- Email: [email protected]
Measurement of polypectomy rate by using administrative claims data with validation against the adenoma detection rate
ORIGINAL ARTICLE: Clinical Endoscopy
Measurement of polypectomy rate by using administrative claims data with validation against the adenoma detection rate Neal C. Patel, MD,1 Rafiul S. Islam, MD,1 Qing Wu, ScD,2 Suryakanth R. Gurudu, MD,1 Francisco C. Ramirez, MD,1 Michael D. Crowell, MD,1 Douglas O. Faigel, MD1 Scottsdale, Arizona, USA
Background: The adenoma detection rate (ADR) is a main quality indicator in colonoscopy but has many challenges for calculating. The polypectomy rate (PR) may be calculable from administrative claims data, but this has not been validated against the ADR. Objective: To determine whether a PR calculated from United States billing claims data is an accurate surrogate for the ADR. Design: A PR was calculated by using billing claims data from Current Procedural Terminology codes. The ADR was calculated for each endoscopist by using an endoscopy report database to which the pathology report data had been added. The relationship between PR and ADR was evaluated with the Pearson correlation coefficient. The ADR was plotted against the PR by individual endoscopist, and a least-squares regression line was created. A t test was used to analyze the differences in lesion detection between endoscopists with a PR above and below the benchmark PR. Setting: Tertiary-care, outpatient endoscopy center. Patients: All ages undergoing colonoscopy. Main Outcome Measurements: PR and ADR. Results: A total of 5382 colonoscopies were reviewed. A significant relationship between endoscopists’ calculated PRs and ADRs was seen (r ⫽ 0.85; P ⬍ .001). Endoscopists needed a PR of 35% to achieve the recommended benchmark ADR of 20%. Endoscopists with PRs of 35% or greater had an ADR of 27% (6.2 standard deviation [SD]) as compared with 19% (1.9 SD) for those with PRs less than 35% (P ⫽ .0029). Limitations: Study population. Conclusion: Calculated PR from billing claims data is an accurate surrogate for ADR and may become an important quality measure for external and internal use. (Gastrointest Endosc 2013;77:390-4.)
As modern medicine has evolved, there has been a distinct move toward an increased emphasis on improvement in the quality of health care. Many organizations such as the U.S. Department of Health and Human Services and the American Medical Association have advocated for the development of more quantitative performance metrics to assess the quality of delivered health
care. This focus on quality is especially important in areas of medicine conducted for preventive purposes, such as colorectal cancer screening by colonoscopy. In 2012, there were an estimated 143,460 new cases of colorectal cancer diagnosed in the United States alone and 51,690 deaths from colorectal cancer.1 In the United States, colonoscopy is the primary method of screening for colo-
Abbreviations: ADR, adenoma detection rate; CPT, current procedural terminology; ICD-9, International Classification of Diseases, Ninth Revision; MERGE, Mayo Electronic Record for Gastrointestinal Endoscopy; PR, polypectomy rate. .
Received August 2, 2012. Accepted September 21, 2012.
DISCLOSURE: All authors disclosed no financial relationships relevant to this publication.
Presented at the American Society of Gastroenterology, Presidential Plenary Session, May 21, 2012, San Diego, California.
Copyright © 2013 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 http://dx.doi.org/10.1016/j.gie.2012.09.032
Reprint requests: Douglas O. Faigel, MD, Division of Gastroenterology, Mayo Clinic, 13400 E Shea Blvd, Scottsdale, AZ 85259.
390 GASTROINTESTINAL ENDOSCOPY Volume 77, No. 3 : 2013
Current affiliations: Division of Gastroenterology (1) and Division of Biostatistics (2), Mayo Clinic, Scottsdale, Arizona, USA.
www.giejournal.org
Patel et al
rectal cancer, and several studies have shown a substantial reduction in colorectal cancer rates among patients who have undergone polypectomy.2 However, studies also have shown that the quality of colonoscopy varies among endoscopists,3-5 which may be associated with variability in the rates of colonic neoplasia detection and removal.6,7
BACKGROUND Many quality measures have been proposed for colonoscopy to try to maximize preventive benefits. The most important of these has been the adenoma detection rate (ADR), which is the proportion of screening colonoscopies in which at least one adenoma is identified. In a previous study, the ADR of an endoscopist was inversely associated with the risk of interval colorectal cancer developing in the patients of that endoscopist.8 Measuring ADR has thus become a priority, and benchmarks for screening colonoscopy have been set at 25% and 15%, respectively, for asymptomatic men and women aged older than 50 years.9 Unfortunately, calculating the ADR of a specific physician presents several problems. First, it is time consuming because pathology reports are not available at the time of the endoscopy and may not be accessible for several days. Second, calculating the ADR requires a review of both the procedural report and the pathology report, with findings then correlated manually, which is difficult when sampling a large number of colonoscopies. Third, the ADR is not amenable to claimsbased reporting using currently available Current Procedural Terminology (CPT) codes, which makes it less effective for value-based purchasing or pay-for-performance programs.10 These limitations have led many researchers to seek a surrogate for ADR. The polypectomy rate (PR), which is the proportion of colonoscopies in which at least one polyp is removed, has been shown to correlate well with ADR.11,12 The PR can be calculated from procedure reports, thus eliminating the time delay inherent in using the ADR. Furthermore, PR potentially can be calculated from administrative claims data, which renders it useful in payfor-performance programs and claims-based reporting. However, to date, a calculated PR abstracted from United States administrative claims data has not been validated against the actual ADR. The aim of this study was to determine whether a PR calculated from U.S. billing claims data is an accurate surrogate for ADR.
METHODS We conducted a retrospective study of outpatient colonoscopies performed at Mayo Clinic, Scottsdale, Arizona, between January 1, 2009 and December 31, 2009. Electronic medical records were reviewed for all patients who underwent outpatient colonoscopies during the study period. Endoscopists reported all colonoscopy findwww.giejournal.org
Polypectomy rate from billing claims
Take-home Message ●
●
A polypectomy rate may be calculated easily from claims data by using existing diagnosis codes, and it correlates well with the actual adenoma detection rate. A rate of 35% correlates with an adenoma detection rate of 20%. A calculated polypectomy rate may be useful for quality improvement and value-based purchasing programs for both external and internal use.
ings by using a standard computerized endoscopy report generator (Mayo Electronic Record for Gastrointestinal Endoscopy [MERGE]). Repeat colonoscopies in the same patient during the study period were excluded if the initial colonoscopy detected adenomas. Procedures also were excluded if the endoscopist performing them had conducted fewer than 50 eligible colonoscopies during the study period. A spreadsheet database from the MERGE endoscopy reporting system was created that included all colonoscopies during the calendar year, and the pathology results were entered as abstracted from the individual pathology reports. Patient identifiers were not included in the final database. The ADR was calculated for each endoscopist. All patients used polyethylene glycol– based bowelcleansing preparations, and all colonoscopies were performed by attending gastroenterologists, by colorectal surgeons, or by gastroenterology fellows under direct supervision of the attending physician. Procedures performed with trainee involvement were attributed to the attending physician. Standard high-definition colonoscopies were used for all procedures, most of which were performed with the patient under moderate sedation by using intravenous fentanyl citrate, midazolam, or meperidine hydrochloride, per the preference of the performing endoscopist. We created a separate database of colonoscopies by conducting a computerized search of billing data from our institution. We identified all patients with billing submitted for colonoscopy by using the CPT codes 45378, 45380, 45381, 45382, 45383, 45384, and 45385 during the same calendar year (2009). The database included the name of the attending physician, the date of the procedure, and all ICD-9 (International Classification of Diseases, Ninth Revision) codes associated with the bill. Patient-specific and identifying information such as age and sex were not recorded. From this database, the PR was calculated for each endoscopist as the proportion of all colonoscopies with at least one of the following CPT codes: 45385 (snare polypectomy), 45384 (hot biopsy), or 45380 (cold biopsy), along with the polyp ICD-9 code 211.3. For patients who had more than one colonoscopy during the study period, only the first colonoscopy was included. Volume 77, No. 3 : 2013 GASTROINTESTINAL ENDOSCOPY 391
Polypectomy rate from billing claims
Patel et al
TABLE 1. Characteristics of 5382 patients undergoing colonoscopy
Indication
Age, Sex, % Proportion Mean of total, % (SD) Male Female
TABLE 2. Colonoscopies sorted by polypectomy rate Total no. of colonoscopies
PR, %a
ADR, %a
Q
527
55.41
38.39
Endoscopist
CRC screening
33
60 (9)
55
45
L
194
51.03
25.12
Polyp surveillance
23
68 (10)
61
39
H
487
49.69
35.45
Altered bowel habit
13
58 (16)
32
68
N
398
44.72
29.29
P
190
42.63
20.77
Bleeding
9
61 (16)
54
46
D
297
42.09
29.58
Family history of colon cancer
6
60 (12)
40
60
E
239
41.84
19.76
Abdominal pain
5
55 (16)
33
67
R
240
40.83
25.19
Anemia
3
68 (14)
44
56
F
157
38.22
24.00
Personal history of colon cancer
3
68 (11)
49
51
J
182
36.81
21.16
S
104
34.62
19.01
Other
5
K
194
34.21
20.19
B
63
33.33
17.11
O
284
32.75
22.98
M
307
32.57
18.56
I
410
32.44
19.34
G
273
30.40
18.48
A
313
27.16
18.10
C
74
27.03
16.46
63 (18)
47
53
SD, Standard deviation; CRC, colorectal cancer.
The association between PR and ADR was evaluated by using the Pearson correlation coefficient. The ADR was plotted against the PR for individual endoscopists, and a least-squares regression line was created. Finally, by using the average benchmark ADR of 20% for both men and women patients together,7 we used the regression line equation to determine a benchmark PR. The t test was used to evaluate the differences in ADR between endoscopists with a PR above or below the benchmark PR. Statistical software (SAS version 9.1; SAS Institute Inc, Cary, NC) was used for data analysis. Statistical significance was defined as P ⱕ .05. The Mayo Clinic Institutional Review Board reviewed the study parameters and deemed the study exempt.
RESULTS A total of 5382 colonoscopy reports were reviewed, as were the pathology reports corresponding to each procedure. A total of 19 endoscopists met the 50-procedure minimum inclusion criterion. Of these 19, most (16 endoscopists) were attending gastroenterologists; 3 were attending colorectal surgeons. The mean number of colonoscopies performed by each endoscopist was 275 (range 63-527). Table 1 summarizes the characteristics of the patients who underwent colonoscopy. The mean (⫾ standard deviation [SD]) age of patients was 62 years (⫾ 14 years), and 46% (2475/5382) were men. Table 2 lists the calculated PR from the claims data for each endoscopist, and it gives the corresponding ADR from the endoscopic database, sorted by decreasing PR. The PR ranged from 27% to 55%, and 392 GASTROINTESTINAL ENDOSCOPY Volume 77, No. 3 : 2013
PR, Polypectomy rate; ADR, adenoma detection rate. aPercentages carried to 2 decimal places as configured by the endoscopic database.
the ADR ranged from 16% to 38%. A significant association was found between PR and ADR (r ⫽ 0.85; P ⬍ .001). Linear regression showed that endoscopists needed a PR of 35% to achieve the recommended benchmark ADR of 20% (Fig. 1). Endoscopists with PRs of ⱖ35% had a mean ADR (⫾ SD) of 27% (⫾ 6.2%), compared with a mean ADR (⫾ SD) of 19% (⫾ 1.9%) for endoscopists with PRs ⬍35% (P ⫽ .003). Most (7/9 [78%]) endoscopists with PRs ⬍35% had ADRs ⬍20%, whereas only 10% (1/10) with PRs of ⱖ35% had ADRs of ⬍20%, and that one endoscopist had an ADR of 19.76%.
DISCUSSION Colonoscopy is the primary method of screening for colorectal cancer in the United States. Although it represents a cost-effective means of screening for colorectal neoplasia, colonoscopy still constitutes an invasive screening examination with inherent costs and risks.13 For this reason, it is imperative to focus on the quality of colowww.giejournal.org
Patel et al
Figure 1. Adenoma detection rate by polypectomy rate. The Pearson correlation coefficient (r) demonstrates a significant correlation between the adenoma detection rate and the polypectomy rate.
rectal cancer screening and surveillance received from colonoscopy. Doing so has become a major point of emphasis for organizations such as the American Society for Gastrointestinal Endoscopy and the American College of Gastroenterology.8 Several studies have shown that the quality of colonoscopy varies among endoscopists and that these variations can lead to changes in polyp detection, including the detection of premalignant lesions.3-6 Consequently, many different quality indicators, such as cecal intubation rates and withdrawal time, have been proposed.8 Because the purpose of screening and surveillance colonoscopy is to find and remove adenomatous polyps, the ADR has emerged as the most important quality metric. An ADR of ⱖ20% has been found to be associated with lower rates of interval colon cancer in a previous study and has become a benchmark for screening colonoscopy.7 However, the ADR has many limitations, the most important being that it is time-consuming to calculate, and it is not amenable to claims-based reporting. As a way to address the limitations of the ADR, the PR has been proposed as a quality measure because it eliminates the need to correlate two separate reports, and it has been found to correlate well with the ADR.10,11,14 The PR also correlates with protection from interval colorectal cancer. Two recent studies in Canada and the United States have described that a PR of ⱖ30% was associated with a lower risk of postcolonoscopy or interval colorectal cancer.15,16 However, to our knowledge, no studies previous to ours have examined whether a calculated PR from U.S. billings claims data might be an accurate surrogate for the true ADR. We found that a calculated PR from billings data has a strong correlation with the ADR. Furthermore, we found that a calculated PR of ⱖ35% was needed to achieve the benchmark ADR of 20%. These findings have broad implications for both external and internal use. A calculated PR is extremely quick www.giejournal.org
Polypectomy rate from billing claims
and easy to determine, and it can be used not only for quality audits but also for pay-for-performance programs or value-based purchasing. A calculated PR could thus be incorporated into programs such as the Centers for Medicare and Medicaid Services Physician Quality Reporting System. There are some potential limitations to using the PR rather than the ADR as a quality metric. The primary concern is of endoscopists possibly “gaming” the system, such as through the removal of small, irrelevant, nonadenomatous polyps in order to achieve certain quality metrics. This could lead to increased costs and risks associated with colonoscopy by exposing patients to unneeded biopsy or tissue removal. However, in our current reimbursement structure, there is already a monetary incentive to remove polyps, because colonoscopy with polypectomy or biopsy is reimbursed at a substantially higher amount than colonoscopy as a diagnostic examination. This potential issue could be addressed by conducting periodic audits of a sample of reports to ensure that the PRs and the ADRs of each endoscopist have a strong correlation. Nonetheless, one strength of the PR is that it can be calculated externally from claims by payers or other auditors without access to patient records. This cannot be done with the ADR, which generally requires an employee of the practice to calculate it by reviewing patient records. Thus, ADR is difficult to audit externally for accuracy. A limitation of our study is that the collected claims data were deidentified, which precluded analysis of the PR by patient demographic information. Thus, patients could not be categorized by sex, which is a determinant in benchmark ADRs. Second, this study was conducted at a tertiarycare, academic referral center, which does not offer financial incentives to its providers based on billing, which is unique from most endoscopists’ compensation models. Also, our center has a separate in-house billing department that reviews all claims, which is potentially more accurate than billing processes in other settings. For example, it is unclear whether the same analysis at an independent endoscopy center would garner similar results. Fortunately, most endoscopy centers use some method of automated billing assistance and audits, which makes the risk of inaccurate billing less of a concern. Finally, it should be noted that the established benchmark ADRs are based on screening colonoscopies, and the database reviewed in this study included colonoscopies for all indications. However, our findings show a strong correlation between the ADR and the PR, and the average adenomato-polyp detection rate quotient has been described previously as approximately 0.6.14 Taking this finding into account, one might predict that achieving an ADR of 20% would require a PR of 30% to 35%. Although a study limited to screening examinations could conclude that a benchmark PR other than 35% is needed, this possibility is unlikely. Volume 77, No. 3 : 2013 GASTROINTESTINAL ENDOSCOPY 393
Polypectomy rate from billing claims
In conclusion, a calculated PR by using billing claims data can be an accurate surrogate for the ADR and is also a much more efficient quality metric to use in colonoscopy. A PR of ⱖ35% is needed to meet the benchmark ADR of 20%, which has been shown to decrease the risk of interval colon cancer. This calculated PR could be used for pay-for-performance programs, internal or external quality audits, or value-based purchasing.
REFERENCES
Patel et al
6.
7. 8. 9.
10.
1. Siegel R, Naishadham D, Jemal A. Cancer statistics. CA Cancer J Clin 2012;62:10-29. Epub 2012 Jan 4. 2. Winawer SJ, Zauber AG, Ho MN, et al. The National Polyp Study Workgroup. Prevention of colorectal cancer by colonoscopic polypectomy. N Engl J Med 1993;30;329:1977-81. 3. Barclay RL, Vicari JJ, Doughty AS, et al. Colonoscopic withdrawal time and adenoma detection during screening colonoscopy. N Engl J Med 2006;355:2533-41. 4. Levin B, Lieberman DA, McFarland B, et al. American Cancer Society Colorectal Cancer Advisory Group; US Multi-Society Task Force; American College of Radiology Colon Cancer Committee. Screening and surveillance for the early detection of colorectal cancer and adenomatous polyps: a joint guideline from the American Cancer Society, the US Multi-Society Task Force on Colorectal Cancer, and the American College of Radiology. Gastroenterology 2008;134:1570-95. Epub 2008 Feb 8. 5. Rex DK, Johnson DA, Anderson JC, et al. American College of Gastroenterology. American College of Gastroenterology guidelines for colorec-
11. 12.
13. 14.
15.
16.
tal cancer screening [corrected]. Am J Gastroenterol 2009;104:739-50. Epub 2009 Feb 24. Erratum in: Am J Gastroenterol 2009;104:1613. Rex DK, Cutler CS, Lemmel GT, et al. Colonoscopic miss rates of adenomas determined by back-to-back colonoscopies. Gastroenterology 1997;112:24-8. Imperiale TF, Glowinski EA, Juliar BE, et al. Variation in polyp detection rates at screening colonoscopy. Gastrointest Endosc 2009;69:1288-95. Kaminski MF, Regula J, Kraszewska E, et al. Quality indicators for colonoscopy and the risk of interval cancer. N Engl J Med 2010;362:1795-803. Rex DK, Petrini JL, Baron TH, et al. ASGE/ACG Taskforce on Quality in Endoscopy. Quality indicators for colonoscopy. Gastrointest Endosc 2006b;63(suppl 4):S16-28. Lin OS. Less stick, more carrot: measuring and improving patient satisfaction with endoscopic procedures. Gastrointest Endosc 2009;69: 892-5. Williams JE, Le TD, Faigel DO. Polypectomy rate as a quality measure for colonoscopy. Gastrointest Endosc 2011;73:498-506. Williams JE, Holub JL, Faigel DO. Polypectomy rate is a valid quality measure for colonoscopy: results from a national endoscopy database. Gastrointest Endosc 2012;75:576-82. Sonnenberg A, Delco F, Inadomi JM. Cost-effectiveness of colonoscopy in screening for colorectal cancer. Ann Intern Med 2000;133:573-84. Francis DL, Rodriguez-Correa DT, Buchner A, et al. Application of a conversion factor to estimate the adenoma detection rate from the polyp detection rate. Gastrointest Endosc 2011;73:493-7. Baxter NN, Sutradhar R, Forbes SS, et al. Analysis of administrative data finds endoscopist quality measures associated with postcolonoscopy colorectal cancer. Gastroenterology 2011;140:65-72. Epub 2010 Sep 18. Cooper GS, Xu F, Barnholtz Sloan JS, et al. Prevalence and predictors of interval colorectal cancers in Medicare beneficiaries. Cancer 2012;118: 3044-52.
Online Audio and Podcasting Audio and Podcasts of article abstracts published in Gastrointestinal Endoscopy are now available online. Recordings are edited by Ian Gralnek, MD, MSHS, Senior Associate Editor, and performed by Deborah Bowman, MFA, Managing Editor of Gastrointestinal Endoscopy. Log on to www.giejournal.org to listen to recordings from the current issue.
394 GASTROINTESTINAL ENDOSCOPY Volume 77, No. 3 : 2013
www.giejournal.org
Measurement of polypectomy rate by using administrative claims data with validation against the adenoma detection rate Neal C. Patel, MD,1 Rafiul S. Islam, MD,1 Qing Wu, ScD,2 Suryakanth R. Gurudu, MD,1 Francisco C. Ramirez, MD,1 Michael D. Crowell, MD,1 Douglas O. Faigel, MD1 Scottsdale, Arizona, USA
Background: The adenoma detection rate (ADR) is a main quality indicator in colonoscopy but has many challenges for calculating. The polypectomy rate (PR) may be calculable from administrative claims data, but this has not been validated against the ADR. Objective: To determine whether a PR calculated from United States billing claims data is an accurate surrogate for the ADR. Design: A PR was calculated by using billing claims data from Current Procedural Terminology codes. The ADR was calculated for each endoscopist by using an endoscopy report database to which the pathology report data had been added. The relationship between PR and ADR was evaluated with the Pearson correlation coefficient. The ADR was plotted against the PR by individual endoscopist, and a least-squares regression line was created. A t test was used to analyze the differences in lesion detection between endoscopists with a PR above and below the benchmark PR. Setting: Tertiary-care, outpatient endoscopy center. Patients: All ages undergoing colonoscopy. Main Outcome Measurements: PR and ADR. Results: A total of 5382 colonoscopies were reviewed. A significant relationship between endoscopists’ calculated PRs and ADRs was seen (r ⫽ 0.85; P ⬍ .001). Endoscopists needed a PR of 35% to achieve the recommended benchmark ADR of 20%. Endoscopists with PRs of 35% or greater had an ADR of 27% (6.2 standard deviation [SD]) as compared with 19% (1.9 SD) for those with PRs less than 35% (P ⫽ .0029). Limitations: Study population. Conclusion: Calculated PR from billing claims data is an accurate surrogate for ADR and may become an important quality measure for external and internal use. (Gastrointest Endosc 2013;77:390-4.)
As modern medicine has evolved, there has been a distinct move toward an increased emphasis on improvement in the quality of health care. Many organizations such as the U.S. Department of Health and Human Services and the American Medical Association have advocated for the development of more quantitative performance metrics to assess the quality of delivered health
care. This focus on quality is especially important in areas of medicine conducted for preventive purposes, such as colorectal cancer screening by colonoscopy. In 2012, there were an estimated 143,460 new cases of colorectal cancer diagnosed in the United States alone and 51,690 deaths from colorectal cancer.1 In the United States, colonoscopy is the primary method of screening for colo-
Abbreviations: ADR, adenoma detection rate; CPT, current procedural terminology; ICD-9, International Classification of Diseases, Ninth Revision; MERGE, Mayo Electronic Record for Gastrointestinal Endoscopy; PR, polypectomy rate. .
Received August 2, 2012. Accepted September 21, 2012.
DISCLOSURE: All authors disclosed no financial relationships relevant to this publication.
Presented at the American Society of Gastroenterology, Presidential Plenary Session, May 21, 2012, San Diego, California.
Copyright © 2013 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 http://dx.doi.org/10.1016/j.gie.2012.09.032
Reprint requests: Douglas O. Faigel, MD, Division of Gastroenterology, Mayo Clinic, 13400 E Shea Blvd, Scottsdale, AZ 85259.
390 GASTROINTESTINAL ENDOSCOPY Volume 77, No. 3 : 2013
Current affiliations: Division of Gastroenterology (1) and Division of Biostatistics (2), Mayo Clinic, Scottsdale, Arizona, USA.
www.giejournal.org
Patel et al
rectal cancer, and several studies have shown a substantial reduction in colorectal cancer rates among patients who have undergone polypectomy.2 However, studies also have shown that the quality of colonoscopy varies among endoscopists,3-5 which may be associated with variability in the rates of colonic neoplasia detection and removal.6,7
BACKGROUND Many quality measures have been proposed for colonoscopy to try to maximize preventive benefits. The most important of these has been the adenoma detection rate (ADR), which is the proportion of screening colonoscopies in which at least one adenoma is identified. In a previous study, the ADR of an endoscopist was inversely associated with the risk of interval colorectal cancer developing in the patients of that endoscopist.8 Measuring ADR has thus become a priority, and benchmarks for screening colonoscopy have been set at 25% and 15%, respectively, for asymptomatic men and women aged older than 50 years.9 Unfortunately, calculating the ADR of a specific physician presents several problems. First, it is time consuming because pathology reports are not available at the time of the endoscopy and may not be accessible for several days. Second, calculating the ADR requires a review of both the procedural report and the pathology report, with findings then correlated manually, which is difficult when sampling a large number of colonoscopies. Third, the ADR is not amenable to claimsbased reporting using currently available Current Procedural Terminology (CPT) codes, which makes it less effective for value-based purchasing or pay-for-performance programs.10 These limitations have led many researchers to seek a surrogate for ADR. The polypectomy rate (PR), which is the proportion of colonoscopies in which at least one polyp is removed, has been shown to correlate well with ADR.11,12 The PR can be calculated from procedure reports, thus eliminating the time delay inherent in using the ADR. Furthermore, PR potentially can be calculated from administrative claims data, which renders it useful in payfor-performance programs and claims-based reporting. However, to date, a calculated PR abstracted from United States administrative claims data has not been validated against the actual ADR. The aim of this study was to determine whether a PR calculated from U.S. billing claims data is an accurate surrogate for ADR.
METHODS We conducted a retrospective study of outpatient colonoscopies performed at Mayo Clinic, Scottsdale, Arizona, between January 1, 2009 and December 31, 2009. Electronic medical records were reviewed for all patients who underwent outpatient colonoscopies during the study period. Endoscopists reported all colonoscopy findwww.giejournal.org
Polypectomy rate from billing claims
Take-home Message ●
●
A polypectomy rate may be calculated easily from claims data by using existing diagnosis codes, and it correlates well with the actual adenoma detection rate. A rate of 35% correlates with an adenoma detection rate of 20%. A calculated polypectomy rate may be useful for quality improvement and value-based purchasing programs for both external and internal use.
ings by using a standard computerized endoscopy report generator (Mayo Electronic Record for Gastrointestinal Endoscopy [MERGE]). Repeat colonoscopies in the same patient during the study period were excluded if the initial colonoscopy detected adenomas. Procedures also were excluded if the endoscopist performing them had conducted fewer than 50 eligible colonoscopies during the study period. A spreadsheet database from the MERGE endoscopy reporting system was created that included all colonoscopies during the calendar year, and the pathology results were entered as abstracted from the individual pathology reports. Patient identifiers were not included in the final database. The ADR was calculated for each endoscopist. All patients used polyethylene glycol– based bowelcleansing preparations, and all colonoscopies were performed by attending gastroenterologists, by colorectal surgeons, or by gastroenterology fellows under direct supervision of the attending physician. Procedures performed with trainee involvement were attributed to the attending physician. Standard high-definition colonoscopies were used for all procedures, most of which were performed with the patient under moderate sedation by using intravenous fentanyl citrate, midazolam, or meperidine hydrochloride, per the preference of the performing endoscopist. We created a separate database of colonoscopies by conducting a computerized search of billing data from our institution. We identified all patients with billing submitted for colonoscopy by using the CPT codes 45378, 45380, 45381, 45382, 45383, 45384, and 45385 during the same calendar year (2009). The database included the name of the attending physician, the date of the procedure, and all ICD-9 (International Classification of Diseases, Ninth Revision) codes associated with the bill. Patient-specific and identifying information such as age and sex were not recorded. From this database, the PR was calculated for each endoscopist as the proportion of all colonoscopies with at least one of the following CPT codes: 45385 (snare polypectomy), 45384 (hot biopsy), or 45380 (cold biopsy), along with the polyp ICD-9 code 211.3. For patients who had more than one colonoscopy during the study period, only the first colonoscopy was included. Volume 77, No. 3 : 2013 GASTROINTESTINAL ENDOSCOPY 391
Polypectomy rate from billing claims
Patel et al
TABLE 1. Characteristics of 5382 patients undergoing colonoscopy
Indication
Age, Sex, % Proportion Mean of total, % (SD) Male Female
TABLE 2. Colonoscopies sorted by polypectomy rate Total no. of colonoscopies
PR, %a
ADR, %a
Q
527
55.41
38.39
Endoscopist
CRC screening
33
60 (9)
55
45
L
194
51.03
25.12
Polyp surveillance
23
68 (10)
61
39
H
487
49.69
35.45
Altered bowel habit
13
58 (16)
32
68
N
398
44.72
29.29
P
190
42.63
20.77
Bleeding
9
61 (16)
54
46
D
297
42.09
29.58
Family history of colon cancer
6
60 (12)
40
60
E
239
41.84
19.76
Abdominal pain
5
55 (16)
33
67
R
240
40.83
25.19
Anemia
3
68 (14)
44
56
F
157
38.22
24.00
Personal history of colon cancer
3
68 (11)
49
51
J
182
36.81
21.16
S
104
34.62
19.01
Other
5
K
194
34.21
20.19
B
63
33.33
17.11
O
284
32.75
22.98
M
307
32.57
18.56
I
410
32.44
19.34
G
273
30.40
18.48
A
313
27.16
18.10
C
74
27.03
16.46
63 (18)
47
53
SD, Standard deviation; CRC, colorectal cancer.
The association between PR and ADR was evaluated by using the Pearson correlation coefficient. The ADR was plotted against the PR for individual endoscopists, and a least-squares regression line was created. Finally, by using the average benchmark ADR of 20% for both men and women patients together,7 we used the regression line equation to determine a benchmark PR. The t test was used to evaluate the differences in ADR between endoscopists with a PR above or below the benchmark PR. Statistical software (SAS version 9.1; SAS Institute Inc, Cary, NC) was used for data analysis. Statistical significance was defined as P ⱕ .05. The Mayo Clinic Institutional Review Board reviewed the study parameters and deemed the study exempt.
RESULTS A total of 5382 colonoscopy reports were reviewed, as were the pathology reports corresponding to each procedure. A total of 19 endoscopists met the 50-procedure minimum inclusion criterion. Of these 19, most (16 endoscopists) were attending gastroenterologists; 3 were attending colorectal surgeons. The mean number of colonoscopies performed by each endoscopist was 275 (range 63-527). Table 1 summarizes the characteristics of the patients who underwent colonoscopy. The mean (⫾ standard deviation [SD]) age of patients was 62 years (⫾ 14 years), and 46% (2475/5382) were men. Table 2 lists the calculated PR from the claims data for each endoscopist, and it gives the corresponding ADR from the endoscopic database, sorted by decreasing PR. The PR ranged from 27% to 55%, and 392 GASTROINTESTINAL ENDOSCOPY Volume 77, No. 3 : 2013
PR, Polypectomy rate; ADR, adenoma detection rate. aPercentages carried to 2 decimal places as configured by the endoscopic database.
the ADR ranged from 16% to 38%. A significant association was found between PR and ADR (r ⫽ 0.85; P ⬍ .001). Linear regression showed that endoscopists needed a PR of 35% to achieve the recommended benchmark ADR of 20% (Fig. 1). Endoscopists with PRs of ⱖ35% had a mean ADR (⫾ SD) of 27% (⫾ 6.2%), compared with a mean ADR (⫾ SD) of 19% (⫾ 1.9%) for endoscopists with PRs ⬍35% (P ⫽ .003). Most (7/9 [78%]) endoscopists with PRs ⬍35% had ADRs ⬍20%, whereas only 10% (1/10) with PRs of ⱖ35% had ADRs of ⬍20%, and that one endoscopist had an ADR of 19.76%.
DISCUSSION Colonoscopy is the primary method of screening for colorectal cancer in the United States. Although it represents a cost-effective means of screening for colorectal neoplasia, colonoscopy still constitutes an invasive screening examination with inherent costs and risks.13 For this reason, it is imperative to focus on the quality of colowww.giejournal.org
Patel et al
Figure 1. Adenoma detection rate by polypectomy rate. The Pearson correlation coefficient (r) demonstrates a significant correlation between the adenoma detection rate and the polypectomy rate.
rectal cancer screening and surveillance received from colonoscopy. Doing so has become a major point of emphasis for organizations such as the American Society for Gastrointestinal Endoscopy and the American College of Gastroenterology.8 Several studies have shown that the quality of colonoscopy varies among endoscopists and that these variations can lead to changes in polyp detection, including the detection of premalignant lesions.3-6 Consequently, many different quality indicators, such as cecal intubation rates and withdrawal time, have been proposed.8 Because the purpose of screening and surveillance colonoscopy is to find and remove adenomatous polyps, the ADR has emerged as the most important quality metric. An ADR of ⱖ20% has been found to be associated with lower rates of interval colon cancer in a previous study and has become a benchmark for screening colonoscopy.7 However, the ADR has many limitations, the most important being that it is time-consuming to calculate, and it is not amenable to claims-based reporting. As a way to address the limitations of the ADR, the PR has been proposed as a quality measure because it eliminates the need to correlate two separate reports, and it has been found to correlate well with the ADR.10,11,14 The PR also correlates with protection from interval colorectal cancer. Two recent studies in Canada and the United States have described that a PR of ⱖ30% was associated with a lower risk of postcolonoscopy or interval colorectal cancer.15,16 However, to our knowledge, no studies previous to ours have examined whether a calculated PR from U.S. billings claims data might be an accurate surrogate for the true ADR. We found that a calculated PR from billings data has a strong correlation with the ADR. Furthermore, we found that a calculated PR of ⱖ35% was needed to achieve the benchmark ADR of 20%. These findings have broad implications for both external and internal use. A calculated PR is extremely quick www.giejournal.org
Polypectomy rate from billing claims
and easy to determine, and it can be used not only for quality audits but also for pay-for-performance programs or value-based purchasing. A calculated PR could thus be incorporated into programs such as the Centers for Medicare and Medicaid Services Physician Quality Reporting System. There are some potential limitations to using the PR rather than the ADR as a quality metric. The primary concern is of endoscopists possibly “gaming” the system, such as through the removal of small, irrelevant, nonadenomatous polyps in order to achieve certain quality metrics. This could lead to increased costs and risks associated with colonoscopy by exposing patients to unneeded biopsy or tissue removal. However, in our current reimbursement structure, there is already a monetary incentive to remove polyps, because colonoscopy with polypectomy or biopsy is reimbursed at a substantially higher amount than colonoscopy as a diagnostic examination. This potential issue could be addressed by conducting periodic audits of a sample of reports to ensure that the PRs and the ADRs of each endoscopist have a strong correlation. Nonetheless, one strength of the PR is that it can be calculated externally from claims by payers or other auditors without access to patient records. This cannot be done with the ADR, which generally requires an employee of the practice to calculate it by reviewing patient records. Thus, ADR is difficult to audit externally for accuracy. A limitation of our study is that the collected claims data were deidentified, which precluded analysis of the PR by patient demographic information. Thus, patients could not be categorized by sex, which is a determinant in benchmark ADRs. Second, this study was conducted at a tertiarycare, academic referral center, which does not offer financial incentives to its providers based on billing, which is unique from most endoscopists’ compensation models. Also, our center has a separate in-house billing department that reviews all claims, which is potentially more accurate than billing processes in other settings. For example, it is unclear whether the same analysis at an independent endoscopy center would garner similar results. Fortunately, most endoscopy centers use some method of automated billing assistance and audits, which makes the risk of inaccurate billing less of a concern. Finally, it should be noted that the established benchmark ADRs are based on screening colonoscopies, and the database reviewed in this study included colonoscopies for all indications. However, our findings show a strong correlation between the ADR and the PR, and the average adenomato-polyp detection rate quotient has been described previously as approximately 0.6.14 Taking this finding into account, one might predict that achieving an ADR of 20% would require a PR of 30% to 35%. Although a study limited to screening examinations could conclude that a benchmark PR other than 35% is needed, this possibility is unlikely. Volume 77, No. 3 : 2013 GASTROINTESTINAL ENDOSCOPY 393
Polypectomy rate from billing claims
In conclusion, a calculated PR by using billing claims data can be an accurate surrogate for the ADR and is also a much more efficient quality metric to use in colonoscopy. A PR of ⱖ35% is needed to meet the benchmark ADR of 20%, which has been shown to decrease the risk of interval colon cancer. This calculated PR could be used for pay-for-performance programs, internal or external quality audits, or value-based purchasing.
REFERENCES
Patel et al
6.
7. 8. 9.
10.
1. Siegel R, Naishadham D, Jemal A. Cancer statistics. CA Cancer J Clin 2012;62:10-29. Epub 2012 Jan 4. 2. Winawer SJ, Zauber AG, Ho MN, et al. The National Polyp Study Workgroup. Prevention of colorectal cancer by colonoscopic polypectomy. N Engl J Med 1993;30;329:1977-81. 3. Barclay RL, Vicari JJ, Doughty AS, et al. Colonoscopic withdrawal time and adenoma detection during screening colonoscopy. N Engl J Med 2006;355:2533-41. 4. Levin B, Lieberman DA, McFarland B, et al. American Cancer Society Colorectal Cancer Advisory Group; US Multi-Society Task Force; American College of Radiology Colon Cancer Committee. Screening and surveillance for the early detection of colorectal cancer and adenomatous polyps: a joint guideline from the American Cancer Society, the US Multi-Society Task Force on Colorectal Cancer, and the American College of Radiology. Gastroenterology 2008;134:1570-95. Epub 2008 Feb 8. 5. Rex DK, Johnson DA, Anderson JC, et al. American College of Gastroenterology. American College of Gastroenterology guidelines for colorec-
11. 12.
13. 14.
15.
16.
tal cancer screening [corrected]. Am J Gastroenterol 2009;104:739-50. Epub 2009 Feb 24. Erratum in: Am J Gastroenterol 2009;104:1613. Rex DK, Cutler CS, Lemmel GT, et al. Colonoscopic miss rates of adenomas determined by back-to-back colonoscopies. Gastroenterology 1997;112:24-8. Imperiale TF, Glowinski EA, Juliar BE, et al. Variation in polyp detection rates at screening colonoscopy. Gastrointest Endosc 2009;69:1288-95. Kaminski MF, Regula J, Kraszewska E, et al. Quality indicators for colonoscopy and the risk of interval cancer. N Engl J Med 2010;362:1795-803. Rex DK, Petrini JL, Baron TH, et al. ASGE/ACG Taskforce on Quality in Endoscopy. Quality indicators for colonoscopy. Gastrointest Endosc 2006b;63(suppl 4):S16-28. Lin OS. Less stick, more carrot: measuring and improving patient satisfaction with endoscopic procedures. Gastrointest Endosc 2009;69: 892-5. Williams JE, Le TD, Faigel DO. Polypectomy rate as a quality measure for colonoscopy. Gastrointest Endosc 2011;73:498-506. Williams JE, Holub JL, Faigel DO. Polypectomy rate is a valid quality measure for colonoscopy: results from a national endoscopy database. Gastrointest Endosc 2012;75:576-82. Sonnenberg A, Delco F, Inadomi JM. Cost-effectiveness of colonoscopy in screening for colorectal cancer. Ann Intern Med 2000;133:573-84. Francis DL, Rodriguez-Correa DT, Buchner A, et al. Application of a conversion factor to estimate the adenoma detection rate from the polyp detection rate. Gastrointest Endosc 2011;73:493-7. Baxter NN, Sutradhar R, Forbes SS, et al. Analysis of administrative data finds endoscopist quality measures associated with postcolonoscopy colorectal cancer. Gastroenterology 2011;140:65-72. Epub 2010 Sep 18. Cooper GS, Xu F, Barnholtz Sloan JS, et al. Prevalence and predictors of interval colorectal cancers in Medicare beneficiaries. Cancer 2012;118: 3044-52.
Online Audio and Podcasting Audio and Podcasts of article abstracts published in Gastrointestinal Endoscopy are now available online. Recordings are edited by Ian Gralnek, MD, MSHS, Senior Associate Editor, and performed by Deborah Bowman, MFA, Managing Editor of Gastrointestinal Endoscopy. Log on to www.giejournal.org to listen to recordings from the current issue.
394 GASTROINTESTINAL ENDOSCOPY Volume 77, No. 3 : 2013
www.giejournal.org