It’s Not Just the Prices: Time-Driven Activity-Based Costing for Initiation of Veno-Venous Extracorporeal Membrane Oxygenation at Three International Sites—A Case Review : Anesthesia & Analgesia

Secondary Logo

Journal Logo

Featured Articles: Special Article

It’s Not Just the Prices: Time-Driven Activity-Based Costing for Initiation of Veno-Venous Extracorporeal Membrane Oxygenation at Three International Sites—A Case Review

Nurok, Michael MBChB, PhD*; Pellegrino, Vin MD; Pineton de Chambrun, Marc MD, MS; Warsh, Jonathan PhD§; Young, Meredith MPH, BNurs Grad Cert (Intensive Care), RN; Dong, Erik DO; Parrish, Neil MBA; Shehab, Syed MD#; Combes, Alain MD, PhD; Kaplan, Robert S. PhD#

Author Information
Anesthesia & Analgesia 135(4):p 711-718, October 2022. | DOI: 10.1213/ANE.0000000000006074


The United States spends more for intensive care units (ICUs) than do other high-income countries. We used time-driven activity-based costing (TDABC) to analyze ICU costs for initiation of veno-venous extracorporeal membrane oxygenation (VV ECMO) for respiratory failure to estimate how much of the higher ICU costs at 1 US site can be attributed to the higher prices paid to ICU personnel, and how much is caused by the US site’s use of a higher cost staffing model. We accompanied our TDABC approach with narrative review of the ECMO programs, at Cedars-Sinai (Los Angeles), Hôpital Pitié-Salpêtrière (Paris), and The Alfred Hospital (Melbourne) from 2017 to 2019. Our primary outcome was daily ECMO cost, and we hypothesized that cost differences among the hospitals could be explained by the efficiencies and skill mix of involved clinicians and prices paid for personnel, equipment, and consumables. Our results are presented relative to Los Angeles’ total personnel cost per VV ECMO patient day, indexed at 100. Los Angeles’ total indexed daily cost of care was 147 (personnel: 100, durables: 5, and disposables: 42). Paris’ total cost was 39 (26% of Los Angeles) (personnel: 12, durables: 1, and disposables: 26). Melbourne’s total cost was 53 (36% of Los Angeles) (personnel: 32, durables: 2, and disposables: 19) (rounded). The higher personnel prices at Los Angeles explained only 26% of its much higher personnel costs than Paris, and 21% relative to Melbourne. Los Angeles’ higher staffing levels accounted for 49% (36%), and its costlier mix of personnel accounted for 12% (10%) of its higher personnel costs relative to Paris (Melbourne). Unadjusted discharge rates for ECMO patients were 46% in Los Angeles (46%), 56% in Paris, and 52% in Melbourne. We found that personnel salaries explained only 30% of the higher personnel costs at 1 Los Angeles hospital. Most of the cost differential was caused by personnel staffing intensity and mix. This study demonstrates how TDABC may be used in ICU administration to quantify the savings that 1 US hospital could achieve by delivering the same quality of care with fewer and less-costly mix of clinicians compared to a French and Australian site. Narrative reviews contextualized how the care models evolved at each site and helped identify potential barriers to change.

See Article, page 708

The United States spends almost 1% of its gross domestic product on intensive care units (ICUs), which is far more than other countries do. Yet, ICU outcomes in the United States are similar.1 Some scholars have argued that savings in ICUs cannot be realized because of their substantial fixed costs, up to 80% in some models.2,3 Others believe that higher US prices for wages, services, and materials explain its much higher health care costs.4,5 Little empiric evidence supports either explanation for why large cross-national differences in ICU expenditures exist.

To clarify this issue, we examined the use of veno-venous extra corporeal membrane oxygenation (VV ECMO) one of the most complex ICU life-support therapies, both in its technical requirements and in the number of personnel deployed to support a patient. VV ECMO involves cannulating the central venous circulation to reroute venous blood through an external gas exchanger for oxygenation and carbon dioxide removal. While VV ECMO is relatively rarely used, its provision has a number of features that make it a compelling model to analyze drivers of ICU cost. It is used only for patients who are unlikely to survive without additional intervention and, therefore, represents the limit of existing life-saving technology. Average survival rates after VV ECMO are only about 60%, and a successful outcome is difficult to predict on an individual basis.6,7 Clinicians are thus not able to decrease costs just by limiting its use only to those most likely to benefit.8 In addition, the methodology we used to analyze VV ECMO costs may also be used to identify potential process improvements when using other ICU-based therapies.

The study’s goals were to (1) measure and compare the costs and efficiencies of delivering VV ECMO in 3 hospitals, each in a different country; (2) apply time-driven activity-based costing (TDABC) as a tool to measure costs in the ICU; and (3) explore opportunities for cost and process improvement by comparing and contextualizing practices in 3 geographically distant hospitals treating a similar disease entity.


Site Selection

We conducted our comparative costing study of VV ECMO programs at 3 sites: Cedars-Sinai Medical Center in Los Angeles (Los Angeles), Hôpital Pitié-Salpêtrière (Paris), and The Alfred Hospital (Melbourne). These sites perform about 90, 260, and 90 total ECMO (veno-arterial [VA] and VV) treatments, respectively, each year. Hospital- and country-specific data are shown in Supplemental Digital Content, eTable 1, The 3 sites were chosen as convenience samples for their high volumes and reputational expertise in providing high acuity and complex cardiothoracic care. Data were obtained between 2017 and 2019. The Cedars-Sinai Institutional Review Board determined that this work did not constitute human research (archive number Pro00042018) and, therefore, waived the need for approval.


Health care has historically estimated treatment costs by allocating departmental expenses to procedures based on cost-to-charge ratios or relative value units (RVUs). These approaches are flawed since they allocate costs only to reimbursable procedures, and a procedure’s charges or RVUs are, at best, weakly correlated with the intensity and mix of resources actually used to treat patients.9 Cost-to-charge and/or RVU allocation are especially uninformative for multidisciplinary critical care where different departments contribute both resources and personnel to a single patient’s episode of care.

Table 1. - Step-by-Step Time-Driven Activity-Based Costing Analysis
1. Develop process maps with the following principles
 a. Each step reflects an activity in patient care delivery
 b. Identify the resources involved for the patient at each step
 c. Identify any supply (disposable) used for the patient at each step
2. Obtain time estimates for each process step through interviews and observations
3. Calculate the CCR for each resource:
C C R   o f   R e s o u r c e   A = E x p e n s e s   a t t r i b u t a b l e   t o   R e s o u c e   A P r a c t i c a l   c a p a c i t y   o f   R e s o u r c e   A
4. Calculate the total direct costs (personnel, equipment, space, and supplies) of all the resources used over the cycle of care
CCR is how much it costs, per hour or per minute, for a resource to be available for patient-related work.
Abbreviation: CCR, capacity cost rate.

We used TDABC procedures (see Table 1; derived from Kaplan and Porter10) to measure each hospital’s personnel and equipment costs for VV ECMO treatment. We limited our analysis to ICU personnel and equipment costs and did not include fees associated with dedicated procedural space (eg, operating rooms). TDABC estimates the costs of a given clinical service by combining information about the process of patient care delivery (specifically, the time and quantity of labor and nonlabor resources utilized to perform each activity) with the unit cost of each resource used to provide the care.10,11 All costs were measured initially in local currencies and translated to US dollars using the current exchange rate as of March 3, 2020.

Process Maps

A process map (steps 1 and 2; Table 1) is a visual representation of all activities performed during the patient’s care cycle along with the average time, the personnel type, and equipment required to complete each activity. Local staff developed process maps for day 1, when VV ECMO was initiated in the ICU, and for day 3, which we considered representative of each subsequent day. Staff focus groups estimated the personnel quantities, staffing, and process times for a typical VV ECMO patient by observing patients through their care cycle and conducting interviews and surveys with clinical and administrative personnel involved in their care. Prevailing staffing ratios for VV ECMO at each institution were applied. These aggregate site-specific estimates were then applied to patients receiving VV ECMO at each institution. Figure 1 shows the process map for VV ECMO cannulation in the ICU, with the steps outlined in red reflecting day 1, the focus of the present work. This work presents data only for day 1 since staffing costs were approximately equal on days 1 and 3; the remaining cost differences were small, relating primarily to the cost of the disposable ECMO circuit.

Figure 1.:
Process map for VV ECMO cannulation and routine on day 1. ECMO indicates extracorporeal membrane oxygenation; ICU intensive care unit; NP, nurse practitioner; OR, operating room; VV, veno-venous.

The goals of the study were to estimate the most likely cost for an average VV ECMO patient at each institution and to quantify the principal sources of cost differences between them. We could not construct confidence intervals around the average per patient cost since actual resource consumption data for individual patients were not available. As a result, no statistical analysis is offered. Where relevant, this article adheres to the Consolidated Health Economic Evaluation Reporting Standards.

Capacity Cost Rates for Personnel and Equipment

The capacity cost rate (CCR), step 3 of TDABC (Table 1), calculates the cost per minute for all personnel and equipment to be available during the patient’s care cycle. Finance managers at each site supplied salary and fringe benefits data on each personnel type. The practical capacity of each personnel type was calculated as the number of clinical minutes available for professional work per year, excluding time for breaks, meetings, training, vacation, and holidays. The annual cost of equipment was estimated from its purchase cost, estimated useful life, annual cleaning and maintenance costs, and depreciation. This cost was divided by its annual available minutes to calculate each equipment’s CCR.

Durables and Disposable Costs

Durables cost included the use of the ECMO machine, ventilator, and the continuous renal replacement therapy (CRRT) machine (if required for the patient). Disposables costs included the VV ECMO circuit and required cannulae.

Total Cost Estimates

The total direct costs to treat patients are calculated in step 4 (Table 1). The CCR for each resource (personnel and equipment) is multiplied by the average minutes that the resource was used for each activity step. The cost of any durable and disposable is added to the calculated personnel and equipment costs for that step.

To preserve the confidentiality of proprietary cost information, we set Los Angeles’ total personnel cost at 100 and measured all costs at the 3 sites relative to this index. We use variance analysis, a conventional cost accounting technique, to separate the total differential in personnel costs into 3 components: differences in compensation paid to personnel, differences in personnel minutes used per patient, and differences in the skill mix of personnel. The variance formulae are provided in Supplemental Digital Content, eMethods 1,

We did not include the indirect costs of hospital staff departments, such as administration, human resources, information technology, finance, and housekeeping. Each hospital has its own complicated and arbitrary method for allocating such overhead costs. Including them would have introduced noisy and unrelated components into the estimated treatment costs.

As an outcomes indicator, we calculated the percentage of patients who received VV ECMO during the study period who survived to removal (decannulation) and to discharge from each hospital. These data were not adjusted for age, comorbidities, or severity of illness.

Narrative Review

We used narrative review based on responses to a set of standardized questions about the history and the current state of the VV-ECMO care delivery model at the 3 sites to contextualize our findings. The 2 principal investigators on the study, an intensivist-sociologist, and a business school professor conducted the interviews and wrote the narratives.


Figure 2A compares the total daily cost of care per VV ECMO patient between Los Angeles and Paris hospitals. Figure 2B shows the daily VV ECMO cost comparison between Los Angeles and Melbourne hospitals. The Los Angeles hospital’s indexed total cost of care at 147, shown as the first vertical bar in Figure 2A, B, was higher than the Paris hospital cost of 39 (26% of Los Angeles cost) and 53 (36% of Los Angeles cost) at the Melbourne hospital.

Figure 2.:
Difference in cost to treat each VV ECMO patient relative to Los Angeles. A, Los Angeles and Paris difference indexed to the total cost of personnel at Los Angeles. Cost per ECMO patient is 3.7× higher at Los Angeles than at Paris. B, Los Angeles and Melbourne difference indexed to the total cost of personnel at Los Angeles. Cost per ECMO patient is 2.8× higher at Los Angeles than at Melbourne. ICU indicates intensive care unit; VV ECMO, veno-venous extracorporeal membrane oxygenation.

The Los Angeles cost of durables was 5 compared to 1 (Paris) and 2 (Melbourne). The Los Angeles hospital’s disposable costs were 42 compared to 26 and 19 at Paris and Melbourne, respectively.

The Los Angeles hospital’s personnel cost, by construction, equaled 100. Personnel costs at the Paris hospital was 12 and at the Melbourne hospital was 32. Differences in personnel costs between the Los Angeles and the 2 non-US sites were attributed to 3 factors. The Los Angeles hospital paid more for its personnel than the 2 non-US sites, had a higher cost mix of personnel, and also used more personnel minutes per patient day.

The personnel price adjustment is shown in the third vertical bar in Figure 2A, B. If Paris personnel had the same CCRs as Los Angeles, its personnel costs would increase from 12 to 38. Using the same calculation, Melbourne’s personnel costs would increase from 32 to 53. The remaining differential, approximately 70% at both Paris and Melbourne, was caused by differences in staffing levels and the lower cost skill mix (task downshifting) at those sites.

The Los Angeles hospital used 2950 personnel minutes per patient day, while Paris used 1500 minutes, about 50% of Los Angeles; Melbourne used 1874 minutes, 63% of Los Angeles, per patient day. If Paris (Melbourne) used the higher staffing levels—and paid its personnel the same—as the Los Angeles hospital, its costs would increase to 88 (89), as shown by the fourth vertical bar in Figure 2A, B.

Paris and Melbourne hospitals had a third source of cost advantage in addition to paying personnel less and having lower staffing levels. These hospitals had a higher percentage of total care minutes delivered by junior and less-skilled personnel. As shown in Figure 3, 82% (86%) of the Paris (Melbourne) minutes were delivered by the 3 lowest paid personnel tiers—fellows, ancillary workers, and nurses. At Los Angeles hospital, only 58% of total minutes were delivered by these lowest paid personnel types. The fifth vertical bar in Figure 2A, B shows that Paris’ personnel costs would increase by 12 if it used the same skill mix as United States, and Melbourne’s cost would increase by 10 if it used the Los Angeles skill mix.

Figure 3.:
Percentage of total direct care time by personnel group. Personnel are sorted from the highest CCR (intensivist) to the lowest CCR (fellow) calculated for Los Angeles. In both Paris and Melbourne, over 80% of minutes were delivered by the lowest paid personnel tiers. Task downshifting gives Paris and Melbourne a cost advantage over Los Angeles. CCR indicates capacity cost rate; NP, nurse practitioner.
Table 2. - ECMO Data 2017 to 2019
Hospital Total ECMO VV ECMO only (excludes conversion to VA) VV ECMO median age (IQR) VV ECMO female VV ECMO requiring renal replacement therapy VV ECMO survived to decannulation VV ECMO survived to hospital discharge
The Alfred Hospital, Melbourne, Australia 272 73 46 (33–59) 31 (41%) 46 (63%) 45 (62%) 38 (52%)
Cedars-Sinai Medical Center, Los Angeles, California 268 54 56 (42–65) 19 (35%) 9 (17%) 31 (57%) 25 (46%)
Hôpital Pitié-Salpêtrière, Paris, France 784 169 51 (39–61) 64 (38%) 91 (54%) 104 (62%) 95 (56%)
Abbreviations: ECMO, extracorporeal membrane oxygenation; IQR, interquartile range; VA, veno-arterial; VV, veno-venous.

The percentage of patients receiving VV ECMO for respiratory failure who survived to decannulation and discharge, between January 1, 2017 and December 31, 2019, was 56% (52%) at Paris (Melbourne) somewhat higher than the 46% at the Los Angeles hospital (Table 2). While not risk-adjusted, these data suggest that the Los Angeles hospital’s higher cost care model (higher personnel pay, staffing levels, and skill mix) did not result in better outcomes.



Los Angeles hospital’s slightly higher durable cost was primarily for CRRT, which it used for only 17% of its VV ECMO patients during the study period. Paris and Melbourne hospitals used CRRT for 54% and 63% of their patients. Los Angeles hospital used an external vendor, which supplied the machine and operator, paying the vendor a fixed daily charge for each CRRT use. Contracting with the vendor was based on longstanding relationships between the vendor, institution, and private (nonemployed) nephrologists. The institution’s nephrologists, mostly nonhospital employees, were required for CRRT delivery and, therefore, had little motivation to “in-source” the service by purchasing the machines and allowing the ICU staff to operate them. In addition, the private vendor provided a bundled service that included dialysis delivery, a complex and highly regulated service, for non-ICU patients. These factors created a barrier for the hospital to in-source CRRT services. Paris and Melbourne hospitals owned their CRRT machines and used the bedside ICU nurse to operate them, producing a lower cost for each day of CRRT. If the Los Angeles hospital had used CRRT as frequently as Paris and Melbourne hospitals, its durable costs would have been even higher relative to those sites.


Paris and Melbourne hospitals achieved lower disposable costs by standardizing on 1 or 2 VV ECMO devices and leveraging their purchasing power to negotiate favorable terms. The Los Angeles hospital purchased ECMO devices from many different vendors and, therefore, had less ability to negotiate prices. The Los Angeles hospital, sensitive to clinician preferences, had little history in standardizing device usage. After the study, Los Angeles clinicians better appreciated how savings could be realized by standardizing medications and other supplies and purchasing from fewer vendors. Los Angeles clinicians also believed that reducing variety of devices and supplies would decrease their costs to order, stock, and train personnel for each device type and lower the risk of error.


The differential prices paid to personnel explained only 30% of the Los Angeles hospital’s higher personnel costs. The remaining 70% was due to Paris and Melbourne hospitals delivering the same quality of care with fewer and lower cost mix of clinicians.

The Los Angeles hospital uses 1 bedside nurse, working a 12-hour shift, for each VV ECMO patient. The Paris hospital’s 50% fewer personnel minutes largely came from using 1 nurse per 2 patients, half the ratio of The Los Angeles hospital, but still higher than the French mandated ratio of 1:2.5. The Paris Hospital also used nurse assistants, 1 per 4 patients, to supplement the regular nurses. Additional personnel savings were realized by having the attending nurse manage the VV-ECMO machine.

The Melbourne hospital uses the provincially mandated 1:1 nurse staffing level (same as Los Angeles hospital) but, similar to the Paris Hospital, has its nurses operate the VV-ECMO machines. The Melbourne hospital had historically used a select group of physicians to manage ECMO with a perfusionist from the operating theater at the bedside 24/7. Knowledge was not disseminated among the physicians, decision-making was nonuniform, survival rates were low, and complication rates were high. In 2002, the hospital transitioned away from including perfusionists in the care model by training and certifying ICU physicians and nurses to manage the ECMO circuit, and training nurses to deliver all subsequent patient care and supervision. With standardized, evidence-based protocols, a dedicated intensivist, and nurses trained and applying their experience in long-term ECMO care (with telephone access to perfusionists at home if needed), survival rates improved and have been maintained.

The Los Angeles hospital had attempted a shift, like Melbourne, from a perfusionist-centric model to a nurse-centric one. The initiative failed for a number of reasons including concern by some members of the cardiac surgery and ICU teams about not having a perfusionist at the bedside. Notably, the Los Angeles program was under the exclusive leadership of a relatively junior cardiac surgeon who felt that the perfusionists were essential to care. The initiative also occurred during a period that care was being provided by a nonemployed group of intensivists working alongside employed intensivists and interpersonal friction added to tensions in communication with cardiac surgery. Eventually, recommendations from an internal task force created a more collaborative leadership model and ultimately a transition to an employed intensivist model. Nevertheless, the memory of the failed attempt to transition to a nurse run ECMO model served as a barrier to subsequent efforts to try a new model.

Paris and Melbourne hospitals also downshifted many tasks to less skilled personnel to reduce their personnel costs. They assign the bedside nurse to operate the VV-ECMO machine, with access to a perfusionist if needed. Los Angeles hospital continues to use an expensive, skilled perfusionist to manage VV-ECMO machines for every 2 patients although at the time of writing this article, the ratio had been decreased to 1:3.

Paris and Melbourne hospitals used a single attending intensivist to manage almost all the care for ECMO patients, with only occasional assistance from cardiac and pulmonary specialists. Melbourne hospital saved additional time by having a lower ratio of intensivists to ECMO devices overnight.

The Los Angeles hospital, in contrast, used an attending intensivist, cardiac surgery and pulmonary medicine attending physicians, and other consultants to oversee the daily care of VV ECMO patients. Los Angeles’ pluralistic model, with both employed and private practice clinicians involved in decision-making, led to a much higher percentage of care minutes delivered by highly compensated physicians. Many of those minutes were used for negotiations and care resolution among the physicians. While the costs of care for nonemployed physicians participating in this model were not incurred by the hospital, they were passed on to the payor as these physicians billed independently for their time. The core ICU group felt that much of the activity of the nonemployed physicians did not add to value, increased costs, and complicated care.

Patient Selection and Outcomes

None of the hospitals believed that the prevailing payment or regulatory model influenced the decision to start or deny ECMO treatments to patients.

The Los Angeles hospital had older patients (median age 56) compared to Paris (51) and Melbourne (46), partly because both the Paris and Melbourne hospitals used explicit protocols and algorithms to screen patients for access to ECMO treatment, a practice only recently implemented at the Los Angeles hospital. The Paris Hospital used criteria for patient selection for ECMO treatment based on research on the efficacy of ECMO for different types of patients and their stage of illness. The unit of the Paris hospital where ECMO is housed cares for primarily ECMO patients and during times of increased acuity can expand ECMO delivery and displace non-ECMO patients to other ICUs. Based on published research12,13 from one of this article’s authors, Paris hospital was introducing ECMO to patients earlier in the course of their respiratory failure.

The Melbourne hospital maintained its high ECMO survival rate by following internally created guidelines that allowed deviation when appropriate. In 2021, a statewide ECMO service was created with similar processes and criteria adopted for ECMO across Victoria.

In summary, Figure 2A, B suggests that the Los Angeles hospital could save 61% (quantity variance + skill mix variance) of its costs per ECMO patient day by adopting the staffing model and skill mix currently used at the Paris hospital. If the Melbourne hospital model was adopted in Los Angeles, savings would be approximately 46%. While the present work was focused just on VV ECMO, its conclusions are also relevant for veno-arterial ECMO treatment for cardiovascular collapse for which all 3 centers used similar staffing models.

After seeing comparisons to the non-US sites, Los Angeles hospital resolved some of these inefficiencies.8,14 These efforts were redoubled during the COVID-19 pandemic, when Los Angeles hospital’s ICUs moved toward Paris and Melbourne’s lower cost staffing model by decreasing staffing ratios, reducing use of consultants, and shifting routine bedside care to less-skilled personnel.


In this case study of ECMO costs in 3 diverse hospital settings, we found that a TDABC approach to cost analysis of complex therapeutic interventions was able to identify potentially addressable differences in health care delivery models used in different settings. Using narrative review, we also highlight the importance of local history and context in explaining how individual care models evolved. Such narratives suggest why high-cost models have persisted and the potential barriers to change. Our findings raise the possibility that attention to organizational care delivery can reduce the cost of high complexity critical care. We found in our analysis that the conventional wisdom, “it’s the prices stupid,”5 explains only 30% of ICU personnel cost differentials between a US site and those in 2 other high-income countries. If future work confirms that ICU outcomes are comparable across the different ICU staffing models, the US site may begin to address its currently much higher spending for critical care. Our findings also contradict the “fixed-cost” claim that savings cannot be achieved in critical care delivery settings.


Because of differences in training, specialty, and experience, ECMO delivery models naturally vary between countries. The generalizability of this work is, therefore, limited to models similar to those we studied. The TDABC model also did not risk-adjust for severity of illness or comorbidities. We did not have patient-by-patient actual resource consumption data and, therefore, had to rely on gross estimates of personnel quantities and staffing for the average ECMO patient. This limitation prevented the construction of confidence intervals based on individual patient variation. In addition, we did not assess the duration of time individual patients spent on VV ECMO, which may vary between centers and has implications for total program costs and the relative proportion of costs spent on durables and disposables. We did not verify that capacity of personnel time or durables that was not allocated to VV ECMO was allocated elsewhere and simply assumed that this capacity was available. Future studies can address these limitations by tracking actual staffing levels, personnel contact time with patients, duration of time on VV ECMO, and durable and disposable consumption at the individual patient level. Better patient level data would also allow risk-adjustments for patient case severity.

The study only examined costs on the day of initiation of VV ECMO, potentially leading to errors if care intensity varied by length of stay. Our mortality outcomes may not fully describe other relevant clinical, functional, and patient-reported outcomes. As a result, potential benefits from Los Angeles hospital’s more resource-intensive staffing model may have been overlooked, though Los Angeles’ lower usage rate of CRRT and lower percentage of patients discharged alive imply otherwise. Finally, US regulations and medical-legal environment may preclude implementing all the efficiencies in the Paris and Melbourne care models.15


In a study of ECMO care in 3 diverse hospital settings, we found that contrary to conventional wisdom, the higher prices paid to health care personnel in 1 site explained only a small fraction of this differential. We estimate that approximately 70% of the difference in cost could be attributed to the non-US sites using lower staffing levels and less-expensive employees to deliver comparable care. The study quantifies the savings that a US hospital could achieve by delivering ICU care with fewer high-level clinicians and greater use of task downshifting. The study’s narrative review helps explain why the care models we describe developed and the potential barriers to change.


The authors thank Mahek Shah, MD, MBA; Mayumi Kharabi, RN, MSN, CNL; Jeffrey Lopez, MSN, RN, PHN, PCCN; Tao Shen, MBBS; Alice Chan, RN, MSN, CNS, CCRN; Chase Coffey, MD, MS; and Francisco Arabia, MD, MBA for their assistance on this project.


Name: Michael Nurok, MBChB, PhD.

Contribution: This author helped with conception, design, acquisition, analysis, and interpretation of data; and drafting, revising, and approval for publication.

Conflicts of Interest: M. Nurok reported receiving compensation as an adviser to Avant-Garde Health.

Name: Vin Pellegrino, MD.

Contribution: This author helped with conception, design, acquisition, analysis, and interpretation of data; and drafting, revising, and approval for publication.

Conflicts of Interest: None.

Name: Marc Pineton de Chambrun, MD, MS.

Contribution: This author helped with acquisition and interpretation of data, revising, and approval for publication.

Conflicts of Interest: None.

Name: Jonathan Warsh, PhD.

Contribution: This author helped with conception, design, acquisition, analysis, and interpretation of data; and drafting, revising, and approval for publication.

Conflicts of Interest: None.

Name: Meredith Young, MPH, BNurs Grad Cert (Intensive Care), RN.

Contribution: This author helped with conception, design, acquisition, analysis, and interpretation of data; and drafting, revising, and approval for publication.

Conflicts of Interest: None.

Name: Erik Dong, DO.

Contribution: This author helped with acquisition and interpretation of data, revising, and approval for publication.

Conflicts of Interest: None.

Name: Neil Parrish, MBA.

Contribution: This author helped with acquisition and interpretation of data, revising, and approval for publication.

Conflicts of Interest: None.

Name: Syed Shehab, MD.

Contribution: This author helped with acquisition, analysis of data, revising, and approval for publication.

Conflicts of Interest: None.

Name: Alain Combes, MD, PhD.

Contribution: This author helped with conception, design, acquisition, analysis, and interpretation of data; and drafting, revising, and approval for publication.

Conflicts of Interest: A. Combes reported receiving grants and personal fees from Maquet, Xenios, and Baxter and serving as the recent past president of the EuroELSO organization and as a member of the executive and scientific committees for ECMONet.

Name: Robert S. Kaplan, PhD.

Contribution: This author helped with conception, design, acquisition, analysis, and interpretation of data; drafting; revising; and approval for publication.

Conflicts of Interest: R. S. Kaplan reported receiving compensation as an adviser to Avant-Garde Health.

This manuscript was handled by: Avery Tung, MD, FCCM.


capacity cost rate
continuous renal replacement therapy
extracorporeal membrane oxygenation
intensive care unit
relative value unit
time-driven activity-based costing


1. Gooch RA, Kahn JM. ICU bed supply, utilization, and health care spending: an example of demand elasticity. JAMA. 2014;311:567–568.
2. Luce JM, Rubenfeld GD. Can health care costs be reduced by limiting intensive care at the end of life? Am J Respir Crit Care Med. 2002;165:750–754.
3. Kahn JM, Rubenfeld GD, Rohrbach J, Fuchs BD. Cost savings attributable to reductions in intensive care unit length of stay for mechanically ventilated patients. Med Care. 2008;46:1226–1233.
4. Papanicolas I, Woskie L, Jha A. Health care spending in the United States and other high-income countries. JAMA. 2018;319:1024–1039.
5. Anderson GF, Reinhardt UE, Hussey PS, Petrosyan V. It’s the prices, stupid: why the United States is so different from other countries. Health Aff (Millwood). 2003;22:89–105.
6. Goligher E, Tomlinson G, Hajage D, et al. Extracorporeal membrane oxygenation for severe acute respiratory distress syndrome and posterior probability of mortality benefit in a post hoc bayesian analysis of a randomized clinical trial. JAMA. 2018;320:2251–2259.
7. Schmidt M, Bailey M, Sheldrake J, et al. Predicting survival after ECMO for severe acute respiratory failure: the respiratory ECMO survival prediction (RESP)-score. Am J Respir Crit Care Med. 2014;189:1374–1382.
8. Nurok M, Warsh J, Dong E, Lopez J, Kharabi M, Kaplan RS. Achieving value in highly complex acute care: lessons from the delivery of extra corporeal life support. NEJM Catalyst. October 31, 2019. Accessed April 24, 2022.
9. Chapko MK, Liu CF, Perkins M, Li YF, Fortney JC, Maciejewski ML. Equivalence of two healthcare costing methods: bottom-up and top-down. Health Econ. 2009;18:1188–1201.
10. Kaplan RS, Porter ME. How to solve the cost crisis in health care. Harv Bus Rev. 2011;89:46–52.
11. Kaplan RS, Witkowski ML, Hohman JA. Boston Children’s Hospital: Measuring Patient Costs. Harvard Business School Case, 2012:112–186.
12. Combes A, Hajage D, Capellier G, et al.; EOLIA Trial Group, REVA, and ECMONet. Extracorporeal membrane oxygenation for severe acute respiratory distress syndrome. N Engl J Med. 2018;378:1965–1975.
13. Goligher EC, Tomlinson G, Hajage D, et al. Extracorporeal membrane oxygenation for severe acute respiratory distress syndrome and posterior probability of mortality benefit in a post hoc Bayesian analysis of a randomized clinical trial. JAMA. 2018;320:2251–2259. DOI: 10.1001/jama.2018.14276. Erratum in: JAMA. 2019;321:2245. PMID: 30347031.
14. Nurok M, Warsh J, Griner T, et al. Extracorporeal membrane oxygenation appropriateness: an interdisciplinary consensus-based approach. Anesth Analg. 2019;128:e38–e41.
15. Holmgren AJ, Downing NL, Bates DW, et al. Assessment of electronic health record use between US and non-US health systems. JAMA Intern Med. Published online December 14, 2020. doi:10.1001/jamainternmed.2020.7071.

Supplemental Digital Content

Copyright © 2022 International Anesthesia Research Society