UICC workshop on the evaluation of screening programmes for cancer: Meeting held in Venice, Italy, on November 14-16, 1983
✍ Scribed by P. C. Prorok; J. Chamberlain; N. E. Day; M. Hakama; A. B. Miller
- Publisher
- John Wiley and Sons
- Year
- 1984
- Tongue
- French
- Weight
- 443 KB
- Volume
- 34
- Category
- Article
- ISSN
- 0020-7136
No coin nor oath required. For personal study only.
✦ Synopsis
A workshop of the Project on Screening for Cancer of the International Union Against Cancer (UICC) was held in Venice, Italy, on November 14-16, 1983. The focus of this workshop was on general principles for evaluation of screening and on three specific cancer sites: lung, bladder and oral cavity.
Evaluation of screening
Considerable attention must be paid to the careful evaluation of cancer screening programmes. Because a screening programme has to be in place before screening can be evaluated, careful planning should precede wide implementation of screening. The circumstances of evaluation may differ from one setting to another, for example in technically advanced compared to developing countries. It should be appreciated that screening presents both benefits and disadvantages (Table I). TABLE 1 -BENEFITS AND DISADVANTAGES OF SCREENING FOR CANCER Benefits Disadvantages ~ _____ Improved prognosis for some cases detected by screening Less radical treatment needed to cure some cases Reassurance for those with negative test results Longer morbidity for cases whose prognosis is unaltered Overtreatment of borderline abnormalities False reassurance for those with false negative results Unnecessary morbidity for those with false positive results Hazard of screening test Resource savings Resource costs
Measurement of the effect of screening
The basic measure of screening benefit is the effect on the population mortality rate. Determination of changes in the incidence rate of invasive cancer is appropriate when screening is aimed at detecting pre-invasive lesions and preventing the occurrence of invasive cancer. The yield of cases gives no information about prognosis and benefit. A shift to a more favorable stage distribution in a screened group is of limited value since such a shift can occur, whether or not outcome is improved, due to lead time and length biases. However, if all cases in a population offered screening and a comparable one not offered screening are ascertained, the rate of advanced cases in the two groups may yield information that reflects possible future changes in mortality. Comparison of survival rates involving screen-detected cases are unreliable because of lead time and length biases. The lack of reliable procedures to estimate and adjust for these biases pinpoints a field where further research is called for. a A randomized trial is the preferred method of evaluation as it removes the necessity for considering confounding and bias. Ideally, the screening method should first be evaluated by means of the randomized trial approach, with randomization of individuals. However, timing is crucial as randomized trials must be carried out before a test is widely disseminated.
Certain information requirements and design standards are desirable for randomized trials. These include inclusion and exclusion criteria, the technique of randomization, adequate sample size, detailed description of the screening, diagnostic, therapeutic and follow-up procedures, blind assessment of critical endpoints, presence of co-morbid conditions, compliance, extent of co-intervention and contamination, details of diagnostic criteria for cases and deaths, including histologic criteria, mortality from all causes, methods of data analysis, and procedures for deciding at what time definitive evaluation might be made.
If a randomized trial is not carried out, it will be necessary to rely on non-randomized cohort or quasiexperimental studies, saturating one area in comparison with others, or, less satisfactorily, correlating intensity of screening with time trends of incidence and mortality. Quasi-experimental studies are more demanding to perform and evaluate than randomized trials and depend on appropriate records being available. All study designs require attention to quality control of intervention manoeuvres and data collection.
It is important to recognize that follow-up in randomized trials and quasi-experimental studies is very demanding, and adequate resources should be made available for this purpose.
Case-control studies have been suggested as an approach to screening evaluation. In principle, such studies can be used to assess screening efficacy by prevention of death or prevention of invasion, through appropriate definition of cases and controls. The approach also has the potential for evaluating the sensitivity of screening tests, and may yield information on the effectiveness of a range of different screening stategies. However, as a novel methodology in this setting, the approach itself requires further evaluation. A number of different designs Methods of evaluation of the effectiveness of screening