Table 3. A comparison of several clinical clues for colorectal cancer diagnosis and the associated median referral times in colorectal cancer patients with and without missed opportunities.
| Clues (κ)a | Patients with at least one missed opportunity (n=161) | Patients with no missed opportunities (n=220) | P valueb | |||
|---|---|---|---|---|---|---|
| Number of missed opportunities | Number of patients | Percentagec | Number of patients | Percentaged | ||
| Median days until first referral | IQRe | Median days until first referral | IQRe | |||
| Clues with high agreement between reviewers (κ > 0.60) | ||||||
|
| ||||||
| Screen positive for barium enema (κ = 0.86) | 6 | 4 | 2.5 | 8 | 3.6 | 0.57 |
| 2,514 | 1,470–3,676 | 0.5 | 0–1.5 | 0.007 | ||
|
| ||||||
| Suspected or confirmed iron deficiency anemia (κ = 0.82) | 357 | 83 | 51.6 | 58 | 26.4 | < 0.0001 |
| 393 | 108–1114 | 2.5 | 0–12 | < 0.0001 | ||
|
| ||||||
| Abnormal imaging suspicious for colon cancer (CT) (κ = 0.81) | 8 | 5 | 3.1 | 22 | 10.0 | 0.01 |
| 71 | 27–274 | 0.5 | 0–1 | 0.001 | ||
|
| ||||||
| BRBPR (in the absence of colonoscopy within previous 1-year period) (κ = 0.80) | 79 | 36 | 22.4 | 48 | 21.8 | 0.90 |
| 161 | 63–321 | 0 | 0–1 | < 0.0001 | ||
|
| ||||||
| Screen positive for flex-sig (κ = 0.78) | 2 | 1 | 0.62 | 23 | 10.5 | < 0.0001 |
| 1,246 | 1,246–1,246 | 0 | 0–0 | < 0.001 | ||
|
| ||||||
| Screen positive for fecal occult blood test (κ = 0.76) | 128 | 64 | 39.8 | 83 | 37.7 | 0.75 |
| 146 | 51–528 | 1 | 0–7 | < 0.0001 | ||
|
| ||||||
| Family history of colon cancer (κ = 0.75) | 12 | 4 | 2.5 | 7 | 3.2 | 0.77 |
| 441 | 210–686 | 0 | 0–0 | 0.003 | ||
|
| ||||||
| Intestinal obstruction (κ = 0.72) | 0 | 0 | 0 | 19 | 8.6 | < 0.0001 |
| n/a | n/a | 1 | 0–2 | —f | ||
|
| ||||||
| Suspected rectal or abdominal mass (κ = 0.69) | 8 | 6 | 3.7 | 7 | 3.2 | 0.78 |
| 14 | 2–44 | 0 | 0–1 | 0.01 | ||
|
| ||||||
| Worsening constipation (κ = 0.66) | 56 | 30 | 18.6 | 19 | 8.6 | 0.01 |
| 281 | 117–487 | 1 | 0–8 | < 0.0001 | ||
|
| ||||||
| Weight loss (κ = 0.64) | 38 | 23 | 14.3 | 29 | 13.2 | 0.76 |
| 144 | 42–395 | 1 | 0–1 | < 0.0001 | ||
|
| ||||||
| Clues with fair or moderate agreement between reviewers (κ between 0.20 and 0.60) | ||||||
|
| ||||||
| History of previous polyps (κ = 0.50) | 61 | 10 | 6.2 | 11 | 5.0 | 0.65 |
| 663 | 277–977 | 0 | 0–8 | < 0.001 | ||
|
| ||||||
| Abdominal pain (κ = 0.49) | 61 | 29 | 18.0 | 34 | 15.5 | 0.58 |
| 140 | 58–333 | 1 | 0–3 | < 0.0001 | ||
|
| ||||||
| Melena (κ = 0.48) | 1 | 1 | 0.62 | 12 | 5.5 | 0.01 |
| 482 | 482–482 | 0 | 0–1 | 0.09 | ||
|
| ||||||
| Other suspicious lower GI symptoms (κ = 0.45) | 38 | 15 | 9.3 | 14 | 6.4 | 0.33 |
| 414 | 64–658 | 1 | 0–31 | 0.002 | ||
|
| ||||||
| Change in stool caliber (κ = 0.39) | 0 | 0 | 0 | 4 | 1.8 | 0.14 |
| n/a | n/a | 0 | 0–1 | —f | ||
a
Kappa (κ) is calculated using all 381 patients.
b
Fisher's exact test was used to test for differences in proportions of patients with a given clue. The nonparametric Wilcoxon's continuity-corrected normal approximation two-sample test was used to test for differences in days until first referral.
c
The percentage of the 161 patients with a missed opportunity who had the given clue.
d
The percentage of 220 patients without a missed opportunity who had the given clue.
e
IQR: interquartile range.
f
Median times could not be compared because there were no patients with missed opportunities.