Exploring adherence to clinical guidelines: application of - PowerPoint PPT Presentation

Title Exploring adherence to clinical guidelines: application of regression tree analysis in diabetes care Carinci F, Pellegrini F, Nicolucci A Monash Institute of Health Services Research, Australia Consorzio Mario Negri Sud, Italy 24 September 2001 4TH INTERNATIONAL CONFERENCE ON THE SCIENTIFIC BASIS OF HEALTH SERVICES Sydney, New South Wales, Australia

Building Evidence Based Health Systems Evidence-based Practice • Heterogeneity • Accuracy Risk Risk • Multidimensionality Outcomes Mortality Hospital length of stay Chronic disease/morbidity Physical functional status Psychosocial functioning Quality of life Resource utilization Disease and medical care complications Cost of care Predictive model building

Outcome definition Mortality Hospital length of stay Chronic disease/morbidity Physical functional status Psychosocial functioning Quality of life Resource utilization Disease and medical care complications Cost of care Predictive model building

Diabetic Complications Study OUTCOME FOR THE ORIGINAL STUDY DESIGN Cases: DIABETIC PATIENTS EXPERIENCING COMPLICATIONS • first amputation in the past 12 months or foot complications • dialysis initiated within the previous 12 months or serum creatinine levels of 3 mg/dl or more • blindness developed within the previous 12 months or proliferative retinopathy/diabetic maculopathy Controls: DIABETIC PATIENTS WITHOUT COMPLICATIONS • patients without any of the previous complications Case - Control rate 1:2 N icolucci A et al. A comprehensive assessment of the avoidability of long term complications of diabetes mellitus: a case-control study, Diabetes Care , 1996; 19 (9): 927-933

National Italian Sample Population Total Cases Controls • N (Subjects) 2774 886 1888 Type 1: 379 185 194 Type 2: 1390 212 1178 Type 2/IT: 1005 489 516 • N (Care centers) Diabetic Outpatient Clinics: 35 General Practitioners: 49

Potential Correlates PREDICTORS • Age* • Diabetes duration* • Sex • Occupation • Marital status • Hypertension • History of myocardial infarction and/or stroke • Co-morbidity • Need for help to reach the health facility • Regular attendance to visits • Compliance with dietary recommendations • Self-adjustment of insulin dose • Self-monitoring • Frequency of educational interventions • Smoke • Alcohol consumption in the past five years *global adjustment terms and tree predictors

Identifying New Outcomes Outcome: quality of care (process) • expressed as the rate of execution of the following procedures in the past 12 months (based on the expected number from the application of St. Vincent Decl.’s guidelines): • use of glycated haemoglobin • monitoring of cardiovascular risk factors (blood pressure, blood lipids) • monitoring of renal function (microalbuminuria,macroalbuminuria, serum creatinine) • eye examination • educational interventions on selected aspects of care (diet, self-monitoring of blood glucose, insulin therapy self-management, hypoglycemic emergencies, foot care, physical activity) • frequency of the educational interventions (regular vs. not regular) cut-off: 50% (lower quartile) low quality ≤ 50% N = 689 good quality >50% N = 2085

RECPAM/SAS /SAS (F.Carinci, A.Ciampi) RECPAM N cases Information Content N controls Engine Engine (Likelihood Ratio) Level 1 Level 2, Level 3, …., Level N Global Confounders + Local Confounders + Var1, Var2,…..,VarN VarN Var1, Var2,….., Determinants + ………………………… N cases (Ordinal, Nominal, N cases N controls U-Shape, Continuous ) N controls Multivariate Analysis Level 1 Level 2, Level 3, …., Level N Leaf 1 Var1, Var2,…..,VarN VarN Var1, Var2,….., Generalized Linear Models (PROC GENMOD) Logistic Regression N cases N cases Poisson Regression N controls N controls Linear Regression GEE Models Leaf 2 Leaf 3 Cox Regression (PROC PHREG) Proportional Hazards Models N cases Multiple Failure Models N controls N cases N controls RECPAM Class I RECPAM Final Model Backward Elimination class II

Quality-based stratification tree + statistical model Type 2 Insulin-Treated 254 Diabetes Mellitus 743 General Practitioner (N=997) Diabetes Outpatient Clinic Setting of Care 95 159 76 667 Low High Class A ( ≤ 6 ) ( ≤ 6 ) ( ≤ 6 ) ( ≤ 6 ) ( >6 ) ( >6 ) ( >6 ) ( >6 ) Socio-economic Status Final Model (Logistic Regression) 126 33 475 192 Variable 95% C.I. OR Class B Class C Not regular visits (R.C.= regular) 2.57 1.80 3.68 Age 1.01 0.99 1.02 Diabetes Duration 0.99 0.97 1.01 Cardiovascular Risk (R.C. = no) 1.49 1.02 2.17 Class A (R.C.= Class C) 5.05 3.05 8.36 Class B (R.C. = Class C) 1.59 1.04 2.45

Disease Management Systems Analysis TYPE 1 TYPE 2/IT TYPE 2/NIT (N=376) (N=997) (N=1371) Bootstrap % Bootstrap % Bootstrap % Variable OR 95% C.I. OR 95% C.I. OR 95% C.I. OR 95% C.I. OR 95% C.I. OR 95% C.I. Regular attend. visits (R.C.=Y) 2.75 1.08 7.01 2.80 1.06 7.30 2.57 1.80 3.68 2.58 1.78 3. 73 1.69 1.25 2.29 1.69 1.24 2.29 Age 1.07 1.03 1.10 1.07 1.03 1.11 1.01 0.99 1.02 1.01 0.99 1.02 1.03 1.02 1.05 1.03 1.02 1.05 Diabetes Duration 0.96 0.92 1.00 0.95 0.90 1.00 0.99 0.97 1.01 0.99 0.97 1.00 0.99 0.97 1.01 0.99 0.97 1.01 Previous Cardiov. Dis, (R.C.= N) - - - - - - 1.49 1.02 2.17 1.49 1.01 2.14 - - - - - - Gender (R.C.= female) - - - - - - - - - - - - 1.39 1.08 1.79 1.39 1.07 1.81 Class A 14.81 4.20 52.28 16.58 4.45 79.84 5.05 3.05 8.36 5.14 3.08 8.76 5.49 3.33 9.05 5.60 3.41 9.64 Class B 3.95 1.37 11.43 4.20 1.33 15.41 1.59 1.04 2.45 1.60 1.06 2.53 1.89 1.19 3.01 1.92 1.23 3.21 Class C 1.00 - - - - - 1.59 1.00 - - - - 1.00 - - - - - 29 254 393 General Diabetes 347 743 Diabetes 978 Diabetes General General Practitioner Outpatient Outpatient Outpatient Practitioner Practitioner Clinic Clinic Setting of Care Clinic Setting of Care Setting of Care 11 18 95 173 220 159 25 322 76 667 161 817 Low High Yes No High Class A Low Class A Class A ( < 6 ) ( ≥ 6 ) Comorbidity ( >6 ) SES ( ≤ ≤ 6 ) ≤ ≤ SES Percentages 11 7 126 33 195 25 (outcome and selected 107 215 475 192 157 660 sub-components) Class B Class C Class B Class C Class B Class C Overall Quality Score ( ≤50 % ≤50 % )……… ≤50 % ≤50 % Glycated Haemoglobin (No)………... Triglicerides, Cholesterol, BP ( ≤50 %) ≤50 %) ≤50 %) ≤50 %) Eye Examination (No)……………..….. Renal Function (None or One)…….... Educational Interventions ( ≤50 % ≤50 % )…… ≤50 % ≤50 % Regular Educational Interv. (No)….....

Evidence-based practice ? Questions exploratory/validation studies New (existing) Databases POPULATION (DIABETICS) Diabetes Modelling tool Outpatient Clinic (RECPAM) 4 5 2 3 4 7 High Setting of Care 9 5 1 ( >6 ) ( >6 ) ( >6 ) ( >6 ) 7 General 6 6 Practitioner Diabetes SES 3 Outpatient 3 Classification Low 2 3 9 9 5 Clinic ( < 6 ) ( < 6 ) ( < 6 ) ( < 6 ) 1 3 9 8 7 6 C 9 7 s s a Setting of Care 0 6 l A C 2 High 2 2 s 1 s 7 a 5 General 1 ( >6 ) ( >6 ) l 7 ( >6 ) ( >6 ) C 8 4 Practitioner Diabetes Outpatient B s s 5 a SES 2 l Clinics C Low 9 7 2 5 3 1 7 ( ≤ 6 ) ( ≤ 6 ) ( ≤ 6 ) ( ≤ 6 ) 7 1 4 C 3 1 6 s 1 s Setting of Care a 8 l 1 5 C A 9 General 1 s 2 No Practitioner s 2 a 0 3 l 6 C 6 B s Comorbidity s a l 7 C Type 1 Yes 5 1 1 2 1 C Class I ..... 5 s 2 s a l C A 1 s 1 Class II .... s a l 7 C 0 1 Type 2 B s s a l C Class I ..... Class II .... Population/Administrative Data Clinical Practice ?? Clinical guidelines