|
|
Patient Panel Size Intensity Adjustment Model for Patient Aligned Care Teams
Background£º
Patients¡¯ access to healthcare is measured by the time between the patient¡¯s desirable appointment date and the actual appointment date for primary care,
which is known as appointment waiting time [1]. One of the important factors influencing appointment waiting time is the size and composition of the patient panel for a particular primary care giver or a Patient Aligned Care Team (PACT team).
In current VHA practice [2], the patient panel size is classified as 1200 patients for physicians, with adjustment calculated by the following factors: a) size of support staff (more supporting staff members, higher the panel size);
b) clinic room space; c) primary care intensity score (score of 1.0 represent the norm, a higher score indicate a sicker patient; and d) newly hired provider (gradual ramp up of panel size for new Providers).
Among these adjustment factors, the primary care intensity score is one of the most important factors and it has a large impact in the adjustment of panel size. The primary care intensity score is calculated based on VHA¡¯s internally developed model [3].
In this model, a regression based method is used. In this regression model, the dependent variable is the count of primary care visit per patient per year; the independent variables are age,
priority, race, marital status, gender, non-primary care visits, registry, insurance status, average distance traveled, and diagnosis (HCC classification). This model is statistically highly significant, but an improved model is wanted..
VHA¡¯s Office of Informatics and Analytics solicited proposals to modify and improve this model to calculate primary care work intensity scores with the following goals:
1. Modify the dependent variable to include primary care work besides clinical visits, such as telephone visit.
2. Modify the independent variables to include better predictors of the work, such as severity.
Our group, together with VA Center of Applied Systems Engineering, responded to this solicitation and won the award from VHA to fund this project.
WSU proposed approach £º
By examining the current internal regression based model [3], our group found one apparent area for improvement in the model is prediction accuracy. The difficulty of using regression models for this kind of analytical activity is that many independent variables,
such as diagnosis, are discrete variables with a large number of categories; and their relationships with the dependent variables (i.e. clinical visit counts, non clinical visit care, relative value unit (RVU)) are complicated and complex.
In this project, we adopted several advanced statistical techniques to improve our model prediction accuracy and achieved significant improvement.
We also did the following:
1) Adding more independent variables, such as can score, inpatient days, age, distance and etc; and adding more dependent variables, namely, number of new visits, relative value unit (RVU),
numbers of established visits, numbers of 0-10 minutes phone calls, and numbers of 11-20 minutes phone calls using government provided data from OIA.
2) Modifying and improving the workload intensity prediction model so that it can not only predict expected levels of dependent variables, (clinical visits, phone calls),
but also the essential measures of variation on dependent variables, such as range (min, max) based on 1 to 99 percentile, or standard deviation.
3) Developing workload models for facility level, provider level, and PACT team level, we did identified significant differences in workload among different types of facilities,
and different providers, all these models can be used to adjust work assignments to different facilities, teams and providers.
References
1. Dixon S, Sampson F, O'Cathain A, Pickin DM. (2006). "Advanced access: more than just GP waiting times?" Family practice 23(2): 233-239.
2. VHA Handbook 1101.02, Primary Care Management Module (PCMM)
3. VHA, 2010 Model for primary care workload intensity
Related funded projects
Patient Panel Size Intensity Adjustment Model for Patient Aligned Care Teams, US Department of Veteran Affairs, (PI, 100%) $74,750, 9/12 - 4/13
|
