Project title:

Health resource use measurement in clinical trials: should we measure general health services or disease-specific health service resource use?

Type of research:

Reanalysis of RCT data and comparison of reporting methods

Background & Scientific Rationale:

In economic evaluation alongside clinical trials, there are alternative ways of obtaining health resource use (HRU) data, and different interpretations of what health resource use data should be included. Obtaining medical records from potentially many different sources is time consuming and costly. As an alternative, one of the most popular methods of obtaining health resource use data is through participant self-reported questionnaires. The DIRUM (Database of Instruments for Resource Use Measurement) initiative has compiled a national database of such health resource use questionnaires, and highlight research gaps in what questions should be asked when measuring health resource use (HRU). One issue highlighted by DIRUM is the lack of consensus on whether all contacts with health services, or only those specific to the medical condition under research, should be measured. One option would be to ask patients only to only report contacts/medication associated with a particular condition, as this may place less burden on the patient. The issue here, however, is whether patients can differentiate as to which contacts/medication are associated with a particular condition. Many complex conditions have wide-ranging consequences the patient may not automatically associate with the primary disease. It is unclear whether patients can differentiate resource ‘specific to their condition’, particularly with multifaceted symptoms such as chronic fatigue or diabetes. The alternative is that all health care contacts/medication are often asked about, where the assumption is that randomization will ensure that non-related contacts/medication do not differ between groups. It is unclear which method will result in less missing data. No guidance exists about the merits of either method, and as a result the volume of research is fairly evenly divided between each approach.  By looking at empirical data, we hope improvements to questionnaire design could be proposed for future researchers. Ultimately, more accurate cost analysis could potentially change cost-effectiveness conclusions

Research questions / aims:

We seek to compare estimates of incremental cost and cost-effectiveness when patients are asked to report only specific condition related health care resource use to the situation when they are asked to report all health care contacts


We propose to conduct secondary analysis of primary data, namely health resource use data from a number of clinical trials conducted within Norwich CRTU and by UEA health economists. Rather than recording new data, this is a low-cost initiative to make better use of trial data which has already been collected. Through a reanalysis of resource use datasets, we will report on the following key areas which have been highlighted as under researched:

  • Definition of general and specific HRU within the context of the trial
  • Mean and total resource use for disease-specific health resource use
  • Mean and total resource use for total (general + specific) HRU
  • A comparison of total versus disease-specific health resource use
  • Estimates of cost-effectiveness from a total and disease specific perspective

We will address the issue of whether patient report questionnaires should ask about disease-specific or overall health resource use. We envisage that with empirical data to underpin our conclusions, we can make practical recommendations as to the future data collection methods for different resource use categories in different patient populations.

Expected Output of Research / Impact and added value:

  • Manuscript for publication in a peer-reviewed journal.
  • Evidence-based recommendations on general versus disease-specific measurement of health service resource use
  • Identification of future research areas

For further information on this project, contact Professor Garry Barton at

Share this article
Tweet about this on Twitter