1476

Factors Associated with the Use of FlexibleSigmoidoscopy as a Screening Test for the Detectionof Colorectal Carcinoma by Primary Care Physicians

BACKGROUND. Despite current recommendations of flexible sigmoidoscopy as aGregory S. Cooper, M.D.

screening test for the detection of colorectal carcinoma, relatively few asymptom-Richard H. Fortinsky, Ph.D.atic patients undergo this procedure. To enhance the use of sigmoidoscopy, differ-Ronald Hapke, M.D.ences in the use of screening, as well as barriers to screening among specificC. Seth Landefeld, M.D.physician groups, should be defined.

METHODS. The authors surveyed 1762 practicing primary care physicians to deter-Department of Medicine, University Hospitals ofmine their self-reported ability to perform sigmoidoscopy and perceived obstaclesCleveland and Cleveland VAMC, Case Westernto either initiating or enhancing screening.Reserve University, Cleveland, Ohio.RESULTS. A total of 884 physicians (50%) responded. Ninety percent of primary

care physicians reported that they offered sigmoidoscopic screening to their pa-

tients, with 46% referring patients and 44% performing the procedure themselves.

Physician characteristics were not associated with the overall use of sigmoidoscopy.

In contrast, compared with physicians who referred patients for the procedure,

physicians who performed sigmoidoscopy themselves were more often board certi-

fied, male, and graduated from medical school after 1970 (Põ 0.001). In a multivari-

ate analysis, these characteristics were also independently associated with the

ability to perform sigmoidoscopy. The barrier to sigmoidoscopy cited most often

was poor patient acceptance, whether or not the physician performed or referred

patients for sigmoidoscopic screening. Other barriers cited were lack of training,

lack of equipment, and time required, each of which was identified most often by

physicians who did not screen at all.

CONCLUSIONS. Most physicians surveyed reported using sigmoidoscopic screeningPresented in part at the American Society of to some degree in their practice, although many did not perform the procedurePreventive Oncology, New Orleans, Louisiana, themselves. Population-based interventions to increase screening may benefitMarch 24, 1997.

from targeting specific physician subgroups and attempting to improve patient

acceptance of the procedure. Cancer 1998;82:1476–81.Supported by an American Cancer Society Insti-q 1998 American Cancer Society.tutional Research Grant to the Case Western

Reserve University Cancer Center.

KEYWORDS: colorectal neoplasms, sigmoidoscopy, physicians’ practice patterns,Dr. Cooper is the recipient of a Clinical Research patient acceptance of healthcare, public health.Training Grant for Junior Faculty from the Amer-ican Cancer Society, and Dr. Landefeld is a Se-

Cnior Research Associate, Health Services Re- olorectal carcinoma is the second most common fatal malignancyin the U.S., with approximately 55,000 deaths expected in 1997.1search and Development Service, Department

of Veterans Affairs. To decrease this burden, the screening of patients older than 50 yearswith flexible sigmoidoscopy has been recommended to detect early

Address for reprints: Gregory S. Cooper, M.D., stage cancers and their precursor lesions, adenomatous polyps.2–5

Division of Gastroenterology, University Hospi- The benefits of sigmoidoscopy in reducing cancer mortality have beentals of Cleveland, 11100 Euclid Avenue, Cleve-

supported in two case–control studies,6,7 and randomized clinicalland, OH 44106.trials are currently ongoing.

Despite the widespread recommendations, sigmoidoscopy ap-Received May 12, 1997; revision received Octo-ber 28, 1997; accepted October 28, 1997. pears to be far underutilized relative to public health goals. For exam-

q 1998 American Cancer Society

/ 7bc0$$0731 03-24-98 08:11:27 cana W: Cancer

Associations with Sigmoidoscopic Screening/Cooper et al. 1477

ple, as recently as 1992, fewer than 10% of persons age reasons included lack of belief in its effectiveness, un-availability of equipment, lack of training in sigmoid-50 years or older reported undergoing sigmoidoscopy

within the past 3 years.8 Also, through the use of case oscopy, poor patient acceptance, time required for theprocedure, and lack of reimbursement. Physicianvignettes, we determined that less than 50% of physi-

cians routinely used sigmoidoscopic screening as part characteristics obtained from the AMA Master File in-cluded specialty, year of medical school graduation,of periodic health assessment.9 There are likely multi-

ple underlying reasons for the low screening rates, in- gender, and board certification status.All 2000 physicians were mailed the survey duringcluding patient refusal, competing demands of prac-

titioners, physician practice patterns, and disagree- the Spring of 1994; if no response was received after6 weeks, a second mailing was sent. If the physicianment by physicians with published guidelines.

Moreover, because the performance of sigmoidoscopy did not respond to either mailing, one of the studypersonnel telephoned their office to verify their cur-entails technical training in the procedure and pur-

chase of endoscopic equipment, lack of prior training rent address, assess their eligibility for study inclusion(i.e., whether they were actively practicing primaryand equipment may be important barriers for some

physicians who may otherwise perform the procedure. care), and determine their willingness to complete thesurvey. A third mailing was sent to those eligible physi-Thus, despite the goal of developing sigmoidoscopy

as a primary care screening test, many physicians may cians who were willing to complete a questionnaire.To obtain evidence about the validity of physicianhave to refer eligible patients to specialists for routine

screening. responses, 18 practitioners in a local managed careorganization were mailed a questionnaire. Their esti-We therefore conducted a survey of practicing pri-

mary care physicians, with the following goals: 1) to mate of the proportion of patients in their practicewhom they tested with sigmoidoscopy was compareddetermine practitioner characteristics associated with

self-reported ability to perform sigmoidoscopy, and 2) with the proportion of patients in their panel age ¢50years who received sigmoidoscopy within the past 3to identify perceived obstacles to either initiating or

enhancing screening. years (determined by a review of health claims). Therewas a high correlation between actual and reportedscreening practices (Spearman correlation Å 0.85; PMETHODS

Sample õ 0.0001).The survey sample was obtained from the AmericanMedical Association (AMA) Master File of Physicians, Analysis

All analyses were limited to physicians who met eligi-which includes both members and nonmembers ofthe AMA and is continuously updated. Physicians were bility criteria and returned a completed questionnaire.

Two-tailed tests of significance were conducted usingeligible for inclusion in the sample if they practicedinternal medicine or family practice and resided in 1 the chi-square test for categoric variables. The associa-

tion of physician characteristics with the use of screen-of 10 states selected to represent a spectrum with re-gard to managed care penetration, proportion of older ing sigmoidoscopy was evaluated; and among physi-

cians screening with sigmoidoscopy, practitionersindividuals, and proportion of patients without healthinsurance. The states included Alaska, Colorado, Flor- were compared according to whether they performed

the procedure themselves or referred patients else-ida, Minnesota, Mississippi, New Jersey, Ohio, RhodeIsland, Texas, and Utah. A random sample of 1000 where. The independent association of physician fac-

tors with performance of sigmoidoscopy as a screen-internists and 1000 family practitioners was selectedfrom the overall sample, using the Nth name tech- ing strategy was determined using multivariable logis-

tic regression analysis.nique.10

Physicians were also divided into three groups:those neither screening with nor referring patients forMeasures

All physicians were mailed a questionnaire that in- sigmoidoscopy, those referring patients but not per-forming the procedure themselves, and those who re-quired about their current sigmoidoscopic screening

practices as well as perceived barriers to increased use. ported performing sigmoidoscopy. Perceived barriersto initiating or increasing screening were comparedPhysicians were asked whether they currently screen

with flexible sigmoidoscopy and whether they perform among these three groups.the procedure themselves or refer to another prac-titioner. Respondents also indicated reasons for not RESULTS

Among the 2000 physicians to whom questionnairesscreening, or, if they were currently screening to somedegree, the major reasons for not screening more. The were sent in the initial mailing, a total of 238 were

/ 7bc0$$0731 03-24-98 08:11:27 cana W: Cancer

1478 CANCER April 15, 1998 / Volume 82 / Number 8

TABLE 1 0.0001), board certification (OR, 2.6; 95% CI, 1.8–3.7;Demographic and Professional Characteristics of Physicians in the P õ 0.0001), and medical school graduation after 1970Survey Sample (OR, 2.1; 95% CI, 1.5–2.9; P õ 0.0001).

The barrier to the performance of sigmoidoscopicRespondents Nonrespondentsscreening cited most frequently was poor patient ac-Characteristics (n Å 884) (n Å 878)ceptance, which was identified by more than 40% of

Medical school graduation ú1970 72% 70% physicians whether or not they performed or referredMale 85% 84% for sigmoidoscopic screening (Table 2). Other barriersBoard certifieda 76% 69%

cited by 11–16% of physicians were lack of training,Family practice 49% 49%lack of equipment, and the time required to performState of practice

Texas 20% 23% sigmoidoscopy, each of which was cited most oftenFlorida 21% 20% by physicians who did not screen at all. Lack of reim-Ohio 20% 16% bursement and questions about the effectiveness ofNew Jersey 13% 17%

sigmoidoscopic screening were cited by less than 10%Minnesota 12% 9%of physicians, although these barriers concerned phy-Colorado 6% 6%

Mississippi 3% 4% sicians who did not screen somewhat more frequentlyUtah 3% 2% than they concerned other physicians.Rhode Island 2% 2%Alaska 1% 1%

DISCUSSIONOffice-baseda 89% 84%In this large national survey of practicing primary care

a P õ 0.005. physicians, we determined reported practices and atti-tudes regarding sigmoidoscopic screening. We em-phasize four findings. First, although almost all physi-cians reported using sigmoidoscopy to some degreeexcluded because they were not engaged in any pri-

mary care practice (n Å 165), were retired from active in their practice, almost half did not perform the pro-cedure themselves and required patient referral forpractice (n Å 65), or were in subspecialty fellowship

training (n Å 8). Of the remaining 1762 physicians, screening. Second, certain physician groups, includingmen, board certified practitioners, and more recentcompleted questionnaires were received from 884, or

50%. Fifty-one percent of respondents were internists, medical school graduates, were more likely to performsigmoidoscopy themselves than other physicians.72% graduated from medical school after 1970, 85%

were male, 76% were board certified, and 89% were Third, among all physicians, poor patient acceptancewas the most frequently cited deterrent to additionalengaged in office-based practice (Table 1). Respon-

dents were more likely than nonrespondents to be screening. Fourth, lack of training and equipmentwere cited as major barriers to screening by a largeboard certified and engaged in office-based practice.

Ninety percent of physicians reported that they number of physicians who did not currently performsigmoidoscopy. The data from this survey are encour-offered sigmoidoscopic screening to their patients,

with 46% of respondents referring patients for the pro- aging in that most respondents report screening withflexible sigmoidoscopy to some degree in their prac-cedure and 44% performing it themselves. Reported

use of sigmoidoscopic screening was not associated tice. However, to utilize interventions that will en-hance screening further, knowledge of the specific bar-(Pú 0.1) with specialty, board certification status, year

of graduation, or gender. However, among physicians riers to screening in the target groups of physicians islikely needed.who reported using sigmoidoscopy, physicians who

performed the procedure themselves differed from Other investigators have established that, al-though most physicians agree with published guide-those who referred to another practitioner. Physicians

who performed sigmoidoscopy themselves were more lines about sigmoidoscopic screening,11–14 actual useof the procedure is much lower than their own ideals.often board certified (55% vs. 27% of non–board certi-

fied physicians, P õ 0.001), male (54% vs. 19% of For example, only 23% of practitioners surveyed in1989 reported meeting or exceeding the Americanwomen, P õ 0.001), medical school graduates after

1970 (53% vs. 36% of graduates before 1971, Põ 0.001), Cancer Society’s sigmoidoscopic screening guide-lines,11 and the most frequent reasons cited for notor, to a lesser degree, family practitioners (52% vs.

44% for internists, Põ 0.05). In a multivariable model, screening included procedural cost to patients, per-ceived patient discomfort and fear, and questionablethree physician characteristics were independently as-

sociated with performance of sigmoidoscopy: male sensitivity of the test. In addition, less than half ofphysicians were trained in the procedure, and onlygender (odds ratio [OR], 5.4; 95% CI, 3.2–9.1; P õ

/ 7bc0$$0731 03-24-98 08:11:27 cana W: Cancer

Associations with Sigmoidoscopic Screening/Cooper et al. 1479

TABLE 2Reasons Cited by Physicians as Barriers to Performance of Sigmoidoscopic Screening, According to Physician Screening Pattern

Physicians not Physicians referring Physicians performingBarrier to screening screening for sigmoidoscopy themselves

Poor patient acceptancea 47% 42% 54%Lack of trainingb 36% 27% 0%Excessive time requireda 28% 17% 13%Lack of equipmentb 32% 15% 1%Lack of reimbursement 10% 7% 9%Questionable effectivenessb 18% 3% 3%

a Proportion of physicians indicating this reason is different among the three groups, P õ 0.005.b Proportion of physicians is different among these groups, P õ 0.001.

35% had the necessary equipment in their offices. In of sigmoidoscopic screening, as recommended by theU.S. Public Health Service.21 However, for interven-a survey of a large cohort of physicians, internists were

less likely than family practitioners to endorse fre- tions to be effective, the predominant reasons for de-ficiencies in screening, including lack of training, pa-quent use of sigmoidoscopy, but the questionnaire did

not determine reported practices across specialties.12 tient refusal, and financial barriers, should first be de-termined. The data from this study suggest that forIn a local survey of 205 primary care physicians, only

one-third reported ‘‘regularly’’ performing sigmoido- many primary care physicians, preceptorships in tech-nical aspects of the procedure may be needed. In onescopic screening, with a perceived low yield of the

procedure cited as the most common barrier to trial that involved ‘‘hands on’’ training in sigmoidos-copy, utilization by practitioners of both this proce-screening.13 Higher reported screening rates were as-

sociated with prior training in sigmoidoscopy, internal dure and barium enemas increased significantly aftertraining.22 Moreover, promotion of sigmoidoscopicmedicine or family practice specialty, and number of

barriers to screening that were cited.13 To our knowl- training may be worthwhile for not only the twogroups least likely to perform sigmoidoscopy (olderedge, our finding of lower reported performance of

sigmoidoscopy among female primary care physicians graduates and non–board certified physicians), but aspart of internal medicine and family practice residencyhas not been noted previously, and it contrasts with

other studies that reported higher rates of Pap testing and postresidency continuing medical education. Inaddition, educational interventions,23 computerizedand mammography among female practitioners.15 If

corroborated with chart audits, determining the un- reminders,24 and the use of local opinion leaders25 and‘‘academic detailing’’26 have all been found to be effec-derlying reasons for physician gender differences in

relation to screening are worthy of future investiga- tive in enhancing cancer screening and could be tar-geted to specific groups of practitioners. Although lacktion.

Prior studies have also determined that patient of reimbursement was only cited by a small proportionof physicians, provider-based interventions, such asawareness and understanding of screening tests is an

important predictor of whether patients will receive current efforts by the Digestive Health Initiative to se-cure Medicare reimbursement for screening sigmoid-cancer screening. Differences in knowledge about can-

cer screening procedures have been reported among oscopy,27 may also facilitate procedure use. Finally,because poor patient acceptance was the most fre-racial groups and different educational levels and are

strongly associated with receipt of screening.16,17 quently cited barrier to enhanced screening, patient-based interventions may also be needed. Patient-fo-Nonetheless, in a study of patients attending a primary

care clinic, physician recommendations regarding sig- cused educational interventions have been shown tobe efficacious in increasing rates of mammographicmoidoscopy were the most important predictor of ac-

tual receipt of the procedure.18 These findings corrob- screening.28,29

Although the benefits of sigmoidoscopy in reduc-orate similar data for screening mammography.19,20

Thus, despite potential patient barriers, promotion of ing mortality from carcinoma of the distal colon andrectum has been demonstrated in two well-designedscreening by the physician may be the most crucial

step in enhancing the use of sigmoidoscopy. case–control studies,6,7 the procedure has a numberof potential limitations. First, because sigmoidoscopyThe current study suggests that specific interven-

tions may be beneficial in enhancing population rates only examines the distal colon, in up to two-thirds of

/ 7bc0$$0731 03-24-98 08:11:27 cana W: Cancer

1480 CANCER April 15, 1998 / Volume 82 / Number 8

patients it may fail to detect adenomas beyond the screening interventions with physicians-in-training toestablish future proactive cancer screening practices.reach of a sigmoidoscope and without an index lesion

in the left colon that would prompt a total colonos- Finally, because most physicians identified poor pa-tient acceptance as a major barrier to screening, pa-copy.30 Second, the rate at which polyps are detected

may be limited by the quality of the bowel preparation tient-based interventions will likely be required aswell.and operator experience. In a recent study with experi-

enced colonoscopists and good bowel preparation,24% of adenomas were not detected by an examiner.31 REFERENCES

1. Parker SL, Tong T, Bolden S, Wingo PA. Cancer statistics,Thus, the effectiveness of the procedure in everyday1997. CA Cancer J Clin 1997;47:5–27.primary care practice may be lower than in clinical

2. Levin B, Murphy GP. Revision in the American Cancer Soci-trials. Third, although the procedure is generally con-ety recommendations for early detection of colorectal can-

sidered to be safe, there is a small complication rate, cer. CA Cancer J Clin 1992;42:296–9.including a 0.01% incidence of perforation and rare 3. National Cancer Institute. Working guidelines for early can-

cer detection: rationale and supporting evidence to decreasecases of electrolyte derangement from phosphate ene-mortality. Bethesda, MD: National Cancer Institute, 1987.mas.32 Fourth, some of the benefits of the procedure

4. American College of Physicians. Clinical efficacy assessmentmay potentially be confounded by detection of slowlyproject. Screening for colorectal cancer. Philadelphia, PA:

advancing cancers, which would lead to length time American College of Physicians, 1990.bias and longer observed survival. Randomized studies 5. U.S. Preventive Services Task Force. Guide to clinical pre-

ventive services. 2nd edition. Washington, DC; 1995.of screening, such as the Prostate, Lung, Colorectal6. Selby JV, Friedman GD, Quesenberry CP Jr., Weiss NS. Aand Ovarian Cancer Screening Trial, are ongoing and

case control study of screening sigmoidoscopy and mortalityare less subject to such biases. Nonetheless, despitefrom colorectal cancer. N Engl J Med 1992;326:653–7.

its limitations, the use of sigmoidoscopy is promoted 7. Newcomb PA, Norfleet RG, Storer BE, Surawicz TS, Marcusby all major cancer prevention groups,2–5 and its cost- PM. Screening sigmoidoscopy and colorectal cancer mortal-

ity. J Natl Cancer Inst 1992;84:1572–5.effectiveness has been shown to be comparable to8. Anderson LM, May DS. Has the use of cervical, breast, andother recommended cancer screening interventions,

colorectal cancer screening increased in the United States?such as mammography (i.e., $42,000–$47,000 per yearAm J Public Health 1995;85:840–2.

of life saved).339. Cooper GS, Fortinsky RH, Hapke R, Landefeld CS. Primary

In interpreting the significance of our findings, we care physician recommendations for colorectal cancerscreening. Arch Intern Med 1997;157:1946–50.recognize several potential methodologic limitations.

10. Fowler FJ. Survey research methods. Revised edition. BeverlyFirst, because the data were derived from physicianHills, CA: Sage, 1988.self-reports, study results may overestimate the actual

11. 1989 survey of physicians’ attitudes and practices in earlyfrequency of screening procedures practiced by re- cancer detection. CA Cancer J Clin 1990;40:77–101.spondents. This pattern was found in a previous study 12. Czaja R, McFall SL, Warnecke RB, Ford L, Kaluzny AD. Pref-

erences of community physicians for cancer screeningthat directly compared self-reported with actual can-guidelines. Ann Intern Med 1994;120:602–8.cer screening practices.34 Second, although there is

13. Schoen RE, Weissfeld JL, Kuller LH. Sigmoidoscopy usealways the possibility that those who respond to aamong primary care physicians. Prev Med 1995;24:249–54.

survey fail to represent the population of interest, the 14. Woo B, Woo B, Cook EF, Weisberg M, Goldman L. Screeningsimilarity between respondents and nonrespondents procedures in the asymptomatic adult. JAMA 1985;

254:1480–4.in this study is somewhat reassuring. However, with-15. Lurie N, Slater J, McGovern P, Ekstrum J, Quam L, Margolisout intensive efforts to contact a sample of nonrespon-

K. Preventive care for women: does the sex of the physiciandents, we cannot assess how this group would havematter? N Engl J Med 1993;329:478–82.

actually answered the questionnaire. If respondents 16. Polednak AP. Knowledge of colorectal cancer and use ofwere more likely than nonrespondents to perform or screening tests in persons 40–74 years of age. Prev Med

1990;19:213–26.refer for sigmoidoscopy, then these findings may rep-17. Jepson C, Kessler LG, Portnoy B, Gibbs T. Black-white differ-resent an overestimate of the actual number of prac-

ences in cancer prevention knowledge and behavior. Am Jtitioners who offer the procedure. Third, we did notPublic Health 1991;81:501–4.

determine physicians’ willingness to initiate or in- 18. Lewis SF, Jensen NM. Screening flexible sigmoidoscopy: fac-crease their use of sigmoidoscopy, and future surveys tors associated with utilization. J Gen Intern Med 1996;

11:542–4.should include items about respondent’s readiness to19. Fox SA, Murata PJ, Stein JA. The impact of physician compli-change.35

ance on screening mammography for older women. ArchWe conclude that, to enhance physician use ofIntern Med 1991;151:50–6.

sigmoidoscopic screening in the United States, tar- 20. NCI Breast Cancer Screening Consortium. Screening mam-geting older, non–board certified practitioners may be mography: a missed clinical opportunity? JAMA 1990;

264:54–8.particularly beneficial. We also recognize the value of

/ 7bc0$$0731 03-24-98 08:11:27 cana W: Cancer

Associations with Sigmoidoscopic Screening/Cooper et al. 1481

21. U.S. Public Health Service. Healthy people 2000: national health 29. Rimer BK, Resch N, King E, Ross E, Lerman C, Boyce A,et al. Multistrategy health education program to increasepromotion and disease prevention objectives. Washington, DC:

U.S. Department of Health and Human Services, 1990. mammography use among women ages 65 and older. PublicHealth Reports 1992;107:369–80.22. Perera DR, LoGerfo JP, Shulenberger E, Ylvisaker JT, Kirz

HL. Teaching sigmoidoscopy to primary care physicians: a 30. Rex DK, Lehman G, Hawes R, Ulbright T, Smith J. Screeningcolonoscopy in asymptomatic average-risk persons withcontrolled study of continuing medical education. J Fam

Pract 1983;16:785–8. negative fecal occult blood tests. Gastroenterology1991;100:64–7.23. Davis DA, Thomson M, Oxman AD, Haynes RB. Evidence

for the effectiveness of CME. JAMA 1992;268:1111–7. 31. Rex DK, Cutler CS, Lemmel GT, Rahmani EY, Clark DW,Helper DJ, et al. Colonoscopic miss rates of adenomas deter-24. McDonald CJ, Hui SL, Smith DM, Tierney WM, Cohen SJ,

Weinberger M, et al. Reminders to physicians from an intro- mined by back-to-back colonoscopies. Gastroenterology1997;112:24–8.spective computer medical record. Ann Intern Med

1984;100:130–8. 32. Waye JD, Kahn O, Auerbach ME. Complications of colonos-copy and flexible sigmoidoscopy. Gastrointest Endosc Clin25. Lomas J, Enkin M, Anderson GM, Hannah WJ, Vayda E,

Singer J. Opinion leaders vs audit and feedback to imple- N Am 1996;6:343–77.33. Wagner JL, Herdman RC, Wadhwa S. Cost effectiveness ofment practice guidelines. JAMA 1991;265:2202–7.

26. Soumerai SB, Avorn J. Principles of educational outreach colon cancer screening in the elderly. Ann Intern Med1991;115:807–17.(‘academic detailing’) to improve clinical decision making.

JAMA 1990;263:549–56. 34. Montano DE, Phillips WR. Cancer screening by primary carephysicians: a comparison of rates obtained from physician27. Levin B, Bond JH. Colorectal cancer screening: recommen-

dations of the U.S. Preventive Services Task Force. Gastroen- self-report, patient survey, and chart audit. Am J PublicHealth 1995;85:795–800.terology 1996;111:1381–4.

28. Nattinger AB, Panzer RJ, Janus J. Improving the utilization 35. Main DS, Cohen SJ, DiClemente CC. Measuring physicianreadiness to change cancer screening: preliminary results.of screening mammography in primary care practices. Arch

Intern Med 1989;149:2087–92. Am J Prev Med 1995;11:54–8.

/ 7bc0$$0731 03-24-98 08:11:27 cana W: Cancer