital finances in more than 1 way.^25–27  Patient finances may also be affected as OBS status stays are billed as outpatient encounters and, depending on the structure of an insurance plan, these stays may require increased out-of-pocket spending by the patient or their family.²⁸  Consideration should be given to the utility of OBS status moving forward, because it is detrimental to hospitals, challenging for researchers, and it may place increased financial burden on patients.

Our study findings present an opportunity to reexamine the utility of patient status designations within the context of transactional relationships between patients, hospitals, physician providers, and insurance and government payers. OBS status, whereas an apparent financial liability for children’s hospitals, may provide an opportunity for hospitals to move toward value-based care in which patient acuity, cost, health care delivery, and financial sustainability come into further alignment. Whereas OBS status initially sustains a financial loss (CCR <1), there is strong pressure by the Centers for Medicare and Medicaid Services and health insurance companies to continue to reduce health care costs. A robust utilization management program may use OBS status within contracting with health insurance plans to deliver a lower-cost service line when needed.

Our study should be viewed in the context of several limitations. First, the IP and OBS cohorts are not identical, and clinical differences between the two may account for the hospital-applied billing designation at discharge. Additionally, because of the lack of individual patient data and the hospital-specific approach to assigning billing designation, we cannot determine if the billing designations were applied to patients in a comparable or consistent way within or between hospitals. Secondly, the CCR should be viewed as an estimation of the net financial impact of a hospital stay. The CCR relies heavily on the determination of hospital cost by the ratio of cost to charge methodology, which is estimated by using 29 department-specific cost-to-charge ratios in PHIS as required by the Centers for Medicare and Medicaid Services. These ratios are updated and collected from each hospital annually; however they may not reflect the true cost of caring for a child in the hospital. Despite this, cost-to-charge ratios have been previously used^29–32  to estimate hospital costs, and errors in these ratios would likely be present across OBS and IP stays; thus, the proportional difference in CCR would be maintained. Next, our focus on medical diagnoses and exclusion of transferred patients and short-stay surgical patients may underestimate the costs of care incurred by hospitals when OBS status is employed. Finally, billing designations are generally finalized at discharge; however, payers may choose to dispute the designation applied by the hospital billing department. A successful challenge by the payer or hospital may result in a stay originally billed as 1 status but ultimately paid as another (or completely denied altogether). The PHIS database captures the billing designation assigned at discharge and does not account for any changes that occur after payer utilization review.

OBS status stays are more likely to result in a poor financial outcome than similar IP stays. Payers, hospital administration, and policy makers must understand the impact of OBS as a billing designation when evaluating the appropriateness of its use for hospitalized children.

OBS

observation

IP

inpatient

DRG

diagnosis related group

LOS

length of stay

RMP

revenue management program

PHIS

Pediatric Health Information System

APR-DRG

All Patient Refined Diagnostic Related Group

CCR

cost coverage ratio

H-RISK

hospitalization resource intensity score for kids

IQR

interquartile range