Trial Information

Randomized Controlled Trial to Study Interventions to Increase Participation in Cervical Cancer Screening Program

Background

Cervical cancer today is considered as a disease that to a very large extent is preventable.
In Sweden cytological screening has been ongoing since the 60:s and this has made cervical
cancer a rare disease. Several new, but costly, methods like HPV-testing and vaccination has
been claimed to further reduce cancer incidence, but this has not been scientifically proven
so far. A recent audit of the Swedish screening program (Andrae 2008) found that the
foremost risk factor for cervical cancer in the context of the program was non
participation. Women who do participate are well protected from cervical cancer.

A Swedish study evaluating different strategies to increase participation found telephone
contact with defaulters to be the superior intervention (Eaker 2004). This study was done in
a region of Sweden with lower coverage (Sparén 2007) and less emphasis on efforts to
increase participation. A study from Kalmar, Sweden, with high coverage did not find an
extended yield to requests from non-participating women to be cost effective (Oscarsson
2007). Self testing for HPV has been advocated as an effective method and there is a kit
commercially available (Stenvall 2007). The procedure is used as a routine in Uppsala county
and marketed to other counties.

Aim To study two possible interventions to increase participation

Design Randomised controlled trials of two interventions and a control arm

Hypothesis

Intervention with a) supportive telephone contact, by a midwife, with women who do not have
a smear registered for two screening rounds will increase participation, increase uptake of
precancerous lesions (CIN2+) and is cost effective. b) selftesting for oncogenic
papillomavirus (HPV) will increase participation and is cost effective.

Method: selection of study base

Selection was done by random among women aged 29 - 63 who did not have a pap-smear
registered within the two recent screening rounds. Women who were excluded from invitation
due to total hysterectomy and women who could be identified as have immigrated into the
region under the period were excluded before randomisation. July 1 2008 there was 52362
women who fulfilled the first criteria (before exclusions) identified through the Register
for Prevention of Cervical Cancer in West Sweden. 8800 selected women will be included and
randomised to one of three arms with the distribution 5:1:5

Methods: Intervention Arm A: 4000 randomised women fulfilling inclusion criteria receive a
letter informing them they will be contacted by a midwife by phone, to be offered an
appointment to take a smear. They can at that stage decline such contact or if they wish
return contact information (telephone number). A week later the women who have not declined
will be called by a midwife and offered an appointment for taking a smear. Abnormal smear
will be followed up with referral to a gynecologist in concordance with normal screening
routine.

Arm B: 800 women will be sent an offer of HPV-self sampling (Aprovix, Uppsala, Sweden). By
regular mail the woman can order a test kit, and return this to the laboratory after
sampling. The samples will be tested with Hybrid Capture II for high risk HPV. A negative
result will be communicated to the women. A positive result will be sent to the
gynecological clinic responsible for the work up of abnormal cytological screening smears in
the area were the woman lives. The afflicted woman will get information from the clinic and
an appointment for colposcopy and cytology. Further investigation will be conducted as
clinical routine. Reminders will be sent to women who have ordered the kit and not returned
any sample. In order to evaluate the effect of a primary reminder women who have not
responded to the primary offer within 60 days will receive a reminder.

Arm C (controls) 4000 women will be controls. No specific action will be taken within the
study outside routine in the screening program (yearly written reminder when smears are not
found in the database)

Methods: Data collection All data about cytology, colposcopy with biopsies and treatment for
CIN are registered as a routine in the West Sweden Registry for Cervical Cancer Prevention
and data will be extracted from that registry. Data about HPV-testing will be transferred to
this registry from the Aprovix laboratory in Uppsala. Man-time needed and costs for material
and analyses for the different interventions will be registered as base for health economic
assessments.

Primary outcomes: Frequency of testing (cytology in arm A and C and HPV-test in arm B).
Frequency of further assessment of abnormal tests (all arms).

Secondary outcomes: (Arm A vs arm C) Frequency of abnormal smears. Frequency of treated CIN,
(CIN1, CIN2 and CIN3, grouped as low grade (CIN1) and high grade (CIN2, CIN3, AIS and
invasive cancer). Number of invasive cancers and FIGO stadium. Cost of interventions. Cost
per CIN2+ found. The study is powered to find a 30% difference in primary outcome based on
an expected 20% participation rate in the control group with 80% power at a significance
level of 5% for both interventions (A and B compared with C). An expected rate of abnormal
smears of 7% among these women with no smear recorded 6 or more years will give a 80% power
to find a 60% relative difference in number of abnormal smears (RR=1,6) when intervention
arm A is compared with the controls in arm C.

The results from work up of abnormal tests in arm A will be considered representative for
arm B as well, given the same level of abnormality.

The cost per participant and per biopsy with CIN found will be calculated. The cost per
found and treated high grade CIN will be compared with a reference of estimated €3500
(preliminary figure) per CIN2+ that is eradicated.

Methods:Statistical analysis All analysis will be based on intention to treat. Pearson's
Chi2 and Fischers exact test will be used to compare distribution of categorical variables
between the groups. One way variances will be used to test differences of means between
groups for continuous variables. The cumulated probability for outcome vs. follow up time
will be calculated with Kaplan-Meier analysis. Multiple Cox regression will be used to
determine relative risks. 95 % confidence intervals will be used throughout and level of
significance will be calculated two sided as 5%