A regional registry for quality assurance and improvement: The Vascular Study Group of Northern New England (VSGNNE)

Article Outline

• Abstract
• Methods
  • Data collection
  • Review of outcomes and quality improvement
  • Data validation
  • Statistical analysis
• Results
  • Carotid endarterectomy
  • Lower extremity bypass
  • Abdominal aortic aneurysm repair
  • Preoperative medications
  • One-year follow-up data
  • Validation
• Discussion
• Conclusion
• Author contributions
• Acknowledgment
• Appendix (online only)
• References
• Copyright

Objective

A regional cooperative data registry was organized for carotid endarterectomy (CEA), lower extremity bypass (LEB), and infrarenal abdominal aortic aneurysm (AAA) repair (open and endovascular) procedures in Northern New England to allow benchmarking among centers for quality assurance and improvement activities.

Methods

Since January 2003, 48 vascular surgeons from nine hospitals in Maine, New Hampshire, and Vermont (25 to 615 beds) have prospectively recorded patient, procedure, and in-hospital patient outcome data. Results plus 1-year follow-up data analyzed at a central site are reported anonymously to each center at semiannual meetings where care processes and regional benchmarks are discussed. Mortality and compliance with procedure entry were validated by independent comparison with hospital administrative data. Initial improvement efforts focused on optimizing preoperative medication usage.

Results

A total of 6143 operations were entered into the registry through December 2006. In-hospital stroke or death after CEA was 1.0%, major amputation or death after LEB was 3.8%, and mortality was 2.9% after elective open and 0.4% after endovascular repair. Variation in results between centers and surgeons provides opportunity for further quality improvement. Any postoperative complication increased median length of stay by ≥3 days. Process improvement efforts initiated in 2004 increased preoperative β-blocker administration from 72% to 91%, antiplatelet agents from 73% to 83%, and statins from 54% to 72% (all P < .001). Procedure volume and discharge status validation with administrative data led to 99% of appropriate operations being reported to the registry. Mortality was accurately reported to the data registry for all patients.

Conclusion

This validated regional data registry within a quality improvement initiative has been associated with improved preoperative medication usage. It provides a potential vehicle for future public and pay-for-performance reporting and has the potential to improve patient outcomes. It has been sustained for >4 years and is a model that could be adopted by other regions.

In his 1994 Presidential Address to the Society for Vascular Surgery, Dr Norman Hertzer urged vascular surgeons to track their outcomes, concluding that “results mean everything.”¹ He specifically encouraged a prospective audit of three index procedures: carotid endarterectomy (CEA), lower extremity bypass, and abdominal aortic aneurysm (AAA) repair. Outcomes analysis is important for hospital credentialing and surgeon selection, but it is critical for quality improvement, especially if it allows regional benchmarking and comparison of processes of care that define best practice. Unfortunately, the insular nature of surgical practice makes it difficult to learn from other surgeons.² In this environment, it is difficult to compare patient selection, decision making, and processes of care that could have an important impact on clinical outcomes and represent an opportunity for improvement. Furthermore, the infrequency of major complications, such as stroke after CEA, makes it difficult for even high-volume surgeons to recognize patterns of events that occur only once every several years. How then can individual surgeons improve their results, even if they carefully record them?

Fortunately, successful models for surgical outcome improvement exist. In 1987, cardiac surgeons and researchers from the five hospitals in Maine, New Hampshire, and Vermont that performed coronary artery bypass grafting (CABG) formed the Northern New England Cardiovascular Disease Study Group (NNECDSG).³ This group prospectively collected data on CABG surgery and compared risk-adjusted outcomes between centers. They identified variation in outcome related not to disease severity but rather to different processes of care. By identifying processes of care associated with CABG-related mortality, the NNECDSG reduced regional operative mortality by 24%.⁴

Stimulated by this success, a group of vascular surgeons from Maine, New Hampshire, and Vermont formed the Vascular Study Group of Northern New England (VSGNNE) in 2002 with the following mission statement:

The VSGNNE is a voluntary, cooperative group of clinicians, hospital administrators and research personnel organized to improve the care of patients with vascular disease. By collecting and exchanging information, the group strives to continuously improve the quality, safety, effectiveness, and cost of caring for patients with vascular disease.

The purpose of this report is to describe the results of this activity during the first 4 years of data collection, from 2003 through 2006, in order to present a model for quality assurance and improvement that might be adopted by other regions.

Back to Article Outline

Methods 

In 2001, 14 vascular surgeons from nine hospitals in Maine, New Hampshire, and Vermont held four initial meetings to develop a regional system to improve the outcomes of patients with vascular disease. The group was substantially assisted by Gerald T. O’Connor, principal investigator of the NNECDSG. The above mission statement was adopted, and a prospective registry was designed to focus on CEA, infrainguinal lower extremity bypass (LEB), and open and endovascular repair (EVAR) of infrarenal AAAs. Variables for data collection were chosen that represented key patient demographic variables, surgical procedure details, and in-hospital outcomes, and a one-page data entry form was devised for each operation. The group also decided to collect follow-up data on key outcomes 1 year after surgery.

Detailed definitions were developed for the approximately 100 variables on each form, including complications. Data forms and definitions are available online at the VSGNNE Web site (http://www.vsgnne.org). Additional surgeons have joined the group over time, such that 48 surgeons (Appendix, online only) have contributed patient, procedural, and outcomes data from the nine hospitals shown geographically in Fig 1. These comprise both academic and community hospitals with bed ranges of 25 to 615.

• 
  • View Large Image
  • Download to PowerPoint

• Fig 1. 

  Geographic location of participating centers in The Vascular Study Group of Northern New England.

Data collection 

Variables on the data forms were grouped so that preoperative patient data could be collected by nurses, research personnel, or surgeons; operative details could be completed by surgeons or assistants; and postoperative outcomes could be entered by nurses, research personnel, or chart abstractors. The personnel used for data collection varied by site according to resource availability. Uniformity of data collection was facilitated by discussions at the biannual meetings and newsletters that addressed questions concerning data entry. Individual questions were also addressed by the central data collection site. Completed forms were transmitted electronically or in paper form to the central data collection site at Dartmouth-Hitchcock Medical Center, where the project manager (M. T. R.), consulting epidemiologist (D. S. L.), and principal investigator (J. L. C.) are located.

Rigorous policies were developed to protect patient confidentiality (available at http://www.vsgnne.org) to comply with Health Insurance Portability and Accountability (HIPAA) regulations. These VSGNNE policies included de-identifying individual patients for all analyses and data presentations while still allowing the procedural audit with hospital administration data.

The study was approved by the institutional review board (IRB) of each participating hospital. Initially, informed consent was obtained from each patient, but this requirement was later eliminated by the IRBs of eight of the nine participating centers because the study was designed primarily as a quality improvement and assurance initiative and was deemed to be of no risk.

Pilot data collection began in selected centers in late 2002, but data in this report were collected from 48 participating surgeons at nine participating hospitals from January 2003 through December 2006. During the course of the study, seven other surgeons from four other hospitals began participating but later terminated their participation owing to lack of interest or resources available to collect data. All costs for data collection were borne by each center, whereas central data collection and analysis was funded by an initial grant (Grant No. 18-C-91674) from the Center for Medicare and Medicaid Services (CMS).

Follow-up data are collected about 1 year after surgery. This is facilitated by the central site sending an electronically automated follow-up form for each operation to each surgeon in advance of the expected 1-year office visit. Follow-up data are reported here for operations through February 2006 because later operations had not completed 1-year follow-up at the time this report was prepared. By recording the date of key events such as death, stroke, and amputation during the first year, event rates at key times such as 30 days can be calculated as desired.

Review of outcomes and quality improvement 

Biannual meetings of the VSGNNE have been held each year, with attendance by surgeons, data collection personnel, researchers, and hospital administrators. Because not all participants can attend each meeting, minutes and slide presentations are prepared after each meeting for debriefing conferences held at each site. At each VSGNNE meeting, data are presented for each surgeon and hospital, but with the names blinded, so that each surgeon can compare his or her data and the data of the hospital with the regional average and also understand the regional variation.

During late 2003, VSGNNE members received instruction in continuous quality improvement techniques and applied these principles to preoperative β-blocker usage in 2004. A protocol was developed that recommended starting β-blocker therapy at least 2 weeks preoperatively and continuing for at least 2 weeks postoperatively, with a target usage of 90% to account for drug intolerance, contraindications, and that very low-risk patients likely do not require β-blockade. Data for each surgeon and each center were displayed anonymously to demonstrate the variation of usage. Each center shared its techniques for improving β-blocker usage, and each site was encouraged to locally audit of 10 consecutive patients to assess β-blocker usage and make changes necessary to improve results. Data on preoperative medication usage have since this time been reported back to the participants at each VSGNNE meeting. Preoperative medications were defined as those being taken ≤36 hours before surgery.

Data validation 

Submitted results from each center were audited by comparison with hospital claims data to be certain that all patients undergoing CEA, LEB, or AAA repair had been entered by each participating surgeon. This was done by extracting patients using International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) procedure codes who had been operated on by a participating surgeon during the interval January 2003 though December 2005 (2006 data have not yet been audited). Primary ICD-9 codes used for each procedure were CEA, 38.12; LEB, 39.29; open AAA, 38.44, and EVAR AAA, 39.71. As the result of an initial pilot study, the following alternative procedure codes were added to ensure complete capture of all patients: LEB, 38.48 (leg aneurysm repair); open AAA, 39.25 (aortofemoral bypass), and 38.46 (AAA resection).

Patients in the VSGNNE registry were then matched to the administrative data using unique identifiers such as Social Security Number. The inclusive nature of this search led to more patients being identified in the administrative data than should have been included in the VSGNNE registry (eg, axillofemoral bypass under ICD-9 39.29), which then required a record review at each site. The audit also allowed validation of hospital deaths and discharge status (ie, discharged home or elsewhere).

Statistical analysis 

Data were entered into a Structured Query Language relational database and analyzed using Stata 9.0 statistical software (StataCorp, College Station, Tex). Data are reported for the overall regional results, and the range of data among the nine centers is also reported for selected variables. For centers who entered data for <10 operations, their contribution to the range reported in tables was censored owing to low accuracy of the estimate. Categoric variables were compared with χ² analysis. Life-table methods were used to calculate time-dependent results during follow-up.

Back to Article Outline

Results 

From January 2003 through December 2006, 6143 operations were entered into the registry: CEA, 3097 (primary, 2984; redo, 62; combined CABG, 51); LEB, 1630 (occlusive disease, 1521; aneurysm disease, 109); open AAA repair, 883 (elective, 667; symptomatic, 63; ruptured, 153); and EVAR, 533 (elective, 495; symptomatic, 22; ruptured, 16). Patient preoperative demographic data are summarized in Table I for the five major procedure subsets of CEA, LEB occlusive disease, open elective, EVAR elective, and ruptured open AAA repair further described in this report.

Table I. Patient demographic information

	CEA (primary)	LEB (occlusive)	Open AAA (elective)	EVAR (elective)	Rupture AAA (open)
No. of procedures	2984	1521	667	495	153
Mean age, y (range)	70(38-97)	69(26-96)	71(43-90)	74(48-97)	73(33-91)
Female sex, %	40	33	26	19	14
White race, %	99	99	99	100	100
Smoking, %
Current	32	39	43	31	45
Ever	79	82	92	87	90
Comorbidities, %
Hypertension	86	86	81	80	82
Diabetes	30	55	15	19	17
Coronary artery disease	35	41	35	41	38
Previous CABG/PTCA	32	34	31	31	28
Congestive heart failure	8	19	7	14	11
COPD	24	31	39	40	44
Creatinine >1.8 mg/dL, %	6	17	9	7	16
Mean mg/dl	1.1	1.5	1.2	1.2	1.3
Cardiac stress test, %
Not done	60	62	21	58	96
Normal result	28	25	59	28	1
Abnormal result	12	14	20	15	3
Previous treatment, %
Arterial bypass	5	32	3	3	1
CEA	13	11	7	6	8
Aneurysm repair	3	3	2	3	3
PTA/stent	5	20	3	3	2
Major amputation	1	4	1	0.4	1
Pre-op medications, %
ASA	83	64	69	65	50
Clopidogrel	17	7	6	6	2
ASA or clopidogrel	86	66	70	67	51
β-Blocker	83	79	83	75	65
Statin	67	54	58	60	31

CEA, Carotid endarterectomy; LEB, lower extremity bypass; EVAR, endovascular repair; CABG, coronary artery bypass grafting, PTCA, percutaneous transluminal coronary angioplasty; COPD, chronic obstructive pulmonary disease; PTA, percutaneous transluminal angioplasty; ASA, acetylsalicylic acid.

Carotid endarterectomy 

The indication for primary CEA was asymptomatic stenosis in 52%, ipsilateral cortical or ocular symptoms in 37%, and contralateral, vertebrobasilar or nonspecific symptoms in 11%. Preoperative imaging included duplex scan in 96%, magnetic resonance angiography (MRA) in 25%, computed tomography angiography (CTA) in 17%, and arteriogram in 7%. Stenosis >80% was present in 75% of operated patients.

CEA was performed by conventional (noneversion) technique in 85% and with general anesthesia in 91%. Among patients undergoing conventional CEA, 89% were patched, most frequently with bovine pericardium (61%) or polyester (23%). Electroencephalographic monitoring was performed in 54%. Shunting was performed routinely in 37% and for a specific indication in an additional 6%. A completion duplex scan was obtained in 34%.

Postoperative complications included re-exploration of the artery after closure, 3%; cranial nerve injury, 6% (most commonly the hypoglossal nerve); intravenous medication for hypotension or hypertension, 27%; and ipsilateral neurologic events, including transient ischemic attack, 0.4%, minor stroke, 0.5%, and major stroke, 0.2%. Nonspecific complications are listed for all operation types in Table II.

Table II. Postoperative complications

CEA, Carotid endarterectomy; LEB, lower extremity bypass; EVAR, endovascular repair; PRBC, packed red blood cell units; NA, not applicable; NR, not recorded.

Lower extremity bypass 

The indication for LEB for occlusive disease was critical ischemia in 77% and claudication in 23%. Previous ipsilateral infrainguinal bypass had been performed in 14% of patients, percutaneous transluminal angioplasty (PTA) or stenting in 7%, and minor amputation in 7%. Preoperative arteriography was performed in 81% of patients, duplex scanning in 40%, MRA in 10%, CTA in 8%, and vein mapping in 57%. General anesthesia was used in 71%. Graft origin was most frequently the common femoral artery (69%), followed by the superficial femoral artery (17%).

Graft recipient was above the above knee popliteal artery in 27%, below knee popliteal in 31%, and infrapopliteal in 42% of cases. Vein grafts were used in 74%, polytetrafluoroethylene (PTFE) grafts in 23%, and polyester grafts in 3%, with 3% having a vein cuff. Vein grafts were used in 83% of patients with below knee bypass, and prosthetic material was used in 61% of above knee bypass grafts. Concomitant iliac bypass, PTA, or stenting was performed in 10% of patients, and femoral endarterectomy was performed in 23%. Minor amputation was performed in 11% of patients, 5% required graft revision, and 2% underwent subsequent major amputation. At discharge, 90% of patients were ambulatory compared with 94% preoperatively, and 98% had patent bypass grafts. Life-table graft patency after discharge is shown in Fig 2.

• 
  • View Large Image
  • Download to PowerPoint

• Fig 2. 

  Life table-graft patency (%) after lower extremity bypass.

Abdominal aortic aneurysm repair 

Elective open repair was performed at a mean AAA diameter of 6.1 cm under general anesthesia alone in 30% and combined with epidural anesthesia in 70%. Retroperitoneal exposure was used in 25% and suprarenal clamping in 24% of cases. Prosthetic grafts were 18 mm median diameter, 88% polyester, 12% PTFE, with the distal anastomosis to the aorta in 52% of cases. Median blood loss was 1100 mL; 507 mL was autotransfused; no packed red blood cells (PRBC) were transfused, and 5 L of crystalloid was administered. Extubation was in the operating room for 66% of patients, and an additional 18% were extubated within 12 hours. Intravenous vasopressors were required by 31% of patients during a median ICU stay of 2 days.

Elective EVAR was performed at a mean AAA diameter of 5.7 cm under general anesthesia in 87% and regional anesthesia in 13%. Thirty-one percent were judged unfit for open repair. Graft manufacturer was Gore (Flagstaff, Ariz) in 38%, Medtronic (Minneapolis, Minn) in 34%, and Cook (Bloomington, Ind) in 24% of patients.

Median graft body and limb diameters were 26 and 15 mm. The internal iliac artery was covered unilaterally in 12% of cases and bilaterally in 2%. On completion, 2% of patients had type I or III endoleak, and 26% had type II endoleak. Conversion to open repair was required in 0.4%.

Median iodinated contrast used was 123 mL, blood loss was 250 mL, crystalloid administered was 2.5 L, and PRBC transfusion was 0 units. Postoperatively, 98% of patients were extubated in the operating room, and 3% required vasopressors during a median ICU stay of 0 days.

Open repair for a ruptured AAA was performed at a mean aneurysm diameter of 7.5 cm. Retroperitoneal exposure was used in 5% of patients and suprarenal clamping in 43%. Distal anastomosis was to the aorta in 63%. Median blood loss was 3000 mL, with 1140 mL autotransfusion, 6 U PRBC, and 7 L crystalloid. Only 4% of patients were extubated in the operating room, and an additional 18% were extubated ≤12 hours. Intravenous vasopressors were required by 58% of patients during a median intensive care unit stay of 4 days.

Key in-hospital outcomes for these procedure subgroups are summarized in Table III, including the overall regional result and the range of outcomes among the nine centers. Length of stay (LOS) from the date of operation to discharge was significantly increased by the occurrence of any of these postoperative complications (P < .001, Fig 3).

Table III. In-hospital outcomes

Outcome	Region results	Center range
CEA (primary), No.	2984	65-993
Any stroke or death, %	1.0	0-2.5
Any complication, %⁎	4.1	1.4-5.8
Unplanned return to OR, %	1.9	0-3.7
Mean length of stay, d†	1.5	1.2-2.7
Not discharged home‡	2.9	1.3-4.2
LEB (occlusive), No.	1521	51-634
Mortality, %	2.0	0.6-7.8
Major amputation, %	2.0	0-3.9
Any complication, %⁎	29	15-32
Unplanned return to OR, %	11.0	4-14
Mean length of stay, d†	6.9	3.7-9.6
Not discharged home, %‡	26	7-51
Open AAA (elective), No.	667	13-148
Mortality, %	2.9	0-5.9
Any complication, %⁎	36	25-45
Unplanned return to OR, %	6	0-23
Mean length of stay, d†	10	8-13
Not discharged home, %‡	17	8-20
EVAR (elective), No.	495	18-246
Mortality, %	0.4	0-2.4
Any complication, %⁎	11	7-30
Unplanned return to OR, %	1.4	0-3.7
Mean length of stay, d†	2.3	2.1-4.3
Not discharged home, %‡	3	0-12
Rupture AAA (open), No.	153	10-53
Mortality	34	19-50
Any complication⁎	76	69-100
Unplanned return to OR, %	29	14-60
Mean length of stay, d†	15	11-20
Not discharged home‡	51	29-58

CEA, Carotid endarterectomy; LEB, lower extremity bypass; EVAR, endovascular repair; OR, operating room.

• 
  • View Large Image
  • Download to PowerPoint

• Fig 3. 

  Preoperative medication usage among all patients. ASA, Acetylsalicylic acid.

Preoperative medications 

For all patients in the database, preoperative β-blocker usage increased from 72% in the first half of 2003 to 91% in the second half of 2006 (P < .001, Fig 4). The most pronounced increased occurred during 2004, when a specific quality-improvement project was initiated. This increase was due to β-blockers specifically initiated preoperatively, which increased from 19% of patients in the first half of 2003 to 34% in the second half of 2006 (P < .001) vs the percentage of patients receiving long-term β-blocker therapy, which did not change significantly over time (53% to 57%). Variation among surgeons who initiated preoperative β-blocker use in their patients was initially 39% to 100% in the first 6 months of 2003. Usage increased among all surgeons to 67% to 100% in the last 6 months of 2006. From 2003 to 2006, the usage of preoperative aspirin or clopidogrel increased from 73% to 83% and preoperative statin usage increased from 54% to 72% (both P < .001, Fig 4).

• 
  • View Large Image
  • Download to PowerPoint

• Fig 4. 

  Hospital length of stay from operation to discharge as a function of any complication listed in Table III. CEA, Carotid endarterectomy; LEB, lower extremity bypass; EVAR, endovascular repair.

One-year follow-up data 

Follow-up data were reported for 83% of patients at a mean interval of 365 days after operation. At the time of follow-up, 92% of patients were alive and living at home, 2% lived in nursing homes, and 6% had died since hospital discharge. Only 20% of patients were currently smoking, compared with 36% preoperatively (P < .001). At follow-up, medications being taken were aspirin in 81%, clopidogrel in 15%, β-blockers in 64%, and statins in 68%. Of patients with LEB, 22% were taking warfarin at follow-up. Key 1-year follow-up results for the five operation subgroups are listed in Table IV.

Table IV. One-year follow-up results

Result	Region result	Center range
CEA (primary), No.	1952	47-657
Mean follow-up, d	366	202-419
With follow-up, %	81	34-97
Persistent cranial nerve injury, %	1.0	0-1.6
Restenosis >70%, %	5.0	0-12
Reoperation or stent, %⁎	1.2	0-4.8
Ipsilateral stroke free, %⁎	98	95-100
Ipsilateral stroke-free survival, %⁎	95	85-96
LEB (occlusive), No.	1045	22-435
Mean follow-up, days	323	216-408
With follow-up, %	83	31-97
Ambulatory, %	91	73-96
Primary patency, %⁎	73	68-90
Primary-assisted patency, %⁎	79	68-85
Secondary patency, %⁎	81	72-86
Limb salvage, %⁎	83	76-100
Survival, %⁎	84	79-94
Open AAA (elective), No.	486	12-121
Mean follow-up, d	362	157-409
With follow-up, %	88	43-100
Reoperation, %⁎	3.3	1.7-8.3
Survival, %⁎	93	89-100
EVAR (elective), No.	320	27-173
Mean follow-up, days	369	364-373
With follow-up, %	84	38-100
Sac increase >5 mm, %	9	5-19
Endoleak, %
Type I or III	3.1	1.5-8.0
Type II	16	3-20
Reintervention, %⁎	4.0	0-8
Survival, %⁎	94	87-98
Rupture AAA (open), No.	108	17-32
Mean follow-up, d	374	328-469
With follow-up, %	89	60-100
Reoperation, %⁎	18	0-33
Survival, %⁎	48	44-58

CEA, Carotid endarterectomy; LEB, lower extremity bypass; EVAR, endovascular repair.

Validation 

The initial audit of administrative claims data and the VSGNNE registry revealed 6182 unique patients with potentially included operations and surgeons. Initially, the computer directly matched 66% of these patients, and an additional 7% were matched by hand after minor errors in data entry were corrected. Individual review of the remaining 1672 patients determined that 13% had operations that were not included in the registry that were identified by broad ICD-9 codes (such as femorofemoral bypass), 2% were errors based on incorrect ICD-9 coding, 5% had been done by nonparticipating surgeons, and 7% were operations that should have been submitted to VSGNNE. Data for these nonsubmitted 408 operations were requested and have been received to date for 368. This yielded complete data capture for 99% of procedures that should have been entered into the VSGNNE database by participating surgeons. Of these procedures, hospital mortality was correctly entered into the VSGNNE database for the 117 patients who died in-hospital. Discharge status disagreement between administrative data and VSGNNE data occurred in only 2.3% of discharges (home vs elsewhere).

Back to Article Outline

Discussion 

This work represents the results of a group of both academic and community vascular surgeons who are motivated to understand and improve their patient outcomes. We were strongly influenced by the successful model of continuous quality improvement demonstrated by the NNECDSG in many of our hospitals.² Our initial surgical results are comparable with single-center reports of outcomes for CEA, LEB, and AAA repair and exceed the results of some population-based reports.⁵, ⁶, ⁷, ⁸, ⁹, ¹⁰ They do not represent population-based results because not all hospitals and surgeons in the region participate. By conducting an audit of our procedure count and discharge status, however, our data represent an accurate accounting of the outcomes of self-selected surgeons who are committed to better understanding and improving their work.

The major accomplishment of the VSGNNE to date is to demonstrate feasibility of a regional quality assurance effort that has persisted for >4 years. This has provided valuable information that is not normally available to individual surgeons who, by nature, want to improve their results. It has provided regional benchmarking data for hospitals and surgeons with sufficient detail to allow analysis of variation in processes of care that could influence outcome. The commitment to 1-year follow-up data brought attention to missed follow-up visits, and in itself provided a vehicle for improvement. The process fulfills hospital needs for quality assurance programs required for continued accreditation and will likely provide an appropriate registry vehicle for surgeons to qualify for CMS payment incentive through the Physician Quality Reporting Initiative.¹¹

An early goal of the VSGNNE was to develop risk-adjustment algorithms to compare observed and expected outcomes at each center to identify variation due to processes of care rather than different patient characteristics.³ Given the low frequency of major adverse events, it has taken longer than expected to accrue sufficient numbers of patients to conduct this analysis, although this remains our long-term goal.

While waiting for future risk-adjusted analysis, we chose to focus on process improvement, initially identifying preoperative medication usage, and specifically β-blockade, as a target for improvement as a result of increasing evidence for benefit of this therapy.¹² In 2004, techniques to implement preoperative β-blockade were shared from each center, and quality improvement principles were discussed with the group by Eugene C. Nelson, a recognized quality improvement expert. Although we cannot prove that the observed improvement in β-blocker usage from 72% in the first 6 months of 2003 to 91% in the most recent 6 months was a result of this method, the most rapid improvement did correspond to the initial 6 months after our specific quality improvement discussions.

It is possible that a Hawthorne effect of providing feedback contributed to this improvement¹³ or that an emphasis on this topic in the medical literature or other hospital initiatives also encouraged the desired clinical behavior. Whatever the mechanism, the use of β-blockade now exceeds the 90% target threshold adopted by the VSGNNE, and antiplatelet and statin usage has also increased substantially during this interval. Maintenance of success and even further improvement appears significant, especially since others have reported a loss of initial improvement in β-blocker administration after completion of an initially successful quality improvement project.¹⁴

Compared with claim-based registries, a substantial advantage of the VSGNNE is the accurate recording of clinical details that may allow an understanding of changes necessary to improve outcome rather than simply reporting outcome alone. As an example, LOS is tracked by most hospitals as an important cost indicator. To safely shorten LOS, however, one must identify the root causes of clinical events—usually complications—that prolong LOS. In our registry, we have identified the substantial and varying impact of individual postoperative complications on LOS for each procedure. Furthermore, we have identified substantial variation in the frequency of these complications among the different centers and individual surgeons. Our next effort will be to benchmark best practices by comparative process analysis using site visits by participating surgeons to map the detailed processes of care performed by surgeons at centers with excellent results. This technique has been effectively used by the NNECDSG to reduce mortality following CABG.¹⁵

In addition to the landmark achievements of the NNECDSG, other successful registries have been developed. The Department of Veterans Affairs National Surgical Quality Improvement Program (NSQIP) is a model for providing risk-adjusted outcomes based on prospectively collected preoperative patient characteristics.¹⁶ These investigators have proven the superiority of this technique compared with the use of administrative claims data for analyzing such outcomes. They have pointed out the difficulty of using ICD-9 codes, as we also have witnessed, because of their lack of specificity. This is particularly important in vascular surgery, where many ICD-9 procedure codes cover a broad range of markedly different procedures.

The Society for Thoracic Surgeons National Cardiac Data Base is an excellent example of a voluntary registry that that provides nationally benchmarked, site-specific feedback and an opportunity for quality improvement.¹⁷ This group demonstrated a significant increase in preoperative β-blocker usage in patients undergoing CABG at sites that received low-intensity quality improvement interventions such as used by the VSGNNE.¹⁷ These included identification of a local opinion leader, providing benchmarking feedback, and techniques for quality improvement.

Several statewide quality improvement efforts have also demonstrated the value of confidential feedback of benchmark outcomes combined with meetings of participating surgeons to discuss variation in care process and outcomes. The Iowa Foundation for Medical Care demonstrated a substantial reduction in stroke or death after CEA,¹⁸ and the Alabama CABG Study Group demonstrated a significant increase in internal mammary artery usage by using these methods.¹⁹ Several vascular societies in Europe have also formed registries that have provided useful data.²⁰, ²¹

Not all vascular registries have been durable, however. The Cleveland Vascular Society, The Eastern Vascular Society, The Kentucky Vascular Society, and The South Carolina Vascular Surgical Society all formed database registries for member participation.²², ²³, ²⁴, ²⁵After initial enthusiasm, these were abandoned because of incomplete participation and missing information from data forms. Among reasons cited for failure was lack of perception of benefit by members and difficulty completing complex data forms.²² In this regard, we have learned the following lessons from our experience in the VSGNNE:

Finally, this effort requires significant resources. Central data collection and analysis was funded by an external grant, at an average cost of $100,000 per year. However, each hospital and group practice has borne the additional expense of data collection and data entry, which has required additional personnel, depending on the volume of the center. Surgeons from several hospitals that did not provide such assistance dropped out of the project after 1 year. This underscores the importance of involving not only surgeons but also research personnel and administrators from each center.

Back to Article Outline

Conclusion 

Our experience leads us to recommend this model of a regional cooperative quality improvement group to others who are interested in analyzing and improving their results. This has allowed nine hospitals and 48 surgeons to compare their outcomes with regional benchmarks for CEA, LEB, and AAA repair and to initiate process improvement efforts. The power of our registry increases with size, and momentum is maintained by feedback of key results to individual surgeons. We have begun the process of risk-adjustment to understand and identify variations in processes of care that could lead to future improvement. Relevant quality measures can now be accurately monitored, which not only allows quality improvement but also helps surgeons and hospitals prepare for pay-for-performance initiatives that are being developed. We are confident that our methodology will meet the demands of public accountability and also improve the quality of care for our patients.

Back to Article Outline

Author contributions 

Back to Article Outline

We gratefully acknowledge Gerald T. O’Connor, DSc, PhD, principal investigator of the Northern New England Cardiovascular Disease Study Group, who provided invaluable assistance in starting this project; Eugene C. Nelson, DSc, MPH, who provided quality-improvement consultation and instruction; and John H. Higgins, MS, who designed the data input system. In addition, we gratefully acknowledge the continued support of nurses, quality assurance staff, and research coordinators who collect data for submission to the registry at each participating hospital. Finally, we appreciate the initial support of two senior vascular surgeons, Carl E. Bredenberg, MD, and David B. Pilcher, MD, who were instrumental in developing this project.

Back to Article Outline

Appendix (online only) 

The following surgeons participated in the Vascular Study Group of Northern New England and contributed data to this report:

Catholic Medical Center, Manchester, New Hampshire

Central Maine Medical Center, Lewiston, Maine

Concord Hospital, Concord, New Hampshire

Cottage Hospital, Woodsville, New Hampshire

Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire

Eastern Maine Medical Center, Bangor, Maine

Fletcher Allen Health Care, Burlington, VT

Lakes Region General Hospital, Laconia, New Hampshire

Maine Medical Center, Portland, Maine

Back to Article Outline

References 

1. Hertzer NR. Outcome assessment in vascular surgery—results mean everything. J Vasc Surg. 1995;21:6–15
   • View In Article
   • Full Text
   • Full-Text PDF (918 KB)
   • CrossRef
2. O’Connor GT, Plume SK, Wennberg JE. Regional organization for outcomes research. Ann N Y Acad Sci. 1993;703:44–50discussion 50-1
   • View In Article
   • MEDLINE
   • CrossRef
3. O’Connor GT, Plume SK, Olmstead EM, Coffin LH, Morton JR, Maloney CT, et al. A regional prospective study of in-hospital mortality associated with coronary artery bypass grafting (The Northern New England Cardiovascular Disease Study Group). JAMA. 1991;266:803–809
   • View In Article
   • MEDLINE
4. O’Connor GT, Plume SK, Olmstead EM, Morton JR, Maloney CT, Nugent WC, et al. A regional intervention to improve the hospital mortality associated with coronary artery bypass graft surgery (The Northern New England Cardiovascular Disease Study Group). JAMA. 1996;275:841–846
   • View In Article
   • MEDLINE
5. Killeen SD, Andrews EJ, Redmond HP, Fulton GJ. Provider volume and outcomes for abdominal aortic aneurysm repair, carotid endarterectomy, and lower extremity revascularization procedures. J Vasc Surg. 2007;45:615–626
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (135 KB)
   • CrossRef
6. Conte MS, Bandyk DF, Clowes AW, Moneta GL, Seely L, Lorenz TJ, et al. 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
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (205 KB)
   • CrossRef
7. Dardik A, Lin JW, Gordon TA, Williams GM, Perler BA. Results of elective abdominal aortic aneurysm repair in the 1990s: A population-based analysis of 2335 cases. J Vasc Surg. 1999;30:985–995
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (85 KB)
   • CrossRef
8. Hallett JW, Pietropaoli JA, Ilstrup DM, Gayari MM, Williams JA, Meyer FB. Comparison of North American Symptomatic Carotid Endarterectomy Trial and population-based outcomes for carotid endarterectomy. J Vasc Surg. 1998;27:845–850discussion 851
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (80 KB)
   • CrossRef
9. Hertzer NR, Mascha EJ, Karafa MT, O’Hara PJ, Krajewski LP, Beven EG. Open infrarenal abdominal aortic aneurysm repair: the Cleveland Clinic experience from 1989 to 1998. J Vasc Surg. 2002;35:1145–1154
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (134 KB)
   • CrossRef
10. Hunink MG, Wong JB, Donaldson MC, Meyerovitz MF, Harrington DP. Patency results of percutaneous and surgical revascularization for femoropopliteal arterial disease. Med Decis Making. 1994;14:71–81
    • View In Article
    • MEDLINE
    • CrossRef
11. http://www.cms.hhs.gov/PQRI
    • View In Article
12. Poldermans D, Boersma E, Bax JJ, Thomson IR, van de Ven LL, Blankensteijn JD, et al. The effect of bisoprolol on perioperative mortality and myocardial infarction in high-risk patients undergoing vascular surgery (Dutch Echocardiographic Cardiac Risk Evaluation Applying Stress Echocardiography Study Group). N Engl J Med. 1999;341:1789–1794
    • View In Article
    • MEDLINE
    • CrossRef
13. Lied TR, Kazandjian VA. A Hawthorne strategy: implications for performance measurement and improvement. Clin Perform Qual Health Care. 1998;6:201–204
    • View In Article
    • MEDLINE
14. Cantor MN, Lavarias V, Lam S, Mount L, Laskova V, Nakhamiyayev V, et al. Barriers to implementing a surgical beta-blocker protocol. Jt Comm J Qual Patient Saf. 2005;31:640–648
    • View In Article
    • MEDLINE
15. Kasper JF, Plume SK, O’Connor GT. A methodology for QI in the coronary artery bypass grafting procedure involving comparative process analysis. QRB Qual Rev Bull. 1992;18:129–133
    • View In Article
    • MEDLINE
16. Best WR, Khuri SF, Phelan M, Hur K, Henderson WG, Demakis JG, et al. Identifying patient preoperative risk factors and postoperative adverse events in administrative databases: results from the Department of Veterans Affairs National Surgical Quality Improvement Program. J Am Coll Surg. 2002;194:257–266
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (93 KB)
    • CrossRef
17. Ferguson TB, Peterson ED, Coombs LP, Eiken MC, Carey ML, Grover FL, et al. Use of continuous quality improvement to increase use of process measures in patients undergoing coronary artery bypass graft surgery: a randomized controlled trial. JAMA. 2003;290:49–56
    • View In Article
    • CrossRef
18. Kresowik TF, Hemann RA, Grund SL, Hendel ME, Brenton M, Wiblin RT, et al. Improving the outcomes of carotid endarterectomy: results of a statewide quality improvement project. J Vasc Surg. 2000;31:918–926
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (49 KB)
    • CrossRef
19. Holman WL, Allman RM, Sansom M, Peterson ED, Hubbard SG, Delong JF, et al. Alabama coronary artery bypass grafting project: results of a statewide quality improvement initiative. JAMA. 2001;285:3003–3010
    • View In Article
    • MEDLINE
    • CrossRef
20. Kantonen I, Lepantalo M, Salenius JP, et al. Auditing a nationwide vascular registry--the 4-year Finnvasc experience (Finnvasc Study Group). Eur J Vasc Endovasc Surg. 1997;14:468–474
    • View In Article
    • Abstract
    • Full-Text PDF (2177 KB)
    • CrossRef
21. Bergqvist D, Troeng T, Einarsson E, Elfstrom J, Norgren L. Vascular surgical audit during a 5-year period (Steering committee on behalf of the Swedish Vascular Registry (Swedvasc)). Eur J Vasc Surg. 1994;8:472–477
    • View In Article
    • MEDLINE
    • CrossRef
22. Taylor SM, Robison JG, Langan EM, Crane MM. The pitfalls of establishing a statewide vascular registry: the South Carolina experience. Am Surg. 1999;65:513–518discussion 518-9
    • View In Article
    • MEDLINE
23. Hertzer NR, Avellone JC, Farrell CJ, Plecha FR, Rhodes RS, Sharp WV, et al. The risk of vascular surgery in a metropolitan community (With observations on surgeon experience and hospital size). J Vasc Surg. 1984;1:13–21
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (918 KB)
    • CrossRef
24. Karmody AM, Blumenberg RM, Wall CA. Preliminary experience with a large scale vascular registry. Am J Surg. 1983;146:162–163
    • View In Article
    • MEDLINE
    • CrossRef
25. Richardson JD, Main KA. Carotid endarterectomy in the elderly population: a statewide experience. J Vasc Surg. 1989;9:65–73
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (865 KB)
    • CrossRef