Table 2.
Accuracy of selected variables related to the length of stay and complications during primary stay in 160 randomly selected patients undergoing elective colorectal surgery at 8 ERAS centers in Sweden in 2017
| Variables | True positive |
False positive |
False negative |
True negative |
Sensitivity (%) | Specificity (%) | PPV (%) |
NPV (%) |
Accuracy (%) | Cohen’s kappaa (95%CI) |
AUCb (95%CI) |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Length of stay (nights) | – | – | – | – | – | – | – | – | 93.75 | 0.84 (0.79, 0.88) | – |
| Complications at all | 45 | 4 | 10 | 101 | 81.82 | 96.19 | 91.84 | 90.99 | 91.25 | 0.80 (0.69, 0.84) | 0.89 (0.84, 0.94) |
| Date of death | – | – | – | – | – | – | – | – | 100.00 | – | – |
| Reoperations | 16 | 1 | 1 | 142 | 94.12 | 99.30 | 94.12 | 99.30 | 98.75 | 0.93 (0.82, 1.00) | 0.97 (0.91, 1.00) |
| Pneumonia | 0 | 0 | 1 | 159 | – | – | – | 99.38 | 99.38 | 0.00 (0.00, 0.00) | – |
| Urinary tract infection | 0 | 0 | 0 | 160 | – | – | – | 100.00 | 100.00 | – | – |
| Intraperitoneal or retroperitoneal abscess | 2 | 1 | 4 | 153 | 33.33 | 99.35 | 66.67 | 97.45 | 96.88 | 0.43 (−0.01, 0.85) | 0.66 (0.46, 0.87) |
| Sepsis | 2 | 0 | 8 | 150 | 20.00 | 100.00 | 100.00 | 94.94 | 95.00 | 0.32 (0.00, 0.66) | 0.60 (0.47, 0.73) |
| Anastomotic leak | 8 | 0 | 2 | 150 | 80.00 | 100.00 | 100.00 | 98.68 | 98.75 | 0.88 (0.66, 1.00) | 0.90 (0.77, 1.00) |
| Mechanical bowel obstruction | 0 | 1 | 1 | 158 | 0.00 | 99.37 | 0.00 | 99.37 | 98.75 | −0.01 (−0.02, 0.00) | 0.50 (0.00, 1.00) |
| Postoperative paralytic ileus | 10 | 9 | 8 | 133 | 55.56 | 93.66 | 52.63 | 94.33 | 89.38 | 0.48 (0.26, 0.68) | 0.75 (0.63, 0.87) |
EMR data was treated as the gold standard. True positive is the number of patients reported to be yes in both EIAS and EMR, false positive is the number of patients reported to be yes in EIAS and no in EMR, false negative is the number of patients reported to be no in EIAS and yes in EMR, and true negative is the number of patients reported to be no in both EIAS and EMR. Sensitivity is the true positive rate (the number of patients reported to be yes in both EIAS and EMR divided by the number of patients reported to be yes in EMR), specificity is the true negative rate (the number of patients reported to be no in both EIAS and EMR divided by the number of patients reported to be no in EMR), PPV is defined as the number of patients reported to be yes in both EIAS and EMR divided by the number of patients reported to be yes in EIAS, NPV is defined as the number of patients reported to be no in both EIAS and EMR divided by the number of patients reported to be no in EIAS, and accuracy is the proportion of patients with accurately recorded value in EIAS (the number of patients whose response from EIAS was confirmed in EMR divided by the total number of patients). AUC and Cohen’s kappa were calculated for variables with dichotomous responses present in both data sources
PPV Positive predictive value; NPV Negative predictive value; CI Confidence interval; AUC Area under the receiver operating characteristic curve; ERAS Enhanced recovery after surgery; EIAS International ERAS Interactive audit system; EMR Electronic medical records
aFor length of stay, the individual intraclass correlation coefficient from one-way random effects model was reported. Cohen’s kappa indicates values < 0 as indicating no agreement and 0–0.20 as slight, 0.21–0.40 as fair, 0.41–0.60 as moderate, 0.61–0.80 as substantial, and 0.81–1 as almost perfect agreement
bAUC indicates 0.5–0.7 as poor, 0.7–0.8 as acceptable, 0.8–0.9 as excellent, and > 0.9 as outstanding