Advertisement

Value-driven cardiac surgery: Achieving “perfect care” after coronary artery bypass grafting

Open ArchivePublished:June 04, 2018DOI:https://doi.org/10.1016/j.jtcvs.2018.03.177

      Abstract

      Objective

      The objective of this study was to determine if the implementation of a value-driven outcomes tool comprising modifiable quality and utilization metrics lowers cost and improves value of coronary artery bypass grafting (CABG) postoperative care.

      Methods

      Ten metrics were defined for CABG patients in 2 temporally separated phases. Clinical care protocols were designed and implemented to increase compliance with these metrics. Clinical outcomes and cost data were harvested from the electronic medical record using a proprietary value-driven outcomes tool and verified by a data management team. “Perfect care” was defined as achieving all 10 metrics per patient episode.

      Results

      Over a 45-month period, data of 467 consecutive patients who underwent isolated CABG were analyzed. “Perfect care” was successfully achieved in 304 patients (65.1%). There were no observed differences in mortality between patient groups. Linear regression analysis showed a negative correlation between percent compliance with “perfect care” and mean cost. When multivariate analysis was used to adjust for preoperative risk score, mean cost for patients with “perfect care” was 37.0% less than for those without “perfect care.”

      Conclusions

      In the context of focused institution-specific interventions to target quality and utilization metrics for CABG care, clinical care pathways and protocols informed by innovative tools that link automated tracking of these metrics to cost data might simultaneously promote quality and decrease costs, thereby enhancing value. This descriptive study provides preliminary support for a systematic approach to define, measure, and modulate the drivers of value for cardiothoracic surgery patients.

      Key Words

      Abbreviations and Acronyms:

      CABG (coronary artery bypass grafting), CMS (Centers for Medicare and Medicaid Services), ICU (intensive care unit), STS (Society of Thoracic Surgeons), VDO (Value-Driven Outcomes)
      Figure thumbnail fx1
      Multivariate analysis of difference in cost with “perfect care” adjusted for preoperative STS risk score.
      Value in cardiothoracic surgery care can be increased by defining and achieving relevant quality and utilization metrics that help to reduce cost while maintaining excellent clinical outcomes.
      Because of the robust tradition of outcomes reporting and strict adherence to quality standards across the specialty, cardiothoracic surgeons are uniquely poised to deliver value, defined as quality and service divided by cost. It is crucial to deliberately develop systems and processes to measure and promote value in modern cardiothoracic surgery practice.
      See Editorial Commentary page 1449.
      Established in 1989 as a quality improvement initiative among cardiothoracic surgeons, the Society of Thoracic Surgeons (STS) database is used to collect data on compliance with quality and process metrics to track clinical outcomes of common cardiac surgery procedures. The outcomes of coronary artery bypass grafting (CABG) surgery are considered the principal performance metric for surgeons and programs.
      • Jin R.
      • Furnary A.P.
      • Fine S.C.
      • Blackstone E.H.
      • Grunkemeier G.L.
      Using Society of Thoracic Surgeons risk models for risk-adjusting cardiac surgery results.
      Since implementing STS database-facilitated outcomes tracking, the specialty has consistently improved outcomes for CABG to the point that achieving a 30-day mortality rate <1% has been articulated as an achievable benchmark goal.
      • LaPar D.J.
      • Filardo G.
      • Crosby I.K.
      • Speir A.M.
      • Rich J.B.
      • Kron I.L.
      • et al.
      The challenge of achieving 1% operative mortality for coronary artery bypass grafting: a multi-institution Society of Thoracic Surgeons database analysis.
      • Mack M.J.
      If this were my last speech, what would I say?.
      • Kimmaliardjuk D.M.
      • Toeg H.
      • Glineur D.
      • Sohmer B.
      • Ruel M.
      Operative mortality with coronary artery bypass graft: where do we stand in 2015?.
      Although these quality and outcomes improvements are laudable, there has been a simultaneous increase in costs.
      • Osnabrugge R.L.
      • Speir A.M.
      • Head S.J.
      • Jones P.G.
      • Ailawadi G.
      • Fonner C.E.
      • et al.
      Prediction of costs and length of stay in coronary artery bypass grafting.
      There has been little motivation to decrease cost while maintaining high quality and good outcomes because treatments have historically been reimbursed in a fee-for-service model, which does not incentivize or reward cost-effective care.
      • Scudeler T.L.
      • Rezende P.C.
      • Hueb W.
      The cost-effectiveness of strategies in coronary artery disease.
      • Osnabrugge R.L.
      • Speir A.M.
      • Head S.J.
      • Jones P.G.
      • Ailawadi G.
      • Fonner C.E.
      • et al.
      Cost, quality, and value in coronary artery bypass grafting.
      As health care reimbursement moves from fee-for-service to bundled disease-specific payments, minimizing care costs on the individual patient and institutional levels will become increasingly more important.
      • Porter M.E.
      • Teisberg E.O.
      Redefining competition in health care.
      • Hawkins R.B.
      • Mehaffey J.H.
      • Yount K.W.
      • Yarboro L.T.
      • Fonner C.
      • Kron I.L.
      • et al.
      Coronary artery bypass grafting bundled payment proposal will have significant financial impact on hospitals.
      In response to this changing environment, the concept of “value-driven” care has recently gained traction. Value is defined as the quality of patient care and service divided by total cost of care (Figure 1).
      • Porter M.E.
      What is value in health care?.
      • Bradley S.M.
      • Strauss C.E.
      • Ho P.M.
      Value in cardiovascular care.
      • Ken Lee K.H.
      • Matthew Austin J.
      • Pronovost P.J.
      Developing a measure of value in health care.
      In this model, if quality and clinical outcomes remain stable, value increases as costs decrease. Although treatment approaches and outcomes for coronary artery disease have become standardized (the numerator of the value equation), cost of care remains widely variable across different health systems and among different physicians (the denominator of the value equation).
      • Osnabrugge R.L.
      • Speir A.M.
      • Head S.J.
      • Jones P.G.
      • Ailawadi G.
      • Fonner C.E.
      • et al.
      Prediction of costs and length of stay in coronary artery bypass grafting.
      • Osnabrugge R.L.
      • Speir A.M.
      • Head S.J.
      • Jones P.G.
      • Ailawadi G.
      • Fonner C.E.
      • et al.
      Cost, quality, and value in coronary artery bypass grafting.
      • Kilic A.
      • Shah A.S.
      • Conte J.V.
      • Mandal K.
      • Baumgartner W.A.
      • Cameron D.E.
      • et al.
      Understanding variability in hospital-specific costs of coronary artery bypass grafting represents an opportunity for standardizing care and improving resource use.
      It can be quite difficult to accurately measure the costs of health care, because hospital and system charges are significantly distinct from actual cost of care delivery.
      • Mandigo M.
      • O'Neill K.
      • Mistry B.
      • Mundy B.
      • Millien C.
      • Nazaire Y.
      • et al.
      A time-driven activity-based costing model to improve health-care resource use in Mirebalais,.
      To objectively measure value, the full cost of care for treatment of a patient's disease process must be known.
      • Lee V.S.
      • Kawamoto K.
      • Hess R.
      • Park C.
      • Young J.
      • Hunter C.
      • et al.
      Implementation of a value-driven outcomes program to identify high variability in clinical costs and outcomes and association with reduced cost and improved quality.
      The University of Utah Value-Driven Outcomes (VDO) program was created specifically to address these concerns.
      • Lee V.S.
      • Kawamoto K.
      • Hess R.
      • Park C.
      • Young J.
      • Hunter C.
      • et al.
      Implementation of a value-driven outcomes program to identify high variability in clinical costs and outcomes and association with reduced cost and improved quality.
      • Kawamoto K.
      • Martin C.J.
      • Williams K.
      • Tu M.C.
      • Park C.G.
      • Hunter C.
      • et al.
      Value driven outcomes (VDO): a pragmatic, modular, and extensible software framework for understanding and improving health care costs and outcomes.
      Using the VDO framework, the current study was designed to determine the effect on value of achieving “perfect” postoperative care in patients who undergo CABG. In this study, “perfect care” is narrowly defined as achieving the 10 quality and utilization metrics that the authors identified as highly relevant to the cardiac surgery practice at the University of Utah. Clinical protocols were designed to increase compliance with these metrics and test the hypothesis that achieving these metrics would decrease costs and thereby increase value. We report the results of this iterative process to improve the value of CABG care by defining and achieving “perfect care” coupled with the measurement of patient direct care cost.

      Methods

      The study is a retrospective observational analysis of the change in value (clinical quality divided by costs) of care before and after an institutional effort to define and achieve CABG-specific quality and utilization metrics. The development and implementation of protocols to increase compliance with these metrics was divided into 2 phases, each with a unique set of parameters derived in an iterative process driven by clinically relevant outcomes and institution-specific utilization data. In this study we compiled the relative total hospitalization costs associated with all patients during and after protocol implementation, grouped according to whether or not all metrics were achieved. Data were extracted from both phases of protocol implementation. This study protocol was reviewed by the University of Utah institutional review board and was judged not to constitute human subjects research. Informed consent was not required.

       Data Source and Study Population

      Cost data were collected using the previously described University of Utah VDO methodology.
      • Lee V.S.
      • Kawamoto K.
      • Hess R.
      • Park C.
      • Young J.
      • Hunter C.
      • et al.
      Implementation of a value-driven outcomes program to identify high variability in clinical costs and outcomes and association with reduced cost and improved quality.
      • Kawamoto K.
      • Martin C.J.
      • Williams K.
      • Tu M.C.
      • Park C.G.
      • Hunter C.
      • et al.
      Value driven outcomes (VDO): a pragmatic, modular, and extensible software framework for understanding and improving health care costs and outcomes.
      This validated proprietary accounting approach links the commercial electronic health record (Epic, Verona, Wis) with clinical outcome data and direct costs. Previously validated against actual cost or time, the VDO tool is designed to capture 98% of total direct costs, including percent attributable facility, professional, personnel, plus actual usage costs (medications, supplies). All data were collected directly from our institution's electronic medical record using automated systems and by a dedicated data entry team that verified all data entries as accurate. The VDO tool synthesized and analyzed data as previously described, including all patient-specific costs.
      • Lee V.S.
      • Kawamoto K.
      • Hess R.
      • Park C.
      • Young J.
      • Hunter C.
      • et al.
      Implementation of a value-driven outcomes program to identify high variability in clinical costs and outcomes and association with reduced cost and improved quality.
      The express goal of this approach is to tabulate actual cost from the health care system perspective to identify and quantify costs that are directly attributable to direct patient care. To that end, certain large groups of cost such as facility space, labor, equipment, and professional time were calculated on the basis of a patient's estimated use of these resources, whereas supplies, medications, and contracted services are directly tabulated on the basis of the health care system's actual acquisition costs.
      • Lee V.S.
      • Kawamoto K.
      • Hess R.
      • Park C.
      • Young J.
      • Hunter C.
      • et al.
      Implementation of a value-driven outcomes program to identify high variability in clinical costs and outcomes and association with reduced cost and improved quality.
      Professional fees were not included in the analysis to allow a focus on health system direct costs. Because costs are individually negotiated with payers, we are institutionally prohibited from reporting actual cost dollar amounts in the results, but instead report percent change.

       Metric Identification, Protocol Development, and Implementation

       Metric identification

      In phase I of the study (May 2014 to December 2016), 7 quality metrics were identified on the basis of the STS CABG Composite Score and National Quality Forum measures. The metrics included 4 CABG Composite Score components (selected on the basis of the ability to be modified through compliance with care protocols and pathways), and 3 additional related quality metrics, determined by stakeholders. For example, mortality is the major component of the STS CABG score, but cannot be directly modified via clinical care protocols, and so was not chosen. Final metrics included: (1) antibiotics delivered within 1 hour before surgery, (2) antibiotics discontinued between 24 and 48 hours postoperatively, (3) antiplatelet medication prescribed at hospital discharge, (4) antilipid medication prescribed at discharge, (5) use of an internal mammary artery, (6) absence of reintubation, and (7) β-blockers prescribed at hospital discharge.
      • Jin R.
      • Furnary A.P.
      • Fine S.C.
      • Blackstone E.H.
      • Grunkemeier G.L.
      Using Society of Thoracic Surgeons risk models for risk-adjusting cardiac surgery results.
      Discharge medications were started during the acute hospitalization, so they were thought to be relevant to the index hospitalization costs.
      In phase II of the study (January 2017 to September 2017), 3 additional utilization metrics were identified, including: (8) discontinuation of inotrope medication use within 24 hours, (9) discontinuation of mechanical ventilation within 24 hours, and (10) use of ≤500 mL of albumin for postoperative resuscitation. Inotrope medications were defined as epinephrine, dobutamine, milrinone, isoproterenol, and dopamine. The rationale for including these additional utilization metrics was to more directly affect total cost of care for each patient. Because of extremely high compliance with the phase I metrics, the total cost of care was disassociated with achievement of these metrics. Therefore, the study designers identified care parameters with high variability and that also directly influenced cost. Although these 3 utilization metrics are not strictly considered quality metrics, we sought to identify inefficiencies in clinical care that contributed to increased costs. This would further support the hypothesis that by achieving compliance with utilization metrics (ie, decreasing resource utilization), total cost of care would be reduced and therefore value would be increased. Specifically, there are few patient outcome data to support the use of albumin versus crystalloid after cardiac surgery,
      • Verheij J.
      • van Lingen A.
      • Raijmakers P.G.
      • Rijnsburger E.R.
      • Veerman D.P.
      • Wisselink W.
      • et al.
      Effect of fluid loading with saline or colloids on pulmonary permeability, oedema and lung injury score after cardiac and major vascular surgery.
      • Gondos T.
      • Marjanek Z.
      • Ulakcsai Z.
      • Szabó Z.
      • Bogár L.
      • Károlyi M.
      • et al.
      Short-term effectiveness of different volume replacement therapies in postoperative hypovolaemic patients.
      although albumin use remained prevalent in our unit. Duration of mechanical ventilation is known to be shorter with protocoled weaning.
      • Lellouche F.
      • Mancebo J.
      • Jolliet P.
      • Roeseler J.
      • Schortgen F.
      • Dojat M.
      • et al.
      A multicenter randomized trial of computer-driven protocolized weaning from mechanical ventilation.
      Last, similar to mechanical ventilation, we believe that focusing attention on iterative reassessment for readiness to wean from inotropes had the potential to reduce the intensive care unit (ICU) length of stay.

       Protocol development

      Clinical care protocols and pathways designed to achieve the identified metrics were drafted by stakeholders. Stakeholders included cardiothoracic surgeons, anesthesiologists, intensive care physicians, cardiovascular ICU nurses, cardiothoracic surgery advanced practice clinicians, and VDO consultants. The protocols were designed to increase compliance with the identified metrics. Our hypothesis was that increased compliance would directly decrease costs, assessed using the VDO tool.

       Protocol implementation

      A 6-month wash-in period followed metric identification for phase I, during which the protocols were widely circulated to nursing staff, cardiothoracic surgeons, surgical and ICU fellows, intensivists, and advanced practice clinicians, and then brought into clinical practice. The wash-in period for phase II was shortened to 1 month because of the familiarity of the staff with the VDO concept. These wash-in periods included silent observation by VDO project staff to assess implementation rates. VDO project staff also conducted interviews with nurses on perceived barriers to facilitate ongoing implementation of the clinical protocols. Compliance with the clinical protocols was tracked on physician and nursing levels to maximize achievement of each metric.

       Protocol Compliance and the Concept of “Perfect Care”

      Compliance with the 7 STS quality metrics identified in phase I were tracked prospectively throughout the study period. Compliance with the phase II utilization metrics were tracked prospectively from the time of their implementation and retrospectively for phase I of the study using the automatic data collection process inherent to the VDO tool and verified by the study data collection team. Direct patient costs were measured using the VDO tool and were associated with metric compliance. Total costs were calculated for each patient encounter. “Perfect care” was tightly defined in the context of this study as achieving all 10 metrics per patient episode.

       Statistical Analysis

      Demographic characteristics were stratified according to compliance with “perfect care.” Counts and column percentages are reported for binary variables whereas means and SDs are included for numeric variables. We used t tests to compare means for continuous variables, whereas χ2 tests of independence were used for categorical variables. Cost data were not normally distributed. Inspection of the cost data versus the theoretical density of the flexible gamma distribution
      • Lellouche F.
      • Mancebo J.
      • Jolliet P.
      • Roeseler J.
      • Schortgen F.
      • Dojat M.
      • et al.
      A multicenter randomized trial of computer-driven protocolized weaning from mechanical ventilation.
      indicated that the data were approximately gamma distributed, as shown in Figures E1 and E2. Thus, as has been recommended for analysis of cost data,
      • Barber J.
      • Thompson S.
      Multiple regression of cost data: use of generalised linear models.
      • Mani K.
      • Lundkvist J.
      • Holmberg L.
      • Wanhainen A.
      Challenges in analysis and interpretation of cost data in vascular surgery.
      generalized gamma regression
      • Manning W.G.
      • Basu A.
      • Mullahy J.
      Generalized modeling approaches to risk adjustment of skewed outcomes data.
      with a log link was used to compare means for patients who did and did not receive “perfect care” adjusted for preoperative STS risk score. STS score was treated as a categorical variable with 3 groups: low, intermediate, and high with preoperative STS of <1.0%, 1.0%-3.0%, and >3.0%, respectively. Linear regression was used to evaluate the relationship of monthly mean costs and “perfect care” compliance rates. Significance was defined as a P value < .05.

      Results

       Patient Characteristics and Outcomes

      Over a 45-month period, data of 467 consecutive patients who underwent isolated CABG were analyzed. Demographic characteristics and comorbidities for the 304 patients who received “perfect care” and 163 patients who did not are shown in Table 1. There were no significant differences in age, body mass index, marital status, or ethnicity between groups. More patients in the group that received “perfect care” had commercial insurance (47.4% vs 31.9%; P = .012) and were male (85.2% vs 68.7%; P < .001), respectively, versus the group that did not receive “perfect care.”
      Table 1Demographic characteristics and outcomes
      CharacteristicWithout “perfect care”With “perfect care”P value
      n163 (34.9)304 (65.1)
      Age group, y.516
       31 to 402 (1.2)5 (1.6)
       41 to 508 (4.9)15 (4.9)
       51 to 6028 (17.2)69 (22.7)
       61 to 7063 (38.6)125 (41.1)
       71 to 8050 (30.7)71 (23.4)
       ≥8112 (7.4)19 (6.2)
      Male sex112 (68.7)259 (85.2)< .001
      Body mass index31.29 (±16.99)29.77 (±5.93).159
      Marital status.155
       Divorced23 (14.1)25 (8.2)
       Married104 (63.8)211 (69.4)
       Other4 (2.8)16 (5.3)
       Single22 (13.5)33 (10.8)
       Widowed10 (6.1)19 (6.2)
      Ethnicity.197
       Hispanic/Latino10 (6.1)27 (8.9)
       Not Hispanic/Latino145 (89.0)270 (88.8)
       Other/unknown8 (4.9)7 (2.3)
       White/Caucasian138 (84.7)247 (81.2).426
      Payer.012
       Utah Commercial52 (31.9)144 (47.4)
       Utah Medicaid13 (8.0)23 (7.6)
       Utah Medicare83 (50.9)117 (38.5)
       Other15 (9.2)20 (6.6)
      Medical history
       Diabetes131 (80.4)225 (74.0).154
       Hypertension136 (83.4)274 (90.1).050
       Heart failure94 (57.7)79 (26.0)< .001
       Pulmonary hypertension30 (18.4)23 (7.6).001
       Previous CVA28 (17.2)27 (8.9).008
       Previous MI45 (27.6)78 (25.7).730
       COPD41 (25.2)35 (11.5)< .001
      Preoperative STS Score3% (±0.04)1% (±0.01)< .001
      Preoperative LVEF< .001
       <30%11 (±16.2)2 (±2.0)
       31% to 45%13 (±19.1)11 (±10.8)
       >45%44 (±64.7)89 (±87.3)
      Surgical priority.041
       Elective128 (78.5)259 (85.2)
       Urgent31 (19.0)44 (14.5)
       Emergency4 (2.5)1 (0.3)
      Outcomes
       30-Day mortality8 (4.9)5 (1.6).080
       Permanent stroke3 (1.8)3 (1.0).726
       Myocardial infarction10 (6.1)20 (6.6)1
       Atrial fibrillation89 (54.6)85 (28.0)< .001
      Total N = 467. Binary variables are presented as counts and percentages, numeric variables are presented as mean ± SD. CVA, Cerebrovascular accident; MI, myocardial infarction; COPD, chronic obstructive pulmonary disease; STS, Society of Thoracic Surgeons; LVEF, left ventricular ejection fraction.
      Preoperative STS risk of mortality was higher in the patients who did not receive “perfect care” compared with the patients who did receive “perfect care” (3% ± 0.04 vs 1% ± 0.01; P < .001). Preoperative myocardial infarction was diagnosed in 26.3% of all patients and was not significantly different between the 2 groups. History of diabetes was not different between the 2 groups. Hypertension was more common in the group that received “perfect care” (90.1% vs 83.4%; P = .023), whereas previous cerebrovascular accident was more common in the group that did not receive “perfect care” (17.2% vs 8.9%; P = .012), as was the incidence of preoperative heart failure (57.7% vs 26.0%; P < .001), the rate of preoperative pulmonary hypertension (18.4% vs 7.6%; P = .001), a history of chronic obstructive pulmonary disease (25.2% vs 11.5%; P < .001), and the incidence of reduced preoperative left ventricular ejection fraction (35.3% vs 12.8%; P < .001). Elective surgical priority was more common in the group that received “perfect care” (85.2% vs 78.5%; P = .041).
      Differences in unadjusted 30-day mortality did not reach statistical significance between patients who did and did not receive “perfect care” (1.6% vs 4.9%; P = .080). The overall incidence of postoperative permanent stroke was not different between groups (1.8% vs 1.0%; P = .726), nor was the incidence of postoperative myocardial infarction (6.1% vs 6.6%; P = 1.0). The incidence of atrial fibrillation was higher in the patient group that did not receive “perfect care” compared with the group that did receive “perfect care” (54.6% vs 28.0%; P < .001).

       Compliance With “Perfect Care” Metrics

      “Perfect care” was achieved for all metrics in 304 patients (65.1%). Figure 2 shows the compliance rate for each metric of perfect care grouped according to phase I and phase II metrics. Perfect care was successfully achieved for the 7 phase I STS database-derived metrics in 421 patients (90.1%). An additional 44 patients (9.4%) achieved 6 of 7 STS metrics. Perfect care was achieved for all 3 of the phase II metrics in 319 patients (68.3%). An additional 97 patients achieved 2 of 3 metrics (20.8%). Table 2 shows a summary of the cost categories tracked according to the VDO tool. Among the phase I STS database-derived metrics, the most common cause of failure was reintubation (16 patients; 4.0%) followed by lack of antiplatelet medication prescribed at discharge (14 patients; 3.0%). Antibiotic delivery and discontinuation, as well as internal mammary artery use, was achieved in 100% of patients. For the phase II metrics, postoperative albumin use ≤500 mL was achieved in 397 patients (84.5%). Inotropes were successfully weaned off within 24 hours in 393 patients (83.6%). Mechanical ventilation was successfully discontinued within 24 hours in 391 patients (83.2%).
      Figure thumbnail gr2
      Figure 2Frequency of study metrics achieved in (A) phase I and (B) phase II. Presented as number of patient care episodes for which the quality and utilization metrics were achieved.
      Table 2“Perfect care” metric pass rates
      MetricFrequency achieved (%)Frequency of lone failure (%)
      Antibiotics delivered 1 h preoperatively467 (100)0 (0)
      Antibiotics discontinued within 24 to 48 h467 (100)0 (0)
      Antiplatelet medication prescribed at discharge453 (97.0)4 (0.9)
      Antilipid medication prescribed at discharge459 (98.3)10 (2.1)
      Use of internal mammary artery467 (100)0 (0)
      No reintubation451 (96.6)2 (0.4)
      β-Blocker medication prescribed at discharge463 (99.1)2 (0.4)
      Use of <500 mL of albumin397 (85.0)28 (6.0)
      Off ventilator within 24 h391 (83.7)50 (10.7)
      Off inotropes within 24 h393 (84.2)28 (6.0)
      The frequency that each metric was achieved is presented as well as how often that metric was the only metric not achieved for a given patient encounter.

       Cost Comparisons

      A comparison of mean total cost between the group that received “perfect care” (phase I + phase II) and the group that did not is shown in Figure 3, A. There was considerably more variability in mean total cost of care across the group that did not receive “perfect care.” In addition to the increased consistency of cost in the “perfect care” group, overall cost of care for patients achieving “perfect care” across all metrics was 45.0% lower than for patients not achieving “perfect care.” Figure 3, B, elucidates the cost savings observed across multiple categories for patients that received “perfect care” and those that did not. Values are presented as relative percentage differences between “perfect care” and non-“perfect care.” The costliest aspect of care was facility fees, which includes cardiovascular ICU, cardiovascular ward, and operating room fees. Facility fees were 42.6% lower in the patient group that received “perfect care.” Supply costs were the second most costly item in total patient care costs and were 41.1% lower in the “perfect care” group. Ancillary services such as technicians and respiratory therapists, in addition to pharmacy, laboratory, and imaging fees were also less costly in the “perfect care” group (53.7% lower, 51.0% lower, 45.6% lower, and 40.6% lower, respectively).
      Figure thumbnail gr3
      Figure 3Total cost of patient care with and without achievement of “perfect care.” A, Mean total cost per patient care episode comparing “perfect care” (blue) with not “perfect care” (red). B, Cost differences according to category when “perfect care” was achieved, presented as percentage change versus not “perfect care.”
      Figure 4, A, shows the correlation between the rate of “perfect care” delivery and the total cost over the time frame of the study. The rate of “perfect care” immediately after implementation of the phase II protocols (dashed line) increased, along with a decrease in mean total cost. During the final 4 months of the study, rates of “perfect care” began to decline, and mean total cost increased. After efforts to increase compliance with the clinical care protocols, the “perfect care” success rate increased and mean total cost decreased. A linear regression analysis confirmed an inverse relationship between “perfect care” success rate and total patient care cost across the entire study period (slope = 2.3; r2 = 0.29; P = .038; Figure 5).
      Figure thumbnail gr4
      Figure 4Timeline of “perfect care” success rate and total cost. A, Combined success rate for all 10 metrics. Dashed line represents the initial roll-out of the phase II protocols. B, Success rate for the single metric of inotrope weaning by 24 hours. C, Success rate for the single metric of ventilator weaned within 24 hours. D, Success rate for albumin use ≤500 mL. Dashed line represents the start point of the phase II clinical protocols.
      Figure thumbnail gr5
      Figure 5Linear regression trend model for mean total cost given the percent “perfect care” achieved. Shaded area represents 95% confidence interval.
      To asses cost differences adjusted for preoperative comorbidities and risk of surgery, a multivariate analysis was performed. Figure 6 shows the risk-stratified cost differences. Patients who were low risk for CABG (STS score <1.0%) and those who were intermediate risk (STS score 1.0%-3.0%) did not have a difference in mean total costs (P = .1). However, high-risk patients (STS score >3.0%) had an associated 155% increase in mean total cost compared with the low-risk patient group (95% confidence interval, 1.32-1.82; P < .001). As presented in Figure 6, even after adjusting for preoperative risk, mean total cost for patients with “perfect care” was 37% lower than for those without (95% confidence interval, 0.56-0.70; P < .001).
      Figure thumbnail gr6
      Figure 6Multivariate analysis of differences in mean total costs. Patients were stratified according to preoperative Society of Thoracic Surgeons (STS) risk of mortality. Low risk was defined as STS score <1.0%, intermediate risk was defined as an STS score of 1% to 3%, and high risk was defined as STS score >3%. Comparisons in mean total cost were adjusted for preoperative risk score and analyzed between groups who received “perfect care” and those who did not receive “perfect care.” CI, Confidence interval.

      Discussion

      Outcomes for coronary artery disease care have dramatically improved in the past 4 decades, with low mortality and morbidity seen across many different health centers and practice structures.
      • Benjamin E.J.
      • Blaha M.J.
      • Chiuve S.E.
      • Cushman M.
      • Das S.R.
      • Deo R.
      • et al.
      Heart disease and stroke statistics-2017 update: a report from the American Heart Association.
      As a fundamental procedure of modern adult cardiac surgery, CABG has become an agreed-upon benchmark with which to measure quality and outcomes.
      • Epstein A.J.
      • Polsky D.
      • Yang F.
      • Yang L.
      • Groeneveld P.W.
      Coronary revascularization trends in the United States, 2001-2008.
      Recently, the Centers for Medicare and Medicaid Services (CMS) have announced their intention to implement a bundled payment model for coronary artery disease that will emphasize quality of care instead of quantity of care.

      Center for Medicare and Medicaid Services: CMS-5519. Medicare Program; Advancing Care Coordination through Episode Payment Models (EPMs). Washington, DC: Department of Health and Human Services; 2016.

      This is a significant departure from the traditional volume-based reimbursement strategies that have been historically used in the United States health care system.
      • Institute of Medicine (IOM)
      • Smith M.
      • Saunders R.
      • Stuckhardt L.
      • McGinnis J.M.
      Best Care at Lower Cost: The Path to Continuously Learning Health Care in America.
      Value of care, defined as quality or outcomes divided by cost has been increasingly used as a conceptual framework to evaluate the performance of health systems.
      • Lee V.S.
      • Kawamoto K.
      • Hess R.
      • Park C.
      • Young J.
      • Hunter C.
      • et al.
      Implementation of a value-driven outcomes program to identify high variability in clinical costs and outcomes and association with reduced cost and improved quality.
      Because of this impending change in the regulatory environment, cardiothoracic surgeons must develop systems and methods to define achievable clinical metrics that are correlated with costs of care. This will allow surgeons to measure, track, and improve value of care to successfully build and maintain a practice within the new CMS payment structure.
      This study represents an iterative process aimed at increasing value of care within a single institution's CABG practice. Mean total cost was significantly less in patients who achieved the postoperative quality and utilization metrics identified in this study, even after adjusting for preoperative risk. In phase I, protocols were designed and implemented in the operating room, ICU, and wards to encourage compliance with 7 STS-database derived quality and process metrics. Mortality rates were not different between the 2 groups and we observed only a modest decrease in cost despite achieving high rates of “perfect care” across the phase I metrics. This suggests that the STS database metrics, although closely related to improved outcomes over the past decade, might have limited effect on total cost of care in the modern era.
      With the addition of phase II protocols and metrics, the total cost of care showed a more consistent inverse relationship with achievement of “perfect care.” This result suggests that the phase II utilization metrics have more influence on cost of care and therefore represent a target for improving value. This iterative process highlights the importance of determining appropriate institution-specific metrics to define “perfect care” to maximize value. An important point is that achieving “perfect care” does not necessarily have to influence patient outcomes to influence value. By preserving outcomes but reducing cost, value is increased. These data show that these interventions have increased value for CABG patients treated within the University of Utah Health System, but identical interventions might not influence costs across other systems, because the drivers of cost for a given procedure or disease might or might not be generalizable to other patient populations. Rather than focusing on the specific metrics tracked in this study, it is crucial to emphasize the framework for intervention as the take-home point from this study. The iterative process of defining “perfect care” metrics and then measuring the cost implications of achieving those metrics allows for analysis of value. Although the specific metrics might not be transferrable, this concept and iterative process have potential applicability to any health care environment if applied systematically. If achievement of specific “perfect care” metrics do not change the result of the value equation for a given institution or health care system, then different metrics must be sought and implemented until cost (and therefore value) is affected.

       Limitations

      This observational study relied on software algorithms that scour the electronic medical record to collect clinical data. Although the VDO system has been validated and manually confirmed for this study, automated data collection represents a real-world challenge. This study represents a single institution's experience in developing a value-based outcomes metric and applying it to CABG surgery practices. Although the current study shows significant cost savings and value improvements by focusing on these 10 metrics, these results might not be generalizable to other centers. A specific factor in this lack of generalizability is the potential for differences in pricing and direct costs among different institutions and health care systems. In addition, this study focused on total cost of the index hospitalization after CABG surgery, but was not designed to evaluate postdischarge or long-term costs incurred after the acute hospitalization.

      Conclusions

      When “perfect care”—defined by institution- and patient population–specific metrics—is achieved in patients who undergo CABG, significant cost reductions can be obtained. Innovative tools linking automated tracking of quality metrics to costing data provide significant opportunities for focused interventions (eg, ventilator weaning protocols) that increase quality, decrease costs, and thereby enhance the value of CABG for individual patients and the health care system. This study provides evidence for the effectiveness of a systematic approach to define, measure, and modulate the drivers of value for cardiothoracic surgery patients. We present our findings as a model for cardiothoracic surgeons to adapt individual practices and health care systems to flourish in this new environment (Video 1).
      Figure thumbnail fx2
      Video 1Dr Glotzbach discusses the main hypothesis, results, conclusions, and context of the study. Video available at: https://www.jtcvs.org/article/S0022-5223(18)31470-3/fulltext.

       Conflict of Interest Statement

      Authors have nothing to disclose with regard to commercial support.

      Appendix

      Figure thumbnail fx3
      Figure E1Histogram data of distribution of cost data. Mean cost for each treatment group is marked by the solid line (not “perfect care”) and dashed line (“perfect care”). The data fit a gamma distribution.
      Figure thumbnail fx4
      Figure E2Additional evidence for gamma distribution of cost data. Probabilistic modeling of data using multiple bootstrapped iterations to account for the inherent uncertainty in the data. CDFs, Cumulative distribution function.

      Supplementary Data

      References

        • Jin R.
        • Furnary A.P.
        • Fine S.C.
        • Blackstone E.H.
        • Grunkemeier G.L.
        Using Society of Thoracic Surgeons risk models for risk-adjusting cardiac surgery results.
        Ann Thorac Surg. 2010; 89: 677-682
        • LaPar D.J.
        • Filardo G.
        • Crosby I.K.
        • Speir A.M.
        • Rich J.B.
        • Kron I.L.
        • et al.
        The challenge of achieving 1% operative mortality for coronary artery bypass grafting: a multi-institution Society of Thoracic Surgeons database analysis.
        J Thorac Cardiovasc Surg. 2014; 148: 2686-2696
        • Mack M.J.
        If this were my last speech, what would I say?.
        Ann Thorac Surg. 2012; 94: 1044-1052
        • Kimmaliardjuk D.M.
        • Toeg H.
        • Glineur D.
        • Sohmer B.
        • Ruel M.
        Operative mortality with coronary artery bypass graft: where do we stand in 2015?.
        Curr Opin Cardiol. 2015; 30: 611-618
        • Osnabrugge R.L.
        • Speir A.M.
        • Head S.J.
        • Jones P.G.
        • Ailawadi G.
        • Fonner C.E.
        • et al.
        Prediction of costs and length of stay in coronary artery bypass grafting.
        Ann Thorac Surg. 2014; 98: 1286-1293
        • Scudeler T.L.
        • Rezende P.C.
        • Hueb W.
        The cost-effectiveness of strategies in coronary artery disease.
        Expert Rev Pharmacoecon Outcomes Res. 2014; 14: 805-813
        • Osnabrugge R.L.
        • Speir A.M.
        • Head S.J.
        • Jones P.G.
        • Ailawadi G.
        • Fonner C.E.
        • et al.
        Cost, quality, and value in coronary artery bypass grafting.
        J Thorac Cardiovasc Surg. 2014; 148: 2729-2735.e2721
        • Porter M.E.
        • Teisberg E.O.
        Redefining competition in health care.
        Harv Bus Rev. 2004; 82: 64-76.136
        • Hawkins R.B.
        • Mehaffey J.H.
        • Yount K.W.
        • Yarboro L.T.
        • Fonner C.
        • Kron I.L.
        • et al.
        Coronary artery bypass grafting bundled payment proposal will have significant financial impact on hospitals.
        J Thorac Cardiovasc Surg. 2018; 155: 182-188
        • Porter M.E.
        What is value in health care?.
        N Engl J Med. 2010; 363: 2477-2481
        • Bradley S.M.
        • Strauss C.E.
        • Ho P.M.
        Value in cardiovascular care.
        Heart. 2017; 103: 1238-1243
        • Ken Lee K.H.
        • Matthew Austin J.
        • Pronovost P.J.
        Developing a measure of value in health care.
        Value Health. 2016; 19: 323-325
        • Kilic A.
        • Shah A.S.
        • Conte J.V.
        • Mandal K.
        • Baumgartner W.A.
        • Cameron D.E.
        • et al.
        Understanding variability in hospital-specific costs of coronary artery bypass grafting represents an opportunity for standardizing care and improving resource use.
        J Thorac Cardiovasc Surg. 2014; 147: 109-115
        • Mandigo M.
        • O'Neill K.
        • Mistry B.
        • Mundy B.
        • Millien C.
        • Nazaire Y.
        • et al.
        A time-driven activity-based costing model to improve health-care resource use in Mirebalais,.
        Haiti. Lancet. 2015; 385: S22
        • Lee V.S.
        • Kawamoto K.
        • Hess R.
        • Park C.
        • Young J.
        • Hunter C.
        • et al.
        Implementation of a value-driven outcomes program to identify high variability in clinical costs and outcomes and association with reduced cost and improved quality.
        JAMA. 2016; 316: 1061-1072
        • Kawamoto K.
        • Martin C.J.
        • Williams K.
        • Tu M.C.
        • Park C.G.
        • Hunter C.
        • et al.
        Value driven outcomes (VDO): a pragmatic, modular, and extensible software framework for understanding and improving health care costs and outcomes.
        J Am Med Inform Assoc. 2015; 22: 223-235
        • Verheij J.
        • van Lingen A.
        • Raijmakers P.G.
        • Rijnsburger E.R.
        • Veerman D.P.
        • Wisselink W.
        • et al.
        Effect of fluid loading with saline or colloids on pulmonary permeability, oedema and lung injury score after cardiac and major vascular surgery.
        Br J Anaesth. 2006; 96: 21-30
        • Gondos T.
        • Marjanek Z.
        • Ulakcsai Z.
        • Szabó Z.
        • Bogár L.
        • Károlyi M.
        • et al.
        Short-term effectiveness of different volume replacement therapies in postoperative hypovolaemic patients.
        Eur J Anaesthesiol. 2010; 27: 794-800
        • Lellouche F.
        • Mancebo J.
        • Jolliet P.
        • Roeseler J.
        • Schortgen F.
        • Dojat M.
        • et al.
        A multicenter randomized trial of computer-driven protocolized weaning from mechanical ventilation.
        Am J Respir Crit Care Med. 2006; 174: 894-900
        • Barber J.
        • Thompson S.
        Multiple regression of cost data: use of generalised linear models.
        J Health Serv Res Policy. 2004; 9: 197-204
        • Mani K.
        • Lundkvist J.
        • Holmberg L.
        • Wanhainen A.
        Challenges in analysis and interpretation of cost data in vascular surgery.
        J Vasc Surg. 2010; 51: 148-154
        • Manning W.G.
        • Basu A.
        • Mullahy J.
        Generalized modeling approaches to risk adjustment of skewed outcomes data.
        J Health Econ. 2005; 24: 465-488
        • Benjamin E.J.
        • Blaha M.J.
        • Chiuve S.E.
        • Cushman M.
        • Das S.R.
        • Deo R.
        • et al.
        Heart disease and stroke statistics-2017 update: a report from the American Heart Association.
        Circulation. 2017; 135: e146-e603
        • Epstein A.J.
        • Polsky D.
        • Yang F.
        • Yang L.
        • Groeneveld P.W.
        Coronary revascularization trends in the United States, 2001-2008.
        JAMA. 2011; 305: 1769-1776
      1. Center for Medicare and Medicaid Services: CMS-5519. Medicare Program; Advancing Care Coordination through Episode Payment Models (EPMs). Washington, DC: Department of Health and Human Services; 2016.

        • Institute of Medicine (IOM)
        • Smith M.
        • Saunders R.
        • Stuckhardt L.
        • McGinnis J.M.
        Best Care at Lower Cost: The Path to Continuously Learning Health Care in America.
        The National Academies Press, Washington, DC2013: 239-247