- ACG System
- About Us
- Update / SIgnout
BACKGROUND: Primary care clinicians know that case-mix varies between different physician practices and geographical areas. Leicester City Clinical Commissioning Group (CCG) is now able to quantify those differences using routinely-collected primary care and hospital data that are grouped and processed by the Johns Hopkins Adjusted Clinical Groups (ACG®) System. Quantification of relative case-mix across a range of practices is facilitating new types of analyses of practices and performance data. However, it is important to understand the impact of data quality and completeness to ensure these analyses are robust.
METHODS AND MATERIALS: Johns Hopkins personnel worked with Leicester City CCG to validate the ACG System’s case-mix-related outputs and evaluate their usefulness to improve primary healthcare quality, equity, and performance. Two metrics were examined: secondary care costs and emergency admission rates. One concern was whether variation in coding quality and completeness negatively affects the validity of case-mix-adjusted outputs. Examining “standardized morbidity ratios” across diagnostic groups confirmed significant coding variation across the CCG’s practices. A computer model was developed to quantify the level of under-coding by practices and further adjust the case-mix-adjusted outputs to take this variation into account.
RESULTS: Results illustrate the variation between expected cost and activity levels (adjusted for case-mix) and observed levels. This allows practices to be compared with their expected levels rather than the average for the CCG. The results also showed the degree of variation attributable to coding quality and its impact on the case-mix-adjusted data. Despite coding quality issues, it was possible to identify the main outliers with regard to higher-than-expected healthcare utilization levels and to identify those practices performing better than expected.
CONCLUSIONS: The ACG® System can be used to case-mix adjust healthcare utilization levels based on morbidity to enable meaningful comparison across physician practices. A computer model was successfully developed to adjust for any local under-coding to create robust observed versus expected levels of activity and performance. Where observed costs or emergency admission rates are lower than expected in a practice, commissioners can investigate and share the learning with those practices where observed levels are higher than expected. This results in improved quality of care and lower costs. In addition, by identifying and quantifying low coding levels, practices can be supported in improving their data quality. Accurate clinical coding is increasingly important in physician records for clinical audit, care planning and decision support, as well as practice performance assessment.
To read a relevant presentation and watch a webinar on this topic click here.