Validation of the PIII CLI risk score for the prediction of amputation-free survival in patients undergoing infrainguinal autogenous vein bypass for critical limb ischemia
Article Outline
Objective
The PREVENT III (PIII) critical limb ischemia (CLI) risk score is a simple, published tool derived from the PIII randomized clinical trial that can be used for estimating amputation-free survival (AFS) in CLI patients considered for infrainguinal bypass (IB). The current study sought to validate this risk stratification model using data from the prospectively collected Vascular Study Group of Northern New England (VSGNNE).
Method
We calculated the PIII CLI risk score for 1166 patients undergoing IB with autogenous vein by 59 surgeons at 11 hospitals between January 1, 2003, and December 31, 2007. Points (pts) were assigned to each patient for the presence of dialysis (4 pts), tissue loss (3 pts), age ≥75 (2 pts), and coronary artery disease (CAD) (1 pt). Baseline hematocrit was not included due to a large proportion of missing values. Total scores were used to stratify each patient into low-risk (≤3 pts), med-risk (4-7 pts), and high-risk (≥8 pts) categories. The Kaplan-Meier method was used to calculate AFS for the three risk groups. Log-rank test was used for intergroup comparisons. To assess validation, comparison to the PIII derivation and validation sets was performed.
Result
Stratification of the VSGNNE patients by risk category yielded three significantly different estimates for 1-year AFS (86.4%, 74.0%, and 56.1%, for low-, med-, and high-risk groups). Intergroup comparison demonstrated precise discrimination (P < .0001). For a given risk category (low, med, or high), the 1-year AFS estimates in the VSGNNE dataset were consistent with those observed in the previously published PIII derivation set (85.9%, 73.0%, and 44.6%, respectively), PIII validation set (87.7%, 63.7%, and 45.0%, respectively), and retrospective multicenter validation set (86.3%, 70.1%, and 47.8%, respectively).
Conclusion
The PIII CLI risk score has now been both internally and externally validated by testing it against the outcomes of 3286 CLI patients who underwent autogenous vein bypass at 94 institutions by a diverse array of physicians (three independent cohorts of patients). This tool provides a simple and reliable method to risk stratify CLI patients being considered for IB. At initial consultation, calculation of the PIII CLI risk score can reliably stratify patients according to their risk of death or major amputation at 1 year.
Critical limb ischemia (CLI) is the most advanced form of peripheral arterial disease and it is associated with a high risk of cardiovascular events, including major limb loss, myocardial infarction (MI), stroke, and death.1, 2, 3, 4 The likelihood of death has been reported to be as high as 20% within 6 months of CLI diagnosis and surpasses 50% at 5 years post-diagnosis.5, 6 These high mortality rates exceed those seen in any other pattern of occlusive disease including patients with symptomatic coronary artery disease (CAD)7, 8 and reflect the severe diffuse atherosclerotic burden associated with a diagnosis of CLI.
Traditionally, open surgical bypass was the only effective treatment strategy for limb revascularization in patients with CLI due to infrainguinal arterial occlusive disease. However, over the last decade, the introduction and evolution of endovascular procedures has significantly increased treatment options.5, 9 This change of treatment paradigm has been driven by technological advances, and by the desire of patients and physicians to reduce procedural risk, albeit with potential trade-offs of inferior durability and greater cost.5 In order to improve clinical decision-making, precise risk stratification for patients who present with CLI has therefore become increasingly important.
The PREVENT III (PIII) CLI risk score is an easy-to-use risk stratification model (Fig 1) developed to predict amputation-free survival (AFS) in CLI patients undergoing open infrainguinal surgical bypass.10 This prediction tool was derived from a cohort of patients who underwent autogenous vein bypass for CLI in the context of the PIII randomized trial.11 The PIII CLI risk score was then validated internally using the trial cohort, and externally using a multicenter retrospective cohort of patients. However, these patients were a highly selected population studied in a clinical trial, and may not represent “real world” community and academic vascular surgery practice. Therefore, the objective of the current investigation was to utilize the prospectively collected Vascular Study Group of Northern New England (VSGNNE) database12 to further validate the PIII CLI risk score. This dataset presents a unique opportunity to validate and assess the utility of the PIII CLI risk score in a heterogeneous population of CLI patients selected to undergo surgical revascularization.
Fig 1.
Score card. PIII, PREVENT III; HCT, hematocrit; CAD, coronary artery disease.
Methods
The Vascular Study Group of Northern New England (VSGNNE)
The VSGNNE is a regional cooperative quality improvement initiative that was developed in 2002 to prospectively collect data on outcomes in patients undergoing vascular surgery. Eleven different teaching and non-teaching hospitals with 59 vascular surgeons (academic and private) currently participate in this program by reporting data into the registry. All data are self-reported and sent to a central data repository were they are aggregated and reviewed. Research analysts are blinded to patient, surgeon, and hospital identity. At the time of discharge after the index operation, a perioperative data sheet containing preoperative, intraoperative, and postoperative data is completed and submitted to the VSGNNE. Similarly, at 1-year follow-up (approximate; mean follow-up for the entire cohort was 307 days), an additional data sheet is completed and submitted to the VSGNNE. On this form, data pertaining to ambulation status, symptom status, patency, ankle-brachial index, bypass graft revisions, or amputations are recorded. A current version of the Social Security Death Index and the 1-year follow-up data are used to confirm survival status. Since the inception of the study, a rigorous audit system has been employed which has consistently demonstrated 99% accuracy in capturing the procedures and their associated outcomes.12 Details relating to the VSGNNE study design have been published previously,12 and are available at www.vsgnne.org.
For the purpose of this study, the VSGNNE dataset was limited solely to patients who underwent autogenous vein lower extremity bypass for CLI (defined as gangrene, nonhealing ischemic ulcer, or ischemic rest pain). In an attempt to be broadly inclusive (and to maximize our ability to assess generalizability), as long as a patient had an autogenous vein infrainguinal bypass for the indication of CLI, regardless of bypass configuration, the patient was included for this study. The study cohort consisted of 1078 patients undergoing 1166 infrainguinal bypass procedures between January 1, 2003, and December 31, 2007.
The PREVENT III cohort
Details of the PIII trial design have been described elsewhere,11 but relevant features are briefly reviewed here. PIII was a prospective, randomized, double-blinded, multicenter trial designed to examine the efficacy of a novel pharmacologic agent (edifoligide) in preventing autogenous vein graft failure in 1404 patients who underwent infrainguinal vein bypass at 83 hospitals exclusively for the treatment of CLI.13 This trial incorporated mandated duplex scan surveillance, independent adjudication of endpoints by a clinical events committee, and external contract research organization monitoring all study data per industry standards. The inclusion criteria specified patients at least 18 years of age who underwent infrainguinal bypass (IB) with autogenous vein for CLI.
The PIII CLI risk score
Details of the derivation and validation of this risk stratification model have previously been published.10 In brief, the PIII risk score utilizes five easily obtainable binary variables—dialysis-dependency, presence of tissue loss, age ≥75 years, hematocrit ≤30%, and a history of advanced CAD—to stratify patients with CLI and surgically correctable infrainguinal disease into three distinct categories of expected AFS. An individual patient is given a point value based on each binary variable. The total sum of points is then converted to a score which places the patient in the low- (score ≤3), medium- (score 4-7), or high-risk (score ≥8) category. Based on this category, the model predicts the likelihood of surviving 1 year after surgery without undergoing a major amputation (low-risk = 86%, medium-risk = 73%, high-risk = 45%).
Validation of the PIII CLI risk score using the VSGNNE
The prediction rule for 1-year AFS was applied to the VSGNNE dataset. Each patient in the VSGNNE dataset was given a net score based on the relevant preoperative variables as dictated by the PIII CLI risk score: dialysis-dependent = 4 points, tissue loss (ulceration or gangrene) = 3 points, age ≥75 = 2 points, and a history of advanced CAD (history of MI or unstable angina) = 1 point. Of note, the baseline hematocrit was not included in the scoring because until 2007, this variable was not routinely collected as part of the VSGNNE dataset. Despite this limitation, we intentionally decided to maintain the PIII risk scoring system as originally described10 in order to properly test the risk stratification ability in a “real world” setting. Based on the total sum of points, each patient was assigned to a PIII risk score category: low-risk (score ≤3), medium-risk (score 4-7), and high-risk (score ≥8).
A Kaplan-Meier 1-year AFS rate estimate was calculated for each VSGNNE risk group. The resulting AFS rates were compared with those obtained from the original PIII derivation and validation sets.10
Additional endpoints
The PIII CLI risk score was specifically designed to stratify patients according to the AFS endpoint and, to date, its discriminative ability has only been tested in this context. Using the identical methodology described above for 1-year AFS, the discriminative ability of the PIII risk score on the following additional 30-day outcomes was assessed: (1) major adverse cardiac event (MACE: death or MI [troponin elevation beyond the normal range or ST changes/new Q waves on electrocardiogram]); (2) limb salvage; and (3) death. Similarly, the discriminative ability of the PIII risk score on the following additional 1-year outcomes was assessed: (1) limb salvage; (2) death; (3) primary patency; (4) primary assisted patency; (5) secondary patency; and (6) ability to ambulate (defined as ambulation with or without assistance).
Statistical analysis
Perioperative events were defined as occurring before hospital discharge after the index operation. Baseline characteristics were compared between groups using Pearson χ2 analysis for categorical variables and t test for continuous variables. Time-to-event endpoint analyses at 1 year were performed using the Kaplan-Meier method and intergroup analyses were compared with the log-rank test. All tests were considered statistically significant at an alpha level of 0.05 (P = .05, two-tailed). All analyses were performed using SAS version 9.1 (Cary, NC). Data submission to the VSGNNE registry was independently reviewed and approved by the institutional review boards at all participating institutions.
Results
The VSGNNE cohort included 1166 bypass procedures for CLI accumulated from 11 different practice settings across northern New England. The estimated 1 year AFS in the VSGNNE dataset as a whole was 79.1%. Baseline patient demographics, medication usage, comorbidities, and surgical characteristics were markedly different from the PIII cohort (Table I). Patients in the VSGNNE group were significantly older (age ≥75; 38.0% vs 32.2%; P = .006) and less likely to be African American (0.7% vs 18.2%; P < .0001). Cardioprotective medications, such as statins (66.3% vs 45.2%; P < .0001) and beta-blockers (84.7% vs 58.1%; P < .0001), were more frequently used in the VSGNNE group. A previous ipsilateral bypass was less common in the VSGNNE group (10.6% vs 14.5%; P = .007). Of the comorbidities examined, patients in the VSGNNE group were more likely to have used tobacco (79.4% vs 74.6%; P = .0009), have hypertension (87.7% vs 81.2%; P < .0001), and have high cholesterol (54.3% vs 46.7%; P < .0001). In the VSGNNE group, a smaller proportion of patients underwent bypasses to the tibial or pedal vessels (58.8% vs 65.5%; P < .0001).
Table I. Patient characteristics in the PREVENT III (PIII) derivation set and the Vascular Study Group of Northern New England (VSGNNE) validation set
| Characteristics | PIII derivation setn = 953 (%) | VSGNNE validation setn = 1166 (%) | P value |
|---|---|---|---|
| DEMOGRAPHICS | |||
 Female | 348(36.5) | 385(33.1) | .095 |
| Age ≥75 | 307(32.2) | 443(38.0) | .006 |
| African American | 173(18.2) | 8(0.7) | <.0001 |
| MEDICATIONS | |||
| Statin | 431(45.2) | 386(66.3) | <.0001 |
| Antiplatelet | 759(79.6) | 883(77.3) | .288 |
| Beta-blocker | 554(58.1) | 494(84.7) | <.0001 |
| RISK FACTORS | |||
| Tissue loss (ulcer or gangrene) | 701(73.6) | 830(71.2) | .225 |
| History of advanced CAD | 393(41.2) | 451(38.7) | .231 |
| Previous ipsilateral bypass | 138(14.5) | 124(10.6) | .007 |
| Smoking (ever) | 709(74.6) | 923(79.4) | .009 |
| Diabetes | 610(64.0) | 714(61.2) | .190 |
| Hypertension | 774(81.2) | 1022(87.7) | <.0001 |
| High cholesterol | 445(46.7) | 632(54.3) | .0005 |
| Dialysis-dependent renal failure | 105(11.0) | 115(10.0) | .456 |
| Ankle brachial index <0.3⁎ | 457(73.2) | 182(24.1) | <.0001 |
| Weight >75 kg | 502(53.5) | 440(37.7) | <.0001 |
| SURGICAL CHARACTERISTICS | |||
| Proximal anastomosis site | |||
| CFA | 464(53.8) | 708(64.0) | <.0001 |
| SFA | 234(27.1) | 240(21.7) | .027 |
| Popliteal | 165(19.1) | 158(14.3) | .017 |
| Distal anastomosis site | |||
| Popliteal | 320(34.5) | 483(41.2) | .001 |
| Tibial | 505(54.4) | 470(40.3) | <.0001 |
| Pedal | 103(11.1) | 205(17.6) | <.0001 |
| Single segment GSV conduit | 781(82.0) | 993(85.3) | .042 |
⁎In the VSGNNE dataset, 567 patients were missing a baseline ABI measurement. |
Discrimination into three strata of risk
The integer score assigned to each covariate was used to calculate each individual patient's risk score for 1 year AFS. The scores ranged from 0 to 10 (median 3, interquartile range 2-5). As shown in Table II, the 1-year Kaplan-Meier estimated AFS rates associated with each risk category were significantly different (P < .0001 for each comparison): 1-year AFS 86.4% in the low-risk group (50.7% of cohort), 1-year AFS 74.0% in the medium-risk group (43.1% of cohort), and 1-year AFS 56.1% in the high-risk group (6.2% of cohort) (Fig 2).
Table II. Observed probability of 1-year amputation-free survival and the associated hazard ratios for death or major amputation at 1 year, stratified by the PREVENT III (PIII) critical limb ischemia (CLI) risk score
| Risk categories | Integer score | Amputation-free survival | HR (95% CI) | P value |
|---|---|---|---|---|
| Low | ≤3 | 86.4 | 1.0(ref) | — |
| Medium | 4-7 | 74.1 | 1.87(1.40-2.50) | <.0001 |
| High | ≥8 | 56.1 | 3.48(2.34-5.18) | <.0001 |
Fig 2.
Kaplan-Meier curves demonstrating amputation-free (AF) survival stratified according to each patient's calculated risk score (Vascular Study Group of Northern New England [VSGNNE] cohort).
Validation
The VSGNNE 1-year AFS rates calculated above using the PIII risk score stratification scheme were compared to the previously published rates obtained from the PIII derivation set (n = 953), PIII validation set (n = 451), and the retrospective validation set (n = 716). The 1-year AFS rates for each risk category were similar when compared between each dataset (Fig 3). In the low-risk group (PIII risk score ≤3), the 1-year AFS rate was 86.4% compared to 85.9%, 87.7%, and 86.3% in the PIII derivation set, PIII validation set, and retrospective multicenter validation set, respectively. In the medium-risk group (PIII risk score 4-7), the 1-year AFS rate was 74.0% compared to 73.0%, 63.7%, and 70.1% in the PIII derivation set, PIII validation set, and retrospective validation set, respectively. In the high-risk group (PIII risk score ≥8), the 1-year AFS rate was 56.1% compared to 44.6%, 45.0%, and 47.8% in the PIII derivation set, PIII validation set, and retrospective validation set, respectively.
Fig 3.
Validation of the PREVENT III (PIII) critical limb ischemia (CLI) risk score in 4 different datasets (3286 patients)—PIII derivation set, PIII validation set, retrospective validation set, prospective Vascular Study Group of Northern New England (VSGNNE) validation set. AF, Amputation-free.
Composition of each risk group (Fig 4)
The entire VSGNNE cohort was analyzed to determine the breakdown of patients in each risk group according to each independent predictor. High-risk classification was assigned to 62% of the patients on dialysis, 9% of the patients with tissue loss, 9% of the patients with age ≥75 years, and 13% of the patients with CAD.
Fig 4.
Risk category breakdown by predictor (Vascular Study Group of Northern New England [VSGNNE] cohort). CAD, Coronary artery disease.
Additional endpoints
In the VSGNNE cohort, the PIII CLI risk score was an effective tool for discriminating three significant strata of risk when tested on the following endpoints: perioperative MACE (P = .003), perioperative mortality (P = .001), 1-year mortality (P < .0001), and 1-year ability to ambulate (P < .0001) (Fig 5). On the contrary, the PIII CLI risk score was not an effective tool for discriminating three distinct strata of risk with regards to perioperative limb salvage (P = .14), 1-year limb salvage (P = .06), or any of the evaluated patency outcomes (primary patency, P = .09; primary assisted patency, P = .59; secondary patency, P = .45).
Fig 5.
PREVENT III (PIII) critical limb ischemia (CLI) risk score was an effective tool for discriminating three significant strata of risk when tested on the following endpoints: perioperative major adverse cardiac event (MACE) (a), perioperative and 1-year mortality (b), and 1-year ability to ambulate (c).
Discussion
The PIII CLI risk score is a reliable and simple tool for stratifying CLI patients selected to undergo bypass surgery into low-, medium-, and high-risk categories. At the time of a patient's initial presentation, five easily obtainable binary variables (dialysis-dependency, tissue loss, advanced age, advanced CAD, and low hematocrit) can be used to provide patients and providers with a valid estimate of the likelihood of AFS at 1 year after surgical revascularization. As a result, we believe that the PIII CLI risk score is a useful clinical tool for surgical decision making.
This model's performance was initially derived and validated in a select population of patients who were participants in a randomized trial.10 It was then further validated in a multicenter retrospective cohort assembled from three different hospitals.10 The present validation study builds on these previous reports by extending the overall generalizability of this risk score. The strength of the VSGNNE dataset stems from its enrollment base—more than 50 different surgeons (private and academic) at 11 hospitals (community and university, ranging in size from 25 to nearly 600 beds)—which provides heterogeneity and a depiction of “real world” practice patterns.12, 14 Despite this heterogeneity and a dramatically different patient cohort than the one from which the model was derived (Table I), the 1-year AFS estimates for each risk category are remarkably similar.
The decision to model 1-year AFS as the primary endpoint for the PIII CLI risk score was a deliberate one. When considering a patient with CLI for surgical bypass, we feel that an estimate of the probability that the patient will be alive at 1 year, with an intact index limb, is of paramount importance. Nonetheless, other endpoints clearly play an important role in the determination of the optimal treatment for a patient with CLI. These include, but are not limited to, quality of life and functional outcomes.15, 16, 17 The PIII CLI risk score was not designed to evaluate these endpoints. Ultimately, the PIII CLI risk score provides an additional piece of information that will complement other elements of clinical judgment to inform appropriate treatment choices for individual patients.
Because endpoints other than 1-year AFS are of value, we investigated the ability of the PIII CLI risk score to stratify patients according to additional perioperative and 1-year endpoints (MACE, death, limb salvage, ability to ambulate, and graft patency). Looking at these results in aggregate, it is clear that the PIII CLI risk score is more effective for predicting “systemic” events (MACE, death, and ambulation) rather than graft-related events (limb salvage and patency). This reflects the focus of the original risk score, which was modeled on the endpoint of 1-year AFS, and which, therefore, prioritized traditional systemic predictors (ie, dialysis)18 above traditional graft predictors (ie, vein diameter).19
It should be noted that the baseline hematocrit variable was not included in this validation study due to a large proportion of missing values in the VSGNNE cohort (not routinely collected in the VSGNNE cohort until 2007). In order to preserve (and test) the model as it was originally described in its published form, we did not impute values or adjust for this missing parameter in any way. As a result, it is possible that a number of patients received lower scores than they would have received had they been included in the other three datasets (any patient with a recorded baseline hematocrit ≤30 would have received two extra points). Despite this limitation, the AFS estimates for each risk group remained consistent across each dataset. This raises three possible considerations: (1) the effect of anemia is not as strong of a predictor as the other included variables and it, therefore, may not be an integral component necessary for AFS risk stratification; (2) the distribution of VSGNNE patients with a low baseline hematocrit was similar, by chance, to the distribution of PIII patients with a low baseline hematocrit; or (3) only a minority of patients in the VSGNNE cohort may have had a baseline hematocrit ≤30 and, therefore, the absence of this variable did not have a profound effect. Further experience using the PIII risk index, with increased patient numbers, will help clarify whether baseline hematocrit remains critical to generating a precise risk estimate. This study seems to suggest that this may not be the case.
In addition to playing a prominent role in clinical decision-making by allowing practitioners and patients to better understand the possible risks of open surgical bypass, we believe that the PIII CLI risk score may also be useful for quality improvement initiatives. Now that this model has been extensively validated in three independent patient populations, the expected 1-year AFS rates can be calculated for a given dataset and these estimates can then be used to create specific risk-adjusted benchmarks for individual practitioners or centers. In the same way that that the National Surgery Quality Improvement Program20, 21 has created observed to expected ratios for individual centers, the PIII CLI risk score can be used to generate accurate comparative outcomes for patients undergoing bypass for CLI. With an enhanced awareness of risk-adjusted outcomes at distinct centers, efforts could be directed at identifying individual factors and processes of care that contribute to these improved outcomes. Once identified, these processes could be promoted and implemented elsewhere in order to improve the overall quality of care for the CLI population. Similar efforts have been applied successfully in a variety of other disease states such as cancer,22 MI,23 CAD,24 and peripheral arterial disease.25
Conclusion
The PIII CLI risk score has now been tested against the outcomes of 3286 CLI patients who underwent infrainguinal autogenous vein bypass at 94 institutions by a diverse array of physicians (three independent patient cohorts). This tool provides a simple and reliable method to risk stratify CLI patients being considered for IB. At initial consultation, patients with a 50% chance of death or major amputation at 1 year can be identified. Future investigation is necessary to determine whether this risk prediction tool is also effective for CLI patients undergoing endovascular therapy.
Author contributions
References
- . Mortality over a period of 10 years in patients with peripheral arterial disease. N Engl J Med. 1992;326:381–386
- . The ratio of ankle and arm arterial pressure as an independent predictor of mortality. Atherosclerosis. 1991;87:119–128
- . Prevalence and clinical correlates of peripheral arterial disease in the Framingham Offspring Study. Am Heart J. 2002;143:961–965
- . Relationship of severity of lower limb peripheral vascular disease to mortality and morbidity: a six-year follow-up study. J Vasc Surg. 1989;9:691–696discussion 696-7
- Bypass versus angioplasty in severe ischaemia of the leg (BASIL): multicentre, randomised controlled trial. Lancet. 2005;366:1925–1934
- . Medical treatment of peripheral arterial disease: a comprehensive review. J Vasc Interv Radiol. 2004;15:1197–1207
- One-year cardiovascular event rates in outpatients with atherothrombosis. JAMA. 2007;297:1197–1206
- . The morbidity and mortality following a diagnosis of peripheral arterial disease: long-term follow-up of a large database. BMC Cardiovasc Disord. 2005;5:14
- Shifting paradigms in the treatment of lower extremity vascular disease: a report of 1000 percutaneous interventions. Ann Surg. 2007;246:415–422discussion 422-4
- . Risk stratification in critical limb ischemia: derivation and validation of a model to predict amputation-free survival using multicenter surgical outcomes data. J Vasc Surg. 2008;48:1464–1471
- . Design and rationale of the PREVENT III clinical trial: edifoligide for the prevention of infrainguinal vein graft failure. Vasc Endovascular Surg. 2005;39:15–23
- . A regional registry for quality assurance and improvement: the Vascular Study Group of Northern New England (VSGNNE). J Vasc Surg. 2007;46:1093–1101discussion 1101-2
- Results of PREVENT III: a multicenter, randomized trial of edifoligide for the prevention of vein graft failure in lower extremity bypass surgery. J Vasc Surg. 2006;43:742–751discussion 751
- . Factors associated with stroke or death after carotid endarterectomy in Northern New England. J Vasc Surg. 2008;48:1139–1145
- . Functional outcome after infrainguinal bypass for limb salvage. J Vasc Surg. 1997;25:287–295discussion 295-7
- . Prospective multicenter study of quality of life before and after lower extremity vein bypass in 1404 patients with critical limb ischemia. J Vasc Surg. 2006;44:977–983discussion 983-4
- . Nottingham health profile scores predict the outcome and support aggressive revascularisation for critical ischaemia. Eur J Vasc Endovasc Surg. 2002;23:495–499
- Refinement of survival prediction in patients undergoing lower extremity bypass surgery: stratification by chronic kidney disease classification. J Vasc Surg. 2007;45:944–952
- Technical factors affecting autogenous vein graft failure: observations from a large multicenter trial. J Vasc Surg. 2007;46:1180–1190discussion 1190
- The National Veterans Administration Surgical Risk Study: risk adjustment for the comparative assessment of the quality of surgical care. J Am Coll Surg. 1995;180:519–531
- The Department of Veterans Affairs' NSQIP: the first national, validated, outcome-based, risk-adjusted, and peer-controlled program for the measurement and enhancement of the quality of surgical care (National VA Surgical Quality Improvement Program). Ann Surg. 1998;228:491–507
- Development and implementation of a medical oncology quality improvement tool for a regional community oncology network: the Fox Chase Cancer Center Partners initiative. J Natl Compr Canc Netw. 2007;5:875–882
- . Optimising care of acute myocardial infarction: results of a regional quality improvement project. J Qual Clin Pract. 2000;20:12–19
- . Alabama coronary artery bypass grafting project: results from phase II of a statewide quality improvement initiative. Ann Surg. 2004;239:99–109
- Case-mix adjustment for evaluation of mortality in cystic fibrosis. Pediatr Pulmonol. 2002;33:99–105
Competition of interest: none.
PII: S0741-5214(09)01191-4
doi:10.1016/j.jvs.2009.05.055
© 2009 Society for Vascular Surgery. Published by Elsevier Inc. All rights reserved.
