Long-term Outcomes Following Positive Fecal Occult Blood Test Results in Older Adults: Benefits and Burdens

Background In the United States, older adults have low rates of follow-up colonoscopy after a positive fecal occult blood test (FOBT) result. The long-term outcomes of these real world practices and their associated benefits and burdens are unknown.

Methods Longitudinal cohort study of 212 patients 70 years or older with a positive FOBT result at 4 Veteran Affairs (VA) facilities in 2001 and followed up through 2008. We determined the frequency of downstream outcomes during the 7 years of follow-up, including procedures, colonoscopic findings, outcomes of treatment, complications, and mortality based on chart review and national VA and Medicare data. Net burden or benefit from screening and follow-up was determined according to each patient's life expectancy. Life expectancy was classified into 3 categories: best (age, 70-79 years and Charlson-Deyo comorbidity index [CCI], 0), average, and worst (age, 70-84 years and CCI, ≥4 or age, ≥85 years and CCI, ≥1).

Results Fifty-six percent of patients received follow-up colonoscopy (118 of 212), which found 34 significant adenomas and 6 cancers. Ten percent experienced complications from colonoscopy or cancer treatment (12 of 118). Forty-six percent of those without follow-up colonoscopy died of other causes within 5 years of FOBT (43 of 94), while 3 died of colorectal cancer within 5 years. Eighty-seven percent of patients with worst life expectancy experienced a net burden from screening (26 of 30) as did 70% with average life expectancy (92 of 131) and 65% with best life expectancy (35 of 51) (P = .048 for trend).

Conclusions Over a 7-year period, older adults with best life expectancy were less likely to experience a net burden from current screening and follow-up practices than are those with worst life expectancy. The net burden could be decreased by better targeting FOBT screening and follow-up to healthy older adults.

In many real-world settings, fewer than 60% of patients receive a colonoscopy within 1 year of a positive FOBT result.^1-4 A recent study found that many older patients without follow-up are in poor health or decline follow-up, which suggests that they should not have been screened in the first place.⁵ In addition, rates of follow-up have shown minimal improvement over the last decade.^3,4,6,7 Despite persistently low rates of follow-up colonoscopy in older adults with positive FOBT results, the long-term outcomes of these real-world screening and follow-up practices have not been described.^3,4,8

The benefit of finding asymptomatic cancer or precancerous polyps that would have caused symptoms years later must be weighed against immediate burdens of follow-up procedures and treatments stemming from a positive screening result.⁹ Randomized trials of FOBT suggest that a person should have a life expectancy of at least 5 years to derive survival benefit from screening; otherwise they are only subject to the potential burdens.^9,10 We are not aware of any studies of real-world practices that observe older patients for more than 3 years after a positive FOBT result to determine whether patients lived long enough to potentially benefit from detecting large adenomas or early stage colorectal cancer, or whether they only experienced burdens from follow-up proceduresand treatments (eg, false-positive results, repeated testing, complications).^3,4 Such information about the benefits and burdens experienced by older adults with positive FOBT results would help to guide individualized screening and follow-up decisions in clinical practice.

We describe 7-year outcomes following a positive FOBT result in older adults in real-world clinical practice. We took a novel approach of combining VA and Medicare claims and medical chart review to follow the downstream outcomes after a positive FOBT result among patients 70 years or older screened at 4 geographically diverse VA facilities. Downstream outcomes included follow-up testing, polypectomies, cancer diagnoses and treatments, procedural complications, and overall mortality. Net benefit or burden resulting from these screening and follow-up practices was determined according to each patient's predicted life expectancy.¹¹ The goal was to inform how clinical practice could improve to maximize the net benefit of FOBT screening and follow-up in older adults.

We conducted a longitudinal cohort study of 212 adults aged 70 years or older who had a positive FOBT result during 2001 at 4 VA facilities (Minneapolis, Minnesota; Durham, North Carolina; Portland, Oregon; and West Los Angeles, California) and observed them for 7 years to determine downstream outcomes. We chose to focus on positive FOBT results as the FOBT is the most common colorectal cancer screening test within the VA.¹² To identify our cohort, we used outpatient claims from the VA National Patient Care Database to identify all patients 70 years or older who had a FOBT between January 1, 2001, and December 31, 2001, at the 4 facilities and met our inclusion and exclusion criteria (n = 2410).⁵ All patients had continuous enrollment in Medicare Parts A and B and fee-for-service coverage from January 1, 2000, through December 31, 2002. Patients were excluded if they had a history of colorectal cancer or polyps, inflammatory bowel disease, colectomy, or colostomy, or were not due for screening.⁵ We used claims from 6 months before their FOBTs and medical chart review to exclude patients with signs or symptoms that would justify performance of FOBT for nonscreening purposes (eg, iron-deficiency anemia, gastrointestinal bleeding, abdominal pain, change in bowel habits, unexplained weight loss).⁵ Two hundred twelve of 2410 patients had positive FOBT results (9%), which were extracted from the Veterans Health Information Systems and Technology Architecture (VISTA) laboratory package. If any FOBT cards indicated a positive finding for occult blood, the FOBT result was considered positive.

We used National VA Data Systems as well as inpatient and outpatient Medicare claims through December 31, 2002, to capture follow-up testing within 1 year of a positive FOBT result inside and outside the VA system.⁵ Patient age was determined on the date of the FOBT. Comorbidity was measured using the Deyo adaptation of the Charlson Comorbidity Index (CCI), derived from administrative data.^13,14 Charlson-Deyo scores were calculated from VA and Medicare inpatient and outpatient claims during the 12 months before the date of the FOBT. Race and sex were obtained from both VA and Medicare data.

Next, 3 of us (C.E.K., M.A.C., and L.C.W.) reviewed VA medical records to determine the long-term outcomes of screening and follow-up through December 31, 2008. The VA computerized medical records system contains notes from all inpatient and outpatient visits. These notes contain information about follow-up that occurs outside the VA per patient report to their VA clinician.¹⁵

First, we categorized patients as having received follow-up colonoscopy related to their positive FOBT result or not. Patients who did not receive colonoscopy over the 7 years or had a colonoscopy for gastrointestinal (GI) tract bleeding or other symptoms unrelated to their positive FOBT result were categorized with the “no follow-up colonoscopy” group. Patients with follow-up colonoscopy were categorized according to whether they were diagnosed as having cancer, significant adenoma, or nonsignificant or normal findings. Significant adenoma was defined as 1 or more large adenomas (≥1 cm), 3 or more small adenomas (<1 cm), or any adenoma with villous pathologic characteristics based on standard GI guidelines.¹⁶ Nonsignificant or normal findings included findings of diverticula, fewer than 3 small adenomas (<1 cm), angiodysplasia, hyperplastic polyps, and normal mucosa.

Next, we determined downstream outcomes over 7 years in those with and without follow-up colonoscopy. Outcomes included cancer diagnoses, complications from treatment, frequency of follow-up testing, and 5-year mortality. Cancer diagnoses included cancer stage and treatment. Complications included those documented in the medical chart as being related to colonoscopy or cancer treatment (eg, pain and/or discomfort, gastrointestinal bleeding, infection, death). Frequency of follow-up testing included the number of repeated FOBTs, sigmoidoscopies, barium enemas, and/or colonoscopies patients received during the 7 years. Five-year mortality was determined from the VA Vital Status File.¹⁷

Next, 2 of us (C.E.K. and L.C.W.) reviewed each patient's outcomes to determine whether they experienced net benefit or net burden from screening and follow-up. Patients were considered to have experienced net benefit if they had a significant adenoma and/or colorectal cancer found on follow-up and lived at least 5 years, even if they had complications or repeated procedures.^18,19 Net benefit or burden was indeterminate for (1) patients with 1 or 2 small adenomas who lived at least 5 years because some clinicians believe that this group potentially benefitted from screening^20-22; or (2) patients who had a colonoscopy outside of the VA without an available pathology report and lived at least 5 years. All other patients, including those with normal findings on follow-up colonoscopy (ie, false-positive FOBT results), those without follow-up colonoscopy (ie, had a positive FOBT result that was not worked up), and those who died within 5 years of their FOBT (ie, subjected to tests for an asymptomatic disease that would never have affected them), were defined as experiencing net burden. While net burden may be small in some cases (eg, those without follow-up who did not develop cancer), some older adults experience embarrassment and discomfort from performing FOBTs,^23,24 and a positive test result may be anxiety provoking.²⁵

To determine downstream outcomes, we observed patients from the date of their positive FOBT result until death or the end of the study period (December 31, 2008). To examine the association between life expectancy and net benefit from screening, we stratified patients into 3 life expectancy subgroups a priori: best—the youngest and healthiest patients (age 70-79 years; CCI, 0) who were expected to live more than 10 years; average—the younger patients with increasing comorbidity and the oldest healthiest patients (age 70-84 years; CCI, 1-3 or age, ≥80 years; CCI, 0) who were expected to live 5 to 10 years; and worst—the sickest and oldest patients (age 70-84 years; CCI, ≥4 or age ≥85 years; CCI, ≥1) who were expected to live less than 5 years.¹³ Differences between patients receiving net burden and net benefit were determined according to worsening life expectancy using the Cochran-Armitage χ² test for trend. We used SAS software, version 9.1 (SAS Institute, Cary, North Carolina) and and Stata/SE, version 10.0 (StataCorp, College Station, Texas) for all analyses.

The committee on human research at the University of California, San Francisco; the committee for research and development at the San Francisco VA Medical Center; and the institutional review board at the Minneapolis VA Medical Center approved this study.

The mean age of the 212 patients was 76.4 years (age range, 70-89 years). Consistent with the elderly veteran population, 99.5% were men (211 of 212), and 84.9% were white (180 of 212) (Table 1). Sixty-six of 212 patients died within 5 years of their FOBT result (31.1%). Five-year mortality was 5.9% for patients with best life expectancy (3 of 51); 38.2% for patients with average life expectancy (50 of 131), and 46.7% for patients with worst life expectancy (14 of 30), which equates to a life expectancy of 5.4 years.

Among the 118 patients who had follow-up colonoscopy over the 7-year period, 6 patients had cancer (5.1%), 34 had significant adenoma (29%), and 70 had neither cancer nor significant adenoma (59%) (Figure 1). One patient with a normal colonoscopy finding developed an interval colorectal cancer 5 years later that was successfully treated, but he died a year later of congestive heart failure. Eight patients had colonoscopies outside the VA without an available pathology report, so it is unclear if they had a significant adenoma. No subsequent notes indicated that any of these patients were ever diagnosed as having colorectal cancer. Seventeen percent of these patients died within 5 years of causes other than colorectal cancer (20 of 118). We also assessed complications of follow-up. Ten percent of patients who underwent colonoscopy or cancer treatment experienced complications (12 of 118), as illustrated in Figure 1. Three of the 6 patients with cancer had complications from surgical treatment of their cancer (Table 2), although all eventually recovered and survived more than 5 years after their positive FOBT result. Finally, 29% of patients with significant adenomas had 3 or more follow-up tests over the 7 years (10 of 34), compared with 20% of patients with nonsignificant or normal colonoscopy results (14 of 70).

Among the 44.3% of patients who had no follow-up colonoscopy over the 7-year study period (94 of 212) (Figure 1), 6% were eventually diagnosed as having colorectal cancer (6 of 94), of whom 4 died within a few months of treatment (Table 2). A total of 46% of patients who did not get a follow-up colonoscopy died within 5 years of causes other than colorectal cancer (43 of 94). A total of 57% of patients underwent some form of follow-up other than colonoscopy (54 of 94), such as repeat FOBT or sigmoidoscopy, and 59% of those patients had more than 1 follow-up test over the study period (32 of 54) but never a colonoscopy.

A total of 15.6% of patients were diagnosed as having cancer or significant adenomas on follow-up colonoscopy (33 of 212) and lived at least 5 years and were defined as receiving net benefit (Table 3). Since it is controversial whether the 22 patients with nonsignificant adenomas (<1 cm) who lived at least 5 years benefitted from screening, we categorized them as receiving indeterminate benefit, as we categorized the 8 patients for whom we did not have pathology reports from their colonoscopy. The remaining 149 patients were defined as receiving net burden, including 10 patients who had adenomas removed but died within 5 years; 45 patients with no adenomas of any kind found on colonoscopy (ie, unequivocal false positives); and 94 patients who did not have a follow-up colonoscopy. The magnitude of net burden varied, ranging from false-positive results in patients without cancer to cancers not found because patients declined follow-up (6.4%).

Also, the net burden of these screening and follow-up practices varied across the 3 prognostic groups (Figure 2). A total of 87% of those with worst life expectancy experienced net burden compared with 70% of those with average life expectancy and 65% of those with best life expectancy (P = .048 for trend). Conversely, 20% of patients with best life expectancy experienced net benefit (10 of 51) vs 15.3% of those with average life expectancy (20 of 131) and 10% of those with worst life expectancy (3 of 30) (P = .25 for trend.).

Even over 7 years after a positive FOBT result, only a little over half of older adults received follow-up colonoscopy. Among those who received follow-up colonoscopy, more than a quarter had significant adenomas or cancer detected and treated and lived more than 5 years, potentially benefitting from screening, while approximately 59% had no significant findings on follow-up (70 of 118), and 10% experienced complications from colonoscopy or cancer treatment. Among patients who did not receive follow-up colonoscopy, nearly half died of other causes within 5 years, while 3% died of colorectal cancer. Patients with worst life expectancy (5.4 years) were more likely to experience net burden than were patients with average or best life expectancy. As current guidelines from the US Preventive Services Task Force²⁶ and other groups encourage individualized decision making in patients older than 75 years,¹⁶ our study provides data about the consequences following the choice to pursue or forego follow-up of a positive FOBT result.

The low rate of follow-up colonoscopy found in this study (56%) is similar to that of other studies in older adults, even though the follow-up period in other studies ranged from only a few months to 3 years, whereas we extended the follow-up period to 7 years.^1-4,7,8 Patients may not get follow-up colonoscopy because of the risks of colonoscopy and treatment and other competing causes of mortality, as perceived either by the physician²⁷ or the patient.⁵ We found that nearly half of those without follow-up colonoscopy died of causes other than colorectal cancer within 5 years, which suggests that the decision to forego follow-up colonoscopy was appropriate for those patients. It also suggests that such decisions are occurring after FOBT results are known rather than following recommendations to avoid the FOBT screening if there is no intention to follow up a positive result with colonoscopy. As the use of screening colonoscopy increases, colonoscopy decisions will be made up front and will need to be better targeted than the FOBT, or the number of colonoscopies performed in people who die within 5 years will increase.

Among our patients with follow-up colonoscopy, about a third had colorectal cancer or significant adenomas, on par with other studies.^10,28,29 Strul et al³⁰ found the overall adenoma rate in older adults to be 26%. Our study indicates that older patients without follow-up had a 3% risk of dying of colorectal cancer in the next 5 years. On the other hand, undergoing follow-up procedures is not without risk. While prior studies have suggested minimal complications from colonoscopy in older adults, these studies only observed patients over a short period.^4,31-35 We found complications of follow-up encompassed more than immediate events. In our cohort, 4 of the 40 older patients with cancer or significant adenomas were hospitalized for several weeks as a result of complications from treatment. In addition, 23% of those with follow-up colonoscopy (24 of 104), excluding those with cancer, had 3 or more additional tests over 7 years. However, while older adults are often at greatest risk for complications from colonoscopy and colorectal cancer treatment owing to their increasing comorbidity,³⁶ they may be most likely to benefit from screening, especially if they have a substantial life expectancy.

While our study is not a randomized trial, the lengthy follow-up allows us to identify patients who most likely received net benefit from real-world screening and follow-up practices, ie, those with significant disease treated as a result of screening and who lived more than 5 years. The choice of a 5-year survival time to achieve benefit is based on the natural history of polyps¹⁸ and on the findings of randomized trials that show survival benefit beginning around 5 years after the start of FOBT screening.¹⁰ In addition, several of the cancers and adenomas found in our study were large (>3 cm) such that it is reasonable to expect that they would have caused symptoms within 5 years. As with all screening tests, FOBT does not benefit most patients because most do not have cancer or significant adenomas. However, more than 15.6% of our patients aged 70 years or older had colorectal cancer or significant adenomas successfully treated and lived more than 5 years (33 of 212), which suggests that a significant minority received net benefit from current practices.

It makes intuitive sense that patients with the best life expectancy are more likely to benefit from screening than those with the worst. This has been shown in cost-effectiveness analyses,^37,38 which are based on numerous methodologic assumptions, whereas our study uses real- world data to describe outcomes of screening according to life expectancy. We found that older patients with the best predicted life expectancy were less likely to experience net burden from screening than those with the worst. Our study supports guidelines that recommend using life expectancy to guide colorectal cancer screening decisions in older adults and argues against 1-size-fits-all interventions that simply aim to increase overall screening and follow-up rates.^16,26 We used the well-validated Charlson-Deyo comorbidity index because it strongly predicts long-term mortality.¹⁴ In our cohort, it effectively stratified patients into groups with widely differing 5-year mortality rates, ranging from 6% for patients with best life expectancy to 47% for those with worst (life expectancy, 5.4 years) for whom most would not recommend screening.³⁶ However, more comprehensive prognostic indices (eg, incorporating functional status) over more than 5 years are needed to better guide physicians as they target screening and follow-up to those older adults with substantial life expectancies.^39,40

This study has several limitations. First, our cohort is primarily composed of men who use VA medical centers, so the generalizability of our findings to nonwhite men, women, and persons who do not use VA medical centers is uncertain. However, the VA medical system is the largest health care system in the United States, so outcomes of screening and follow-up in this system are likely to have generalizable lessons for US health care. Second, our sample size was small because only 212 patients had positive FOBT results at the 4 participating facilities in 2001 (9%). Third, although we used Medicare claims data for the first year of follow-up, we relied on medical chart review for later outcomes, which might have missed some complications or tests that occurred outside the VA health system, and we were unable to find pathology results for 8 patients with a colonoscopy outside the VA system, although all 8 remained alive without colorectal cancer. Fourth, we defined patients as having benefitted from screening if they had colorectal cancer or significant adenomas detected on follow-up colonoscopy and were treated and survived more than 5 years. Others may argue for a longer or shorter survival length to define benefit in older adults.^10,41

In conclusion, systematic reviews of colorectal cancer screening have called for more studies to assess the net benefit of real-world colorectal cancer screening practices to improve appropriate use and minimize burdens of screening.³⁹ Our study used a novel method of observing patients with a positive FOBT result for 7 years to determine the net benefit and burden of real-world screening and follow-up practices in older adults. We demonstrated that older adults with substantial life expectancies were less likely to experience net burden than those with limited life expectancies. Therefore, through individualized decision making, the percentage of patients experiencing net burden could be decreased by better targeting FOBT screening and follow-up to healthy older adults.

Correspondence: Christine E. Kistler, MD, MASc, Department of Family Medicine, University of North Carolina at Chapel Hill, 590 Manning Dr, CB 7595, Chapel Hill, NC 27599-7595 (Christine_Kistler@med.unc.edu).

Accepted for Publication: March 3, 2011.

Published Online: May 9, 2011. doi:10.1001/archinternmed.2011.206

Author Contributions: Drs Kistler and Walter had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Walter. Acquisition of data: Kistler, Lee, and Walter. Analysis and interpretation of data: Kistler, Kirby, Lee, Casadei, and Walter. Drafting of the manuscript: Kistler and Walter. Critical revision of the manuscript for important intellectual content: Kirby, Lee, Casadei, and Walter. Statistical analysis: Kistler and Kirby. Obtained funding: Walter. Administrative, technical, and material support: Casadei.

Funding/Support: Work for this research was supported by the VA Quality Scholars Program, National Institute on Aging grant T32 AG000212-16, Agency for Healthcare Research and Quality Mentored Career Development Program in Comparative Effectiveness Development grant K12 HS19468-01, the Cecil G. Sheps Center for Health Services Research, and the Lineberger Cancer Center at the University of North Carolina at Chapel Hill (Dr Kistler); and by VA Health Services Research and Development grant IIR-04-427 and National Cancer Institute grant 1R01CA134425 (Dr Walter).

Role of the Sponsors: The funding institutions had no role in the design, conduct, or analysis of this study or in the decision to submit the manuscript for publication.