Performance and effectiveness of organised breast cancer screening in Finland

Abstract

Rintasyövän väestöseulonta on vaikuttavaa

Väitöstutkimuksessa selvitettiin maksuttoman rintasyövän väestöseulonnan laatua ja seulonnan vaikutusta rintasyöpäkuolleisuuteen. Tutkimus perustui Suomen Syöpäyhdistyksen Joukkotarkastusrekisterin kutsu- ja tarkastustietoihin ja Suomen Syöpärekisterin seurantatietoihin vuosilta 1991-2003. Aineisto sisälsi lähes 400 000 suomalaisnaisen tiedot.

Väestöseulonta täytti Euroopan Unionin laatuvaatimukset ja oli vaikuttavaa. Seulontaan kutsuttujen naisten rintasyöpäkuolleisuus 50-69-vuotiaana oli noin viidenneksen alhaisempi kuin se olisi ollut tilanteessa, jossa seulontaa ei olisi järjestetty. Voimakkain kuolleisuusvaikutus 60-79-vuoden iässä havaittiin kunnissa, jotka kutsuivat 50-69-vuotiaita naisia seulontaan.

Satunnaistetut kokeet ovat osoittaneet seulonnan alentavan rintasyöpäkuolleisuutta. Koska rutiinitoiminnan vaikutus kuolleisuuteen voi olla erilainen kuin koeasetelmassa, väestöseulonnan arviointi on tärkeää. Väitöstutkimus oli ensimmäinen, joka selvitti seulonnan kuolleisuusvaikutusta vuoden 1991 jälkeen, jolloin seulonnan piiriin saatiin kaikki 50-59-vuotiaat suomalaisnaiset.

Vaikka seulonnan laatu täytti Euroopan Unionin vaatimukset, havaittiin eräissä laatutekijöissä, kuten lisätutkimusten määrissä, syöpien osuudessa, sekä seulonnan syöväntunnistuskyvyssä eroja seulontayksiköiden välillä. Laatutunnuksissa havaittu vaihtelu ei kuitenkaan ollut yhteydessä kuolleisuusvaikutuksen tasoon.

Mammografiaseulonnan kyky tunnistaa rintasyöpä parani naisten ikääntyessä. Kun myös kuolleisuusvaikutus oli paras 50-69-vuotiaina seulotuilla naisilla, voitiin todeta, että kutsuiän laajentaminen nykyisestä 50-59-vuodesta 50-69-vuoteen tulee parantamaan seulonnan vaikuttavuutta. Estettyjen rintasyöpäkuolemien määrä tosin jäänee melko pieneksi ainakin alkuvuosina. Laajennuksen kustannusvaikuttavuus on toistaiseksi arvailujen varassa. Useat kunnat ovat kutsuneet 60-69-vuotiaita naisia maksullisiin seulontoihin. Suomessa tehdään myös mammografiatutkimuksia, joihin hakeudutaan lääkäriltä saadun lähetteen avulla. Näitä mammografioita ei ole rekisteröity, eikä niiden laajuudesta tai vaikuttavuudesta ole tietoja.

Background and objective. Randomised trials have shown that invitation to screening reduces mortality from breast cancer. Since the results from rigorously conducted trials may not be directly convertible into routine service, and the validity of some trials has been questioned, evaluation of population-based service screening is essential. This study retrospectively evaluated the performance and the effectiveness of organised breast cancer screening in Finland by estimating the process and validity indicators of screening, the effect of screening on breast cancer mortality, and the relationships between process, validity and outcome of screening.

Material and methods. The study subjects were 40-74 year old women who had been invited to breast cancer screening in ten centres of the Cancer Society of Finland (CSF) from 1991 onwards. Invitations from CSF centres accounted for 50-60% of screening invitations in Finland in the study period.

The indicators for the screening process, e.g. screening coverage and attendance, recall rate, breast cancer detection rate, and positive predictive value of mammography, were examined among the 40-74 year old invitees and participants. The indicators were measured separately for the first and subsequent screens by five-year age groups, invitational year, and screening centre in 1991-2000. The indicator for screening validity, episode sensitivity by detection and incidence methods, was examined among the 50-64 year old participants. The sensitivity was measured separately for the first and for the subsequent screens over the whole study period 1991-2001, during the first and the second year after the screening visit, and by five-year age groups, three-year periods, and three centre categories. The categorisation of screening centres was based on recall rates and visits for each centre at the subsequent screens.

The overall incidence-based breast cancer mortality was analysed in 1992-2003 among the invitees and participants of screening aged 50-69 at death. Mortality was measured by five-year age groups at death and by the same centre categories that were used in the validity analysis. Observed deaths from breast cancer were compared with expected breast cancer deaths without screening. The observed deaths were obtained from a cohort of individual invitees (n=361 848). The expected deaths were defined by modelling population level breast cancer mortality from 1974 to 1985 and 1992 to 2003. The population data were derived from the same municipalities (n=260) that were incorporated into the cohort. The analysis of incidence-based breast cancer mortality was further extended to ages 60-79 by comparing observed and expected breast cancer deaths among all female and among screened and non-screened women in three invitational policies (regular screening at ages 50-59; regular screening at ages 50-69; irregular screening at ages 50-69).

Results. In 1991-2000, the CSF centres sent more than one million invitations, and processed 930 000 visits. The coverage of invitations was more than 95% among women aged 50-59 years, 20-40% among women aged 60-69 years, and less than 10% among 40-49 and 70-79 year old women. The overall attendance at ages 50-64 was 90%. Per one subsequent screen, less than 3% of attendees were sent for further examinations. Their breast cancer detection rate was 0.4%. The positive predictive value of mammography increased towards the end of the study period and with age.

In 1991-2001, the overall episode sensitivity at the subsequent screens was 54% by incidence, and 65% by detection method. The sensitivity estimates 0-11 and 12-23 months after the screening visit were 70% and 38% by the incidence method. The sensitivity increased 13% per 1 % absolute increase in the recall rate.

In 1992-2003, the overall reduction in the incidence-based breast cancer mortality among the screening invitees aged 50-69 at death was 22% (relative risk 0.78, 95% confidence interval 0.70-0.87). After adjusting for self-selection, the reduction among the participants became 28% (0.72, 0.56-0.88). Among all females aged 60-79 at death, the greatest reduction in breast cancer mortality, 28% (0.72, 0.51-0.97), was estimated in municipalities inviting 50-69 year old women on a regular basis. In municipalities restricting their invitations to 50-59 year old women, no reduction in breast cancer mortality at ages 60-79 was observed.

The overall performance indicators in Finland were at a level similar to those in other European screening programmes, but the centre-specific differences were wide. Variability in centre-specific episode sensitivity was associated with the variability in recall. In further analyses, no association between the level of recall and the screening outcome could be found.

Conclusion. Organised breast cancer screening in Finland has been effective. Extension of invitations to women aged 60-69 will decrease breast cancer deaths among the elderly. Due to ambiguity in the relationships between the estimates of performance and outcome, evaluation of breast cancer mortality remains the most reliable means to ascertain the success of service screening.