Projected impact of colorectal cancer screening with computerized tomographic colonography on current radiological capacity in Europe

Authors


Dr C. Hassan, Ospedale Nuovo Regina Margherita, Gastroenterologia ed Endoscopia Digestiva, Via Morosini 30, 00153 Rome, Italy.
E-mail: cesareh@hotmail.com

Summary

Background  The impact of a primary colorectal cancer screening with computerized tomographic colonography on current radiological capacity is unknown. The multispecialty needs for computerized tomographic examinations raise some doubts on the feasibility of a mass colorectal cancer screening with computerized tomographic colonography.

Aim  To assess whether the number of available computerized tomographic units in Europe is adequate to cover population screening with computerized tomographic colonography.

Methods  A mathematical and a Markov model were, respectively, used to assess the number of computerized tomographic colonography procedures needed to be performed each day in the start-up and in the steady-state phases of a colorectal cancer screening programme in Europe. Such outcome was divided for the total number of computerized tomographic machines aged <10 years estimated to be present in the European hospitals.

Results  At a simulated 30% compliance, 28 760 130 European people would need to be screened by the 3482 available computerized tomographic units in a 5-year start-up period, corresponding to 6.6 CTC/CT unit/day. Assuming a 10-year repetition of computerized tomographic colonography between 50 and 80 years, the number of computerized tomographic colonography needed to be performed in the steady-state period appeared to be 4.3/CT unit/day.

Conclusions  The current radiological capacity may cover the need for a primary colorectal cancer screening with computerized tomographic colonography in a steady state. On the other hand, a substantial implementation of the current computerized tomographic capacity or a synergistic approach with other techniques seems to be necessary for the start-up period.

Introduction

Computerized tomographic colonography (CTC) has already been shown to be a feasible procedure for colorectal cancer (CRC) screening, being able to detect precancerous adenomatous polyps and early stage CRCs.1 Although early studies have offered mixed results,2, 3 CTC may be regarded as the second most accurate technique for colorectal neoplasia detection after colonoscopy.4 Moreover, two meta-analyses have shown a relatively high sensitivity of CTC for clinically meaningful lesions, such as large polyps and CRC.5, 6 The expected advantage of CTC over colonoscopy in CRC screening is represented by a potential improvement in population compliance, as CTC, also named as ‘virtual colonoscopy’, is perceived as a less invasive procedure than colonoscopy.7 The need for a more accepted CRC screening strategy is also underlined by the still low CRC screening uptake, especially when compared with breast, cervical and prostate screening programmes, and by the evidence that even after being properly informed, a substantial portion of asymptomatic people still refuse colonoscopy.8–10 When dealing with a mass population programme, cost-effectiveness is also important. Although previous studies have shown conflicting results on the relative incremental cost-effectiveness between colonoscopy and CTC, all of them agreed on an acceptable cost-effectiveness of CTC when compared to no screening, being the cost per life-year saved consistently less than the $100 000 threshold widely accepted for mammography or cervical cancer screening.11–14 On the other hand, no study has dealt with the potential impact of a population CRC screening programme with CTC on the current radiological capacity. Different from endoscopy, CT units are necessary for virtually all the medical and surgical specialties, and it is, therefore, extremely unlikely that radiological centres may provide most of the CT time-machine to gastroenterological screenings. Previous studies have shown that endoscopy capacity may not be sufficient for a whole population CRC screening with colonoscopy in the United States,15 suggesting the need for CTC capacity to match, at least in part, the impressive screening demand.

A periodic survey is conducted by the European Coordination Committee of the Radiological, Electromedical and Medical IT Industries (COCIR) to monitor the age profile of medical equipment in Europe.16 Briefly, survey data are provided by the main companies involved in CT installation in the selected countries, and they entail the total number of CT units installed in each country by these companies, their setting (hospital and nonhospital) and their age.

The aim of this simulation study was to compare the potential demand of CTC examinations with the current CT capacity in Europe, when primary CRC screening with CTC is simulated in the general population.

Methods

The COCIR survey provides data on the CT equipment installed in the nine main European countries (Belgium, Finland, France, Germany, Italy, the Netherlands, Spain, Sweden and UK). Being the survey data necessary for evaluating the current CT capacity, the analysis was applied only to the cumulative population of these countries. When population screening is performed, two demands on the available healthcare resources should be clearly distinguished:17 (i) a start-up period in which all the unscreened 50- to 80-year population will undertake their first screening examination; (ii) a steady-state period in which only those people turning 50 years require the first testing, the other needing only to be periodically retested up to 80 years with a CTC every decade. To address the first issue, we constructed a simple mathematical model to predict the CTC demand, considering CTC as the only screening strategy available to the compliant population. The outcome measured was the total number of predicted CTC in 1 year, assuming a 5-year interval as a realistic period to screen all the millions of unscreened European people between 50 and 80 years of age. To deal with the second issue, we used a previously published Markov model to estimate the number of CTC needed to be performed each year once the screening programme, based on a CTC every decade, is in a steady state.14, 18 The outcomes of the two phases were therefore divided for the number of CT units available in the considered countries as estimated by the COCIR, and for the number of working days in 1 year to establish the number of CTC procedures per day that each CT unit should perform to meet all the simulated demand.

Model inputs

Population

Demographic characteristics were separately assessed for each of the nine included countries. In detail, the number of individuals between the ages of 50 and 80 years were obtained from the international database of the US census, and the corresponding life expectancy from the World Health Organization.19, 20 Previous studies on the feasibility of CRC screening with flexible sigmoidoscopy in European countries assessed the rate of people eligible to CRC screening, ranging between 89% and 95%.21, 22 These studies also indicated an overall compliance between 18% and 35%. As flexible sigmoidoscopy is usually perceived as a non-invasive test when compared to colonoscopy, we assumed in the reference case scenario, a similar compliance of 30% for the simulated screening with CTC. Further data are provided in Table 1.

Table 1.   Characteristics and parameters used in the two mathematical models on which the analysis is based
VariableReference case value (range)*
  1. The range shown for each variable represents the interval used for the sensitivity analysis.

  2. CRC, colorectal cancer; CTC, computerized tomographic colonography; DIS, distant; LOC, localized; REG, regional.

  3. * It was distributed as follows: Belgium (3 215 000), Finland (1 719 000), France (17 988 000), Germany (27 200 000), Italy (19 005 000), the Netherlands (4 825 000), Spain (12 053 000), Sweden (2 879 000) and UK (17 635 000).

Population
 Number in Europe, 50–80 years old*106 519 00019
 CTC CRC screening compliance30% (10–90)21, 22
 Eligible for CRC screening90%21, 22
Natural history
 Adenoma prevalence at age 50 (%) 15 (10–45)25–37
  Fraction of adenomas <10 mm (%)9525–37
  Fraction of adenomas ≥10 mm (%)525–37
 New adenomatous polyp rate (% per year)Age specific 2–346
 Annual transition rate from ≤5 to 6–9 mm (%)Age specific 2–4 (0.02–8)47, 48
 Annual transition rate from 6–9 to ≥10 mm (%)Age specific 2–4 (0.02–8)47, 48
 Annual transition rate from ≥10 mm to LOC CRCAge specific 3–4 (1–10)49, 50
 Annual transition rate from LOC CRC to REG CRC (%)33 (25–10)51
 Annual transition rate from REG CRC to DIS CRC (%)40 (20–60)51
Screening
 CTC sensitivity for ≤5 mm polyps (%)48 (25–60)5, 6
 CTC sensitivity for 6–9 mm polyps (%)70 (45–90)5, 6
 CTC sensitivity for ≥10 mm polyps (%)85 (35–100)5, 6
 CTC sensitivity for CRC (%)95 (70–100)5, 6
 CTC specificity (%)86 (35–100)5, 6
CT capacity
 Number of CT units in Europe611216
 Hospital-based CT units (%)6416
 CT units aged <10 years (%)1216

Markov model

To evaluate the number of CTC examinations performed in the steady-state phase of a population screening, we used a Markov model on a hypothetical cohort of 100 000 subjects ageing from 50 to 100 years, as previously suggested in papers dealing with colonoscopy capacity.23, 24 Details of the model are reported elsewhere.14, 18 Briefly, the natural history of colorectal neoplasia was calibrated to reproduce the age- and sex-specific prevalence at autopsy and screening studies of both adenomas and large adenomas,25–37 and the incidence and mortality rate of CRC.38 The health intervention superimposed on the natural history model was CTC repeated every 10 years between 50 and 80 (included) years, the number of polyps detected being purely related to the polyp and CRC detection according to the accuracy values presented in Table 1. In the CTC strategy, detection of a lesion was followed by a colonoscopy with polypectomy or biopsy as necessary. Two surveillance schedules were simulated: (i) patients in follow-up for a diminutive polyp were allowed to go back to the CTC screening compartment at the first negative endoscopy and (ii) those in follow-up for a >6 mm polyp remained in the 5-year surveillance endoscopic programme until the end of the simulation (without requiring further CTC screening).

To project the outcomes of our simulation upon the cumulative population of the included countries, we assumed a steady state for population size and age distribution, represented by the year 2005 census data.19 Each age-specific output (per person) of the model was multiplied by the number of people of that age in the European population. Adding the results for all ages yielded national estimates on the number of CTC needed to be performed each year.

CT capacity

The COCIR survey separately offers data on the number of CT machines installed in a hospital and in a nonhospital setting.16 For the purpose of our analysis, we included only the CT units installed in the European hospitals, which represent 64% of all the equipment available in the included countries. Also, we considered only those units aged <10 years, excluding those older because not technically feasible for a CTC examination. We assumed each CT unit to be working on average 250 days/year.23

Sensitivity analysis

One- and two-way sensitivity analysis was performed for all the variables of the model (Table 1), the results being reported for those most relevant.

Results

Start-up phase

At the simulated compliance and eligibility rate, 28 760 130 European people would need to be screened with CTC. Considering a 5-year period to meet this potential demand, the number of CTC to be performed each year is 5 752 026. The number of hospital based, aged <10 years CT units in the included countries was 3482. Consequently, the number of CTC examinations/year needed to be performed by each CT unit in Europe was equal to 1652, corresponding to 6.6 CTC/CT unit/day.

At sensitivity analysis, compliance to screening, CT capacity, population size and the years on which the start-up phase is distributed appeared to be the key variables. As illustrated in Figure 1, a reduction of the expected compliance from 30% to 15% would reduce the number of CTC needed to be performed each day from 6.6 to 3.3, whilst a compliance increase to 55% or 75% would lead to 12 and 16.5 CTC/CT unit/day. Simulating an increase of CTC capacity by 30%, corresponding to a CTC number of 4527, the CTC screening procedures would be reduced to 5/CT unit/day, whilst a 50% reduction (1741 available CT) would lead to the need of 13.2 procedures. Limiting the population only to those younger than 65, 60 or 55 years would reduce the burden of examinations from 6.6 to 3.9, 2.7 and 1.4 CTC/CT unit/day. As illustrated in Figure 2, a progressive increase of the years on which the screening is distributed also affects the outcome of the analysis. Assuming a 7-year period, the number of CTC/CT unit/day would be reduced to 4.7, being 3.3 when simulating a 10-year distribution.

Figure 1.

 Sensitivity analysis (start-up period). Variation of the number of computerized tomographic (CT) colonography examinations needed to be performed by each CT unit per day to screen all the European population according to compliance and CT capacity.

Figure 2.

 Sensitivity analysis (start-up period). Variation of the number of computerized tomographic (CT) colonography examinations needed to be performed by each CT unit per day to screen all the European population according to the number of years over which the start-up period of the screening is distributed and compliance.

At two-way sensitivity analysis, assuming a 30% increase of CTC capacity, the number of CTC/CT unit/day at compliance rates of 15%, 55% and 75% would be 2.5, 9.3 and 12.7 (Figure 1). On the other hand, simulating a 50% decrease of CTC capacity, the number of CTC/CT unit/day would be 6.6, 24.2 and 33 at compliance rates of 15%, 55% and 75%, respectively. Including only the 50–65 years population and varying the compliance rates, the number of CTC/CT unit/day would be 1.9, 7.1 and 9.6 at compliance rates of 15%, 55% and 75%, respectively.

Steady-state phase

When simulating a screening with CTC every decade and an initial compliance of 30%, the overall number of CTC procedures needed to be performed in 1 year in the steady-state period of the European population between 50 and 75 years appeared to be 3 774 437. Dividing such value for the estimated CT capacity, it corresponds to 1084 procedures/CT unit/year and 4.3/CT unit/day. At sensitivity analysis, the potential CTC demand was heavily sensitive to variation in screening compliance. As illustrated in Figure 3, a reduction of compliance to 15% resulted in a demand drop to 2.2 CTC/CT unit/day, whilst an increase to 55% and 75% would lead to a need for 8 and 10.8 CTC/CT unit/day, respectively. Increasing CT capacity by 30% would reduce the number of CTC/CT unit/day to 3.3, whilst a 50% decrease resulted in the need of 8.7 procedures per day. At two-way sensitivity analysis, assuming a 30% increase in CT capacity, at compliance rates of 15%, 55% and 75%, the number of CTC/CT unit/day would be 1.7, 8 and 10.8, respectively, whilst, simulating a 50% decrease in CT capacity, at the same compliance rates, the corresponding numbers would be of 4.3, 16 and 21.7 (Figure 3). Limiting the population only to those people younger than 65 or 55 years would decrease the number of examinations per CT unit/day from 4.3 to 2.8 and 1.6, respectively. If CTC had to be repeated not every decade, but every 5 years between 50 and 80 years, the corresponding number of CTC needed to be performed each day would rise from 4.3 to 7.6.

Figure 3.

 Sensitivity analysis (steady-state period). Variation of the number of computerized tomographic (CT) colonography examinations needed to be performed by each CT unit per day to screen all the European population according to compliance and CT capacity.

Discussion

Our study clearly shows that the current CT capacity in Europe is able to satisfy a substantial demand for CRC screening with CTC. However, if all the compliant population had to be screened with CTC, the burden of demand would strenuously stress the current CT capacity during the start-up period. Indeed, to perform in all the European CT units, nearly 7 CTC/day in addition to the everyday CT routine would seem unlikely, unless a large implementation had to be accomplished. On the other hand, if we assume that modern CT may perform 20 examinations per day, the potential demand for four examinations per day in the steady-state period would seem, although significant, reasonable. This evidence strictly mirrors an analogous analysis performed in the US for colonoscopy. Seef et al. clearly showed that the current colonoscopy capacity would be inadequate to meet all the potential demand in the start-up phase of a US population screening with colonoscopy,15 whilst Ladabaum’s and Vijan’s showed that it is enough to match the simulated demand during the steady-state period.23, 24 These data altogether suggest that there is neither radiological nor endoscopic capacity to undertake a primary screening with CTC or colonoscopy. On the other hand, it is possible that a mix of these procedures could satisfy the impressive demand during the initial phase of the programme. For instance, we have shown that only 3.3 CTC/CT unit/day should be performed when reserving CTC to only 15% of the total population, corresponding to 50% of the compliant people, and only 2 CTC/day would be sufficient to repeat the procedure every decade in a life-time period. Moreover, a 30% increase of current CT capacity would practically halve these values, rendering the estimate extremely realistic. Intriguingly, such 30% corresponds to the CT machines already installed in Europe in a nonhospital setting. Health authorities could decide to employ these machines for the population screening more than to install new hospital-based CT units that would be necessary only in the first part of the screening programme. To reduce the potential demand on CT capacity, CTC could be reserved to those people more likely to gain a benefit from this procedure, while reserving colonoscopy to the others. Interestingly, it has been shown that restricting CTC only to low-risk people (women and men below 60 and 55 years, respectively, without family history) and proposing colonoscopy to the high-risk subjects would substantially improve the overall detection of advanced neoplasia, mainly through an increase in the negative predictive value of CTC.39 Our analysis showed that restricting CTC only to people younger than 60 or 55 years heavily reduces the burden of examinations, being a realistic alternative to a whole population screening with CTC. The aim of a non-invasive test is to recruit people who are unwilling to undertake the invasive procedure, but who may be presumed to accept the invasive procedure once a polyp or a cancer is detected. In this case, the non-invasive test will ultimately improve the CRC prevention rate achieved by colonoscopy. On the other hand, if the non-invasive test is only able to pick up people who would have undertaken the invasive procedure in the first place, it will only result in a uselessly costly duplication of the screening procedures, with a net reduction of efficacy, since, even in the best scenario of accuracy, a non-invasive test will never be more sensitive than colonoscopy in identifying colonic polyps.

All these speculations are based on the assumptions that all the available CT capacity may be dedicated to CTC. We clearly showed that if only 50% of the CT units adhered to a CTC screening, there would not be enough capacity to meet a large demand. It should be clarified whether the entire radiological community is willing to be involved in CRC screening and, more in general, in CTC technique. Moreover, an adequate training is needed. The European Society of Gastrointestinal and Abdominal Radiology has been offering official CTC workshops in a 2-day format since 2003. In these years, seven courses have been organized and 758 radiologists have been trained. It is still unclear how much training is necessary. Fifty cases with colonoscopic correlation seem to be adequate for some individuals, especially those with previous experience in gastrointestinal radiology.40 This would increase the throughput of training courses. In the ACRIN 6664 CTC multicentric trial, readers were obligated to have read at least 500 cases or attend a 1.5-day training course and also to pass a certified examination on 50 cases, with a detection rate of at least 90% for adenomas 1 cm or larger. Noteworthy, more than 50% of the readers were needed to undergo additional training to pass the certified examination.41

In this study, we simulated a strategy based on a 10-year repetition of CTC in the steady-state period. Indeed, it has already been shown that such schedule may be a cost-effective alternative to colonoscopy and does effectively reduce the number of post-CTC colonoscopy examinations compared to a primary colonoscopy screening.13, 14, 42 On the other hand, current radiological and gastroenterological guidelines mainly support a higher frequency of examinations such as CTC every 5 years.43, 44 Our data suggest that such an intensive schedule is not technically feasible at the baseline assumptions, implying nearly eight extra-CT procedures per day in the steady-state period of the programme. Similarly, the present analysis also indicates that, at the present time, there is no enough radiological capacity to meet a potentially very high compliance from the general population, such as those reached by prostate and cervical cancer screenings. If this was the case, the creation of radiological centres mainly dedicated to CTC examinations should be implemented, as it has already been successfully proved to be possible.1 Alternatively, the number of years on which the start-up phase is distributed could be shifted by 5–10 years, as it has already been suggested for a primary colonoscopy screening.15 However, such a wide interval between the first and the last person to be screened may appear disputable from an ethical point of view. No study has ever addressed the feasibility of any screening programme in Europe. Our analysis estimates for the first time the potential demand of CRC prevention in these countries. Although we used these data to ascertain the feasibility of a radiological screening, they could also be used to estimate the feasibility of other programmes in Europe. For instance, we could reasonably presume that the number of endoscopy units in Europe at least equals that of CT units, taking into consideration that the cost of an endoscopic unit is much lower than that of a CT unit. Unfortunately, no database similar to the CORI system in US is available in Europe to have a better idea on the endoscopic capacity.24 However, if this was the case, colonoscopy would have the same capacity limitations as CTC, making a synergistic approach with the two techniques mutually necessary. The European endoscopic capacity being still unknown, we were also unable to assess the impact of a population CTC screening on the endoscopic resources. However, a recent study showed a 8% post-CTC referral rate for polyps larger than 6 mm.45 Such ability of CTC in rationalizing the access to endoscopy may be useful whether the endoscopic capacity for a primary colonoscopic screening was shown to be insufficient.

In conclusion, our study shows that the available CTC capacity in the European countries is large enough to satisfy the potential demand of a primary population screening with CTC in the steady-state period. On the other hand, the initial start-up phase of CRC screening may pose a significant burden on the current CT units that a large implementation of the CT capacity or a synergistic approach with other techniques is clearly necessary.

Acknowledgement

Declaration of personal and funding interests: None.

Ancillary