Ding et al.

International Journal for Equity in Health (2020) 19:212

https://doi.org/10.1186/s12939-020-01323-z

RESEARCH Open Access

Coverage determinants of breast cancer

screening in Flanders: an evaluation of the
past decade
L. Ding1,2, S. Jidkova3,4, M. J. W. Greuter5,6, K. Van Herck3,4, M. Goossens4, P. Martens4, G. H. de Bock1* and
G. Van Hal2,4

Abstract
Background: Breast cancer (BC) is the most common cancer in women in the developed world. In order to find
developing cancers in an early stage, BC screening is commonly used. In Flanders, screening is performed in and
outside an organized breast cancer screening program (BCSP). However, the determinants of BC screening
coverage for both screening strategies are yet unknown.
Objective: To assess the determinants of BC screening coverage in Flanders.
Methods: Reimbursement data were used to attribute a screening status to each woman in the target population
for the years 2008–2016. Yearly coverage data were categorized as screening inside or outside BCSP or no
screening. Data were clustered by municipality level. A generalized linear equation model was used to assess the
determinants of screening type.
Results: Over all years and municipalities, the median screening coverage rate inside and outside BCSP was 48.40%
(IQR: 41.50–54.40%) and 14.10% (IQR: 9.80–19.80%) respectively. A higher coverage rate outside BSCP was
statistically significantly (P < 0.001) associated with more crowded households (OR: 3.797, 95% CI: 3.199–4.508),
younger age, higher population densities (OR: 2.528, 95% CI: 2.455–2.606), a lower proportion of unemployed job
seekers (OR: 0.641, 95% CI: 0.624–0.658) and lower use of dental care (OR: 0.969, 95% CI: 0.967–0.972).
Conclusion: Coverage rate of BC screening is not optimal in Flanders. Women with low SES that are characterized
by younger age, living in a high population density area, living in crowded households, or having low dental care
are less likely to be screened for BC in Flanders. If screened, they are more likely to be screened outside the BCSP.
Keywords: Breast neoplasms, Mammography, Mass screening, Coverage rate, Social determinants of health

* Correspondence: [email protected]
1
Department of Epidemiology, University Medical Center Groningen,
University of Groningen, Groningen, The Netherlands
Full list of author information is available at the end of the article

© The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License,
which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if
changes were made. The images or other third party material in this article are included in the article's Creative Commons
licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons
licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain
permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the
data made available in this article, unless otherwise stated in a credit line to the data.
Ding et al. International Journal for Equity in Health (2020) 19:212 Page 2 of 9

Background being an immigrant [19], physical disability [20], and

Belgium is among the countries with the highest having one or more chronic diseases [21]. Only a rela-
female breast cancer (BC) incidence and mortality tively small amount of studies are dedicated to explor-
worldwide [1]. In 2018, the age-standardized BC inci- ation of the determinants of screening outside a BCSP.
dence and mortality rate of Belgium women were Regular visits to a gynecologist, being employed and
113.2/100,000 and 22.2/100,000 person-years, respect- low esteem of the quality of the population screening
ively, which is higher than the estimate for the whole program are associated with an increased attendance
Western European region (92.6 and 15.5/100,000 to screening outside a BCSP [6, 7, 22]. However, these
person-years, respectively) [1, 2]. While sufficient evi- studies only depend on self-reported data from health
dence has indicated that mammography screening has surveys or focus group discussions.
the potential to initiate early diagnosis and treatment There is a paucity of studies that have investigated
for BC and lower BC mortality, the effect of mammog- the determinants of screening coverage in a setting
raphy screening relies on the degree to which women that has BC screening in and outside BCSP. The aim of
participate in screening [3]. this study therefore was to evaluate the factors associ-
In most of the high income countries, women are rec- ated with the coverage rate of mammography screen-
ommended to participate in an organized BC screening ing and factors that contribute to women’s choice of
program (BCSP), where quality is warranted by system- screening in and outside the BCSP using municipality
atic quality control measures [4]. Outside this program, level aggregated data.
with the aim to screen more of the eligible women,
spontaneous screening is also endorsed in some countries, Methods
such as in Belgium [5], France [6] and Switzerland [7]. Screening in Flanders
The coverage of BC screening, defined as the percentage Flanders, the most populated region of Belgium, estab-
of screened women in the total eligible population within lished a BCSP for women aged 50–69 in 2001 [5]. The
the specific interval of routine screening [8], is an import- organization and implementation of mammography
ant indicator for the evaluation of the effectiveness of screening in and outside the BCSP in Flanders have
screening [4, 9]. However, the average coverage rate in been described in detail elsewhere [5, 23, 24]. Briefly,
2016 across OECD countries was only 57.4% [10]. As for in BCSP, every 2 years, eligible women aged 50–69 are
Flanders, the coverage rate of BCSP in 2017 was only 49 actively recruited through a personalized invitation let-
and 13% was screened outside the BCSP [11]. ter sent by the Center for Cancer Detection in Flanders
Many factors have previously been shown to be with a fixed time and place for a digital mammography
associated with a reduced coverage level of BCSP. A screening fully and directly paid by the health insur-
systematic review summarized that the barriers to BC ance system in Flanders. The Flemish program follows
screening fell into three categories: 1) health care sys- the European quality assurance guidelines [9]. Mam-
tem level barriers, such as lack of health care providers mography screening outside the context of the BCSP
and economic barriers; 2) social barriers, such as lack can be accessed by a referral from a general practi-
of social support and cultural norms opposed to BC tioner (GP) or a gynecologist, is not fully covered by
screening;, and 3) individual level barriers, such as lack health insurance [5], and does not systematically in-
of cancer knowledge and beliefs, negative expectations clude quality-control activities (e.g. double reading)
of screening, and distrust of the medical system [12]. [5]. Since 2016, women who received reimbursement
However in this review, the majority of the included for mammography in the last 2 years from the health
studies relied on self-reported data, studies with insurance or have been diagnosed with BC in the last
random and convenience samples were pooled, and 10 years in the Flemish health care system are not in-
evidence was only qualitatively synthesized. Many vited for the population screening program.
other studies have also provided quantitative evidence
on these hampering factors. Among them, economic Data description
related barriers were the most commonly studied Municipality level screening coverage in 2008–2016
factors and results showed that low income [13], was calculated using data from the Center for Cancer
crowded housing condition [14], unemployment [15] Detection in Flanders [25]. Municipalities that have no
and residing in social-economically deprived areas missing values of the number of screened and non-
[16] are predictors of a lower BC screening coverage screened women were included in the study. Independ-
rate. Lack of a regular health care provider is associ- ent variables at the municipality level of 2008–2016
ated with a reduced coverage rate of screening both were derived from the database of the Flemish provin-
inside [17] and outside a BCSP [7]. Other individual cial authorities and linked to data of the screening
level characteristics include residential instability [18], coverage. We included only the variables that were
Ding et al. International Journal for Equity in Health (2020) 19:212 Page 3 of 9

publicly available in order to reduce the bias that may defined as the average number of residents per house-
be induced by the selection of variables [26]. holds as a proxy for crowded housing conditions.
(2)Women with equivalent living wages was defined as
Privacy considerations the percentage of women with equivalent living wages
Privacy was warranted since only aggregated data were which is the minimum income awarded by the social
available at municipality level and for municipalities with welfare center. (3)Share of borrowers with at least one
less than 5 screened women overall or in one of the four overdue loan was defined as the percentage of bor-
age groups (50–54, 55–59, 60–64, and 65–69), a missing rowers with at least one overdue loan per municipal-
value was used. ity where a high percentage was considered as a
proxy for poverty; and (4)Job seekers were defined as
Main outcome the percentage of unemployed residents with waiting
The main outcome of our analysis was the screening allowance or bridging allowance per municipality.
coverage rate inside and outside the BCSP from 2008 to Health status was indicated by residents aged 18–64
2016. The coverage rate was presented overall as a me- with physical disability or status of diabetes, and de-
dian value over all years and municipalities and stratified fined as the percentage of handicapped residents aged
by age groups and the two screening strategies. 18–64 and the percentage of residents with diabetes
recognized by the health insurance system, respect-
Determinants considered ively. Healthy behavior was indicated by dental visit
For an overview of the variables considered in the defined as the percentage of residents having at least
analysis, see Table 1. Number of residents and popu- 2 visits at the dentist in 2 different years within a
lation density were defined as the total number of period of 3 calendar years per municipality.
residents and the number of residents per km2 per
municipality, respectively. Natural balance was de- Statistical analysis
fined as the natural growth per 1000 residents per Median value and interquartile range (IQR): p25-p75
municipality. Residential stability was indicated by the were calculated for all continuous variables which
percentage of residents having the same address as were not normally distributed. The annual screening
the year before. Non-Belgian nationality was defined coverage rate inside and outside the BCSP was calcu-
as the percentage of residents without a Belgian na- lated as a median value over all years and municipal-
tionality per municipality. The socioeconomic status ities and presented overall and stratified by four age
(SES) of residents was characterized by the following groups: 50–54, 55–59, 60–64, and 65–69. To evalu-
four proxy variables: (1) Average household size was ate which determinants were related to the annual

Table 1 Social demographic parameters of Flanders per municipality in the period 2008–2016. In total 295 municipalities were
included
Median (P25-P75)
Population and households
number of residents (105 residents) 0.15 (0.10–0.22)
population density (1000 residents per km2) 0.41 (0.27–0.66)
natural balance (per 1000 residents) 0.86 (−0.76–2.43)
same address as last year (compared to all residents)% 92.50 (91.40–93.30)
non-Belgian nationality (compared to all residents)% 3.30 (2.10–6.00)
average household size 2.44 (2.37–2.51)
Welfare and poverty (%)
women with equivalent living wages (compared to all women residents) 0.26 (0.18–0.41)
share of borrowers with at least one overdue loan (compared to all borrowers) 3.00 (2.50–3.80)
job seekers (compared to all residents)% 1.80 (1.40–2.20)
Health and handicap (%)
person with physical disability18-64y (compared to all residents in 18-64y) 1.96 (1.57–2.58)
diabetes (compared to all residents) 5.10 (4.60–5.60)
dental visit (compared to all residents) 54.50 (51.30–57.70)
Ding et al. International Journal for Equity in Health (2020) 19:212 Page 4 of 9

coverage of the two screening strategies, a logistic the BCSP was positively associated with the average
regression model with generalized estimating equa- household size (OR: 1.282, 95% CI: 1.138–1.444),
tions (GEE) was constructed to account for the cor- while negatively associated with the percentage of
relation of repeated measurements of municipality women with equivalent living wages (OR: 0.899, 95%
level screening coverage rate and social demographic CI: 0.855–0.945), the percentage of unemployed job
parameters. In the GEE model, the dependent vari- seekers (OR: 0.961, 95% CI: 0.936–0.987) and popula-
able was the municipality level coverage rate and the tion density (OR: 0.918, 95% CI: 0.888–0.949).
independent variables were the municipality level (Table 3). After the adjustment for social demographic
social demographic parameters as given in Table 1. A parameters in the multivariate analysis, the probability
binary variable that indicated the type of screening of being screened inside or outside the BCSP was only
strategy that the coverage rate referred to was negatively associated with average household size (OR:
provided and used as an independent variable. Odds 0.894, 95% CI: 0.809–0.988), population density (OR:
ratios (OR) were reported with 95% confidence inter- 0.929, 95% CI: 0.906–0.952), and diabetes prevalence
val (CI). The effect of social demographic parameters (OR: 0.964, 95% CI: 0.952–0.976) whereas positively
was investigated by assessing a two-way interaction associated with the percentage of unemployed job
between the two screening strategies and the significant seekers (OR: 1.073, 95% CI: 1.051–1.095), and the per-
independent variables. All statistical analyses were per- centage of residents with proper dental care (OR:
formed using R version 3.6.0, and a two-sided P < 0.05 was 1.005, 95% CI: 1.003–1.007) (Table 4).
considered statistically significant. Contrary to the BCSP, the probability of being
screened outside the BCSP was positively associated with
Results being in a younger age group, a high population density
We included 295 of the 308 municipalities in Flanders (OR: 2.528, 95% CI: 2.455–2.606), and a larger house-
that reported full data of the number of screened holds size (OR: 3.797, 95% CI: 3.199–4.508), and nega-
women of the two screening strategies in all age tively associated with the diabetes prevalence (OR: 0.942,
groups in 2008–2016. The median percentages of all 95% CI: 0.921–0.962), the percentage of unemployed job
included social demographic parameters over all years seekers (OR: 0.641, 95% CI: 0.624–0.658) and the per-
and municipalities are shown in Table 1.The overall centage of residents with proper dental care (OR:
median coverage of all years and municipalities of both 0.969, 95% CI: 0.967–0.972) (Table 4).
screening strategies combined was 60.90%. The median
coverage rates of all years and municipalities inside
and outside the BCSP were 48.40% (IQR: 41.50– Discussion
54.40%) and 14.10% (IQR: 9.80–19.80%) respectively, In the present study, we assessed the coverage deter-
Table 2. The median coverage of screening outside the minants of screening inside and outside the BCSP in
BCSP decreased from 2008 to 2016, especially in the Flanders. A median 48.4% of women aged 50–69 are
youngest age group, while an increase of screening screened by the BCSP which is significantly higher
coverage inside the BCSP was seen in all age groups, than the 14.1% of women screened outside the pro-
Fig. 1. gram. Working women in younger age group (50–54
From the univariate analysis it followed that signifi- years of age), and women living in crowded house-
cantly less women were screened outside the BCSP holds with low dental care go less frequently to the
than inside the BCSP (OR: 0.184, 95% CI: 0.180– screening, and if they go, they tend to be screened
0.189). The probability of being screened in or outside more frequently outside the context of the BCSP.

Table 2 Median screening coverage (P25-P27) in Flanders

Screening coverage (%): Median (P25-P75)
Population BC screening Non-population BC screening
Overall 48.40 (41.50–54.40) 14.10 (9.80–19.80)
Age group
50–54 year 45.40 (37.50–51.30) 17.50 (13.00–24.10)
55–59 year 50.10 (42.80–56.10) 14.30 (10.00–20.50)
60–64 year 50.10 (43.60–56.20) 13.50 (9.30–18.70)
65–69 year 47.80 (42.40–53.40) 11.40 (8.20–16.00)
Ding et al. International Journal for Equity in Health (2020) 19:212 Page 5 of 9

Fig. 1 Median (P25-P75) screening coverage rate of 295 municipalities by age groups in 2008–2016

The total median coverage rate of 60.90% of screen- the BCSP. A similar pattern is also observed in coun-
ing inside and outside the BCSP is within the range of tries like France [6, 31] and the United States of
coverage levels of European countries (average: 48.2% America [32] where both screening strategies are
(range: 19.4–88.9%)) [27]. The median coverage rate provided in large scale. A potential explanation can
of the BCSP in Flanders of 48.4% is close to the cover- be that older women are more likely to attend the
age rate of the BCSP in countries such as France relatively fixed time and place of the BCSP than
(52.8%) [6] and Switzerland (46.7%) [7, 27] and higher younger working women.
than in Serbia (38.0%) [28]. In these three countries We found that living in crowded households, living
there is screening in and outside the context of the in an area with high population density, and having a
BCSP. However, it is much lower than the coverage low dental care are associated with a lower probability
rate of the BCSP in some western and northern of being screened. These three characteristics are all
European countries like the United Kingdom (78.0%), indicators for a low SES. People living in areas with a
the Netherlands (78.5%), and Norway (72.1%) [27] high population density tend to have a lower SES [33].
where only the BCSP is endorsed as the population People living in crowded household are more likely to
screening strategy. fall into income poverty [34]. As dental care is not
From 2006 to 2016, the coverage rate of BCSP fully covered by the health insurance system in Flan-
increased while the coverage rate outside the BCSP ders [35], a low dental care indicates a lower SES [36].
decreased. This effect might be explained by public Similar associations are also available in the literature
health campaigns via mass media and community regarding the increased BCSP coverage and increased
education programs [24, 29], which increased the dental care [19], less crowded household condition
visibility and awareness of BCSP for the target popu- [14], and decreased population density [37].
lation and their doctors [29, 30]. A decrease in Interestingly, women that are characterized by living
screening coverage rate was observed from 17.50 to in an area with high population density, living in a
11.40% for the individuals from age 50–54 to 65–69 more crowded households, or having a low dental care
years old in the screening outside the BCSP, whereas tend to go more frequently for screening outside the
this pattern was not observed for the individuals in BCSP. The reverse SES gradient in the use of screening
Ding et al. International Journal for Equity in Health (2020) 19:212 Page 6 of 9

Table 3 Univariate analysis of the determinants of screening in or outside the population BC screening
Variable Crude OR (95% CI) P value
Year 1.002 (0.996–1.008) 0.534
Age group < 0.001
50–54 year ref
55–59 year 1.034 (0.993–1.077)
60–64 year 1.014 (0.973–1.057)
65–69 year 0.933 (0.894–0.973)
BC screening < 0.001
Population BC screening ref
Non-population BC screening 0.184 (0.180–0.189)
Population and households
number of residents (105 residents) 0.962 (0.941–0.983) < 0.001
population density (1000 residents per km2) 0.918 (0.888–0.949) < 0.001
natural balance (per 1000 residents) 0.996 (0.990–1.002) 0.194
same address as last year (compared to all residents) 1.024 (1.013–1.035) < 0.001
non-Belgian nationality (compared to all residents) 0.996 (0.993–0.999) 0.016
average household size 1.282 (1.138–1.444) < 0.001
Welfare and poverty
women with equivalent living wages (compared to all women) 0.899 (0.855–0.945) < 0.001
share of borrowers with at least one overdue loan (compared to all borrowers) 0.970 (0.958–0.982) < 0.001
job seekers (compared to all residents) 0.961 (0.936–0.987) 0.004
Health and handicap
physical disability18-64y (compared to all residents of 18-64y) 1.003 (0.986–1.021) 0.701
diabetes (compared to all residents) 0.972 (0.954–0.991) 0.003
dental visit (compared to all residents) 1.009 (1.006–1.012) < 0.001

in and outside BCSP was also seen in other settings The strength of this study is that we examined deter-
where both screening strategies coexist [6, 7, 37]. An minants of coverage rate of screening in and outside
explanation for this phenomenon is that women with a the BCSP with longitudinal administrative data instead
higher SES are more likely to have a higher level of of self-reported screening uptake, which may induce
health literacy [38]. For these women, information recall bias. For that, regular collected and maintained
regarding the importance of mammography screening administrative data of screening coverage outside the
and the systematic quality control is more likely to mo- BCSP were applied. This enabled us to evaluate the
tivate them to participate in the BCSP [5] [29]. Another determinants of the two coexisting screening strategies
explanation is that poor employed women could have for BC and to better understand which further efforts
less flexible working time, which can conflict with the are needed to improve the coverage of the BCSP in
fixed working time of organized screening units [6, 7, Flanders. However, our study had some limitations as
37, 39]. It has also been mentioned that areas with a well. First, a limitation of this study was the use of
higher population density have a lower population BC aggregated data, which reduced the options to evaluate
screening capacity (defined as the number of mammog- correlation structures in the data [41]. Similarly, due to
raphy facilities per 10,000 women) [40] and that in the aggregated data, a variation of coverage rate and
these areas there are more private clinics for opportunistic the associated determinants within a municipality can
screening [37]. As a lower capacity of screening units can be concealed. However, the association between the
induce a longer waiting time and therefore a lower satis- determinants and screening uptake in our study is con-
faction of screening experience [5], low SES women living sistent with other studies that applied neighborhood or
in these areas might be more likely to have negative individual level factors [13, 18, 19]. Second, proxy vari-
screening experience and as a consequence prefer to ables for SES were applied instead of income which
go for screening outside the BCSP [22]. can directly characterize SES of women. However, the
Ding et al. International Journal for Equity in Health (2020) 19:212 Page 7 of 9

Table 4 Multivariable analysis of the determinants of screening in or outside the population BC screening
Variable Model 1a Model 2b
Adjusted OR (95% CI) P value Adjusted OR (95% CI) P value
Age group < 0.001 < 0.001
50–54 year ref ref
55–59 year 1.039 (1.015–1.065) 1.207 (1.179–1.235)
60–64 year 1.018 (0.994–1.042) 1.217 (1.190–1.244)
65–69 year 0.928 (0.907–0.949) 1.142 (1.117–1.167)
BC screening < 0.001 < 0.001
Population BC screening ref ref
Non-population BC screening 0.224 (0.220–0.229) 0.303 (0.295–0.312)
Population and households
number of residents (105 residents) 0.983 (0.965–1.001) 0.059 0.996 (0.985–1.007) 0.444
population density (1000 residents per km2) 0.929 (0.906–0.952) < 0.001 0.647 (0.634–0.660) < 0.001
same address as last year (compared to all residents) 1.008 (0.998–1.018) 0.123 1.005 (0.995–1.015) 0.309
non-Belgian nationality (compared to all residents) 1.0005 (0.9984–1.0025) 0.654 0.9997 (0.9982–1.0012) 0.719
average household size 0.894 (0.809–0.988) 0.028 0.580 (0.522–0.645) < 0.001
Welfare and poverty
women with equivalent living wages 0.972 (0.934–1.012) 0.164 0.987 (0.947–1.029) 0.532
(compared to all women)
share of borrowers with at least one overdue loan 0.989 (0.976–1.002) 0.092 0.989 (0.978–1.001) 0.067
(compared to all borrowers)
job seekers (compared to all residents) 1.073 (1.051–1.095) < 0.001 1.250 (1.226–1.273) < 0.001
Health can handicap
diabetes (compared to all residents) 0.964 (0.952–0.976) < 0.001 0.985 (0.973–0.997) 0.016
dental visit (compared to all residents) 1.005 (1.003–1.007) < 0.001 1.016 (1.015–1.018) < 0.001
Interaction terms
age group × BC screening < 0.001
NPS × 50–54 year ref
NPS × 55–59 year 0.668 (0.642–0.694)
NPS × 60–64 year 0.612 (0.589–0.636)
NPS × 65–69 year 0.554 (0.533–0.576)
NPS × population density 2.528 (2.455–2.606) < 0.001
NPS × average household size 3.797 (3.199–4.508) < 0.001
NPS × job seekers 0.641 (0.624–0.658) < 0.001
NPS × status of diabetes 0.942 (0.921–0.962) < 0.001
NPS × dental visit 0.969 (0.967–0.972) < 0.001
a
model 1: multivariable regression model including all significant covariates of the univariate regression
b
model 2: multivariable regression model including two-way interaction terms between screening strategies and the significant covariates in model 1
NPS Non-population BC screening

proxy variables used are commonly applied and the characterized by younger age, living in a high popula-
magnitude and direction of the association between tion density area, living in crowded households, or
variables is consistent with the literature [6, 14, 18]. having low dental care, go less frequently to screen-
ing. If they go to screening, they are more likely to
Conclusion be screened outside the BCSP. Further efforts targeted
A sizeable part of women attend screening outside on this group of women are needed to improve the
the BCSP in Flanders. Women with low SES that are coverage rate of the BCSP in Flanders.
Ding et al. International Journal for Equity in Health (2020) 19:212 Page 8 of 9

Abbreviations 7. Eichholzer M, et al. Breast cancer screening attendance in two Swiss regions
BC: Breast cancer; BCSP: Breast cancer screening program; GP: General dominated by opportunistic or organized screening. BMC Health Serv Res.
practitioner; SES: Socioeconomic status; IQR: Interquartile range; 2016;16(1):519.
GEE: Generalized estimating eqs.; OR: Odds ratios; CI: Confidence interval 8. Massat N, et al. Variation in cervical and breast cancer screening coverage in
England: a cross-sectional analysis to characterise districts with atypical
Acknowledgements behaviour. BMJ Open. 2015;5:e007735.
L. Ding is supported by the China Scholarship Council (CSC) PhD scholarship 9. Perry N, et al. European guidelines for quality assurance in breast cancer
(file NO. 201808320439) for his research and study at the University of screening and diagnosis. Fourth edition--summary document. Ann Oncol.
Groningen, University Medical Center Groningen, Groningen, The 2008;19(4):614–22.
Netherlands. The scholarship had no role in study design, data analysis and 10. OECD.Stat, Health Care Utilisation. https://stats.oecd.org/Index.
interpretation, the writing of the manuscript and the decision to submit the aspx?DataSetCode=HEALTH_PROC. Accessed 3 Feb 2020.
article for publication. 11. Annual Report Population Screening for Cancer 2018. Center for Cancer
detection-Belgian Cancer registry, Bruges, 2018.
Authors’ contributions 12. Sarma EA. Barriers to screening mammography. Health Psychol Rev. 2015;
L. Ding: Conceptualization, Methodology, Writing- Original draft preparation. 9(1):42–62.
S. Jidkova: Data curation, Writing- Reviewing and Editing, Validation. M.J.W. 13. Katz D, et al. Patient and physician characteristics affect adherence to
Greuter: Conceptualization, Methodology, Writing- Reviewing and Editing. screening mammography: a population-based cohort study. PLoS One.
G.H. de Bock: Supervision. K. Van Herck: Writing- Reviewing and Editing, 2018;13(3):12.
Validation. M. Goossens: Writing- Reviewing and Editing, Validation. P. 14. Zackrisson S, et al. Non-attendance in breast cancer screening is associated
Martens: Writing- Reviewing and Editing, Validation. G. Van Hal: with unfavourable socio-economic circumstances and advanced carcinoma.
Conceptualization, Methodology, Writing- Reviewing and Editing. The Int J Cancer. 2004;108(5):754–60.
author(s) read and approved the final manuscript. 15. Jensen LF, et al. Identifying specific non-attending groups in breast cancer
screening - population-based registry study of participation and socio-
Funding demography. BMC Cancer. 2012;12:9.
None. 16. Smith D, et al. The breast cancer paradox: a systematic review of the
association between area-level deprivation and breast cancer screening
Availability of data and materials uptake in Europe. Cancer Epidemiol. 2019;60:77–85.
Breast cancer screening coverage dataset is available at https:// 17. Pornet C, et al. Socioeconomic and healthcare supply statistical
bevolkingsonderzoek.incijfers.be/, variables regarding the determinants of determinants of compliance to mammography screening programs: a
screening coverage can be requested by contacting the Center for Cancer multilevel analysis in Calvados, France. Cancer Epidemiol. 2010;34(3):309–15.
Detection in Flanders at www.bevolkingsonderzoek.be. 18. Zackrisson S, et al. Social predictors of non-attendance in an urban
mammographic screening programme: a multilevel analysis. Scand J Public
Ethics approval and consent to participate Health. 2007;35(5):548–54.
Not applicable. 19. von Euler-Chelpin M, et al. Socio-demographic determinants of participation
in mammography screening. Int J Cancer. 2008;122(2):418–23.
Consent for publication 20. Zha N, et al. Beyond Universal Health Care: Barriers to Breast Cancer
All authors have approved the manuscript and agree with this submission. Screening Participation in Canada. J Am Coll Radiol. 2019;16(4 Pt B):
570–9.
Competing interests 21. Miles RC, et al. Chronic medical illness as a risk factor for poor
None. mammography screening adherence. J Women's Health (Larchmt). 2019;
28(10):1378–83.
Author details 22. Ferrat E, et al. Understanding barriers to organized breast cancer screening
1
Department of Epidemiology, University Medical Center Groningen, in France: women's perceptions, attitudes, and knowledge. Fam Pract. 2013;
University of Groningen, Groningen, The Netherlands. 2Department of Social 30(4):445–51.
Epidemiology and Health Policy, University of Antwerp, Antwerp, Belgium. 23. Van Oyen H, Verellen W. Breast cancer screening in the Flemish region,
3 Belgium. Eur J Cancer Prev. 1994;3(Suppl 1):7–12.
Department of Public Health and Primary Care, Ghent University, Ghent,
Belgium. 4Center for Cancer Detection, Flanders, Belgium. 5Department of 24. Willems B, Bracke P. The impact of regional screening policies on the
Radiology, University Medical Center Groningen, University of Groningen, diffusion of cancer screening participation in Belgium: time trends in
Groningen, The Netherlands. 6Department of Robotics and Mechatronics, educational inequalities in Flanders and Wallonia. BMC Health Serv Res.
University of Twente, Enschede, The Netherlands. 2018;18(1):943.
25. Goossens M, et al. Flemish breast cancer screening programme: 15 years of
Received: 13 April 2020 Accepted: 11 November 2020 key performance indicators (2002-2016). BMC Cancer. 2019;19(1):1012.
26. Portnov BA, Dubnov J, Barchana M. On ecological fallacy, assessment errors
stemming from misguided variable selection, and the effect of aggregation
References on the outcome of epidemiological study. J Expo Sci Environ Epidemiol.
1. Bray F, et al. Global Cancer Statistics 2018: GLOBOCAN estimates of 2007;17(1):106–21.
incidence and mortality worldwide for 36 cancers in 185 countries: global 27. Giordano L, et al. Mammographic screening programmes in Europe:
Cancer statistics 2018. CA Cancer J Clin. 2018;68(6):394–424. organization, coverage and participation. J Med Screen. 2012;19(Suppl
2. International Agency for Research on Cancer. CANCER TODAY: Population 1):72–82.
fact sheets-Belgium, http://gco.iarc.fr/today/data/factsheets/populations/56- 28. Jovicevic A, et al. Factors influencing participation in breast cancer
belgium-fact-sheets.pdf; 2018[Accessed 3 February 2020]. opportunistic screening in Belgrade, Serbia. J Buon. 2018;23(3):706–12.
3. Myers ER, et al. Benefits and harms of breast Cancer screening: a systematic 29. Wouters M, Vleminckx F, Van Hal G. How to reach a higher participation
review. JAMA. 2015;314(15):1615–34. rate for breast cancer screening in Flanders? Arch Public Health Brussels.
4. Lauby-Secretan B, et al. Breast-Cancer screening — viewpoint of the IARC 2006;64(2/3):109–21.
working group. N Engl J Med. 2015;372(24):2353–8. 30. Rogers, E.M. Diffusion of innovations.5th ed. New York: Free Press; 2003.
5. Goossens M, et al. Quantifying independent risk factors for failing to 31. Ouedraogo S, et al. Breast cancer screening programmes: challenging the
rescreen in a breast cancer screening program in Flanders, Belgium. Prev coexistence with opportunistic mammography. Patient Educ Couns. 2014;
Med. 2014;69:280–6. 97(3):410–7.
6. Duport N. Characteristics of women using organized or opportunistic breast 32. Beaber EF, et al. Multilevel predictors of continued adherence to breast
cancer screening in France. Analysis of the 2006 French health, health care cancer screening among women ages 50-74 years in a screening
and insurance survey. Rev Epidemiol Sante Publique. 2012;60(6):421–30. population. J Women's Health. 2019;28(8):1051–59.
Ding et al. International Journal for Equity in Health (2020) 19:212 Page 9 of 9

33. Zhu Z, et al. The impact of urban characteristics and residents’ income on
commuting in China. Transp Res Part D: Transp Environ. 2017;57:474–83.
34. Melki IS, et al. Household crowding index: a correlate of socioeconomic
status and inter-pregnancy spacing in an urban setting. J Epidemiol
Community Health. 2004;58(6):476.
35. Lambert MJ, et al. Socioeconomic inequalities in caries experience, care
level and dental attendance in primary school children in Belgium: a cross-
sectional survey. BMJ Open. 2017;7(7):e015042.
36. Kailembo A, et al. Income and wealth as correlates of socioeconomic
disparity in dentist visits among adults aged 20 years and over in the
United States, 2011–2014. BMC Oral Health. 2018;18(1):147.
37. Jensen A, et al. Do nonattenders in mammography screening programmes
seek mammography elsewhere? Int J Cancer. 2005;113(3):464–70.
38. Guo Y, et al. Health literacy: a pathway to better oral health. Am J Public
Health. 2014;104(7):e85–91.
39. Collie-Akers VL, et al. Assessment of characteristics of capacity among breast
cancer screening facilities. J Community Health. 2012;37(3):626–31.
40. Elkin EB, et al. Changes in the availability of screening mammography, 2000-
2010. Cancer. 2013;119(21):3847–53.
41. Russo L, Beauguitte L. Aggregation level matters: evidence from French
electoral data. Qual Quant. 2014;48(2):923–38.

Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.