European Journal of Cancer (2014) 50, 434– 446
Available at www.sciencedirect.com
ScienceDirect journal homepage: www.ejcancer.com
Review
Sorting out measures and definitions of screening participation to improve comparability: The example of colorectal cancer Jean-Luc Bulliard a,⇑, Montse Garcia b, Johannes Blom c, Carlo Senore d, Verna Mai e, Carrie Klabunde f a
Cancer Epidemiology Unit, University Institute of Social and Preventive Medicine, Lausanne, Switzerland Cancer Prevention and Control Program, IDIBELL, Catalan Institute of Oncology, L’Hospitalet de Llobregat, Spain c Department of Molecular Medicine and Surgery, Karolinska Institutet, Karolinska University Hospital, Stockholm, Sweden d AO Citta` della Salute e della Scienza, CPO Piemonte, Turin, Italy e Canadian Partnership Against Cancer, Toronto, Canada f National Cancer Institute, Bethesda, MD, USA b
Available online 18 October 2013
KEYWORDS Participation Definition Colorectal cancer Screening Programme
Abstract Participation is a key indicator of the potential effectiveness of any population-based intervention. Defining, measuring and reporting participation in cancer screening programmes has become more heterogeneous as the number and diversity of interventions have increased, and the purposes of this benchmarking parameter have broadened. This study, centred on colorectal cancer, addresses current issues that affect the increasingly complex task of comparing screening participation across settings. Reports from programmes with a defined target population and active invitation scheme, published between 2005 and 2012, were reviewed. Differences in defining and measuring participation were identified and quantified, and participation indicators were grouped by aims of measure and temporal dimensions. We found that consistent terminology, clear and complete reporting of participation definition and systematic documentation of coverage by invitation were lacking. Further, adherence to definitions proposed in the 2010 European Guidelines for Quality Assurance in Colorectal Cancer Screening was suboptimal. Ineligible individuals represented 1% to 15% of invitations, and variable criteria for ineligibility yielded differences in participation estimates that could obscure the interpretation of colorectal cancer screening participation internationally. Excluding ineligible individuals from the reference population enhances comparability of participation measures. Standardised measures of cumulative participation to compare screening protocols with different intervals and inclusion of time since invitation in definitions are urgently needed to improve international comparability of colorectal cancer screening participation. Recommendations to improve comparability of participation indicators in cancer screening interventions are made. Ó 2013 Elsevier Ltd. All rights reserved.
⇑ Corresponding author: Address: Unite´ d’e´pide´miologie du cancer, Institut universitaire de me´decine sociale et pre´ventive, Biopo ˆ le 2, Rte de la Corniche 10, 1010 Lausanne, Switzerland. Fax: +41 21 314 7373. E-mail address:
[email protected] (J.-L. Bulliard).
0959-8049/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ejca.2013.09.015
J.-L. Bulliard et al. / European Journal of Cancer 50 (2014) 434–446
1. Introduction Participation is a key indicator of the potential effectiveness of any screening intervention since it reflects the degree to which a population is exposed to the intervention. For cancer screening programmes, minimal thresholds of participation are recommended to ensure the intended benefits of the intervention can be achieved [1–3]. Monitoring participation is paramount to screening programmes and other groups, including cancer control agencies and public health authorities, as a benchmarking parameter for comparing screening performance across jurisdictions. Over the past decades, more countries have implemented cancer screening programmes in increasingly diverse settings. For instance, the prevalence of opportunistic screening (screening performed outside the setting of an organised programme) varies considerably across countries and influences participation since recent opportunistic testing usually renders an individual ineligible for the screening programme [3]. The growing number of programmes and factors affecting the analysis and interpretation of participation rates heighten the need to be more precise and comprehensive in approaches taken to monitor screening participation [4,5]. The International Cancer Screening Network (ICSN) is a consortium encompassing 33 countries with population-based cancer screening programmes, sponsored by the U.S. National Cancer Institute for the purpose of fostering international efforts to evaluate the quality and effectiveness of screening programmes [6,7]. Recent ICSN projects have determined how performance parameters could most suitably be compared for breast cancer [8– 10]. With the lessons learned from well-established breast and cervix screening programmes and the emerging need to monitor performance of more recently established colorectal cancer (CRC) screening programmes, it is an opportune time to consider a common set of indicators that can be used to provide an accurate picture of participation rates in CRC screening internationally. The objectives of this paper are to examine literature to (1) highlight the issues related to measuring, reporting and comparing CRC screening participation across programmes and countries, (2) identify definitions of participation that have been used and (3) assess the impact of varying definitions on estimates of participation. Although the focus is on colorectal, most of the issues addressed are relevant to other population-based cancer screening programmes. 2. Materials and methods To identify definitions and reports of CRC screening participation, a multistep search strategy was adopted. A search of articles was first conducted on Medline
435
(National Library of Medicine, Bethesda, United States of America (USA)) using the PubMed query interface, and included Medical Subject Headings (MeSH) corresponding to ‘colon cancer’, ‘colorectal cancer’ or ‘bowel cancer’ and ‘screening participation’, ‘screening uptake’, ‘screening compliance’ or ‘screening adherence’. The search, restricted to publications in English (at least the abstract) between 1st January 2005 and 30th June 2012, identified 1497 potentially relevant papers. The retrieved articles were reviewed independently by two investigators (J.L.B. and M.G.) and their references were manually checked. We also searched the Web for monitoring and evaluation reports and guidelines addressing eligibility criteria and definition of participation in order to include important unpublished material (J.L.B.). This study focused on screening interventions with a defined target population (i.e. designated age groups and geographic catchment areas) and a systematic invitation scheme, because a precise assessment of participation rate is otherwise problematic. For example, cross-sectional studies of CRC screening use cannot assess medical eligibility, are often based on self-report of attendance and render difficult the distinction between screening and diagnostic tests; screening recruitment through GP visits does not allow accurate estimation of the population fraction exposed to the intervention [11–14]; and interventions targeting selected sub-populations such as health plan members [15,16], volunteers [17], veterans [18] or reporting on two-stage recruitment strategy, where non-compliers or screennegative individuals are subsequently offered an alternative screening test [19–21], limit international comparability. Where several publications were available for a country or a programme, the most recent or complete one, with respect to participation indicators, was used. For countries with national screening programme delivered at a regional level, nationwide reports were given greater weight than region- or centre-based reports (i.e. Italy, Spain and United Kingdom (UK)), but additional information from complementary sources was included. National participation estimates were computed from unpooled regional data when necessary [22]. Pilot and trial studies were reported separately as their particular settings generally favour higher participation rates. The definitions of coverage and participation proposed in the first edition of the European Guidelines (EG) for Quality Assurance in CRC Screening were used [3]. The coverage rate by invitation (C: % screeneligible subjects that are invited in the target population during a given time frame) reflects a programme’s ability to invite its target population. Although the participation rate (P: % invited people during a time frame that were screened/tested during this time frame) is independent of the coverage rate, both indicators should be assessed concomitantly since they contribute to
436
J.-L. Bulliard et al. / European Journal of Cancer 50 (2014) 434–446
determining the fraction of the entire eligible population that is screened (P C). In other words: (1) at similar levels of participation, the fraction of the target population screened can differ between programmes because of different coverage rates, and (2) a programme characterised by higher participation but lower coverage (P1: 60%, C1: 80%, P1 C1 = 48%) can screen a smaller fraction of its target population than a programme with lower participation but higher coverage by invitation (P2: 50%, C2: 98%, P2 C2 = 49%). Bearing that in mind, this study focused on the impact of using different definitions of participation, regardless of the coverage rate. Grounds for exclusion were investigated and grouped into administrative (moved out of target area, deceased, unreachable, under guardianship, signed consent form not returned) and medical criteria (personal or family history, comorbidity, recent bowel examination or other), whether they were applied before or after invitation, or were permanent or temporary. Symptoms were seldom mentioned among exclusion criteria. We assumed that, in other cases, screen-eligible subjects were considered asymptomatic, regardless of their invitation. As far as possible, each exclusion criterion was quantified and expressed as a proportion of the invited population. Data extracted from each study and quantification of exclusion criteria were independently validated by two investigators (J.L.B. and J.B.). Nonrespondents of unknown eligibility status were considered eligible. International comparisons were based on first-round participation as this was the most frequently available indicator and the most consistent for this purpose. From the articles and monitoring/evaluation reports retrieved, we also identified all terms used to describe participation, their respective definitions, as well as the dimension of participation aimed at by these various measures. 3. Results Programmes with a systematic invitation scheme for their target population at the end of the selection process originated from Europe and the Western Pacific region (Table 1). The screening interval used for faecal occult blood test (FOBT) and faecal immunological test (FIT) was 2 years, bar Asian programmes and a Swiss pilot project that used an annual interval. With the exception of Japan (40 years) and Finland and England (60 years), FOBT/FIT-based national programmes invited subjects from age 50. Endoscopic procedure (flexible sigmoidoscopy or colonoscopy) and computed tomography (CT) colonography were offered as a primary screening tool in some European countries, most often within a pilot trial or concomitantly with a noninvasive test, and generally on a once-only basis. Seven different terms were used to describe participation
(Table 1); more than 10 when including specific dimensions of participation (see Table 4). Criteria used for measuring first-round participation and participation rates for organised CRC screening programmes and pilot trials are reported in Tables 2 and 3, respectively. About one third of programmes using faecal tests counted only individuals with valid returned test kits as participants. Three programmes restricted participation to those attending within a few months after invitation [23–25], and one used a statistical projection to provide a prompt participation rate that accounts for the time lag between invitation and attendance [26,27]. Criteria used to define screen-eligible people varied widely across programmes, and more so for medical than administrative criteria. Whereas quantification of ineligibility was poorly documented in organised programmes, pilot studies showed that ineligible subjects could represent up to about 15% of all invitations [28]. Besides those who had moved out of the target area, were deceased or unreachable (invitations undelivered), some interventions also excluded institutionalised people and subjects who did not sign a consent form. All ineligible people could be identified when the context permitted checking screening eligibility prior to the invitation. However, this information was inconsistently used across studies (number of invited versus number of invited and eligible people as denominator for participation rate). The coverage rate, when reported, was generally high; lower coverage rates reflected implementation of new programmes and difficulty with systematic dissemination of invitations (i.e. Italy [29]). First-round participation in organised programmes ranged from 34% to 82% with stool-based tests, and was 24% with flexible sigmoidoscopy (Table 2). The participation rate with structural screening examination (flexible sigmoidoscopy, colonoscopy or CT colonography) was generally higher in pilot studies than in organised programmes (Tables 2 and 3). All trials comparing FOBT and FIT yielded higher participation with FIT. Fig. 1 shows the impact on participation of case exclusion from the target population, assuming ineligible subjects were not counted as screenees. Withdrawing ineligible subjects from the denominator restricts the target population, hence increasing the participation rate. The higher the participation, the greater is the impact of excluding a given fraction of ineligible cases. For instance, if 10% of the target population were deemed ineligible, then a participation rate of 45% results in a corrected participation rate of 50%, a 5% increase in absolute terms. To reach the same absolute effect with participation rates of 35% (35 to 40%) and 25% (25 to 30%), one would need to have 12% and 18% of subjects excluded as ineligible, respectively. Use of different definitions of, and terms for, participation can reflect different aims and interpretations of
Table 1 Characteristics of selected organised colorectal cancer (CRC) screening programmes by country within regions defined by the World Health Organisation. Country
Western Pacific Australia Japan South Korea Taiwan
Finland France Ireland Italy
Programme type
Primary screening test – screening interval
Age range**
Term(s) used for participation
Programme sources
Website
All Rural Queensland All All Keelung county
OrgP Pilot OrgP OrgP OrgP
FIT-2 FIT-2/FOBT-2* FIT-1 FIT-1 FIT-1
50, 55, 65 50–74 40+ 50+ 50–79
Participation Participation Participation Participation Compliance
[26,27] [48,49] [50] [51,52] [53]
www.aihw.gov.au/bowel-cancer-screening/
Flanders (three areas) All Vejle and Copenhagen counties All
Pilot Pilot Pilot
FIT FOBT-2 FOBT
50–74 71,72,74,75 50–74
Participation Response Participation
[54,55] [30] [56]
FOBT-2
60–67
Uptake
[31]
23 districts Haut-Rhin Dublin area 89 reg prog
OrgP (RCT) Pilot Pilot Pilot OrgP
FOBT-2 FOBT-2 FIT FIT-2
50–74 50–74 50–74 50–69
[57] [58] [59] [29]
Nine reg prog
OrgP
FS-once
58–60
14 centres throughout Italy
Pilot (RCT)
FOBT/CO
55–64
Six centres
Pilot (RCT) Pilot (RCT) OrgP
FIT-2/FS-once/COonce FS-once/FIT+FSonce FIT-2/FOBT-2*
55–64
Participation Participation Participation Compliance/ attendance Compliance/ attendance Attendance/ participation/ compliance Attendance
50–64
Attendance
[61,62]
50–69
Participation/ rescreening
[22,34]
Pilot (RCT) Pilot
FS
59–61
Participation
[63]
FOBT-1/FS-5/ FOBT-1 + FS-5/ CO-once FOBT/FIT
50–80
Participation/ recruitment
[32]
50–75
Participation
[24]
FOBT/FIT/FS
50–74
Participation
[64]
CO/CT
50–75
Participation
[65]
Norway
Oslo and Telemark county
Spain
Catalonia, Valencia, Murcia and Basque Country
Sweden
Uppsala and Lund areas
Switzerland
Uri and Glarus cantons, Valle´e de Joux
The Netherlands
Amsterdam, Nijmegen and surroundings Greater Rotterdam area Amsterdam and Rotterdam
Pilot (RCT) Pilot (RCT) Pilot (RCT)
www.invs.sante.fr
www.osservatorionazionalescreening.it
[29] [28]
[60]
www.cribadocancer.es www.murciasalud.es/recursos/ficheros/142015Ingles_Recommendations_National_Panel_Experts.pdf
J.-L. Bulliard et al. / European Journal of Cancer 50 (2014) 434–446
Europe Belgium Croatia Denmark
Region(s)
(continued on next page) 437
OrgP: organised programme; RCT: randomised controlled trial; FOBT: (guaiac) faecal occult blood test; FIT: faecal immunochemical test; FS: flexible sigmoidoscopy; CO: colonoscopy; CT: CT colonography. * FIT only in subsequent rounds. ** Age at initial round for which screening participation was evaluated (not necessary the whole age range targeted by a programme).
Pilot Grampian, Tayside and Fife
UK-Scotland
FOBT-2
50–69
Uptake
[40,68]
www.isdscotland.org/Health-Topics/Cancer/ Publications/2011-08-30/KPI_report.pdf [36] [67] [23] Uptake Uptake Uptake 50–69 58–59 50–74 Pilot Pilot OrgP West Midlands London boroughs All
FOBT-2 FS-once FOBT-2/FIT-2
www.cancerscreening.nhs.uk/bowel/pilot-2nd-roundevaluation.pdf [25,66] Uptake 60–69 All UK-England
FOBT-2 OrgP
Region(s) Country
Table 1 (continued)
Programme type
Primary screening test – screening interval
Age range**
Term(s) used for participation
Website
J.-L. Bulliard et al. / European Journal of Cancer 50 (2014) 434–446
Programme sources
438
this indicator (Table 4). The participation rate has been used as an early indicator of a programme’s impact, a process indicator reflecting organisational and communication-related features of a programme, a measure of the efficiency of the screening process or an indicator of long-term adherence and satisfaction with a programme. Beyond differences in definition, participation indicators, particularly when assessing repeated participation, can vary considerably depending on the dimension captured. For instance, first-time participation between 2000 and 2007 in Scotland was 13.1% (% non-attendees from previous rounds who participated in the current (third) round) while the rescreening rate (% screenees who participated in at least one (of the two) previous rounds) was 83% [40]. In a regional programme in Italy, round-specific participation rates ranged between 56% and 63%, whereas longitudinal adherence (% subjects participating in all four biennial rounds) was 38% [39].
4. Discussion Comparing screening participation across settings has become increasingly complex. Our study documented substantial variability in measuring, defining and reporting participation across CRC screening programmes worldwide. The magnitude of these differences could impact sizeably on participation estimates and distort underlying, real variations across programmes. Some contributing factors are the repetition of tests and multiplicity of modalities, the need to concomitantly monitor recently-implemented and older programmes alike, and the mix of programmatic and non-programmatic delivery systems, with often uneven quality control and monitoring. Accounting for contextual differences when comparing screening performance indicators internationally has proven challenging [8–10]. A previous effort to standardise CRC screening indicators has left out participation measures owing to difficulty in gathering consistent and complete information [41]. We addressed this comparability issue by starting with the operational definitions provided in the EG [3] and by restricting our analysis to pilot studies and programmes that employed active invitation of a defined target population. Applying similarly defined indicators should enable genuine differences in participation estimates across programmes to be distinguished from artefactual ones. Participation (or uptake) rate as defined in the EG [3] (number of invited people screened within a given time frame as a proportion of eligible people invited during this time frame) implicitly assumes knowledge of eligibility status for each subject. This assumption does not hold for programmes that cannot ascertain eligibility before sending out invitations. In addition, eligibility can also be assessed after the invitation, with some programmes
Table 2 Criteria used for measuring participation across organised colorectal cancer (CRC) screening programmes and their quantification. Country, time frame
Numerator
All screenees
Finland, 2004–7
Valid stool test
Italy, 2009
All screenees
Japan, 2004 South Korea, 2008 Spain, NS
All screenees Valid stool test Valid stool test
Taiwan, 2000–2 UK-England, October 2006–Janaury 2009 UK-Scotland, November 2008–October 2010
All screenees 60
Yes
CO < 5 FS < 5
4.2
Yes
Yes
GP patient
Yes
%
Participation (first round) %
40***
3.2
Yes GP patient
Coverage
100
70
40
43+
100
18++ 21++ 34
96
82 54 54
FIT < 2 FS < 5
Rel: first-degree relative; FIT: faecal immunochemical test; FS: flexible sigmoidoscopy; CO: colonoscopy. * expressed in % of invited. ** Personal history of CRC, polyps, IBD or intestinal surgery. *** Estimated 2008 participation (up to January 2010) for first round based on Kaplan–Meier method to account for lag time between invitation and attendance. + Annual (2009) participation was 50%. ++ First and subsequent rounds combined.
24
J.-L. Bulliard et al. / European Journal of Cancer 50 (2014) 434–446
FOBT/FIT Australia, 2008
Denominator
439
Country, time frame
Numerator
Denominator
Eligibility check
Admin. ineligibility %*
Medical ineligibility (%*)
Source
Invited
Post-inv
Invited & eligible
Post-inv
All screenees All screenees
Invited
None
Invited & eligible
Pre-inv
National Pathology Registry
Yes
All screenees
Invited & eligible
Post-inv
Patient, GP
Yes
Valid stool test
Invited & eligible
Post-inv
patient, GP
3.6
Valid stool test All screenees
Invited
NS
Invited & eligible
Italy, October 2002– January 2004
All screenees
Invited & eligible
Norway, January 1999– 2001 Sweden, 1996
All screenees
Invited & eligible
Pre-inv: med crit Post-inv: admin crit Pre-inv: med crit Post-inv: admin crit NS
All screenees All screenees
Invited
Pre-inv
Invited & eligible
Post-inv
Belgium, March– November 2009 Croatia, 2008– March 2009 Denmark, August 2005– December 2006 France, 2 years*** France, September 2003– September 2006 Ireland, 2008–2010 Italy, November 2003–November 2006
Switzerland, June 2000– June 2001
All screenees All screenees
Pers. hist**
Fam. hist
Patient, GP
Yes
Yes
Patient, GP
Yes
Comorbidity
Recent exam
Other
Ineligibility, total %*
Coverage % 100
Yes
3.7 CO < 10
5.4
Participation (first round) % FIT: 39 FOBT: 30 45+
21++ CO < 2
1 rel < 65, 2 rel > 65 1 rel < 60, 2 rel > 60
Yes (if reversible)
CO < 5 FIT < 2
0.5%
6.8 CO < 5 FOBT < 2
6.4
Refusals
9.0
100
42
10.5
98
55
99
51
100
FOBT: 27 CO: 10
Patient, GP
3.0
0.6
0.6 2 rel
6.3
4.8
15.3
Patient, GP
1.8
1.1
2 rel
2.7
4.2 CO < 2 FS < 2 FOBT < 2
>8
Patient, GP, relatives
1.7
Yes
Yes^
Yes
0.7 Patient
Yes
CO < 5 FS < 5
48
FIT: 32 FS: 32 CO: 27
3.7
100
FS: 67 FIT + FS: 63
0.7
100
39
1.3
12
J.-L. Bulliard et al. / European Journal of Cancer 50 (2014) 434–446
Pre/ post-inv. Australia, NS
440
Table 3 Criteria used for measuring participation across colorectal cancer (CRC) screening pilot studies and their quantification.