Screening Colonoscopies by Primary Care Physicians: A Meta-Analysis

The American College of Gastroenterology recently published a critique of "Screening Colonoscopies by Primary Care Physicians: A Meta- Analysis' Ann. Fam. Med, Jan 2009; 7: 56 - 62. In that critique the following statements were made:

“The College believes that the findings of the Wilkins study are potentially misleading.” “According to Dr. Rex, all three population-based studies that have determined miss rates for colorectal cancer during colonoscopy by specialty have found that primary care physicians are more likely than gastroenterologists to miss colorectal cancer during colonoscopy (Rex et al Gastroenterology; 1997: 112:17-23; Bressler et al Gastroenterology 2007; 132:96-102; Singh et al Gastroenterology 2007; 132:A-149).”

The American College of Gastroenterology’s response relies heavily upon the three studies cited above and a closer look at them is warranted.

The first article by Rex was a retrospective comparison of barium enema (BE) and colonosocpy. Sensitivity of BE was 82%, colonosocpy was 95% overall, (by GI 97%, by non GI – 87%) Reach the cecum rate was not reported. Non GI included both surgeons and primary care. Only 5% of cases were “screening”. The Singh article was published only in abstract. The author reports a relative risk that is “twice as high” for developing colon cancer if the colonoscopy was performed in a rural setting. But the absolute risk is only 0.4%. So one patient out of 250 could be affected by this difference. When location was excluded, specialty of the endoscopist did not affect outcome.

Of the 3 articles cited, the most recent by Bressler is the only one to compare primary care physicians with others. This study indicates colonoscopy in anyone’s hands can miss 3-6% of cancers. Surgeons did most of the cases - 49%, GI did 31%, primary care plus “other” did 20%. “Other” had the best detection rate. “Other” should be non-GI. If we combine primary care and “other” the rate is 95.5%. Reach the cecum rates were not reported. Office location had the worst outcome and could be due to several factors. The percentage of primary care in the office setting was not reported. Using the results of this latter study, a patient with a colon cancer has a 3% chance of having that cancer missed on a colonoscopy performed by a gastroenterologist. The chance of missing the cancer by a non gastroenterologist may be as high as 6%. Based on Rex’s study, the chance of missing the cancer with a barium enema is 18%. The chance of missing a cancer with sigmoidoscopy is 22-32% and with fecal occult blood testing the chance of missing a cancer is greater than 50%. No screening is currently the most common cause of screening failure since only 1/3 of patients have any screening. If no screening is performed, the chance of screening failure is 100%.

Based on good scientific data, preventive medicine experts recommend that physicians should screen patients for colon cancer by one of several accepted methods including colonoscopy, sigmoidoscopy, or fecal occult blood testing. If the conclusions from three studies cited above are accurate, colonoscopy by anyone is far superior to any other method. More impact on public health would occur if screening by any method was increased. The difference between primary care and subspecialty care in missing a colon cancer on colonoscopy is not fully known, but it is insignificant when one considers the miss rates of other methods.

If the American College of Gastroenterology wants to improve population health, their emphasis should be toward improving effectiveness of screening methods by all physicians regardless of specialty. (These comments do not necessarly reflect the opinion of the AAPCE)

Competing interests:   None declared

Quality measures in all aspects of medicine including colonoscopy have recently been the focus of multiple studies. For colonoscopy to be effective in preventing colon cancer the entire colon should be visualized and all polyps removed. Colonoscopy completion rates, withdrawal times and adenoma detection rates are considered as important markers of quality in colonoscopy.

These issues must be considered when critically assessing studies such as that recently published by Wilkins and colleagues (Annals of Family Medicine 2009:7:56-62) which claims to demonstrate outcomes equivalent to those of gastroenterologists when primary care physicians perform screening colonoscopy. The finding contrasts with those of previous studies that found miss rates for colon cancer are higher when colonoscopy is performed by primary care physicians compared to gastroenterologists. 1, 2, 3 Since seventy-three percent (13,363 out of 18,292) of the colonoscopies included in the analysis were derived from a single study that has not been published in a peer-reviewed journal, one’s interpretation of these findings must be cautious. In addition, colonoscopies from this same study should also have been excluded because the center involved a gastroenterologist and colorectal surgeon, one of whom was on site full-time to rescue procedures when primary care physicians failed to reach the cecum.

The overall reported rate for cecal intubation of 89 percent in this meta-analysis is 6 percent below the minimum recommended threshold for screening colonoscopy by the U.S. Multi-Society Task Force on Colorectal Cancer. In contrast, all published studies of screening colonoscopy by gastroenterologists in the United States exceed 95 percent. When the data from the non-peer reviewed single-center study with specialist rescue are excluded, only 4929 procedures from the Wilkins study are available for analysis. In these remaining cases, the cecal intubation rate was 83.5 percent, more than 10 percent lower than the recommended 95 percent threshold. Given the low rate of perforation expected with screening colonoscopy, these 4929 procedures are not enough to determine whether there is a true difference in perforation rate compared to gastroenterologists.

A recent article published in Annals of Internal Medicine reported that nearly 60 percent of right-sided lesions were missed by doctors performing colonoscopy who participated in a Canadian study to determine the sensitivity of colonoscopy in detecting polyps and colon cancers4. However, nearly 70 percent of the doctors who participated in this study were general/family practitioners or internists and not trained gastroenterologists.

Finally, other data indicate that primary care physicians are less likely than gastroenterologists to recommend colonoscopy at appropriate intervals according to widely accepted guidelines5

Continued efforts to measure and improve the quality of screening colonoscopy will serve to improve outcomes for our patients and reduce the burden of colon cancer. The bottom line: colonoscopy is a highly valuable screening and prevention tool when performed to the highest standards.

1. Rex DK, Rahmani E, Hasseman JH, et al. Relative sensitivity of colonoscopy and barium enema for detection of colorectal cancer in clinical practice Gastroenterology. 1997;112:17-23.
2. Bressler B et al. Rates of new or missed colorectal cancers after colonoscopy and their risk factors: A population-based analysis. Gastroenterology. 2007; 132:96-102.
3. Singh H, Turner D, Xue L, Targownik LE, Bernstein CN. Colorectal cancers after a negative colonoscopy. Gastroenterology. 2007;132:A-149.
4. Baxter NN, Goldwasser MA, Paszat LF, Saskin R, Urbach DR, Rabeneck L. Association of colonoscopy and death form colorectal cancer: A population-based case-control study. Ann Intern Med. 2009;150:1-8.
5. Boolchand V, Olds G, Singh J, Singh P, Chak A, Cooper GS. Colorectal screening after polypectomy: a national survey study of primary care physicians. Ann Intern Med. 2006;145:654-9.

Correspondence:
Samir A. Shah, MD, FACG, Chair, Board of Governors, American College of Gastroenterology; Clinical Associate Professor of Medicine, Alpert Medical School of Brown University; Gastroenterology Associates, Inc., 44 West River Street, Providence RI, 02904, phone: (401) 274-4800, fax: (401) 454-0410, email: samir@brown.edu

Competing interests:   None declared

Dr. Rex’s comments on our study are particularly welcome due to his internationally recognized expertise in colorectal cancer (CRC) screening and quality indicators in colonoscopy. All colonoscopists should strive for nationally recognized quality indicators as proposed by Dr. Rex et al [1], and we look forward to working towards this goal to ensure access to safe, high-quality CRC screening for all Americans in order to decrease the deaths and disability due to this insidious disease. Several of the comments merit further examination to clarify our study methodology and to clarify the validity of the studies that Dr. Rex cites.

We disagree that our meta-analysis does not demonstrate adequate cecal intubation rates. First, while many of the colonoscopies were done for screening, some procedures were done for other indications such as abdominal pain, weight loss, gastrointestinal bleeding, and abnormal x- ray. Therefore, a cecal intubation rate of 95% would not be expected as is recommended when the indication is for CRC screening [1]. Furthermore, Dr. Rex recommends that colonoscopists exclude cases in which procedures are aborted secondary to inadequate bowel preparation or severe colitis [1]. In our extraction of the data, authors did not report reasons for incomplete exams. We suspect that many of the procedures that did not reach the cecum were secondary to reasons described above and these cases would have been excluded by other colonoscopists in calculations of cecal intubation rate. When all indications are included and cases excluded secondary to poor bowel preparation or other reasons, the cecal intubation goal is 90% or greater [1]. We reported a cecal intubation rate of 89.2% (95% confidence interval 80.1% - 94.4%) [2].

Significant variability exists in regards to cecal intubation rates in clinical practice. In 2006, cecal intubation data from 5,477 (average age was 57.9 years and population was 53.6% female) colonoscopies were analyzed [3]. Colonoscopies were completed by 10 fully trained gastroenterologists from the University of Maryland. In this study, the mean cecal intubation rate was 90.3% with significant variation among the individual colonoscopists in regards to cecal intubation rates (63% to 97%) [3].

Experts disagree on the use of unpublished data in meta-analysis. However, the strength of using unpublished data is that negative trials, which are less likely to be published, are included. Our inclusion of unpublished studies was designed to mitigate this publication bias and would be more likely to under-estimate colonoscopy outcomes, e.g. cecal intubation rates and adenoma detection rates. Additionally, we performed a comprehensive search strategy of MEDLINE and EMBASE, we reviewed bibliographies, and we contacted expert primary care endoscopists in the US via a listserv to find all published and unpublished studies of primary care physicians performing colonoscopies. Our meta-analysis included 10 published and 2 unpublished studies. All studies met our inclusion criteria. While in our paper, we were constrained by word count and could not describe each study in detail. In our response, we would like to expand on the details of Dr. Lloyd’s endoscopy center.

Dr. Stephen Lloyd’s Primary Care Endoscopy center is located outside of Columbia, S.C. and has been in operation since 1999. He has federal ($2m) and state ($1m) grants to support CRC screening. His center keeps meticulous records of colonoscopy quality indicators (cecal intubation rate, adenoma detection rate, complications, and insertion and withdrawal times) and every chart is peer-reviewed. Dr. Lloyd’s center hired a colorectal surgeon for 1 year (June 2007 through July 2008) during which approximately 3,000 screening colonoscopies were completed [4]. The colorectal surgeon was called to provide technical assistance in only 5% of these procedures [5, 6]. In a sub-group analysis, quality indicators (cecal intubation rate, complication rate, and adenoma detection rate) were not affected by the number of prior colonoscopies or the volume of monthly colonoscopy procedures [5].

Although Dr. Rex is correct in pointing out that a large percentage of cases come from one center, it is important to emphasize that random effects modeling was used to account for larger sample sizes in some studies and to adjust for the variability seen across sites. Rather than reporting simple averages, random effects modeling yields point estimates and 95% confidence intervals for cecal intubation rates, adenoma detection rates, and adenocarcinoma detection rates that are less influenced by sample size and the observed variability. Nevertheless, we agree that excluding the two major outliers [7, 8] might be warranted, which would yield an adjusted reach-the-cecum rate of 90.5% (rather than the raw average of 83.5% suggested by Dr. Rex), which is still clearly within the recommended ranges by U.S. Multi-Society Task Force on Colorectal Cancer [1].

The “population-based” studies referenced by Dr. Rex all have issues with validity and relevance [9-11]. For example, Dr. Rex’s own study identified cases between 1988-1993, when it is known that equipment and sedation techniques in office-based practices might adversely affect cecal intubation rates and adenoma detection rates compared to colonoscopies completed hospital-based endoscopy units [10]. Unfortunately, while this study implies that CRC miss rates are higher in “non-gastroenterologists,” the study does not statistically adjust for “non-gastroenterologist” confounders such as training, specialty, or location where the procedures were performed [10]. The Canadian study to which Dr. Rex refers has similar limitations where a different level of training for primary care physicians may exist compared with US-trained primary care physicians [9]. This study did not provide sufficient information on the location of procedures (e.g. office-based or hospital-based) by providers, to determine if this is a location bias, instead of a training/specialty determinant [9]. Finally, the third reference is an abstract presented at a Gastroenterology conference, so complete information on the methods and analysis is unavailable [11]. However, this abstract concludes that colonoscopy volume had no effect on the detection of miss rates of CRC [11]. This abstract reported that missed cancers occurred more likely in rural settings versus urban settings. Finally, no morbidity or mortality data on the patients was provided. There are clearly issues that influence miss rates of colonoscopy that should be addressed, as well as the importance of rigorous training and proper logistical support, but the primary specialty of the physician does not seem to be one of these.

When debating these issues across specialty lines, it is easy to lose focus of the problem. The problem is that nearly 50,000 Americans died in 2008 from CRC [12] and many of these deaths could have been prevented through a comprehensive, national CRC screening program. However, there are insufficient endoscopic resources in the U.S. to enact a national colonoscopy-based CRC screening program [13, 14]. With a projected shortage of gastroenterologists over the next decade, the numbers of unscreened Americans will only increase. Our study provides evidence that primary care physicians can safely and effectively perform colonoscopy, thus contributing to the endoscopic workforce, increasing access to screening colonoscopy, and saving the lives of many Americans from CRC. In the future American health care environment, where shortages of physicians (especially gastroenterologists) will be a challenge, increasing the capability of CRC screening in the primary care specialties will allow gastroenterologists to focus their considerable training and expertise on more complex cases that require their cognitive and technical skills.

[1] Rex DK, Petrini JL, Baron TH, Chak A, Cohen J, Deal SE, et al. Quality indicators for colonoscopy. Gastrointest Endosc. 2006 Apr;63(4 Suppl):S16-28.
[2] Wilkins T, LeClair B, Smolkin M, Davies K, Thomas A, Taylor ML, et al. Screening colonoscopies by primary care physicians: a meta- analysis. Annals of family medicine. 2009 Jan-Feb;7(1):56-62.
[3] Aslinia F, Uradomo L, Steele A, Greenwald BD, Raufman JP. Quality assessment of colonoscopic cecal intubation: an analysis of 6 years of continuous practice at a university hospital. Am J Gastroenterol. 2006 Apr;101(4):721-31.
[4] Lloyd SC. Description of Primary Care Endoscopy center. Personal Communication. January 26, 2009.
[5] Sweeney WB. Description of "expert" primary care colonoscopy model. Personal communication. January 27, 2009.
[6] Sweeney WB, Lloyd SC. Primary Care Colonoscopy: An Effective Model for the Prevention of Colorectal Cancer. Gastrointestinal Endoscopy. 2007;65(5):AB312-AB.
[7] Lloyd SC. Seeking Cecums II, A Success Story Addressing the Manpower Needs Towards Achieving Colon Cancer Screening Goals of Healthy People 2010. http://2006.confex.com/uicc/uicc/techprogram/P6194.HTM. Accessed 2-15-2007. UICC World Cancer Congress 2006 Bridging the Gap: Transforming Knowledge into Action; 2006 7-11-2006; Washington D.C.; 2006.
[8] Rodney WM, Dabov G, Cronin C. Evolving colonoscopy skills in a rural family practice: the first 293 cases. Fam Pract Res J. 1993 Mar;13(1):43-52.
[9] Bressler B, Paszat LF, Chen Z, Rothwell DM, Vinden C, Rabeneck L. Rates of new or missed colorectal cancers after colonoscopy and their risk factors: a population-based analysis. Gastroenterology. 2007 Jan;132(1):96 -102.
[10] Rex DK, Rahmani EY, Haseman JH, Lemmel GT, Kaster S, Buckley JS. Relative sensitivity of colonoscopy and barium enema for detection of colorectal cancer in clinical practice. Gastroenterology. 1997 Jan;112(1):17-23.
[11] Singh H, Turner D, Xue L, Targownik LE, Bernstein CN. Colorectal cancers after a negative colonoscopy. Gastroenterology. 2007;132:A149.
[12] Jemal A, Siegel R, Ward E, Hao Y, Xu J, Murray T, et al. Cancer statistics, 2008. CA: a cancer journal for clinicians. 2008 Mar- Apr;58(2):71-96.
[13] Seeff LC, Manninen DL, Dong FB, Chattopadhyay SK, Nadel MR, Tangka FK, et al. Is there endoscopic capacity to provide colorectal cancer screening to the unscreened population in the United States? Gastroenterology. 2004 Dec;127(6):1661-9.
[14] Seeff LC, Richards TB, Shapiro JA, Nadel MR, Manninen DL, Given LS, et al. How many endoscopies are performed for colorectal cancer screening? Results from CDC's survey of endoscopic capacity. Gastroenterology. 2004 Dec;127(6):1670-7.

Competing interests:   None declared

I believe your meta-analysis of primary care endoscopy illustrates that adequately trained primary care physicians can indeed help with the ever increasing demand for endoscopy. I have also just completed a quality assurance study for my first 1949 endoscopies performed in rural Alberta, Canada (which will be published in Canadian Family Physician Feb 2009) which yielded similar results.

I do agree with Dr. Rex's comment that it would be great to see the largest trial (Lloyd 2006) formally published, but disagree with his comment that the cecal intubation rate is inadequate. Many of the studies reviewed included patients that were being investigated for symptoms and therefore were not screening colonoscopies, and therefore the appropriate target reach the cecum rate would be 90%^1,2. Recent literature literature does however suggest that reach the cecum rates targets may in fact be difficult for many endoscopists to achieve.^3,4,5. Continuing to train family physicians in gastrointestinal medicine may be one avenue of improving access to timely gastrointestinal care and endoscopic evaluations for many rural patients.

Michael Kolber, Peace River, Alberta, Canada

1. Rex DK, Bond JH, Winawer S, Levin TR, Burt RW, Johnson DA, et al. Quality in the technical performance of colonoscopy and the continuous quality improvement process for colonoscopy: Recommendations of the U.S. Multi-Society Task Force on Colorectal Cancer. Am J Gastroenterol 2002; 97(6): 1296-308.
2. Rex DK, Petrini JL, Baron TH, Chak A, Cohen J, Deal SE, et al. Quality indicators in colonoscopy. Gastrointes Endosc 2006; 63(4 Suppl): S16-S28.
3. Aslinia F, Uradomo L, Steele A, Greenwood BD, Raufman JP. Quality assessment of colonoscopic cecal intubation: An analysis of 6 years of continuous practice at a university hospital. Am J Gastroenterol 2006; 101(4): 721-31.
4. Cotton PB, Connor P, McGee D, Jowell P, Nickl N, Schultz S, et al. Colonoscopy: Practice variation among 69 hospital-based endoscopists. Gastrointest Endosc 2003; 57(3): 352-357.
5. Armstrong D, Hollingworth R, Gardiner T, Klassen M, Smith W, Hunt RH, et al. Practice Audit in Gastroenterology (PAGE) program: A novel approach to continuing professional development. Can J Gastroenterol 2006; 20(6): 405- 10.

Competing interests:   None declared

In response to an invitation from Annals of Family Medicine, I have the following comments regarding the study by Wilkins et al on colonoscopy by primary care physicians (1). Colonoscopy quality is emerging as an important issue in colorectal cancer prevention. Two documents generated largely by gastroenterologists address quality indicators that should be used by all colonoscopists regardless of specialty. Review of those documents will establish that they harbor no bias against primary care physicians, but rather are designed to support quality by all colonoscopists (2,3).

Population based studies (4-6), which are of greater value than the meta-analysis of single center studies performed by Wilkins et al, indicate that on average primary care physicians are more likely than gastroenterologists to miss colorectal cancer during colonoscopy.

The study by Wilkins et al does not demonstrate adequate cecal intubation rates by primary care physicians. Of the 18, 292 cases, 13,363 (73%) come from one study published as an abstract only and not in a peer reviewed journal. This study claims a 98% cecal intubation rate but does not state that the center employs a gastroenterologist and a colorectal surgeon who are available to rescue failed procedures by primary care physicians. Since that practice model is not widely available, the study should have been excluded by Wilkins et al. In the remaining cases summarized by Wilkins et al the combined cecal intubation rate was 83.5%, well below the 95% standard recommended for screening colonoscopy and achieved consistently by gastroenterologists (2,3).

Thus, available data indicate that on average, colonoscopy by primary care physicians is of lower quality compared to gastroenterologists. Substantive steps to improve the quality of colonoscopy by primary care physicians are needed. All physicians performing colonoscopy should measure and strive to reach quality targets if we are to maximize the impact of colonoscopy on colorectal cancer prevention (2,3).

References:
1. Wilkins T, LeClair B, Smolkin M, Davies K, Thomas A, Taylor ML, Strayer S. Screening colonoscopies by primary care physicians: a meta- analysis. Ann Fam Med 2009;7:56-62.
2. Rex DK, Bond JH, Winawer S, Levin TR, Burt RW, Johnson DA, Kirk LM, Litlin S, Lieberman DA, Waye JD, Church J, Marshall JB, Riddell RH. Quality in the technical performance of colonoscopy and the continuous quality improvement process for colonoscopy: recommendations of the U.S. Multi-Society Task Force on Colorectal Cancer. Am J Gastroenterol 2002;97:1296-308.
3. Rex DK, Petrini JL, Baron TH, Chak A, Cohen J, Deal SE, Hoffman B, Jacobson BC, Mergener K, Petersen BT, Safdi MA, Faigel DO, Pike IM. Quality indicators for colonoscopy. Gastrointest Endosc 2006;63:S16-28.
4. Rex DK, Rahmani EY, Haseman JH, Lemmel GT, Kaster S, Buckley JS. Relative sensitivity of colonoscopy and barium enema for detection of colorectal cancer in clinical practice. Gastroenterology 1997;112:17-23.
5. Bressler B, Paszat LF, Chen Z, Rothwell DM, Vinden C, Rabeneck L. Rates of new or missed colorectal cancers after colonoscopy and their risk factors: a population-based analysis. Gastroenterology 2007;132:96-102.
6. Singh H, Turner D, Xue L, Targownik LE, Bernstein CN. Colorectal cancers after a negative colonoscopy. Gastroenterology 2007;132:A149.

Competing interests:   None declared

Primary care endoscopists need to develop and utilize quality improvement systems to monitor important and clinically relevant outcomes like reach-the-cecum rates, adenoma and adenocarcinoma detection rates, perforations, major bleeding episodes, and other complications. Only when this kind of data is collected prospectively on a national level by many primary care colonoscopists can we address some of these issues in regards to developing health care policy and national screening guidelines. This type of quality improvement project is a major research initiative of the American Association of Primary Care Endoscopy (www.aapce.org).

Competing interests:   None declared

We appreciate Dr. Foxx-Orenstein's comment. She gives sound advice to patients considering a colonoscopy. Her advice is to ensure that your endoscopist performs many of these tests and has a record of performing a complete examination. We echo her advice to patients.

A recent New York Times article, "Gastroenterologist Shortage Is Forecast" (January 9, 2009) reports that there are 10,390 practicing gastroenterologists in the US and predicts that there will be a shortage of 1,050 to 1,550 gastroenterologists in the next decade.

As America's aging population expands over the next decade, we are likely to need more colonoscopists to meet the increasing demand. Our paper demonstrates that a properly trained primary care physician can provide a safe, complete, and high quality colonoscopy.

Competing interests:   None declared

The training and experience of the endoscopist are critical to the quality of a colonoscopy. Gastroenterologists receive special training in colonoscopy and perform more colonoscopies by far than any other specialty. Their training emphasizes attention to detail and incorporates comprehensive knowledge of the entire GI tract to provide the highest quality endoscopy and consultative services. The American College of Gastroenterology advises patients to seek a well-trained endoscopist who performs many of these tests and has a record of performing a complete examination of the colon.

Many GI physicians participate in rating and improvement programs such as the one being piloted by the American College of Gastroenterology and the American Society for Gastrointestinal Endoscopy. Such a quality benchmarking program permits gastroenterologists to measure their performance on issues such as reaching the cecum, an anatomical landmark in the colon that represents a thorough exam.

Competing interests:   None declared