Appraisal Standard dashboard - ASSEHS

The Appraisal Standard (AS) dashboard has been developed within Work Package 4 (WP4) of the European Project ASSEHS.

WP4 is one of the Core WPs within ASSEHS and it is responsible for the design and development of a consolidated standard for appraising stratification techniques: ASSEHS AS. The AS will facilitate critical comparisons of different risk stratification techniques.

The AS is based upon a data model whose design was inspired by a Scoping review. As a consequence, it was possible to isolate those features that help to univocally describe a risk stratification model and the scenario where it has been deployed and tested. Therefore, not only the performance (e.g. discriminative power, predictive capacity, etc.) of the models are considered but also the implementation requirements (e.g. data and health information systems' specifications), as well as the predictors (e.g. clinical, socio-demographic, pharmacy data, etc.) used to derive the risk score.

The Scoping review was based on published literature in academic journals and magazines, but also reports and white papers from commercial or governmental organizations and other unpublished, but publicly available material (so-called "grey" literature). The selected documents were read and synthesized according to the AS data model and the information stored in a database. The latter represents the data source for the AS dashboard. Therefore the information provided by the dashboard is restricted within the scope of the Scoping review.

In order to allow a more informative and homogeneous comparison, the performances belonging to risk stratification models which represent extensions of the same "reference" model were attributed to the "reference" model. In case the extension represents a significant deviance from the "reference" model in terms of scope and technical aspects, then a new entry in the "Models" data table was created (e.g. "Patients at Risk of Re-admission within 30 days" and "Patients at Risk for Re-hospitalisation 1").

The AS dashboard has several "Summary" tables reporting mean as well as minimum and maximum values of performances per model including the results achieved in different scenarios (i.e. different datasets, different predictors' subsets, etc.). As a consequence, the user has an indication of the performance' range per model according to the studies stored in the AS database. The single performance's entries can also be visualized for a deeper investigation of the specific scenario.

Predicted outcome

Risk Stratification models

Evaluation metric

Evaluation metric

Diamonds show the mean performance of the model assessed through the selected evaluation metric, while the lines range between the minimum and maximum performance

Predictors' categories

Healthcare system classification

Boehm and colleagues (Boehm et al., 2013) classified 30 OECD healthcare systems into five categories considering three dimensions (i.e. regulation, financing and service provision) and three types of actor (i.e. state, societal and private actors).

30

Healthcare systems

5

Categories

Healthcare System category

Models

Studies