Women registered with a GP with a negative attitude towards breast cancer screening were more likely to be non-participants compared with women registered with a positive GP. Controlling for women’s socio-demographic characteristics and for distance to screening site reduced the association, but it remained statistically significant.
Although women were invited based on the GP practice with which they were registered, these results were still somewhat surprising given the fact that bookings, investigations and follow-up of the screening programme did not involve general practice. Although the confidence interval was fairly close to 1.0 in the multivariate analyses, fairly similar prevalence ratios were observed across different statistical models. One explanation could be that the women feel more comfortable consulting their GP than the booking service for advice if they are uncertain whether or not to take part in the screening programme due to the central role of the primary health care system in Denmark. The attitude of the GP is therefore more likely to influence the women’s choice of participation.
Sub-analyses showed no statistical difference between active and passive non-participants and their GPs’ attitudes towards screening. Hence, no association was observed between GPs’ attitudes and whether or not the women were more likely to call and cancel their appointment (active non-participation) or not to show up (passive non-participation). It should be noted, however, that since these analyses were conducted only on non-participants (n
2,515), some of the groups were rather small.
The results of this study are consistent with findings from studies in other countries indicating that the GPs' influence on screening participation seems to be universal despite different ways of organising the programme worldwide [13
]. However, the results of this study are not as clear as those seen in the USA where advice from health care professionals is regarded as one of the most important determinants for screening participation [20
]. Results similar to those of the present study have been seen in Sweden [18
]. This supports the conclusion of the Swedish study that the GP has an influence on participation in population-based outreach mammography screening programmes.
It should be noted that we do not know to what extent the women consulted their GP for advice on screening participation. Also, on the basis of this observational study, it is not possible to make causal inferences as the actual interaction between the GPs and the women remains unknown. One study, however, has indicated that GPs with a positive attitude are more likely to recommend screening than GPs with a negative attitude [10
The strength of this study is the large population-based design where information about screening participation and the women’s characteristics were obtained from valid and complete registers. This minimises the risk of selection and information bias. An additional strength is that data on screening participation was collected during the first screening round in the region, which makes the population ideal for studying the effect of GPs’ attitudes since no women were excluded on the grounds that they had previously actively chosen not to take part in the programme. The response rate among the GPs was high although only 67 GPs participated, of which only five were negative towards screening. Selection bias may be present since it cannot be ruled out that GPs with special characteristics were more likely to respond or not respond to the questionnaire, e.g. it is plausible that negative GPs were more likely to be non-responders. In addition, women registered with singlehanded GPs may represent women with special characteristics. This is indicated by a somewhat higher participation rate (81.1%) among the population included in our study compared with the entire population of women invited to the first screening round (78.7% excluding the study population – data not shown).
The method used to measure the GPs’ attitudes may be a limitation of the study. Data on the GPs’ attitudes were collected in another study which included only a single categorical question about the GPs’ attitudes. It might have been advantageous to use a more comprehensive measure to assess the GPs’ attitude. Furthermore, due to the data collection method used in this study, 80% of the women were offered screening before their GP answered the questionnaire assessing their attitudes. It cannot be excluded that the attitude of the GP may have been affected and even ultimately changed as a consequence of the screening round, which could lead to information bias.
Only singlehanded GPs were included in this study. If we had included women registered with partnership GPs, the number of included GPs would have been larger and the statistical accuracy higher. However, including partnership GPs could seriously bias the study, as we could not link a GP to the women in partnership practices and therefore also do not know the attitude of the particular GP whom a woman sees. By only including singlehanded GPs we knew the attitude of the GP the woman had seen if she had sought advice from her GP during this period. Future studies should be designed to measure the association between the attitude towards breast cancer screening among partnership practices and women’s screening participation.