Pilot of Decision Support to Individualize Colorectal Cancer Screening Recommendations

ABSTRACT

Objectives: To test the feasibility of using an electronic medical record (EMR)-based decision support system (DSS) that incorporates morbidity and frailty information to individualize colorectal cancer (CRC) screening recommendations.

Study Design: Our framework used the payoff time, defined as the minimum time until the benefits of screening exceed the harms.

Methods: Subjects were 24 patients eligible for CRC screening and 22 primary care providers (PCPs). Measures included PCP satisfaction with existing reminder systems and with decision support.

Results: The run-in phase, during which the intervention was inactive but its performance was verified, had 14 patients enrolled. The intervention phase, during which payoff time and life expectancy calculations were used to recommend for or against CRC screening, had 10 patients enrolled. Of the 10 patients enrolled in the intervention phase, the DSS recommended in favor of CRC screening for 6 patients. (The PCPs also recommended it for those 6 patients, although 3 refused the screening.) The DSS recommended against CRC screening for 4 patients, while the PCPs recommended against it for 3 of those 4 and ordered the screening for 1 patient. PCPs who had patients enrolled in the intervention phase indicated interest in having payoff time information for all patients eligible for CRC screening. This pilot study was small and was not powered to determine the effect of the intervention on screening behavior.

Conclusions: Colorectal cancer screening involves balancing immediate harms with longer-term benefits; EMR decision support may facilitate personalized benefit/harm assessment. The payoff time framework is feasible for implementation in EMR decision support.

Am J Manag Care. 2015;21(7):e439-e446

Take-Away Points

We tested the feasibility of using an electronic medical record-based decision support that incorporates morbidity and frailty information to individualize colorectal cancer screening recommendations.

• Colorectal cancer screening involves balancing immediate harms with longer-term benefits.
• Our framework used the payoff time: the minimum time until the benefits of screening exceed the harms.
• Electronic medical record decision support may facilitate personalized benefit/harm assessment.

Colorectal cancer (CRC) is the third most common cancer and the second leading cause of cancer deaths in the United States.¹ CRC has a long preclinical phase from the development of adenomatous polyps to symptomatic cancer.² Randomized controlled trials of fecal occult blood tests (FOBTs) and sigmoidoscopy and case-control studies of colonoscopy have shown that screening reduces CRC mortality,^3-8 with the interval between screening and mortality benefit thought to be at least 5 to 7 years.⁹

While screening for CRC has considerable benefits, it also has harms. The possible harms of colonoscopy include inconvenience, cardiopulmonary complications from sedation, and complications from the colonoscopy itself.^9,10 The decision to screen for CRC involves balancing the immediate harms with longer-term benefits; often, advanced age is used as a surrogate for an unfavorable harm-to-benefit profile. The US Preventive Services Task Force (USPSTF) recommends screening beginning at age 50 years and continuing until age 75 years using FOBT, sigmoidoscopy, or colonoscopy, noting that “The lead time between detection and treatment of colorectal neoplasia and a mortality benefit is substantial, and competing causes of mortality make it progressively less likely that this benefit will be realized with advancing age.”¹¹

Comorbidity and frailty may be more important than age in determining life expectancy (LE) and the individualized benefit-to-harm profile. Despite the importance of incorporating comorbidity and frailty into screening decisions, guidelines do not specify how to do this.^11-13 In practice, CRC screening is as common among those with severe comorbidities as among healthy persons.^14-16

Most research regarding implementation of CRC screening has focused on increasing use, with minimal evaluation of strategies to reduce overuse and misuse.¹⁷ Overuse may occur because physicians have difficulty estimating LE, few validated accessible tools for doing so, or little time to do so in the context of competing demands. Because screening is offered largely in the context of clinic visits, ill patients with frequent visits may be inappropriately targeted.¹⁵ Quality measures and clinical reminder systems often promote CRC screening regardless of comorbidity. The clinical reminder in the Veterans Administration (VA) electronic medical record (EMR), or Computerized Patient Record System (CPRS), is activated for patients aged 50 to 75 years, regardless of comorbidities. At the time this study was conducted, the VA’s External Peer Review Program chart audits excluded only patients with esophagus, liver, or pancreatic cancer; enrollment in hospice; or estimated LE less than 6 months from the CRC screening measure.¹⁸

Individualized CRC screening decisions might be facilitated by incorporating explicit consideration of comorbidity in screening guidelines, the development of tools to estimate LE and incorporate this information in screening decisions, and inclusion of LE and comorbidity in clinical reminder systems. Our objective was to develop a decision support system (DSS) for using comorbidity and frailty information to individualize screening recommendations.

METHODS

Our DSS incorporated the “payoff time” framework: the minimum time until the benefits attributable to a guideline exceed its harms.¹⁹ We estimated the payoff time for CRC screening based on individual characteristics. We then integrated this calculation into an EMR and pilot-tested this DSS.

Applying the Payoff Time to CRC Screening Decisions

We used previously published methods to estimate the mortality payoff for CRC screening.^19,20 This requires estimating LE for a particular patient, then estimating the payoff time for CRC screening, adjusting that payoff time (based on comorbidities, frailty, and other factors that influence the balance of benefits and harms), and then comparing the adjusted payoff time to the patient’s LE.

Four-year mortality was estimated from a prognostic index developed in a population-based sample of community-dwelling US adults older than 50 years.²¹ The index incorporates both comorbid conditions and measures of frailty, and has good discrimination (C statistic = 0.84, derivation cohort; 0.82, validation cohort).

To adapt the prognostic index’s estimates of 4-year mortality to estimates of LE, we used the declining exponential approximation of life expectancy. Payoff time is calculated based on estimated age- and sex-specific CRC incidence and death rates from Surveillance, Epidemiology, and End Results data; the estimated risk reduction for CRC mortality based on data sources used by the USPSTF (assuming benefit from screening begins after 5 years); and estimates of adverse events following colonoscopy.^19,20 Next, the payoff time is adjusted based on 18 patient characteristics that may affect the risk for CRC death or colonoscopy-induced death, including smoking, use of anticoagulants, American Society of Anesthesiologists class, and family history of CRC. The individual’s risk for benefit (based on the risk of CRC death) is divided by the individual’s risk for harm (based on the risk of death from colonoscopy-related complications); this yields an individualized risk-to-benefit adjustor. The adjusted payoff time is then compared with the patient’s estimated LE to yield a recommendation to pursue or avoid CRC screening.²⁰

Incorporating the Payoff Time Within a Clinical DSS

We adapted the VA clinical reminder system so that CRC screening recommendations were tailored to individual patients’ likelihood of benefit.

Data to estimate LE and payoff time were obtained from CPRS and supplemented by patients’ self-report. Medical comorbidities, medication use, smoking and alcohol status, body mass index, and demographic information were extracted from CPRS (see eAppendix, available at www.ajmc.com). Data regarding frailty and family history of CRC were obtained by having patients complete a 7-question survey prior to seeing their primary care provider (PCP). Responses were input into CPRS by the research assistant prior to calculation of the payoff time; survey completion and entry generally took about 2 or 3 minutes.

The DSS “turned off” the usual CRC screening reminder when LE was less than the payoff time (see Figures 1a and 1b), modifying the VA reminder algorithm by: a) basing the recommendation to screen (or not) on the payoff time, rather than a blanket “screen” recommendation for patients in the aged 50 to 75 years group, and b) adding an option for resolving the reminder by selecting “Due to comorbidities and/or frailty, colorectal cancer screening is no longer indicated.”

A note explaining the rationale for the screening recommendation was also created in CPRS (see Figure 1c). An explanatory document with full details of the calculations was available to PCPs, but not made a part of the EMR. The final decision regarding screening remained with the PCP and patient (Figure 2). PCPs could screen for CRC using colonoscopy, sigmoidoscopy, or FOBT.

Pilot-Testing the Clinical DSS

This study was approved by the Institutional Review Board at the VA Connecticut Healthcare System (VACHS). PCPs and patients consented to participate.

Primary Care Providers

We enrolled staff PCPs at VACHS in West Haven, Connecticut. PCP focus groups indicated a desire for information regarding LE to guide screening decisions; that the intervention be integrated into CPRS and not interrupt clinical work flow; and that patient-specific information regarding the details of the payoff-time calculation be available.

We surveyed PCPs regarding the existing CRC clinical reminder before the study, and at the conclusion of the study surveyed those PCPs whose patients were enrolled in the intervention phase.

Patients

We enrolled a convenience sample of patients from primary care clinics at VACHS in West Haven, CT; patients had clinic appointments that coincided with research assistant availability. Inclusion criteria were being aged 50 to 75 years, being “due” for CRC screening, and having at least 2 comorbid conditions from the mortality index: diabetes, cancer, chronic lung disease, congestive heart failure, or current tobacco use. Exclusion criteria were first visit to the primary care clinic, on panel of a medical resident, and impairment precluding informed consent. Patients undergoing surveillance for adenomatous polyps were not included.

Run-in Phase

The payoff time and clinical reminder intervention was inactive during the run-in phase of the study, during which the decision support parameters were validated, and active during the subsequent intervention phase. Validation involved confirming that all data were accurately extracted from CPRS and the patient survey, and that the decision support algorithm was able to accurately generate LE and payoff time data.

RESULTS

Primary Care Providers

Of 28 PCPs invited to complete our pre-intervention survey, 22 did so. The mean age was 45.5 years (SD = 8.4) with 17 females (77%) and an average of 13.9 years (SD = 7.7) in practice; 20 of 22 were physicians. All indicated that they consider comorbidities in deciding whether to recommend CRC screening. Nonparticipating PCPs mostly had minimal clinical duties or were planning to leave employment in the clinic soon.

PCPs indicated moderate satisfaction (54.5%) with clinical reminders and high ratings of effectiveness (72.7%); 77.3% indicated “The colorectal cancer clinical reminder helps me to screen patients for colorectal cancer.” However, 91% agreed or strongly agreed that “The colorectal cancer clinical reminder sometimes prompts me to screen patients for colorectal cancer who are too sick or frail to benefit from screening.” While 81.8% agreed or strongly agreed that “I feel comfortable determining which patients are appropriate for colorectal cancer screening,” only 31.8% indicated that they “feel confident estimating patient life expectancy to determine appropriateness for colorectal cancer screening.” Most (91%) indicated, “I would like the colorectal cancer clinical reminder to allow me to exclude patients with limited life expectancy” (Table 1).

PCPs showed engagement with the intervention. The note containing the rationale for the screening recommendation was always signed by the PCP. PCPs often commented on the rationale for screening or not screening in their notes, or selected “due to comorbidities and/or frailty, colorectal cancer screening is no longer indicated” in the tailored CRC screening clinical reminder.

Six of the 22 PCPs had subjects enrolled in the intervention phase of the study. These PCPs were surveyed regarding their experience with the intervention. Five of the 6 PCPs reported that the intervention did not interfere with work flow and 4 agreed that the intervention provided “useful information about whether patients would benefit from colorectal cancer screening.” All stated, “I would like to have more information about which patients would benefit from colorectal cancer screening, of the type provided by the CRC payoff time study,” and 5 agreed that “The VA should apply the CRC payoff time study algorithm to all patients potentially eligible for colorectal cancer screening” (Table 2). PCPs largely used the clinical decision support tool to formulate their screening recommendation, which they then presented to the patient; patients were largely unaware of the detailed calculations regarding payoff time and LE. No patients reported discomfort or confusion regarding the recommendations, either to their PCPs or the research team.

Patients

Fourteen subjects enrolled in the control phase of the study, during which the intervention was inactive. Ten subjects, all men, were successfully enrolled in the intervention phase of the study; they were aged 52 to 72 years. Subjects had to have at least 2 of the 5 comorbidities (ie, diabetes, cancer, congestive heart failure, chronic lung disease, and smoking) used in the calculation of LE, some did have more than 2. Subjects reported difficulty with, on average, 1.9 of the 4 frailty items (ie, difficulty with bathing/showering, managing money, walking, and pushing/pulling large objects).

DISCUSSION

For 6 intervention subjects, the clinical DSS recommended CRC screening; 3 had screening ordered and 3 refused to be screened. For 4 intervention subjects, the clinical DSS recommended not to screen (“The risks may outweigh the benefits of colorectal cancer screening”); the PCP ordered screening for 1 of these patients and did not request screening for 3 of these patients.We pilot-tested the feasibility of using the payoff time framework to individualize CRC screening recommendations in a clinical DSS. We integrated calculation of the payoff time for individual patients in CPRS and adapted the VA clinical reminder system to tailor CRC recommendations to individual patients’ likelihood of benefit.

PCPs showed engagement with the intervention and indicated that they would want the payoff time information to be available when making decisions regarding CRC screening in the future.

Existing clinical reminder systems promote CRC screening regardless of comorbidity, and do not provide an effective means to “opt out” of screening. There has been little evaluation of strategies to reduce overuse of CRC screening; our study adds to this emerging literature on personalization of decision support.²²

Limitations

The limitations of this pilot study include its small size and the fact that it was not designed to determine the effect of the payoff time intervention on screening behavior; this effect would be important to characterize in future studies. PCPs at an academic VA facility may not be representative of PCPs in the general community, and may be more likely to accept computerized decision support. In addition, use of the payoff time should increase confidence regarding the benefit (or lack thereof) of screening; however, estimates of payoff time and life expectancy are indeed estimates and are therefore imprecise when applied to individual patients. If a particular patient has a strong preference to receive screening even though it decreases his life expectancy, it could be considered in the context of shared decision making.

CONCLUSIONS

The payoff time framework can be used as part of a DSS to inform screening decisions. Future studies should evaluate the DSS in a larger sample with sufficient power to detect whether it influences screening decisions and decision quality. Author Affiliations: Yale University School of Medicine (ARS), New Haven, CT; VA Connecticut Healthcare System (ARS, FLL), West Haven, CT; Department of Veterans Affairs (JAO), VISN 1, Providence, RI; New York University School of Medicine (RSB), New York, NY.

Source of Funding: Robert Wood Johnson Foundation, Grant Number 66922.

Author Disclosures: The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (RSB); acquisition of data (ARS, JAO, FLL); analysis and interpretation of data (ARS, FLL, RSB); drafting of the manuscript (ARS, RSB); critical revision of the manuscript for important intellectual content (ARS, JAO, RSB); obtaining funding (RSB); administrative, technical, or logistic support (JAO, FLL); and supervision (RSB).

REFERENCES

Address correspondence to: Amy R. Schwartz, MD, Assistant Professor, Yale University School of Medicine, VA Connecticut Healthcare System, 950 Campbell Ave, 11ACSL, West Haven, CT 06516. E-mail: amy.schwartz2@va.gov.

1. American Cancer Society. Cancer facts and figures 2011. Atlanta, GA: American Cancer Society; 2011.

2. Stryker SJ, Wolff BG, Culp CE, Libbe SD, Ilstrup DM, MacCarty RL. Natural history of untreated colonic polyps. Gastroenterology. 1987;93(5):1009-1013.

3. Hewitson P, Glasziou P, Irwig L, Towler B, Watson E. Screening for colorectal cancer using the faecal occult blood test, Hemoccult. Cochrane Database Syst Rev. 2007;1:CD001216. doi:10.1002/14651858.CD001216.pub2.

4. Atkin WS, Edwards R, Kralj-Hans I, et al; UK Flexible Sigmoidoscopy Trial Investigators. Once-only flexible sigmoidoscopy screening in prevention of colorectal cancer: a multicentre randomized controlled trial. Lancet. 2010;375(9726):1624-1633.

5. Baxter NN, Goldwasser MA, Paszat LF, Saskin R, Urbach DR, Rabeneck L. Association of colonoscopy and death from colorectal cancer. Ann Intern Med. 2009;150(1):1-8.

6. Brenner H, Chang-Claude J, Seiler CM, Rickert A, Hoffmeister M. Protection from colorectal cancer after colonoscopy: a population case-control study. Ann Intern Med. 2011;154(1):22-30.

7. Schoen RE, Pinsky PF, Weissfeld JL, et al; PLCO Project Team. Colorectal-cancer incidence and mortality with screening flexible sigmoidoscopy. N Engl J Med. 2012;366(25):2345-2357.

8. Zauber AG, Winawer SJ, O’Brien MJ, et al. Colonoscopic polypectomy and long-term prevention of colorectal-cancer deaths. N Engl J Med. 2012;366(8):687-696.

9. Whitlock EP, Lin JS, Liles E, Beil TL, Fu R. Screening for colorectal cancer: a targeted, updated systematic review for the U.S. Preventive Services Task Force. Ann Intern Med. 2008;149(9):638-658.

10. Gatto NM, Frucht H, Sundararajan V, Jacobson JS, Grann VR, Neugut AI. Risk of perforation after colonoscopy and sigmoidoscopy: a population-based study. J Natl Cancer Inst. 2003;95(3):230-236.

11. U.S. Preventive Services Task Force. Screening for colorectal cancer: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med. 2008;149(9):627-637.

12. 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, 2008: a joint guideline from the American Cancer Society, the US Multi-Society Task Force on Colorectal Cancer, and the American College of Radiology. CA Cancer J Clin. 2008;58(3):130-160.

13. Department of Veterans Affairs. Colorectal cancer screening. VHA Directive 1015. http://www.va.gov/vhapublications/ViewPublication.asp?pub_ID=3068. Published December 30, 2014. Accessed July 15, 2015.

14. Fisher DA, Galanko J, Dudley TK, Shaheen NJ. Impact of comorbidity on colorectal cancer screening in the Veterans Healthcare System. Clin Gastroenterol Hepatol. 2007:5(8):991-996.

15. Walter LC, Lindquist K, Nugent S, et al. Impact of age and comorbidity on colorectal cancer screening among older veterans. Ann Intern Med. 2009;150(7):465-473.

16. Goodwin JS, Singh A, Reddy N, Riall TS, Kuo YF. Overuse of screening colonoscopy in the Medicare population. Arch Intern Med. 2011;171(15):1335-1343.

17. Holden DJ, Jonas DE, Porterfield DS, Reuland D, Harris R. Systematic review: enhancing the use and quality of colorectal cancer screening. Ann Intern Med. 2010;152(10):668-676.

18. Veterans Health Administration Office of Analytics and Business Intelligence. EPRP Technical Manual. FY2012 Quarter 4. http://vaww.oqp.med.va.gov/programs/pm/pmPolicyTechManual.aspx. Accessed July 19, 2012.

19. Braithwaite RS, Concato J, Chang CC, Roberts MS, Justice AC. A framework for tailoring clinical guidelines to comorbidity at the point of care. Arch Intern Med. 2007;167(21):2361-2365.

20. Braithwaite RS, Fiellin D, Justice AC. The payoff time: a flexible framework to help clinicians decide when patients with comorbid disease are not likely to benefit from practice guidelines. Med Care. 2009;47(6):610-617.

21. Lee SJ, Lindquist K, Segal MR, Covinsky KE. Development and validation of a prognostic index for 4-year mortality in older adults. JAMA. 2006;295(7):801-808.

22. Saini SD, van Hees F, Vijan S. Smarter screening for cancer: possibilities and challenges of personalization. JAMA. 2014;312(21):2211-2212. 

Download PDF: Pilot of Decision Support to Individualize Colorectal Cancer Screening Recommendations