Need to add title
This is the overall summary of the paper - the key messages so there isn’t any confusion with writing it later
This needs to get to Adam ASAP
When we started our own project on measuring instances of low value care within administrative data sets, we realised that the population we selected for the denominator had an enormous impact on our conclusions. Basically - we can set the population as the patient sample or the service sample.
This may be obvious when you consider it - we are stating we have x% of y or w% of z. However, after searching the literature for an extensive set of measured low-value services we found that each of these result types were used and that this distinction had only been briefly touched on. To make general comparisons between published results, it was necessary to categorise these results based on these distinctions.
There is something with these categories in that they are appealing or necessary for different types of decision making about clinical care.
There are three categories of results for low-value care measurement, distinguishable by the description of the denominator.
One of these categories is patient-centric, another is service-centric, and the third looks at all admissions or total low-value care.
The service-centric measure will give more insight to the level of inappropriately used services for health care decision makers. Patient-centric measures will give more insight to the level of patients receiving inappropriate care
The population in each study must be considered in order to compare reported levels of nominated low-value service use.
While the interpretation of low-value care use estimates from individual studies may be clear on their own, comparing results across studies highlights the impact that categorizing these interpretations has.
When measuring service use there are typically two subsets of observations: instances where the specific service is used, and the patients that are the ‘target’ for the service (that is, they have the indications of the service).
A second difference in sample selection is the original intent of the investigation. Low-value care measurement studies can be classed as being aimed at one of two audiences: a clinical audience interested in patient selection for services and safety, or a resource-allocation conscious audience concerned about efficiency and opportunity costs in health care.