HOULD WE SCREEN FOR COLORECTAL CANCERSPeople with a known genetic predisposition for colorectal...
Embed Size (px)
Transcript of HOULD WE SCREEN FOR COLORECTAL CANCERSPeople with a known genetic predisposition for colorectal...
HOULD WE SCREEN FOR COLORECTAL CANCERS
Bernie Towier, Public Health Officer, Epidemiology andHealth Services Evaluation Branch;Les Irwig, A/Professor (Epidemiology), Department of PublicHealth, University of Sydney;Marion Hems, Public Health Officer, Epidemiology andHealth Services Evaluation Branch;Angela Piunkett, Associate Lecturer, Department of PublicHealth, University of Sydney;Glenn Sal/reid, Lecturer (Health Economics), Departmentof Public Health, University of Sydney.
I n NSW in 1991 colorectal cancer was the second mostcommon cancer affecting women and men, after breast and
prostate cancer, respectively. Because of the burden of thisdisease and because prevention is not possible, population-based screening to detect and remove colorectal adenomasand early asymptomatic cancers has been advocated bysome2. However, the decision to implement population-basedscreening is a comp]ex one. Early detection and treatment ofcolorectal neoplasms (adenomas and cancers) must beshown to be effective in reducing disease mortality and anybenefits of screening must outweigh potential morbiditycaused. Other considerations include the feasibility, cost andacceptability of screening to Australians and the likelihoodof patient and physician compliance with a screeningprogram.We examined the evidence about the effectiveness ofscreening and related issues to inform screening policy.This included a comprehensive and critical review of the
literature, preliminary economic analysis of screeningin the Australian health care setting and correspondencewith Australian investigators for information about localresearch relevant to screening. A summary of this reviewand its recommendations are presented below and inTable 4. The full report, including the tabulated results ofthe critical appraisal of the literature, is available fromthe Epidemiology and Health Services Evaluation Branch.
Is SCREENING EFFECTIVE?Whether screening is effective is the first and crucialquestion to ask before embarking on a screening program.Randomised controlled trials provide the best qualityevidence about screening effectiveness. There are five largetrials (one non-randomised) investigating the effectivenessof screening using the faecal occult blood test (FOBT)Hemoccult'°'°'''°''". The New York trial'20 is evaluatingHemoccult additional to sigmoidoscopy which was offeredto all study participants. The major characteristics andmortality findings of the trials are given in Table 5. In alltrials, 1-Jemoccult positive subjects were referred for furtherinvestigation which was mainly colonoscopy, oralternatively, sigmoidoscopy with double contrast bariumenema. Colorectal neoplasms detected were removed.
Three of the trials have reported mortality findings,analysed by intention to treat'1'. Only the Minnesota trial'has reported a statistically significant 33 per cent reductionin colorectal cancer mortality with annual screening (Table5)- However, the findings of the Minnesota trial must beviewed with caution for several reasons. There areinconsistencies in the numbers of colorectal cancers
Hospital-related mortality in NSW
Continued from page 26
While our development of risk-adjusted indicators is still atan early stage, there are a few excellent examples of clinicalgroups in NSW that have produced and implemented suchindicators to monitor the quality of their services. One suchexample is the Australian and New Zealand Intensive CareSociety (AINZICS). ANZICS, partially funded by the NSWHealth Department's Health Outcomes Program, isinstalling a standard clinical information system inintensive care units throughout Australia to provide risk-adjusted mortality data to the participating units. Theinformation from this system will allow individual units tocompare their overall performance and condition-specificmortality to an international benchmark and facilitatemonitoring of trends in performance. The development ofsuch systems is complementary to the work being carriedout in the Epidemiology and Health Services EvaluationBranch.
The next steps in our exploration of hospital mortality willbe to:
•recalculate in-hospital mortality rates adjustingfor age, sex and casemix, and condition-specificmortality rates adjusting for age and sex;
• evaluate methods for using existing inpatient datain the measurement of co-morbidity and severity ofillness and in risk adjustment;
compare the use of routinely reported inpatient dataand these algorithms against results derived frommore comprehensive clinical databases, e.g. TraumaRegistries;demonstrate the feasibility and usefulness of linkingroutinely reported inpatient data to mortality datato extend this analysis to include post-dischargedeaths; andassess the usefulness of additional data items fromexisting hospital information systems and processeswhich may be of value in the measurement of cc-morbidity, severity of illness and risk adjustment.This evaluation will determine whether suitabledata items are captured by hospital records, theextent and consistency of recording acrossinstitutions and the feasibility of computerisation.
This work will be assisted by a grant from theCommonwealth Government and will include collaborationwith the Victorian Department of Health and CommunityServices.
1. Us Congress, Office of Technology AasessrnenL The Quality of MedicalCare: Information for Consumers, OTA-H-386 Washington, DC; USGovernment Printing Office, June 1988) Chapter 4 p71.2. International Classification of Diseases and Causes of Death, 9th Revision,Clincial Modification Vol 1. 1986. Ann Arbor Michigan Library of CongressNo. 77-94472.3. NSW Public Hospital Comparison Data 1991/2. NSW Health Department,State Health Publication No. CCI 93.134.4. Elixhauser A, Andrews RM, and Fox S. Clinical Classification for HealthPolicy Research: Discharge Statistics by Principal D,agriosis and Procedure.Division of Provider Studies Research Note 17 Agency for Health Care Policyand Receurcli, 1.993 Rockville. MD Public Health Service )ACHCPRPuhlie,itior No, 93.0043),
SUMMARY OF REVIEW AND RECOMMENDATIONSABOUT COLORECAL CANCER SCREENING
1. Population screening is not recommended as screeningeffectiveness has not yet been adequately established.Follow-up continues in the major trials and review of thisposition will be necessary in a few years time when furtherinformation is available.2. Although evidence for screening persons at high risk,such as those with a known genetic susceptibility forcolorectal cancer, is also lacking, it seems prudent to screensuch groups on the basis of the increased risk.3. Further work is needed in Australia to determine theimplications of introducing screening for health servicesplanning and costs, to determine the most accurate andfeasible screening test and the likely compliance ofAustralians with screening, and to determine and caterfor the psychosocial consequences of screening.
reported, information about vital status for all studyparticipants at follow-up is not given, there is no intennediatebiennial screening benefit and about one-third of all peoplescreened had colonoscopy with its attendant risks and costs.It is possible that using colonoscopy alone on a randomsample of one-third of people from either group would haveresulted in this mortality reduction.
The New York trial" has reported a borderline significant43 per cent reduction in colorectal cancer mortality withannual screening in one of the subgroups examined (Table5). But the New York trial was a non-randomised studyand study groups were demonstrably not comparable,suggesting bias in the allocation of subjects to study groups.The Danish trial" has reported a non-significant 17 percent reduction in colorectal cancer mortality with biennialscreening (Table 5). Follow-up continues in four of the fivetrials. In the interim, evidence for the effectiveness ofscreening using Hemoccult remains inconclusive.
Screening using flexible sigmoidoscopy has been suggested2.The only evidence about screening effectiveness usingsigrnoidoscopy comes from two recent case-controlstudies'4". The authors reported 60-80 per cent reductionsin colorectal cancer mortality with sigmoidoscopic screeningbut these results must be interpreted cautiously becauseof the biases inherent in case-control studies to evaluatescreening. In addition, sigmoidoscopy is more invasive andexpensive than Hemoccult and has an estimated rate ofbowel perforation of 0.02 per cent'6. Randomised trialsevaluating the effectiveness of screening usingsigrnoidoscopy are under way irs the United Kingdom'7and the United States". There are no studies investigatingthe effectiveness of screening using colonoscopy. However,colonoscopy is probably unsuitable as a screening testas it is expensive, invasive, requires sedation and bowelpreparation and has an estimated complication rate ofperforation, haemorrhage and death of 0.17 per cent,0.03 per cent and 0.02 per cent respectively.
FAECAL OCCULT BLOOD TESTS (FOBTS)The major trials are evaluating screening using Hemoccult,an inexpensive guaiac-based FOBT which detects bloodproducts in faeces. Estimates of the sensitivity of Hemoccultfor detecting colorectal cancer in asymptomatic populationsvary greatly (22-92 per cent"°). More recent
immi.mochemical FOBTs appear to have greater sensitivitythan Hemoccult without loss of specificity"", but the newertests need further evaluation in asymptomatic populations.The value of detection of colorectal adenomas with FOBTsin reducing mortality from colorectal cancer remainsunclear although researchers from the Minnesota trial'suggest this may become clearer with further follow-up.
GROUPS AT INCREASED RISK OF COLORECTAL CANCERPeople with a known genetic predisposition for colorectalcancer such as those with familial adenomatous polyposis orhereditary non-polyposis colorectal cancer (HNPCC) have amarkedly increased risk of colorectal cancer: 50 per cent ofthe children of people affected with HNPCC are reported todevelop colorectal cancer". Thus, despite uncertainty aboutthe trial evidence, screening seems prudent for these peoplein the light of their high risk. People with a family history ofcolorectal cancer with no known genetic basis have a two- tofour-fold increased risk of colorectal cancer". The decision toscreen or not screen such people is less clear and should beleft to the individuals and their medical practitioners.
WOULD AUSTRALIANS PARTICIPATE IN SCREENING?Assuming that colorectal cancer screening is effective,compliance of the Australian population with screeningwould be critical to the ability of a screening program toreduce mortality rates. The notion of compliance is complex,encompassing initial and continued participation inscreening and adherence to follow-up, treatmentrecommendations and post-treatment surveillance.
Compliance of screen-positive people with follow-upinvestigations is essential to maximise the effectiveness ofscreening. Compliance with follow-up was high in the majortrials (77-93 per cent). However, the numbers of eligiblepeople who choose to participate in screening will influencethe public health impact of the screening program in itsability to reduce colorectal cancer mortality. Compliancewith initial Hemoccult testing ranged from 53 per cent to80 per cent in the major trials (Table 5) and was positivelyassociated with younger age and female gender. Compliancewith rescreening, where offered to all those in the screengroup, not just previous compliers, was lower (20-58 percent, Table 5). Personalised invitations to screening,reminder letters or telephone calls increased compliancein the trials and other studies"". Preliminary Australianresearch on compliance with colorectal cancer screening hasbeen done"7" but more work is needed to assess barriers toscreening participation and to test strategies for overcomingthese barriers.
PSYCHOSOCIAL IMPACT OF SCREENINGIn the major trials, screening with Hemoccult resulted in apositive predictive value for cancer ranging from 2 per centto 18 per cent and 2 to 4 per cent if Hemoccult slides wererehydrated (Table 5). Therefore, more than 80 per cent ofpositives are false positives: such large numbers of peoplereceiving falsely positive results is important since apositive result is not unexpectedly associated withdistress'1". All these people would require colonoscopy, aninvasive procedure which, in addition to the physical risk,may cause pain and embarrassment' and significantanxiety'. There is also evidence from other screeningprograms that anxiety aroused by a false positive result
SUMMARY OF CHARACTERISTICS AND MORTALI1Y RESULTS OF TRIALSOF HEMOCCULT SCREENING FOR COLORECTAL CANCER
Minnesota, Nottingham. Goteborg, Funen, New York,USA England Sweden Denmark USA
Study population 46,551 Mirr'iesota 107349 Nottingham All 27700 61,938 inhabitants 21,756 clentsvolunteers aged subjects aged 50-74 inhabitants of of the island of attending New York50-80 years. yrs identified from Goteborg, Sweden, Funen, Denmark, medical clinic, aged
OP records. aged 60-64 years. aged 45-74 years. 40 years and older.
Assgnmentto study groups Volunteers randomly Subjects randomly All subjects Subjects identified Subjects presenting atassigned to annual allocated either to randomly allocated from the Central clinic from 1975-1979or biennial screen receive HO test to either receive HO Person Register were stratified into twogroup or control packs with letter tests, letter of randomly allocated groups according togroup, from OP or to instructions plus to biennial HO dinic attendance and
control group. questionnaire or screening (tests, allocated to screen orto control group. instructions plus control groups
questionnaire) or according to monthto controL group. of presentation. Both
groups offered annualSO; screen gp alsooffered annual HO.
Screening test(s) HO offered to screen HO - unhydr. HO - most rebydr. HO - unhydr. HO - most unhydr.groups.HO - most rehydr.
Screening and follow-up Screening: 1975- Screening started Screening started Screening: 1985- Authors state: 'Follow-periods 1982 and 1986-1992. 1981, offered 1982, rescreening 1990, three screens up ceased 1984'.
Follow-up continues, biennially. offered to all screen done. Follow-upRecruitment and group 16-22 months continues.follow-up continues, later. Follow-up
Compliance with screening Percent of (Rescreening offered 1st round: (Rescreening offered Compliance with HOscreenings only to compliers.) 66% completed only to compliers.) on first round:completed: 1st round:
HO test and 1st round: Regular attenders: 70%Annual gp: 75.2% 53% completed HO
questionnaire. 67% completed First presenters: 80%Biennial yp; 78,4%
test. 2nd round:55% completed HO
screening.Compliance in next
2nd round:77% completed
and questionnaire.2nd round:93% completed
2 rounds decreased to
tests, tests.20% and then i6% forfirst presenters.
Positive Predictive Value (PPV) 2.2% (rehydr HO) - (NB: Proportion 1st round: 1st round: 17.7% Overall: 10.7%of test for colorectal cancer 5.6% (unhydr HO).
of HO positives (most unhydr) 5.0% 2nd round: 8.4%investigated not 2nd round:given.) (all rehydr) 4.2%1st round: 710.2%
2nd round: 78.5%
Mortality findings reported At 13 years follow- Mortality data not Mortality data not At 38 months At 9 years follow-up:up: yet available, yet available, follow-up: CRC mortalityll000lyr:CRC mortality/bOG: Follow-up continues. Follow-up continues. CRC mortality/bOO: 'regular attenders':Annual: 5.9 (4.67.2)* Screen gp: 2.4 Screen gp: 0A7Biennial: 8.3 (6.8-9.8) Control gp: 2.9 Control gp: 0.41Control: 8.8 (7.3-10.4) (p=0.24)
'first presenters':All cause All cause Screen gp: 0.36mortality/i 000: mortality/lOOt): Control gp: 0.63Annual: 216 Screen gp: 86.2 (p=O.O53)
(209-22 2) Control gp: 89.6Biennial: 218
HO - Hemoccult Ii; CRC - Colorectal Cancer; SO - Signoidoscopy; rehydr - Rehydrated, unhydr - Unfrydrated.*95% Confidence Intervals in brackets.
I ___jVol.5JNo.3 30
Continued from page 29
may be long term32 and negative results may reassurepeople who then do not seek medical care for rectalbleeding20. Implementation of screening requires attentionto the provision of appropriate information about itspm-pose and pitfalls. There also need to be support servicesfor people receiving, and undergoing investigation for,positive test results.
ECONOMIC CONSIDERATIONSThe risks, benefits and costs are major factors in decisionsabout screening. Information on screening benefits in termsof survival with and without the disease remains unclearwithout conclusive randomised trial evidence butassumptions about survival, the natural history of thedisease, the screening procedures and clinical history can bemade to compare the costs and benefits of screening. In theUS, Eddy33 has evaluated three screening strategies whichresemble proposed Australian screening guidelinesz2O:annual FOBT, annual FOBT plus sigmoidoscopy every fiveyears and annual FOBT plus colonoscopy every five years.The estimated cost-effectiveness ratio for each strategy was$US8,800, $IJS7,760 and $US2O,700 per life year gainedrespectively. However, based on the findings of theMinnesota trial, the cost-effectiveness of FOBT screeningmay be considerably tmderestimated. We have begunpreliminary estimates of the cost-effectiveness of screeningin an Australian context using Australian cancer incidencedata and Australian health care costs. The results of thisanalysis will be reported when available.
CONCLUSION AND RECOMMENDATIONSContrary to the recommendations of The Gut Foundation2,we recommend that screening for colorectal cancer shouldnot be implemented on a population basis as there is notsufficient evidence of its effectiveness. This position willneed review when further trial evidence becomes available.Screening for groups at high risk of colorectal cancer such asthose with a known genetic predisposition seems prudent onthe basis of the increased risk. If evidence accruing from thetrials over the next few years warrants a decision to screen,it must be with the support of general practitioners, withquality control for FOBTs and with adequate linkage toensure appropriate management and follow-up of peoplewith positive results23. In addition, there must bemonitoring and evaluatioll of the screening program'sability to detect early cancers and minimise the intervalcancer rate. In the interim, further investigation in theareas of compliance, psychosocial impact and betterscreening tests, as well as more detailed economicevaluation, is needed.
1. Coates M, McCredie M, Taylor H. Cancer in New South Wales.Incidence and Mortality 1991. Sydney: NSW Central Cancer Registry,1994.2. The Gut Foundation Working Party. Colorectal Cancer: Prevention,Diagnosis, Treatment. Randwick: The Gut Foundation ResearchInstitute, 1993.3. Mandel JS, Bond JH, Bradley M, Snover DC, Church TB, Williams Set al. Sensitivity, Specificity, and Positive Predictivity of the HemoccultTest in Screening for Colorectal Cancers. Ga.so-centeroiogy 1989;97(3):597-600.4. Mandel JS, Bond ,IH, Church TR, Snover DC, Bradley GM, SchumanLM, Ederer F. Reducing mortality from colorectal cancer by screening forferal occult blood. N Engl J Med 1993; 3281191:1365-71,
5. Hardcastle JD, Chir M, Armitage NC, Chamberlain J, Amar SS,James PD, Balfour TW. Focal Occult Blood Screening for ColorectalCancer in the General Population. Results of a Controlled Trial. Cancer1986; 58(21:397-403.6. Hardcastle ID, Chamberlain J, Sheffield J, Balfour TW, Arinitage NC,Thomas 'llIrvI et al. Randomised, Controlled Trial of Faecal Occult BloodScreening for Colorectal Cancer. Results for the First 107,349 Subjects.The Lancet 1989; May 27:1160-4.7. Kewenter J, Bjork S, Haghnd E, Smith L, Svanvik J, Ahren C.Screening and Rescreening for Colorectal Cancer. A Controlled Trial ofFocal Occult Blood Testing in 27,700 Subjects. Cancer 1988; 62131:645-51.8. Kronborg 0, Fenger C, Olsen J, Bach K, Sondergoard 0. RepeatedScreening for Colorectal Cancer with Fecal Occult Blood Test. AProspective Randomized Study at Funen, Denmark. Scand JGastro,enterol 1989; 24:599-606.9. Illaaborg K, Madsen MS, Sondergaard 0, Kronborg 0. Participationin Mass Screening for Colorectal Cancer with Feral Occult Blood Test.&andJGastroenterol 1986; 21:1180-4.10. Kronborg A, Fenger C, Sondergaard 0, Pederen KM Olsen 3. InitialMass Screening for Colorectal Cancer with Focal Occult Blood Teat.Second J Gaslroenterol 1987; 22:677-86.11. Kronborg 0, Fenger C, Worm J, Pedersen SA, Hem J, Bertelsen K,Olsen J. Causes of Death during the First 5 Years of a Randomized Triulof Maes Screening for Colorectal Cancer with Fecal Occult Blood Test.Scand J Ga.ot,'oentsroi 1992; 27:47-52.12. Fiehinger BJ, Herbert B, Winawer SJ, Miller DG. Screening forcolorectal cancer with focal occult blood test and sigmoidoscopy:Preliminary Report of the Colon Project of Memorial Sloan-KetteringCancer Center and PMI-Sto'ang Clinic. In Chamberlain J, Miller AR, eds.Screening for Gastrointestinal Cancer (International Union AgainstCancer). Toronto: Hans Huber, 1988; 9-16.13. Winawer SJ, Flehinger BJ, Schottenfeid D, Miller DO. Screening forColorectal Cancer With Focal Occult Blood Testing and Sigmoidoscopy.JNat Cancer Inst 1993; 85(161:1311-8.14. SeIbyJV, Friedman G, Quesenberry CP, Weiss NS. A Case-ControlStudy of Screening Sigmoidoscopy and Mortality from Colorectal Cancer.NEngi JMecl 1992; 326:653-7.15. Newcomb PA, Norfleet RG, Storer BE, Surawicz TS, Marcus PM.Screening Sigmoidoscopy and Colorectal Cancer Mortality. JN0t CancerInst 1992; 84(201:1752-5.16. Diagnostic and Therapeutic Technology Assessment (DATFA). Rigidand Flexible Sigmoidoscopies. JAMA 1990; 264111:89-92.17. Robinson MHE, Berry DP, Vellacott KD, Moshakie V, Hardcastle JD.A Randomised Trial of Flexible Sigmoidoscopy and Haemoccult VsHaemoccult Alone in Colorectal Cancer Population Screening. Get 1993;34(1):S40.18. Kramer B, Gohagan J, Prorok PC, Smart C. A National CancerInstitute Sponsored Screening Trial for Prostatic, Lung, Colorectal, andOvarian Cancers. Cancer 1993; 71:589-93.19. Hahr-Gama A, Waye JD. Complications and Hazards ofGastrointestinal Endoscopy. World JSurg 1989; 13(21:193-201.20. Allison JE, Feldman R, Tekawa IS. Hemoccult Screening inDetecting Colorectal Neoplasms: Sensitivity, Specificity, and PredictiveVolmAnn Intern Med 1991; 112151:328-33.21. St John DJB, Young GP, Alexeyelf MA, Deacon MC, CuthbertsonAM, Macrae PA, Penfold JOB. Evaluation of New Occult Blood Tests forDetection of Colorectal Neoplasia. Gostroonterolngy 1.993; 104(61:1661-8.22. Thomas DW, Colon cancer detection based on the radialimmunodiffusion test, 'Detectacol'. In: Young GP, Saito H, eds. FaecalOccult Blood Test.s. San Jose: SmithKline Diagnostics Inc 1992; 76-81.23. Australian Gastroenterology Institute and the Auetralian CancerSociety. Guidelines for Early Detection and Prevention of ColorectalCancer. Sydney, 1994. In press.24. Arveux F, Durand G, Milan C, Bedenne L, Levy B, Bui DHD, FaivreJ. Views of a general population on mass screening for colorectal cancer:the Burgandy study. Proc Med 1992; 21:574-81.25. Mant D, Fuller A, Northover J, Astrop P, Chivers A, Crockett A, at al.Patient compliance with colorecta] cancer screening in general practice.Brit J Gee Prod 1992; 42:18-20.26. Fairbrother G, KingJ, Morris DC. The effect of a local communitymedia educational campaign on compliance with faecal occult bloodscreening. Health Prom JAnet. In press.27. King J, Fairbrother G, Thompson C, Morris DC. The influenceof socio-economic status, ethnicity and an educational brochure oncompliance with faecal occult blood testing in Australia. Acost J PublicHealth. In press.28. Weller Dl', Hiller JE, Wilison K, Wilson B, Owen N. Colorecta]cancer and its prevention: knowledge, attitudes and beliefs in the SouthAustralian Population. tJnpublished.29. Mant D, Fitzpatrick R, Hogg A, Fuller A, Farmer A, Verne J,Northover J. Experiences of patients with false positive results fromcolorectai cancer screening. Brit J Can Pract 1990; 40:423-5.30. McCarthy BD, Moskowitz MA. Screening flexible sigmoidoscopy:patient attitudes and compliance, J Can Intern Med 1993; 8:120-5.31. Fox E, O'Boyle C, Lennon 3, Keeling PWN. Trait anxiety and copingstyle as predictors of pre-operative anxiety. Brit J Psychol 1989; 28:89-90.32. Turnbull D, Irwig L, Simpson J. Long-term psychological morbidityafter false positive screening mammograms. Unpublished.33. Eddy DM. Screening for colorertal cancer, Annals ofInternalMedicine 1990; 113:373-84.34. Brown ML. Screening for colorectal cancer etterl. N Engi J Med1993; 3291 181:1352-3.
Vol. 5/No. 3