Clinical Investigation and Outcomes Research Health Outcomes Research Marcia A. Testa, MPH, PhD Department of Biostatistics Harvard School of Public Health 1 The Discipline of Health Outcomes Research • Scientific inquiry evaluating the results of medical interventions and health care services. • Outcomes data are direct measures of whether medical treatments are beneficial 2 How Are Outcomes Results Used? • “Outcomes research seeks to understand the end results of particular health care practices and interventions. • Used to provide information on the quality of care so that it can be improved by determining – which health care services influence the probability of optimal patient outcomes – which produce optimal improvement in the patient's physiologic status, physical function, emotional and intellectual performance and comfort (comparative effectiveness research) 1. Outcomes Research Resource Guide, 1996/97 Edition, American Medical Association 3 Measuring Outcomes • Outcomes are a function of – – – – – – baseline health status patient clinical characteristics patient demographics psychosocial characteristics treatment health care setting • The GOAL of health outcomes research analysis is to isolate the relationship between outcomes and treatment. 4 Outcome Measures • Mortality – Survival time, Death event • Morbidity – Time to an Occurrence of a Clinical Events (e.g., Stroke, MI, Cancer) • Health Status – Physical, Mental and Emotional Health Functioning • Quality of Life – Health Status as Perceived by the Individual • Patient Satisfaction – Distance between Quality of Life and Individual Expectations • Health Economic Outcomes – Cost utility, cost effectiveness 5 Hematologic Malignancies & Anemia Comparative Effectiveness Trial 6 Hb Changes Figure 2. Mean hemoglobin (Hb) change (intent-to-treat; n = 269). aP < .0001 early versus late group. Postrandomization Months 1, 2, 3, and 4 values correspond with mean Hb between Weeks 0 (baseline) to Weeks 4, 5-9, 10-14, and 15-20, respectively. a P < 0.0001 7 Treatment, Fatigue, Symptoms bP < .01 late vs. early treatment. cP < .05 late vs. early treatment. 8 Treatment and Health Status bP < .01 late vs. early treatment. cP < .05 late vs. early treatment. 9 Treatment and QOL bP < .01 late vs. early treatment. cP < .05 late vs. early treatment. 10 Measuring Outcomes Death Outcome Full Health 11 Measuring Outcomes, Measuring Performance Improving Process (how we perform) Improving Outcomes (the results of our performance) Measurement 12 The Consequences of Health Care and Medical Intervention Clinical Economic Patient Centered 13 The Consequences of Health Care and Medical Intervention –Types of Outcomes Clinical Labs, Clinical Events, Physician Assessments 14 The Consequences of Health Care and Medical Intervention –Types of Outcomes Symptom reports, health status, quality of life, patient satisfaction Patient Centered 15 Multi-faceted QOL Domains Well-being General Perceived Health Mental and Emotional Physical/ Symptoms Cognitive Work/Social 16 Outcomes Research - 1996 QOL is Recognized as Important 17 Outcome Measures Functional Health Coverage • Generic health instruments are address larger health constructs and hence their causal links to specific treatment events may be more difficult to detect • Condition-specific instruments will vary with the condition being treated, and hence are typically more sensitive to treatment effects 18 Outcome Measures Format Influencing Coverage • Fixed-Length or “Static” – number of questions is fixed – greater coverage requires greater number of questions • Dynamic instruments - use computer adaptive testing to restrict items based upon a Bayesian approach which selects the next question based upon the answer to the previous question • Combines the practicality of short form instruments with the sensitivity and target coverage of condition-specific instruments 19 Questionnaires and Surveys Generic Instruments • Some outcomes survey questions, commonly referred to as “instruments”, focus on describing how individuals rate their health overall – or generic instruments • General health surveys such as the SF-36 are now used in research studies, population surveys, and some health plans to assess patients' overall level of functioning. • Translation of SF-36 into Arabic and the translation methods references are given in the Additional Resources Section of the Website. 20 Questionnaires and Surveys Condition-Specific Instruments • Developing outcome instruments for specific diseases has been an especially prolific research area • Such instruments are more likely than general health survey measures to be able to detect changes in the disease due to treatment 21 Steps in Designing an Outcome Research Study • Define a researchable question • Develop a conceptual model • Identify the critical dependent and independent variables • Identify appropriate measures for each • Develop an analysis plan • Indicate what is believed to cause the outcome • Identify critical pathways and what other factors are likely to affect the outcome • Identify which variables (in the outcomes function equation) are relevant to your hypothesis 22 The Conceptual Health Model Environmental Behavioral, Social •Income •Social Support •Education •Health Access •Lifestyle Patient Factors •Age •Gender •Occupation Medical Interventions •Specific medications •Surgery •Diet and Exercise •Case Management Outcome Measures •Lab values •Symptoms •Functioning •Quality of life •Employment/Work 23 The Outcomes Model Patient Characteristics Risk Adjustment OUTCOME Structure (Setting) Change Change Process (Treatment) Measurement 24 Outcomes Research - 1998 Health Economic Benefits 25 Diabetes Outcomes Model Clinical Factors •Severity •Duration •Etiology •Comorbidity Patient Factors •Age •Race •Gender •Occupation Treatment •Specific medications •Diet •Exercise •Case Management Outcomes •HbA1c •Symptoms •Function •Complications •Quality of life •Employment/Work 26 Understanding Causal Pathways REGIMEN BURDEN • Pain, Life-style, discomfort HEALTH ECONOMICS EFFICACY • Productivity • Clinical 1c Treatment Satisfaction HEALTH STATUS • Self-Reported • Health Care Utilization Adherence Symptom distress QUALITY OF LIFE SIDE EFFECTS •Adverse reactions • Functional status • Physical, Mental, Cognitive, Social 27 Study Design 62 sites in the United States Glipizide GITS + diet vs Placebo + diet 16-week multicenter, randomized, double-blind, placebo-controlled, parallel titration study o o • 1-week washout o o o ooo o o • 3-week single-blind o ooooo oo oo o o o o o o oo placebo o oo o o • 4-week dose titration o o o o o o o oo • 8-week maintenance o o o o o o Testa MA, Simonson DC. JAMA 1998; 280:1490-1496. 28 Study Design Maintenance Screening / Single-blind Dose titration washout placebo 5-20 mg ■ Week -1 0 1 2 ■ 3 4 5 6 Maintenance ■ 7 8 9 10 ■ 11 12 13 ■ 14 15 Randomization to placebo & diet or Glipizide GITS & diet (1:2) Clinical and Laboratory Assessment ■ Health Economic Assessment Testa MA, Simonson DC. JAMA 1998; 280:1490-1496. 29 Multicenter, Randomized Clinical Trial Screened/Placebo Wash N=1007 Eligible for Clinical Study n=594 Diet & Placebo Inelig. QOL: n=9 Eligible QOL: n = 192 Week 4,8, n=178,169 Early Withdrawal Week 12 n=158 Completers Diet and GLipizide GITS Inelig QOL: n=16 Eligible: n = 377 Week 4,8 n=349,340 Early Withdrawal WeeK 12 n=333 Completers 30 Patient Population Baseline Clinical Characteristics Number of Patients Placebo 201 Glipizide GITS 393 Gender (M / F) (%) 60 / 40 55 / 45 Race (W / B / O) (%) 70 / 18 / 12 70 / 18 / 12 Emp / Ret / Unemp(%) 50 / 38 / 12 48 / 40 / 12 Age (yrs) 58 ± 12 59 ± 11 Duration DM (yrs) 5±5 6±6 FPG (mg / dL) HbA1c (%) 231 ± 66 8.7 ± 1.4 218 ± 62 8.5 ± 1.5 Testa MA, Simonson DC. JAMA 1998; 280:1490-1496. 31 Clinical Results End of Week 15 Placebo Glipizide GITS P-Value FPG (mg / dL) 224 ± 66 161 ± 41 < 0.001 HbA1c (%) 9.3 ± 1.9 7.5 ± 1.2 < 0.001 Hypoglycemic Symptoms (%) 4.8 6 NS Glucose < 55 mg / dL (%) 0 0 NS Testa MA, Simonson DC. JAMA 1998; 280:1490-1496. 32 HbA1c and Symptom Distress EFFICACY • HbA1c HEALTH STATUS • Self-Reported Symptom distress 33 Pharmacological Side Effects and Symptom Distress EFFICACY • HbA1c HEALTH STATUS • Self-reported Symptom distress SIDE EFFECTS •Adverse reactions 34 Mean HbA1c: 9.3 1.9% Trembling Low blood sugar reaction Fast pulse, rapid heartbeat, palpitations Heartburn Wheezing or difficulty breathing Gaining weight Cold hands or feet Feeling overweight Constipation Swelling of feet or ankles Abdominal cramps Skin rash Decrease in appetite Mood swings Losing weight Flushing, sensation of heat Shortness of breath or breathing hard Increase in hunger Inability to sleep, insomnia Pains in legs or calves Lightheadedness Lethargy, no energy to do things Numbness or tingling of hands or feet Vomiting Nauseous, queasy, sick to stomach Diarrhea Itching, scratching Heart pounding, beating hard Nightmares Headaches Impaired or worsening vision Tired, feeling weary Muscle cramps Vertigo, sensation of spinning Dizziness when standing up Tightness,pain in chest Drowsy or sleepy Cold sweat, clammy skin Sugar in urine Foot cramps, foot pain Sweating, perspiring Confusion General weakness or fatigue Numbness of lips or mouth Blurred or double vision Crabby, short-tempered Getting up often during the night to urinate Being Thirsty Having to urinate frequently Sweet taste in mouth Drinking a lot of fluids Dryness of mouth, eyes or nose High blood sugar -0.2 7.5 1.2% Hypoglycemic symptoms, weight gain, feeling overweight Symptom Worsening With Glucose Lowering P = N.S. Symptom Improvement With Glucose Lowering P < .05 P < .01 P < .001 0 0.2 0.4 SD Units Testa MA, Simonson DC. JAMA 1998; 280:1490-1496. Hyperglycemic symptoms, thirst, nocturia, blurred vision, fatigue 0.6 0.8 1 35 Impact on Quality of Life EFFICACY • HbA1c HEALTH STATUS • Self- Reported Symptom distress QUALITY OF LIFE SIDE EFFECTS • Adverse reactions • Functional status • Physical, Mental, Cognitive, Social 36 Change In QOL Scales With Therapy in Type 2 Diabetes 0.4 QOL Score (SD units) 0.3 * 0.2 *** ** ** Overall Rating Mental Health Cognitive Function Perceived Health Symptom Distress 0.1 0 * P < 0.05 ** P < 0.01 *** P < 0.001 -0.1 -0.2 -0.3 Placebo Glipizide GITS Testa MA, Simonson DC. JAMA 1998; 280:1490-1496. 37 QOL Global Outcome (Z-score) Change in HbA1c and QOL 0.2 In Patients with NIDDM Favorable QOL Response 0 * * -0.2 * * No Change in QOL No Change in HbA1c -0.4 * Unfavorable QOL Response -0.6 log-linear regression, r = .95, p < .01 -0.8 >1.5 Worsened HbA1c 1.5 to .5 .5 to -.5 -.5 to -1.5 HbA 1c (%) Change from Baseline Testa MA, Simonson DC. JAMA 1998; 280:1490-1496. <-1.5 Improved HbA1c 38 Understanding Causal Pathways REGIMEN BURDEN • Pain, Life-style, discomfort HEALTH ECONOMICS EFFICACY • Productivity • Clinical 1c Treatment Satisfaction HEALTH STATUS • Self-Reported • Health Care Utilization Adherence Symptom distress QUALITY OF LIFE SIDE EFFECTS •Adverse reactions • Functional status • Physical, Mental, Cognitive, Social 39 Change (Week 15 – Baseline) Production Losses Absenteeism Bed Days Restricted Activity Days (US $ / worker / month) (US $ / 1000 person days) (US $ / 1000 person days) $1000's 100 1000 4 50 500 2 0 0 0 P < 0.001 -50 P = 0.01 P = 0.05 -2 -500 Placebo Glipizide GITS $ Loss Placebo Glipizide GITS Placebo Glipizide GITS $ Savings Testa MA, Simonson DC. JAMA; 1998;280:1490-1496. 40 Health Care Utilization Percent of Patients 40 Percent of Patients Reporting 1 or More Non-studyrelated Ambulatory Visits (clinic, ER, physician office) per Month 40 P = NS P = 0.002 30 30 20 20 Baseline Placebo Week 15 Baseline Week 15 Glipizide GITS Estimated savings = $11 per patient per month (assuming cost of $66 per ambulatory visit) 41 Testa MA, Simonson DC. JAMA; 1998;280:1490-1496. Summary • Use Outcomes Research to – Improve the quality of health care by changing treatment and services and by promoting preventive strategies • Outcomes are “probability statements” – Multifaceted – Requires integrating and consolidating many different components of health functioning • Outcomes may take time to develop, use intermediate outcomes if necessary 42 Summary • Use Outcomes Research to – Improve the quality of health care by changing treatment and services and by promoting preventive strategies • Outcomes are “probability statements” – Multifaceted – Requires integrating and consolidating many different components of health functioning • Outcomes may take time to develop, use an intermediate outcomes 43 Additional Resources and References • Online Questionnaires: SF12, LASA, EQ-5D. • SF-36 translated into Arabic by Saud Abdulaziz bin Al Abdulmoshin, 1997 • Coons SJ, Reliability of an Arabic Version of the RAND-36 Health Survey and Its Equivalence to the US English Version • Testa and Simonson, NEJM, 1996 • Testa and Simonson, JAMA, 1998 • Straus, Testa, Sarokhan et al., Cancer 2006 44