Public health impact of colonoscopy use on colorectal cancer mortality in Germany and the United States

Public health impact of colonoscopy use on colorectal cancer mortality in Germany and the United States

Accepted Manuscript Public health impact of colonoscopy use on colorectal cancer mortality in Germany and the United States Chen Chen, MSPH, Christian...

398KB Sizes 0 Downloads 38 Views

Accepted Manuscript Public health impact of colonoscopy use on colorectal cancer mortality in Germany and the United States Chen Chen, MSPH, Christian Stock, PhD, Michael Hoffmeister, PhD, Hermann Brenner, MD, MPH PII:

S0016-5107(17)31807-2

DOI:

10.1016/j.gie.2017.04.005

Reference:

YMGE 10528

To appear in:

Gastrointestinal Endoscopy

Received Date: 13 January 2017 Accepted Date: 4 April 2017

Please cite this article as: Chen C, Stock C, Hoffmeister M, Brenner H, Public health impact of colonoscopy use on colorectal cancer mortality in Germany and the United States, Gastrointestinal Endoscopy (2017), doi: 10.1016/j.gie.2017.04.005. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Public health impact of colonoscopy use on colorectal cancer mortality in Germany and the United States Running title: Impact of colonoscopy use on CRC mortality in the general population

RI PT

Authors: Chen Chen, MSPH,1 Christian Stock, PhD,1, 2 Michael Hoffmeister, PhD,1 Hermann Brenner, MD, MPH1, 3, 4

Division of Clinical Epidemiology and Aging Research, German Cancer Research Center

SC

1

2

M AN U

(DKFZ), Im Neuenheimer Feld 581, D-69120 Heidelberg, Germany

Institute of Medical Biometry and Informatics, University of Heidelberg, Im Neuenheimer

Feld 130.3, D-69120 Heidelberg, Germany 3

Division of Preventive Oncology, German Cancer Research Center (DKFZ) and National

Center for Tumor Diseases (NCT), Im Neuenheimer Feld 460, D-69120 Heidelberg,

4

TE D

Germany

German Cancer Consortium (DKTK), German Cancer Research Center (DKFZ), Im

EP

Neuenheimer Feld 280, D-69120 Heidelberg, Germany

AC C

Corresponding author:

Hermann Brenner, MD, MPH Division of Clinical Epidemiology and Aging Research German Cancer Research Center Im Neuenheimer Feld 581 D-69120 Heidelberg Phone +49-6221-421300 Fax

+49-6221-421302

ACCEPTED MANUSCRIPT E-mail [email protected]

Grant support: CC was supported by the Helmholtz International Graduate School for

RI PT

Cancer Research at the German Cancer Research Center (DKFZ).

Author contributions: Conception and design: HB; analysis and interpretation of the data: CC, CS, HB; drafting of the article: CC; critical revision of the article for important

AC C

EP

TE D

M AN U

SC

intellectual content: CC, CS, MH, HB; final approval of the article: CC, CS, MH, HB.

ACCEPTED MANUSCRIPT Abstract

2

Background and Aims: Colonoscopy has been demonstrated to be effective in reducing

3

colorectal cancer (CRC) incidence and mortality, and has been widely used for primary CRC

4

screening in Germany and the United States. We performed a population-based analysis to

5

evaluate and compare the public health impact of recent colonoscopy use on CRC deaths

6

among adults aged 55 to 79 years in Germany and the United States from 2008 to 2011.

7

Methods: The epidemiologic metrics of attributable fraction and prevented fraction as well

8

as the impact numbers were calculated using colonoscopy utilization data from nationally

9

representative health surveys, relative risk estimates from medical literature, and CRC death

M AN U

SC

RI PT

1

registry data.

11

Results: Overall, 36.6% (95% credible interval [CrI], 27.3%-45.5%) of CRC deaths in

12

Germany were estimated to be attributable to nonuse of colonoscopy, compared with the U.S.

13

estimates of 38.2% (95% CrI, 28.6%-47.1%) and 33.6% (95% CrI, 24.8%-42.2%) for years

14

2008 to 2009 and 2010 to 2011, respectively. The proportion of CRC deaths theoretically

15

prevented by colonoscopy use within 10 years was 30.7% (95% CrI, 24.8%-35.7%) in

16

Germany, whereas in the United States this proportion ranged from 29.0% (95% CrI, 23.4%-

17

33.6%) for 2008 to 2009 to 33.9% (95% CrI, 27.4%-39.2%) for 2010 to 2011.

18

Conclusions: Recent colonoscopy use is likely to have prevented a considerable fraction of

19

CRC mortality in both countries, and more deaths could be avoided by increasing

20

colonoscopy utilization in the target population. Attributable and prevented fraction can

21

provide valuable information on the public health impact of colonoscopy use and guide the

22

policy making.

AC C

EP

TE D

10

1

ACCEPTED MANUSCRIPT Background

24

Colorectal cancer (CRC) is the third most commonly diagnosed cancer worldwide, and a

25

significant cause of cancer morbidity and mortality in North America and Europe.1 Because

26

of the slow progression from adenomas to invasive cancers, substantial proportions of CRC

27

cases and deaths could be prevented by screening.2 Through detection and removal of

28

precancerous lesions and early cancers, lower gastrointestinal endoscopy can effectively

29

reduce the incidence of advanced disease and thus lead to a decrease in CRC mortality.3-11

30

Lower gastrointestinal endoscopy, especially colonoscopy, is recommended by several

31

guidelines to be used as a screening tool for persons aged 50 and older at average risk for

32

CRC,12-14 and has become one of the principal screening tests in countries like the US and

33

Germany.15, 16

M AN U

SC

RI PT

23

34

The continuous and widespread use of colonoscopy screening is believed to be a major

36

contributory factor in the decreasing CRC incidence and mortality observed in the United

37

States during the past decade.17, 18 Using the epidemiologic metrics of prevented fraction (PF)

38

and attributable fraction (AF), Stock et al19 estimated that approximately 7300 to 11,700 CRC

39

deaths in the United States in 2005 were prevented by colonoscopy use, accounting for 13%

40

to 19% of total CRC deaths among persons aged over 50 years, and about twice as many

41

deaths could be avoided if nonuse of colonoscopy were to be eliminated. In another study,

42

Meester et al20 used both the standard epidemiological method and a microsimulation model

43

to estimate the AF, and reported about 46% to 63% of CRC deaths in the United States in

44

2010 were attributable to not having any of the screening tests.

AC C

EP

TE D

35

45 46

In Germany, colonoscopy has become more commonly accepted and frequently used by the

47

general population over the past decade. In 2008 to 2011, more than half of the screening-

2

ACCEPTED MANUSCRIPT eligible population reported having undergone colonoscopy within the past 10 years.16

49

Although modeling studies have previously evaluated the expected effects of screening

50

colonoscopy on future CRC incidence,21, 22 the actual public health benefits of colonoscopy

51

on deaths from CRC in the general German population have not been explored and remain

52

largely unclear. In this study, we aimed to estimate the CRC mortality prevented by use of

53

colonoscopy and the further potential of colonoscopy use in reducing CRC mortality in

54

Germany, and to compare these results with the estimates of the United States over the same

55

time period.

SC

RI PT

48

56

Methods

58

Data sources

59

The epidemiologic metrics of AF and PF were estimated to reflect the impact of colonoscopy

60

use within the past 10 years on CRC deaths in the general population. Data from several

61

sources were combined for the calculation of AF and PF estimates for Germany and the

62

United States.

TE D

63

M AN U

57

Colonoscopy effectiveness estimates were derived from the medical literature. Studies

65

assessing the effects of colonoscopy versus no colonoscopy on CRC mortality in the general

66

average-risk population were searched at PubMed and Web of Science from inception to June,

67

2016 and combined using meta-analysis to obtain a pooled estimate that also includes the

68

most recent evidence (that is not yet covered in published meta-analyses).

AC C

EP

64

69 70

Data on colonoscopy use in the general population were derived from nationally

71

representative health surveys. The German Health Interview and Examination Survey for

72

Adults (DEGS) is a periodically repeated nationwide health survey of non-institutionalized

3

ACCEPTED MANUSCRIPT German population aged between 18 to 79 years.23 The most recent estimates of endoscopic

74

CRC screening adherence were available from the first wave of DEGS (2008-2011).

75

Respondents aged 55 to 79 years were asked if they had ever undergone colonoscopy and the

76

time since the last examination. This age range was selected because screening colonoscopy

77

is offered from age 55 on only in Germany. Indication for colonoscopy use was not

78

ascertained. The proportion of people who have undergone colonoscopy within the past 10

79

years as well as the age- and sex-specific prevalence estimates, which were weighted to be

80

representative of the resident population of Germany as of 31 December 2010, were directly

81

extracted from the study report.16

M AN U

82

SC

RI PT

73

Endoscopy use data in the United States were drawn from the National Health Interview

84

Survey (NHIS), a household, multistage probability sample survey conducted annually since

85

1957.24, 25 The NHIS is one of the principle sources of information on the health of the

86

civilian, non-institutionalized U.S. population. To enable comparisons of estimates from the 2

87

countries, data among adults aged 55 to 79 years from 2008 and 2010, when CRC screening

88

test utilization was ascertained, were used as estimates for 2008 to 2009 and 2010 to 2011,

89

respectively. In the 2008 NHIS, respondents aged 40 and older were asked if they have ever

90

had sigmoidoscopy, colonoscopy or proctoscopy. If utilization was reported, respondents

91

were then asked to specify the type of the most recent examination they had (sigmoidoscopy,

92

colonoscopy or proctoscopy) as well as the time and reason for this most recent exam. In the

93

2010 NHIS, respondents aged 40 and older were asked separate questions on colonoscopy,

94

sigmoidoscopy, CT colonography and fecal occult blood test (FOBT). A brief description of

95

each test was provided. Regarding colonoscopy use, participants were asked if they have ever

96

had colonoscopy and the time and reason for the most recent examination.

AC C

EP

TE D

83

97

4

ACCEPTED MANUSCRIPT Numbers of deaths from CRC, defined as cancers of colon and rectum (C18-20 by

99

International Classification of Diseases-10 Version), were extracted from the German Centre

100

for Cancer Registry Data (ZfKD)26 and the United States Cancer Statistics (USCS) report,27

101

for the time period 2008 to 2011. The cancer mortality data on the ZfKD platform are

102

provided by the Federal Statistical Office Germany, which collects data on all deaths in

103

Germany. The USCS is a Web-based report that includes the official federal statistics on

104

cancer mortality provided by the National Vital Statistics System, with 100% of the U.S.

105

population covered.27

SC

RI PT

98

106

Statistical analysis

108

The colonoscopy effectiveness parameter was obtained by combining the effect estimates of

109

observational studies that evaluated the impact of colonoscopy on CRC mortality. Studies

110

that specifically and exclusively addressed screening colonoscopy were excluded. Risk ratios

111

and hazard ratios from cohort studies as well as odds ratios from case-control studies that

112

were used to measure effects were all referred to as relative risk. The pooled effect size and

113

95% confidence intervals (CIs) were calculated using a random-effects model.28

TE D

EP

114

M AN U

107

Prevalence of colonoscopy use in the United States was calculated from NHIS 2008 and 2010

116

data. All colonoscopies were included in the analysis regardless of indication. Respondents

117

with a history of CRC or missing information on CRC history were excluded to enable an

118

analysis in the average-risk population. To obtain nationally representative estimates, sample

119

weights were applied to reflect the selection probabilities as well as adjustments for non-

120

response and post-stratification.29, 30 Variance estimates were computed by Taylor series

121

linearization method to adjust for complex sample design.29, 30

AC C

115

122

5

ACCEPTED MANUSCRIPT We used 2 measures of attribution and prevention to quantify the impact of colonoscopy use

124

on CRC mortality. The metric of AF was first introduced in a study on the association of

125

smoking and lung cancer, and has gradually become widely used for measuring the impact of

126

an exposure on an outcome at the population level.31 AF is often interpreted as the proportion

127

of those with the outcome that might be attributable to the exposure, or the maximum

128

proportion of those with the outcome that might be reduced if the exposure is eliminated.32, 33

129

In our study, the AF for colonoscopy use refers to the proportion of CRC mortality that is

130

likely to be attributable to not having had a colonoscopy within 10 years, and also means the

131

maximum proportional reduction in CRC mortality if nonuse of colonoscopy were to be

132

eliminated.19 PF is an alternative public health impact measure that is more commonly used

133

in the cases of protective exposure or intervention. It represents the proportion of those

134

avoiding the outcome because of the protective exposure.34 In our analysis, the PF for

135

colonoscopy use is defined as the proportion of hypothetical CRC deaths that have been

136

prevented by recent colonoscopy use.19 Absolute numbers of CRC deaths attributable to

137

nonuse and prevented by use of colonoscopy were derived by multiplying the estimated AFs

138

and PFs with the observed number of CRC deaths and estimated hypothetical CRC deaths if

139

colonoscopy use were eliminated. (See Appendix for detailed methods.) We used 2 different

140

metrics in an effort to reflect both the benefits of colonoscopy use in the past and the

141

potential gains of further increasing colonoscopy use in the future. The 95% credible intervals

142

(CrIs) of AFs, PFs, and absolute impact numbers account for the uncertainty in both

143

prevalence of colonoscopy use and colonoscopy effectiveness, and were calculated by Monte

144

Carlo simulation. The probability distribution for included parameters was built based on

145

their point estimates and standard errors. Distributions of outcomes were then generated by

146

running simulations with 10,000 iterations, from which the quantiles 2.5% and 97.5% were

AC C

EP

TE D

M AN U

SC

RI PT

123

6

ACCEPTED MANUSCRIPT 147

extracted to derive 95% CrIs. Monte Carlo errors were smaller than 5% of the standard

148

deviation of the outcomes under 10,000 iterations.

149

Sex- and age-specific estimates were calculated for each of the parameters and outcomes

151

except colonoscopy effectiveness. Stata 13.1 (StataCorp, College Station, Tex, USA) and

152

SAS version 9.4 (SAS Institute Inc., Cary, NC, USA) were used to perform meta-analyses

153

and the estimation of colonoscopy use prevalence, respectively. R version 3.2.5 (R

154

Foundation for Statistical Computing, Vienna, Austria) was used for Monte Carlo

155

simulation.35

156

Results

157

Estimates of parameters

158

Estimates of the effects of colonoscopy use on CRC mortality are summarized in Table 1. Of

159

the included 8 observational studies that evaluated the relative risk, 4 were cohort studies36-39

160

and 4 case-control studies.40-43 All studies were carried out in North America. Except in one

161

study in which persons who had colonoscopy use were compared with the general

162

population,36 all other studies conducted the comparison between those with colonoscopy use

163

and without. The 8 studies comprised colonoscopy use and CRC deaths data collected

164

between 1981 and 2010, and covered colonoscopy performed up to 10 or more years before

165

CRC diagnosis or death. All studies were at least matched by or adjusted for age and sex.

166

Despite the heterogeneity in populations, settings, and study design, colonoscopy use for any

167

indication was consistently associated with a reduction in CRC mortality in all studies, with a

168

pooled estimate of 56% risk reduction (RR 0.44; 95% CI, 0.35-0.55).

AC C

EP

TE D

M AN U

SC

RI PT

150

169 170

Prevalence of colonoscopy use in Germany and the United States are shown in Table 2.

171

During 2008 to 2011, 54.8% (95% CI, 52.5%-57.0%) of German adults aged 55 to 79 years 7

ACCEPTED MANUSCRIPT reported completing a colonoscopy in the last 10 years, the proportions being much higher in

173

age groups 60 to 69 (58.4%) and 70 to 79 (56.6%) than in age group 55 to 59 (46.1%). Data

174

from the NHIS yielded comparable estimates. In 2008, 55.8% (95% CI, 54.2%-57.4%) of the

175

U.S. population aged 55 to 79 had undergone a colorectal endoscopy (colonoscopy,

176

sigmoidoscopy, or proctoscopy) in the past 10 years (data not shown), from which over 90%

177

identified colonoscopy as the most recent endoscopy (51.7% [95% CI, 50.1%-53.2%] of the

178

overall 55-79 population). In 2010, 60.4% (95% CI, 59.0%-61.8%) of respondents indicated

179

that they have had this procedure in the preceding 10 years. For both calendar years, the

180

estimated proportion was found to be increasing with age.

SC

RI PT

172

M AN U

181

During 2008 to 2011, a total of 109,479 and 57,591 persons aged 55 to 79 years died of CRC

183

in the United States and Germany, respectively (Supplementary Table 1). Figure 1 displays

184

the trend in CRC deaths with age. For both countries the death count and mortality rate

185

increased with age. This increase in mortality with age was much larger in Germany than in

186

the United States. When examining the trends in CRC deaths during the 4 years, a decreasing

187

trend for the mortality rate was observed for both countries.

EP

188

TE D

182

Estimates of AFs and PFs

190

Overall from 2008 to 2011, 21,086 (95% CrI, 15,720-26,215) CRC deaths in Germany were

191

estimated to be attributable to nonuse of colonoscopy within the previous 10 years,

192

accounting for 36.6% (95% CrI, 27.3%-45.5%) of total CRC deaths among people aged 55-

193

79 years (Table 3). The AFs in the United States were 38.2% (95% CrI, 28.6%-47.1%) and

194

33.6% (95% CrI, 24.8%-42.2%) for years 2008 to 2009 and 2010 to 2011, respectively,

195

which implies that a maximum of 20,947 (95% CrI, 15,724-25,864) and 18,344 (95% CrI,

196

13,525-23,052) estimated deaths would have been avoided if all adults 55 to 79 years of age

AC C

189

8

ACCEPTED MANUSCRIPT 197

had had a colonoscopy within 10 years. In both countries, more CRC deaths among men were

198

attributable to nonuse of colonoscopy than among women. In Germany, the vast majority of

199

attributable deaths occurred in age group 70 to 79 years, whereas such estimates were more

200

equally distributed across age groups in the United States.

RI PT

201

Over the same period of time, 25,564 (95% CrI, 18,983-32,002) CRC deaths in Germany

203

were theoretically prevented by recent use of colonoscopy, accounting for 30.7% (95% CrI,

204

24.8%-35.7%) of the estimated hypothetical CRC deaths of the population in the absence of

205

colonoscopy (Table 4). The percentage of CRC deaths theoretically prevented by

206

colonoscopy use in the United States for years 2008 to 09 and 2010 to 11 were 29.0% (95%

207

CrI, 23.4%-33.6%) and 33.9% (95% CrI, 27.4%-39.2%), respectively. These translated into

208

22,422 (95% CrI, 16,786-27,794) and 27,980 (95% CrI, 20,595-35,235) prevented deaths. In

209

stratified analyses, although no systematic differences emerged between male and female in

210

PFs, substantially more deaths could be prevented by the use of colonoscopy among men

211

than among women, and the largest number of deaths was prevented in age group 70 to 79 in

212

both countries.

EP

213

TE D

M AN U

SC

202

Discussion

215

In this study, we evaluated the public health impact of colonoscopy use on CRC deaths in

216

Germany and the United States. Our results indicate that in Germany from 2008 to 2011

217

approximately 25,600 deaths among adults aged 55 to 79 years were prevented by

218

colonoscopy use within the past 10 years, and about 37% of the actual 57,591 CRC deaths at

219

best could have been avoided if 100% colonoscopy utilization were achieved. Comparable

220

with the estimates of Germany, colonoscopy is likely to have saved around 50,400 persons

221

aged 55 to 79 from CRC death in the United States over the same time period, and about

AC C

214

9

ACCEPTED MANUSCRIPT 222

39,300 out of 109,479 actual CRC deaths (36%) at the most could have been prevented by

223

100% use of colonoscopy within 10 years.

224

The observed trend in AF and PF estimates with age primarily reflects the differences in

226

colonoscopy use. Persons aged 55 to 59 years, as the group with the lowest prevalence of

227

recent colonoscopy use, had the highest AFs and the lowest PFs, though absolute numbers of

228

attributable and prevented deaths were largest in age group 70 to 79 as a result of the

229

substantially higher CRC mortality rate in this age group. Compared with the United States,

230

Germany had a particularly high proportion of attributable CRC deaths in age group 70 to 79.

231

This can be explained by a much larger increase of CRC mortality with age and much smaller

232

reduction in population size from age 55 to 79 in Germany,26, 27 which might serve as

233

evidence for including the elderly aged 75 to 79 years as screening target in this country.

234

Because of the similar proportions of persons who have undergone colonoscopy within 10

235

years, sex differences in AF and PF estimates were small. However, with males having a

236

higher CRC mortality rate in both countries, especially Germany, where the CRC mortality

237

estimates of men are about 70% to 80% higher than those of women in the target age group

238

for CRC screening,26 more men can be prevented from dying from CRC by recent use of

239

colonoscopy. Estimates of 2008 and 2010 from the United States confirmed the expectation

240

of Stock et al. that with increasing use of colonoscopy and decreasing CRC mortality rate in

241

recent years, AFs would decrease and PFs would increase over time.19

SC

M AN U

TE D

EP

AC C

242

RI PT

225

243

Colonoscopy is recommended as one of the primary screening tools to persons at average risk

244

for CRC in both Germany and the United States.13, 44 CRC screening by FOBT, followed by

245

colonoscopy in case of a positive FOBT result, has been offered in Germany since 1977.

246

Since 2002, colonoscopy has been recommended for primary CRC screening to persons over

10

ACCEPTED MANUSCRIPT 55 years. This change, along with increased use of colonoscopy for diagnostic purposes, has

248

contributed to the increase in colonoscopy use prevalence in Germany over the past decade,

249

which has gradually risen to the level of the United States and is much higher than in other

250

European countries.15, 16, 45, 46 With the increased use of colonoscopy, over 30% of CRC

251

deaths in Germany were avoided from 2008 to 2011, similar to the estimate of the United

252

States. Our findings thereby underline the large potential for reducing the mortality from

253

CRC through colonoscopy, even though even larger reductions would be possible by higher

254

utilization of the screening offers.

SC

RI PT

247

255

The accuracy of AF and PF estimates is dependent on how well the colonoscopy

257

effectiveness parameter we used represents the true association between the test and CRC

258

mortality in the target population. There have been many recent cohort and case-control

259

studies investigating the effect of colonoscopy on CRC incidence and mortality in multiple

260

populations.47, 48 Our summary estimate of colonoscopy effectiveness is in line with estimates

261

derived in 2 other previously reported meta-analyses that summarized the effect of

262

colonoscopy for any indication (relative risk, 0.40; 95% CI, 0.32-0.49, and 0.43 95% CI,

263

0.33-0.58, respectively),47, 48 but had not yet included the most recently published studies

264

addressing this research question.39, 43

TE D

EP

AC C

265

M AN U

256

266

Although the metrics of AF and PF have been widely used to quantify the impact of an

267

exposure on an outcome in epidemiological studies, there have been relatively few attempts

268

to estimate the public health impact of colonoscopy as a protective factor on CRC.19, 20 Our

269

results are generally consistent with the findings of these analyses that nonuse of colonoscopy

270

or other screening tests contributed to a considerable fraction of CRC deaths in the United

271

States, though there are some variations that generate from the different colonoscopy

11

ACCEPTED MANUSCRIPT effectiveness parameter used, screening tests considered and years explored. In the study of

273

Stock et al,19 colonoscopy effectiveness from 2 case-control studies, one optimistic and one

274

conservative estimate were used for analysis, which led to 2 estimates for AF and PF each for

275

2005: 28% and 44%, and 13% and 19%, respectively.19 Meester et al20 evaluated the impact

276

of all CRC screening tests on CRC deaths in the United States in 2010, and obtained an AF of

277

46% when a relative risk of 0.32 from one specific prospective cohort study was applied.

RI PT

272

278

It is worth noting that use of colonoscopy is not a preventive measure by itself, but that its

280

preventive effect comes through removal of precancerous lesions or treatment of CRC at an

281

early stage when these lesions are detected at colonoscopy. In the majority of colonoscopies,

282

no such lesions are detected, and those colonoscopies do not have any preventive effect.

283

Calculation of AFs and PFs for use of colonoscopy is nevertheless meaningful in countries,

284

such as Germany and the United States where colonoscopy is offered and widely used for

285

primary CRC screening and to a large extent reflects screening coverage. In countries where

286

CRC screening is primarily based on noninvasive methods, such as fecal immunochemical

287

test for hemoglobin (FIT), such as the Netherlands,49 and colonoscopy is primarily used to

288

follow-up positive results of such noninvasive tests, comparable levels of prevention should

289

be achievable with much lower rates of use of colonoscopy, as the proportion of

290

colonoscopies that yield a preventive effect, ie, those with relevant findings, should be much

291

larger under such circumstances.

M AN U

TE D

EP

AC C

292

SC

279

293

Some limitations should also be noted in considering the findings of our study. First, only 2

294

of the studies evaluating colonoscopy effects included procedures performed in the past

295

decade,36, 43 whereas others considered colonoscopy completed many years ago, with the

296

furthest dating back to 1980s.40 This might underestimate the strength of the association

12

ACCEPTED MANUSCRIPT between colonoscopy and CRC death, as better test performance is expected nowadays with

298

the advances in technology. Second, an overall relative risk instead of age- and sex-specific

299

estimates was used in stratified analyses due to lack of reporting of subgroup data and

300

different categorization of ages. This is not expected to bias our estimates greatly, especially

301

sex-specific estimates, because colonoscopy is found to provide a similar magnitude of

302

protection towards men and women.48 Third, questions regarding CRC screening tests

303

utilization in the NHIS across different survey years were not phrased in a standardized way,

304

which makes it hard to conduct direct comparisons over time. In 2008, respondents were

305

asked to report the type of the most recent colorectal endoscopy from colonoscopy,

306

sigmoidoscopy, or proctoscopy, which might underestimate the proportion of colonoscopy

307

use within 10 years; however, as the vast majority identified colonoscopy as the most recent

308

test, the impact should not be major. Another limitation relates to the self-reporting nature of

309

colonoscopy use data for both countries, which might result in an overestimation of

310

utilization; however, high sensitivity and specificity between self-reports and medical record

311

data were found.50

SC

M AN U

TE D

312

RI PT

297

In summary, colonoscopy use within 10 years prevented a considerable fraction of CRC

314

mortality in Germany and the United States during 2008 to 2011, and there is a large

315

potential for further reducing the mortality by increasing colonoscopy utilization, either as

316

primary screening examination or to follow-up positive results of promising noninvasive

317

screening tests, such as FITs, in the target population. Our findings confirm the values of AF

318

and PF metrics in evaluating the public health impacts of colonoscopy in screening settings,

319

which can serve as evidence in the development and prioritization of intervention programs

320

and strategies regarding endoscopic CRC screening. Although how to set up an attainable and

AC C

EP

313

13

ACCEPTED MANUSCRIPT 321

realistic goal requires more exploration, our results underline the need to further encourage

322

CRC screening in the general population.

323

References

324

1.

RI PT

Ferlay J, Soerjomataram I, Dikshit R, et al. Cancer incidence and mortality worldwide:

325

sources, methods and major patterns in GLOBOCAN 2012. Int J Cancer

326

2015;136:E359-86.

327

2.

Knudsen AB, Zauber AG, Rutter CM, et al. Estimation of Benefits, Burden, and Harms of Colorectal Cancer Screening Strategies: Modeling Study for the US

329

Preventive Services Task Force. JAMA 2016;315:2595-609. 3.

M AN U

330

SC

328

Atkin W, Wooldrage K, Parkin DM, et al. Long term effects of once-only flexible

331

sigmoidoscopy screening after 17 years of follow-up: the UK Flexible Sigmoidoscopy

332

Screening randomised controlled trial. Lancet 2017, DOI:10.1016/s0140-

333

6736(17)30396-3. 4.

Lin JS, Piper MA, Perdue LA, et al. Screening for Colorectal Cancer: A Systematic

TE D

334

Review for the U.S. Preventive Services Task Force. Evidence Synthesis No. 135.

336

AHRQ Publication No. 14-05203-EF-1. Rockville, MD: Agency for Healthcare

337

Research and Quality, 2016. 5.

colorectal cancer incidence and mortality: a randomized clinical trial. JAMA

339

2014;312:606-15.

340 341

Holme O, Loberg M, Kalager M, et al. Effect of flexible sigmoidoscopy screening on

AC C

338

EP

335

6.

Brenner H, Stock C, Hoffmeister M. Effect of screening sigmoidoscopy and screening

342

colonoscopy on colorectal cancer incidence and mortality: systematic review and

343

meta-analysis of randomised controlled trials and observational studies. BMJ

344

2014;348:g2467.

14

ACCEPTED MANUSCRIPT 345

7.

with screening flexible sigmoidoscopy. N Engl J Med 2012;366:2345-57.

346 347

Schoen RE, Pinsky PF, Weissfeld JL, et al. Colorectal-cancer incidence and mortality

8.

Manser CN, Bachmann LM, Brunner J, et al. Colonoscopy screening markedly reduces the occurrence of colon carcinomas and carcinoma-related death: a closed

349

cohort study. Gastrointest Endosc 2012;76:110-7.

350

9.

RI PT

348

Elmunzer BJ, Hayward RA, Schoenfeld PS, et al. Effect of flexible sigmoidoscopybased screening on incidence and mortality of colorectal cancer: a systematic review

352

and meta-analysis of randomized controlled trials. PLoS Med 2012;9:e1001352. 10.

following a colonoscopy. Dig Liver Dis 2011;43:899-904.

354 355

Strock P, Mossong J, Scheiden R, et al. Colorectal cancer incidence is low in patients

11.

M AN U

353

SC

351

Segnan N, Armaroli P, Bonelli L, et al. Once-only sigmoidoscopy in colorectal cancer

356

screening: follow-up findings of the Italian Randomized Controlled Trial--SCORE. J

357

Natl Cancer Inst 2011;103:1310-22. 12.

Bibbins-Domingo K, Grossman DC, Curry SJ, et al. Screening for Colorectal Cancer:

TE D

358 359

US Preventive Services Task Force Recommendation Statement. JAMA

360

2016;315:2564-75.

the American College of Physicians. Ann Intern Med 2015;162:718-25.

362 363

14.

Burt RW, Cannon JA, David DS, et al. Colorectal cancer screening. J Natl Compr Canc Netw 2013;11:1538-75.

364 365

Wilt TJ, Harris RP, Qaseem A. Screening for cancer: advice for high-value care from

EP

13.

AC C

361

15.

Shapiro JA, Klabunde CN, Thompson TD, et al. Patterns of colorectal cancer test use,

366

including CT colonography, in the 2010 National Health Interview Survey. Cancer

367

Epidemiol Biomarkers Prev 2012;21:895-904.

15

ACCEPTED MANUSCRIPT 368

16.

Starker A, Sass AC. Participation in cancer screening in Germany. Results of the

369

German Health Interview and Examination Survey for Adults (DEGS1).

370

Bundesgesundheitsblatt-Gesundheitsforschung-Gesundheitsschutz 2013;56:858-867.

2014;64:104-17.

372 373

18.

Zauber AG. The impact of screening on colorectal cancer mortality and incidence: has it really made a difference? Dig Dis Sci 2015;60:681-91.

374 375

Siegel R, Desantis C, Jemal A. Colorectal cancer statistics, 2014. CA Cancer J Clin

19.

RI PT

17.

Stock C, Knudsen AB, Lansdorp-Vogelaar I, et al. Colorectal cancer mortality

SC

371

prevented by use and attributable to nonuse of colonoscopy. Gastrointest Endosc

377

2011;73:435-443 e5.

378

20.

M AN U

376

Meester RG, Doubeni CA, Lansdorp-Vogelaar I, et al. Colorectal cancer deaths

379

attributable to nonuse of screening in the United States. Ann Epidemiol 2015;25:208-

380

213 e1. 21.

Brenner H, Altenhofen L, Stock C, et al. Expected long-term impact of the German

TE D

381 382

screening colonoscopy programme on colorectal cancer prevention: analyses based on

383

4,407,971 screening colonoscopies. Eur J Cancer 2015;51:1346-53.

endoscopy on colorectal cancer incidence: a modelling study. Oncotarget 2016, DOI:

385

10.18632/oncotarget.10178.

386 387

23.

first data collection wave. BMC Public Health 2012;12:730.

389

391

Scheidt-Nave C, Kamtsiuris P, Gosswald A, et al. German health interview and examination survey for adults (DEGS) - design, objectives and implementation of the

388

390

Brenner H, Kretschmann J, Stock C, et al. Expected long-term impact of screening

EP

22.

AC C

384

24.

National Center for Health Statistics. Data File Documentation, National Health Interview Survey, 2010 (machine readable data file and documentation). National

16

ACCEPTED MANUSCRIPT 392

Center for Health Statistics, Centers for Disease Control and Prevention, Hyattsville,

393

Maryland. 2011.

394

25.

National Center for Health Statistics. Data File Documentation, National Health Interview Survey, 2008 (machine readable data file and documentation). National

396

Center for Health Statistics, Centers for Disease Control and Prevention, Hyattsville,

397

Maryland. 2009.

398

26.

RI PT

395

The German Centre for Cancer Registry Data (ZfKD). Available at:

www.krebsdaten.de/Krebs/EN/Database/databasequery_step1_node. Accessed July

400

19, 2016. 27.

U.S. Cancer Statistics Working Group. United States Cancer Statistics: 1999–2013

M AN U

401

SC

399

402

Incidence and Mortality Web-based Report. Atlanta: U.S. Department of Health and

403

Human Services, Centers for Disease Control and Prevention and National Cancer

404

Institute; 2016. Available at: www.cdc.gov/uscs. Accessed July 19, 2016. 28.

1986;7:177-88.

406 407

DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials

29.

TE D

405

National Center for Health Statistics, Centers for Disease Control and Prevention. 2008 National Health Interview Survey (NHIS) Public Use Data Release: NHIS

409

Survey Description. Available at: ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/Dataset_Documentation/NHIS/2008/srv

410

ydesc.pdf. Accessed May 27, 2016.

411 412

AC C

EP

408

30.

National Center for Health Statistics, Centers for Disease Control and Prevention.

413

2010 National Health Interview Survey (NHIS) Public Use Data Release: NHIS

414

Survey Description. Available at:

415

ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/Dataset_Documentation/NHIS/2010/srv

416

ydesc.pdf. Accessed May 27, 2016.

17

ACCEPTED MANUSCRIPT 31.

Epidemiol 2015;25:147-54.

418

32.

prevented fractions, and causation probabilities. Ann Epidemiol 2015;25:155-61.

420 421

33.

34.

Benichou J. A review of adjusted estimators of attributable risk. Stat Methods Med Res 2001;10:195-216.

424 425

Levine B. What does the population attributable fraction mean. Prev Chronic Dis 2007;4:A14.

422 423

Greenland S. Concepts and pitfalls in measuring and interpreting attributable fractions,

RI PT

419

Poole C. A history of the population attributable fraction and related measures. Ann

35.

SC

417

R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing, 2016. URL https://www.R-

427

project.org/. 36.

after colonoscopy varies by site of the cancer. Gastroenterology 2010;139:1128-37.

429 430

Singh H, Nugent Z, Demers AA, et al. The reduction in colorectal cancer mortality

37.

Jacob BJ, Moineddin R, Sutradhar R, et al. Effect of colonoscopy on colorectal cancer

TE D

428

M AN U

426

431

incidence and mortality: an instrumental variable analysis. Gastrointest Endosc

432

2012;76:355-64 e1.

screening effectiveness: an example of colonoscopy. J Med Screen 2013;20:198-207.

434 435

39.

wide cohort of screening age. Gastrointest Endosc 2016, DOI:

437

10.1016/j.gie.2015.12.035.

438

440

Stock D, Paszat LF, Rabeneck L. Colorectal cancer mortality reduction is associated with having at least 1 colonoscopy within the previous 10 years among a population-

436

439

Eldridge RC, Doubeni CA, Fletcher RH, et al. Uncontrolled confounding in studies of

EP

38.

AC C

433

40.

Muller AD, Sonnenberg A. Protection by endoscopy against death from colorectal cancer. A case-control study among veterans. Arch Intern Med 1995;155:1741-8.

18

ACCEPTED MANUSCRIPT 441

41.

from colorectal cancer. Ann Intern Med 2009;150:1-8.

442 443

Baxter NN, Goldwasser MA, Paszat LF, et al. Association of colonoscopy and death

42.

Baxter NN, Warren JL, Barrett MJ, et al. Association between colonoscopy and colorectal cancer mortality in a US cohort according to site of cancer and

445

colonoscopist specialty. J Clin Oncol 2012;30:2664-9.

446

43.

RI PT

444

Samadder NJ, Curtin K, Pappas L, et al. Risk of Incident Colorectal Cancer and Death After Colonoscopy: A Population-based Study in Utah. Clin Gastroenterol Hepatol

448

2016;14:279-286.e2.

449

44.

SC

447

German Guideline Program in Oncology (German Cancer Society, German Cancer Aid, AWMF): Evidenced-based Guideline for Colorectal Cancer, long version 1.0,

451

AWMF registration number: 021-007OL,

452

http://leitlinienprogrammonkologie.de/Leitlinien.7.0.html. Accessed August 28, 2016.

453

45.

M AN U

450

Spaeth A, Zwahlen M. Use of lower gastrointestinal endoscopy and fecal occult blood test in the 2007 Swiss Health Interview Survey respondents aged 50 years and older.

455

Endoscopy 2013;45:560-6.

456

46.

TE D

454

Benning TM, Dellaert BGC, Severens JL, et al. The Effect of Presenting Information about Invasive Follow-Up Testing on Individuals' Noninvasive Colorectal Cancer

458

Screening Participation Decision: Results from a Discrete Choice Experiment. Value Health 2014;17:578-587.

459 460

47.

Elmunzer BJ, Singal AG, Sussman JB, et al. Comparing the effectiveness of competing tests for reducing colorectal cancer mortality: a network meta-analysis.

461

Gastrointest Endosc 2015;81:700-709 e3.

462 463

AC C

EP

457

48.

Pan J, Xin L, Ma YF, et al. Colonoscopy Reduces Colorectal Cancer Incidence and

464

Mortality in Patients With Non-Malignant Findings: A Meta-Analysis. Am J

465

Gastroenterol 2016;111:355-65.

19

ACCEPTED MANUSCRIPT 49.

overview of existing programmes. Gut 2015;64:1637-49.

467 468

Schreuders EH, Ruco A, Rabeneck L, et al. Colorectal cancer screening: a global

50.

Dodou D, de Winter JC. Agreement between self-reported and registered colorectal cancer screening: a meta-analysis. Eur J Cancer Care (Engl) 2015;24:286-98.

469

RI PT

466

470

Figure legends

472

Figure 1. CRC death counts and mortality rates in Germany and the United States, 2008 to

473

2011

474

Death count is the total CRC death count during 2008 to 2011.

475

CRC = colorectal cancer

476

Data source:

477

1. U.S. Cancer Statistics Working Group. United States Cancer Statistics: 1999 to 2013

478

Incidence and Mortality Web-based Report. Atlanta: U.S. Department of Health and Human

479

Services, Centers for Disease Control and Prevention and National Cancer Institute; 2016.

480

Available at: www.cdc.gov/uscs.

481

2. The German Centre for Cancer Registry Data (ZfKD). Available at:

482

www.krebsdaten.de/Krebs/EN/Database/databasequery_step1_node.

AC C

EP

TE D

M AN U

SC

471

20

ACCEPTED MANUSCRIPT

Table 1. Effects of colonoscopy use on colorectal cancer mortality from observational studies Country Study design Years US Case-control 1981-1992

No. Cases: 4358; controls: 16,531

Colonoscopy time frame* Up to 10 years before diagnosis

Covariates adjusted/matched Matched by age, sex, race; adjusted for other colorectal procedures, procedures other than colorectal, length of coverage by the Department of Veterans Affairs, arthritis-related diseases

Relative risk (95% CI) 0.45 (0.30-0.66)

Baxter, 2009 41

Canada

Case-control

1992-2003

Cases: 10,292; controls: 51,460

6 months-10 years before diagnosis

Matched by age, sex, geographic location, socioeconomic status; adjusted for comorbid conditions

0.63 (0.57-0.69)

Singh, 2010 36

Canada

Cohort†

1987-2008

54,803

6 months before diagnosis21 years before CRC deaths

Age, sex, calendar year

0.71 (0.61-0.82)

Baxter, 2012 42

US

Case-control

1991-2007

Cases: 9458; controls: 27,641

6 months-12 years before diagnosis

Matched by sex, year of birth, race, SEER registry; adjusted for socioeconomic status, comorbidities, urban/rural status

0.40 (0.37-0.43)

Jacob, 2012 37

Canada

Retrospective 1996-2005 cohort

1,089,998

3 years before diagnosis-10 years before CRC deaths

Primary care physician (age, sex, country of medical training) and patient characteristics (age, sex, income, comorbidity)

0.19 (0.07-0.47)

Eldridge, 2013 38

US

Cohort

1996-2008

68,531

Before diagnosis-16 years before CRC deaths

Age, sex, education, race, diabetes, family history of CRC, hormone replacement therapy use, healthy lifestyle score

0.41 (0.30-0.55)

Samadder, 2016 43

US

Case-control

1996-2010

Cases: 5128; controls: 20,512

6 months-10 years before diagnosis

Matched by age, sex; adjusted for family history of CRC, church affiliation status

0.33 (0.28-0.39)

Stock, 2016 39

Canada

Cohort

1992-2009

1,509,423

6 months before diagnosis10 years before CRC deaths

Age, sex, socioeconomic status, comorbidity

0.36 (0.33-0.38)

SC

M AN U

TE D

EP AC C

Pooled estimate from random effects model

RI PT

First author, year Muller, 1995 40

0.44 (0.35-0.55)

* Colonoscopy time frame for colorectal cancer cases † Comparison with the general population 21

ACCEPTED MANUSCRIPT Table 2. Prevalence of colonoscopy use within 10 years in Germany and the United States* Germany

United States 2008-09 6022

2010-11 7366

Overall†

54.8 (52.5-57.0)

51.7 (50.1-53.2)

60.4 (59.0-61.8)

Sex Male Female

54.1 (50.8-57.3) 55.4 (52.6-58.2)

52.0 (49.7-54.3) 51.4 (49.4-53.4)

60.8 (58.7-63.0) 60.0 (58.3-61.8)

Age (years) 55-59 60-69 70-79

46.1 (41.4-50.8) 58.4 (55.0-61.7) 56.6 (52.9-60.2)

47.6 (44.9-50.4) 52.6 (50.3-54.9) 54.9 (52.2-57.6)

55.5 (52.8-58.2) 61.7 (59.9-63.6) 63.6 (61.2-65.9)

SC

M AN U

Sample size

RI PT

2008-11 1889

* Prevalence estimates were based on DEGS1 for Germany and NHIS 2008 and 2010 for the US [% (95% CI)]

AC C

EP

TE D

† Overall prevalence of colonoscopy use among persons aged 55 to 79 years

22

ACCEPTED MANUSCRIPT

Table 3. Estimated fractions and numbers of CRC deaths attributable to nonuse of colonoscopy in Germany and United States

RI PT

United States 2008-09 AF, % (95% CI) No. of deaths (95% CI)† 38.2 (28.6-47.1) 20,947 (15,724-25,864)

37.0 (27.5-46.0) 36.3 (27.0-45.2)

13,069 (9728-16,259) 8074 (6003-10,062)

38.0 (28.5-47.0) 38.3 (28.8-47.3)

Age 55-59 60-69 70-79

40.8 (30.8-50.1) 34.7 (25.6-43.6) 35.7 (26.4-44.7)

2040 (1538-2504) 6475 (4770-8133) 12,107 (8948-15,158)

40.1 (30.3-49.2) 37.7 (28.3-47.0) 36.6 (27.2-45.5)

3227 (2439-3959) 7889 (5909-9766) 9479 (7058-11,798)

2010-11 No. of deaths (95% CI)† AF, % (95% CI) 33.6 (24.8-42.2) 18,344 (13,525-23,052)

33.4 (24.6-42.0) 33.8 (24.9-42.5)

10,435 (7676-13,143) 7891 (5818-9912)

36.3 (27.0-45.2) 32.9 (24.1-41.4) 31.7 (23.2-40.3)

3007 (2236-3747) 7083 (5203-8933) 7856 (5733-9968)

TE D

AF = attributable fraction; CRC = colorectal cancer.

11,907 (8928-14,723) 9029 (6780-11,145)

M AN U

Sex Male Female

SC

Overall*

Germany 2008-11 AF, % (95% CI) No. of deaths (95% CI)† 36.6 (27.3-45.5) 21,086 (15,720-26,215)

AC C

EP

* Overall estimates for persons aged 55 to 79 years. † Sex- and age-specific estimates do not necessarily sum up to the overall estimate.

23

ACCEPTED MANUSCRIPT

Table 4. Estimated fractions and numbers of CRC deaths prevented by colonoscopy use in Germany and United States

2008-09 No. of deaths (95% CI)† PF, % (95% CI) 29.0 (23.4-33.6) 22,422 (16,786-27,794)

Sex Male Female

30.4 (24.4-35.5) 31.1 (25.0-36.2)

15,404 (11,392-19,453) 10,029 (7421-12,633)

29.2 (23.5-33.9) 28.8 (23.3-33.5)

Age 55-59 60-69 70-79

25.9 (20.4-30.8) 32.8 (26.3-38.3) 31.8 (25.5-37.2)

1745 (1283-2228) 9089 (6664-11,570) 15,789 (11,597-20,094)

26.7 (21.5-31.2) 29.5 (23.8-34.3) 30.8 (24.8-35.9)

SC

M AN U

12,899 (9627-16,083) 9549 (7143-11,862)

2931 (2197-3649) 8755 (6527-10,925) 11,539 (8551-14,510)

2010-11 No. of deaths (95% CI)† PF, % (95% CI) 33.9 (27.4-39.2) 27,980 (20,595-35,235)

34.1 (27.5-39.6) 33.7 (27.2-39.0)

16,185 (11,878-20,482) 11,836 (8709-14,917)

31.1 (25.1-36.3) 34.6 (28.0-40.1) 35.7 (28.8-41.4)

3750 (2776-4720) 11,411 (8362-14,441) 13,726 (10,003-17,489)

TE D

CRC = colorectal cancer; PF = prevented fraction.

United States

RI PT

Overall*

Germany 2008-11 PF, % (95% CI) No. of deaths (95% CI)† 30.7 (24.8-35.7) 25,564 (18,983-32,002)

AC C

EP

* Overall estimates for persons aged 55 to 79 years. † Sex- and age-specific estimates do not necessarily sum up to the overall estimate.

24

ACCEPTED MANUSCRIPT

Supplementary Table 1. Colorectal cancer death counts ages 55-79 years in Germany and the United States, 2008-2011* Germany 2009 2010 14,319 14,194

2011 14,101

2008 27,675

9157 5820

8774 5545

8683 5511

8736 5365

15,726 11,949

Age 55-59 60-64 65-69 70-74 75-79

1269 1754 3291 4194 4469

1256 1711 3148 4083 4121

1211 1688 2740 4362 4193

1265 1805 2517 4167 4347

4003 4927 5514 6170 7061

15,593 11,617

15,652 11,522

15,615 11,805

4042 5063 5410 6036 6659

4047 5230 5498 5865 6534

4248 5456 5371 5960 6385

M AN U

EP

TE D

Deaths from cancers of colon and rectum (C18-20 by International Classification of Diseases-10 Version)

AC C

*

2011 27,420

SC

Sex Male Female

United States 2009 2010 27,210 27,174

RI PT

Total

2008 14,977

25

ACCEPTED MANUSCRIPT Appendix. Statistical analysis of AF and PF We applied the method described in detail in the study of Stock et al. to calculate the estimates of AF, PF and the corresponding death numbers, Naf and Npf.19 Naf and Npf were

RI PT

derived by multiplying the estimated AF and PF with the observed number of CRC deaths and estimated hypothetical CRC deaths in the absence of colonoscopy, respectively. Briefly, the following formulas were used:

SC

AF = P0(RR0/1-1)/[ P0(RR0/1-1)+1]

Naf = AF×N Npf = PF×N/[P1(RR1/0-1)+1]

M AN U

PF = P1(1-RR1/0)

(1) (2) (3) (4)

AF: attributable fraction PF: prevented fraction

TE D

Symbols and acronyms were defined as follows:

EP

P0: prevalence of nonuse of colonoscopy (unexposed) P1: prevalence of colonoscopy use (exposed); P1=1-P0

AC C

RR1/0: relative risk of deaths from CRC in people with use of colonoscopy (exposed) versus people without (unexposed) RR0/1: inverse of the relative risk for exposure to colonoscopy use; RR0/1=1/ RR1/0 Naf: Number of CRC deaths attributable to nonuse of colonoscopy Npf: Number of CRC deaths prevented by use of colonoscopy N: actual CRC death number of the year

26

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT Acronyms List AF, attributable fraction; CI, confidence interval; CRC, colorectal cancer; CrI, credible interval; DEGS, German Health Interview and Examination Survey for Adults; FIT, fecal

RI PT

immunochemical test for hemoglobin; FOBT, fecal occult blood test; NHIS, National Health Interview Survey; PF, prevented fraction; USCS, United States Cancer Statistics; ZfKD,

AC C

EP

TE D

M AN U

SC

German Centre for Cancer Registry Data