- Email: [email protected]
Determinants of participation in organized colorectal cancer screening in Isère (France)
Clinics and Research in Hepatology and Gastroenterology (2013) 37, 193—199
Available online at
www.sciencedirect.com
ORIGINAL ARTICLE
Determinants of participation in organized colorectal cancer screening in Isère (France) Florence Poncet a,∗, Patricia Delafosse a, Arnaud Seigneurin a, Catherine Exbrayat b, Marc Colonna a a b
Registre du cancer de l’Isère, CHU de Grenoble, Pav. E, BP 217, 38043 Grenoble cedex 9, France Office de lutte contre le cancer (ODLC), 38240 Meylan, France
Available online 14 June 2012
Summary Background and aims: In France, participation in organized colorectal cancer screening remains low. The objective of this study was to identify the determinants of participation in colorectal cancer screening in Isère, a French administrative entity. Methods: This study examined the target population invited for screening between 2007 and 2008 in Isère. The statistical analysis method was based on a two-level logistic regression model: the first was the individual level relative to the individuals invited for screening and the second was an aggregate level corresponding to the socioeconomic level of an invited person’s residence area (IRIS: ‘‘Ilot regroupé pour l’Information Statistique’’; Regrouped statistical information block). The evaluation of the socioeconomic level was based on the Townsend deprivation. Results: Participation varied depending on sex, age, and health insurance plan. The people residing in the least deprived IRISes participated more than individuals residing in the most deprived IRISes. The multilevel analysis showed a 24% difference in participation between the least and the most deprived IRISes. Conclusions: The use of socioeconomic data on the IRIS geographical unit has identified, socially and geographically, the populations that participate the least, although this reflects ‘‘mean’’ behaviors. These results could be used to set up targeted actions to encourage participation in these populations. © 2012 Elsevier Masson SAS. All rights reserved.
Introduction
∗ Corresponding author. Tel.: +33 4 76 90 76 10; fax: +33 4 76 41 87 00. E-mail address: [email protected] (F. Poncet).
In France in 2011, the number of new cases of colorectal cancer was estimated at 40,500 and the number of deaths at 17,500 [1]. In terms of incidence, this cancer location ranks third after prostate and breast cancer and ranks second in terms of mortality, behind lung cancer.
2210-7401/$ – see front matter © 2012 Elsevier Masson SAS. All rights reserved. http://dx.doi.org/10.1016/j.clinre.2012.04.011
194 Studies have shown that the mortality of this cancer can be reduced by 15—20% by performing a fecal occult blood test every 2 years in individuals who present no particular risk for this cancer [2—5]. In Isère department (a French administrative entity), an experimental organized screening campaign was set up in 1990 for women 50—69 years of age in association with breast cancer screening [6]. In 2002, this test was proposed to both men and women 50—74 years of age in accordance with the national screening program [7]. Between 2002 and 2005, 23 geographical departments set up this organized screening program. It was then generalized in 2008 and 2009 to the entire population in France [8]. This organized screening program is designed for a population with a moderate risk aged 50—74 years [9]. Individuals are identified from health insurance files. The healthcare management organization invites them by post every 2 years to take the home screening test (Hemoccult® test). The test can be obtained from their family physician or by post [7]. If the test is positive, a colonoscopy is necessary. The participation rate of the target population is one of the fundamental parameters conditioning the organized screening program’s efficacy. The French program has been evaluated by the National Health Watch Institute (Institut de Veille Sanitaire [InVS]) based on data collected by the different management organizations. The assessments established in December 2009 and December 2010 report insufficient participation (37% in 2007—2008) and the need to set up priority actions to improve this indicator [10,11]. The objective of this study was to conduct a detailed geographic analysis of participation in Isère, a French administrative entity, so as to identify determinants of this participation and locate the areas in which specific actions could be adapted in relation to adherence to the program and the socioeconomic context.
Material and methods Study design We analyzed the original data from an ongoing population based colorectal cancer screening program in Isère, a French administrative entity with nearly 1.2 million inhabitants.
F. Poncet et al.
Data collection Data on the study population were obtained from the Office De Lutte Contre le Cancer (ODLC), which coordinates cancer screening in Isère. Information on individuals characteristics included sex, age, health insurance plan. The precise residential address, participation or non-participation in the organized screening program, as well as any causes for not taking the screening test were known for every person invited for screening. In France, there are different health insurance plans and the most common is the general plan. The agricultural plan covering agricultural occupations. The special plan concerns persons belonging to the SNCF (national railways), the RATP (Parisian transport), the electrical and gas companies (EDF and GDF) as well as some employees whose functions are directly related to the State such as the military, French National Police. . . The other coverage including civil service plan and schemes for the self-employed. Individuals who had taken the screening test during the period were defined as participants. In absence of individual socioeconomic data, the aggregate data at the ‘‘Ilôts Regroupés pour l’Information Statistique’’ (IRIS, or regrouped statistical information block) from the 2007 national population census of the French National Institute for Statistics and Economic Studies (INSEE) [12] provided the social environment of the subjects invited. The IRIS is the smallest geographical unit for which INSEE data are available. An IRIS corresponds to a small neighborhood, which is defined as a group of contiguous districts of approximately 2000 inhabitants. All towns with more than 10,000 inhabitants and most towns with 5000—10,000 inhabitants are broken down into several IRISes. The other towns form one IRIS. Isère department is broken down into 749 IRISes. Each person invited was assigning to an IRIS by geocoding the exact residential address. To define the socioeconomic level of an IRIS, the Townsend Index was used [13]. This index is used in the Anglo-Saxon countries to measure, in an aggregate fashion, the socioeconomic levels when individual data is missing [14]. This index was constructed based on four socioeconomic variables: the unemployment rate, the proportion of households without a car, the proportion of non-home owner households, and the proportion of overcrowded housing units (more than one person per room).
Study population
Statistical analyses
This study examined the target population for organized screening (50—74 years of age) invited for screening between January 2007 and December 2008 in Isère. The individuals whose address was incomplete or imprecise were excluded, as were those for whom medical contraindications were identified in accordance with the national program (presence of digestive symptoms, high risk of colorectal cancer based on coloscopic screening, another serious extra-intestinal disease, another particular contraindication, or a complete normal coloscopic exam less than 5 years before).
Given the hierarchical structure of the data, the statistical analysis method was based on a random-intercept two-level logistic regression [15,16]. Individual and aggregate data were used. The relevant variable was participation in the organized screening campaign: for each IRIS, the participation rate corresponds to the number of participants over the number of invited individuals. The first level was the individual level (sex, age, health insurance plan) compared to the invited individuals. The second level was the aggregate level corresponding to the
Participation in colorectal screening Table 1
195
Description of the target population in the organized colorectal cancer screening program in Isère (France, n = 247,776). Non-participation
Participation
n
%
n
%
Sex Male Female
74732 73678
62.1 57.8
45,667 53,699
Age 50—54 years 55—59 years 60—64 years 65—69 years 70—74 years
43875 35418 30407 18780 19930
63.4 61.2 58.3 54.3 58.5
107993 3418 18450 12282 6267 18319 21401 24430 42021 42239
Health insurance plan General plan Special plan Civil service plan Self-employed plan Agricultural plan Townsend Index Quintile 1 (least deprived IRISes) Quintile 2 Quintile 3 Quintile 4 Quintile 5 (most deprived IRISes)
Crude OR
95% CI
37.9 42.2
1 1.19
1.17—1.21
25,289 22,442 21,721 15,788 14,126
36.6 38.8 41.7 45.7 41.5
1 1.10 1.24 1.46 1.23
1.07—1.12 1.21—1.27 1.42—1.50 1.20—1.26
60.3 51.7 53.4 68.3 66.3
71,200 3190 16,082 5704 3190
39.7 48.3 46.6 31.7 33.7
1 1.42 1.32 0.70 0.77
1.35—1.49 1.29—1.35 0.68—0.73 0.74—0.81
56.8 58.1 58.4 59.8 63.4
13,960 15,428 17,370 28,196 24,412
43.2 41.9 41.6 40.2 36.6
1 0.95 0.93 0.88 0.76
0.91—0.98 0.91—0.96 0.86—0.90 0.74—0.78
socioeconomic level (Townsend Index) of an invited person’s residential area (IRIS). The Townsend Index was broken down by quintile. Quintile 1 represented the least deprived IRISes and quintile 5 the most deprived IRISes. Three two-level logistic regression models were successively estimated. The first model estimated was a model with no parameters (the empty model). Since it contains only random effects at the IRIS level, this empty model can detect the existence of a contextual effect on participation. The individual factors (characterizing the first level) were introduced into the second model and tested whether the variations between the IRISes may be explained by the differences in how the groups were composed in terms of individual characteristics. Finally, in the last model, the socioeconomic characteristics defined by the Townsend Index of each IRIS (second level) were added to the individual factors. This model can search to determine whether the differences in participation between the IRISes were explained by the characteristics of the IRISes. The analyses were performed with STATA 10.0 software.
Results The organized colorectal cancer screening invitations in Isère in 2007 and 2008 involved 281,987 people, including 26,037 who were secondarily excluded from the organized screening either for medical reasons (16,688) or because of an erroneous address and non-receipt of the invitation (9349).
A residential address that was not sufficiently precise that did not allow geocoding also led to excluding 2.7% of the individuals invited. In the end, the study included 247,776 individuals. The characteristics of the target population are described in Table 1. The crude participation rate was 40.1%, varying depending on sex, age, and health insurance plan. Women participated more than men, with an OR = 1.19 (95%CI [1.17—1.21]). The participants were on average slightly older (60.4 years) than the non-participants (59.7years) (P < 0.01) with a different participation depending on the age group. The 65- to 69-year-old group presented the highest participation rate. The 50- to 54-year-old age group presented the lowest participation rate. The beneficiaries of special plans participated the most (OR = 1.42; 95%CI [1.35—1.49]), whereas the beneficiaries of the agricultural healthcare plan participated the least (OR = 0.77; 95%CI [0.74—0.81]). The participation rate also varied in relation to the Townsend Index. The more deprived the IRIS, the lower participation was (Fig. 1). The participation rate in the least deprived IRISes (quintile 1 of the Townsend Index) was 43% and 37% in the most deprived IRISes (quintile 5 of the Townsend Index). The results of the multilevel analysis (Table 2) show that participation varied significantly between the residential IRISes (empty model + Fig. 2). The significant individual variables in univariate analysis were introduced into the multilevel analysis.
196
F. Poncet et al.
Table 2
Factors associated with participation in colorectal cancer screening in Isère (France) — multilevel analysis (n = 247,776). Empty model
Model 1 OR
Model 2 95%CI
OR
95%CI
Level 1: individuals Sex Male Female
1 1.16
1.14
1.18
1 1.16
1.14
1.18
Age 50—54 years 55—59 years 60—64 years 65—69 years 70—74 years
1 1.11 1.25 1.48 1.26
1.08 1.22 1.44 1.22
1.13 1.28 1.52 1.29
1 1.11 1.25 1.48 1.26
1.08 1.22 1.44 1.23
1.13 1.28 1.52 1.29
Health insurance plan General plan Special plan Civil service plan Self-employed plan Agricultural plan
1 1.43 1.29 0.72 0.73
1.36 1.26 0.70 0.69
1.51 1.32 0.75 0.76
1 1.43 1.29 0.72 0.71
1.36 1.26 0.70 0.68
1.5 1.32 0.74 0.75
1 0.96 0.95 0.90 0.75
0.91 0.90 0.85 0.71
1.02 1.00 0.95 0.79
Level 2: IRIS Townsend Index Quintile 1 (least deprived IRISes) Quintile 2 Quintile 3 Quintile 4 Quintile 5 (most deprived IRISes) Random effects Level 2 variance P-value
0.038 < 0.01
0.039 < 0.01
0.027 < 0.01
differences in participation between the IRISes. Introducing the Townsend Index to the model (Table 2, model 2) reduced the inter IRIS variance by 30.8%, signifying that socioeconomic status explained almost third of the disparities between IRISes. The people residing in the most deprived IRISes participated less than those living in the least deprived IRISes. After adjusting for individual variables, the multilevel analysis showed a 25% difference in participation between the least deprived IRISes (IRIS in the 1st quintile of the Townsend Index) and the most deprived IRISes (IRIS in the 5th quintile of the Townsend Index; OR = 0.75 [0.71—0.79]).
Discussion
Figure 1 Participation rate in the colorectal cancer screening program in Isère (France) in relation to the Townsend Index and sex (n = 247,776).
Between the empty model and model 1, in which the subjects’ characteristics were introduced (sex, age, and healthcare plan), the variance increased slightly, which indicates that the individual factors do not explain the
This study conducted on the entire target population invited for an organized colorectal cancer screening program between 2007 and 2008 in Isère showed that socioeconomic characteristics influence the participation in the organized colorectal screening program. The multivariate analysis demonstrated that there was less participation in the IRISes with higher levels of socioeconomic disadvantage. The role of unfavorable socioeconomic status as a factor of lower participation has been demonstrated in many studies but most of these studies used self-administered
Participation in colorectal screening
Figure 2 unit).
197
Crude participation rate in organized colorectal cancer screening program in Isère (n = 749 IRISes, French geographical
questionnaires and were therefore subject to participation bias. Age, sex, marital status, and educational level also had an impact on participation [17—22]. The present study used the same analysis strategy as a study conducted in the Calvados department (France), thus avoiding selection bias. In Calvados, the study investigated only a representative sample of the target population. The results of these two studies are in agreement and show that participation in organized screening is higher in the least deprived IRISes [23]. These results are also in agreement with two British studies, which also used aggregate data. However, these studies present methodological differences (multivariate analysis with no level 2 random effect). They differed in their method for inviting subjects to participate in screening. In England, the screening test is directly sent by mail to the target population, whereas in France, the target population is first invited to consult their family physician to obtain the screening test. The first study examined the results of the pilot organized screening program in England [24]. In the most disadvantaged areas, participation was lower than in higher socioeconomic level areas, with an OR of 0.68 (95%CI [0.59—0.79]). The second UK study examined the participation inequalities in the first campaign in the national colorectal cancer screening program in England [25]. It showed a linear association between participation and the deprivation index, participation in the least deprived areas was 50% higher than in the most deprived areas.
A Danish study [26] also analyzed the effect of several socioeconomic factors on participation. Based on individual data, it showed that a low socioeconomic level measured through educational level, employment, and income was strongly related to low participation. The present study presents some limitations. In absence of individual socioeconomic data, we chose to use aggregate data on the IRIS geographical unit from the INSEE census, which could result in a classification bias of the subjects [27,28]. The Townsend Index is considered to be identical for all people residing in the same IRIS, whereas large disparities may exist within an IRIS: consequently, these represent ‘‘mean’’ behaviors. The Townsend Index was used to characterize the socioeconomic level because no other valid, published French deprivation index exists based on the IRISes and covering the entire country. This index was drawn up in the context of the United Kingdom and probably only partially accounts for the French socioeconomic level. However, it has allowed us, most notably, to compare our results with the Calvados study. Some people targeted by the organized screening campaign were excluded from the analysis because their residential address was too imprecise for geocoding to the IRIS geographical unit. These exclusions for the most part concerned men between 50 and 54 years of age who did not participate in the screening program. Their inclusion in the study would probably have reinforced the results, since
198 men in the 50- to 54-year-old age group where those who participated the least. Among the non-participants, there were probably people who should not participate in organized screening because of a contraindication, but this information was declarative; the absence of a response by the person invited or by the physician in the healthcare management organization increased the number of non-participants. Conversely, some people underwent the screening test when they should not have (medical contraindications). Since 2007, the ODLC had used the databases of the medical data of healthcare facilities so as not to invite, for example, people who had undergone a colonoscopy in less than 5 years. Knowledge of these exclusions is important to optimally evaluate the participation rate [11]. The different studies on participation show that incentive actions are necessary, aimed at both the target population and practitioners. Involvement of family physicians can increase participation. The test is more often taken when given by the subject’s doctor than when sent directly to the subject [29,30]. The family physician plays an essential role in the screening program because he or she assesses what to do in relation to the level of risk. The consultation is the opportunity to give the subject the test or, if need be, to point out a medical contraindication to organized screening and orient the patient toward more adapted care. The physician’s role is also fundamental in explaining the procedure to the patient as well as the consequences of the result. He or she can answer questions and respond to worries associated with this procedure. A number of studies have underscored the importance of identifying how to optimize participation so as to have an impact on the ‘‘social’’ inequalities concerning colorectal cancer [31,32]: some authors emphasize the need for specific actions to reduce the differences in participation [33—35]. This study has provided a social and geographic identification of populations that participate the least using socioeconomic data at the IRIS unit, even if only ‘‘mean’’ behaviors are reflected. These results could be used to develop targeted incentive actions in the areas thus identified as low participators. Conducting exhaustive studies on the motivations for participation in individuals sharing a priori similar socioeconomic characteristics could contribute to identifying specific initiatives to improve the adherence of populations to this screening campaign.
Disclosure of interest The authors declare that they have no conflicts of interest concerning this article.
F. Poncet et al.
[2]
[3]
[4]
[5]
[6]
[7]
[8] [9]
[10]
[11]
[12]
[13] [14]
[15] [16]
Acknowledgments This study was funded by the Ligue Nationale Contre le Cancer — Comité départemental de l’Isère and by the Cancéropôle Lyon Auvergne Rhône-Alpes (CLARA).
[17]
[18]
References [19] [1] Hospices civils de Lyon/Institut de veille sanitaire/Institut national du cancer/Francim/Institut national de la santé et de la recherche médicale. Projections de l’incidence et de
la mortalité par cancer en France en 2011. Rapport technique; juin 2011.*Detailed results and comments [online] http://www.invs.sante.fr/surveillance/cancers [accessed 7th June 2011]. Kronborg O, Fenger C, Olsen J, Jørgensen OD, Søndergaard O. Randomised study of screening for colorectal cancer with faecal occult-blood test. Lancet 1996;348:1467—71. Hardcastle JD, Chamberlain JO, Robinson MH, Moss SM, Amar SS, Balfour TW, et al. Randomised controlled trial of faecal-occult-blood screening for colorectal cancer. Lancet 1996;348:1472—7. Faivre J, Dancourt V, Lejeune C, Tazi MA, Lamour J, Gerard D, et al. Reduction in colorectal cancer mortality by fecal occult blood screening in a French controlled study. Gastroenterology 2004;126:1674—80. Mandel JS, Bond JH, Church TR, Snover DC, Bradley GM, Schuman LM, et al. Reducing mortality from colorectal cancer by screening for fecal occult blood. Minnesota Colon Cancer Control Study. N Engl J Med 1993;328:1365—71 [Erratum in: N Engl J Med 1993;329:672]. Exbrayat C, Garnier A, Bolla M, Winckel P, Salicru B, Marron J, et al. Simultaneous screening of cancers of the breast, cervix, colon and rectum. The Isère experience. Bull Cancer 1996;83:641—8. Ministère chargé de la Santé et des Solidarités. Arrêté du 29 septembre 2006 relatif aux programmes de dépistage des ¸aise cancers, annexes. Journal Officiel de la République franc no 295 du 21 décembre 2006. Viguier J. L’organisation du dépistage du cancer colorectal en France. BEH 2009;2-3:19—21. Consensus conference: Prevention screening and management of the colonic cancers. Paris, France, January 29—30, 1998. Proceedings. Gastroenterol Clin Biol 1998;22:S1—295. Goulard H, Boussac-Zarebska M, Salines E, Bloch J. Évaluation épidémiologique du dépistage organisé du cancer du côlon et du rectum en France — Bilan du programme pilote : actualisation des données au premier novembre 2007. Saint-Maurice (Fra): Institut de veille sanitaire, avril 2009, 8 p. Available http://www.invs.sante.fr/publications/2009/plaq eval at: depistage cancer colon/index.html. Goulard H, Jezewski-Serra D, Duport N, Danzon A. Évaluation épidémiologique du dépistage organisé du cancer colorectal en France — résultats des programmes pilotes au-delà de la première campagne. Saint-Maurice: Institut de veille sanitaire; décembre 2010. 12 p. Available at: http://www.invs. sante.fr/publications/2010/plaquette colorectal/index.html. Recensement de la population 2007 : bases de données infra-communales — IRIS — 2007 [fichier électronique], Insee [producteur], centre Maurice Halbwachs (CMH) [diffuseur]. Townsend P. Deprivation. J Soc Pol 1987;16:125—46. Krieger N, Williams DR, Moss NE. Measuring social class in US public health research: concepts, methodologies, and guidelines. Annu Rev Public Health 1997;18:341—78. Diez-Roux AV. Multilevel analysis in public health research. Annu Rev Public Heath 2000;21:171—92. Chaix B, Chauvin P. The contribution of multilevel models in contextual analysis in the field of social epidemiology: a review of literature. Rev Epidemiol Sante Publique 2002;50:489—99. O’Malley AS, Forrest CB, Feng S, Mandelblatt J. Disparities despite coverage: gaps in colorectal cancer screening among medicare beneficiaries. Arch Intern Med 2005;165:212—35. Cokkinides VE, Chao A, Smith RA, Vernon SW, Thun MJ. Correlates of underutilization of colorectal cancer screening among U.S. adults, age 50 years and older. Prev Med 2003;36:85—91. Seeff LC, Nadel MR, Klabunde CN, Thompson T, Shapiro JA, Vernon SW, et al. Patterns and predictors of colorectal cancer test use in the adult U.S. population. Cancer 2004;100: 2093—103.
Participation in colorectal screening [20] Wee CC, McCarthy EP, Phillips RS. Factors associated with colon cancer screening: the role of patient factors and physician counseling. Prev Med 2005;41:23—9. [21] Weber MF, Banks E, Ward R, Sitas F. Population characteristics related to colorectal cancer testing in New South Wales, Australia: results from the 45 and Up Study cohort. J Med Screen 2008;15:137—42. [22] Goulard H, Boussac-Zarebska M, Duport N, et al. Facteurs d’adhésion au dépistage organisé du cancer colorectal: étude Fado-colorectal, France, décembre 2006—avril 2007. BEH 2009;2—3:25—9. [23] Pornet C, Dejardin O, Morlais F, Bouvier V, Launoy G. Socioeconomic determinants for compliance to colorectal cancer screening. A multilevel analysis. J Epidemiol Community Health 2010;64:318—24. [24] Weller D, Coleman D, Robertson R, Butler P, Melia J, Campbell C, et al. The UK colorectal cancer screening pilot: results of the second round of screening in England. Br J Cancer 2007;97:1601—5. [25] von Wagner C, Good A, Wright D, Rachet B, Obichere A, Bloom S, et al. Inequalities in colorectal screening participation in the first round of the national screening programme in England. Br J Cancer 2009;101(Suppl 2):S60—3. [26] Frederiksen BL, Jørgensen T, Brasso K, Holten I, Osler M. Socioeconomic position and participation in colorectal cancer screening. Br J Cancer 2010;103:1496—501. [27] Galobardes B, Shaw M, Lawlor DA, Lynch JW, Davey Smith G. Indicators of socioeconomic position (part 2). J Epidemiol Community Health 2006;60:95—101.
199 [28] Galobardes B, Lynch J, Smith GD. Measuring socioeconomic position in health research. Br Med Bull 2007;81—82:21—37. [29] Tazi MA, Faivre J, Lejeune C, Benhamiche AM, Dassonville F. Performance of the Hemoccult test in the screening of colorectal cancer and adenoma. Results of 5 screening campaigns in Saône-et-Loire. Gastroenterol Clin Biol 1999;23: 475—80. [30] Launoy G, Veret JL, Richir B, Reaud JM, Ollivier V, Valla A, et al. Involvement of general practitioners in mass-screening. Experience of a colorectal cancer mass screening program in the Calvados region (France). Eur J Cancer Prev 1993;2: 229—32. [31] Emmons K, Puleo E, McNeill LH, Bennett G, Chan S, Syngal S. Colorectal cancer screening awareness and intentions among low income, sociodemographically diverse adults under age 50. Cancer Causes Control 2008;19:1031—41. [32] Ioannou GN, Chapko MK, Dominitz JA. Predictors of colorectal cancer screening participation in the United States. Am J Gastroenterol 2003;98:2082—91. [33] James TM, Greiner KA, Ellerbeck EF, Feng C, Ahluwalia JS. Disparities in colorectal cancer screening: a guideline-based analysis of adherence. Ethn Dis 2006;16:228—33. [34] McGregor SE, Hilsden RJ, Li FX, Bryant HE, Murray A. Low uptake of colorectal cancer screening 3 yrs after release of national recommendations for screening. Amj Gastroenterol 2007;102:1727—35. [35] McGregor SE, Bryant HE. Predictors of colorectal cancer screening: a comparison of men and women. Can J Gastroenterol 2005;19:343—9.
Available online at
www.sciencedirect.com
ORIGINAL ARTICLE
Determinants of participation in organized colorectal cancer screening in Isère (France) Florence Poncet a,∗, Patricia Delafosse a, Arnaud Seigneurin a, Catherine Exbrayat b, Marc Colonna a a b
Registre du cancer de l’Isère, CHU de Grenoble, Pav. E, BP 217, 38043 Grenoble cedex 9, France Office de lutte contre le cancer (ODLC), 38240 Meylan, France
Available online 14 June 2012
Summary Background and aims: In France, participation in organized colorectal cancer screening remains low. The objective of this study was to identify the determinants of participation in colorectal cancer screening in Isère, a French administrative entity. Methods: This study examined the target population invited for screening between 2007 and 2008 in Isère. The statistical analysis method was based on a two-level logistic regression model: the first was the individual level relative to the individuals invited for screening and the second was an aggregate level corresponding to the socioeconomic level of an invited person’s residence area (IRIS: ‘‘Ilot regroupé pour l’Information Statistique’’; Regrouped statistical information block). The evaluation of the socioeconomic level was based on the Townsend deprivation. Results: Participation varied depending on sex, age, and health insurance plan. The people residing in the least deprived IRISes participated more than individuals residing in the most deprived IRISes. The multilevel analysis showed a 24% difference in participation between the least and the most deprived IRISes. Conclusions: The use of socioeconomic data on the IRIS geographical unit has identified, socially and geographically, the populations that participate the least, although this reflects ‘‘mean’’ behaviors. These results could be used to set up targeted actions to encourage participation in these populations. © 2012 Elsevier Masson SAS. All rights reserved.
Introduction
∗ Corresponding author. Tel.: +33 4 76 90 76 10; fax: +33 4 76 41 87 00. E-mail address: [email protected] (F. Poncet).
In France in 2011, the number of new cases of colorectal cancer was estimated at 40,500 and the number of deaths at 17,500 [1]. In terms of incidence, this cancer location ranks third after prostate and breast cancer and ranks second in terms of mortality, behind lung cancer.
2210-7401/$ – see front matter © 2012 Elsevier Masson SAS. All rights reserved. http://dx.doi.org/10.1016/j.clinre.2012.04.011
194 Studies have shown that the mortality of this cancer can be reduced by 15—20% by performing a fecal occult blood test every 2 years in individuals who present no particular risk for this cancer [2—5]. In Isère department (a French administrative entity), an experimental organized screening campaign was set up in 1990 for women 50—69 years of age in association with breast cancer screening [6]. In 2002, this test was proposed to both men and women 50—74 years of age in accordance with the national screening program [7]. Between 2002 and 2005, 23 geographical departments set up this organized screening program. It was then generalized in 2008 and 2009 to the entire population in France [8]. This organized screening program is designed for a population with a moderate risk aged 50—74 years [9]. Individuals are identified from health insurance files. The healthcare management organization invites them by post every 2 years to take the home screening test (Hemoccult® test). The test can be obtained from their family physician or by post [7]. If the test is positive, a colonoscopy is necessary. The participation rate of the target population is one of the fundamental parameters conditioning the organized screening program’s efficacy. The French program has been evaluated by the National Health Watch Institute (Institut de Veille Sanitaire [InVS]) based on data collected by the different management organizations. The assessments established in December 2009 and December 2010 report insufficient participation (37% in 2007—2008) and the need to set up priority actions to improve this indicator [10,11]. The objective of this study was to conduct a detailed geographic analysis of participation in Isère, a French administrative entity, so as to identify determinants of this participation and locate the areas in which specific actions could be adapted in relation to adherence to the program and the socioeconomic context.
Material and methods Study design We analyzed the original data from an ongoing population based colorectal cancer screening program in Isère, a French administrative entity with nearly 1.2 million inhabitants.
F. Poncet et al.
Data collection Data on the study population were obtained from the Office De Lutte Contre le Cancer (ODLC), which coordinates cancer screening in Isère. Information on individuals characteristics included sex, age, health insurance plan. The precise residential address, participation or non-participation in the organized screening program, as well as any causes for not taking the screening test were known for every person invited for screening. In France, there are different health insurance plans and the most common is the general plan. The agricultural plan covering agricultural occupations. The special plan concerns persons belonging to the SNCF (national railways), the RATP (Parisian transport), the electrical and gas companies (EDF and GDF) as well as some employees whose functions are directly related to the State such as the military, French National Police. . . The other coverage including civil service plan and schemes for the self-employed. Individuals who had taken the screening test during the period were defined as participants. In absence of individual socioeconomic data, the aggregate data at the ‘‘Ilôts Regroupés pour l’Information Statistique’’ (IRIS, or regrouped statistical information block) from the 2007 national population census of the French National Institute for Statistics and Economic Studies (INSEE) [12] provided the social environment of the subjects invited. The IRIS is the smallest geographical unit for which INSEE data are available. An IRIS corresponds to a small neighborhood, which is defined as a group of contiguous districts of approximately 2000 inhabitants. All towns with more than 10,000 inhabitants and most towns with 5000—10,000 inhabitants are broken down into several IRISes. The other towns form one IRIS. Isère department is broken down into 749 IRISes. Each person invited was assigning to an IRIS by geocoding the exact residential address. To define the socioeconomic level of an IRIS, the Townsend Index was used [13]. This index is used in the Anglo-Saxon countries to measure, in an aggregate fashion, the socioeconomic levels when individual data is missing [14]. This index was constructed based on four socioeconomic variables: the unemployment rate, the proportion of households without a car, the proportion of non-home owner households, and the proportion of overcrowded housing units (more than one person per room).
Study population
Statistical analyses
This study examined the target population for organized screening (50—74 years of age) invited for screening between January 2007 and December 2008 in Isère. The individuals whose address was incomplete or imprecise were excluded, as were those for whom medical contraindications were identified in accordance with the national program (presence of digestive symptoms, high risk of colorectal cancer based on coloscopic screening, another serious extra-intestinal disease, another particular contraindication, or a complete normal coloscopic exam less than 5 years before).
Given the hierarchical structure of the data, the statistical analysis method was based on a random-intercept two-level logistic regression [15,16]. Individual and aggregate data were used. The relevant variable was participation in the organized screening campaign: for each IRIS, the participation rate corresponds to the number of participants over the number of invited individuals. The first level was the individual level (sex, age, health insurance plan) compared to the invited individuals. The second level was the aggregate level corresponding to the
Participation in colorectal screening Table 1
195
Description of the target population in the organized colorectal cancer screening program in Isère (France, n = 247,776). Non-participation
Participation
n
%
n
%
Sex Male Female
74732 73678
62.1 57.8
45,667 53,699
Age 50—54 years 55—59 years 60—64 years 65—69 years 70—74 years
43875 35418 30407 18780 19930
63.4 61.2 58.3 54.3 58.5
107993 3418 18450 12282 6267 18319 21401 24430 42021 42239
Health insurance plan General plan Special plan Civil service plan Self-employed plan Agricultural plan Townsend Index Quintile 1 (least deprived IRISes) Quintile 2 Quintile 3 Quintile 4 Quintile 5 (most deprived IRISes)
Crude OR
95% CI
37.9 42.2
1 1.19
1.17—1.21
25,289 22,442 21,721 15,788 14,126
36.6 38.8 41.7 45.7 41.5
1 1.10 1.24 1.46 1.23
1.07—1.12 1.21—1.27 1.42—1.50 1.20—1.26
60.3 51.7 53.4 68.3 66.3
71,200 3190 16,082 5704 3190
39.7 48.3 46.6 31.7 33.7
1 1.42 1.32 0.70 0.77
1.35—1.49 1.29—1.35 0.68—0.73 0.74—0.81
56.8 58.1 58.4 59.8 63.4
13,960 15,428 17,370 28,196 24,412
43.2 41.9 41.6 40.2 36.6
1 0.95 0.93 0.88 0.76
0.91—0.98 0.91—0.96 0.86—0.90 0.74—0.78
socioeconomic level (Townsend Index) of an invited person’s residential area (IRIS). The Townsend Index was broken down by quintile. Quintile 1 represented the least deprived IRISes and quintile 5 the most deprived IRISes. Three two-level logistic regression models were successively estimated. The first model estimated was a model with no parameters (the empty model). Since it contains only random effects at the IRIS level, this empty model can detect the existence of a contextual effect on participation. The individual factors (characterizing the first level) were introduced into the second model and tested whether the variations between the IRISes may be explained by the differences in how the groups were composed in terms of individual characteristics. Finally, in the last model, the socioeconomic characteristics defined by the Townsend Index of each IRIS (second level) were added to the individual factors. This model can search to determine whether the differences in participation between the IRISes were explained by the characteristics of the IRISes. The analyses were performed with STATA 10.0 software.
Results The organized colorectal cancer screening invitations in Isère in 2007 and 2008 involved 281,987 people, including 26,037 who were secondarily excluded from the organized screening either for medical reasons (16,688) or because of an erroneous address and non-receipt of the invitation (9349).
A residential address that was not sufficiently precise that did not allow geocoding also led to excluding 2.7% of the individuals invited. In the end, the study included 247,776 individuals. The characteristics of the target population are described in Table 1. The crude participation rate was 40.1%, varying depending on sex, age, and health insurance plan. Women participated more than men, with an OR = 1.19 (95%CI [1.17—1.21]). The participants were on average slightly older (60.4 years) than the non-participants (59.7years) (P < 0.01) with a different participation depending on the age group. The 65- to 69-year-old group presented the highest participation rate. The 50- to 54-year-old age group presented the lowest participation rate. The beneficiaries of special plans participated the most (OR = 1.42; 95%CI [1.35—1.49]), whereas the beneficiaries of the agricultural healthcare plan participated the least (OR = 0.77; 95%CI [0.74—0.81]). The participation rate also varied in relation to the Townsend Index. The more deprived the IRIS, the lower participation was (Fig. 1). The participation rate in the least deprived IRISes (quintile 1 of the Townsend Index) was 43% and 37% in the most deprived IRISes (quintile 5 of the Townsend Index). The results of the multilevel analysis (Table 2) show that participation varied significantly between the residential IRISes (empty model + Fig. 2). The significant individual variables in univariate analysis were introduced into the multilevel analysis.
196
F. Poncet et al.
Table 2
Factors associated with participation in colorectal cancer screening in Isère (France) — multilevel analysis (n = 247,776). Empty model
Model 1 OR
Model 2 95%CI
OR
95%CI
Level 1: individuals Sex Male Female
1 1.16
1.14
1.18
1 1.16
1.14
1.18
Age 50—54 years 55—59 years 60—64 years 65—69 years 70—74 years
1 1.11 1.25 1.48 1.26
1.08 1.22 1.44 1.22
1.13 1.28 1.52 1.29
1 1.11 1.25 1.48 1.26
1.08 1.22 1.44 1.23
1.13 1.28 1.52 1.29
Health insurance plan General plan Special plan Civil service plan Self-employed plan Agricultural plan
1 1.43 1.29 0.72 0.73
1.36 1.26 0.70 0.69
1.51 1.32 0.75 0.76
1 1.43 1.29 0.72 0.71
1.36 1.26 0.70 0.68
1.5 1.32 0.74 0.75
1 0.96 0.95 0.90 0.75
0.91 0.90 0.85 0.71
1.02 1.00 0.95 0.79
Level 2: IRIS Townsend Index Quintile 1 (least deprived IRISes) Quintile 2 Quintile 3 Quintile 4 Quintile 5 (most deprived IRISes) Random effects Level 2 variance P-value
0.038 < 0.01
0.039 < 0.01
0.027 < 0.01
differences in participation between the IRISes. Introducing the Townsend Index to the model (Table 2, model 2) reduced the inter IRIS variance by 30.8%, signifying that socioeconomic status explained almost third of the disparities between IRISes. The people residing in the most deprived IRISes participated less than those living in the least deprived IRISes. After adjusting for individual variables, the multilevel analysis showed a 25% difference in participation between the least deprived IRISes (IRIS in the 1st quintile of the Townsend Index) and the most deprived IRISes (IRIS in the 5th quintile of the Townsend Index; OR = 0.75 [0.71—0.79]).
Discussion
Figure 1 Participation rate in the colorectal cancer screening program in Isère (France) in relation to the Townsend Index and sex (n = 247,776).
Between the empty model and model 1, in which the subjects’ characteristics were introduced (sex, age, and healthcare plan), the variance increased slightly, which indicates that the individual factors do not explain the
This study conducted on the entire target population invited for an organized colorectal cancer screening program between 2007 and 2008 in Isère showed that socioeconomic characteristics influence the participation in the organized colorectal screening program. The multivariate analysis demonstrated that there was less participation in the IRISes with higher levels of socioeconomic disadvantage. The role of unfavorable socioeconomic status as a factor of lower participation has been demonstrated in many studies but most of these studies used self-administered
Participation in colorectal screening
Figure 2 unit).
197
Crude participation rate in organized colorectal cancer screening program in Isère (n = 749 IRISes, French geographical
questionnaires and were therefore subject to participation bias. Age, sex, marital status, and educational level also had an impact on participation [17—22]. The present study used the same analysis strategy as a study conducted in the Calvados department (France), thus avoiding selection bias. In Calvados, the study investigated only a representative sample of the target population. The results of these two studies are in agreement and show that participation in organized screening is higher in the least deprived IRISes [23]. These results are also in agreement with two British studies, which also used aggregate data. However, these studies present methodological differences (multivariate analysis with no level 2 random effect). They differed in their method for inviting subjects to participate in screening. In England, the screening test is directly sent by mail to the target population, whereas in France, the target population is first invited to consult their family physician to obtain the screening test. The first study examined the results of the pilot organized screening program in England [24]. In the most disadvantaged areas, participation was lower than in higher socioeconomic level areas, with an OR of 0.68 (95%CI [0.59—0.79]). The second UK study examined the participation inequalities in the first campaign in the national colorectal cancer screening program in England [25]. It showed a linear association between participation and the deprivation index, participation in the least deprived areas was 50% higher than in the most deprived areas.
A Danish study [26] also analyzed the effect of several socioeconomic factors on participation. Based on individual data, it showed that a low socioeconomic level measured through educational level, employment, and income was strongly related to low participation. The present study presents some limitations. In absence of individual socioeconomic data, we chose to use aggregate data on the IRIS geographical unit from the INSEE census, which could result in a classification bias of the subjects [27,28]. The Townsend Index is considered to be identical for all people residing in the same IRIS, whereas large disparities may exist within an IRIS: consequently, these represent ‘‘mean’’ behaviors. The Townsend Index was used to characterize the socioeconomic level because no other valid, published French deprivation index exists based on the IRISes and covering the entire country. This index was drawn up in the context of the United Kingdom and probably only partially accounts for the French socioeconomic level. However, it has allowed us, most notably, to compare our results with the Calvados study. Some people targeted by the organized screening campaign were excluded from the analysis because their residential address was too imprecise for geocoding to the IRIS geographical unit. These exclusions for the most part concerned men between 50 and 54 years of age who did not participate in the screening program. Their inclusion in the study would probably have reinforced the results, since
198 men in the 50- to 54-year-old age group where those who participated the least. Among the non-participants, there were probably people who should not participate in organized screening because of a contraindication, but this information was declarative; the absence of a response by the person invited or by the physician in the healthcare management organization increased the number of non-participants. Conversely, some people underwent the screening test when they should not have (medical contraindications). Since 2007, the ODLC had used the databases of the medical data of healthcare facilities so as not to invite, for example, people who had undergone a colonoscopy in less than 5 years. Knowledge of these exclusions is important to optimally evaluate the participation rate [11]. The different studies on participation show that incentive actions are necessary, aimed at both the target population and practitioners. Involvement of family physicians can increase participation. The test is more often taken when given by the subject’s doctor than when sent directly to the subject [29,30]. The family physician plays an essential role in the screening program because he or she assesses what to do in relation to the level of risk. The consultation is the opportunity to give the subject the test or, if need be, to point out a medical contraindication to organized screening and orient the patient toward more adapted care. The physician’s role is also fundamental in explaining the procedure to the patient as well as the consequences of the result. He or she can answer questions and respond to worries associated with this procedure. A number of studies have underscored the importance of identifying how to optimize participation so as to have an impact on the ‘‘social’’ inequalities concerning colorectal cancer [31,32]: some authors emphasize the need for specific actions to reduce the differences in participation [33—35]. This study has provided a social and geographic identification of populations that participate the least using socioeconomic data at the IRIS unit, even if only ‘‘mean’’ behaviors are reflected. These results could be used to develop targeted incentive actions in the areas thus identified as low participators. Conducting exhaustive studies on the motivations for participation in individuals sharing a priori similar socioeconomic characteristics could contribute to identifying specific initiatives to improve the adherence of populations to this screening campaign.
Disclosure of interest The authors declare that they have no conflicts of interest concerning this article.
F. Poncet et al.
[2]
[3]
[4]
[5]
[6]
[7]
[8] [9]
[10]
[11]
[12]
[13] [14]
[15] [16]
Acknowledgments This study was funded by the Ligue Nationale Contre le Cancer — Comité départemental de l’Isère and by the Cancéropôle Lyon Auvergne Rhône-Alpes (CLARA).
[17]
[18]
References [19] [1] Hospices civils de Lyon/Institut de veille sanitaire/Institut national du cancer/Francim/Institut national de la santé et de la recherche médicale. Projections de l’incidence et de
la mortalité par cancer en France en 2011. Rapport technique; juin 2011.*Detailed results and comments [online] http://www.invs.sante.fr/surveillance/cancers [accessed 7th June 2011]. Kronborg O, Fenger C, Olsen J, Jørgensen OD, Søndergaard O. Randomised study of screening for colorectal cancer with faecal occult-blood test. Lancet 1996;348:1467—71. Hardcastle JD, Chamberlain JO, Robinson MH, Moss SM, Amar SS, Balfour TW, et al. Randomised controlled trial of faecal-occult-blood screening for colorectal cancer. Lancet 1996;348:1472—7. Faivre J, Dancourt V, Lejeune C, Tazi MA, Lamour J, Gerard D, et al. Reduction in colorectal cancer mortality by fecal occult blood screening in a French controlled study. Gastroenterology 2004;126:1674—80. Mandel JS, Bond JH, Church TR, Snover DC, Bradley GM, Schuman LM, et al. Reducing mortality from colorectal cancer by screening for fecal occult blood. Minnesota Colon Cancer Control Study. N Engl J Med 1993;328:1365—71 [Erratum in: N Engl J Med 1993;329:672]. Exbrayat C, Garnier A, Bolla M, Winckel P, Salicru B, Marron J, et al. Simultaneous screening of cancers of the breast, cervix, colon and rectum. The Isère experience. Bull Cancer 1996;83:641—8. Ministère chargé de la Santé et des Solidarités. Arrêté du 29 septembre 2006 relatif aux programmes de dépistage des ¸aise cancers, annexes. Journal Officiel de la République franc no 295 du 21 décembre 2006. Viguier J. L’organisation du dépistage du cancer colorectal en France. BEH 2009;2-3:19—21. Consensus conference: Prevention screening and management of the colonic cancers. Paris, France, January 29—30, 1998. Proceedings. Gastroenterol Clin Biol 1998;22:S1—295. Goulard H, Boussac-Zarebska M, Salines E, Bloch J. Évaluation épidémiologique du dépistage organisé du cancer du côlon et du rectum en France — Bilan du programme pilote : actualisation des données au premier novembre 2007. Saint-Maurice (Fra): Institut de veille sanitaire, avril 2009, 8 p. Available http://www.invs.sante.fr/publications/2009/plaq eval at: depistage cancer colon/index.html. Goulard H, Jezewski-Serra D, Duport N, Danzon A. Évaluation épidémiologique du dépistage organisé du cancer colorectal en France — résultats des programmes pilotes au-delà de la première campagne. Saint-Maurice: Institut de veille sanitaire; décembre 2010. 12 p. Available at: http://www.invs. sante.fr/publications/2010/plaquette colorectal/index.html. Recensement de la population 2007 : bases de données infra-communales — IRIS — 2007 [fichier électronique], Insee [producteur], centre Maurice Halbwachs (CMH) [diffuseur]. Townsend P. Deprivation. J Soc Pol 1987;16:125—46. Krieger N, Williams DR, Moss NE. Measuring social class in US public health research: concepts, methodologies, and guidelines. Annu Rev Public Health 1997;18:341—78. Diez-Roux AV. Multilevel analysis in public health research. Annu Rev Public Heath 2000;21:171—92. Chaix B, Chauvin P. The contribution of multilevel models in contextual analysis in the field of social epidemiology: a review of literature. Rev Epidemiol Sante Publique 2002;50:489—99. O’Malley AS, Forrest CB, Feng S, Mandelblatt J. Disparities despite coverage: gaps in colorectal cancer screening among medicare beneficiaries. Arch Intern Med 2005;165:212—35. Cokkinides VE, Chao A, Smith RA, Vernon SW, Thun MJ. Correlates of underutilization of colorectal cancer screening among U.S. adults, age 50 years and older. Prev Med 2003;36:85—91. Seeff LC, Nadel MR, Klabunde CN, Thompson T, Shapiro JA, Vernon SW, et al. Patterns and predictors of colorectal cancer test use in the adult U.S. population. Cancer 2004;100: 2093—103.
Participation in colorectal screening [20] Wee CC, McCarthy EP, Phillips RS. Factors associated with colon cancer screening: the role of patient factors and physician counseling. Prev Med 2005;41:23—9. [21] Weber MF, Banks E, Ward R, Sitas F. Population characteristics related to colorectal cancer testing in New South Wales, Australia: results from the 45 and Up Study cohort. J Med Screen 2008;15:137—42. [22] Goulard H, Boussac-Zarebska M, Duport N, et al. Facteurs d’adhésion au dépistage organisé du cancer colorectal: étude Fado-colorectal, France, décembre 2006—avril 2007. BEH 2009;2—3:25—9. [23] Pornet C, Dejardin O, Morlais F, Bouvier V, Launoy G. Socioeconomic determinants for compliance to colorectal cancer screening. A multilevel analysis. J Epidemiol Community Health 2010;64:318—24. [24] Weller D, Coleman D, Robertson R, Butler P, Melia J, Campbell C, et al. The UK colorectal cancer screening pilot: results of the second round of screening in England. Br J Cancer 2007;97:1601—5. [25] von Wagner C, Good A, Wright D, Rachet B, Obichere A, Bloom S, et al. Inequalities in colorectal screening participation in the first round of the national screening programme in England. Br J Cancer 2009;101(Suppl 2):S60—3. [26] Frederiksen BL, Jørgensen T, Brasso K, Holten I, Osler M. Socioeconomic position and participation in colorectal cancer screening. Br J Cancer 2010;103:1496—501. [27] Galobardes B, Shaw M, Lawlor DA, Lynch JW, Davey Smith G. Indicators of socioeconomic position (part 2). J Epidemiol Community Health 2006;60:95—101.
199 [28] Galobardes B, Lynch J, Smith GD. Measuring socioeconomic position in health research. Br Med Bull 2007;81—82:21—37. [29] Tazi MA, Faivre J, Lejeune C, Benhamiche AM, Dassonville F. Performance of the Hemoccult test in the screening of colorectal cancer and adenoma. Results of 5 screening campaigns in Saône-et-Loire. Gastroenterol Clin Biol 1999;23: 475—80. [30] Launoy G, Veret JL, Richir B, Reaud JM, Ollivier V, Valla A, et al. Involvement of general practitioners in mass-screening. Experience of a colorectal cancer mass screening program in the Calvados region (France). Eur J Cancer Prev 1993;2: 229—32. [31] Emmons K, Puleo E, McNeill LH, Bennett G, Chan S, Syngal S. Colorectal cancer screening awareness and intentions among low income, sociodemographically diverse adults under age 50. Cancer Causes Control 2008;19:1031—41. [32] Ioannou GN, Chapko MK, Dominitz JA. Predictors of colorectal cancer screening participation in the United States. Am J Gastroenterol 2003;98:2082—91. [33] James TM, Greiner KA, Ellerbeck EF, Feng C, Ahluwalia JS. Disparities in colorectal cancer screening: a guideline-based analysis of adherence. Ethn Dis 2006;16:228—33. [34] McGregor SE, Hilsden RJ, Li FX, Bryant HE, Murray A. Low uptake of colorectal cancer screening 3 yrs after release of national recommendations for screening. Amj Gastroenterol 2007;102:1727—35. [35] McGregor SE, Bryant HE. Predictors of colorectal cancer screening: a comparison of men and women. Can J Gastroenterol 2005;19:343—9.