Development of a risk score to stratify symptomatic adults referred for colonoscopy.

BACKGROUND AND AIM

With an increasing burden on overstretched colonoscopy services, a simple risk score for significant pathology in symptomatic patients may aid in the prioritization of patients.

METHODS

A derivative study of a risk score model for colonic neoplasia (colorectal carcinoma [CRC] and advanced adenoma) and CRC alone was conducted in symptomatic adults referred for an index colonoscopy. The accuracy of the final model was assessed by the area under the curve (AUC) of the receiver operating characteristic curve and the Hosmer-Lemeshow goodness-of-fit statistic.

RESULTS

A total of 1013 subjects (mean age 59.9 ± 13.7 years, 52.3% females) from a multi-ethnic Asian background (Chinese 56%, Malay 20.4%, Indian 21.5%) were recruited. Colonic neoplasia and CRC were identified in 175 (17.3%) and 114 (11.3%) cases, respectively. Risk scores were assigned to individual factors identified in a logistic regression model of both demographic (age, gender, ethnicity, education level, smoking history, Aspirin use) and clinical symptoms (change in bowel habit, bloody stool, weight loss, appetite loss, lethargy). The risk score for each patient was the sum of their individual risk factors. The AUC of the risk score for colonic neoplasia and CRC was 0.76 (Hosmer-Lemeshow goodness-of-fit statistic of P = 0.745) and 0.83 (Hosmer-Lemeshow goodness-of-fit statistic of P = 0.982), respectively.

CONCLUSION

A simple risk score for colonic neoplasia and CRC may be able to prioritize colonoscopy referrals in symptomatic subjects from a multi-ethnic background. A further study to validate this scoring system is required.