Intended for healthcare professionals

Education And Debate

An economic view of high compliance as a screening objective

BMJ 1994; 308 doi: https://doi.org/10.1136/bmj.308.6921.117 (Published 08 January 1994) Cite this as: BMJ 1994;308:117
  1. D J Torgerson,
  2. C Donaldson
  1. Health Economics, Research Unit, University of Aberdeen, Polwarth Building, Aberdeen AB9 2ZD.

    Accepted wisdom holds that high compliance is essential for a screening programme to be successful. Indeed, a reason that the national breast screening programme is not routinely offered to women aged 65 or more is on the grounds of predicted poor compliance by older women. Increasing compliance is often associated with increased costs. These costs represent a lost opportunity for screening alternative target populations. We question the need for screening programmes to achieve high compliance, and we argue that a screening programme can be efficient with very low levels of compliance. Adopting compliance as a screening objective and as a measure of the success of screening may be detrimental to the efficiency of a screening programme.

    The level of compliance in a screening programme is often seen as one measure of its success.1 Compliance can be defined as the proportion of a target population that actually has the screening test, and compliance rates feature strongly in policy recommendations regarding screening. The new general practitioner contract has targets of 50% and 80% compliance for cervical screening, and, for people aged over 75, 100% compliance for annual health checks is expected.2 Forrest recommended that routine breast screening should not be extended to women aged over 65 because of expected poor compliance in this age group.3 This emphasis on the importance of compliance as a measure of success in breast screening was repeated in a recent report on the NHS breast screening programme.4

    The aim of this paper is to assess whether compliance is a good measure of success in screening. This is important because compliance in itself is not, of course, the real objective of screening. Compliance is merely a proxy objective that may be more easily measurable than the real screening objective. In order to assess whether compliance is a good proxy for the real screening objectives, however, we need to ask what the true objectives are and whether achieving high compliance helps to meet these objectives.

    Screening objectives

    A screening programme's objective might be a reduction in the morbidity and mortality associated with the target disease. As resources are always limited this objective needs to be couched in terms of resource availability. Thus a screening objective of a maximum reduction in morbidity and mortality within available resources is more realistic. This is an efficiency goal.

    Compliance and efficiency

    The efficiency of an intervention such as screening may be measured in terms of cost per unit of health gain relative to other uses of resources. The efficiency of increasing compliance is thus determined by its expected health gains relative to health gains from other uses of the resources used to increase compliance. It is therefore important to know both the costs and health gains of increasing compliance, and we need to ask questions such as what is the cost and what are the expected health benefits of increasing compliance from 60% to 70%, and do the benefits outweigh the costs?

    Screening programmes tend to be biased towards the non-manual classes in the sense that such programmes are used more by these social groups.5 Thus, increasing compliance may be expected to attract more people from the manual groups. Some diseases are associated with social class: the incidence of breast cancer is higher in non-manual groups6 whereas the incidence of cervical cancer is higher in manual groups.7 Increasing compliance for breast cancer might therefore be associated with a declining rate of case detection and hence a declining rate of health gain. Such gains may not be worth the cost incurred. For cervical cancer, however, increasing compliance might be associated with an increasing rate of case detection with consequent increases in the rate of health gain. Hence, the decision whether to spend extra resources to increase compliance must be made in the context of targeted disease if efficiency is to be considered.

    Compliance and screening for breast cancer

    Efficiency rather than compliance ought to be the goal of a screening programme. It has to be realised that increasing compliance may result in foregone benefits that are greater than those achieved by the increased compliance. We show how using resources to increase compliance may result in less health benefit compared with an alternative use of the resources in the context of screening for breast cancer.

    Methods Of Recruitment

    The Department of Health recommends that recruitment for breast screening should be by fixed appointment rather than open invitation.8 With a fixed appointment the letter of invitation includes a set date and time for the screening test to take place. The open invitation places the onus on the recipient to contact the screening programme to arrange their appointment. The recommendation to use the fixed appointment for breast screening is based on the appointment method having achieved a 10% higher level of compliance than the open invitation method in a randomised trial of the two appointment methods.9 However, this study ignored the costs of achieving the higher compliance. It is possible, using data from the original paper, to calculate the opportunity cost - that is, the benefit foregone - of increasing compliance by 10%. This supports the argument that, in some cases, it may not be worth attempting to increase compliance.

    Cost Of Higher Compliance

    Although the study of recruitment methods for breast screening did not present any financial cost data, the authors did reveal the numbers of screening appointment slots that were wasted due to non-attendance.9 As the proportion of costs which varies with screening activity is small,10 slot wastage is a good representation of the costs of screening. Slot wastage can then be translated into opportunity cost in the form of screening opportunities foregone.

    The table shows the cost of each recruitment method. The average cost (calculated by dividing the number of wasted slots by the number of women screened) is 0.28 and 0.026 wasted slots per woman screened for the fixed and open invitation methods respectively.

    Costs of screening for breast cancer associated with different methods of recruitment

    View this table:

    However, the important cost to quantify is the opportunity cost of increasing compliance from, in this instance, 76% to 86%. If the target population of the fixed appointment method is assumed to be 188 women then, had the open invitation method been used, 76% (143) of these women would have been screened at a cost of 3.7 slots anyway (143 x 0.026). With the fixed appointment method, the total cost of screening 162 women was 45 wasted slots. This means that 41.3 (45-3.7) slots were used to screen an extra 19 (162-143) women. If, however, these 41.3 slots could have been used to screen an alternative group of women by means of the open invitation method about 40 extra women would have been screened (41.3-(41.3 x 0.026)). Therefore, the opportunity cost of increasing compliance by 19 women using the fixed appointment method is the benefit foregone by the 40 women who lose the opportunity of a screening test had the open invitation method been used. The question which now arises is whether the health benefits that accrue to these 19 women outweigh the possible health benefits gained by the 40 women from an alternative target population.

    Comparison Of Benefits

    Whether the benefits of increased compliance are greater than its costs depends partly on the characteristics of those whose compliance is enhanced relative to women whose screening opportunity is foregone. It might be that women who need successive reminders to attend screening have characteristics that place them at increased risk of the target disease. Thus, while costs rise so might the rate of case detection.

    If the objective of a screening programme is to maximise the numbers of cases detected how may the current allocation of screening resources be best used? If an alternative target population exists then resources presently used to raise compliance from 76% to 86% might detect more cases if the target population is redefined. In the case of screening for breast cancer, the target population could be redefined by reducing the interval between screenings or by extending screening to older women. In this case study the opportunity cost is 2.1 (40/19), which means that 2.1 women from an alternative target population lost the opportunity of a screening test for every extra woman screened by increasing compliance. It follows that, for increased compliance to fulfil the objective of maximising cases detected, the rate of case detection in this marginal compliant population must be at least 2.1 times greater than that in any alternative target population. Thus, committing resources to increasing compliance may not be the most cost effective method of achieving an overall reduction in mortality and morbidity in the general population.

    Discussion

    It has been argued that low compliance in breast screening will render the screening programme ineffective.1 It is unclear from the literature at what level of compliance a screening programme should be judged successful. Is 70% sufficient or 80%? If it is 70%, what happens if a screening programme achieved a compliance level of only 69%? If 90% compliance produces a 30% fall in breast cancer,11 should a screening programme be judged a failure because it only has 45% compliance and so produces only a 15% fall in breast cancer but at half the cost? If a screening programme is judged purely in efficiency terms - that is, cost per unit of health benefit generated - then a screening programme can be judged efficient whatever the compliance rate.

    Compliance is only a proxy objective for screening. The real objective of screening might be lives saved or morbidity reduced or avoided. Setting screening objectives in term of compliance is simpler and clearer than setting more complicated objectives such as maximising life years saved, but it can lead to very different policy recommendations.12 Indeed, it could be argued that setting objectives in terms of compliance has led to breast screening not being offered to women aged 65 or more.

    Compared with the cost of screening women aged 50-64, screening of women aged 65-70 has a lower cost per life year gained and screening of women aged 71-75 has a similar cost.13 Despite this the screening programme is not extended to these women because it might be difficult to fulfil compliance objectives. It might, however, be more efficient to put screening resources into expanding the target population rather than pursuing increased compliance in younger women.

    Increasing compliance may be justified in terms of efficiency if there is good reason to suppose that the marginal benefit of the increased compliance equals or exceeds the cost. This would mean that those resources could not be redirected to another health activity that would generate superior health gain. In addition, aiming for high compliance can be justified when a screening service is in its trial stage. Clearly, in a randomised trial of screening versus no screening a significant difference is more likely to be found if compliance is maximised. When a screening service is implemented on a national basis, however, the value of the intervention should already be proved, and the need for high compliance disappears.

    It is important to couch screening objectives in terms as close as possible to the true objectives of screening.12 High compliance may seem to be an attractive screening objective because it is easily measurable, but as those running screening programmes seek to meet compliance targets they may inadvertently be denying screening resources to other populations at risk. This in turn may lower net health benefit achievable by screening. Compliance as a screening objective needs to be reassessed.

    We thank John Cairns, Gavin Mooney, and Phil Shackley for helpful comments. DJT is funded by the Wolfson Foundation, and CD is supported by the Chief Scientist Office of the Home and Health Department of the Scottish Office. The views expressed are not necessarily those of the funding bodies.

    References

    View Abstract