sM
Women’s Health Issues 15 (2005) 249–257
CP
COLORECTAL CANCER SCREENING BEHAVIOR IN WOMENATTENDING SCREENING MAMMOGRAPHY:LONGITUDINAL TRENDS AND PREDICTORS
Ruth C. Carlos, MD, MSa,b�, A. Mark Fendrick, MDc,d, Paul H. Abrahamse, MAc,Qian Dong, MDb, Stephanie K. Patterson, MDb, and Steven J. Bernstein, MD, MPHa,b,c
aVA Center for Practice Management and Outcomes Research, Ann Arbor, MichiganbDepartment of Radiology, University of Michigan, Ann Arbor, Michigan
cDepartment of Internal Medicine, University of Michigan, Ann Arbor, MichigandDepartment of Health Management and Policy, University of Michigan, Ann Arbor, Michigan
Received November 8, 2004; received in revised form January 1, 2005; accepted June 3, 2005
Purpose. Nationally representative surveys demonstrate that the adherence to screening mam-mography guidelines are associated with increased prevalence of colorectal cancer (CRC)screening; however, the incidence of CRC screening in the screening mammography populationis unknown. Our purpose was to describe non-fecal occult blood test (FOBT) CRC screeningutilization by women prior to and subsequent to screening mammography at a large academicmedical center.
Materials and methods. Using the institutional administrative data base, 17,790 women aged50 and older who underwent screening mammography between 1998 and 2002 were retro-spectively identified. We determined that women were current with non-FOBT CRC screen-ing at the time of mammography if they had undergone flexible sigmoidoscopy or double-contrast barium enema in the 5 years or colonoscopy since 1995, the earliest for which data areavailable. We excluded FOBT as a form of CRC screening because the administrative database did not adequately capture episodes of FOBT. Women who were not current wereconsidered eligible for non-FOBT CRC screening. We then assessed the number of womenwho underwent flexible sigmoidoscopy, barium enema, or colonoscopy within 12 monthsfollowing mammography. Age, insurance status, Breast Imaging Reporting and Data Systemclassification, recommendations after screening mammography and year of mammographywere examined as potential predictors of non-FOBT CRC screening completion.
Results. At the time of mammography, 13.3% women were current with non-FOBT CRCscreening. Of women eligible for non-FOBT CRC screening at the time of mammography,1.1% completed non-FOBT CRC screening within 12 months after mammography. The rate ofnon-FOBT CRC screening completion increased over time. After multivariate analysis, beinginsured by a commercial managed care organization or by Medicaid remained significantpredictors of non-FOBT CRC screening.
Conclusion. The prevalence of non-FOBT CRC screening is low in the population of womenundergoing screening mammography, with an incidence of 1.0%. Future studies shouldexamine whether delivering CRC screening interventions at a screening mammography visit
increase adherence to non-FOBT CRC screening.I
Ta
� Correspondence to: Ruth C. Carlos, MD, MS, Assistant Profes-or, Department of Radiology, University of Michigan, 1500 E.
iedical Center Drive, Ann Arbor, MI 48109-0030.
opyright © 2005 by the Jacobs Institute of Women’s Health.ublished by Elsevier Inc.
ntroduction
he American Cancer Society estimated that therewere 74,700 new cases of colorectal cancer (CRC)
nd an estimated 28,800 deaths from CRC in women
n 2003 (Cancer Facts and Figures, 2003). Previous1049-3867/05 $-See front matter.doi:10.1016/j.whi.2005.06.001
sioCcs(atyar(fdWcdtrccw
ti(rFo&sgevisfawb
toSpHgFamS
sdyc
mCssaspdts(cdmsCr
s“2ccacuCeicpycessrafwanca
coran(2ettp
R. C. Carlos et al. / Women’s Health Issues 15 (2005) 249–257250
tudies have illustrated the potential benefit of screen-ng using fecal occult blood testing (FOBT), sigmoid-scopy, and colonoscopy to decrease mortality fromRC. An older, randomized trial demonstrated de-
reased mortality rates from CRC in individuals as-igned to annual FOBT compared to no screening33% decrease after 18 years of follow-up) (Mandel etl., 1993). More recent trials have found 15–18% mor-ality reductions from biennial screening after 8–10ears of follow-up (Hardcastle et al., 1996; Kronborg etl., 1996). A small randomized trial of sigmoidoscopyeported relative risk reduction in mortality of 20%Thiis-Evensen et al., 1999). Other case-control studiesound that sigmoidoscopy screening reduced risk ofeath by 59% (Selby, Friedman, Quesenberry, &eiss, 1992). No identified studies evaluated double-
ontrast barium enema using decreased incidence oreath from CRC as endpoints. There is no screening
rial that has reported the ability of colonoscopy toeduce CRC incidence or mortality; however, a case-ontrol study found decreased mortality fromolonoscopy screening (odds ratio [OR] 0.43) (Ke-enter & Brevinge, 1996).Despite ample evidence demonstrating the effec-
iveness of screening in reducing colon cancer mortal-ty, the Behavioral Risk Factor Surveillance SurveyBRFSS) performed in 1999 revealed that only 40% ofespondents stated that they had ever submitted aOBT, and 44% had undergone either sigmoidoscopyr colonoscopy (Bolen, Rhodes, Powell-Griner, Bland,
Holtzman, 2000; Janes et al., 1999). Using moretringent criteria to assess adherence to publisheduidelines revealed that only 34% of respondents hadither sigmoidoscopy or colonoscopy within the pre-ious 5 years, and only 21% had FOBT in the preced-
ng year. The BRFSS did not differentiate betweenigmoidoscopy or colonoscopy use and did not assessor double-contrast barium enema use. Further, therere no reliable estimates of the proportion of womenho were screened for CRC using double-contrast
arium enema.The rate of CRC screening continues to lag behind
he rate of breast cancer and cervical cancer screeningver the past 5 years (Seeff, Shapiro, & Nadel, 2002;hapiro, Seeff, & Nadel, 2001), even with increasedromotional initiatives and public attention. The-ealthy People 2010 nationwide health initiative has aoal that 50% of adults 50 years and older undergo aOBT within a 2-year period and that at least 50% ofdults 50 years and older will have ever had a sig-oidoscopy (U.S. Department of Health and Human
ervices, 2000).Cross-sectional analyses of the BRFSS, based on
urveys of noninstitutionalized participants, haveemonstrated that the vast majority of women 50ears and older participate in breast and cervical
ancer screening (82% and 70%, respectively), with a varked discrepancy in the rate of participation ofRC screening (Carlos, Fendrick, Patterson & Bern-
tein, 2005). In addition, women who participated increening mammography within the past year werelmost three times likely to be current with CRCcreening guidelines compared to women who did notarticipate in screening mammography (Carlos, Fen-rick, Patterson & Bernstein, 2005). However, even in
his highly compliant group, the prevalence of CRCcreening adherence remained much lower than 50%Carlos, Fendrick, Patterson & Bernstein, 2005). Be-ause these analyses were based on cross-sectionalata, it could not be determined if screening mam-ography preceded CRC screening. Although an as-
ociation was found between mammography andRC screening adherence, the direction of the effect
emains unknown.Given its near universal acceptance by women,
creening mammography represents a potentialteachable moment”(i.e., unique opportunity) (Carlos,004) for educating patients about the risk of colonancer. Using the setting of screening for one type ofancer(such as screening mammography) to encour-ge compliance with screening for another type ofancer(such as colorectal cancer) represents a novelse of the diagnostic imaging experience (Carlos, 2004;arlos & Fendrick, 2004). Other investigators havevaluated the use of diagnostic testing as a method tomprove health behaviors, most notably in smokingessation. Hepper et al. (1980) demonstrated thatatients were more likely to quit smoking within 2–3ears of the first spirometry test if the test is abnormal,ompared to patients with a normal first-time spirom-try test result. Women participating in comprehen-ive lung cancer screening protocol that includedpiral computed tomography documented a 16% quitate after the procedure (Schnoll et al., 2002). Theseuthors further demonstrated that 59% of current orormer smokers queried about lung cancer screening
ould be interested in receiving smoking cessationdvice with screening. In the same population, 52%oted that they would quit if the scan was positive,ompared to 19% if the scan was negative (Schnoll etl., 2002).Many reasons exist as to why a woman may not
omplete CRC screening. Inadequate understandingf the health benefits screening provides, no doctorecommendation, poor insurance coverage, and evenpsychological dread of the procedures may or may
ot all constitute factors that reduce CRC screeningHarewood, 2002; Walsh, Posner, & Perez-Stable,002). Educational efforts have been shown to mod-stly improve FOBT compliance. Coupling educa-ional intervention with a teachable moment increaseshe effect of the intervention. Screening mammogra-hy represents a potential teachable moment for pro-
iding CRC education designed to improve overall
kpi
CtwadawM
M
PW5meubropPCgaiwsaervuaavl
titaa
cRdhWtzS
iata
v
AWswfyco1CpsyAwpbs
basctc
iGacbfccdoumse
1db
IPbB
R. C. Carlos et al. / Women’s Health Issues 15 (2005) 249–257 251
nowledge of CRC screening and to address specificsychosocial concerns, similar to smoking cessation
nterventions (McBride, Emmons, & Lipkus, 2003).Before we can study the effectiveness of promoting
RC screening at these teachable moments, we needo estimate the rate of CRC screening compliance in
omen undergoing screening mammography whore eligible for CRC screening. In this paper, weescribe non-FOBT CRC screening utilization prior tond subsequent to screening mammography foromen attending breast cancer screening at a largeidwestern academic medical center.
ethods
atient population and data collectione retrospectively identified 17,790 women between
0 and 75 years of age who participated in screeningammography(CPT code 76092) at a large Midwest-
rn academic medical center between 1998 and 2002sing the radiology information system (RIS) dataase.Screening mammography is defined as mammog-aphy performed in women who do not have a historyf breast cancer, a palpable mass, or an active breastroblem. This age cohort was chosen as both the U.S.reventive Services Task Force and the Americanancer Society (ACS) recommend CRC screening be-inning at age 50 for asymptomatic individuals ofverage risk who do not have risk factors predispos-ng to or a family history of CRC (Bolen, 1997; Ke-
enter & Brevinge, 1996). Further, there are no datahowing that screening beyond 80 years is efficaciousnd randomized controlled trials suggest that lifexpectancy of at least 5 years may be required toealize the benefits of screening (U.S. Preventive Ser-ices Task Force, 2002). Because we specifically eval-ated double-contrast barium enema, colonoscopy,nd sigmoidoscopy as methods of CRC screeningdherence, where the minimum recommended inter-al between screenings is 5 years, we chose an upper
imit of 75 years for our population.For each subject, we recorded their health insurance
ype, mammogram result (based on the Breast Imag-ng Reporting and Data System [BIRADS] classifica-ion system) and the radiologist’s recommendationfter screening mammography using the institutionaldministrative data bases.Mammography quality assurance (QA) data, ac-
essed through the RIS, was used to determine BI-ADS classification and mammographer’s recommen-ation for screening mammography follow-up. Dataoused within the institution’s Health System Dataarehouse were used to determine insurance status at
he time of screening mammography and CRC utili-ation after screening mammography. The Health
ystem Data Warehouse is an information system antegrating data from the Health System’s hospitalnd professional and clinical and financial data sys-ems. The HSDW includes data from the hospital costccounting system and the professional billing system.The study was approved by the Institutional Re-
iew Board.
scertaining eligibility for colorectal cancer screeningomen who were not current with non-FOBT CRC
creening at the time of screening mammographyere considered eligible for CRC screening. To screen
or CRC, the ACS recommends that individuals 50ears and older undergo FOBT annually, sigmoidos-opy or double-contrast barium enema every 5 years,r colonoscopy every 10 years (Kewenter & Brevinge,996). We considered women current with non-FOBTRC screening at the time of screening mammogra-hy if they had any of the following tests: 1) flexibleigmoidoscopy or barium enema within the last 5ears; or 2) any previous colonoscopy. Although theCS recommends colonoscopy every 10 years, weere not able to determine colonoscopy utilizationrior to 1995. Therefore, we considered the subject toe current with CRC screening if she had colonoscopyince 1995.
We excluded FOBT as a method of CRC screeningecause of limitations in identifying this test throughdministrative data bases. Institutional QA demon-trated that FOBT is usually performed in the clini-ian’s office and although the results are recorded inhe patient’s medical chart, they are not consistentlyaptured by administrative data bases.
CRC screening utilization was determined using thenstitution’s administrative data base using CPT codes0104 (screening flexible sigmoidoscopy) and G0121
nd G0105 (screening colonoscopy). Because the Godes are a recent addition to the list of approved CPTilling codes we also used existing procedure codesor flexible sigmoidoscopy (45.24 or 45330), double-ontrast barium enema (76092 or 74280), or colonos-opy (45378) in combination with V76.41 and V76.51iagnostic codes (screening for malignant neoplasmsf the rectum or colon). These diagnostic codes weresed to differentiate between receiving flexible sig-oidoscopy, barium enema, and colonoscopy for
creening rather than for diagnostic evaluation of anxisting medical condition.
Completion of CRC screening in eligible women in2 months following screening mammography wasetermined using the institution’s administrative dataase, as described.
dentifying predictors of colorectal cancer screeningotential predictors of non-FOBT CRC screeningehavior included age, type of health insurance,IRADS classification of the mammogram result,
nd recommendations after screening mammogra-
psemBpp(tmof
cio
STtfrTmMcomdsootm
fTwvtwlOaiwviaotFsCs
QEmppsiwpm
ap
Ts
C
NA
Y
H
B
R
M
C*†
R. C. Carlos et al. / Women’s Health Issues 15 (2005) 249–257252
hy (Table 1). Previous investigators have demon-trated that patients who undergo mammographyxperience anxiety from a positive screening mam-ography necessitating further evaluation (Lowe,
alanda, Del Mar, & Hawes, 1999). Even if the initialositive mammogram is determined to be a falseositive, residual distress and anxiety can exist
Aro, de Koning, Absetz, & Schreck, 1999). Al-hough there is no direct evidence that false-positive
ammography results in enhanced CRC adherence,ne may infer that this elevated concern may trans-
able 1. Characteristics of women 50–75 years old receivingcreening mammography between 1998 and 2002
Current WithCRC
Screening*
Not CurrentWith CRCScreening*
haracteristic N (%) (N)
2370 15420ge*(y)50–59 1276 (53.8) 9186 (59.6)60–69 769 (32.4) 4389 (28.5)�70 325 (13.7) 1845 (12.0)
ear of screeningmammography
1998 420 (17.7) 2923 (19.0)1999 507 (21.4) 3072 (19.9)2000 380 (16.0) 2498 (16.2)2001 464 (19.6) 3151 (20.4)2002 599 (25.3) 3776 (24.5)ealth insurance type†
Commercial, nonmanagedcare
381 (16.1) 4048 (26.3)
Commercial, managedcare
1202 (50.7) 7037 (45.6)
Medicare 738 (31.1) 4103 (26.6)Medicaid 49 (2.1) 232 (1.5)
IRADS classification ofmammography results
Normal or benignfindings
2330 (98.3) 15232 (98.8)
Indeterminate or probablymalignant findings
32 (1.4) 151 (1.0)
Not specified 8 (0.3) 37 (0.2)ecommendation for follow-
upRoutine screening
mammography in 12months
2238 (94.4) 14470 (93.8)
Recommend additionalimaging/tissuesampling/clinicalconsultation
132 (5.6) 950 (6.2)
ethod of CRC screening*Flexible sigmoidoscopy 1741 (75.5) —Colonoscopy 432 (18.9) —Double-contrast barium
enema132 (5.6) —
RC, colorectal cancer.Fecal occult blood testing was not assessed.p � .05.
er to CRC adherence behavior from Aro et al., t
oupled with data from Fendrick et al., indicatingncreased CRC adherence in women with a historyf breast cancer (Fendrick, unpublished data).
tatistical analysishe study’s primary outcome was the rate of comple-
ion of non-FOBT CRC screening in the 12 monthsollowing screening mammography. Age was catego-ized into three age groups: 50–59, 60–69, and �70.ype of health insurance was categorized into com-ercial nonmanaged care, commercial managed care,edicare, or Medicaid. We dichotomized the BIRADS
lassification of mammographic findings into normalr benign findings and indeterminate or probablyalignant findings. We also dichotomized recommen-
ations after screening mammography into routinecreening mammography recommended in 12 monthsr no further screening mammography recommendedwing to age, and requiring additional comparison,esting, or consultation. The year of screening mam-
ography was treated as a categorical variable.Each potential predictor variable was first screened
or its relationship with cancer screening adherence.he univariate analysis was evaluated using �2 testhen the predictors were dichotomous or categorical
ariables and using Student’s t-test when the predic-ors were continuous variables. Multivariate analysis
as subsequently performed, using a constructedogistic regression model to determine the adjustedRs of each of the predictor variables. Multivariate
nalysis was conducted and the presence of confound-ng and effect modification was assessed using for-
ard selection. Given the few number of predictorariables examined and the size of the population, we
ncluded all predictor variables in the multivariatenalysis. A nonparametric test for trend, an extensionf the Wilcoxon rank-sum test, was used to determinehe presence of a significant temporal trend in non-OBT CRC screening behavior (Cuzick, 1985). Alltatistical analyses were done using Stata 7.0 (Stataorp., College Park, TX). Statistical significance was
et at .05.
uality assurance of data collectionach woman may have participated in screeningammography from one to five times during the
eriod of evaluation. To reduce bias attributable toreferences of women who underwent multiple epi-odes of annual screening mammography at the samenstitution, we conducted a subanalysis on unique
omen undergoing screening mammography in theeriod of evaluation, where the last screening mam-ogram represented the index mammogram.Many women who obtain screening mammography
t the academic medical center reported on in thisaper receive their routine clinical care elsewhere;
hus all non-FOBT CRC screening procedures may not
becitptred
R
CTmOwic5.ncct
CSwtambwrytccctcmms1nc
PAc.ac
tli
QIuwttCitccrwmp
(tiFppi
Fsdstosoo
R. C. Carlos et al. / Women’s Health Issues 15 (2005) 249–257 253
e captured within its administrative data base. Tonsure that the administrative data base adequatelyaptured all instances of non-FOBT CRC screening, wedentified the subset of women who were enrolled inhe academic medical center’s large managed carelan at the time of screening mammography, under
he assumption that these women would receive theiroutine clinical care within the institution, and allncounters would be captured by the administrativeata bases.
esults
urrent with colon cancer screeninghere were 17,790 women who underwent screeningammography between 1998 and 2002 (see Table 1).f these 17,790 women, 2,370 (13.3%) were currentith non-FOBT CRC screening at the time of screen-
ng mammography. A smaller proportion of womenurrent with non-FOBT CRC screening were aged0–59 compared to women who were not current (p �01). A greater proportion of women current withon-FOBT CRC screening had commercial managedare insurance compared to women who were noturrent (p � .01). No other differences in characteris-ics between the two groups were demonstrated.
ompletion of colon cancer screening in eligible womenubsequent analysis was limited to the 15,420 womenho were eligible of non-FOBT CRC screening at the
ime of screening mammography. Length of follow-upfter screening mammography ranged from 0–72onths (median 30 months). Only 175 (1.1%) of eligi-
le women completed non-FOBT CRC screeningithin 12 months of screening mammography. The
ate of non-FOBT CRC screening completion in theear after screening mammography increased overime (Figure 1), the trend toward increasing CRCompletion was statistically significant (p � .05). Weompared the characteristics of eligible women whoompleted non-FOBT CRC screening compared tohose who did not (Table 2). More women whoompleted non-FOBT CRC screening had commercialanaged care insurance or Medicaid (p � .01). Theajority of women who completed non-FOBT CRC
creening received mammography screening after998 compared to women who did not completeon-FOBT CRC screening (p � .01). No other signifi-ant differences in characteristics were identified.
redictors of colon cancer screening completionfter multivariate analysis (Table 3), having commer-
ial managed care insurance (adjusted OR 2.22, p �001) or Medicaid(adjusted OR 2.93, p � .01) remained
significant predictor of non-FOBT CRC screening
ompletion within 1 year after mammography. Fur- .her, compared to 1998, the first year of our study, theikelihood of completion of non-FOBT CRC screeningncreased over time (p � .01).
uality assurance of data collectionn the subanalysis of the 8,132 unique women whonderwent screening mammography, 1,241 (15.0%)ere current with non-FOBT CRC screening at the
ime of the index mammogram, compared to 13.3% inhe overall population. The average yearly non-FOBTRC screening completion rate was 1.1%, the same as
n the overall population. After multivariate analysis,he significant predictors of non-FOBT CRC screeningompletion included having commercial managedare insurance and the year of screening mammog-aphy(see Table 3). If bias toward preferences ofomen who more frequently participated in screeningammography influenced non-FOBT CRC screening
revalence and incidence, the effect is slight.In the subanalysis of 5,773 institutional insurees, 824
14.3%) were current with non-FOBT CRC screening atime of screening mammography compared to 13.3%n the overall population. The average yearly non-OBT CRC screening completion rate was 1.4%, com-ared to 1.1% in the overall population. Significantredictors of non-FOBT CRC screening completion
ncluded having managed care (adjusted OR 4.40, p �
igure 1. Colorectal (CRC) utilization in women undergoingcreening mammography. CRC utilization is reflected in the calen-ar year of mammography utilization. There is a statisticallyignificant trend toward increasing rates of CRC screening comple-ion over time.(p � .05). Prevalence of CRC screening is the percentf women who were current with CRC screening at the time ofcreening mammography. Incidence of CRC screening is the percentf eligible women who completed CRC screening within 12 monthsf screening mammography.
03) and the year of screening mammography.
D
ImmFpFmCniFt
WciccigetKowurDcCnbsfa
hpB1putmevoscssmi
itwortp
wttdsha
Tcm
C
NA
H
Y
B
R
M
C*†
R. C. Carlos et al. / Women’s Health Issues 15 (2005) 249–257254
iscussion
n the current study, we demonstrated, at the time ofammography, a baseline prevalence of approxi-ately 13% of women who were current with non-
OBT CRC screening. The current study extends therevious analyses by evaluating the incidence of non-OBT CRC screening completion after screeningammography. Of the more than 85% eligible forRC screening, the average rate of completion ofon-FOBT CRC screening was 1% in the year follow-
ng screening mammography. Further, although non-OBT CRC screening completion increased over time,
able 2. Characteristics of eligible women who completedolorectal cancer screening one year after screeningammography compared to noncompleters
haracteristicCompleted CRCScreening, N(%)
Did Not CompleteCRC Screening,
N(%)
umber 175 15245ge (y)50–59 112 (64.0) 9074 (59.5)60–69 52 (29.7) 4337 (28.4)�70 11 (6.3) 1834 (12.0)ealth insurance type†
Commercial, nonmanagedcare
28 (16.0) 4020 (26.4)
Commercial, managedcare
109 (62.3) 6928 (45.4)
Medicare 38 (21.7) 4070 (26.7)Medicaid 5 (2.9) 227 (1.5)
ear of screeningmammography†
1998 13 (7.4) 2910 (19.1)1999 41 (23.4) 3031 (19.9)2000 28 (16.0) 2470 (16.2)2001 39 (22.3) 3112 (20.4)2002 54 (30.9) 3722 (24.4)
IRADS classification ofmammography results
Normal or benignfindings
174 (99.4) 15054 (98.7)
Indeterminate or probablymalignant findings
1 (0.6) 191 (1.3)
ecommendation for follow-up
Routine screeningmammography in 12months
165 (94.3 14333 (94.8)
Recommend additionalimaging/tissuesampling/clinicalconsultation
10 (5.7 912 (1.3)
ethod of CRC screening*Flexible sigmoidoscopy 79 (45.1) —Colonoscopy 26 (14.9) —Double contrast barium
enema70 (40.0) —
RC, colorectal cancer.Fecal occult blood testing was not assessed.p � .05
he maximum rate of completion did not exceed 2%. a
e speculate that the relative improvement in CRCompletion rate over time may be an effect of increas-ng practitioner and public attention on colorectalancer screening in 2000. In 1999, Medicare beganovering CRC screening services. In 2000, the ACSssued its first set of comprehensive cancer screeninguidelines in nearly 20 years (ACS, 1980), the cost-ffectiveness of CRC screening in the general popula-ion was demonstrated (Frazier, Colditz, Fuchs, &untz, 2000) and a screening colonoscopy was shownn national television. In particular, the latter event, inhich a national correspondent on The Today Shownderwent a colonoscopy has been linked to a tempo-al increase in CRC screening rates (Cram et al., 2003).espite the improvement, the rate of CRC screening
ompletion in this population remains quite low.learly, there is room for improvement. Given theear universal acceptance of screening mammographyy women, the mammography appointment repre-ents a potential “teachable moment” (Carlos, 2004)or educating women about the risk of colon cancernd benefits of CRC screening.There are psychosocial factors that account for a
igh level of compliance with screening mammogra-hy (Aro et al., 2000; Lauver, Henriques, Settersten, &umann, 2003; Zapka, Harris, Stoddard, & Costanza,991). Understanding these psychosocial factors canotentially increase compliance with CRC screening,sing the mammography suite as a delivery setting for
argeted education. We posit that the screening mam-ography encounter can be used as a gateway to
ncourage women to partake in CRC screening ser-ices, regardless of the type of CRC screening methodr which clinical service will be providing CRCcreening. Within the mammography encounter, wean potentially close the loop between intention tocreen when the woman is made aware of CRCcreening need and actually scheduling an appoint-ent with her clinician to receive a referral for screen-
ng (Carlos & Fendrick, 2005).Previous analyses of cancer screening behavior us-
ng the BRFSS data have demonstrated that adherenceo breast and cervical cancer screening is associated
ith CRC screening adherence; however, the directionf the association could not be determined. The cur-ent study addresses this limitation by evaluating theemporal association between screening mammogra-hy and CRC screening.Our study has several limitations. The prevalence ofomen who were current with CRC screening at the
ime of screening mammography is significantly lesshan has been reported from analyses of the BRFSSata. Our institutional data collection relied on retro-pective analysis of administrative data, which mayave underestimated the true prevalence. Others havelso noted the limitations of claims data in the evalu-
tion of cancer screening and cancer detection (Fini-
sZB
CedarCsCoWioqdra
dcliFsop3
pma
otchirnsFil
ii2Tc
tjhaic
T
V
A
H
Y
B
R
MppO†
* g; thus
R. C. Carlos et al. / Women’s Health Issues 15 (2005) 249–257 255
on, Wellins, Wennberg, & Lucas, 1999; Freeman,hang, Freeman, & Goodwin, 2000; Nattinger, Laud,ajorunaite, Sparapani, & Freeman, 2004).In addition, we adopted a stringent definition of
RC screening using only double-contrast bariumnema, flexible sigmoidoscopy, or colonoscopy toetermine adherence. Although FOBT remains anccessible and recommended form of CRC screening,ecent work supports a more stringent definition ofRC screening as up to one third of physicians use
ingle-sample in-office testing as the sole means ofRC screening (Nadel et al., 2005), an ineffective formf CRC screening (Collins, Lieberman, Durbin, &eiss, 2005). We likely underestimated CRC screen-
ng adherence because we excluded FOBT as a meansf CRC screening; compliance with FOBT was inade-uately captured in the institutional administrativeata base. Our estimate of CRC utilization, therefore,epresents the lower bound of institutional compli-nce with CRC screening.In a previous analysis of the BRFSS 2001 data, we
emonstrated that 22% of respondents who wereurrent with CRC screening had received FOBT (Car-os, Fendrick, Patterson & Bernstein, 2005). Othernvestigators have demonstrated 5% incidence ofOBT in the 6 months following identification of CRCcreening need. Even assuming that similar utilizationf FOBT was achieved in our current population, therevalence of CRC screening would be approximately
able 3. Correlates of completing CRC screening within 1 year aft
ariable
ge(y)50–5960–69�70ealth insurance typeCommercial managed careMedicareMedicaid
ear of screening mammography19981999200020012002
IRADS classification of mammography resultsNormal or definitely benign findings
ecommendation for follow-upRoutine screening mammography in 12 months or no additional
recommended due to age
ultivariate analysis was performed using a constructed logisticredictor variables. The presence of confounding and effect modifredictor variables examined and the size of the population, all preR, odds ratio; CI, confidence interval.
p � .05.There were no women on Medicaid who completed CRC screenin
6% with an incidence of 6%, markedly lower com- w
ared to screening mammography rates. There re-ains a tremendous opportunity to improve CRC
dherence.Some women may have received their clinical care
utside our institution, leading to underestimation ofhe completion rate of CRC screening. Therefore, weonducted a subanalysis of women whose primaryealth insurance was the institution’s own health
nsurance plan. Among these women, although theate of CRC completion was statistically greater thanoninstitutional insurees, the absolute rate of CRCcreening completion remained extremely low (1.4%).urthermore, the significant predictors of CRC screen-
ng completion were the same as in the larger popu-ation.
Previous work has demonstrated that CRC screen-ng was influenced by demographic factors such asncome, education or race/ethnicity (Shapiro et al.,001; Carlos, Fendrick, Patterson & Bernstein, 2005).hese data are not routinely collected as part of thelinical encounter and are unavailable for analysis.
We were unable to determine colonoscopy utiliza-ion prior to 1995. Therefore, we considered the sub-ect to be current with CRC screening if she had everad colonoscopy. This minor limitation may havellowed underestimation of prevalence of CRC screen-ng. The study was conducted at one large academicenter in the Midwest and may not generalize to other
ning mammography in eligible women
Adjusted OR (95% CI)in All Cases
Adjusted OR (95% CI)in Unique Cases
— —1.33 (0.95–1.86) 1.36 (0.83–2.21)1.12 (0.59–2.15) 0.93 (0.43–2.00)
2.22 (1.46–3.38)† 3.36 (1.59–7.10)†
1.03 (0.56–1.89) 3.27 (1.36–7.89)†
2.93 (1.11–7.72)† —*
—2.99 (1.60–5.60)† 1.85 (0.51–6.74)2.43 (1.26–4.71)† 1.73 (0.39–7.78)2.68 (1.43–5.04)† 5.30 (1.59–17.6)†
3.06 (1.66–5.64)† 3.56 (1.10–11.5)†
2.23 (0.30–16.7) 0.98 (0.13–7.35)
-up 1.11 (0.57–2.16) 2.18 (0.78–6.07)
sion model to determine the adjusted odds ratios of each of thewas assessed using forward selection. Given the few number of
variables (above) were included in the multivariate analysis.
, this variable was excluded in the multivariate analysis.
er scree
follow
regresicationdictor
omen.
IAcrrsttribli
tvaArrw&
spjuiibhotscc
oataassatitp
ATC
RA
A
A
A
B
B
C
C
C
C
C
C
C
F
F
F
H
H
H
R. C. Carlos et al. / Women’s Health Issues 15 (2005) 249–257256
mplicationslthough nationally representative survey data indi-
ate that women who adhere to screening mammog-aphy are more likely to adhere to CRC screening, theate of CRC screening after women undergoingcreening mammography is quite low. We have iden-ified at least two predictors of CRC screening, namelyhe type of insurance and year of screening mammog-aphy. These factors are not readily modifiable. Find-ngs suggest that although CRC screening appears toe improving with time, they still remain relatively
ow. Therefore, other interventions are necessary tomprove CRC screening in this population.
Why should radiologists be interested in recogni-ion of potential teachable moments in imaging? Someiew radiologists’ participation in CRC screening onlys providers of the double-contrast barium enema.lthough the barium enema is currently the only
ecommended radiologic imaging test, the role ofadiologists in CRC screening will continue to expandith the maturation of virtual colonography (BromerWeinberg, 2005; Rex, 2002).Furthermore, the use of existing imaging-based
creening programs, such as screening mammogra-hy, represents a novel use of imaging technology, not
ust as a diagnostic tool, but as a means for improvingtilization of other proven but underemployed screen-
ng services. Recognition of such teachable momentsn radiology extends the underlying value of imagingeyond its contribution to clinical management, andighlights the potential of radiology services as meansf improving overall patient care. This shift in valua-ion further improves the cost-effectiveness of thecreening mammography program by taking into ac-ount the life savings that accrue from decreasingolon cancer mortality (Carlos, 2004).
Mammography, a widely accepted test, can be anpportunity for improving colon cancer screeningdherence. Screening mammography may represent aeachable moment for behavior-related interventionimed at reducing risk of colon cancer. Targeting thislready compliant group can potentially increase CRCcreening adherence to levels approximating that ofcreening mammography. Future studies should ex-mine whether delivering CRC screening interven-ions at a screening mammography visit may yieldncreased adherence to CRC screening, particularly inhe vast majority of women who receive mammogra-hy as part of their routine care.
cknowledgmenthe project is supported in part by the NIH/NCI 1 K07
A108664 01A1.eferencesmerican Cancer Society (ACS). Guidelines for the cancer-relatedcheckup: recommendations and rationale. (1980). CA: A CancerJournal for Clinicians, 30, 4–50.
merican Cancer Society. (2003). Cancer facts and figures. 4. Atlanta:American Cancer Society.
ro, A. R., de Koning HJ, Absetz P, & Schreck M. (1999). Psycho-social predictors of first attendance for organised mammographyscreening. Journal of Medical Screening, 6, 82–88.
ro, A. R., Absetz, S. P., van Elderen, T. M., van der Ploeg, E., & vander Kamp, L. J. Th. (2000). False-positive findings in mammog-raphy screening induces short-term distress—Breast cancer-spe-cific concern prevails longer. European Journal of Cancer, 36,1089–1097.
olen, J. C., Rhodes, L., Powell-Griner, E. E., Bland, S. D., &Holtzman, D. (2000). State-specific prevalence of selected healthbehaviors, by race and ethnicity—Behavioral Risk Factor Surveil-lance System, 1997. MMWR CDC Surveillance Summary, 49, 1–60.
romer,M. Q., & Weinberg, D. S. (2005). Screening for colorectalcancer—now and the near future. Seminars in Oncology, 32, 3–10.
arlos, R. C. (2004). The added value of screening mammography inimproved screening for other cancers. Journal of the AmericanCollege Radiology, 1, 597–600.
arlos, R. C., & Fendrick, A. M. (2004). Improving cancer screenadherence: Using the “teachable moment” as a delivery settingfor educational intervention. American Journal of Managed Care, 10,247–248.
arlos, R. C., & Fendrick, A. M. (2005). Value added radiology:Using existing radiology services to improve colorectal cancerscreening. Academic Radiology, 12, 475–478.
arlos, R. C., Fendrick, A. M., Patterson, S. K., & Bernstein S. J.(2005). Associations in breast and colon cancer screening behav-ior in women. Academic Radiology, 12, 451–458.
ollins, J. F., Lieberman, D. A., Durbin, T. E., & Weiss, D. G. for theVeterans Affairs Cooperative Study #380 Group. (2005). Accu-racy of screening for fecal occult blood on a single stool sampleobtained by digital rectal examination: A comparison with rec-ommended sampling practice. Annals of Internal Medicine, 142,81–85.
ram, P., Fendrick, A. M., Inadomi, J., Cowen, M. E., Carpenter, D.,& Vijan, S. (2003). The impact of a celebrity promotional cam-paign on the use of colon cancer screening: The Katie Couriceffect. Archives of Internal Medicine, 163, 1601–1605.
uzick, J. A. (1985). Wilcoxon-type test for trend. Statistics inMedicine, 4, 87–90.
inison, K. S., Wellins, C. A., Wennberg, D. E., & Lucas, F. L. (1999)Screening mammography rates by specialty of the usual carephysician. Effective Clinical Practice, 2, 120–125.
razier, A. L., Colditz, G. A., Fuchs, C. S., & Kuntz, K. M. (2000).Cost-effectiveness of screening for colorectal cancer in the gen-eral population. Journal of the American Medical Association, 284,1954–1961.
reeman, J. L., Zhang, D., Freeman, D. H., & Goodwin, J. S. (2000).An approach to identifying incident breast cancer cases usingMedicare claims data. Journal of Clinical Epidemiology, 53, 605–614.
ardcastle, J. D., Chamberlain, J. O., Robinson, M. H., Moss, S. M.,Amar, S. S., Balfour, T. W., et al. (1996). Randomised controlledtrial of faecal-occult-blood screening for colorectal cancer. Lancet,348, 1472–1477.
arewood, G. C., Wiersema, M. J., & Melton, L. J. (2002). Aprospective, controlled assessment of factors influencing accep-tance of screening colonoscopy. American Journal of Gastroenterol-ogy, 7, 3186–3194.
epper, N. G., Drage, C. W., Daview, S. F., Rupp, W. M., LaMothe,J., Schoenfelder, P. F., et al. (1980). Chronic obstructive pulmo-nary disease: A community-oriented program including profes-sional education and screening by a voluntary health agency.
American Review of Respiratory Disease, 121, 97–104.
J
K
K
L
L
M
M
N
N
R
S
U
S
S
S
T
U
W
Z
A
meh
btfb
gc
otm
R. C. Carlos et al. / Women’s Health Issues 15 (2005) 249–257 257
anes, G. R., Blackman, D. K., Bolen, J. C., Kamimoto, L. A., Rhodes,L., Caplan, L. S., et al. (1999). Surveillance for use of preventivehealth-care services by older adults, 1995–1997. Morbidity andMortality Weekly Report CDC Surveillance Summaries, 48, 51–88.
ewenter, J., & Brevinge, H. (1996) Endoscopic and surgical com-plications of work-up in screening for colorectal cancer. Diseasesof the Colon & Rectum, 39, 676–680.
ronborg, O., Fenger, C., Olsen, J., Jorgensen, D., & Sondergaard, O.(1996). Randomised study of screening for colorectal cancer withfaecal-occult-blood test. Lancet, 348, 1467–1471.
auver, D. R., Henriques, J. B., Settersten. L., & Bumann, M. C.(2003). Psychosocial variables, external barriers, and stage ofmammography adoption. Health Psychology, 22, 649–653.
owe, J. B., Balanda, K. P., Del Mar, C., & Hawes, E. (1999).Psychological distress in women with abnormal findings in massmammography screening. Cancer, 85, 1114–1118.
andel, J. S., Bond, J. H., Church, T. R., Snover, D. C., Bradley,G. M., Schuman, L. M., et al. (1993). Reducing mortality fromcolorectal cancer by screening for fecal occult blood. MinnesotaColon Cancer Control StudyNew England Journal of Medicine, 328,1365–1371.
cBride, C. M., Emmons, K. M., & Lipkus, I. M. (2003). Understand-ing the potential of teachable moments: The case of smokingcessation. Health Education Research, 18, 156–170.
adel, M. R., Shapiro, J. A., Klabunde, C. N., Seeff, L. C., Uhler, R.,Smith, R. A., et al. (2005). A national survey of primary carephysicians’ methods for screening for fecal occult blood. Annals ofInternal Medicine, 142, 86–94.
attinger, A. B., Laud, P. W., Bajorunaite, R., Sparapani, R. A., &Freeman, J. L. (2004). An algorithm for the use of Medicare claimsdata to identify women with incident breast cancer. HealthServices Research, 39, 1733–1749.
ex, D. K. (2002) Current colorectal cancer screening strategies:Overview and obstacles to implementation. Review of Gastroen-terological Disorders, 2 (Suppl 1), S2–11.
chnoll, R. A., Miller, S. M., Unger, M., McAleer, C., Halbherr, T., &Bradley, P. (2002). Characteristics of female smokers attending alung cancer screening program: A pilot study with implicationsfor program development. Lung Cancer, 37, 257–265.
. S. Preventive Services Task Force. (2002). Screening for colorectalcancer, recommendations and rationale. Agency for Healthcare Re-search and Quality, 03-510A. Available: http://www.ahrq.gov/clinic.
eeff, L. C., Shapiro, J. A., & Nadel M. R. (2002). Are we doingenough to screen for colorectal cancer? Findings from the 1999Behavioral Risk Factor Surveillance System. Journal of Family
Practice, 51, 761–766.w
elby, J. V., Friedman, G. D., Jr., & Weiss, N. S. (1992). A case-controlstudy of screening sigmoidoscopy and mortality from colorectalcancer. New England Journal of Medicine, 326, 653–657.
hapiro, J. A., Seeff, L.C., & Nadel, M. R. (2001). Colorectal cancer-screening tests and associated health behaviors. American Journalof Preventive Medicine, 21, 132–137.
hiis-Evensen, E., Hoff, G. S., Sauar, J., Langmark, F., Majak, B. M.,& Vatn, M. H. (1999). Population-based surveillance by colonos-copy: Effect on the incidence of colorectal cancer: Telemark PolyStudy I. Scandinavian Journal of Gastroenterology, 34, 414–420.
. S. Department of Health and Human Services. (2000). Healthypeople 2010. With understanding and improving health and objectivesfor improving health (2nd ed., 2 vols). Washington, D. C.: U. S.Government Printing Office.alsh, J. M., Posner, S. F., & Perez-Stable, E. J. (2002). Colon cancerscreening in the ambulatory setting. Preventive Medicine, 35,209–218.
apka, J. G., Harris, D. R., Stoddard, A. M., & Costanza, M. E. (1991).Validity and reliability of psychosocial factors related to breastcancer screening. Evaluation and the Health Professions, 14, 356–367.
uthor DescriptionsRuth C. Carlos is a MRI radiologist interested pri-arily in the use of diagnostic imaging test experi-
nces as teachable moments for improving women’sealth behavior.A. Mark Fendrick is a general internist who has
een extensively involved in assessment of emergingechnologies and more recently has focused on bene-it-based copayment as a means of improving healthehavior.Paul H. Abrahamse is a staff research scientist.Qian Dong is an MRI radiologist.Stephanie K. Patterson is a breast imaging radiolo-
ist interested in racial and ethnic disparities in breastancer detection and treatment.
Steven J. Bernstein is a general internist focusingn appropriateness of use of medical technologies,he roles doctors and patients play in decision
aking, the effects of patient decision aids, and
ays to improve adherence with clinical guidelines.Recommended
