2017-18 HSC Section 3 Green Book

Original article

come. A P value of < 0.05 was considered statistically signifi- cant.

Results A total of 73 patients presented to us during the study period (2009 – 2014) with caustic ingestion. Patients who reported la- ter than 24 hours after ingestion (n = 2) or had clinical or radio- logical features of perforation (n = 3) were excluded from the study group. After the initial clinical and radiological evaluation had been completed, 68 patients underwent day 1 EGD. Of these patients, four refused to undergo a relook EGD on day 5 and two were taken for surgery for impending perforation. Therefore, 62 patients underwent day 1 and day 5 EGDs, and subsequently formed the study population. The group consis- ted of 35 men, 27 women, with a mean age of 33 ± 15 years, of whom 58 had ingested acid and four had ingested alkali. Endoscopy findings ( ▶ Figs. 1 – 3 ) ▶ Table 1 shows the numbers of patients with gradings of se- vere esophageal and gastric injury on the day 1 and day 5 EGDs. While 29 patients (46.8 %) were deemed to have a severe esophageal injury on day 1, only 20 (32.3 %) had severe esoph- ageal injury on day 5 EGD. Similarly, day 1 EGD was interpreted as showing severe gastric injury in 35 patients (56.5 %), but only 18 patients (29.0 %) had severe gastric injury on day 5. Compar- ison of the two endoscopic evaluations showed that day 1 EGD overestimated esophageal injury in nine patients (14.5 %) and gastric injury in 17 (27.4 %). No patients had an upgrade of their mucosal injury on day 5. There were no complications of EGD, either on day 1 or on day 5. On follow-up endoscopic and radiological evaluation, a total of 16 patients had developed isolated antropyloric stenosis, nine had developed isolated esophageal strictures, and four had developed both. Therefore, there were 13 patients with esophageal strictures and 20 with antropyloric stenosis ( ▶ Ta- ble 2 ). Compared with day 1, endoscopic grading of severe in- jury on day 5 had higher specificity (83 % vs. 65 %), higher PPV (60 % vs. 41 %), and higher positive LR (5.65 vs. 2.66) for pre- dicting the development of esophageal stricture ( ▶ Table 3 ). However, the sensitivity, NPV, and negative LR were similar for day 1 and day 5 endoscopy. Similarly, endoscopic grading of se- vere injury on day 5 compared with day 1 had higher specificity (95 % vs. 61 %), higher PPV (88 % vs. 54 %), and higher positive LR (16 vs. 2.5) for predicting the development of antropyloric stenosis ( ▶ Table 4 ). However, the sensitivity, NPV, and nega- tive LR were similar.

▶ Fig. 3 Gastric findings: a on day 1 endoscopy showing edematous gastric folds with mucosal hemorrhages (arrow) and superficial ul- ceration with hemorrhagic fluid in between the folds (arrowheads), graded as 2a; b on day 5 endoscopy showing edematous folds with patches of hyperemia, interpreted as grade 1.

Statistical analysis

Data were analyzed using SPSS software (U23; IBM Corp., Ar- monk, New York, USA). Categorical data were described as counts/proportions; continuous data were described as mean ± standard deviation (SD), or median and interquartile range (IQR) if the data had a non-parametric distribution. Categorical data were analyzed using the chi-squared test. Numerical data were compared using the Student ’ s t test or Mann – Whitney U test, as appropriate. The primary outcome variable was analyzed using the chi- squared test. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and positive and nega- tive likelihood ratios (LRs) were calculated. Receiver operating characteristic (ROC) curves were plotted and the area under the curve (AUC) was compared with the prediction of key out-

▶ Table 1 Presence of severe endoscopic injury following caustic ingestion assessed on days 1 and 5 (n = 62).

Day 1

Day 5

Esophagus

29 (46.8 %)

20 (32.3 %)

Stomach

35 (56.5 %)

18 (29 %)

Kochhar Rakesh et al. Relook endoscopy predicts … Endoscopy

215