The cost burden to the NHS of acute cough in pre-school children, at over £30 m, is substantial and the majority of this cost arises from consultations with general practitioners. This cost represents 1.3 million consultations each year in the UK. Parents experience some personal cost through travel and expenditure on over-the-counter preparations, and may suffer significantly if loss of earnings is experienced.
We have highlighted a number of reasons why our estimate may be conservative. A further factor is the limitation of our recruitment strategy. We were only able to recruit children who presented to primary care during normal office hours, thus children presenting initially to A&E or to out-of-hours primary care providers were not included. These children are likely to have more serious illness and may represent a greater cost burden. Also, we only included resource use from the time of the first GP consultation so any use of healthcare prior to that point will have been excluded. In addition, our cost estimate is limited by the poor data on indirect costs. Whilst we were able to estimate the direct cost to parents of time off work, the societal cost of that time is unknown. In total, the time off work due to childhood cough is substantial though this composed of a large number of short periods of absenteeism. The valuation of productivity loss is contentious and several methods have been proposed , the most appealing for pragmatic research being the friction cost method, a variation of the human capital approach, which includes only the resources required to replace the employee. Because of the disparate nature of the absenteeism here it is impossible to estimate the cost of lost productivity using the friction cost method without more detailed information on how the absenteeism was dealt with by employers. This conservative approach means that the true cost to society, including the cost of lost productivity, would also inflate the estimate. Despite this, our results are useful in indicating the cost burden of acute cough to the NHS and to parents and caregivers, and there is face validity to these estimates when compared with those of a recent study of the minimum cost of preschool asthma and wheeze . Other studies of the cost of respiratory tract infections concur. Although they address different patient groups in different settings, the studies by Ehlken  and Lambert  also suggest our estimates may be conservative.
There is very little information available to compare the cost of cough with other reasons children consult in general practice. The most common symptoms are: stomach ache, fever, rash, and diarrhoea and vomiting . Unpublished results of a current study indicate the cost to the NHS in the UK of fever in pre-school children is likely to be less than that of cough due to lower consultation rates per episode of illness and lower prevalence. Another common reason for children to use primary care is for routine immunisations. The cost of these is around £40 per child over the first five years of life, suggesting an annual cost to the NHS similar to that found here.
As recommended by the Medical Research Council , we chose a symptom (cough) rather than diagnosis or disease-based entry criterion for this study because of the inconsistency of diagnostic label use for respiratory tract infections in primary care . Our study could therefore be regarded as a 'cost of symptom' study (rather than 'cost of illness'). Nevertheless, interpretation of the results is unaffected by this, as it is the symptom rather than the 'illness' that instigates the consultation and any intervention designed to reduce consultation rates would need to target the symptom (cough). Cost-of-illness studies can be viewed as being 'incidence based' or 'prevalence based'  and each has a different value to planners of health care services. Here we have performed both types of study by estimating the cost per episode of illness and the annual cost burden to the NHS. The prevalence approach indicates the overall scale of the cost burden, allows for comparison with other areas of health care, and can inform decisions as to whether cost savings are possible and/or desirable. Although as Fleurence and Torgerson  point out, one disadvantage of cost of illness studies is the limited extent to which they can inform the research agenda, this study shows that information provided by such a study is necessary if not sufficient to evaluate the need for future research.
The incidence approach is useful from an intervention perspective, indicating the scope for cost savings as a result of different treatment alternatives. There is evidence that delayed prescribing supported by patient information leaflets can affect healthcare-seeking behaviour , and information leaflets alone can affect reconsultation rates and patient behaviour [23, 24]. Although no study has focussed on young children with cough there is an indication that such interventions could reduce antibiotic use and reconsultation rates considerably without affecting clinical outcomes . In this study, if antibiotic use and reconsultations had been 50% lower, our estimate of cost per episode would be £2.04 lower and the annual cost to the NHS reduced by £2.3 m to £29.1 m. Therefore, an intervention costing less than £2.3 m, which reduces re-consultations by 50%, would be cost saving. Furthermore, delayed prescribing can change patient attitudes and lessen the need to consult in the future thus reducing costs to the NHS even further, though possibly at the expense of parents.
This study raises several questions that could be addressed by future research. Cost of illness studies address only the cost of a disease, (or as in this case, a symptom). They do not address outcomes and do not compare the cost and outcomes of alternative strategies, as an economic evaluation would. In order to assess the full significance of the results the costs need to be considered in the light of benefits arising from the expenditure. Most of the cost is general practitioner time so the benefits of a consultation must be identified. Two main outcomes are likely to be: reassurance for parents anxious to confirm that the illness is self-limiting; and the identification of children who are seriously ill, for example, have pneumonia, an infectious illness such as whooping cough, or have asthma. The value of these benefits is unknown and subjective.
Efficiency in delivering healthcare can be addressed by improving benefits or reducing the level of resource use or both. Exploiting the skill-mix within a practice could reduce resources used in delivering primary care to young children with cough. Practice nurses, trained to identify symptoms of serious illness, could be used to provide reassurance and deliver an educational intervention designed to inform parents about the likelihood of acute cough developing into a more serious condition and the symptoms that indicate deterioration. Although there is uncertainty about the overall relative cost-effectiveness of nurses and doctors in primary care , matching tasks to skills and experience may result in more efficient use of staff resources.