A Martin, G Purcell, J Roberts-Thomson, D Brockwell
A Martin, G Purcell, J Roberts-Thomson, D Brockwell. Colonoscopy Key Performance Indicators in a Rural General Australian Hospital. The Internet Journal of Gastroenterology. 2013 Volume 12 Number 1.
We evaluated colonoscopic neoplastic yield (adenomas and cancers) in a rural endoscopic unit alongside several key performance indicators known to influence yield. A second aim was to compare neoplastic yield according to colonoscopic indication.
Colonoscopy is without dispute the gold standard for clinical evaluation of the colon.1 However its effectiveness as a tool for the detection of significant neoplastic lesions, (adenomas and cancers) – here referred to as “SNL”, has been shown to be closely correlated with several key performance indicators. These form the basis of current screening guidelines and include: 1). Caecal Intubation rate2, 2). Adequacy of bowel preparation3 and 3). Withdrawal time to anus4-5.
A caecal intubation rate of ≥ 90% is the recommended standard by the U.S. Multi-Society Task Force on Colorectal Cancer among other advisory bodies6, while published rates vary widely from 55-98.8%2. Successful caecal intubation has been shown to be heavily dependent on the clinical experience of the operator7-8; although factors independent of the operator may limit outcomes including previous pelvic surgery, female gender and a long transverse colon9-10. Longer withdrawal times furthermore are associated with higher adenoma detection rates (ADR) after adjustment of other variables, with the U.S. Multi Society Task Force on Colorectal Cancer recommending a minimum of 6-10 minutes; that being the mean time for operators with low rates of missed adenomas4 11-12. The Task Force (2002) also recommended a benchmark ADR of 25% for males and 15% for females over 5013.
We evaluated colonoscopies undertaken at the endoscopic unit of a rural general hospital for: 1). SNL detection rate (including ADR). 2). Caecal intubation rate, 3). Adequacy of bowel preparation and 4). Withdrawal time to anus. Secondly, we evaluated SNL detection rates for different colonoscopic indications with relevance to the wider body of knowledge contributing to the sometimes difficult question of when to screen.
Patients and Methods
Data was retrieved retrospectively. Colonoscopic indication, adequacy of bowel preparation, endoscopic findings and success/failure of caecal intubation were entered from medical records. Withdrawal time to anus was retrieved from a log book kept in the endoscopy unit and histopathology from an electronic database.
All patients who had a colonoscopy at the Mersey General Hospital, Tasmania over six months (Nov 2007- April 2008) were included.
Prior to colonoscopy, all patients received a leaflet entailing directions for fasting and bowel preparation. Patients were instructed to restrict intake to clear fluids after breakfast from the day before colonoscopy with units of Picoprep® or Picolax® at 15:00, 17:30 and 19:00 followed by fasting from 6:30 on the day of colonoscopy. Bowel Preparation was assessed semi-subjectively as good (>75% bowel wall visualised), fair (35-75% visualised) and poor (<35% visualised)14
All patients received propofol for sedation plus a narcotic (fentanyl or alfentanil) and/or midazolam. There were nine endoscopists of which seven were General Surgeons, one registrar and one physician. Completion was defined as visualisation of one or all of the Triradiate Fold, Appendiceal Orifice and the Ileo-caecal Valve.
371 patients received colonoscopies (163 male, 208 female) for 453 indications. The vast majority were in age categories 41-60 and 61-80.
153 procedures (41%) were normal. 77 polyps were detected in 73 patients of which 52 were adenomas. 15 cancers were detected in 15 patients, all adenocarcinomas. Diverticular Disease was detected in 99 patients (26.7%)
Colonoscopy Key Performance Indicators
ADR fell short of benchmark standards at 11.86% of procedures overall (males 18.4%, females 6.7%). Inclusion of cancers (SNL detection rate) yielded 15.9% (males 20.86%, females 12.02%) with a statistically significantly higher proportion for males (p < 0.0001), whereas there were no significant differences in SNL detection rates between age categories of 21-40, 41-60, 61-80, >80 (p = 0.3145)
Overall caecal intubation rate was 85.6% (males 90.5%, females 81.4%) not including those not recorded (0.9%). Most common causes for failure were poor bowel preparation (n = 18), obstructing mass/stricture (n = 9) and excessive bowel looping or difficult angulation (n = 9).
Bowel preparation was judged as adequate in 79.24% (good = 56.87 %, fair = 22.37%), poor = 13.4%, not recorded = 7.28%.
Mean withdrawal time to anus was 5.76 minutes. ADR for withdrawal time categories of 0-5, 6-10 and > 11 minutes were compared defining ADR as number of lesions detected rather than positive procedures yielding adenomas (or SNL) consistent with previous studies4 11-12. There were no significant differences in male: female ratio (p = 0.3749) or the major colonoscopic indications listed in
SNL detection rates according to indication
Below we compare Number Needed to Screen (NNS) for detection of SNL against colonoscopic indication using
Colonoscopy, alongside Faecal Occult Blood testing has been shown to lower mortality. The Australian Institute of Health and Welfare (AIHW) statistics for 2006 listed Colorectal Cancer as the second most common cause of cancer at an annual incidence of 13.1% of all newly diagnosed cancers15. Similarly, 9.7% of all mortality attributable to cancer was due to colorectal cancer16. Detection of adenomas remains a good surrogate for malignancy, with a 5% transformation rate17. ADR in our study fell below benchmark standards, especially in women; although the wide range of ADR in the literature must be appreciated.
ADR is subject to numerous variables. Male sex as shown in our study is a recognised independent risk factor11. Notably our study had a higher proportion of female participants. There were no statistically significant differences across age categories; although the majority of studies point to age being an independent risk factor18. Adequacy of bowel preparation ranked reasonably well at 79.24% alongside a 77% adequacy rate found across 93,004 cases in the US19 without including 7.28% of cases for which quality of bowel preparation was not recorded. Importantly, the semi-subjective nature of assessing quality of bowel preparation must be recognised, especially given that no universally-recognised system for evaluating quality of bowel preparation exists19.
Our finding of lower ADR at withdrawal times 0-5 minutes is significant and consistent with a wide body of literature12. Importantly we have accounted for the confounding effects of sex ratio and relative proportions of screening indications on withdrawal time categories. Similarly, all patients with multiple adenomas detected belonged to longer withdrawal time categories. The significance of the latter point is uncertain, but it may be speculated that initial detection of adenomas may cause the endoscopist to slow down. However, an important confounder to higher ADR at longer withdrawal times is the time taken to snare polyps during withdrawal which intuitively would increase with number of polyps detected. Importantly we have not recorded for adenoma size. The malignant potential of adenomas < 10 mm remains controversial20.
Few studies have been performed comparing the diagnostic yield of presenting indications. Importantly there was a large amount of overlap with 453 indications across the 371 patients, especially for abdominal pain and PR bleeding. Our findings were somewhat comparable to a significantly larger Dutch trial18 (n = 4623) which found SNL detection rates of 15% for PR bleeding; 21% for abdominal pain and/or change in bowel habit; 8% for Anaemia; 10% for past history of Colorectal Cancer and 13% for past history of polyps. Significantly their SNL detection rate for family history of colorectal cancer was 10% compared to our finding of 20.69%. Faecal Occult Blood remains a useful screening tool for asymptomatic patients with published meta-analyses showing reduction in mortality from colorectal cancer by 14-16%21-22; hence the federal initiative (2006) of the largely successful National Bowel Screening Program with detection rates of 5.2% for cancers and 12.4% for adenomas23.