Cervical screening: the optimum visit plan for contacting users and non-users in Scotland.

Open Access
Publisher Website
Google Scholar
Abstract
OBJECTIVE--To investigate the numbers of visits required to obtain interviews with users and non-users of cervical screening, and to determine the workload involved to enable an optimum visit plan to be developed. DESIGN--Case-control study of users and non-users of cervical screening using a flexible visit plan that involved up to eight attempts at contact. Visits were made in mornings, afternoons, and evenings, the visit pattern being determined by information gained from local sources. PATIENTS--Altogether 660 non-users of cervical screening (cases), aged 20-64 and registered with 23 randomly selected general practitioners (GPs), were identified from the Tayside computerised register of cervical smears. These women were selected from the computerised lists of 18 GPs in Dundee and five in Perth. A total of 417 women recorded as having a smear within the previous three years (controls), matched by age and GP, were also identified from the computerised register. RESULTS--Altogether 1834 attempts were made to contact the cases, of whom 339 were interviewed, giving a workload of 18 interviews per 100 attempts. For the controls 1359 attempts were made at contact to yield 339 interviews, a workload of 25 interviews per 100 attempts. Refusals (19%) and incorrect addresses (23%) were the two major reasons for failing to achieve interview. Only for four (0.6%) of the cases and one (0.2%) of the controls was no information gained. The proportion of attempts which led to interview remained constant with increasing numbers of call-backs (up to six for the cases and eight for the controls). CONCLUSIONS--A flexible approach to visit scheduling that takes account of local knowledge can lead to interviews with 66% of non-users of health screening, when incorrect addresses are removed. It is preferable to plan for many (up to six) visits to achieve interview. This will minimise non-response bias without increasing the workload per successful interview.