SEARCH

SEARCH BY CITATION

Keywords:

  • colonoscopy;
  • colorectal cancer screening;
  • fecal occult blood test;
  • flexible sigmoidoscopy;
  • screening evaluation

Abstract

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. References
  7. APPENDIX – INTERNATIONAL COLORECTAL CANCER SCREENING NETWORK
  8. Supporting Information

The International Colorectal Cancer Screening Network was established in 2003 to promote best practice in the delivery of organized colorectal cancer screening programs. To facilitate evaluation of such programs, we defined a set of universally applicable colorectal cancer screening measures and indicators. To test the feasibility of data collection, we requested data on these variables and basic program characteristics from 26 organized full programs and 9 pilot programs in 24 countries. The size of the target population for each program varied considerably from a few thousand to 36 million. The majority of programs used fecal occult blood tests for primary screening, with more using guaiac than immunochemical tests. There was wide variation in the ability of screening programs to report the requested measures and in the values reported. In general, pilot programs were more likely to provide screening measure values than were full programs. As expected, detection rates for polyps and neoplasia were substantially higher in programs screening with endoscopy than in those using fecal occult blood tests. It is hoped that the screening measures and indicators, once revised in the light of this survey, will be adopted and used by existing programs and those in the early planning stages, allowing international comparison with the goal of improved colorectal cancer screening quality.

The International Colorectal Cancer Screening Network (ICRCSN) is a global consortium public health professionals who are focussed on planning and delivering colorectal cancer (CRC) screening to their populations. The ICRCSN is sponsored by the Centers for Disease Control and Prevention and the American Cancer Society, Atlanta, GA. The ICRCSN aims to facilitate the international sharing and comparison of CRC screening data to benefit both existing programs and those in the planning stages. During Phase 1 (2003–2005) organized screening initiatives were identified worldwide and their characteristics documented.1 It became clear that if CRC screening programs were to be usefully compared, a minimum set of universally applicable CRC screening indicators would be required.

The aim of Phase 2 of the ICRCSN was to establish a minimum set of universally applicable CRC screening measures and indicators to evaluate and compare screening programs on an international level, while contributing to program evaluation at a local level; and to investigate the feasibility of data collection for these screening measures. In this paper, we define these measures and indicators and report data collected in 2008 from 35 screening programs.

Material and Methods

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. References
  7. APPENDIX – INTERNATIONAL COLORECTAL CANCER SCREENING NETWORK
  8. Supporting Information

In Phase 1, we surveyed CRC screening initiatives with varying levels of organization, including full programs, pilot programs, and research projects. In Phase 2, we restricted our survey to full or pilot programs fulfilling at least four of the International Agency for Research on Cancer (IARC) criteria for an organized screening program. These include (i) an explicit policy with specified eligibility categories, method and interval for screening; (ii) a defined target population; (iii) a management team responsible for implementation; (iv) a health care team for decisions and care; (v) a quality assurance structure and (vi) a method for identifying cancer occurrence in the target population.2

The Expert Working Group for Phase 2, consisting primarily of public health researchers and those responsible for developing and running screening programs, met in September 2007 and drafted an initial set of screening indicators based on those developed for breast cancer screening by such agencies as the IARC,2, 3 the International Breast Screening Network (http://appliedresearch.cancer.gov/icsn), and the UK NHS Breast Screening Programme (http://www.cancerscreening.nhs.uk/breastscreen). Each indicator is a rate calculated as a ratio of two conventional “screening measures” (see Table 1 for a list of all screening measures and their definitions, and Table 2 for a list of screening indicators). For example, the indicator “fecal occult blood test (FOBt) positivity rate” is calculated from the two screening measures “FOBt positive population” and “tested population” (FOBt positivity rate = number of people with positive FOBts/number in the tested population).

Table 1. Screening measures and their definitions, as used in the survey
inline image
Table 2. Screening indicators, as defined for the survey
inline image

Our aim was to define each screening measure, so that it would be applicable to and easily understood in all health care contexts in all countries. The screening measures and indicators were piloted (using a written questionnaire) by seven screening programs representing a broad spectrum of different health systems and different modalities, including FOBt, flexible sigmoidoscopy (FS) and total colonoscopy (TC). Subsequently, a consensus on the definition of each screening measure was reached by the Expert Working Group, and the indicator questionnaire was refined.

A representative from each eligible program was emailed the questionnaire in May 2008 and invited to participate in the Phase 2 survey. In addition to the screening measures, we collected information on program characteristics. Programs were asked to provide their most recent data for a minimum period of 12 months. Separate questionnaires were completed for each screening modality within a program. Of the 11 screening measures, eight were applicable to all modalities, one only to either FS or FOBt, and two only to FOBt.

For each modality, programs were asked to provide a value for each screening measure for first (prevalent) screens, subsequent (incident) screens and total screens (the sum of first and subsequent screens) within the reporting period for their screening program. Programs (or modalities within a program) were subsequently classified as those in their first screening round, and therefore only able to report first screens; those that had undergone more than one screening round and were able to report data on first and subsequent screens and those unable to distinguish between first and subsequent screens. Programs (or modalities within a program) for which only total screens could be reported were excluded from screening indicator calculations, as were those with a tested population <100 people. Unclear responses to the questionnaire were clarified by subsequent communication with the program representatives.

Preliminary results from the survey were discussed at a meeting for survey respondents and the Expert Working Group in Oxford, England in September 2008, and proceedings of this meeting were published elsewhere.4

Results

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. References
  7. APPENDIX – INTERNATIONAL COLORECTAL CANCER SCREENING NETWORK
  8. Supporting Information

In Phase 2, 43 organized screening programs were identified. Of these, eight programs had started screening too recently to be able to provide data for the survey: four in Canada and one in each of China (Hong Kong), Lithuania, Slovenia and Sweden. A total of 35 programs had been collecting data for at least 12 months and were thus eligible for survey. Of the 35 eligible programs, from 24 different countries, 26 were full programs and 9 were pilot programs (Tables 3 and 4). Those programs with published references or websites describing their program details are listed on the ICRCSN website (Supporting Information, E-tables 1a and 1b: http://icrcsn.ceu.ox.ac.uk/).

Table 3. Characteristics of screening programs using fecal occult blood tests as the primary screening modality, presented alphabetically by country within regions defined by the World Health Organization
inline image
Table 4. Characteristics of screening programs using endoscopy as the primary screening modality, presented alphabetically by country within regions defined by the World Health Organization
inline image

Four programs used different screening modalities in separate target groups (Tables 3 and 4): New York used both guaiac FOBt (gFOBt) and immunochemical FOBt (iFOBt), Maryland used gFOBt and TC, the Dutch program screening the regions of Amsterdam and Nijmegen used both gFOBt and iFOBt, and the Dutch program screening the Greater area of Rotterdam used gFOBt, iFOBt and FS. Thus, a total of 40 modality-specific survey questionnaires were completed. All results are presented separately by modality within programs, as appropriate.

Characteristics of surveyed screening programs

The size of the target population for each program (or modalities within programs) varied considerably, from a few thousand for the three modalities within the Dutch pilot program screening the Greater area of Rotterdam (gFOBt, iFOBt and FS) and the TC program in the USA (Delaware) to 36 million in the Japanese program. The earliest program began in 1971 in Germany (FOBt program) and the most recent in Croatia in 2008. As shown in Table 3, 28 of the programs used FOBt as their primary screening modality. Of these, 16 used guaiac tests, 9 used immunochemical tests and 3 used both kinds of tests. Most programs (20 [69%]) developed their FOBts in a central laboratory, four [14%] developed tests in the primary care practitioner's office and four (14%) used both sites. Most programs using gFOBt collected six stool samples (two samples from three consecutive bowel movements), whereas programs using iFOBt collected only one or two stool samples (one sample per bowel movement). Regardless of the number of samples taken, most programs defined a test as positive when any one of the samples was considered positive. However, England and Spain defined a test as positive when at least five of the six samples were positive on first tests, or for borderline tests (1–4 samples positive), on repeat testing (any one of 12 samples positive on two further tests). Scotland defined a test as positive when at least five of the six samples taken were positive, or when 1–4 of the six samples and a subsequent iFOBt were positive. Croatia and Israel used a sensitive gFOBt, and three programs using gFOBt in the USA (Maryland, New York and New Jersey) used a combination of standard and sensitive tests on their populations.

The age group invited to screening is influenced by national guidelines, but in some nations also by the test, commitment to on-going versus one-time screening, and the screening interval (Tables 3 and 4). For example, in all USA programs included in the survey, screening started at age 50 with criteria for not screening defined by the individual's health status rather than age, and the screening interval varied by the test that was used. In Europe, screening typically started at either age 50 or 60, with upper age limits typically between ages 66 and 75. Upper and lower age limits in the Western Pacific were more variable.

Screening interval varied by screening modality. The two programs using FS screened “once only”, whereas the seven programs using TC screened every 10 years (Table 4). For programs using FOBt activities (Table 3), the interval was “once only” for seven programs (six of which were pilot programs), annually for 10, biennially for 12, a combination of both annually and biennially specified by age for the German program, and every 51 months for the Slovak Republic program.

Screening measures

Table 5 shows the number of programs (or modalities within programs) able to give values for each screening measure, separately for those that were able to provide a value for first screens only (no subsequent screens having yet been done), those that were able to provide values for both first and subsequent screens, and those that were able to provide a value only for total screens (including first and subsequent screens). Table 6 shows values for reported screening measures, by program (or modalities within a program).

Table 5. Number of screening programs (or modalities within a program) able to provide values for each screening measure: only for first screens, separately for first and subsequent screens, and only for total screens
inline image
Table 6. Screening measures reported for each program (or modalities within a program), presented alphabetically by country within regions defined by the World Health Organization
inline image

Reporting of the measures “target population” and “invited population” was inconsistent and incomplete so these measures are not included in Table 6 (or used for indicator calculations, all of which take the tested population as the baseline). All 35 programs were able to report the size of their tested population and almost all (33 [94%]) were able to report number of CRC detected. Around half of the programs were able to provide data on 30-day mortality related to an endoscopic procedure, and 13 programs stated that all or part of their data on screening measures were estimated (with the majority of these programs estimating their target population). Only three programs were able to provide a value for all applicable screening measures (Chile, Hungary and the Dutch program using three modalities); a further five programs could provide values for all except for the mortality measures (Australia, Finland, Romania, Spain and USA [New York] FOBt). Ability to report screening measures did not appear to vary substantially by geographical region (Europe, The Americas and Western Pacific) or by modality (FOBt and endoscopy). Pilot programs were more likely than full programs to provide values for any screening measure.

Screening indicators

Tables 7 and 8 show the screening indicator values calculated from data reported for first (n = 30 programs or modalities within programs) and subsequent (n = 10 programs or modalities within programs) screens, respectively. Excluded from these tables are the eight programs that did not distinguish between first and subsequent screens (Czech Republic, Germany [FOBt], Slovakia, Japan, Uruguay, Scotland, Singapore and USA [New Jersey]), and two very small programs (both from USA [Missouri]). Indicator rates (such as participation rate) which depend on the target or invited populations are not given, because these measures were inconsistently reported. Mortality rates are also not reported, because only one program (USA – Colorado) reported a value greater than 0 cases (two cases per 40,000 screened).

Table 7. Screening indicators for first (prevalent) screens, presented alphabetically by country within regions defined by the World Health Organization: includes programs able to provide data only for first screens, and those able to distinguish between first and subsequent screens
inline image
Table 8. Screening indicators for subsequent (incident) screens, presented alphabetically by country within regions defined by the World Health Organization, in programs able to provide data
inline image

Detection rates for CRC at the reported screening round were substantially higher for programs using endoscopy as their primary screening method compared with those using FOBt. For first screens, the mean cancer detection rate per 1,000 people tested by FOBt was 2.4 (standard deviation [SD] 1.6); range = 0–7.1), whereas for FS it was 4.9 (SD = 0.5; range = 4.6–5.3), and for TC it was 9.1 (SD = 1.2; range = 7.4–10.7). Results did not vary by type of FOBt (gFOBt: 2.7 [SD 1.6]; iFOBt: 2.2 [SD 1.9] per 1,000 people tested). Polyp and adenoma detection rates were slightly higher for programs using iFOBt than those using gFOBt: for polyp detection, mean rates were 23.6 (SD 16.2) and 11.2 (SD 4.5) per 1,000 people tested using iFOBt and gFOBt, respectively, and for adenoma detection, mean rates were 17.1 (SD 14.1) and 9.1 (SD 3.3) per 1,000 people tested using iFOBt and gFOBt, respectively.

Programs using iFOBt tended to report higher test positivity rates at first screens than those using gFOBts (iFOBt: 6.1% [SD 1.7]; gFOBt: 4.6% [SD 3.3]). Within gFOBts, mean test positivity rate for programs using sensitive gFOBts only was 6.2% [SD 3.6], whereas for programs using standard gFOBts was 4.1% [SD 3.2] (with only one program having a positivity rate greater than 2.7%). Once modality was taken into account, there were no obvious differences in indicator values by region or between full programs and pilot programs, although these analyses were limited by the small number of programs evaluated.

Positive predictive values (PPVs) for polyp, adenoma and cancer detection for FOBt screening are shown in Table 9 for first screens and Table 10 for subsequent screens. For first screens, PPVs of a positive FOBt (in those undergoing subsequent diagnostic endoscopy) ranged from 14% (Romania) to 69% (The Netherlands – Amsterdam and Nijmegen, iFOBt) for polyp detection, 9.5% (Australia) to 62% (The Netherlands – Greater area of Rotterdam, iFOBt) for adenoma detection, and 0% (New York FOBt) to 16% (Spain) for cancer detection.

Table 9. Positive predictive values for colorectal neoplasia: programs using fecal occult blood tests as their primary screening modality that provided data for first screens, presented alphabetically by country within regions defined by the World Health Organization
inline image
Table 10. Positive predictive values for detection of colorectal neoplasia: programs using fecal occult blood tests as their primary screening modality that provided data for subsequent screens, presented alphabetically by country within regions defined by the World Health Organization
inline image

Discussion

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. References
  7. APPENDIX – INTERNATIONAL COLORECTAL CANCER SCREENING NETWORK
  8. Supporting Information

We identified 43 organized CRC screening programs during Phase 2 of the ICRCSN, of which 35 were eligible to participate in the survey. The eligible programs represented 24 countries and consisted of 26 full programs and nine pilot programs. As in Phase 1, the majority of programs conducted primary screening using FOBt with slightly more using guaiac than immunochemical tests.

Of the eligible programs, 26 provided sufficient data to allow calculation of most indicator values based on the tested population, though only three programs were able to report observed values for all the measures requested. A variety of reasons were given for inability to report measures: the screening program was in its early stages; measures were calculated elsewhere outside the screening program and were not easily available; or reporting such measures was not a current priority of the screening program.

On the basis of the survey definitions, many programs had difficulty in reporting invited and target populations. In some cases, no population register was available; others were able to report data for the program screening interval only, but not for the requested reporting period and in some cases, the distinction between target and invited population was misunderstood. It is clear that the definitions of these screening measures in particular need to be refined or more clearly stated to avoid confusion and to allow the calculation of fundamental indicators of screening program performance, such as coverage rate. Measures of 30-day mortality resulting from an endoscopic procedure were also commonly not available, and in many cases were not routinely collected and linked to the program. These important measures of program performance require particular attention.

At the time of data collection, eight programs (or modalities within programs) which had completed both first and subsequent screens were unable to report data separately for the initial and subsequent screening rounds. For the 22 programs (or modalities within programs) still conducting their first round of screening, their ability to distinguish data from future subsequent rounds is not yet known. Distinguishing between previously screened and unscreened groups is important for indicator comparisons, as detection rates for neoplasia are likely to be higher in those who have not been screened previously. This aspect of routine screening program data collection should be emphasized, because the expected cancer detection rate is an important metric in on-going program evaluation and comparison of program performance.

The survey suggests that the ability of a program to report screening measures was somewhat higher for pilot programs than for full programs, and this may reflect a higher emphasis on evaluation in the pilot phase followed by reduced organizational capacity for data collection and analysis in routine practice.

There was sufficient data to allow comparison between indicator values for first screens in 30 programs (or modalities within programs; Table 7), and for subsequent and first screens separately for 10 programs (or modalities within programs; Table 8). As expected, polyp, adenoma and cancer detection rates (and the related PPVs) were higher for endoscopy-based than FOBt-based programs; and FOBt positivity rates and cancer detection rates tended to be lower for subsequent than for first screens within the same program. While mean values for the indicators were essentially similar when considered by region and between full programs and pilot programs, for many indicators the range of values between individual programs was very wide. Further information on tests will be needed to make reliable comparisons between programs. As this was a first attempt to collect these data on such a large scale, many of the values for the screening measures were either estimated or incomplete.

Some of the variation in FOBt positivity between programs is explicable in terms of known factors relating to the FOBt used or to the tested population. For example, those programs using iFOBts tend to report higher values for test positivity than those using gFOBts. In addition, the higher than average positivity rates seen for the gFOBt programs in Croatia (9.2%) and the USA (New York) (7.0%) may reflect the use of more sensitive gFOBts. The low positivity rates for gFOBts in Spain (1.1%) and England (1.7%) may be due to more stringent criteria for defining test positivity. Information on iFOBt cut-off levels for test positivity was not routinely collected during the survey, but it may help explain the differences between positivity rates among programs using this modality. The wide variation in follow-up rates for diagnostic endoscopy after positive FOBt (35–96% for programs using gFOBt, and 47–95% for programs using iFOBt) may reflect differences in organization of the screening programs and availability of endoscopy services. It should also be noted that the values given are for follow-up within the period reported only; the time lag between FOBt and follow-up colonoscopy means that for some subjects, a positive FOBt but not the subsequent colonoscopy is included in this cross-sectional data.

Using common data indicators in screening programs provides potential for comparing performance across programs with similar designs. The availability of directly comparable performance and outcome data also holds the potential for on-going comparative effectiveness research based on differences in technology, screening intervals, criteria for a positive test result and so forth. Insofar as additional prospective randomized trials of CRC screening are unlikely, policy makers may need to rely on observational studies and the evaluation of service screening for data to inform program design. As in all aspects of clinical practice, screening programs have a responsibility to ensure that the service is of benefit to the targeted population: on-going audit and evaluation is an integral part of their remit. The results of this survey suggest that many programs do not yet have adequate processes in place to collect fundamental measures, such as population coverage and CRC mortality.

We intend to refine some of the screening measure definitions, based on our survey experience and if possible to add other measures such as cancer stage, in the future. It is hoped that existing programs that were unable to complete the survey will use the revised measures and that programs in the planning stages will be able to benefit from the availability of these precise definitions. We would also like to extend an invitation to any interested screening program not already part of the network to contact the ICRCSN.

References

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. References
  7. APPENDIX – INTERNATIONAL COLORECTAL CANCER SCREENING NETWORK
  8. Supporting Information
  • 1
    Benson VS, Patnick J, Davies AK, Nadel MR, Smith RA, Atkin WA, on behalf of the International Colorectal Cancer Screening Network. Colorectal cancer screening: a comparison of 35 initiatives in 17 countries. Int J Cancer 2008; 122: 135767.
  • 2
    International Agency for Research on Cancer. Use of breast cancer screening. In: International Agency for Research on Cancer. IARC handbooks of cancer prevention: breast cancer screening. Lyon, France: IARC Press, 2002. 4786.
  • 3
    Perry N, Broeders M, de Wolf C, Törnberg S, Holand R, von Karsa L, Puthaar E, editors. European guidelines for quality assurance in breast cancer screening and diagnosis. Belgium: European Communities, 2006.
  • 4
    Atkin WA, Benson VS, Green J, Monk CR, Nadel MR, Patnick J, Smith RA, Villain P. Improving colorectal cancer screening outcomes: proceedings of the second meeting of the International Colorectal Cancer Screening Network, a global quality initiative. J Med Screen 2010; 17: 1527.

APPENDIX – INTERNATIONAL COLORECTAL CANCER SCREENING NETWORK

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. References
  7. APPENDIX – INTERNATIONAL COLORECTAL CANCER SCREENING NETWORK
  8. Supporting Information

Principal investigators

J. Patnick, W.S. Atkin.

Expert Working Group for Phase 2 (in alphabetical order by last name)

L. Altenhofen, R. Ancelle-Park, W.S. Atkin, V.S Benson, J. Green, T.R. Levin, S.M. Moss, M. Nadel, J. Patnick, D. Ransohoff, N. Segnan, R.A. Smith, P. Villain, D. Weller.

Participants in the Phase 2 survey (in alphabetical order by country and last name)

Australia: A. Koukari, G. Young; Chile: F. López-Kostner; Croatia: N. Antoljak; Czech Republic: Š. Suchánek, M. Zavoral; Denmark: I. Holten; Finland: N. Malila; France: R. Ancelle-Park, E. Salines; Germany: L. Altenhofen; G. Brenner; Hungary: L. Herszényi, Z. Tulassay; Israel: G. Rennert; Italy: N. Segnan, C. Senore, M. Zappa, M. Zorzi; Japan: H. Saito; Latvia: M. Leja; The Netherlands, Amsterdam and Nijmegen: E. Dekker, J. Jansen; The Netherlands, Greater area of Rotterdam: L. Hol, E. Kuipers; Poland: M.F. Kaminski, J. Regula; Romania: C. Sfarti, A. Trifan; Singapore: C-L Tang; Slovak Republic: R. Hrcka; Spain: G. Binefa, JA. Espinàs, M. Peris; Taiwan: TH. Chen; UK: J. Patnick, R. Steele; Uruguay: G. Pou; USA, Missouri: D. Bisges; USA, Maryland: D. Dwyer, C. Groves; USA, Northern California: S. Courteau, T.R. Levin; USA, New York, all of State: R. Kramer, K. Siegenthaler; USA, New York, Suffolk County: D. Lane; USA, Delaware: C. Herrera, J. Rogers; USA, New Jersey: M. Rojewski; USA, Colorado: H. Wolf.

Participants that attended the Phase 2 international meeting 2008 in addition to those listed above (in alphabetical order by country and last name)

China, Hong Kong: J.J. Sung, K Ling; Canada: H. Bryant, L. Rabeneck, J Dale, L. Sware, H. Yang; France: J. Viguier, L. Von Karsa; Lithuania: L Kupcinskas; The Netherlands, Amsterdam and Nijmegen: M. Deutekom; Sweden: S. Törnberg; UK: J. Austoker (deceased), V. Beral, C. Monk, R. Valori, J. Watson; USA: S. Kobrin, M. Pignone, S. Taplin.

Supporting Information

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. References
  7. APPENDIX – INTERNATIONAL COLORECTAL CANCER SCREENING NETWORK
  8. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
IJC_26310_sm_SuppInfo.doc125KSupporting Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.