Predicting reattendance to the second round of the Maltese national breast screening programme: an analytical descriptive study

Breast cancer continues to be a major cause of female morbidity and mortality worldwide [1]. Regular breast screening (BS) for breast cancer (BC) could lead to a 25–30% reduction in mortality rates in the population of women invited for screening [2]. This evidence led to the implementation of screening programmes across Europe based on the EU Council recommendations, which recommend biennial screening mammography in average-risk women aged 50–69 years [3]. A Maltese breast screening programme (MBSP) was set up in Malta in 2009 to screen women (aged 50 – 60 years) through mammography every 3 years [4], aiming gradual expansion to reach women until the age of 69 years while reducing the screening time interval as in other countries [5]. In 2015, this age cohort (50–60 years) was in its second BS round.

Breast screening uptake is defined as “the proportion of women invited who attend for screening within 6 months of their invitation” ([6], p.1). In Malta, the overall attendance rate was below the acceptable target of 70% in the first BS round [7]. Similarly, substantial variation in BC uptake exists across other countries [1]. An important factor that merits exploration includes regular screening adherence, because of its significant impact on morbidity and mortality reductions [8, 9]. Detecting BC early is not ensured by a one-off BS attendance [10], but on the consistency of attendance in line with recommended time intervals [11]. The literature suggests that previous mammography use is highly associated with future use [12] because women believe in the effectiveness of screening which in turn increases their intentions to go for screening, resulting in their adherence to subsequent screens [13]. An earlier study by Cockburn and colleagues [14] found that those having weakest intentions to attend for their first screening are less likely to attend for their second screening (OR = 0.44, CI 0.23, 0.85). Therefore, it has been suggested that programs should focus on reaching those who have underutilized mammography in their past [12] as this would feed into attendance in subsequent BS invitation rounds. Nonetheless, many studies have focused on the reasons for one-time screening rather than subsequent use [9, 15] and limited studies have sought to specifically explore predictors of uptake to second round BS invitations [14, 16, 17]. In a comparative study of 200 re-attenders and 200 non-re-attenders for second triennial National Breast Screening Programme appointments in Nottingham [10], the 200 women who failed to accept their invitation implicated their negative initial screening experience in their decision, with 41% implicating pain, 6% stress and 3% embarrassment.

To our knowledge, no research yet exists on women’s reattendance at the MBSP or on screening predictors to the second BS round in Malta. In 2017, we reported on first screening uptake of the MBSP [4]. We found that Health Belief Model (HBM) constructs were the strongest predictors of uptake of first invitation to the MBSP, though the inclusion of illness representation dimensions improved the predictive accuracy for non-attendance. In the present study, we follow a sub-sample of Maltese women with three main objectives: (1) to determine their re-attendance at the second BS invitation round; (2) to explore whether sociodemographic factors, health status, knowledge, health beliefs and illness perceptions are significant predictors of second BS uptake; and (3) to determine if uptake of first invitation to attend the MBSP is a significant predictor of uptake to the second invitation in Malta.

BS behaviour is influenced by a number of factors, including health beliefs [8], illness representations [8], knowledge of BC and BS [4], socio-demographic factors [4], and health status (medical factors) [8]. These factors were found to influence BS attendance in the first invitation round in Malta [4]. The Health Belief Model (HBM) was used as a theoretical framework for this study, due to its application in myriad preventive health behaviours, including BS [8, 18‐20]. The HBM consists of six main variables (constructs), namely perceived susceptibility, perceived severity, perceived benefits, perceived barriers, cue to action, and self-efficacy [19]. These variables play an important role in an individual’s perception about BS, such that women are more likely to perform BS if they feel susceptible to BC or the risks of contracting BC (perceived susceptibility), believe in the seriousness and consequences of the disease (perceived severity), perceive more benefits than barriers from undergoing mammography, have higher confidence for obtaining mammography, and take heed in cues to action [19]. Based on the HBM, women’s perceptions about BC are derived from their knowledge and perceptions about the disease, which predicts whether women will attend for mammography [18]. However, the HBM does not include the impact of emotions to predict BS behaviours, which is why the Common-Sense Model (CSM) [20], also used to explain BS behaviour in a small number of studies [8, 21‐25], was utilised in this study. The CSM considers the cognitive and emotional representations of a health threat [8, 26, 27] and guides subsequent coping behaviour [28]. Illness perceptions comprise the following constructs: BC identity, cause, cancer timeline (acute/chronic, cyclical), consequences, personal and treatment control, illness coherence and emotional representations. [8, 21].

Similar to the analysis carried out in our previous papers to analyse the associations between attendees and non-attendees in relation to lifetime mammography practices [30] and timely adherence [31], the following analysis focused on the associations between attendees and non-attendees to a second BS invitation.

We further explored health beliefs and illness perceptions as ‘constructs’ and which of these constructs were statistically significant with second BS uptake. Table 4 shows the relationship between all 14 constructs respectively (perceived susceptibility, perceived benefits, perceived barriers, cues to action, self-efficacy, breast cancer identity, causes of breast cancer, cancer timeline: acute/chronic, cancer timeline: cyclical, consequences, personal control, treatment control, illness coherence, emotional representations), and attendance and non-attendance to second BS round.

The ‘perceived barrier’ construct was the only statistically significant variable that described the variance between attendees and non-attendees.

Similar to the analyses carried out in our previous paper to predict lifetime mammography attendance in Malta [30], the following analyses focused on predicting attendance and non-attendance to the second BS invitation. The most significant variables for all logistic regression models are presented in Table 5. Logistic regression models 1 and 2 incorporated all items related to demographic and health status variables respectively, though none were found to be significant predictors of second BS uptake.

Models 3 to 7 include the psychosocial variables. Model 3 incorporated all HBM variables, of which three variables were found to be good predictors of second BS uptake: ‘there is no possibility of getting breast cancer’, ‘fear of the unknown procedure’, and ‘GP advice to attend’ (Table 5). For this model, attendance was predicted with an accuracy of 93.2% and non-attendance with 30.8% accuracy. When removing ‘fear of the unknown procedure’ from the model, the accuracy decreased from 30.8 to 19.2% and hence was retained even though p > 0.05.

Model 4 included all IPQ-R variables, of which four IPQ-R variables were found to be good predictors: ‘breast swelling, dimpling, redness or soreness of the skin’, ‘diet’, ‘altered immunity’, and ‘if you had BC, your whole life would change’ (Table 5). The accuracy for this model was 89.2% for attendance and 69.2% for non-attendance.

Model 5 focused on the significant variables found in model 3 and 4 together; i.e. on 12 variables (five Health Beliefs and seven Illness Perception variables). The final model (Model 5) retained the same significant predictors as in Model 4, excluding the HBM variables, hence showing that illness perceptions are important predictors for second BS uptake. The model accuracy, when combining both scores, was identical to that of Model 4, predicting attendance by 89.2% and non-attendance by 69.2%.

A logistic regression model (Model 7) was constructed with all 14 constructs (i.e. HBM and IPQ-R as constructs not individual items). The following constructs were the most significant predictors of second screening uptake: ‘perceived barriers’, ‘breast cancer identity’, ‘causes of breast cancer’ and ‘consequences’. The ‘perceived barriers’ construct was found to be the strongest predictor. However, non-attendance was predicted with an accuracy of 30.0%, which is inferior when compared to Models 5 and 6 (69.2%). Moreover, when removing the constructs ‘causes of breast cancer’ and ‘consequences’ from the model (p-value slightly greater than 0.05), the accuracy to predict non-attendance would decrease from 30.0 to 15.4% even though p > 0.05.

Although ‘perceived barriers’ remains the most important construct to describe the variance between attendees and non-attendees, illness perception constructs (breast cancer identity, its causes and consequences) can also be considered as strong predictors of second BS uptake; a result further echoed in Model 5, where the predictors are all related to illness perceptions (breast cancer identity and consequences). Hence, although Health Beliefs are important predictors of BS uptake, the model accuracy improved with the inclusion of illness perception items into one logistic regression model (Model 6 vs Model 3).

When a logistic regression model was applied to predict second (subsequent) BS uptake using the first BS uptake as the predictor (Table 6, Model 8), non-attendance was predicted with an accuracy of 65.4% and attendance was predicted with an accuracy of 83.8%.

Another model (Model 9) incorporated the Health Beliefs and Illness Perception constructs as covariates, together with the first screening uptake variable as the main independent variable. Model 9 shows that, following the inclusion of all the constructs, ‘Perceived barriers’ and ‘Breast cancer identity’ were found to be important covariates to improve the accuracy of predicting the attendance to the second screening cycle i.e. 83.8% (Model 8) to 91.9% (Model 9). On the other hand, model accuracy dropped from 65.4% (Model 8) to 61.5% (Model 9) when predicting non-attendance to the second invitation at the MBSP.

It is not enough to getting women to initiate BS, but it is essential to encourage them to maintain use over time. Previous studies have examined re-attendance rates [6, 16, 32]. However, BS statistics are not available specifically for a second BS invitation. The available routine BS statistics largely provide cross-sectional estimates of coverage rather than information on women’s ongoing attendance at BS [33]. Moreover, data has long relied on self-reports [34], are measured by area deprivation (using either residential postcodes or general practice postcodes) rather than individual characteristics [35] and can be affected by inflation based on registered general practice lists [33].

Our preliminary rates for Maltese women invited to their subsequent call seem to be lower than those in other countries (Fig. 1: from the total data (n = 100), 62% of Maltese women attended the first and second round, 9% had attended the first but not the second round, 17% attended neither call). In a recent study in London among different ethnic groups [6], white British women were most likely to attend their first call (67%) and routine recall (78%). Mixed White and Asian women had the next highest uptake of routine recall invitations (75%), followed by Indian women (first call (61%) or routine recall (74%) appointments), Pakistani (52 and 67%, respectively) and Bangladeshi women (43 and 61%, respectively). The lower subsequent uptake rates in Malta could be due to the programme being relatively new. However, we could not find previous studies that explored re-attendance. We therefore sought to explore the associations and predictive psychosocial factors to second screening based on social cognitive theory.

All sociodemographic characteristics were not significantly associated with second screening uptake and the latter were also found to be non-significant predictors of second screening to the MBSP. These findings are substantiated by similar predictors of returning to a second screen in Australia [16].

More specifically, earlier studies found that women with a breast problem were more likely to undergo clinical breast examination (CBE) and mammography than those who had none [36, 37], while poor self-rated general health status was not associated with ever having had a mammogram in another study [38]. This relationship is likely driven by higher knowledge levels on the benefits of BS [36]. In contrast, we did not observe significant differences between attendees and non-attendees for self-reported health status in this study, except for those who claimed they had a breast condition. The latter women were more likely not to attend for second screening when compared with those who said they did not have any breast problem. We are aware that self-reports may not be as accurate as clinically documented mammographic reports. Nonetheless, in view of the nature and organization of the cognitive and emotional representations of such a health threat, as proposed by CSM [8, 26, 27], women may have opted to attend for private mammography [4], possibly as our earlier study suggests, to obtain an earlier result [7]. Such reasons for non-attendance to second round screening merit further investigation in the local context.

This study also provides evidence that women who were less knowledgeable of the recommended screening frequency were less likely to reattend at the MBSP for a second invitation. There is a similar widespread lack of knowledge of the recommended screening guidelines for a first BS invitation in Malta [4] and in other countries [39, 40]. The impact of the physician-patient relationship may be a reason for the latter finding. In this respect, it is important to note the variation in the order of invitation from service to service. For example, in the UK, women are invited to BS through GP practices [41] whereas in Malta, women are not registered with GP practices but are invited to BS according to age cohorts. Possibly, this infrequent encounter between GPs and Maltese women (Table 1) may be the reason for physicians’ lack of opportunity to address knowledge gaps and to recommend regular BS practices.

Factors associated with BS behaviours in other populations [42, 43] were similar to the second BS round at the MBSP. Namely, Maltese women who believed that they were susceptible to BC, considered the personal consequences of the disease, believed that mammography was not painful, and considered cues to action to motivate them to attend, were more likely than others to re-attend. Similarly, items for BC consequences and cues to action as well as mammography pain were also significantly associated with first BS uptake at the MBSP. Evidence has shown that women’s responsibilities within the family may conflict with self-care and limit screening attendance and re-attendance [44]. Efforts to educate health care providers, particularly physicians, should emphasize the importance of mammography referral and regular physical check-ups [43].

Of the remaining cognitive variables, perceived BC susceptibility and attitude towards use predicted attendance to the second BS round, echoing predictors of repeat use in other studies [45] while contrasting other evidence [32]. Attitudes towards BS behaviour and risk perceptions may be necessary components of why Maltese women contemplate maintaining a behaviour, but may not be prime motivators to influence the initiation of that behaviour.

Self-efficacy was not significantly associated with maintaining BS practices in our study but played a key role in explaining why some women were unable to initiate BS at the MBSP [4]. Although previous research has shown that self-efficacy may be particularly central in moving women from contemplating about undergoing mammography to actually obtaining it [46], in the context of a second BS invitation, the rising challenges or barriers women experience when trying to maintain that behaviour may buffer the intentions that prompt planning for that behaviour [32].

Although several variables differentiated those women who returned for second round screening in the programme from those who did not, the ‘perceived barrier’ construct was found to consistently explain the differences between attendees and non-attendees and to improve the accuracy of predicting attendance to the second screening cycle; a concurrent finding for first invitation to the MBSP [8]. A trial by Farhadifar and colleagues [47] also suggests that regular BS practices are related to fewer barriers. The fact that some women did not return to their second invitation implies that based on their first contact with the screening service, they were less likely to return for routine calls. In the extant literature, this is attributable to a previous negative experience [48], possibly due to pain or discomfort [7, 49], embarrassment, distress, unhelpful staff and/or lower reassurance during the first screen [50]. Our national data also provides evidence that fear of pain was a major factor for not attending the first BS invitation [4] and similarly, this study found pain to be a significant determinant to non-adherence. Preparing women through the invitation letter or in screening campaigns [51] and improving the mammography experience so that it matches women’s expectations could help to increase and maintain BS uptake [6].

The inclusion of both Health Beliefs and Illness Perception items in our study improved the accuracy of predicting non-attendance to the second cycle (Model 6, 84.6%). This finding coincides with findings for first invitation [4]. Given the importance of both theories and the higher predictive accuracy for both BS cycles, it is likely that psychosocial variables, women’s perception of BC and its related risk, and the enactment of cancer control measures predict regular BS behaviours [52] in an attempt to gain control over the disease [53]. This sense of internal control has been proven among disease-effected individuals across diverse health contexts [22, 54, 55] and healthy individuals [8, 52].

Finally, when considering that only one variable was utilised to predict second screening uptake, Model 8 can be considered as an extreme improvement over all the previous logistic regression models (Table 5). Furthermore, the fact that Model 8 utilised only one predictor to predict second uptake makes the model more efficient and easy to use. Moreover, first screening uptake is a very important and significant predictor to predict future uptake as this variable does not require any health beliefs and/or illness perception variables or any other information to predict second BS uptake. Our results further support the evidence that women who obtain at least one mammogram are more likely to obtain subsequent screening [8, 12, 56]. Attendees may be more aware of the possibility for BC to occur, possibly because a BC threat (BC identity, causes and consequences found as CSM predictors for a second BS invitation in this study), coupled by BS benefits, may have helped the individual to ascertain what factors pose barriers to BS attendance and adjust these in the first place. Therefore, what matters most is the social and psychological characteristics, health behaviours and attitudes which women bring to screening the first time, such as a positive attitude towards BS, its value, health behaviours and the belief that an individual would be able to overcome any obstacles to attendance [8, 56]. The most important implication is that if women can be recruited successfully on the first occasion, they will probably stay in the programme. Hence, efforts could focus on identifying and encouraging attendance among women who have never participated in BS and who are reluctant or unsure to participate initially [14] because if women are persuaded to change their beliefs, attendance rates will increase and reattendance will become a matter of routine [17].

Researchers have focused more often on promoting behaviour change than on sustaining change. We explored, for the first time, the psychosocial associations, differences and predictors to the second screening invitation based on demographic factors, health status, knowledge, health beliefs, illness perceptions, and actual previous health behaviours. The combination of HBM and CSM variables provided improved prediction of attendance and non-attendance to the second screening call. Perceived barriers, breast cancer identity, causes of BC, and consequences contributed most to the regression models, though perceived barriers was consistently significant across the analyses. Interventions should particularly target non-attendance to first screening. If non-attendees can be persuaded to attend once, they are likely to re-attend, unless their screening experience has been a negative one. The implications of these results are considered for theory, policy and practice to improve the limited understanding of second round screening and to aid the design of culturally sensitive interventions to improve breast screening uptake in Malta.