E ec i eness o sc eening o colo ec al
cance wi h a aecal occul -blood es ,
in Finland
J Pi käniemi,
12
K Seppä,
1
M Hakama,
13
O Malminiemi,
4
T Pal a,
5
M-S Vuo is o,
5
H Jä inen,
6
H Paimela,
7
P Pikka ainen,
8
A An ila,
1
L Elo ainio,
1
T Hakulinen,
1
S Ka jalainen,
9
L Pylkkänen,
9
M Rau alah i,
10
T Sa keala,
1
H Ve io,
9
N Malila
13
To ci e: Pi käniemi J ,
Seppä K, Hakama M , e al.
E ec i eness o sc eening o
colo ec al cance wi h a
aecal occul -blood es ,
in Finland. BMJ Open Gas o
2015;2:e000034.
doi:10.1136/bmjgas -2015-
000034
▸Addi ional ma e ial is
a ailable. To iew please isi
he jou nal (h p://dx.doi.o g/
10.1136/bmjgas -2015-
000034).
Recei ed 4 Ma ch 2015
Accep ed 13 May 2015
Fo numbe ed a ilia ions see
end o a icle.
Co espondence o
D J Pi käniemi;
janne.pi kaniemi@cance . i
ABSTRACT
Backg ound: Sc eening o colo ec al cance (CRC)
wi h guaiac-based aecal occul -blood es (FOBT) has
been epo ed o educe CRC mo ali y in andomised
ials in he 1990s, bu no in ou ine sc eening, so a .
In Finland, a la ge andomised s udy on biennial FOB
sc eening o CRC was g adually nes ed as pa o he
ou ine heal h se ices om 2004. We e alua e he
e ec i eness o sc eening as a public heal h policy in
he la ges popula ion so a epo ed.
Me hods: We andomly alloca ed (1:1) men and
women aged 60–69 yea s o hose in i ed o
sc eening and hose no in i ed (con ols), be ween
2004 and 2012. This esul ed in 180 210 subjec s in
he sc eening a m and 180 282 in he con ol a m. In
2012, he p og amme co e ed 43% o he a ge age
popula ion in Finland.
Resul s: The median ollow-up ime was 4.5 yea s
(maximum 8.3 yea s), wi h a o al o 1.6 million
pe son-yea s. The CRC incidence a e a io be ween he
sc eening and con ol a m was 1.11 (95% CI 1.01 o
1.23). The mo ali y a e a io om CRC be ween he
sc eening and con ol a m was 1.04 (0.84 o 1.28),
espec i ely. The CRC mo ali y isk a io was 0.88
(0.66 o 1.16) and 1.33 (0.94 o 1.87) in males and
emales, espec i ely.
Conclusions: We did no ind any e ec in a
andomised heal h se ices s udy o FOBT sc eening
on CRC mo ali y. The subs an ial e ec di e ence
be ween males and emales is inconsis en wi h he
e idence om andomised clinical ials and wi h he
ecommenda ions o se e al in e na ional
o ganisa ions. E en i ou indings a e s ill
inconclusi e, hey highligh he impo ance o
andomised e alua ion when new heal h policies a e
implemen ed.
T ial egis a ion: 002_2010_augus .
INTRODUCTION
Accumula ed e idence om la ge ando-
mised ials
1–6
has shown a mo ali y e ec
o colo ec al cance (CRC) sc eening using
aecal occul -blood es (FOBT). The
a e age educ ion in CRC mo ali y is es i-
ma ed o be 12%, a ying om 10% o
21%, based on he mos ecen me a-
analysis.
7
These ials sugges e en a bigge
educ ion in CRC mo ali y wi h annual
sc eening.
12
In he fi s epo o he UK
s udy,
3
wi h a median ollow-up o 7.8 yea s,
he di e ence in CRC mo ali y be ween
sc eening and con ol a ms was 15%, and
he e ec began o eme ge a e 3–4 yea s
om s udy en y. In he ollow-up o he
same s udy,
4
wi h a 15-yea sc eening
pe iod and almos 15 yea s o ollow-up
a e he sc eening pe iod, a 12% educ ion
in CRC mo ali y was obse ed. In he US
ial,
1
he biennial sc eening a m had a
Summa y box
Wha is al eady known abou his subjec ?
▸Randomised ials wi h guaiac-based aecal
occul -blood es (FOBT) ha e been epo ed o
educe colo ec al cance (CRC) mo ali y in he
1990s.
▸The EU and he US bo h ecommend sc eening
om 50 un il 74 yea s o age.
▸E idence o he e ec i eness o guaiac-based FOB
in sc eening as a pa o ou ine heal h se ice
wi h unselec ed s udy subjec s is missing.
Wha a e he new indings?
▸Ou andomised heal h se ices s udy o FOBT
sc eening on CRC mo ali y ound no e ec .
▸We obse ed a subs an ial e ec di e ence
be ween males and emales, s ill inconclusi e
howe e .
How migh i impac on clinical p ac ice in
he o eseeable u u e?
▸Re-e alua ion o exis ing sc eening p ac ices o
CRC wi h FOBT in cu en heal h se ice p o-
g ammes migh be needed.
▸Be o e applying any new es as ou ine heal h
se ice a andomised e alua ion o he e ec i e-
ness is wa an ed.
Pi käniemi J , Seppä K, Hakama M , e al.BMJ Open Gas o 2015;2:e000034. doi:10.1136/bmjgas -2015-000034 1
Epidemiology
highe CRC mo ali y han he con ol a m, om 4–5
o 8–9 yea s a e s udy en y, and, o e all, a modes
6% educ ion in CRC mo ali y was de ec ed. In he
analysis o he same ial wi h 30 yea s o ollow-up,
2
14 yea s du ing he sc eening pe iod and 16 yea s a e
ha , a 12% educ ion in CRC mo ali y was obse ed. In
he Danish and Swedish ials, he di e ence in he
cumula i e CRC mo ali y eme ged a abou 8 yea s o
ollow-up o la e .
58
Resul s on CRC mo ali y wi h
epea ed FOB-based es ing ha e also been epo ed
om non- andomised s udies in F ance
9
(a 33% educ-
ion) and Ge many
10
(a 34% educ ion).
The US ial (Minneso a) wi h 30 yea s o ollow-up
ound a significan ly di e en educ ion in CRC mo al-
i y be ween males and emales in he biennial sc eening
a m, RR=0.63 (95% CI 0.48 o 0.82) and RR=0.92 (95%
CI 0.72 o 1.18), espec i ely.
2
On he o he hand, in he
UK ial, he educ ion in CRC mo ali y wi h app oxi-
ma ely 20 yea s o ollow-up was simila in males
(RR=0.91, 95% CI 0.82 o 1.02) and emales (RR=0.90,
0.80 o 1.01).
4
Globally, 1.2 million new CRC cases a e diagnosed and
600 000 dea hs a e due o CRC annually.
11
Cance s o
he colon and ec um emain he hi d mos common
( anked by si e specific incidence) in Finland. The e
we e 2904 new CRC cases (ICD-10 classifica ion: C18–
C21) in Finland o 5.42 million inhabi an s and 1161
CRC dea hs in 2012.
12
In Finland, be ween 2001–2012,
CRC mo ali y has dec eased annually on a e age by
1.5% pe yea in males and 0.7% pe yea in emales.
13
E ec i e ools o p ima y p e en ion a e limi ed, and
p e en i e e o s ha e ocused on he de ec ion o CRC
in he ea ly s ages o he umou g ow h p ocess. CRC is
conside ed o de elop ia an adenoma o ca cinoma
sequence.
14
E ec i e sc eening esul s in de ec ion o
adenomas and p eclinical cance s and he ea e in
dec eased CRC mo ali y. A he momen , he EU
15
and
he US
16
bo h ecommend sc eening o CRC s a ing
om 50 un il 74 yea s o age.
A new sc eening p og amme should allow unbiased
e alua ion o he e ec i eness in he a ge popula ion.
17
A se ice p og amme has mo e challenges han ando-
mised ials, and he expec ed e ec is usually smalle
han ha obse ed in andomised ials.
18 19
Se ice p o-
g ammes a e un wi hin he no mal heal h ca e sys em
including mo e a ia ion in he p ocess and wi h limi ed
esou ces and o en less de o ed human esou ces as
compa ed o scien ific ials. Thus, he eal li e applica-
ion, se ice p og ammes, should be e alua ed igo ously
including andomisa ion. This is possible only du ing
he implemen a ion pe iod o he new sc eening p o-
g amme, wi h disease-specific mo ali y as he end
poin .
17
We epo he e he fi s esul s on mo ali y o an indi-
idually andomised communi y-based CRC sc eening
p og amme (a andomised heal h se ices (RHS) s udy)
wi h he FOB based biennial es among 360 000 men
and women in Finland.
METHODS
S udy design
The Finnish popula ion-based sc eening p og amme was
indi idually andomised (1:1) in he implemen a ion
phase by egion, gende and bi h yea o hose o be
in i ed o sc eening (sc eening a m) and hose no
in i ed (con ol a m), based on Cen al Popula ion
Regis e (CPR) da a and consen applied as ca ied ou
in he Finnish heal h se ices. The a ge g oup
included men and women om 60 o 69 yea s o age.
De ails o he s udy design ha e been epo ed
ea lie .
20 21
The wo king g oup o sc eening a he
Finnish Minis y o Social A ai s and Heal h decided
fi s o ecommend a 6-yea implemen a ion pe iod wi h
andom alloca ion o he popula ion o be able o e alu-
a e he e ec s eliably. In 2009, due o low co e age o
he p og amme, he andomisa ion pe iod was ex ended
up o he yea 2014. Thus, he es ima ion o e ec i eness
is needed o heal h policy planning, a e he end o
he ex ended andomisa ion pe iod. A his poin we
need o ake a decision on ei he o apply o con inu-
a ion o he implemen a ion pe iod o o apply o
closing he andomisa ion pe iod.
In sho , om 2004 un il 2012, al oge he 362 165
pe sons we e indi idually andomised ei he o sc een-
ing (181 080 subjec s) o con ol (181 085 subjec s) a m.
All indi iduals in he sc eening a m we e in i ed i hey
had a alid add ess a ailable om he CPR. The p esen
s udy popula ion co e ed app oxima ely 43.5% (in i ees
21.8%) o he whole Finnish a ge popula ion aged 60–
69 yea s a he end o 2012.
22
The CPR has a legisla i e manda e o collec eco ds
on esiden s o Finland including a pe sonal iden ifica-
ion code ha can be used o link da a om a ious
heal h egis e s. The egis e also includes he name,
bi h da e and a alid home add ess; and da es o emi-
g a ion and dea h in case he pe son has mo ed ou side
Finland o died, espec i ely. S a is ics Finland ecei es
dea h ce ifica es o all dea hs and codes he o ficial
cause o dea h na ionally. The Finnish Cance Regis y
(FCR) collec s na ional da a on cance cases since 1953
wi h high co e age, close o 99% o solid umou s.
23
The pe sonal iden ifica ion code is used o link people
be ween hese egis e s.
Assessmen o s udy subjec s and end poin s
Subjec s who died a e he e ie ing o he popula ion
sample om he CPR bu be o e o a he da e o an-
domisa ion we e excluded (94 in o al; 49 in i ees and
45 con ols) om analysis (figu e 1). Simila ly also we
excluded hose who emig a ed (in o al 7; 4 in i ees and
3 con ols). Subjec s who we e diagnosed wi h CRC
be o e o a he da e o he andomisa ion we e also
excluded (1572; 817 in i ees and 755 con ols). In
Janua y 2007, some con ols (109 subjec s) ecei ed a
sc eening in i a ion due o p oblems in he so wa e—in
he ac i i y i sel hese people we e kep in he sc eening
a m. Fo he p esen analysis, hese people we e
2Pi käniemi J , Seppä K, Hakama M , e al.BMJ Open Gas o 2015;2:e000034. doi:10.1136/bmjgas -2015-000034
Open Access
included in he o iginal con ol a m. In he final analysis
we had a o al o 360 492 subjec s (180 210 in he sc een-
ing a m and 180 282 in he con ol a m).
Inciden CRC cases and dea hs om CRC we e
e ie ed by linkage wi h he FCR and defined acco ding
o he ICD-O-3 classifica ion opog aphy codes C18.0–
C21.2 and C26.0 i malignan beha iou , excluding
lymphomas (mo phology codes ≥9590) and anal epi-
de moid cance s ( opog aphy C21.0–C21.2 and mo ph-
ology code 8070). In o ma ion abou he i al s a us a
he end o 2012, including da e o dea h and emig a-
ion, was ecei ed om he CPR, and cause o dea h was
ob ained om S a is ics Finland h ough eco d linkage
wi h he unique pe sonal iden ifica ion numbe .
Ou p ima y end poin in he cu en s udy was dea h
om CRC defined by he o ficial classifica ion o dea h
by S a is ics Finland. Dea h due o any cause was he end
poin in assessing he excess (all-cause) mo ali y among
pa ien s wi h CRC (diagnosed a e he da e o andom-
isa ion). We define a sc een-de ec ed CRC e en o be a
cance diagnosed wi hin 6 mon hs om he sc eening
es . An in e al cance is one ha is de ec ed a e he
6 mon h pe iod om he sc eening es bu be o e he
nex sc een.
Sc eening p ocess
Those in i ed o sc eening e e y second yea , we e admi-
nis e ed he guaiac-based non- ehyd a ed FOBT a home
a e ecei ing h ee es -ca ds (Hemoccul ) and ins uc-
ions by pos al mail. A aecal sample was ins uc ed o be
collec ed h ee imes wi hin 1 week o he fi s sample by
ob aining wo smea s om di e en loca ions o he
aeces. Die a y es ic ions consis ed o a oiding aw
mea , blood and li e dishes 3 days be o e sample- aking
and du ing he ime o sampling. Also, i amin C supple-
men s wi h mo e han 250 mg o i amin we e no o be
used. Tes ca ds we e e u ned by mail o be analysed a
he na ional sc eening cen e a Pi kanmaa Cance
Socie y in Tampe e. One cen al labo a o y co e ed all o
Finland and samples we e analysed wi hin 14 days o
sample- aking. Resul s om es ing we e no ified ia
pos al mail o all a ende s. In case o blood in any o he
samples ( es posi i e), a egional con ac nu se was also
in o med. The con ac nu se in e iewed (phone mos ly)
he es -posi i e pa icipan and he ea e o ganised a
ull colonoscopy examina ion. Colonoscopies we e pe -
o med egionally ei he a he heal h cen e, p i a e
clinics o hospi als, by expe ienced physicians ha ing
ex ensi e aining in colonoscopy (mos ly specialised gas-
oen e ologis s). The o e all compliance o colonoscopy
was 84% and he annual compliance a ied om 81% o
90% by calenda yea . Colonoscopies o con ols and in
in i ees no a ending sc eening as well as hose esul ing
as in e al cance s we e pe o med by he same p o ide s
as he sc een induced ones.
S a is ical analysis
O iginal sample size calcula ion was based on 90%
powe and a 5% ype I e o i he ue e ec was 20%
20
in CRC mo ali y. These assump ions implied ha we
would need o accumula e 1.6 million pe son-yea s in
he cu en s udy. Incidence and mo ali y a es we e
es ima ed by di iding he espec i e numbe s o dea h
om any cause, om CRC and om causes o he han
CRC, by he numbe o pe son-yea s. The numbe o
pe son-yea s is he sum o each subjec ’s ime a isk. In
Figu e 1 Flowcha o he numbe o subjec s andomised, excluded and in he analysis da ase .
Pi käniemi J , Seppä K, Hakama M , e al.BMJ Open Gas o 2015;2:e000034. doi:10.1136/bmjgas -2015-000034 3
Open Access
he es ima ion o incidence, he ime a isk was calcu-
la ed om andomisa ion o CRC diagnosis, dea h, emi-
g a ion, o o he end o 2012, whiche e came fi s . In
es ima ion o he mo ali y, he ime a isk was calcu-
la ed om andomisa ion un il dea h, emig a ion, o o
he end o 2012, whiche e came fi s . The a ios o he
mo ali y a es and o he excess mo ali y a es we e es i-
ma ed wi hin he in en ion- o- ea p inciple o measu e
he e ec o he se ice sc eening. Confidence in e als
(CIs) we e es ima ed assuming he obse ed numbe o
dea hs o ollow a Poisson p obabili y law. The di e ence
in he e ec o mo ali y om CRC be ween males and
emales was es ed using he Cox p opo ional haza ds
model and he classical likelihood a io es . Res ic ed
cubic spline unc ions
24 25
wi h a ime-dependen e ec
o in i a ion we e fi ed o indi idual-le el ollow-up da a
o desc ibing a m-specific CRC mo ali y a es (ie, he
haza d o dea h om CRC) and hei a io o e
ollow-up ime. Cumula i e p opo ion o dea hs om
CRC was es ima ed in a compe ing isk se ing using a
weigh ed empi ical cumula i e dis ibu ion unc ion.
26
The excess mo ali y a e was es ima ed by di iding he
excess numbe o dea hs obse ed in pa ien s wi h CRC
by he o al numbe o pe son-yea s in each a m.
27 28
De ails o es ima ion o excess mo ali y a e and i s a i-
ance using he del a me hod a e desc ibed in online
supplemen a y appendix. When ollow-up was s a ed
1 yea a e andomisa ion in o de o emo e he
pe iod wi h no po en ial e ec o sc eening, ou esul s
did no change and, hus, he esul s a e p esen ed wi h
ull ollow-up s a ing om he da e o andomisa ion.
E hics
The RHS s udy on implemen a ion o CRC sc eening
was app o ed by he Minis y o Social A ai s and
Heal h in 2004 (STM/42/07/2004) and upda ed in
2010 by he o ficial au ho i y, he Na ional Ins i u e o
Heal h and Wel a e (THL/619/5.05.00/2010). The
s udy has been egis e ed in he egis y o RHS s udies
main ained by he Cance Socie y o Finland (h p://
www.cance .fi/ hs/002_2010_augus /).
RESULTS
The backg ound cha ac e is ics be ween he sc eening
and con ol a ms we e in balance ( able 1). App oxima ely
20 000 subjec s we e assigned annually ei he o he
sc eening o he con ol a m esul ing in 180 000 in i ees
by he end o 2012 and a simila numbe o con ols. The
median ollow-up ime in ou s udy was 4.5 yea s ( ange
0.0–8.3), wi h 25% o subjec s (app oxima ely 90 000
subjec s) wi h a leas 6.5 yea s o ollow-up. The longes
ollow-up ime was 8.3 yea s.
In all, close o 440 000 in i a ions we e sen be ween
2004 and 2012, wi h an up ake o 68.8% (61.5% among
males and 76.0% among emales; able 2). The p opo -
ion o FOB posi i e es s was 3.6% o all es s (4.7%
among males and 2.7% among emales). The p opo -
ion o sc een-de ec ed CRCs was 42.7% o all CRCs
(41.7% among males and 43.9% in emales) ( able 2).
Colonoscopy was pe o med in 84% o sc een posi i es.
Al oge he , 73% ou o he o al numbe o 1.6 million
pe son-yea s we e accumula ed du ing he fi s 4 yea s
a e andomisa ion ( able 3). The incidence a e o
CRC was highe in he sc eening a m compa ed o he
con ol a m: 112.4/100 000 pe son-yea s in he sc een-
ing a m and 100.9/100 000 pe son yea s in he con ol
a m ( able 3). The CRC incidence a e a io be ween
he sc eening and con ol a m was 1.11 (95% CI 1.01 o
1.23). In bo h a ms, he incidence a e o CRC was
highe in males (133.1/100 000 in he sc eening a m
and 120.8/100 000 in he con ol a m) han in emales
(92.4 and 81.7/100 000). The CRC incidence a e a io
o males was 1.10 (0.97, 1.25) and o emales, 1.13
(0.98, 1.31).
The numbe s o new CRC cases showed a subs an ial
di e ence in he fi s 2-yea in e al a e andomisa ion:
84 mo e people we e diagnosed wi h CRC in he sc een-
ing han in he con ol a m ( able 4).
A o al o 15 963 dea hs occu ed du ing he ollow-up
( able 3), 8000 dea hs in he sc eening a m and 7963 in
he con ol a m. We did no find any di e ence be ween
s udy a ms in o e all mo ali y a e ( a e a io, RR, 1.00;
95% CI 0.97 o 1.04) o in mo ali y a e om o he
Table 1 Numbe s and p opo ions (%) o subjec s aged 60–64 yea s and andomised o sc eening and con ol a ms
acco ding o calenda yea , sex and s udy a m
Sc eening Con ol
Calenda yea Men Women To al Men Women To al
2004 2132 (2.4) 2389 (2.6) 4521 (2.5) 2136 (2.4) 2377 (2.6) 4513 (2.5)
2005 11 581 (12.9) 11 853 (13.1) 23 434 (13.0) 11 578 (12.9) 11 872 (13.1) 23 450 (13.0)
2006 12 391 (13.8) 12 368 (13.7) 24 759 (13.7) 12 403 (13.8) 12 378 (13.7) 24 781 (13.7)
2007 10 685 (11.9) 10 691 (11.8) 21 376 (11.9) 10 689 (11.9) 10 694 (11.8) 21 383 (11.9)
2008 11 873 (13.2) 12 264 (13.6) 24 137 (13.4) 11 875 (13.2) 12 235 (13.5) 24 110 (13.4)
2009 12 036 (13.4) 12 153 (13.4) 24 189 (13.4) 12 078 (13.4) 12 141 (13.4) 24 219 (13.4)
2010 9919 (11.1) 9898 (10.9) 19 817 (11.0) 9946 (11.1) 9887 (10.9) 19 833 (11.0)
2011 8841 (9.9) 8807 (9.7) 17 648 (9.8) 8830 (9.8) 8815 (9.7) 17 645 (9.8)
2012 10 254 (11.4) 10 075 (11.1) 20 329 (11.3) 10 272 (11.4) 10 076 (11.1) 20 348 (11.3)
To al 89 712 (100) 90 498 (100) 180 210 (100) 89 807 (100) 90 475 (100) 180 282 (100)
4Pi käniemi J , Seppä K, Hakama M , e al.BMJ Open Gas o 2015;2:e000034. doi:10.1136/bmjgas -2015-000034
Open Access
causes han CRC (RR 1.00; 0.97 o 1.04). The o e all
mo ali y a e in sc eening and con ol a ms was simila
in emales (RR 1.00; 0.95 o 1.06) and in males (RR
1.01; 0.97 o 1.05). The e was no di e ence be ween
sc eening and con ol a ms in o he cause mo ali y in
emales (RR 0.99; 0.94 o 1.05) o in males (RR 1.01;
0.97 o 1.05). CRC was he cause o dea h o 170
pe sons in he sc eening a m and 164 pe sons in he
con ol a m ( able 3). We did no find any di e ence in
CRC mo ali y a e be ween he sc eening and he
con ol a m (RR 1.04; 0.84 o 1.28). In males, he CRC
mo ali y a e a io was 0.88 (0.66 o 1.16) and in
emales, 1.33 (0.94 o 1.87). The in e ac ion be ween
s udy a m and gende in CRC mo ali y a e was o bo -
de line significance (p=0.06). The excess mo ali y a es
ga e simila es ima es: RR 1.05 (0.84 o 1.30) in males
and emales combined, and 0.94 (0.71 o 1.24) in males
and 1.26 (0.88 o 1.80) in emales. Figu e 2 shows he
cumula i e CRC mo ali y p opo ion pe 100 000
pe sons in sc eening and con ol a ms o up o
8.3 yea s om s udy en y. The e was no di e ence in
cumula i e CRC mo ali y be ween sc eening and
con ol a m. In emales he cumula i e CRC mo ali y in
he sc eening a m was consis en ly la ge han in he
con ol a m, while in males cumula i e CRC mo ali y in
he sc eening a m was smalle han in he con ol a m.
Smoo hed CRC mo ali y a es and CRC mo ali y a e
a ios as a unc ion o ollow-up ime o sc eening and
con ol a ms by gende a e shown in figu e 3. The e was
no consis en pa e n in he a e a ios by sex and
ollow-up ime. The di e ence be ween he a ms in he
o e all numbe s o CRC dea hs was small; a mos , h ee
dea hs pe 2 yea in e al (excep ea ly dea hs in in e al
0–2 yea s) ( able 4).
DISCUSSION
Ou communi y-based RHS s udy did no show any di -
e ence in CRC mo ali y be ween a ms (RR=1.04).
Despi e he es ima ed 12% (RR=0.88) educ ion in CRC
mo ali y a e ound in males and 33% (RR=1.33)
inc ease in emales, nei he o hese a e a ios was s a is-
ically significan . The di e ence be ween males and
emales in CRC mo ali y a e a io was o bo de line sig-
nificance. Ou es ima es o he excess mo ali y a e due
o CRC we e simila o he abo e esul s.
Ou s udy is a RHS s udy in con as o a scien ific an-
domised con olled ial (RCT). Majo di e ence
be ween he wo ypes o s udies is in he o igin o he
obse a ions, hey a e ei he a by-p oduc o a ou ine
ac i i y (RHS) o hey a e specifically designed o a
esea ch pu pose (RCT). The e o e, RHS is pa icula is-
ic and based on ou ine heal h se ices, whe eas he
objec i e o a RCT is abs ac and gene al, ha is, scien-
ific. Randomisa ion and iming a e simila bu e hics,
unding, blinding and se e al o he aspec s di e
be ween RHS and RCT.
17
Based on in o ma ion om he ea lie RCTs,
29
he
e ec o sc eening in educing CRC mo ali y is likely o
be obse ed a e a longe ollow-up han ha in he
cu en s udy, wi h a median 4.5 yea s (maximum o
8.3 yea s) o ollow-up. This is suppo ed by he esul s o
Mandel e al,
1
who ound only a 6% educ ion in CRC
mo ali y a e 13 yea s o ollow-up wi h biennial sc een-
ing, and he educ ion was 21% in a la e analysis.
2
O iginally, ou s udy was planned o find a s a is ically sig-
nifican educ ion in CRC mo ali y i he ue e ec had
been 20%. Acco ding o he ecen me a-analysis, he es i-
ma e o he mo ali y e ec was 15%.
7
A longe ollow-up
would educe unce ain y in ou e ec es ima e and
na ow he CI. The e o e, i is possible ha he beneficial
e ec o biennial sc eening wi h FOBT on CRC mo ali y
eme ges only a e 6–10 yea s o ollow-up.
1
The excess mo ali y a e measu es bo h he di ec
and indi ec mo ali y due o CRC.
27
Unlike CRC mo -
ali y, he excess mo ali y does no depend on he eli-
abili y o classifica ion o dea hs, as i compa es all-cause
mo ali y in pa ien s wi h CRC wi h ha in compa able
CRC- ee pe sons. Because he es ima e o CRC mo al-
i y a e was simila o ha o he excess mo ali y, he
non-exis ing sc eening e ec is unlikely o be due o mis-
classifica ion o he causes o dea h.
The need o analysis o he sc eening e ec a hand
was spelled ou wi h he Minis y o Social A ai s and
Heal h in Finland, in an ag eemen on e alua ion a e
he implemen a ion phase o he p og amme and
be o e making any decision on i s u u e. The imple-
men a ion phase o he sc eening p og amme was based
on g adual expansion in he fi s 6 yea s (2004–2009)
and on ollow-up o fi e mo e yea s (2010–2014).
Ou s udy is a pa o Finnish public heal h and he e-
o e many ac o s di e compa ed o ea lie sc eening
ials.
13
In ou ine se ice sc eening, he e is mo e a i-
a ion in a ailable esou ces (bo h human and ma e ial),
he popula ion may be less mo i a ed, and expe s a e
no as de o ed and do no ollow he guidelines as
s ic ly as in scien ific ials. Despi e he high compliance
(70%), selec ion among ou s udy subjec s canno be
Table 2 Compliance o sc eening and es esul s by sex in 2004–2012
Colo ec al cance s
Gende In i a ions Sc eens (%) Posi i e es s (%) Sc een de ec ed (%) O he
Males 216 393 133 029 (61.5) 6215 (4.7) 220 (41.7) 305
Females 222 281 168 871 (76.0) 4528 (2.7) 166 (43.9) 212
To al 438 674 301 900 (68.8) 10 743 (3.6) 386 (42.7) 517
Pi käniemi J , Seppä K, Hakama M , e al.BMJ Open Gas o 2015;2:e000034. doi:10.1136/bmjgas -2015-000034 5
Open Access

Table 3 Desc ip i e and compa a i e s a is ics o colo ec al cance incidence and mo ali y, by s udy a m and sex
All Males Females
Sc eening Con ol Sc eening Con ol Sc eening Con ol
Numbe o pe sons 180 210 180 282 89 712 89 807 90 498 90 475
Pe son-yea s 805 480 805 693 395 614 395 851 409 866 409 843
Dea hs
All causes 8000 7963 5486 5453 2514 2510
Colo ec al cance 170 164 93 106 77 58
Pa ien s wi h colo ec al cance
Numbe o pa ien s 903 811 525 477 378 334
Incidence a e pe 100 000 pe son-yea s 112.4 100.9 133.1 120.8 92.4 81.7
Pe son-yea s 2285 1805 1318 1016 967 790
Dea hs 202 190 123 127 79 63
Expec ed numbe o dea hs* 25.0 21.1 18.6 15.7 6.4 5.5
Excess numbe o dea hs†177.0 168.9 104.4 111.3 72.6 57.5
CRC incidence a e a io (95% CI) 1.11 (1.01 o 1.23) 1 1.10 (0.97 o 1.25) 1 1.13 (0.98 o 1.31) 1
Mo ali y a es pe 100 000
All causes 993 988 1387 1378 613 612
Non-colo ec al cance causes 972 968 1363 1351 595 598
Colo ec al cance 21.1 20.4 23.5 26.8 18.8 14.2
Excess mo ali y due o CRC 21.7 21.0 26.4 28.1 17.7 14.0
Mo ali y a e a ios
All causes 1.00 (0.97 o 1.04) 1 1.01 (0.97 o 1.05) 1 1.00 (0.95 o 1.06) 1
Non-colo ec al cance causes 1.00 (0.97 o 1.04) 1 1.01 (0.97 o 1.05) 1 0.99 (0.94 o 1.05) 1
Colo ec al cance 1.04 (0.84 o 1.28) 1 0.88 (0.66 o 1.16) 1 1.33 (0.94 o 1.87) 1
Excess mo ali y due o CRC 1.05 (0.84 o 1.30) 1 0.94 (0.71 o 1.24) 1 1.26 (0.88 o 1.80) 1
*Calcula ed acco ding o he mo ali y o colo ec al cance - ee pe sons s a i ied by sex, age, calenda yea , a m and pa icipa ion s a us.
†The di e ence be ween he obse ed and he expec ed numbe o dea hs.
CRC, colo ec al cance .
6Pi käniemi J , Seppä K, Hakama M , e al.BMJ Open Gas o 2015;2:e000034. doi:10.1136/bmjgas -2015-000034
Open Access
uled ou . In andomised scien ific ials, he eligibili y
o s udy subjec s is gua an eed ei he by inclusion o
exclusion c i e ia as pa o he s udy p o ocol. These
selec i e p ocesses also in oduce a possibili y o incom-
pa abili y when he es is applied o an unselec ed
a ge popula ion.
The e is also subs an ial di e ence conce ning a ious
pa ien and heal h se ices cha ac e is ics be ween he
old ials and he cu en s udy. Su i al o pa ien s wi h
CRC has imp o ed subs an ially om he ime o he
ea ly clinical ials o he beginning o 2000, om 40%
o close o 60% in all No dic coun ies,
13
and a s eady
imp o emen o 5-yea su i al om colon and ec al
cance has been obse ed in de eloped coun ies.
30 31
Such imp o emen in su i al o he con ols leads o a
smalle di e ence in CRC mo ali y be ween ando-
mised g oups and hus lowe s a is ical powe o de ec
i .
The up ake o sc eening was ela i ely high in ou p o-
g amme. Usually, he up ake is lowe in ou ine applica-
ion (RHS) han in con olled ials (RCT). In Finland,
he up ake was 69%. I was less han ha epo ed om
he US ial (Minneso a Colon Cance Con ol S udy)
1
(78%), bu be e han ha o he No ingham S udy
3
(60%). The e o e, i is no likely ha ou lack o signifi-
can e ec can be explained by lowe pa icipa ion.
In he cu en s udy, he sensi i i y o FOBT imp o ed
a e he fi s yea s o sc eening.
21
The mos conclusi e
da a a e p o ided by he longes ollow-up. Un o una ely,
hese da a had, a he same ime, he poo es sensi i i y.
This indica es a lea ning cu e. Up ake in he Bowel
Cance Sc eening P og am
32
in England was 57% in he
fi s ound, 61% in he second and 66% in he hi d
Table 4 Numbe o pe sons ali e a he beginning o ou consecu i e ollow-up ime in e als and numbe s o pe son-yea s,
colo ec al cance cases and dea hs om colo ec al cance , and om o he causes, du ing each ollow-up ime in e al, by sex
and a m
All Males Females
Follow-up in e al Sc eening Con ol Sc eening Con ol Sc eening Con ol
Pe sons ali e a he beginning o ollow-up in e al
[0, 2) 180 210 180 282 89 712 89 807 90 498 90 475
[2, 4) 139 696 139 765 68 865 68 950 70 831 70 815
[4, 6) 100 060 100 034 48 706 48 693 51 354 51 341
[6, 8.3) 60 993 61 022 29 537 29 566 31 456 31 456
Pe son-yea s
[0, 2) 335 031 335 127 166 096 166 222 168 935 168 905
[2, 4) 251 879 251 992 123 648 123 745 128 231 128 247
[4, 6) 157 381 157 374 76 607 76 603 80 774 80 771
[6, 8.3) 61 189 61 201 29 264 29 282 31 925 31 920
Colo ec al cance cases
[0, 2) 371 287 218 162 153 125
[2, 4) 259 264 148 167 111 97
[4, 6) 194 185 114 107 80 78
[6, 8.3) 79 75 45 41 34 34
Dea hs om colo ec al cance
[0, 2) 42 35 27 21 15 14
[2, 4) 50 51 20 33 30 18
[4, 6) 49 52 23 33 26 19
[6, 8.3) 29 26 23 19 6 7
Dea hs om o he causes
[0, 2) 2815 2781 1965 1948 850 833
[2, 4) 2426 2417 1671 1669 755 748
[4, 6) 1787 1797 1239 1195 548 602
[6, 8.3) 802 804 518 535 284 269
Figu e 2 Cumula i e colo ec al cance mo ali y p opo ion
(pe 100 000 pe sons) by s udy a m (sc eening a m wi h solid
line and con ol a m wi h dashed line) and sex.
Pi käniemi J , Seppä K, Hakama M , e al.BMJ Open Gas o 2015;2:e000034. doi:10.1136/bmjgas -2015-000034 7
Open Access
ound. A simila inc ease in up ake by ound was also
obse ed in ou p og amme.
33
The imp o ed sensi i i y
and be e up ake bo h indica e a po en ial imp o emen
in he e ec in he u u e.
The ideal design would ha e been o compa e FOBT,
aecal immunochemical es (FIT) and no sc eening.
The epo s on FIT posi i i y p opo ion a y om 4% o
10%,
34 35
high o ou ine sc eening wi h colonoscopy in
sc een posi i es. I is possible ha FIT would ha e shown
a mo ali y educ ion e en a he same posi i i y p opo -
ion h eshold as ou FOBT (3%), because in such a case
mo e indi iduals wi h bleeding had been iden ified. The
eason o bleeding is, howe e , unknown. The e is no
di ec e idence on imp o ed e ec i eness o FIT com-
pa ed o FOBT, o he ime being.
Con amina ion o he con ol a m by wild sc eening is
likely o be small in Finland. CRC sc eening o a e age
isk indi iduals wi hou symp oms is no common and
egula GP checking o heal hy indi iduals is no a
common p ac ice. Faecal occul blood es s a e used
mainly in clinical se ings o ollow-up o bowel diseases
o as pa o diagnos ics, no in sc eening pu poses. In
summa y, he e ec i eness o ou ine sc eening o CRC
wi h FOBT s ill emains open.
Se e al in e na ional bodies, including he Ame ican
Cance Socie y
29
and he EU,
15
ecommend ou ine
sc eening o CRC wi h FOBTs. The ecommenda ions
a e based on e idence om andomised s udies,
13568
bu hey do no include he e alua ion o he p ocess as
pa o he ou ine heal h se ice sys em. The Finnish
p og amme is consis en in p ocess esul s and no sig-
nifican ly di e en in he mo ali y ou come wi h e i-
dence a ailable so a .
The di e ence be ween males and emales in he
e ec on CRC mo ali y was o bo de line significance.
Th ee ials
246
and wo case–con ol s udies
910
epo ed
educ ion in CRC mo ali y o males and emales sepa -
a ely. The educ ion in CRC mo ali y in males was
be ween 5% and 37%, and in emales be ween 8% and
26%. In ou s udy, he educ ion in CRC mo ali y in
males (12%) is consis en wi h he p e ious s udies.
Howe e , we a e conce ned wi h he obse ed 33%
inc ease in CRC mo ali y in emales, which has no
been epo ed by any o he o he s udies. Rega ding
biennial FOBT, he e is conce n on poo sensi i i y o
he guaiac FOBT in Finland, especially in women.
21 36
Also, he No wegian sc eening s udy on flexible sigmoid-
oscopy and FOBT ei he combined o wi h flexible
Figu e 3 Smoo hed colo ec al cance mo ali y a e (pe 100 000 pe son-yea s) in he sc eening and con ol a m o e ollow-up
ime, and a io o he mo ali y a es by s udy a m and sex. Do ed lines show he limi s o 95% CIs o he a e a io.
8Pi käniemi J , Seppä K, Hakama M , e al.BMJ Open Gas o 2015;2:e000034. doi:10.1136/bmjgas -2015-000034
Open Access
sigmoidoscopy alone, epo ed a di e en e ec on mo -
ali y be ween men and women
37
: sc eening was e ec i e
in men (RR o CRC mo ali y 0.58, 95% CI 0.40 o 0.85)
bu no in women (RR 0.91, 0.64 o 1.30) a e a median
ollow-up o 10.9 yea s. A simila e ec in CRC mo ali y
was obse ed a e adenoma emo al in males and
emales: he SMR o CRC was 0.86 (0.74 o 1.00) and
1.06 (0.93 o 1.22), espec i ely.
38
The e a e di e en hypo heses as o why women may
no benefi as much as men om sc eening wi h
FOBTs,
2
including p opo ionally mo e adenomas and
CRCs in he p oximal colon in men han in women, and
ha he biology may be di e en (sessile se a ed
adenoma pa hway) be ween men and women.
2
Also,
men ha e mo e como bidi y om o he se ious heal h
p oblems han women a his age, and i has been specu-
la ed ha i may lead o di e ences in cause o dea h
coding be ween men and women. This was no he case
in ou ma e ial, howe e .
We ound a subs an ial di e ence in up ake be ween
males (62%) and emales (76%). I is ob ious ha pa -
icipa ion is di e en ly selec i e in men and in women.
We ha e ea lie epo ed, o example, ha ma i al
s a us was o high impo ance in up ake, especially in
men; ma ied men pa icipa e mo e o en han hose
who a e single.
39
Howe e , his di e ence in up ake
should inc ease a he han accoun o he di e ence
in e ec by gende . The di e ence in he CRC mo ali y
e ec be ween males and emales highligh s he need
o a eanalysis wi h longe ollow-up.
We did no find any e ec in a RHS s udy o FOBT
sc eening on CRC mo ali y. The subs an ial e ec di e -
ence be ween males and emales is inconsis en wi h he
e idence om andomised clinical ials and wi h he
ecommenda ions o se e al in e na ional o ganisa ions.
E en hough ou findings a e inconclusi e, hey high-
ligh he impo ance o andomised e alua ion when
new heal h policies a e implemen ed.
Au ho a ilia ions
1
Finnish Cance Regis y, Ins i u e o S a is ical and Epidemiological Cance
Resea ch, Helsinki, Finland
2
Depa men o Public Heal h, Uni e si y o Helsinki, Finland
3
School o Heal h Sciences, Uni e si y o Tampe e, Tampe e, Finland
4
Fimlab Labo a o ies, Tampe e, Finland
5
Pi kanmaa Cance Socie y, Tampe e, Finland
6
Depa men o Gas oen e ological Su ge y, Helsinki Uni e si y Hospi al,
Helsinki, Finland
7
Depa men o Gas oin es inal Su ge y, Uni e si y Hospi al o No he n
No way, Na ik, No way
8
Depa men o Medicine, Tampe e Uni e si y Hospi al, Tampe e, Finland
9
The Cance Socie y o Finland, Helsinki, Finland
10
Finnish Medical Socie y Duodecim, Helsinki, Finland
Con ibu o s All au ho s ead and app o ed he inal e sion o he
manusc ip . Speci ic au ho con ibu ions we e as ollows: NM, MH we e
in ol ed in he s udy concep and design. NM, TP we e in ol ed in he
pa icipan ec ui men and cha ac e isa ion. OM, TP, M-SV we e in ol ed in
he implemen a ion o sc eening es . KS, JP we e in ol ed in he da a and
s a is ical analyses. All au ho s we e in ol ed in he manusc ip d a ing.
Funding This esea ch ecei ed no speci ic g an om any unding agency in
he public, comme cial o no - o -p o i sec o s.
Compe ing in e es s None decla ed.
P o enance and pee e iew No commissioned; ex e nally pee e iewed.
Da a sha ing s a emen No addi ional da a a e a ailable.
Open Access This is an Open Access a icle dis ibu ed in acco dance wi h
he C ea i e Commons A ibu ion Non Comme cial (CC BY-NC 4.0) license,
which pe mi s o he s o dis ibu e, emix, adap , build upon his wo k non-
comme cially, and license hei de i a i e wo ks on di e en e ms, p o ided
he o iginal wo k is p ope ly ci ed and he use is non-comme cial. See: h p://
c ea i ecommons.o g/licenses/by-nc/4.0/
REFERENCES
1. Mandel JS, Bond JH, Chu ch TR, e al. Reducing mo ali y om
colo ec al cance by sc eening o ecal occul blood. Minneso a
Colon Cance Con ol S udy. N Engl J Med 1993;328:1365–71.
2. Shauka A, Mongin SJ, Geisse MS, e al. Long- e m mo ali y
a e sc eening o colo ec al cance . N Engl J Med 2013;369:
1106–14.
3. Ha dcas le JD, Chambe lain JO, Robinson MH, e al. Randomised
con olled ial o aecal-occul -blood sc eening o colo ec al cance .
Lance 1996;348:1472–7.
4. Schole ield JH, Moss SM, Mangham CM, e al. No ingham ial o
aecal occul blood es ing o colo ec al cance : a 20-yea ollow-up.
Gu 2012;61:1036–40.
5. Lindholm E, B e inge H, Haglind E. Su i al bene i in a andomized
clinical ial o aecal occul blood sc eening o colo ec al cance .
B J Su g 2008;95:1029–36.
6. Jø gensen OD, K onbo g O, Fenge C. A andomised s udy o
sc eening o colo ec al cance using aecal occul blood es ing:
esul s a e 13 yea s and se en biennial sc eening ounds. Gu
2002;50:29–32.
7. Massa NJ, Moss SM, Hallo an SP, e al. Sc eening and p ima y
p e en ion o colo ec al cance : a e iew o sex-speci ic and
si e-speci ic di e ences. J Med Sc een 2013;20:125–48.
8. K onbo g O, Fenge C, Olsen J, e al. Randomised s udy o
sc eening o colo ec al cance wi h aecal-occul -blood es . Lance
1996;348:1467–71.
9. Fai e J, Dancou V, Lejeune C, e al. Reduc ion in colo ec al
cance mo ali y by ecal occul blood sc eening in a F ench
con olled s udy. Gas oen e ology 2004;126:1674–80.
10. Wah endo J, Rob a BP, Wiebel H, e al. E ec i eness o colo ec al
cance sc eening: esul s om a popula ion-based case-con ol
e alua ion in Saa land, Ge many. Eu J Cance P e O J Eu
Cance P e O gan ECP 1993;2:221–7.
11. B enne H, Kloo M, Pox CP. Colo ec al cance . Lance
2014;383:1490–502.
12. Finnish Cance Regis y. h p://www.cance . i/syopa ekis e i/en/
13. Engholm G, Fe lay J, Ch is ensen N, e al. NORDCAN—a No dic
ool o cance in o ma ion, planning, quali y con ol and esea ch.
Ac a Oncol S ockh Swed 2010;49:725–36.
14. Fea on ER, Vogels ein B. A gene ic model o colo ec al
umo igenesis. Cell 1990;61:759–67.
15. Segnan N, Pa nick J, on Ka sa L, Eu opean Commission,
Di ec o a e Gene al o Heal h & Consume s, In e na ional Agency
o Resea ch on Cance . Eu opean guidelines o quali y assu ance
in colo ec al cance sc eening and diagnosis. Luxembou g:
Publica ions O ice o he Eu opean Union, 2010.
16. U.S. P e en i e Se ices Task Fo ce. Sc eening o colo ec al
cance . Rock ille, MD: Agency o Heal hca e Resea ch and Quali y,
2008. h p://www.usp e en i ese ices ask o ce.o g/usps 08/
colocance /colo s.h m
17. Hakama M, Malila N, Dillne J. Randomised heal h se ices s udies.
In J Cance 2012;131:2898–902.
18. Vainio H, Bianchini F, eds. In e na ional Agency o Resea ch on
Cance , Wo ld Heal h O ganiza ion. IARC Handbooks on Cance
P e en ion Vol 7. B eas cance sc eening. Lyon: IARC P ess, 2002.
19. Ma mo MG, Al man DG, Came on DA, e al. The bene i s and
ha ms o b eas cance sc eening: an independen e iew.
B J Cance 2013;108:2205–40.
20. Malila N, An ila A, Hakama M. Colo ec al cance sc eening in
Finland: de ails o he na ional sc eening p og amme implemen ed in
Au umn 2004. J Med Sc een 2005;12:28–32.
21. Malila N, Oi anen T, Malminiemi O, e al. Tes , episode, and
p og amme sensi i i ies o sc eening o colo ec al cance as a
public heal h policy in Finland: expe imen al design. BMJ 2008;337:
a2261.
Pi käniemi J , Seppä K, Hakama M , e al.BMJ Open Gas o 2015;2:e000034. doi:10.1136/bmjgas -2015-000034 9
Open Access