Evaluation of Possibilities for Mass Screening for Colorectal Cancer with Hemoccult| Fecal Blood Test ULF CARLSSON, M.D., GORAN EKELUND, M.D., RUTGER ERIKSSON, M.D., THOMAS FORK, M.D., LARS JANZON, M.D., LENNART LEANDOER, M.D., CLAS LINDSTROM, M.D., ERIK TRELL, M.D.
From the Maim6 General Hospital, Malta6, Sweden
Carlsson U, Ekelund G, Eriksson R, Fork T, J a n z o n L, Leandoer L, L i n d s t r 6 m C, Trell E. Evaluation of possibilities for mass screening for colorectal cancer with H e m o c c u l t ~ fecal blood test. Dis Colon R e c t u m 1986;29:553-557.
length time bias associated with screening, i.e., that screening preferentially might lead to the detection of slowly growing, and therefore less malignant tumors, the data from noncontrolled trials cannot be used to prove the benefits associated with screening. 6 Sweden has, in comparison with many other countries, a high incidence of colorectal cancer. 7 Malm6, a city in southern Sweden of about 235,000 inhabitants, is one of the areas in Sweden with the highest incidence of colorectal cancer. A pilot study was performed to assess the potential for prevention of death from colorectal cancer by screening with the occult fecal blood test. Adherence to the planned test procedure and the prevalence of colorectal cancer and other neoplastic pathology were assessed. T h e potential screening population would be men and women aged 45 to 69 years. These age limits were chosen to find and remove potentially precancerous lesions a m o n g those who are relatively young, and to achieve the highest possible acceptance rate and the longest followup time. This study assesses the feasibility of a controlled trial to prove that screening with the occult blood test leads to a reduced mortality from colorectal cancer. T o do this, we needed an estimate of the expected rate of participation and the expected number of deaths from colorectal cancer (cause specific mortality) and other disorders among 45to 69-year-old men and women followed for varying lengths of time.
Data from a h e a h h survey i n c l u d i n g the H e m o c c u l r g fecal blood test, together with official cause-speclfic death rates, were used to assess the m a g n i t u d e of a controlled trial that would be required to p r o v e a 25 percent reduction of the mortality from colorectal cancer associated with screening. All m e n in three age groups in the city of Malm6, Sweden, were invited, but 46 percent did not participate in, the Hemoccult screening. O n e carcinoma and 89 adenomas were detected in 56 of the 2422 w h o did. With the risk function used in o u r calculation and a compliance rate of 60 percent, a study p o p u l a t i o n a m o n g 45- to 69year-olds of 605,000 is required to prove a n expected 25 percent reduction of the mortality with 90 percent power. Considering the size of such a trial, we question whether a controlled trial is feasible. With k n o w n risk functions for death from all causes and death from eolorectal cancer, the study p o p u l a t i o n was calculated u s i n g variable statistical power, participation rate, and risk reduction. Statistical methods and c o m p u t e r p r o g r a m s are given. In addition, alternative study models to assess the benefits associated with s c r e e n i n g are discussed. [Key words: Colonic neoplasm; diagnosis; o c c u r r e n c e ; R e c t a l neoplasm; diagnosis; occurrence; Mass screening; Epidemiologic methods]
T H E G U A I A C T E S T WAS I N T R O D U C E D IN 1976 by Greegor as a screening method for the early detection of colorectal cancer.1 Subsequently, this screening test has been adopted in many countries. A c o m m o n experience from these studies is that screening often leads to the detection of carcinomas at a favorable stage. Furthermore, a high frequency of adenomas, which can be considered to be potentially precancerous, also have been detected at screening.2-~ T h e results from screening activities have been viewed with optimism, and regular examinations using occult blood tests have been recommended. Due to the potential
Materials and Methods Pilot Study: Invitations to participate in a general health screening were mailed to all 53-, 56-, and 60-yearold male inhabitants in Malm6~ Altogether 4495 men were invited. Information about the Hemoccult II Test| with three double samples for the detection of occult fecal
Read at the meeting of the American Society of Colon a n d Rectal Surgeons, New Orleans, Louisiana, May 6-11, 1984. Address reprint requests to Dr. Ekelund: Department of Surgery, M a l m 6 General Hospital, IS--214 01 Malm6, Sweden.
553
554
Dis. Col. & Rect.
September1986
CARLSSON, ET AL.
blood was given to the participants after the general health examination. Instructions did not include meat restriction. Samples, which were returned by mail, were rehydrated before analysis. Participants with a positive test were offered complete clinical workups, including physical examination, rigid sigmoidoscopy, and double-contrast b a r i u m enema. If no bleeding source was identified, colonoscopy was performed. If bleeding persisted, patients were studied for possible bleeding sources in the u p p e r gastrointestinal tract. Statistical Methods: T h e expected number of deaths from colorectal cancer a m o n g 1000 45- to 69-year-old subjects chosen at r a n d o m from the p o p u l a t i o n can be calculated if one knows: 1) the risk function for deaths from colorectal cancer, C(t); 2) the risk function for deaths from all causes, D(t); and 3) the age distribution a m o n g 45- to 69-year-old men and women, B(t). In these calculations the official statistics from the Swedish National Bureau of Statistics were used for computing the risk functions B, C, and D. T h e risk function D is based on the published causes of death in 1980. 8 T h e n u m b e r of expected deaths from colorectal cancer after L (number) years of follow-up was computed from the following risk function: 70
L
t
f{f[C(v + t)(exp --fD(v + u)du)ldt B(v)/B }dv 45
0
0
where:
7o
B = f B(t)dt/1000 45 T h e expected number of deaths from colorectal cancer according to this risk function a m o n g 1000 men and women, respectively, followed for L (number) years is given in Table 1. Assuming 25 percent and 50 percent mortality reductions, respectively, and a 5 percent significance level using a two-sided test, the necessary n u m b e r of subjects to prove this effect was calculated assuming variable rates of participation and statistical power. Computations are based
on a comparison of the proportion of deaths from colorectal cancer in the invited g r o u p and in the control g r o u p using a modification of Fisher's exact testy Formulas and a computer program for these computations are presented in the appendix.
Results
T w o thousand four hundred twenty-two men participated in the H e m o c c u h screening, an acceptance rate of 54 percent. T h e test was positive in 270 men (11 percent). Sixteen refused further examination (Fig. 1). Further workups of the remaining 254 participants revealed one carcinoma (Dukes' Stage A), and 89 histopathologically proven adenomas, including seven carcinomas in situ. Neoplastic tumors were found in 56 men. Plausible bleeding sources were identified in all but 55 of the remaining 214 participants, including the 16 w h o refused workups. T h e probability that a positive Hemoccult test represents a neoplastic t u m o r is 21 percent (56/270). In Table 2, the number of participants required to prove 25 and 50 percent risk reductions, respectively, when followed for five and ten years, respectively, using a two-sided test with 5 percent significance level, is given. If one assumes a participation rate of 90 percent, a duration of 10 years, and a risk reduction of 50 percent, the n u m b e r required (using a power of 90 percent) would be approximately 23,000 people. According to our o w n experience, gained from the pilot study, however, computations should instead be based on an expected 60 percent participation rate. If one assumes this rate of participation, a risk reduction of 25 percent, five years' duration, and 90 percent power, 302,613 individuals would be needed in each of the two groups.
4495 i
3107 (69%) Received tests
1388 Nonparticipation
2422 (78o/o) Returned tests
685 Tests not returned
270 ( 1 1 % ) Positive tests
2152 Negative tests
56 (21o/0) Neoplasms proven
214 Neoplasm not proven [
TABLE1. Expected Number of Deaths from Colorectal Cancer among 1000 45- to 69-year-old Men and Women*
Follow-up (Years)
Men
Women
1 2 3 4 5 6 7 8 9 10
0.55 1.13 1.76 2.44 3.17 3.93 4.74 5.59 6.49 7.42
0.43 0.9 1.4 1.95 2.53 3.15 3.8 4.51 5.26 6.05
*Computer program can be requested from L. Janzon.
159 Non-neoplastic possible bleeding sources
FIG. 1. Breakdown of participants in study.
55 Negative findings
Volume29
Number9
FECAL BLOOD TEST
Discussion T h e detection rate and pathologic findings are similar to the results of other screening projects in comparable age groups.l~ T h e number of in situ carcinomas and the early stage of the one detected cancer in the present study, suggest the benefits associated with H e m o c c u h screening. When the history of an individual tumor is not known, however, we cannot, with certainty, claim that the detection of tumors will increase the probability of cure. This can only be proved by a controlled trial. T h e proper question to ask in such a trial would be, Does invitation to participate in a screening program for the early detection of colorectal cancer lead to reduced mortality from the disease in the invited group? In the present analysis, it appears that it will be difficult to prove such a benefit from screening in a controlled trial. This conclusion was reached based on the model we chose, but other study models may be used. ~3 A number of variables may be changed in the calculations, and will influence the result of the analysis. These are ages of participants, acceptance rate, follow-up (observation) time, expected degree of reduction of the specific cause of mortality, and finally, significance level and power of the test, respectively. Ages of Participants: This study focused on people 45 to 69 years old. In this age group, the mortality rate from colorectal cancer is low, and the majority of the participants are relatively young. Among 45- to 49-year-olds, the mortality from colorectal cancer in 1980 was only 9.6 • -5 a m o n g men and 11.1 • 10-5 a m o n g women. Among 65- to 69-year-olds, the colorectal cancer mortality is nine times higher in men and six times higher in women, but the number of deaths from colorectal cancer accounts for only 2 to 5 percent of the total mortality. 8 Furthermore, the participation rate is inversely related to age.~4 T h e expected lower rate of acceptance rate among older participants would reduce the validity of such a trial. Compliance Rate: In the present study, a compliance rate slightly higher than that found in the pilot study was used for calculations. T h e higher the compliance rate, the lower the required number of participants. Rates from 15 to 98 percent have been reported. 15 Most of these figures are based on small studies or selected populations. Results from large-scale studies of unselected population groups are rare. Therefore, a significantly higher compliance rate seems unrealistic. If the test is promoted to the public in mass-marketing campaigns, the risk that the persons in the control group will demand the test increases. If so, the control group will not remain pure. T h e question of how tests for occult blood performed outside the screening project influence the results of the study and the estimation of sample size is thus of great interest, but cannot be answered in this paper. Follow-up Time: T h e "adenoma-carcinoma sequence"
555
TABLE 2. Required Number of Participants at Different Levels of
Compliance, Risk Reduction, and Power Duration
Compliance
(Years)
Risk Reduction
Power
Number of
(Percent) (Percent) Participants
60 60 60
5 5 5
25 25 25
70 80 90
355,554 452,121 605,226
60 60 60
5 5 5
50 50 50
70 80 90
81,173 73,344 138,176
60 60 60
10 10 10
25 25 25
70 80 90
150,453 191,321 256,109
60 60 60
10 10 10
50 50 50
70 80 90
34,349 43,679 58,471
90 90 90
5 5 5
25 25 25
70 80 90
151,251 192,336 257,468
90 90 90
5 5 5
50 50 50
70 80 90
32,440 41,252 55,221
90 90 90
10 10 10
25 25 25
70 80 90
64,004 81,389 108,951
90 90 90
10 10 10
50 50 50
70 80 90
13,727 17,456 23,367
is known to be a slow process, 16 and because colorectal carcinomas grow slowly, ~7a long observation time will enhance the possibilities of evaluating the influence of screening in the study group. Considering the magnitude of such a trial and the practical problems associated with it, however, one questions the design of a study based on a follow-up period exceeding five years. A five-year study in this age group means that the oldest participants at the end of the study will be 75 years of age, and still thus may be evaluated. Furthermore, with a duration exceeding five years, the likelihood increases that other methods for screening may be marketed and will, if used among participants in the program, compromise the original study design.
Degree of Reduction of Mortality, Significance Level, and Power: T h e number of persons required to prove that benefits are associated with screening is heavily influenced by these assumptions. Our computations are based on an expected 100 percent sensitivity, which is optimistic. 15 In the trials evaluating mammographic screening, a 30 percent reduction of the mortality from breast cancer associated with the screening was found. ~8 T h e occult fecal blood test is unlikely to be more sensitive
556
Dis, Col. & Rect.
CARLSSON, ET AL.
t h a n m a m m o g r a p h y ; therefore, w e consider 25 p e r c e n t to be a m o r e l i k e l y e s t i m a t e of the risk r e d u c t i o n t h a n 50 percent. As to the p o w e r of the test, o n e c a n discuss w h a t level to choose. A s c r e e n i n g p r o j e c t of this m a g n i t u d e w o u l d need s u b s t a n t i a l f u n d i n g . It is d i f f i c u l t to justify the a l l o c a t i o n of resources to a n d the e n r o l l m e n t of p e o p l e in a project of this m a g n i t u d e if the p o w e r were to be less t h a n 90 percent. T h u s , different study m o d e l s that m a y reduce the n u m b e r of p a r t i c i p a n t s m a y be used, b u t such studies will result in c o n c l u s i o n s of less p r a c t i c a l value, a n d there w i l l be a n increased risk that the effect of the screening will r e m a i n u n p r o v e n despite s u b s t a n t i a l e c o n o m i c i n p u t . W i t h the m o d e l used in the p r e s e n t study, however, the m a g n i t u d e of the trial w o u l d i m p o s e h u g h practical p r o b l e m s . S u c h a study m i g h t n o t be feasible even, if o n a n a t i o n a l level, a study p o p u l a t i o n of the necessary size m i g h t be recruited. Therefore, o t h e r m e t h o d s a n d sources of i n f o r m a t i o n to assess the benefits of screening s h o u l d be considered. O n e a p p r o a c h m i g h t be to l o o k for a h i g h - r i s k p o p u l a t i o n w i t h a h i g h m o r t a l i t y rate from colorectal cancer, u s i n g a m o d e l s i m i l a r to that used i n the L i p i d Research Clinics C o r o n a r y P r i m a r y P r e v e n t i o n Trial.19, 20 I n d i v i d u a l s at p o t e n t i a l l y h i g h risk for colorectal cancer are, however, rare, a n d w h e n identified, m u s t be f o l l o w e d by o t h e r m e a n s t h a n o c c u l t b l o o d testing. H i g h risks for colorectal cancer are k n o w n to be associated w i t h f a m i l i a l a d e n o m a t o s i s , 21 the C a n c e r F a m i l y S y n d r o m e , 22 l o n g s t a n d i n g i n f l a m m a t o r y b o w e l disease, 23 a n d prev i o u s colorectal cancer or a d e n o m a s . 24,25 T h e relative risk of the last factor is n o t of a m a g n i t u d e that can be used for a study m o d e l based o n a selective screening of a h i g h - r i s k p o p u l a t i o n . 26 T h e present c a l c u l a t i o n s are based o n the cause specific m o r t a l i t y in colorectal cancer. O n l y those p a t i e n t s d y i n g of a n d n o t w i t h colorectal cancer were considered. A n o t h e r p o s s i b l e a p p r o a c h is the case c o n t r o l m e t h o d , w i t h the n u m b e r of patients w h o have died from colorectal cancer a n d the c o r r e s p o n d i n g n u m b e r of l i v i n g c o n trols, m a t c h e d for age, sex, a n d l i v i n g area. A n " O d d s R a t i o , " w h i c h w o u l d d e t e r m i n e the n u m b e r of dead p a t i e n t s w h o h a d been screened for colorectal cancer vs. the n u m b e r w h o h a d n o t been screened, a n d c o m p a r e the result w i t h the n u m b e r of l i v i n g c o n t r o l s to e s t i m a t e h o w m u c h a s c r e e n i n g p r o g r a m u s i n g the o c c u l t fecal b l o o d test w o u l d reduce the relative risk of d y i n g from colorectal cancer. S u c h a retrospective study can be m a d e o n l y if a n a d e q u a t e n u m b e r of p e o p l e have been tested a n d if the records of the tests can be traced. I n Sweden, n e i t h e r of these two r e q u i r e m e n t s has been fullfilled. cancer. S u c h a retrospective study can be m a d e o n l y if an a d e q u a t e n u m b e r of p e o p l e have been tested a n d if the
September1986
records of the tests c a n be traced. I n Sweden, n e i t h e r of these two r e q u i r e m e n t s has been fullfilled. Conclusion
W e h a v e n o evidence that a s c r e e n i n g p r o g r a m w i t h fecal o c c u l t b l o o d test leads to a r e d u c e d m o r t a l i t y rate from colorectal cancer. T o prove that a 25 percent reduction of the m o r t a l i t y from colorectal cancer results f r o m screening, several h u n d r e d t h o u s a n d p e o p l e m u s t be foll o w e d for at least five years. A l t h o u g h this m i g h t be achieved o n a n a t i o n a l level, the p r a c t i c a l p r o b l e m s associated w i t h a m u h i c e n t e r trial w o u l d be immense. Alternative m e t h o d s are discussed, b u t o n l y c o n t r o l l e d trial can prove the benefit of screening. Appendix Statistical Methods:
In a screening study where the aim is to show a reduced mortality from colorectal cancer associated with invitation to the screening program, the proportion of deaths from colorectal cancer in the screened group and in the control group may be compared. A modification of Fisher's exact test9may be used. The number of patients in each group at a certain significance level and power can be computed according to the formula n = {(K + Kl.~/(arcsin(vrP-~l)...... in(x/~2))}Z/2 where P~ and Pz represent the proportion of deaths from colorectal cancer in the two groups and ke is defined as = (fesp(-xz/Z)dx)/~/2rr K
The computer program for this computation is as follows: 10 DEF FNT(X)=l/(l+.3326*ABS(X)) 20 DIM F(20) 30 FOR I=l TO 20:READ F(I):NEXT I 40 FOR I=l TO 20:READ A:F(I)=(F(I)+A)/2:NEXT I 50 DATA .55,1.13,1.76,2.44,3.17,3.93,4.74,5.59,6.49,7.42,8.38,9.36,10.38 60 DATA 11.43,12.49,13.56,14.63,15.70,16.76,17.79 70 DATA. 43,.90,1.40,1.95,2.53,3.15,3.80,4.51,5.26,6.05,6.88,7.75,8.66, 9.61 80 DATA 10.60,11.60,12.63,13.68,14.72,15.77 90 DEF FNQ(X)=(SGN(X)+ 1)/2 100 DEF FNP(X)=(.436183*FNT(X) --.120167*FNT(X)ii2+ .937298*FNT(X)ii3)*(I--2*FNQ(X))*EXP(--X*X/2)/SQR (2*3.14159265#)+FNQ(X) 110 INPUT "POWER (50-99%) ";Q:Q=Q/IOO 120 INPUT "DURATION OF THE STUDY (1-20 YEARS) ";Y 130 INPUT "RISK REDUCING EFFECT (1-99%) ";R:R=R/100 140 INPUT "PARTICIPATING RATE (1-100%) ";S:S=S/IO0 150 GOSUB 260 160 C=.0001*((1--S)+(1--R)*S) | 70 Z=ATN(.01/SQR(.9999)) 180 Z=Z--ATN(SQR(C)/SQR(l-C)) 190 N=(1.96+A)/Z 200 N=N*N/2 210 N=2*.I/F(Y)*N 220 PRINT :PRINT "TOTAL NUMBER OF PATIENTS NEEDED ";INT(N+.5) 230 PRINT "(TWO-SIDED TEST AT THE LEVEL .05)" 240 PRINT 250 GOTO 110 260 D=.I:A=I:B=0
V o l u m e 29
Number9
FECAL BLOOD TEST
270 T=A:P=FNP(T) 280 T=A+D:U----FNP(T) 290 A=A+(Q--P)*D/(U--P) 300 D=D*.8:B=B+I 310 IF B=I THEN 330 320 IF ABS(F--A) < .001 THEN 340 330 F=A:GOTO 270 340 RETURN Data represent the number of expected deaths from col| cancer among 1000 45- to 69-year-old men and women chosen at random from the population followed 1, 2.... to 20 years. These numbers were calculated from the following risk function: 70 L
t
fff[C(v + t)(exp --fD(v + u)du)]dt B(v)/B}dv 45 0
700
B = f B(t)dt/1000 45
C(t) represents the risk function for death from col| cancer, D(t) the risk function for death from all causes, and B(t) the age distribution among 45- to 69-year-old men and women. These risk functions are based on the official statistics from the Swedish National Bureau of Statistics, published in 1980.
References 1. Greegor DH. Diagnosis of large-bowel cancer in the asymptomatic patient. JAMA 1967;201:123-5. 2. Gilbertsen VA, McHugh R, Schuman L, Williams SE. The earlier detection of col| cancers: a preliminary report of the results of the occult blood study. Cancer 1980;45:2899-901. 3. Winawer SJ, Andrews M, Flehinger B, Sherlock P, Schottenfeld D, Miller DG. Progress report on controlled trial of fecal occult blood testing for the detection of col| neoplasia. Cancer 1980;45:2959-64. 4. Gnauck R. Screeningtests nach Darmkrebs: erneuter Vergleich yon Haemoccult| mit hemoFEC | und Rfickblick auf 9 Jahre klinische Erfahrung mit Haemoccuh| Z Gastroenterol 1982;20: 84-92. 5. Hardcastle JD, Farrands PA, Balfour TW, Chamberlain J, Amar SS, Sheldon MG. Controlled trial of faecal occult blood testing in the detection of col| cancer. Lancet 1983;2:1-4. 6. Schottenfeld D. Fundamental issues in cancer screening. In: Winawer S, Schottenfeld D, Sherlock P, eds. Col| cancer: prevention, epidemiology and screening. New York: Raven Press, 1980:167-74. 7. Cancer incidence in Sweden 1980. Stockholm, National Board of Health and Welfare. The Cancer Registry, 1983. 8. Causes of death 1980: official statistics of Sweden. Stockholm, National Bureau of Statistics, 1982. 9. Owen DB. Handbook of statistical tables. London: Pergamon Press, 1962:497.
557
10. Farrands PA, Griffiths RL, Britton DC. The Frome experiment: value of screening for col| cancer. Lancet 1981;1:1231-2. 11. Kurnick JE, Walley LB, Jacob HH, Nakayama L. Col| cancer detection in a community hospital screening program. JAMA 1980;243:2056-7. 12. Gnauck R. Okkultes Blut im Stuhl als Suchtest nach kolorektalem Krebs und prfikanzer6sen Polypen. Z Gastroentero11974; 12:23950. 13. Taylor WF, Fontana RS. Biometric design of the Mayo lung project for early detection and localization of bronchogenic carcinoma. Cancer 1972;30:1344-7. 14. Kronborg O. Mass screening for col| cancer. Scand J Gastroenterol 1984;19:1-5. 15. Simon JB. Occult blood screening for col| carcinoma: a critical review. Gastroenterology 1985;88:820-37. 16. Day DW. The adenoma-carcinoma sequence. Scand J Gastroenterol 1984;104(suppl):99-107. 17. Welin S, Youker J, Spratt JS Jr, et al. The rates and patterns of growth of 375 tumors of the large intestine and rectum observed serially by double contrast enema study (Malta6 technique). AJR 1963;90:673-87. 18. Tabt~r L, Gad A, Holmberg LH, et al. Reduction in mortality from breast cancer after mass screening with mammography. Lancet 1985; 1:829-32. 19. Lipid Research Clinics Program. The Lipid Research Clinics Coronary Primary Prevention Trial Results: I. Reduction in incidence of coronary heart disease. JAMA 1984;251:351-64. 20. Lipid Research Clinics Program. The Lipid Research Clinics Coronary Primary Prevention Trial Results: II. The relationship of reduction in incidence of coronary heart disease to cholesterol lowering. JAMA 1984;251:365-74. 21. Alm T, Licznerski G. The intestinal polyposes. Clin Gastroenterol 1973;2:577-602. 22. Lynch PM, Lynch HT, Harris RE. Hereditary proximal colonic cancer. Dis Colon Rectum 1977;20:661-8. 23. Devroede G J, Taylor WF, Sauer WG, Jackman R J, Stickler GB. Cancer risk and life expectancy of children with ulcerative colitis. N Engl J Med 1971;285:17-21. 24. Heald RJ, Lockhart-Mummery HE. The lesion of the second cancer of the large bowel. Br J Surg 1972;59:16-9. 25. Ekelund GR. Cancer risk with single and multiple adenomas: synchronous and metachronous tumors. In: Winawer S, Schottenfeld D, Sherlock P, eds. Col| cancer: prevention, epidemiology and screening. New York: Raven Press, 1980:151-5. 26. T6rnqvist A, Ekelund G, Leandoer L. Early diagnosis of metachronous col| carcinoma. Aust NZ J Surg 1981;5:442-5. 27. Clarke EA, Anderson TW. Does screening by "Pap" smears help prevent cervical cancer? A case-control study. Lancet 1979;2:1-4. 28. Verbeek AL, Hendriks JH, Holland R, Mravunac M, Sturmans F, Day NE. Reduction of breast cancer mortality through mass screening with modern mammography: first results of the Nijmegen project, 1975-1981. Lancet 1984;1:1222-4. 29. Collette H J, Day NE, Rombach J J, de Waard F. Evaluation of screening for breast cancer in a non-randomized study (the DOM project) by means of a case-control study. Lancet 1984;1:1224-8.