ARTICLE IN PRESS
Association between quality of care and complications after abdominal surgery Simon Bergman, MD, MSc, FACS, FRCSC,a,b Melina Deban,a Vanessa Martelli,a Michele Monette, OT, MSc,b Nadia Sourial, MSc,b Fadi Hamadani, MD,a Debby Teasdale, RN,b Christina Holcroft, ScD,c Helena Zakrzewski, BSc,a and Shannon Fraser, MD, MSc, FACS, FRCSC,a Montreal, Quebec, Canada
Background. Measuring the quality of surgical care is essential to identifying areas of weakness in the delivery of effective surgical care and to improving patient outcomes. Our objectives were to (1) assess the quality of surgical care delivered to adult patients; and (2) determine the association between quality of surgical care and postoperative complications. Methods. This retrospective, pilot, cohort study was conducted at a single university-affiliated institution. Using the institution’s National Surgical Quality Improvement Program database (2009– 2010), 273 consecutive patients $18 years of age who underwent elective major abdominal operations were selected. Adherence to 10 process-based quality indicators (QIs) was measured and quantified by calculating a patient quality score (no. of QIs passed/no. of QIs eligible). A pass rate for each individual QI was also calculated. The association between quality of surgical care and postoperative complications was assessed using an incidence rate ratio, which was estimated from a Poisson regression. Results. The mean overall patient quality score was 67.2 ± 14.4% (range, 25–100%). The mean QI pass rate was 65.9 ± 26.1%, which varied widely from 9.6% (oral intake documentation) to 95.6% (prophylactic antibiotics). Poisson regression revealed that as the quality score increased, the incidence of postoperative complications decreased (incidence rate ratio, 0.19; P = .011). A sensitivity analysis revealed that this association was likely driven by the postoperative ambulation QI. Conclusion. Higher quality scores, mainly driven by early ambulation, were associated with fewer postoperative complications. QIs with unacceptably low adherence were identified as targets for future quality improvement initiatives. (Surgery 2014;j:j-j.) From the Department of Surgery, aCentre for Clinical Epidemiology and Community Studies,b the SolidageMcGill University/Universite de Montre al Research Group on Frailty and Aging, Lady Davis Institute for Medical Research, and the Centre for Clinical Epidemiology and Community Studies,c Lady Davis Institute for Medical Research, Jewish General Hospital, McGill University, Montreal, Quebec, Canada
IN SURGERY, there is an increasing recognition that patient outcomes vary by provider, with surgeons and hospitals being asked to provide evidence for the quality of care that they deliver.1-4 Measuring Funded by the Lady Davis Institute for Medical Research, Jewish General Hospital, Start-up Fund. Presented at the American College of Surgeons 97th Annual Clinical Congress, Medical Student Program, San Francisco, California, October 24, 2011; and the 10th Annual L.D. Maclean Day, General Surgery, Montreal General Hospital, Montreal, Quebec, Canada, April 18, 2012. Accepted for publication December 27, 2013. Reprint requests: Simon Bergman, MD, MSc, FACS, FRCSC, Jewish General Hospital, Pavilion A-515, 3755 C^ ote-Sainte-Catherine, Montr eal, Quebec, H3T 1E2 Canada. E-mail: simon. [email protected]. 0039-6060/$ - see front matter Ó 2014 Mosby, Inc. All rights reserved. http://dx.doi.org/10.1016/j.surg.2013.12.031
quality of surgical care allows clinicians and administrators to identify areas of weakness or of low performance.5 Such identification may lead to improved patient outcomes and to safer surgical services.5-8 Despite the widespread use of outcome measurement in surgery (what happens to the patient), process-based measurement (how care is delivered) may be a more desirable means of quality of care assessment.1,9 Processes are directly related to the care delivered to patients, thereby representing ideal targets for developing or for monitoring quality improvement initiatives. The literature on perioperative process measurement is sparse. Most often, the impact of a single operative process is measured, such as the impact of b-blockade. When adherence to a set of evidence-based processes (or bundle, as defined by the Institute for Healthcare Improvement)10 is measured, it is SURGERY 1
ARTICLE IN PRESS 2 Bergman et al
usually compared with an isolated, directly related outcome, such as surgical site infections in the Surgical Infection Prevention Project.11 By combining survey methodology and administrative databases, the relationship between quality of care and outcomes has been indirectly described, demonstrating that adhering to several postoperative care processes was associated with lower than expected mortality.12 A single indicator could not summarize the overall quality of care delivered.13 From a clinical standpoint, superior outcomes are most likely owing to a greater adherence to a broad array of processes and variability in quality of care is likely to be reflected in more than what can be expressed by a single outcome.14 The Assessing Care of Vulnerable Elders (ACOVE-3) group9 and McGory et al15 previously developed a set of perioperative quality indicators (QIs). Although these QIs are intended for the elderly population, many of the processes of care defined by these QIs may also be applicable to a younger population. For the purposes of this pilot study, we selected a comprehensive set of 10 QIs that define a broad range of processes of care and may have an impact on several different outcomes. The objectives of this pilot study were to (1) describe the quality of surgical care at our institution by measuring adherence to a comprehensive set of perioperative process-based QIs; and (2) assess the association between quality of care and postoperative complications. METHODS Study design and patient population. This retrospective pilot cohort study was conducted at a single, university-affiliated institution. Inclusion criteria were patients $18 years who underwent elective major abdominal operations. Major abdominal operation included any operation of the abdominal wall or cavity performed by a general or colorectal surgeon and requiring postoperative hospitalization. To provide sufficient time to evaluate the selected QIs, duration of stay had to be $2 days. Those undergoing emergency operations, who had a previous operation within 30 days of elective admission, or who were transferred from another institution were excluded. The institution’s National Surgical Quality Improvement Program (NSQIP) database, which contains a random sample of patients and procedures, was used to select 300 consecutive patients who met these criteria as of November 1, 2009. After reviewing the selected cases, 27 patients were excluded from the analysis because
Surgery j 2014
they were found to be misclassified. A review of the electronic medical records was done for 273 patients. Ethics approval was obtained from our institutional Research Ethics Office. Study parameters. Patient baseline characteristics included age, gender, smoker/nonsmoker status, preoperative functional status (independent, partially dependent, or totally dependent), Charlson Comorbidity Index (CCI),16 and wound class. The CCI is a weighted score of 23 conditions, with assigned weights of 1, 2, 3, or 6 depending on the risk of dying from each condition. The CCI has been shown to predict prognosis and health service utilization.16 Wound class is defined intraoperatively as clean, clean/contaminated, contaminated, or dirty/infected. The primary outcome for the study was the number of postoperative complications. Other outcomes included duration of stay and mortality. Patient baseline characteristics and outcomes were abstracted from the hospital’s NSQIP database. Process measurement. A total of 60 QIs developed by ACOVE-3 and9 McGory et al15 and one inhouse QI (surgical safety checklist QI) were first selected by the principal investigator. A Delphi consensus survey was then conducted to determine those QIs most relevant to our patient population and those that could be feasibly extracted from our institution’s electronic medical records. Twentyfive experts from our institution were involved in the Delphi consensus survey: 8 surgeons (general, colorectal, and vascular surgeons, chief and executive director), 4 geriatricians, 2 perioperative medicine physicians, 9 nurses (chief, head, geriatric, preoperative screening, and operating room nurses), the Chair of the Quality Committee, and the NSQIP manager. Thirty-five QIs were considered to be both relevant and feasible by these experts. Given that this study included patients $18 years of age, a subset of 10 QIs was selected as these were considered by our group to be applicable to all general surgery patients regardless of age (Table I). These 10 QIs could also be abstracted in a retrospective fashion using our institution’s electronic medical records, whereas other potential QIs would have required review of paper charts or direct observation. We developed a standardized abstraction manual and data abstraction form according to validated best practice guidelines for retrospective medical record review.17,18 To measure quality of surgical care, 2 investigators reviewed the institution’s electronic medical records for 273 patients. The eligibility of each QI was first assessed for each patient (eg, a patient without diabetes
ARTICLE IN PRESS Bergman et al 3
Surgery Volume j, Number j
Table I. Process-based quality indicators Quality indicator Prophylactic antibiotics Postoperative euglycemia Prophylactic VTE therapy Central venous line Urinary catheter Postoperative ambulation Medication list Pressure ulcer risk assessment Oral intake documentation Surgical safety checklist
Description Peri-operative administration of prophylactic antibiotics within 1 hour before incision and discontinued within 24 hours after surgery Postoperative treatment of diabetes (or documentation to attempt) to keep blood sugar below 10 mmol/L on day of surgery and the first 2 postoperative days Treatment with deep venous thrombosis prophylaxis Documentation of daily examination of line site for signs of infection and continued need for the central line Documentation of the continued need for the catheter every day until its removal Ambulation should be performed by postoperative day 2 Presence of an up-to-date medication list in the medical record Performance of a risk assessment for pressure ulcers using a standardized scale upon admission Documentation of oral intake during the hospitalization Use of the surgical safety checklist
VTE, Venous thromboembolism.
mellitus would not be eligible for a QI pertaining to this disease). If the patient was eligible, adherence to that QI was then assessed. To test the inter-rater reliability, a third investigator independently evaluated QIs for 39 patients randomly selected from the cohort. Statistical analysis. Patient baseline characteristics (age, gender, smoker/nonsmoker status, CCI, functional status, wound class) and outcomes (patient complications, duration of stay, mortality) were summarized. The pass rate for each individual QI was calculated as the number of patients who passed each particular QI divided by the number of patients eligible for each particular QI. The Chi-square or Fisher’s exact test was used to test for an association between the occurrence of complications and QI adherence. P values were adjusted for multiple testing.19 The patient quality score was calculated for each patient as the number of QIs passed divided by the number of QIs for which each patient was eligible. A Poisson regression was used to test for an association between patient quality score and the occurrence of complications, which was adjusted for other patient characteristics. The model was initially adjusted for age, sex, smoker/nonsmoker status, CCI, functional status, and wound class. Model selection was performed using the Akaike information criteria to select the best model, balancing model complexity and model fit. No overdispersion was detected in the model. Final adjustment variables were age, sex, CCI, and wound class. An incidence rate ratio (IRR) for each model parameter was estimated from Poisson regression, which represented the change in the incidence of complications for a 1-point increase
in the model parameter. For example, for the patient quality score, if the rate ratio was >1, the incidence of complications increased with a 1point increase in the patient quality score, and if the rate ratio was 50%. In unadjusted analyses testing for an association between adherence to each individual QI and the occurrence of complications, adherence to the postoperative ambulation QI was associated with fewer complications (P < .001; Table IV). No significant associations were found between adherence to the other QIs and the occurrence of complications. Quality of care and postoperative complications. The mean overall patient quality score was 67.2 ± 14.4% (range, 25–100%). Poisson regression showed that as the quality score increased, the incidence of postoperative complications decreased (IRR, 0.19; P = .011). Conversely, as the CCI and wound class increased, the incidence of complications also increased (IRR, 1.09 [P = .002] and IRR, 1.68 [P < .001], respectively) (Table V). To explore whether the association between patient quality score and the occurrence of complications was being driven by the postoperative ambulation QI (the only QI found to be significantly associated with the occurrence of complications; Table IV), a sensitivity analysis was performed removing the postoperative ambulation QI from the overall patient quality score in the multivariable Poisson regression. Results of the model demonstrated that the patient quality score
Surgery j 2014
Table II. Baseline patient (n = 273) Characteristic Age, mean (SD) Male, n (%) CCI, mean (SD) Functionally independent, n (%) Nonsmoker, n (%) Wound class, n (%) Clean Clean/contaminated Contaminated Dirty/infected
Value 64.0 133 3.4 267 237
(15.3) (48.7) (3.3) (97.8) (86.8)
39 214 13 7
(14.3) (78.4) (4.8) (2.6)
CCI, Charlson Comorbidity Index; SD, standard deviation.
Table III. Patient complications Complication
n (%)
Any complication No. of complications 0 1 2 $3 Type of complication Bleeding (transfusion) Organ space infection Superficial incisional infection Urinary tract infection Sepsis Intubation Ventilator Myocardial infarction Deep vein thrombosis Progressive renal failure Wound disruption Pneumonia Pulmonary embolus Septic shock Deep incisional infection Acute renal failure Cardiac arrest (cardiopulmonary resuscitation)
82 (30.0) 191 53 18 11
(70.0) (19.4) (6.6) (4.1)
35 20 19 14 8 6 3 3 3 2 2 2 2 2 1 1 1
(12.8) (7.3) (7.0) (5.1) (2.9) (2.2) (1.1) (1.1) (1.1) (0.7) (0.7) (0.7) (0.7) (0.7) (0.4) (0.4) (0.4)
without the postoperative ambulation QI was not associated with complications (IRR, 0.86; P = .851). The inter-rater analysis showed a mean percent agreement of 79.4% (range, 53.8–94.9%) between the 2 raters and a mean reliability coefficient of 0.71 (range, 0.351–0.940; Table VI). In general, reliability was higher for the QIs rated based on objective measures or check boxes (prophylactic antibiotics, postoperative euglycemia, reconciliation of medication list, and surgical safety checklist).
ARTICLE IN PRESS Bergman et al 5
Surgery Volume j, Number j
Table IV. Adherence to quality indicators and association with postoperative complications
Quality indicator (n = number triggered) Prophylactic antibiotics (n = 273) Postoperative euglycemia (n = 47) Prophylactic venous thromboembolism therapy (n = 273) Central venous line (n = 32) Urinary catheter (n = 241) Postoperative ambulation (n = 273) Medication list (n = 273) Pressure ulcer risk assessment (n = 214) Oral intake documentation (n = 273) Surgical safety checklist (n = 253)
Passed/ failed
n
%
Patients with complications (%)
Passed Failed Passed Failed Passed Failed Passed Failed Passed Failed Passed Failed Passed Failed Passed Failed Passed Failed Passed Failed
261 12 18 28 193 80 26 6 175 66 168 105 219 54 117 97 26 247 235 18
95.6 4.4 38.3 61.7 70.7 29.3 81.3 18.8 72.6 27.4 61.5 38.5 80.2 19.8 54.7 45.3 9.5 90.5 92.9 7.1
30.7 16.7 38.9 58.6 30.1 30.0 69.2 66.7 30.9 34.9 21.4 43.8 30.6 27.8 33.3 28.9 11.5 32.0 31.5 11.1
Unadjusted P value
Adjusted P value*
.520
.993
.188
.993
.993
.993
.999y
.999
.554
.993
Association between quality of care and complications after abdominal surgery Simon Bergman, MD, MSc, FACS, FRCSC,a,b Melina Deban,a Vanessa Martelli,a Michele Monette, OT, MSc,b Nadia Sourial, MSc,b Fadi Hamadani, MD,a Debby Teasdale, RN,b Christina Holcroft, ScD,c Helena Zakrzewski, BSc,a and Shannon Fraser, MD, MSc, FACS, FRCSC,a Montreal, Quebec, Canada
Background. Measuring the quality of surgical care is essential to identifying areas of weakness in the delivery of effective surgical care and to improving patient outcomes. Our objectives were to (1) assess the quality of surgical care delivered to adult patients; and (2) determine the association between quality of surgical care and postoperative complications. Methods. This retrospective, pilot, cohort study was conducted at a single university-affiliated institution. Using the institution’s National Surgical Quality Improvement Program database (2009– 2010), 273 consecutive patients $18 years of age who underwent elective major abdominal operations were selected. Adherence to 10 process-based quality indicators (QIs) was measured and quantified by calculating a patient quality score (no. of QIs passed/no. of QIs eligible). A pass rate for each individual QI was also calculated. The association between quality of surgical care and postoperative complications was assessed using an incidence rate ratio, which was estimated from a Poisson regression. Results. The mean overall patient quality score was 67.2 ± 14.4% (range, 25–100%). The mean QI pass rate was 65.9 ± 26.1%, which varied widely from 9.6% (oral intake documentation) to 95.6% (prophylactic antibiotics). Poisson regression revealed that as the quality score increased, the incidence of postoperative complications decreased (incidence rate ratio, 0.19; P = .011). A sensitivity analysis revealed that this association was likely driven by the postoperative ambulation QI. Conclusion. Higher quality scores, mainly driven by early ambulation, were associated with fewer postoperative complications. QIs with unacceptably low adherence were identified as targets for future quality improvement initiatives. (Surgery 2014;j:j-j.) From the Department of Surgery, aCentre for Clinical Epidemiology and Community Studies,b the SolidageMcGill University/Universite de Montre al Research Group on Frailty and Aging, Lady Davis Institute for Medical Research, and the Centre for Clinical Epidemiology and Community Studies,c Lady Davis Institute for Medical Research, Jewish General Hospital, McGill University, Montreal, Quebec, Canada
IN SURGERY, there is an increasing recognition that patient outcomes vary by provider, with surgeons and hospitals being asked to provide evidence for the quality of care that they deliver.1-4 Measuring Funded by the Lady Davis Institute for Medical Research, Jewish General Hospital, Start-up Fund. Presented at the American College of Surgeons 97th Annual Clinical Congress, Medical Student Program, San Francisco, California, October 24, 2011; and the 10th Annual L.D. Maclean Day, General Surgery, Montreal General Hospital, Montreal, Quebec, Canada, April 18, 2012. Accepted for publication December 27, 2013. Reprint requests: Simon Bergman, MD, MSc, FACS, FRCSC, Jewish General Hospital, Pavilion A-515, 3755 C^ ote-Sainte-Catherine, Montr eal, Quebec, H3T 1E2 Canada. E-mail: simon. [email protected]. 0039-6060/$ - see front matter Ó 2014 Mosby, Inc. All rights reserved. http://dx.doi.org/10.1016/j.surg.2013.12.031
quality of surgical care allows clinicians and administrators to identify areas of weakness or of low performance.5 Such identification may lead to improved patient outcomes and to safer surgical services.5-8 Despite the widespread use of outcome measurement in surgery (what happens to the patient), process-based measurement (how care is delivered) may be a more desirable means of quality of care assessment.1,9 Processes are directly related to the care delivered to patients, thereby representing ideal targets for developing or for monitoring quality improvement initiatives. The literature on perioperative process measurement is sparse. Most often, the impact of a single operative process is measured, such as the impact of b-blockade. When adherence to a set of evidence-based processes (or bundle, as defined by the Institute for Healthcare Improvement)10 is measured, it is SURGERY 1
ARTICLE IN PRESS 2 Bergman et al
usually compared with an isolated, directly related outcome, such as surgical site infections in the Surgical Infection Prevention Project.11 By combining survey methodology and administrative databases, the relationship between quality of care and outcomes has been indirectly described, demonstrating that adhering to several postoperative care processes was associated with lower than expected mortality.12 A single indicator could not summarize the overall quality of care delivered.13 From a clinical standpoint, superior outcomes are most likely owing to a greater adherence to a broad array of processes and variability in quality of care is likely to be reflected in more than what can be expressed by a single outcome.14 The Assessing Care of Vulnerable Elders (ACOVE-3) group9 and McGory et al15 previously developed a set of perioperative quality indicators (QIs). Although these QIs are intended for the elderly population, many of the processes of care defined by these QIs may also be applicable to a younger population. For the purposes of this pilot study, we selected a comprehensive set of 10 QIs that define a broad range of processes of care and may have an impact on several different outcomes. The objectives of this pilot study were to (1) describe the quality of surgical care at our institution by measuring adherence to a comprehensive set of perioperative process-based QIs; and (2) assess the association between quality of care and postoperative complications. METHODS Study design and patient population. This retrospective pilot cohort study was conducted at a single, university-affiliated institution. Inclusion criteria were patients $18 years who underwent elective major abdominal operations. Major abdominal operation included any operation of the abdominal wall or cavity performed by a general or colorectal surgeon and requiring postoperative hospitalization. To provide sufficient time to evaluate the selected QIs, duration of stay had to be $2 days. Those undergoing emergency operations, who had a previous operation within 30 days of elective admission, or who were transferred from another institution were excluded. The institution’s National Surgical Quality Improvement Program (NSQIP) database, which contains a random sample of patients and procedures, was used to select 300 consecutive patients who met these criteria as of November 1, 2009. After reviewing the selected cases, 27 patients were excluded from the analysis because
Surgery j 2014
they were found to be misclassified. A review of the electronic medical records was done for 273 patients. Ethics approval was obtained from our institutional Research Ethics Office. Study parameters. Patient baseline characteristics included age, gender, smoker/nonsmoker status, preoperative functional status (independent, partially dependent, or totally dependent), Charlson Comorbidity Index (CCI),16 and wound class. The CCI is a weighted score of 23 conditions, with assigned weights of 1, 2, 3, or 6 depending on the risk of dying from each condition. The CCI has been shown to predict prognosis and health service utilization.16 Wound class is defined intraoperatively as clean, clean/contaminated, contaminated, or dirty/infected. The primary outcome for the study was the number of postoperative complications. Other outcomes included duration of stay and mortality. Patient baseline characteristics and outcomes were abstracted from the hospital’s NSQIP database. Process measurement. A total of 60 QIs developed by ACOVE-3 and9 McGory et al15 and one inhouse QI (surgical safety checklist QI) were first selected by the principal investigator. A Delphi consensus survey was then conducted to determine those QIs most relevant to our patient population and those that could be feasibly extracted from our institution’s electronic medical records. Twentyfive experts from our institution were involved in the Delphi consensus survey: 8 surgeons (general, colorectal, and vascular surgeons, chief and executive director), 4 geriatricians, 2 perioperative medicine physicians, 9 nurses (chief, head, geriatric, preoperative screening, and operating room nurses), the Chair of the Quality Committee, and the NSQIP manager. Thirty-five QIs were considered to be both relevant and feasible by these experts. Given that this study included patients $18 years of age, a subset of 10 QIs was selected as these were considered by our group to be applicable to all general surgery patients regardless of age (Table I). These 10 QIs could also be abstracted in a retrospective fashion using our institution’s electronic medical records, whereas other potential QIs would have required review of paper charts or direct observation. We developed a standardized abstraction manual and data abstraction form according to validated best practice guidelines for retrospective medical record review.17,18 To measure quality of surgical care, 2 investigators reviewed the institution’s electronic medical records for 273 patients. The eligibility of each QI was first assessed for each patient (eg, a patient without diabetes
ARTICLE IN PRESS Bergman et al 3
Surgery Volume j, Number j
Table I. Process-based quality indicators Quality indicator Prophylactic antibiotics Postoperative euglycemia Prophylactic VTE therapy Central venous line Urinary catheter Postoperative ambulation Medication list Pressure ulcer risk assessment Oral intake documentation Surgical safety checklist
Description Peri-operative administration of prophylactic antibiotics within 1 hour before incision and discontinued within 24 hours after surgery Postoperative treatment of diabetes (or documentation to attempt) to keep blood sugar below 10 mmol/L on day of surgery and the first 2 postoperative days Treatment with deep venous thrombosis prophylaxis Documentation of daily examination of line site for signs of infection and continued need for the central line Documentation of the continued need for the catheter every day until its removal Ambulation should be performed by postoperative day 2 Presence of an up-to-date medication list in the medical record Performance of a risk assessment for pressure ulcers using a standardized scale upon admission Documentation of oral intake during the hospitalization Use of the surgical safety checklist
VTE, Venous thromboembolism.
mellitus would not be eligible for a QI pertaining to this disease). If the patient was eligible, adherence to that QI was then assessed. To test the inter-rater reliability, a third investigator independently evaluated QIs for 39 patients randomly selected from the cohort. Statistical analysis. Patient baseline characteristics (age, gender, smoker/nonsmoker status, CCI, functional status, wound class) and outcomes (patient complications, duration of stay, mortality) were summarized. The pass rate for each individual QI was calculated as the number of patients who passed each particular QI divided by the number of patients eligible for each particular QI. The Chi-square or Fisher’s exact test was used to test for an association between the occurrence of complications and QI adherence. P values were adjusted for multiple testing.19 The patient quality score was calculated for each patient as the number of QIs passed divided by the number of QIs for which each patient was eligible. A Poisson regression was used to test for an association between patient quality score and the occurrence of complications, which was adjusted for other patient characteristics. The model was initially adjusted for age, sex, smoker/nonsmoker status, CCI, functional status, and wound class. Model selection was performed using the Akaike information criteria to select the best model, balancing model complexity and model fit. No overdispersion was detected in the model. Final adjustment variables were age, sex, CCI, and wound class. An incidence rate ratio (IRR) for each model parameter was estimated from Poisson regression, which represented the change in the incidence of complications for a 1-point increase
in the model parameter. For example, for the patient quality score, if the rate ratio was >1, the incidence of complications increased with a 1point increase in the patient quality score, and if the rate ratio was 50%. In unadjusted analyses testing for an association between adherence to each individual QI and the occurrence of complications, adherence to the postoperative ambulation QI was associated with fewer complications (P < .001; Table IV). No significant associations were found between adherence to the other QIs and the occurrence of complications. Quality of care and postoperative complications. The mean overall patient quality score was 67.2 ± 14.4% (range, 25–100%). Poisson regression showed that as the quality score increased, the incidence of postoperative complications decreased (IRR, 0.19; P = .011). Conversely, as the CCI and wound class increased, the incidence of complications also increased (IRR, 1.09 [P = .002] and IRR, 1.68 [P < .001], respectively) (Table V). To explore whether the association between patient quality score and the occurrence of complications was being driven by the postoperative ambulation QI (the only QI found to be significantly associated with the occurrence of complications; Table IV), a sensitivity analysis was performed removing the postoperative ambulation QI from the overall patient quality score in the multivariable Poisson regression. Results of the model demonstrated that the patient quality score
Surgery j 2014
Table II. Baseline patient (n = 273) Characteristic Age, mean (SD) Male, n (%) CCI, mean (SD) Functionally independent, n (%) Nonsmoker, n (%) Wound class, n (%) Clean Clean/contaminated Contaminated Dirty/infected
Value 64.0 133 3.4 267 237
(15.3) (48.7) (3.3) (97.8) (86.8)
39 214 13 7
(14.3) (78.4) (4.8) (2.6)
CCI, Charlson Comorbidity Index; SD, standard deviation.
Table III. Patient complications Complication
n (%)
Any complication No. of complications 0 1 2 $3 Type of complication Bleeding (transfusion) Organ space infection Superficial incisional infection Urinary tract infection Sepsis Intubation Ventilator Myocardial infarction Deep vein thrombosis Progressive renal failure Wound disruption Pneumonia Pulmonary embolus Septic shock Deep incisional infection Acute renal failure Cardiac arrest (cardiopulmonary resuscitation)
82 (30.0) 191 53 18 11
(70.0) (19.4) (6.6) (4.1)
35 20 19 14 8 6 3 3 3 2 2 2 2 2 1 1 1
(12.8) (7.3) (7.0) (5.1) (2.9) (2.2) (1.1) (1.1) (1.1) (0.7) (0.7) (0.7) (0.7) (0.7) (0.4) (0.4) (0.4)
without the postoperative ambulation QI was not associated with complications (IRR, 0.86; P = .851). The inter-rater analysis showed a mean percent agreement of 79.4% (range, 53.8–94.9%) between the 2 raters and a mean reliability coefficient of 0.71 (range, 0.351–0.940; Table VI). In general, reliability was higher for the QIs rated based on objective measures or check boxes (prophylactic antibiotics, postoperative euglycemia, reconciliation of medication list, and surgical safety checklist).
ARTICLE IN PRESS Bergman et al 5
Surgery Volume j, Number j
Table IV. Adherence to quality indicators and association with postoperative complications
Quality indicator (n = number triggered) Prophylactic antibiotics (n = 273) Postoperative euglycemia (n = 47) Prophylactic venous thromboembolism therapy (n = 273) Central venous line (n = 32) Urinary catheter (n = 241) Postoperative ambulation (n = 273) Medication list (n = 273) Pressure ulcer risk assessment (n = 214) Oral intake documentation (n = 273) Surgical safety checklist (n = 253)
Passed/ failed
n
%
Patients with complications (%)
Passed Failed Passed Failed Passed Failed Passed Failed Passed Failed Passed Failed Passed Failed Passed Failed Passed Failed Passed Failed
261 12 18 28 193 80 26 6 175 66 168 105 219 54 117 97 26 247 235 18
95.6 4.4 38.3 61.7 70.7 29.3 81.3 18.8 72.6 27.4 61.5 38.5 80.2 19.8 54.7 45.3 9.5 90.5 92.9 7.1
30.7 16.7 38.9 58.6 30.1 30.0 69.2 66.7 30.9 34.9 21.4 43.8 30.6 27.8 33.3 28.9 11.5 32.0 31.5 11.1
Unadjusted P value
Adjusted P value*
.520
.993
.188
.993
.993
.993
.999y
.999
.554
.993
