Episodes of IllnessObjectivesExisting ApproachesNew ApproachAdvantage: Built on Existing DataA Mathematical TheoryNo ClustersNot a Measure of Treatment IntensityTerminologyTheorySlide 11Slide 12Slide 13Slide 14Slide 15Slide 16Severity of an EpisodeWhy Multiply Severity Scores?Evaluation of the TheoryConstructing Episode MeasuresResults of Test of TheoryConclusions of Pilot TestTake Home LessonEpisodes of IllnessFarrokh Alemi, [email protected] presentation trains you in using our procedures for measuring episodes of illnessBased on United States patent application 10/054,706 filed on 1/24/2002 by George Mason University. We grant permission to individual scientists within university, Federal and State governments settings to use these procedures free of licensing fees. Permission is also granted to all students using this procedure as part of an educational class.Existing ApproachesProspective Risk AdjustmentAmbulatory Visit GroupsDisease StagingProducts of Ambulatory CareAmbulatory Diagnosis GroupsAmbulatory Care Groups.New ApproachEasy to implementBuilt using Standard Query Language operations on existing data within your organizationTailored to the special populations served by your organizationDynamically changing Changing as the nature of diseases changeAdvantage: Built on Existing DataSimple database manipulations can produce the desired episodes of illness from Existing Organization’s DataCan be used within electronic health recordsWorks on any administrative database, which has information on date of visit and diagnosesA Mathematical TheoryNot a black box, shows in detail how episodes are measuredMakes it possible for researchers to build on each other’s workNo ClustersExisting approachesSchneeweiss and colleagues classified all diagnoses into 92 clusters. Otitis media infection not same as wound infection Not limited to the etiology of the diseaseAll operations are defined on individual diagnoses without need for broad clustersNot a Measure of Treatment IntensityNot intended to classify patients into homogenous resource use groups All short visits do not belong to same episode Intensity-based measures can measure if length of visit is appropriate but not if number of visits are appropriate.TerminologyEpisode of careDoes not depend on the nature of services Does not assume that temporally contiguous Anchor diagnosisTrigger diagnosisStopping point Rate of progressionPeak severityOutcomesTheoryPia= function {Tia, Sia} Probability of diagnosis i and a being part ofsame episodeTheoryPia= function {Tia, Sia} Time betweendiagnosis i and aSimilarity of diagnosis i and aTheoryPia=Sia/(1+βTia) Probability of diagnosis i and a being in same episodePia= function {Tia, Sia}TheoryPia=Sia/(1+βTia) Similarity of Diagnosis i and aPia= function {Tia, Sia}TheoryPia=Sia/(1+βTia) A constantTime between diagnosis i and aPia= function {Tia, Sia}TheoryPia=Sia/(1+βTia) Pia= function {Tia, Sia}TheoryWhen a patient presents with several diagnoses …Probability that any two of the diagnoses may belong to an episode is calculatedPair-wise probabilities are used to classify diagnosis into groupsSeverity of an EpisodeOverall severity of episode=1-пi (1-Sevi)Severity of diagnosis iWhy Multiply Severity Scores?Overall severity of episode=1-пi (1-Sevi)Symbol for multiplicationEvaluation of the Theory565 Developmentally delayed children who were enrolled in the Medicaid program of one Southeastern State Randomly sampled Included both in-patient and outpatient Medicaid payments for the patient State paid $9,296 per patient per year. The standard error of the cost was $2,238Constructing Episode MeasuresTime between two diagnoses Severity of each diagnosisSimilarity of the two diagnosesThe number of times the two diagnoses co-occur within a specific time frame Mean number of episodes was 147 (standard error = 320).Results of Test of Theory Coefficients P-value Intercept -7297 0.003 Average severity of episodes -33.58 0.000 Number of episodes 444971 0Interaction between number of episodes & severity of episodes 756 0 Regression of "Amount paid by the State" on severity and number of episodesNumber of observations = 565, Adjusted R Squared = 53.11%Conclusions of Pilot TestEpisodes of care can be constructed Explained a large percentage of variance in cost of care 53% versus typical 10%-20%Take Home LessonSimple database queries can create a measure of episodes of illness that could explain a large portion of variation in