Cost-Benefit Analysis of a Medical Emergency Team in a Children’s Hospital
OBJECTIVES: Medical emergency teams (METs) can reduce adverse events in hospitalized children. We aimed to model the financial costs and benefits of operating an MET and determine the annual reduction in critical deterioration (CD) events required to offset MET costs.
METHODS: We performed a single-center cohort study between July 1, 2007 and March 31, 2012 to determine the cost of CD events (unplanned transfers to the ICU with mechanical ventilation or vasopressors in the 12 hours after transfer) as compared with transfers to the ICU without CD. We then performed a cost-benefit analysis evaluating varying MET compositions and staffing models (freestanding or concurrent responsibilities) on the annual reduction in CD events needed to offset MET costs.
RESULTS: Patients who had CD cost $99 773 (95% confidence interval, $69 431 to $130 116; P < .001) more during their post-event hospital stay than transfers to the ICU that did not meet CD criteria. Annual MET operating costs ranged from $287 145 for a nurse and respiratory therapist team with concurrent responsibilities to $2 358 112 for a nurse, respiratory therapist, and ICU attending physician freestanding team. In base-case analysis, a nurse, respiratory therapist, and ICU fellow team with concurrent responsibilities cost $350 698 per year, equivalent to a reduction of 3.5 CD events.
CONCLUSIONS: CD is expensive. The costs of operating a MET can plausibly be recouped with a modest reduction in CD events. Hospitals reimbursed with bundled payments could achieve real financial savings by reducing CD with an MET.
- cost-benefit analysis
- health care financing
- hospital rapid response team
- intensive care units
- Patient Protection and Affordable Care Act
- CD —
- critical deterioration
- CI —
- confidence interval
- LOS —
- length of stay
- MET —
- medical emergency team
- PHIS —
- Pediatric Health Information System
What’s Known on This Subject:
Numerous studies have demonstrated clinical benefits of medical emergency team (MET) implementation, including reductions in mortality, cardiac arrests, and critical deterioration events. No studies have evaluated the financial costs and benefits of METs.
What This Study Adds:
The costs of operating an MET can plausibly be recouped with a modest reduction in critical deterioration events. Hospitals reimbursed with bundled payments could see real financial savings by reducing critical deterioration events with a MET.
Medical emergency teams (METs) have been implemented throughout the world with the aim of quickly assessing, stabilizing, and transferring deteriorating hospital patients to a higher level of care if needed.1 Although numerous studies have examined the clinical impacts of pediatric2–10 and adult11,12 METs, no studies have evaluated their financial costs and benefits.
Most studies have targeted the reduction of rare events including mortality and cardiac arrests. Recently, pre-arrest proximate outcomes have been described as targets for pediatric MET evaluation.10,13,14 One of these measures, the critical deterioration (CD) metric, was developed13 and validated10 by our research team. Included in this measure are rare catastrophic outcomes (arrests and deaths), as well as pre-arrest events that required ICU transfer and life-sustaining interventions. In a previous study, we showed that implementation of a rapid response system that included an MET was associated with a reduction in the rate of CD compared with the pre-intervention trend.10
METs require financial investment for staffing and administration. However, if METs are effective in preventing deterioration, they are likely to result in cost savings, because resuscitating and treating critically ill patients is expensive. In this study we estimated the financial costs and benefits of operating an MET. We first determined the excess hospital costs associated with CD events and then performed a cost-benefit analysis to determine the parameters in which a MET is cost-saving. We used the perspective of a hospital with a bundled payment model providing a fixed reimbursement for a hospitalization, similar to those being evaluated by the US Centers for Medicare and Medicaid Services as part of the Affordable Care Act.15,16
The methods are described in 3 sections. Section 1 describes how excess costs associated with CD were determined. Section 2 outlines how MET staffing and administrative costs were estimated. Section 3 describes how the cost-benefit analysis was done. The Children’s Hospital of Philadelphia Institutional Review Board approved the study.
(1) Cost of Critical Deterioration
Design and Setting
A retrospective cohort study was performed by using data from July 1, 2007 through March 31, 2012 at The Children’s Hospital of Philadelphia, an urban, tertiary care children’s hospital with 535 beds, including a 55-bed pediatric ICU and an 85-bed neonatal ICU.
Unplanned transfers to the pediatric or neonatal ICU during the study period had already been identified for a previous project.10 Transfers that were planned were excluded. “Planned” was defined as: (1) documented as elective or planned in advance, (2) arrived directly from operating room or sleep laboratory, and (3) ICU level of care attributed to need to recover from sedation. For patients who had multiple admissions during the study period, we used only the first admission, and for admissions with multiple unplanned transfers to the ICU, we used only the first transfer.
The time from ICU arrival to any life-sustaining interventions, including initiation of noninvasive ventilation (continuous or bi-level positive airway pressure), invasive mechanical ventilation via endotracheal tube or tracheostomy, and vasopressor infusion, was abstracted from ICU flow sheets. We classified events requiring any of these interventions in the first 12 hours after ICU transfer as CD.10,13
Financial Data Source
The Pediatric Health Information System (PHIS), a national database that contains all billed charges for each hospitalization from freestanding children’s hospitals, was the source of charge data for this study. Children’s Hospital Association maintains the database and performs systematic monitoring on an ongoing basis to ensure data quality. We linked patient identifiers in our clinical data to the PHIS charge data for our hospital using an encryption key.
After identifying the ICU transfers to include in the study, we queried total charges occurring within 2 date-of-service intervals: (1) “post-event ICU stay,” defined as the date of transfer-in to ICU through the date of transfer-out of ICU, and (2) “post-event hospital stay,” defined as the date of transfer-in to ICU through the date of hospital discharge.
Standardizing Hospital Charges Over Time
Hospital accounting systems, including ours, routinely change the way they itemize and combine units of service. For this reason, we had to adjust these units to standardize charges across the time periods included in the cohort. We first addressed the standardizing of charges by replacing all room and board charges before Fiscal Year 2012 (FY12, July 1, 2011 through June 30, 2012) with the corresponding FY12 room and board charge. We then excluded charges for 53 high-volume, low-cost items (examples include suture removal kits, hot packs, and multivitamins) that were rolled into the room and board rate beginning in FY12 and were no longer billed individually.
We standardized the remaining charges by replacing each line item charge in the dataset with the actual charge that same item generated in FY12. For line item charges that did not have a corresponding amount available for FY12, we standardized the charges using the hospital’s routine annual charge increase of 6% (which accounts for inflation as well as price increases).
However, we also identified changes in charges across years that appeared extreme and therefore suspect. To address these, we examined all charges that increased by more than twofold or decreased by half in any 1-year interval. We then reviewed the charge descriptions to differentiate between charges that changed substantially owing to differences in pricing structure versus generic unspecified charge codes such as “clinical service unspecified” that were used to charge variable amounts for a range of different services, drugs, or supplies of varying costs. For these items, we applied the standard annual 6% increase to these items rather than replacing with the 2012 charge.
Estimating Costs From Hospital Charges
The amount that hospitals charge for services provided is consistently an overestimate of the actual cost they incur to provide that service.17 Multiplying charges by an institution-specific cost-to-charge ratio is a method that can be used to better estimate costs.18 We multiplied charges from PHIS by our hospital’s overall 2012 ratio of costs-to-charges of 0.3566.
We addressed the following question: Among unplanned transfers to the ICU, do patients with CD cost more to care for compared with patients who did not have CD (1) while they are in the ICU (the post-event ICU stay), and (2) from the time they are transferred-in to ICU until they are discharged (the post-event hospital stay, including their stay on the wards after transfer-out of ICU, if applicable). Because the cost data were highly right-skewed, we performed the analysis at the patient level using generalized linear models with the log link, γ family, and robust estimator option in Stata 13.1 (Stata Corp, College Station, TX).19,20 The primary outcome in the model was the cost for each patient. The primary exposure was whether the patient met CD criteria. We adjusted for potential confounders including gender, age group, season, transferring ward type (general medical, general surgical, oncology), destination ICU, and the hospital’s monthly case-mix index. We included the transferring ward type as an interaction term in the model.
We performed a secondary analysis excluding patients who died before hospital discharge because we hypothesized that patients who died could impact the results in 2 different ways. CD patients who died soon after transfer to ICU would have fewer costs generated than a patient who was slightly less ill and survived, biasing results toward the null. CD patients who survived their initial critical illness but deteriorated again later in the hospitalization and died would generate more costs, biasing away from the null.
(2) Cost of MET Staffing and Administration
To estimate the cost of MET staffing, we used a 3-member team (the most common team size in the United States21) and modeled different membership consisting of combinations of a nurse, a respiratory therapist, a nurse practitioner, a critical care fellow, and a critical care attending physician. We estimated staffing wages as: $47/hour for the nurse (estimated by using annual salary of $80 390 based on 75th percentile nationally for registered nurses,22 48 weeks of work per year, and 36 hours per week), $38/hour for the respiratory therapist (estimated by using annual salary of $65 080 based on 75th percentile nationally for respiratory therapists,23 48 weeks of work per year, and 36 hours per week), $56/hour for the nurse practitioner (estimated by using annual salary of $97 000 based on 75th percentile of state medians for nurse practitioners,24 48 weeks of work per year, and 36 hours per week), $23/hour for the critical care fellow (estimated using annual salary of $65 607 based on 75th percentile of postgraduate year 6 fellows in 1 of 39 critical care fellowships with published salary data,25 48 weeks of work per year, and 60 hours per week), and $124/hour for the critical care attending physician (estimated by using annual salary of $296 000 based on 75th percentile nationally for Pediatric Critical Care attending physician in a medical school at the Associate Professor level,26 48 weeks of work, and 50 hours per week). We added our hospital’s fringe benefit rate of 25.8% to obtain the total cost per hour to staff the MET. For MET administration, we estimated 4 hours/week for a critical care attending physician team leader and 8 hours/week for a registered nurse team leader for 48 weeks of work per year at the hourly rates provided above (total $29 950 per year, including fringe, for the physician leader, and $22 704 for the nurse leader). These leaders would be responsible for managing the team, tracking call volume, outcomes, and participating in continuous quality improvement. We added those fixed costs to the model.
(3) Cost-Benefit Analysis
Next, we performed a cost-benefit analysis27 to model the balance between the costs of staffing and administering a MET with the potential cost savings achievable if the MET leads to reductions in the incidence of CD. We calculated the reduction in CD events necessary to pay for MET staffing and administration in 2 different scenarios: (1) a freestanding MET with no other clinical responsibilities, and (2) a MET whose members have concurrent clinical responsibilities while responding to 3 events per day and spending 2 hours on each response (allotting time for the initial response and a follow-up visit). For the base-case, we used a team with concurrent clinical responsibilities comprised of a nurse, a respiratory therapist, and a critical care fellow (roles included in at least 70% of pediatric METs in the United States28 and the same model as our hospital’s MET). We also modeled the net cost-benefit in a sensitivity analysis29 evaluating the impacts of altering team composition/hourly staffing cost and staffing model (freestanding or concurrent responsibilities) on the absolute annual reduction in the number of CD events compared with pre-MET implementation needed to offset MET costs.
A total of 1759 unplanned transfers occurred during the study period. After applying the exclusion criteria (Fig 1), we included 1396 patients in the analysis. Overall, 378 of 1396 unplanned transfers (27.1%) met CD criteria. Patient characteristics are presented in Table 1.
Cost of Post-Event ICU Stay
In unadjusted and multivariable models, we found that patients who have CD cost more to care for overall while they are in the ICU. In unadjusted models, patients not meeting CD criteria generated an average of $48 927 in costs, and patients meeting CD criteria generated an average of $138 187 in costs (difference of $89 260 per patient, 95% confidence interval [CI], $60 293 to $118 228; P < .001). After adjustment for the potential confounders (gender, age group, season, transferring ward type, destination ICU, and the hospital’s monthly case-mix index), patients not meeting CD criteria generated an average of $49 593 in costs, and patients meeting CD criteria generated an average of $130 760 in costs (difference of $81 167 per patient, 95% CI, $56 095 to $106 239; P < .001).
Cost of Post-Event Hospital Stay
In unadjusted and multivariable models, we found that patients who have CD cost more to care for overall between the time they are transferred-in to an ICU and the time they are discharged from the hospital. In unadjusted models, patients not meeting CD criteria generated an average of $83 511 in costs, and patients meeting CD criteria generated an average of $197 529 in costs (difference of $114 018 per patient, 95% CI, $79 803 to $148 232; P < .001). After adjustment for the potential confounders, patients not meeting CD criteria generated an average of $85 278 in costs, and patients meeting CD criteria generated an average of $185 051 in costs (difference of $99 773 per patient, 95% CI, $69 431 to $130 116; P < .001). Multivariable model results stratified by transferring ward type are presented in Fig 2, demonstrating that the cost differences for the post-event ICU and hospital stays are significant in patients transferred from medical and oncology, but not surgical wards.
When we excluded patients who died before hospital discharge in a secondary analysis, the differences between CD and non-CD patients were smaller. The difference in the cost of the post-event ICU stay between CD and non-CD patients, adjusted for the potential confounders, was $67 102 (95% CI, $42 178 to $92 027; P < .001). The difference in costs of the post-event hospital stay, adjusted for the potential confounders, was $85 678 (95% CI, $55 216 to $116 141; P < .001).
Hourly MET staffing costs including fringe benefits ranged from $107/hour for a team consisting only of a nurse and respiratory therapist to $263/hour for a team consisting of a nurse, a respiratory therapist, and a critical care attending physician (Table 2). Using the base-case scenario of a team with concurrent clinical responsibilities comprised of a nurse, a respiratory therapist, and a critical care fellow, the annual staffing cost is $298 044 and the administrative cost is $52 654, for an annual total cost of $350 698, equivalent to the cost savings associated with reducing CD events by 3.5 per year. Table 2 lists the “break even” point for several MET scenarios as a sensitivity analysis, and includes a sample calculation.
Depending on the clinical impact of the MET, the cost savings from reducing CD events could plausibly exceed the cost of a MET. For example, in a hospital like ours with approximately 300 unplanned transfers from ward to ICU per year and a 30% CD proportion, reducing that proportion to 25% (an absolute reduction of 15 CD events per year) by implementing a MET comprised of a nurse, respiratory therapist, and critical care fellow with concurrent clinical responsibilities would result in eliminating $1 496 595 in excess costs per year for a net savings of $1 145 897 annually.
The primary findings of our study are (1) unplanned transfers that meet CD criteria have much costlier post-event ICU and hospital stays than unplanned transfers that do not meet CD criteria, and (2) in some situations, the financial benefits of a MET can exceed the costs to operate the MET.
No studies have previously evaluated the excess costs associated with proximate outcomes intended for use in evaluating pediatric METs. Duncan and colleagues performed related work, evaluating the short-term health service costs of in-hospital acute life-threatening events among children in the United Kingdom’s National Health Service.30 Inflated to 2012 pounds31 and converted to US dollars,32 they estimated the mean cost of the post-event length of stay (LOS) to be $51 948 for cardiac arrest, $60 635 for other acute life-threatening events, and $60 181 for urgent ICU admissions. We have added to their work by determining the excess costs associated with CD in a US children’s hospital, revealing differences in costs between medical, oncology, and surgical patient populations, and examining program expenses in a cost-benefit analysis.
Payment models in the US health care system are changing. The Centers for Medicare and Medicaid Services recently launched the Bundled Payments for Care Improvement Initiative, a model that replaces traditional fee-for-service models with bundled payment models that provide a fixed reimbursement for an episode of care (hospitalization only, post-hospital care only, or a combination of hospitalization and post-hospital care).15 As this initiative spreads, hospitals throughout the country will have clear financial incentives to provide high quality, cost-efficient care. In this study, we have demonstrated situations in which a MET could reduce CD events and provide substantial cost savings for hospitals in a bundled payment system.
Our study has several limitations. First, we used the dates of transfer-in and -out of ICU as our analysis intervals rather than times, because the times of service were not in the dataset. Therefore, all charges for the day of transfer would be included, even if the transfer occurred at 11:59 pm. We found significant differences in transfer-in time between CD and non-CD groups, with a higher proportion of non-CD patients transferred later in the day (Table 1). There were not significant differences in transfer-out time (Table 1). As a result, non-CD patients have more non-ICU charges included from the day of transfer-in to ICU. This biases our findings toward the null, because including more non-ICU charges from the day of transfer among non-CD patients will result in an attenuation of the difference between groups. Second, we did not model the costs of team members with concurrent patient care responsibilities in the ICU leaving their primary patients. These costs, which could be financial, opportunity, or safety costs, are important considerations when making the decision to implement a team with concurrent responsibilities. Third, not all CD events are preventable. Hospitals should work to identify and target MET-preventable CD events. Fourth, we evaluated costs by converting charges to costs using a ratio. Better methods to determine actual costs of care, for example using activity-based methods, should be a focus of future work.
CD events not only represent adverse events for patients; they are also very costly. In a bundled payment reimbursement model, the costs of staffing and administering a MET can plausibly be recouped with a modest reduction in CD events. Work to improve patient safety and quality should highlight not only the clinical benefits, but also the financial impacts for hospitals reimbursed with bundled payments.
- Accepted April 23, 2014.
- Address correspondence to Christopher P. Bonafide, MD, MSCE, The Children’s Hospital of Philadelphia, 34th St and Civic Center Blvd, Suite 12NW80, Philadelphia, PA 19104. E-mail:
Dr Bonafide conceptualized and designed the study, analyzed the data and interpreted the results, and drafted the article; Drs Keren and Nadkarni assisted and mentored Dr Bonafide in the study design and interpretation of the results and reviewed and revised the manuscript; Dr Localio assisted and mentored Dr Bonafide in the study design and statistical analysis and reviewed and revised the manuscript; Mr Song acquired and managed the data and reviewed and revised the manuscript; Ms Lutts assisted in study design from a finance perspective and reviewed and revised the manuscript; Ms Roberts, Dr Priestley, Ms Paine, Ms Zander, and Dr Brady participated in the interpretation of the results and reviewed and revised the manuscript; and all authors approved the final manuscript as submitted.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: Health Research Formula Fund Award from the Pennsylvania Department of Health (Dr Keren, Principal Investigator).
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
COMPANION PAPER: A companion to this article can be found on page 375, and online at www.pediatrics.org/cgi/doi/10.1542/peds.2014-1417.
- Brilli RJ,
- Gibson R,
- Luria JW,
- et al
- Hanson CC,
- Randolph GD,
- Erickson JA,
- et al
- Kotsakis A,
- Lobos A-T,
- Parshuram C,
- et al.,
- Ontario Pediatric Critical Care Response Team Collaborative
- ↵Centers for Medicare & Medicaid Services. Bundled Payments for Care Improvement (BPCI) Initiative: General Information | Center for Medicare & Medicaid Innovation. Available at: http://innovation.cms.gov/initiatives/bundled-payments/. Accessed December 19, 2013
- Muennig P
- Glick HA,
- Doshi JA,
- Sonnad SS,
- Polsky D
- Chen JG,
- Kemper AR,
- Odetola F,
- Cheifetz IM,
- Turner DA
- ↵Registered Nurse Job Overview | Best Jobs | US News Careers. Available at: http://money.usnews.com/careers/best-jobs/registered-nurse. Accessed December 9, 2013
- ↵Respiratory Therapist Job Overview | Best Jobs | US News Careers. Available at: http://money.usnews.com/careers/best-jobs/respiratory-therapist. Accessed December 9, 2013
- ↵Nurse Practitioner Salary by State. salarybystate.org. Available at: http://salarybystate.org/healthcare/nurse-practitioner-salary-by-state. Accessed December 30, 2013
- ↵American Medical Association. FREIDA Online | Program Search. Available at: https://freida.ama-assn.org/Freida/user/search/programSearch.do. Accessed December 30, 2013
- Association of American Medical Colleges
- Drummond MF,
- Sculpher MJ,
- Torrance GW,
- O'Brien BJ,
- Stoddart GL
- Drummond MF,
- Sculpher MJ,
- Torrance GW,
- O'Brien BJ,
- Stoddart GL
- ↵UK Office of National Statistics. Dataset Selector. Consum Price Indices - Ser D7FC CPI Index 063 Hosp Serv. 2013. Available at: www.ons.gov.uk/ons/datasets-and-tables/data-selector.html?cdid=D7FC&dataset=mm23&table-id=1.1. Accessed November 15, 2013
- ↵OANDA Corporation. Historical Exchange Rates: GBP to USD for January 1, 2012. OANDA. Available at: www.oanda.com/currency/historical-rates/. Accessed November 15, 2013
- Copyright © 2014 by the American Academy of Pediatrics