Eighty-eight patients in the observation group received the allocated intervention in the DIAMOND trial. For the present study, four patients were excluded due to protocol violations, as we did not have their final AAS recorded. Also, one patient with radiologically diagnosed appendicitis treated nonoperatively was excluded as we did not have the histopathological confirmation of acute appendicitis. In the end, 83 patients were available for analysis in this study.
The median age of the patients was 29 (i.q.r. 25–38), and 43 (52%) were female. Table 1 describes the demographics and clinical characteristics of the patients.
Table 1
Demographics and clinical characteristics
Age, years* | 29 (25–38) |
Female | 43 (52) |
Duration of symptoms, hours† | 12.7 (5.6) |
Baseline CRP, mg/L* | 6.0 (3.0–14.0) |
Baseline WBC count, E9/L* | 12.3 (10.0–15.0) |
Baseline AAS* | 13 (12–14) |
Values in parentheses are percentages unless indicated otherwise; values are *median (i.q.r.) and †mean (s.d.). CRP = C-reactive protein, WBC = white blood cell, AAS = Adult Appendicitis Score.
Twenty-five (30%) patients had a final AAS of 16 or higher, and the decision to operate was made without imaging. All these patients had appendicitis; two of them had complicated appendicitis. Thirty-one (37%) patients underwent abdominal imaging after the re-scoring, resulting in 24 surgeries for acute appendicitis, one of which was a negative appendectomy. Twenty-seven patients (33%) were discharged without imaging. None of these patients returned due to appendicitis within the first 30 days. In the end, forty-eight (58%) patients were diagnosed with histologically confirmed appendicitis.
The receiver operating characteristic curves (Fig. 1) and the AUROC values (Table 2) show that both the ΔAAS and final AAS perform better in diagnosing acute appendicitis (ROC area values 0.932 (95%CI 0.868–0.996) and 0.936 (95%CI 0.886–0.987)) compared to baseline AAS and ΔCRP (ROC area values 0.629 (95%CI 0.498–0.760) and 0.796 (95%CI 0.696–0.897)).
Table 2
The values of AUROC for the 𝛥AAS and the reference standards
| AUROC | SE | p |
Baseline AAS | 0.629 (0.498–0.760) | 0.067 | 0.046 |
Final AAS | 0.936 (0.886–0.987) | 0.026 | < 0.001 |
𝛥AAS | 0.932 (0.868–0.996) | 0.033 | < 0.001 |
𝛥CRP | 0.796 (0.696–0.897) | 0.051 | < 0.001 |
95% confidence interval in parenthesis. AUROC = area under the receiver operating characteristics, SE = standard error, AAS = Adult Appendicitis Score, 𝛥AAS = change in Adult Appendicitis Score, 𝛥CRP = change in C-reactive protein. P-values were determined with Mann-Whitney U-test.
To identify the low, intermediate, and high probability of appendicitis after observation, two cut-off values for ΔAAS were determined using the ROC curves: ΔAAS ≥ -1 and ΔAAS ≥ 1. For comparison, the reference standards' cut-off values were also determined: baseline AAS ≥ 14, final AAS ≥ 16, and ΔCRP ≥ 9, Fig. 1. The sensitivity, specificity, likelihood ratios, and diagnostic odds ratios of acute appendicitis were calculated for specific ranges of baseline ASS, final AAS, ΔAAS, and ΔCRP, as shown in Table 3. These cut-off values are indicated in Fig. 1.
Table 3
Defining the cut-off values for ΔAAS and the reference standards
| ΔAAS ≥ -1 | ΔAAS ≥ 1 | Baseline AAS ≥ 14 | Final AAS ≥ 16 | ΔCRP ≥ 9 |
Sensitivity | 97.9% (89.1–99.6%) | 70.8% (56.8–81.8%) | 52.1% (38.3–65.5%) | 52.1% (38.3–65.5%) | 75.0% (61.2–85.1%) |
Specificity | 80% (64.1–90.0%) | 94.3% (81.4–98.4%) | 68.6% (52.0–81.4%) | 100.0% (90.1–100.0%) | 80.0% (64.1–90.0%) |
+LR | 4.90 (2.52–9.51) | 12.40 (3.19–48.20) | 1.66 (0.95–2.90) | infinity | 3.75 (1.90–7.42) |
-LR | 0.03 (0.004–0.18) | 0.31 (0.20–0.48) | 0.70 (0.51–0.96) | 0.48 (0.36–0.64) | 0.31 (0.19–0.52) |
DOR | 188.00 (21.97–1608.95) | 40.07 (8.45–190.14) | 2.37 (0.95–5.90) | infinity | 12.00 (4.18–34.46) |
95% confidence interval in parenthesis. 𝛥AAS = change in Adult Appendicitis Score, AAS = Adult Appendicitis Score, 𝛥CRP = change in C-reactive protein, +LR = positive likelihood ratio, -LR = negative likelihood ratio, DOR = diagnostic Odd's ratio.
Based on the chosen cut-off values of ΔAAS, patients can now be classified into three groups corresponding to the probability of appendicitis: low (ΔAAS ≤ -2), intermediate (ΔAAS − 1 − 0), and high (ΔAAS ≥ 1 ) probability. The proportions of acute appendicitis were then calculated in these three groups, Table 4. There were 29 (35%) patients in the low probability group, and only 1 (3%) of them had acute appendicitis. Eighteen (22%) patients fell into the intermediate probability group, and 13 (72%) of them had acute appendicitis. There were 36 (43%) patients in the high probability group, and 34 (94%) of them had acute appendicitis. The negative predictive value of the low (ΔAAS ≤ -2) probability of acute appendicitis is 97 per cent, and the positive predictive value of the high (ΔAAS ≥ 1) probability of acute appendicitis is 94 per cent. The flow of patients into these groups is presented in Fig. 2.
Table 4
Three groups were formed: low, intermediate, and high probability of acute appendicitis
| Low: ΔAAS ≤ -2 | Intermediate: ΔAAS − 1 − 0 | High: ΔAAS ≥ 1 |
N | 29 (35) | 18 (22) | 36 (43) |
Acute appendicitis diagnosis | 1 (3) | 13 (72) | 34 (94) |
Values in parentheses are percentages. 𝛥AAS = change in Adult Appendicitis Score.
Thus, utilising these groups based on ΔAAS, only the 18 (22%) patients in the intermediate probability group would have benefited from imaging studies. Compared with the 31 (37%) patients actually imaged during the trial according to the trial protocol, this is a significantly lower number of patients, p < 0.001.