Abstract
Objectives
To evaluate the direct costs of first and repeat colorectal cancer screening by immunochemical faecal occult blood testing (I-FOBT).
Methods
Florence district residents aged 50–70 were invited to undergo one-time I-FOBT every two years. Full colonoscopy was recommended for FOBT-positive subjects. Direct cost analysis was carried out separately for the first and repeat screening. All relevant resources consumed by the programme were calculated.
Results
Among 25,428 or 62,369 subjects invited to the first or repeat screening, respectively, the corresponding participation rate was 47.8% or 52.3%, and the positivity rate was 4.4% and 3.3%. Corresponding detection rates and positive predictive values for cancer and advanced adenoma were 11.3% or 8.9% and 32.4% or 32.8%, respectively. The assessment phase accounted for the major cost, as compared with recruitment and screening. All cost indicators were slightly higher in the first screening compared with repeat screening. Cost per cancer and advanced adenoma detected was similar in the first or repeat screening. A higher than observed participation rate would have substantially reduced screening cost.
Conclusion
Analysis of I-FOBT-organized population-based screening cost demonstrates lower cost at repeat compared with first screening and provides reference for decision-making in screening implementation.
INTRODUCTION
Colorectal cancer is the second leading cause of cancer deaths in developed countries. 1 Randomized controlled trials have shown that screening by faecal occult blood testing (FOBT) reduces colorectal cancer mortality. Based on this evidence, FOBT screening is recommended by the European Community. 2 In Italy, full national coverage by a population-based organized screening has been planned and is being implemented. 3
Screening efficacy has been proven by randomized trials employing three-day guaiac-based FOBT (G-FOBT), but several studies have suggested that single-day or two-day immunochemical FOBT (I-FOBT), based on specific antibodies for human haemoglobin (Hb), might be more sensitive and specific than G-FOBT. 4,5 Moreover, I-FOBT requires no dietary restriction, and might substantially improve screening cost-effectiveness. All recent screening programmes implemented in Italy are based on I-FOBT. 6
A population-based FOBT screening programme has been ongoing in Florence District since 1982, run by the International Society for Preventive Oncology (ISPO) – Cancer Prevention and Research Institute. In our previous studies, 5,7 I-FOBT resulted in being more cost-effective than G-FOBT for screening purpose, and was associated with good sensitivity. 8,9 Therefore, a latex agglutination test (LAT) with a positivity threshold of 100 ng Hb/mL of sample solution has been adopted as the standard test in our programme since January 2000.
In the present study we evaluated direct costs of the first and repeat screening in the Florence programme, as we previously did for breast and cervical cancer screening programmes. 10,11 In 1997, we evaluated screening costs using two different FOBT assays, namely a guaiac test (Haemoccult) compared with a reversed passive haemoagglutination test (RPHA). 7 LAT-based screening programme costs were estimated in the present study. As far as we know, this is one of the few cost-outcome analyses on I-FOBT population-based screening.
PATIENTS AND METHODS
Screening programme organization
The main features of the Florence screening programme have already been reported. 12 Briefly, all subjects aged 50–70 years living in Florence district were invited by mail every two years to undergo a one-time immunochemical FOBT, without any dietary restriction. Subjects eligible for screening were identified from the resident lists of District municipalities. Based on ISPO database and on the information provided by general practitioners, subjects who had performed a FOBT test in the last two years or had a double-contrast X-ray in the last three years, or a total colonoscopy in the last five years, were excluded from invitation. In addition, subjects with a personal history of cancer or adenoma under endoscopy surveillance were also excluded.
I-FOBT was based on a LAT test (OC-Hemodia, developed with the OC-Sensor instrument, Eiken, Tokyo, Japan), positivity cut-off being set at 100 ng/mL of sample solution. Negative FOBT reports were mailed together with a biennial screening recommendation. Non-responders to first invitation received a reminder within six months. FOBT-positive subjects were invited to undergo total colonoscopy in accredited assessment clinics during dedicated sessions. In case of incomplete colonoscopy, a double-contrast X-ray in the existing NHS facilities was recommended. Subjects with screen-detected neoplasms were referred for proper surgical or endoscopy treatment. A set of quality indicators are routinely calculated in order to control screening performance 10 and were available for cost analysis. All data concerning screening or assessment examinations were available on ISPO-computerized archives.
Cost analysis
For the economic evaluation we adopted the method described by Drummond. 13 Direct cost analysis was carried out separately for first and repeat screening, as the latter is expected to have a greater effect on overall costs in a stable biennial screening programme, accounting for nine of 10 biennial rounds from age 50–70.
The screening process was divided into three phases:
Recruitment (including costs for mail invitations, organization and evaluation of the screening programme); Screening (including costs for distribution and collection of kits, registration of participants, FOBT assay, mailing of results for FOBT-negative subjects); Assessment (including costs for FOBT-positive subjects recall, colonoscopy, polipectomy or biopsy, reporting of results).
Costs of follow-up for patients with screen-detected neoplasms were not included. For each phase, all relevant resources consumed by the programme (costs) were calculated.
The following cost items were considered:
Staff (endoscopists, epidemiologists, anaesthesiologists, biologists, nurses, laboratory technicians, secretaries, etc.); Equipment (computers, fax, printers, photocopier, portable refrigerators for transport of faecal samples, all endoscopy and laboratory instruments and accessories, etc.); Consumables (stationary, consumable for FOBT assay [e.g. sample kits, reagents] and endoscopy [e.g. loops, rectal probes, bowel preparation solution, drugs]); Outsourcing services (mailing of invitations and reminder letters, double-contrast barium enema X-ray tests, histological examination of biopsies or polipectomies, laundry and sterilization service, disposable material supply service, etc.); General expenses (electricity, phone calls, cleaning, heating, building rent and water).
When a single resource was shared for clinical activities other than screening, the resource percentage attributable to the screening programme was determined. Cost analysis involved the screening activity carried out in the years 2004 and 2005. During the study period, colonoscopies were performed at ISPO Endoscopic Unit and at two other referral centres. As definition of costs in the latter centres was unreliable due to inexact systematic expenditure recording, we assumed all endoscopic examinations being performed at ISPO for study purposes. Endoscopy under deep sedation was offered to all patients, and approximately 80% accepted. An anaesthesiologist was present throughout the whole endoscopy session.
General expenses were calculated by dividing the total expenses of our centre by the proportion of the total building area currently occupied by the screening programme: occupation time was considered in case of areas used for both screening and clinical purposes. Administrative and management costs of ISPO were not considered for assessing general expenses. Software cost was determined on the basis of market prices. Costs of pathological examinations (biopsies and polipectomies) and radiological exams, both performed outside the ISPO, were determined according to Regional Health Authority tariffs. Information on the actual costs of each resource was obtained from the Budget Control Department of ISPO. Prices included taxes. As far as the staff was considered, 2004 salaries were taken into account. Amortization for each piece of equipment ranged between five years and 10 years.
The cost per examined person was defined as the ratio between the total cost of the screening programme and the number of responders. The costs per invited, screened and assessed person were defined as the ratio between the total cost of recruitment, screening and assessment phase and the number of invited, screened and assessed subjects, respectively. Cost per screen-detected cancer or advanced adenoma was defined as the ratio between the total cost of the screening programme and the number of screen-detected cancers or advanced adenomas. Advanced adenoma was defined as adenoma with a diameter ≥ 1 cm, or with at least 20% of villous component or with high-grade dysplasia. Costs were expressed in euros (€) at the 2004 currency.
Henceforth, when the term ‘cost’ is not further specified (i.e. per invited subjects, per screened subjects, per cancer-detected, etc.) it relates to the overall costs.
As participation rate is another important factor that may influence cost-outcome indicators, we estimated cost also to a 70% participation rate assumption, the best of a wide range of results reported in the last Italian national survey. 6
RESULTS
Table 1 shows invited subjects, participants, positive results of FOBT, colonoscopies performed, and screen-detected cancers and advanced adenomas in the first and repeat screening.
Participation was 47.8% or 52.3% in the first or repeat screening, respectively. As expected, corresponding positivity rates were 4.4% and 3.3%, respectively. Detection rates for cancer and advanced adenomas in repeat screening were substantially lower when compared with first screening (8.9‰ vs. 11.3‰), whereas the positive predictive value (PPV) for cancer and advanced adenoma were similar in both the settings.
Number of invited subjects, number of participants and number of colonoscopy assessments in the Florence district screening programme during 2004–2005, detection rate (per 1000 screened subjects) for cancer and for advanced adenomas, positive predictive value (PPV) for cancer and advanced adenomas, by screening round (first or repeat) and in the overall series
Table 2 shows costs according to the screening phase and round. Recruitment phase proportional cost was higher in the first when compared with repeat screening, whereas cost of screening and assessment was lower in the first compared with repeat screening.
Cost (in Euros) by cost item and screening programme phase, at first or repeat screening
Changes in cost by assuming a 70% participation rate when compared with actual participation are reported in square brackets
When cost is analysed by item, staff accounted for the major cost, especially in the first compared with repeat screening (49.7% vs. 45.5%), whereas equipment was associated with lesser cost (4.2% in the first, 4.1% in repeat screening). Major differences between first and repeat screening were recorded for staff, consumables and general expenses, particularly for screening and assessment phases. Significant differences were also observed for outsourcing service costs in recruitment and screening phases. These differences in total cost are due to the higher number of subjects examined in repeat when compared with the first screening (58,929 vs. 25,428 invited subjects, 2.3:1 ratio; see also Table 1).
By assuming a 70% participation rate, a higher cost was observed for the screening (staff involved in kit delivery and report mailing) and assessment phase (consumables and staff costs for colonoscopy, counselling for FOBT-positive subjects), and lower cost was observed for the recruitment phase (reduced number of reminder letters), whereas general expenses increased for all phases. Recruitment phase in the first or repeat screening accounted for 20.8% or 18.2% of total costs, compared with 25.8% or 21.9% with a lower participation rate. Assessment phase in the first or repeat screening accounted for 47.5% or 47.3% of total cost, compared with 43.9% or 42.8% with a lower participation rate (Table 2).
Table 3 shows the cost indicators according to screening phase. All indicators are higher in the first when compared with repeat screening, particularly for the cost per examined person (26 € in the first, 20.1 € in repeat screening), with the only exception of cost per assessed person. By assuming a 70% participation rate, cost indicators are lower in comparison with a lower participation scenario.
Cost indicators according to screening round (in Euros)
Changes in cost by assuming a 70% participation rate when compared with the actual participation are reported in square brackets
A detailed description of costs for staff, equipment, consumables and for outsourcing services is provided in Appendix A.
DISCUSSION
This is a cost-outcome analysis of colorectal cancer screening. Several previous studies have addressed the cost issue, particularly, cost-effectiveness 14–19 but they are dealing with G-FOBT screening: I-FOBT screening is the standard policy in Italy, whereas it is much less adopted in Europe. To our knowledge this is the first study to evaluate direct cost of I-FOBT-based colorectal cancer screening, and may provide important information to decision-makers involved in the implementation of new screening programmes, or in changing the screening test from G-FOBT to I-FOBT, a choice which is becoming increasingly popular for its advantages in terms of acceptability and diagnostic accuracy but is strongly influenced by cost considerations.
The main indicators shown in Table 1 outline the performance of our screening programme. Overall uptake, positivity rate and FOBT PPV for cancer and advanced adenomas are comparable with the results of a recent Italian survey. 6 A higher value of positivity rate in the first screening is probably due to a higher disease prevalence in an unscreened population. Similarly, a decrease in detection rates for cancer and adenomas in repeat screening probably depends on a lower underlying disease prevalence due to a screening effect.
Screening cost is often defined without differentiating between first and repeat screening. We introduced such a differentiation as cost-outcome is expected to vary between the first or repeat screening scenarios, mostly for difference in the prevalence and detection rates of cancer and advanced adenomas. Moreover, the population is expected to undergo one first and nine repeat biennial screening rounds (from 50 to 69 years of age). Thus, repeat screening will have a greater impact on the average cost of the whole screening process: if average cost is not estimated on the whole process, e.g. one first and one repeat round, and no distinction between the first and repeat screening is provided, a bias might be introduced.
According to cost-outcome indicators, our programme was slightly cheaper at repeat when compared with the first screening, and cost per subject examined or per lesion detected were lower at repeat screening. A lower cost per subject examined at repeat screening was expected, as screening cost depends mostly from the proportion of subjects undergoing endoscopy assessment, the latter being lower at repeat screening due to a lower positivity rate. A lower cost per lesion detected was somehow unexpected, considering the lower detection rate, and was due mainly to the fact that in repeat screening recall rate was lower and PPV was slightly higher when compared with the first screening.
Positivity rate has a relevant impact in determining screening cost and efficiency. As mentioned above, the positivity cut-off of screening may be changed, based on cost-effectiveness and cost-analysis considerations. A multicentric study comparing two-sample I-FOBT with different cut-off thresholds compared with the traditional one-sample testing with 100 ng/mL cut-off is in progress in Italy. A cost-outcome analysis will be performed, taking into account the cost-benefit ratio of different levels in positivity cut-off.
A screening programme with limited workload may be less cost-efficient. When the first and repeat screening is compared in our study, it is evident that increasing the workload (screened subjects) by 2.3-fold, total costs increased only 1.9-fold. This is particularly evident when considering phase-specific costs. Staff costs in recruitment phase were reduced by increasing the number of invitations in repeat screening. This finding may indicate that excess staff resources were allocated to the first recruitment phase, as they could successfully cope with a higher screening workload at repeat screening.
Participation rate may influence cost-outcome indicators by improving efficiency. In fact, based on a 70% participation rate assumption, cost indicators were lower when compared with a lower participation rate, in both the first and repeat screening. These data suggest that increasing participation rate is not only likely associated with higher efficacy, but is also more cost-effective. A similar reduction of cost per screened subject by increasing attendance rate was observed also in a previous cost analysis of our programme, 7 and a similar finding was reported by the Nottingham trial. 19 Increasing participation rate is a crucial problem in cancer screening. In the Florence programme participation has slowly but progressively increased over time, probably due to increasing awareness of the population and to promotional campaign at a regional and national level, but also with the help of general practices which were progressively involved in the invitation process. Participation rates as high as 70% will be probably reached in the future, as is observed in northern Italy, 6 but the process is likely to be slow. In order to speed-up, we are presently testing a new invitation procedure foreseeing mail delivery and retrieval of FOBT, which will be the object of a future report.
Cost per screened subject and cost per cancer detected in the first screening were substantially higher when compared with a previous cost analysis of our programme, 7 employing RPHA testing at a higher threshold (26 € vs. 25.0–27.6 US$, and 11,270 € vs. 9020–9940 US$, respectively). Such a comparison may be questionable but also for differences in study design. First, cost assessment in the previous study was based on NHS tariffs, whereas real expenditures were considered in the present study. Second, the current protocol foresees a large use of sedation (anaesthesiology staff), which was not in use in the previous study period. Third, cost per adenoma detected in the present study was based on advanced adenomas, whereas in the previous study it was based on adenomas larger than 9 mm only. This may account for observed cost differences between present and previous analysis (e.g. adjusted cost per adenoma detected would be comparable in both studies).
In Italy, a cost analysis of different colorectal cancer screening models has been carried out in Turin, 20 and cost estimate per subject examined was consistent with our findings (26.21 €), even if a lower participation rate was considered.
In the Nottingham trial cost analysis, the cost per cancer detected calculated at 2002 prices was £ 5290 (7778 €), substantially lower than our estimates in the first or repeat screening (11,270 € and 11,084 €, respectively). This may be explained by several differences in the two study designs. The Nottingham trial used a different screening modality (biennial 3-days G-FOBT, mailed), had a higher participation rate (58%), a lower positivity rate (1.5%) and the cost per cancer detected was calculated over five screening rounds. As far as cancers are concerned, the Nottingham trial was more specific than our programme, but had a lower cancer detection rate (1.7‰), which might suggest that our screening policy is more sensitive.
Results in cost-outcome indicators consistent with our findings have been reported by Denis, 21 estimating direct cost in the first screening based on G-FOBT in a French District. Cost per subject screened was 29.30 €, whereas cost per cancer detected was 13,466€. Diagnostic indicators reported for this French experience are similar to those reported for Florence programme, which may explain similar cost estimates.
CONCLUSION
Present estimates suggest acceptable costs for an organized population-based screening programme. Direct cost analysis suggests that cost per screened subject and per detected lesion is slightly higher in the first when compared with repeat screening. This may be explained by a decrease in positivity rate, a stable PPV for cancer and advanced adenomas, more efficient services with a larger screening workload and an improved participation rate.
The results of the present study may help decision-makers involved in the implementation of a new screening programme and may be used for comparison with other ongoing programmes.
Footnotes
ACKNOWLEDGEMENTS
We are grateful to Mrs Samanta Degl'Innocenti (Secondary Prevention Unit ISPO) and Mrs Manola Turci (Budget Control Department ISPO) for their help in the conduction of the study.