 |
VALMER - Project Methodology |
 |
|
| |
3 HMR Outcomes Assessed: |
|
A sample of 180 reviews (18% of the total) will be assessed by four, four member panels of medication therapy experts using a modified Delphi technique. Four panels, each consisting of two pharmacists, one GP and one specialist will be assigned 90 HMRs. These will consist of a “common” group of 60 HMRs, and an additional 30 HMRs that are “panel-specific”. This will mean that there will be 16 expert opinions regarding 60 of the reviews and 4 opinions on each of a further 120 reviews. The assessments of the “common” reviews can be used to determine inter-panel reliability. By using this method, a larger sample of the HMRs can be assessed by the experts, minimising costs associated with this aspect of the project.
The assessment involves the use of a unique internet-based assessment system, which allows the assessor to access the details of the HMR remotely from the database. Assessors from throughout Australia can therefore be used, without the expense of a “bringing-them-all-together” process.
Fundamentally, the assessment requires the assignment of probabilities of multiple possible consequences, at different levels of severity, for the clinical situation before and after the HMR. The “before intervention” probability estimates can then be considered as the “counterfactual” state and the difference between the after and before probability estimates will be the estimated risk reduction. There are obviously multiple potential consequences of the recommendations made in an HMR, and our assessment process is unique in allowing multiple outcomes to be attributed to the HMR.
As an example, if aspirin was recommended for a patient with diabetes, there are several possible outcomes, each of a different level of severity, and each with a different probability before and after the addition of the aspirin. This is shown in Figure 2: Assessment of HMRs by Experts. |
| |
One outcome may be a reduced their risk of a stroke (severe) or of a less severe event such as a TIA (mild). On the other hand there may be an increased risk of severe gastrointestinal bleeding (Consequence B, severe) or simple gastrointestinal upset (Consequence B, mild). The addition of aspirin does not, however, completely remove the risk of stroke, so the assessor is asked to estimate the probability of stroke with and without the aspirin (Probability before the HMR). In this way, each combination of consequence and severity is assigned an absolute probability difference (reduction or increase), which is used to estimate the costs associated with the intervention. The cost estimates (positive and negative) are summed together for the various possibilities in an HMR and the net result is used as the estimated benefit see: Figure 3: Calculation of Total Value of HMR |
|
Our project team has previously developed and validated a table of clinical consequences at three levels of severity, each with an assigned level of disability and expense.
There is also the possibility that a particular DRP would be detected by another health professional (for example on a future visit to the General Practitioner). This will be accounted for by asking each assessor to determine the “Attribution” of each DRP being assessed. The total value of the DRP will therefore be “discounted” by the attribution. In addition, the actual uptake of each recommendation will be determined when the data is collected (see Phase 2). In this way, the realized attributable value can be calculated for each DRP, and then summed with the realized attributable value of any other DRPs identified in the HMR (up to a maximum of three). This process can be summarised as in Figure 4 (overleaf).
The parameters relating to each consequence (at each of three levels of severity) are:
| a) |
Level of impact on health status |
| b) |
Duration of impact on health status |
| c) |
Cost and Duration of Hospital Admission |
| d) |
Number and cost of GP consultations |
| e) |
Number and cost of specialist consultations |
| f) |
Cost of additional investigations |
These consequence parameters were validated in two previous projects, Med-E-Support and PROMISe II, and are based on current AR-DRG costings, MBS Schedules and the Manual of Health Resource Utilisation.
|
|
|
|
|
|
|
 |
|
 |
