Evaluation of Surveillance Bias and the Validity of the Venous Thromboembolism Quality Measure. Bilimoria KY, Chung J, Ju MH, Haut ER, Bentrem DJ, Ko CY, Baker DW. JAMA. 2013 Oct 9; 310(14):1482-9.
Several quality improvement groups have attempted to standardize quality of care metrics to objectively inform hospitals and institutions about how they are performing in comparison to their peers. One such quality measure adopted over a decade ago by the Agency for Healthcare Research and Quality (AHRQ) is the PSI-12 indicator for presence of a postoperative VTE complication. The PSI-12 is frequently utilized for public reporting initiatives by the Centers for Medicare and Medicaid Services (CMS) and other reporting agencies, often serving for the basis for penalization or reward. Studies involving the PSI-12 in trauma populations had already found that increased surveillance for VTEs (PSI-12 rates) did not result in reduced VTE rates.
In this study, a large sample of 2838 hospitals derived from merging 2 major national databases, (Hospital Compare and the American Hospital Association) was cross-referenced to the Medicare claims of 954 926 postoperative patients during the same years. Patients with 11 operations were studied including: total knee arthroplasty (50% of patients), colectomy (16.7%), CABG (13.8%) and 8 other major operations. Publicly reported SCIP-VTE-2 rates were used to indicate adherence to VTE prophylaxis protocols (chemical/mechanical) and structural measures of hospital quality level (≥300beds bed size, Joint Commission accreditation, Level 1 trauma designation and others) were used to rate the structural quality of hospitals. Imaging rates were defined as the use of duplex ultrasonography, chest computed tomography/MRI and VQ scans.
Compared to hospitals possessing the lowest structural quality characteristics, those with highest quality characteristics demonstrated the highest VTE prophylaxis adherence (Low-93.3% vs. high-95.5%, p<0.001) but paradoxically significantly higher risk-adjusted VTE rates (Low-4.8 vs. high-6.4 VTE rate/1000pt discharges, p<0.001). Hospitals with the highest Imaging rates (167.5 imaging rate /1000 pt discharges) had greatest VTE prophylaxis adherence (93%) but also the highest mean risk-adjusted VTE rate (13.5/1000pt discharges) as compared to hospitals with the lowest imaging rates (31.5 imaging rate /1000 pt discharges; 91.8% adherence; 5.0 VTE rates/1000pt discharges, p<0.001 for all comparisons). This stepwise relationship scored true whether imaging was for VTE, DVT or PE. Further supporting the concept of surveillance bias in the realm of VTE rate as a quality indicator, the authors found a significant, albeit weak, correlation between higher VTE prophylaxis adherence and higher risk-adjusted VTE rates (r2=4.2%; p =0.03).
The authors conclude that even though metrics in quality are crucial, some may carry a significant surveillance bias as demonstrated in the PSI-12 AHRQ indicator of postoperative VTE rates. This bias may lead to paradoxically misrepresenting the quality of administered care by reporting greater complication rates in institutions with greater surveillance and interest in finding their complications.
Association of Coworker Reports About Unprofessional Behavior by Surgeons With Surgical Complications in Their Patients. Cooper WO, Spain DA, Guillamondegui O, Kelz RR, Domenico HJ, Hopkins J, Sullivan P, Moore IN, Pichert JW, Catron TF, Webb LE, Dmochowski RR, Hickson GB.
JAMA Surg. 2019 Sep 1;154(9):828-834.
The current ever-increasing efforts of improving patient safety and quality of care have led certain groups to strive to become highly reliable organizations to specifically minimize threats to delivery of reliable care. Reliable care depends on well-functioning teams that require optimal communication and dialogue, mutual respect and situational awareness of team members and leaders. Surgeons have culturally been associated with specific personality traits and behaviors which have been deemed as unprofessional and yet, are the team leaders in the care of surgical patients. The authors hypothesized that perceived surgeon unprofessional behavior may directly affect team dynamics and patient care as well as outcomes.
In this retrospective cohort study the authors sought to cross-reference patient data (1/2012-12/2016) from 2 large, geographically diverse academic health systems participating in NSQIP and their electronic reports of coworker concerns about surgeons’ unprofessional behaviors (1/2009-12/2016). Independent coders classified coworker reports as 1) concerns for unsafe care, 2) clear and respectful communication, 3) integrity and 4) responsibility. Patient and surgeon identifiers were not utilized. Complications were the standard NSQIP medical (pneumonia, reintubation, AKI, stroke etc.) or surgical (SSSI, wound disruption) complications occurring within 30 days of operation. 13 553 patients (54% female, mean 57 y/o) undergoing an operation (77% general surgery, 8% ortho, 6% vascular, 1% OBGYN and others) performed by 202 surgeons (70% male) in the study time period were included. Mean number of coworker unprofessionalism reports per surgeon was 1.3 (range 0-2) reports in the 36 months preceding the operation. Surgeons in the group with highest number of reports (>4) had a mean of 6.1 reports. Those in the group with the lowest number of reports (0) were mostly women (p<0.001). The cases of surgeons who had the highest number of reports had higher ASA classification, were more likely to have a contaminated wounds and had longer operative times. 11.6% (1 583) of postoperative patients suffered complications (6% surgical, 8% medical). In the observed 30 days, 140 (1%) patients died, 473 (3.5%) returned to the OR and 1053 (7.7%) were readmitted to hospital. Highest coworker report surgeons had the highest overall complications as compared to lowest report surgeons (14% vs. 10%, p<0.001). These results were mirrored with both medical and surgical complications, separately. Death, reoperation and readmission rates were no different among surgeons ranked by number of coworker reports. Analyses controlling for patient factors (age, sex, ASA class, operative wound class, operative time etc.) revealed a stepwise increase in estimated risk of complications per surgeon (0 reports: baseline; 1-3 reports: 18.1% higher; >4 reports: 31.7% higher, p<0.001). Sensitivity analyses controlling for surgeon’s years of experience, study site revealed identical findings.
While not demonstrating causality, the authors report and verifiable association between surgeons receiving greater than 4 professionalism-related reports and greater postoperative complications. While it may be self-evident that better team collaboration and communication occurs within care groups lead by a professional and courteous surgeon, this study provides further evidence that this may also directly impact patients’ postoperative outcomes.