Economic Evaluation of Three Populational Screening Strategies for Cervical Cancer

Background A high percentage of cervical cancer cases have not undergone cytological tests within 10 years prior to diagnosis. Different population interventions could improve coverage in the public system, although costs will also increase. The aim of this study was to compare the effectiveness and the costs of three types of population interventions to increase the number of female participants in the screening programmes for cancer of the cervix carried out by Primary Care in four basic health care areas.

Methods A cost-effectiveness analysis will be performed from the perspective of public health system including women from 30 to 70 years of age (n=20,994) with incorrect screening criteria from four basic health care areas in the Valles Occidental, Barcelona, Spain. The patients will be randomly distributed into the control group and the three Intervention Groups (IG1: invitation letter to participate in the screening; IG2: invitation letter and informative leaflet; IG3: invitation letter, informative leaflet and a phone call reminder) and followed for three years. Clinical effectiveness will be measured by the number of Human Papillomavirus (HPV), epithelial lesions and cancer of cervix cases detected. The number of deaths avoided will be secondary measures of effectiveness. The temporal horizon of the analysis will be the life expectancy of the female population in the study. Costs and effectiveness will be discounted at 3 %. In addition, univariate and multivariate sensitivity analysis will be carried out.

Expected results IG3 is expected to be more cost-effective intervention than IG1 and IG2, with greater detection of HPV infections, epithelial lesions and cancer than other strategies, albeit at a greater cost.

METHODS

An economic evaluation of three populational screening strategies for cervical cancer will be performed. Particularly, a cost-effectiveness analysis will be conducted. These interventions will be compared to the current opportunistic screening strategy using data of the multicentre randomised trial (CRICERVA). The analysis will be conducted from the National Health Care System perspective.

Design: Pragmatic, blinded, multicentre, randomised, controlled clinical trial with four branches, and a three years follow-up. The randomisation unit was Basic Health Care Area (BHCA).

Setting: Primary Health Care Services (PCS) of Cerdanyola-Ripollet, province of Barcelona, comprising 4 municipalities and 5 BHCA. The population covered by this Primary Care Service (PCS) is 120,293 inhabitants over 14 years. As there are four study groups and 5 BHCAs, only 4 BHCAs with most homogeneous socioeconomic criteria will be considered.

Population: 20,994 women from 30 to 70 years of age with incorrect screening criteria (data obtained from Data Base of Primary Health Care Services) ascribed to the BHCA will be included in the study. Incorrect screening will be defined as [9]: 1.- No cytology in the last 3 years from women between 30 to 40 years, 2.- No cytology in the previous 5 years from women between 40 to 65 years, 3.- No previous cytology history for females older than 65 years or women who have not had their last cytology before the age of 60. The exclusion criteria will be: (a) hysterectomised women, with a current history of pre-malignant lesions [Atypical Glandular Cells of Undetermined Significance (AGUS), Atypical Squamous Cells of Undetermined Significance (ASCUS), Low-grade Squamous Intraepithelial Lesions (LSIL), High-grade Squamous Intraepithelial Lesions (HSIL)], carcinoma in situ and cervical-uterine cancer, Human Immunodeficiency Virus (HIV) positive or other causes of immunosuppression (since these women follow a specific protocol); (b) those residing outside the study setting for more than 6 months; and (c) those ascribed to the study BHCA but with a physician assigned in an UBA of another zone different from the one considered in the study.

Sample size: The sample size has been calculated based on the detection of a difference in effectiveness compared with the Non Intervention Group (NIG). It has been calculated by multiplying the size of a simple randomised design by the design effect or factor of inflation. For the simple randomised design, on accepting an alpha risk of 0.05 and a beta risk of 0.20 in a bilateral contrast, 59 subjects will be required in the first group and 59 in the second group to detect a difference greater than or equal to 28.4 % in the screening coverage of the 41.6 % in the Non Intervention Group (NIG). The lost to follow up rate has been estimated at 20 %. The calculation of the sample has been performed with the Granmo 5.2 computer programme for Windows. According to a review of the literature [23-25], considering an intraclass correlation coefficient of 0.05 and a mean number of 3,500 women from 30 to 70 years of age with incorrect screening by BHCA, the design effect will be 176 and thus, 20,768 women with incorrect screening will be required.

ETHICAL ASPECTS The investigators are committed to respect the prevailing norms of Good Clinical Practice as well as the requisites of the Declaration of Helsinki and the clauses of general and particular ethical conditions related to the right to privacy, anonymity and confidentiality. Neither the first name nor surname or any other type of data indicating the identification of the women will be registered. Therefore, identification will be made by numeric codes. Since this type of study is developed in the usual clinical setting, authorisation and support must be and has already been granted by the representatives and authorities of the collectives involved and thus, individualised informed consent is not necessary. Nonetheless, the research team decided that women attending the consultation for the cytology should sign the consent form. The protocol has been evaluated by the Clinical Investigation Ethics Committee of the IDIAP Jordi Gol.

LIMITATIONS Randomisation by groups will avoid the potential introduction of selection bias which may be produced among the interventions performed at the same site. Since the characteristics of the study do not allow the application of the double-blind masking technique, the masked response evaluation will be used to ensure that the measurement and interpretation of the dependent variables is carried out the same way in all groups. The possible loss of information, which may be produced in women doing screening outside public health care if they are not contacted by the research team, will be minimised with a phone call reminder. This will be made when the women do not attend the appointment. The language difficulties in women from other countries will be solved with cultural mediators at each site. Within the setting of the study, the administrative personnel have been updating the postal addresses of the users attending the BHCA since 2007 and, therefore the postal registry is quite precise, thereby reducing the potential loss of letters.

List of abbreviations

AETM Agency for Evaluation of Technology and Medical Investigation BHCA Basic Health Care Area AGUS Atypical glandular cells of undetermined significance ASCUS Atypical squamous cells of undetermined significance PCC Primary Care Centre ECCI Ethical Committee of Clinical Investigation eCCN Electronic Data Collection Notebook PCT Primary Care Team c-CH Computerised Clinical History IT Investigative Team IG Intervention Group NIG Non intervention Group HC2 Hybrid Capture 2 HSIL High grade Squamous Intraepithelial Lesion CIO Catalan Institute of Oncology CIH Catalan Institute of Health IIPC Institute of Investigation in Primary Care LSIL Low grade Squamous Intraepithelial Lesion SRHC Sexual and Reproductive Health Care PCS Primary Care Service ISU Investigation Support Unit HSV Herpes Simple Virus HIV Human Immunodeficiency Virus HPV Human Papilloma Virus UBA Unitat Bàsica Assistencial (General Practicioner and Nurse Team)

Financial support

This protocol has received financial support from the Fondo de Investigación Sanitaria del Instituto Carlos III de Madrid (exp PI10/01275).