Comparative Effectiveness 

Most medical evidence is derived from randomized controlled trial (RCT). However, there is growing recognition that most questions relevant to everyday clinical practice are not addressed by RCTs and are unlikely to be addressed in the future. Because of the cost and time required to conduct RCTs, they are unlikely to be widely used to address the many questions that arise in routine inpatient and outpatient care. A major interest of the CHR is in exploring other means of creating evidence that is both timely and directly relevant to clinical care. In particular,  electronic health record (EHR) and other sources (e.g., claims, billing, self reported) of data offer the potential to evaluate the comparative effectiveness (CE) of dianostics, interventions, and prediction models on outcomes.

Effectiveness studies based on EHR data reflect how care is actually delivered to all patients, regardless of co-morbidities or other possible risks. CE studies can be used to complement and bridge gaps in evidence that will not be addressed by RCTs. Geisinger investigators are conducting observational studies to answer questions directly relevant to the needs of patients and clinicians. Investigators at the CHR are actively engaged in both testing and validating predictive models and in CE research.  For example, research is currently underway to compare the effectiveness of the long-term use of different antihypertensive therapies on risk of stroke and coronary artery disease. 

Predicting Heart Failure (HF) Diagnosis in Primary Care

PI: Walter (Buzz) Stewart, PhD, MPH

HF is a common, severely disabling disease. It is the most costly disease for CMS. HF is usually detected too late by primary care physicians to change the natural history of the disease, to prevent its occurrence, or to substantially slow progression. To address this gap, we used longitudinal electronic health record data (EHR) on primary care patients to determine if it was possible to detect HF 12 to 30 months before it is usually diagnosed. In logistic regression modeling, selected diagnoses (e.g., diabetes, AF, PVD, hypertension), use of anti-hypertension medications, and lab measures (i.e., HDL, BUN) predicted diagnosis of HF.  The area under the curve for the receiver operator curve for the 6 to 18 month (i.e., prediction window before diagnosis) was validated at 0.80.  The model results offer an opportunity to implement an early HF detection program in primary care

Bridging Knowledge Gaps in the Comparative Effectiveness of ACE

PI: Nirav Shah, MD, MPH

Inhibitors & Angiotensin II Receptor Blockers (ARBs)

Evidence on the comparative effectiveness of Angiotensin Converting Enzyme Inhibitors (ACEIs) and Angiotensin II Receptor Blockers (ARBs) leaves much to be desired for specific populations and outcomes for which no data exist. We propose to extend known results from randomized clinical trials related to ACEIs and ARBs, to fill gaps in evidence for the large and diverse patient populations with hypertension (including Medicare and Medicaid recipients) who would not have been eligible for an ACEI or ARB clinical trial. Using a retrospective cohort study design with data from electronic health records, our study objectives are to determine if long-term (> 1 year) use of ACEIs or ARBs is related to differences in rates of death, new-onset diabetes, stroke, or coronary artery disease (CAD) among demographically- and comorbidity-defined subgroups.

Couples Colorectal Cancer Screening Study      

PI: Joseph Boscarino, PhD

This study is a randomized trial that seeks to develop a couple-based intervention whose goal is to increase uptake of colorectal cancer screening (CRCS) among couples in which neither partner is on-schedule with regard to CRCS. This intervention is based on the Interdependence Model, which is a dyad-level social psychological theory proposed to understand the interpersonal context of social situations. Based on the Interdependence Model, the goal of this intervention is to transform partners’ motivation for CRCS to a couple-focused motivation. This study is being conducted in collaboration with Fox Chase Cancer Center. This study is funded by the National Cancer Institute.

Garden Gang

PI: Margaret Rukstalis, MD

Geisinger “Garden Gang” is a National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) R34 grant to conduct a pediatric overweight/obesity intervention in a two-group randomized clinical trial in primary care.  Participants will be 60 children ages 4-8 who are obese or at-risk for obesity (> 85th % BMI for age, sex) and their caregiver. The children will be randomized to one of two conditions: group Family-based Behavioral Modification (FBM; N= 30) sessions or Minimal Nutrition Information (MNI; N=30). FBM includes primary care group meetings where parents receive skills training to reduce sedentary behavior, increase physical activity, reduce energy-dense and increase nutrient-dense food consumption. MNI includes mailed handouts on nutrition and exercise. Primary outcomes measures will be 5-month changes in child weight status, diet and physical activity, and cardiovascular outcomes (for effect size estimation), and measures of treatment acceptability by primary care staff and families. This study will provide investigators pilot data necessary to map a larger-scale multi-site intervention study, via an R18 funding mechanism.  

 Patient-Reported Outcomes in Urology

 PI: Jove Graham, PhD

A large number of self-reporting urinary questionnaires exist, several of which have been validated in clinical literature for measuring and tracking symptom severity. Different questionnaires, however, are rarely compared directly against each other to assist a physician in choosing the best one for a particular clinical need. We have directly compared several clinical instruments:  the IPSS, UDI-6, MESA and a score developed by Geisinger and previously validated for population-based studies, the Bladder Health Survey (BHS).  We analyzed 628 patients in a prospective surgical registry to compare the scores’ ability to distinguish between patients with mild symptoms who differ by severity. BHS and IPSS scores had similar distributions, and were less skewed than MESA or UDI-6, suggesting they may be better at detecting change and distinguishing between specialty clinic patients with mild symptoms. BHS may also be more responsive than IPSS to incontinence symptoms as captured by MESA.  More comparisons of different scales are needed to gauge their relative responsiveness and help physicians choose the best tools.

Converting Between Versions of Erectile Function Symptom Scores

PI:Jove Graham, PhD

It can be difficult for a physician to compare symptoms between patients or studies if different scoring questionnaires are used for each.  The 15-item International Index for Erectile Function (IIEF) and a 5-item short form version (IIEF-5) have both been validated for measuring erectile dysfunction symptoms.  We performed a retrospective analysis of 396 paired responses from male patients to test for a significant correlation between scores and find a conversion factor for calculating one score from the other.  Results:  IIEF scores correlated very strongly with IIEF-5 score (r = 0.91), and the IIEF score could be predicted very accurately by multiplying the short form score by 2.84 and rounding.  Cross-validation showed that using this formula accurately predicted IIEF score (on a scale of 5-75) within 5 points for over 50% of patients and within 16 points for over 90% of patients.  Despite the fact that one questionnaire covers additional domains (i.e., sexual desire) that the other does not, our results suggest that a clinician can infer with reasonable certainty what a patient’s IIEF score would be by administering only the 5-question form.

The Adolescent Weight Management Study

PI: Margaret Rukstalis, MD

The Adolescent Weight Management study, funded by the PA Department of Health, is a randomized clinical trial comparing two deliveries of a primary care manual guided Family Based Lifestyle Modification (FBM) intervention. This study is being conducted at two sites, The Children’s Hospital of Pennsylvania in Philadelphia and Geisinger Clinic in Danville, PA. The primary aim of the study is to assess the relative effectiveness of the group versus self-guided interventions to reduce body mass index (BMI) for the sample as a whole and within the two target groups (urban African-American and rural Caucasians). Participants (N=158) are adolescents (13-17 years of age) with a BMI of 28 or greater kg/m2 and a caregiver. The self-guided intervention consists of six meetings. The group intervention consists of 17 multi-family group meetings plus six adolescent-parent-interventionist sessions. Secondary analyses will evaluate changes in risk factors for cardiovascular disease and diabetes, including lipids (total-, HDL-, LDL-cholesterol and triglycerides), glucose and insulin, waist circumference, blood pressure, diet and activity.

Comparative Effectiveness of Common 1st and 2nd Line Antihypertensive Therapies In Moderating Hypertension-related Changes in Renal Function  

PI: Jove Graham, PhD

This is an observational study using electronic health record (EHR) data to analyze creatinine and glomerular filtration rate (GFR) outcomes from different antihypertensive medications used as first-line single-drug and two-drug therapy.

Decreasing Long Term Complications and Cost Following Hip Fracture using a Medical Home Concept (MHC)

Co-PI: Jove Graham, PhD

This is a prospective, non-randomized controlled study to compare rates of hospital readmissions, healthcare costs, and serious adverse events between hip fracture patients whose post-discharge chronic care is managed using a patient-centered medical home model versus standard care.

Expanding Comparative Effectiveness Research in Orthopedics by Capturing Uniform Measures of Patient-Reported Functional Outcomes at Two Institutions

Site PI: Jove Graham, PhD

This is a prospective study to design and implement an electronic infrastructure for collecting functional outcome data from patients with osteoarthritis of the knee using a uniform set of questionnaires and touchscreen computers at Geisinger and NYU. This work will support future funding proposals to establish a knee replacement registry at the two institutions.

Treatment Seeking & Treatment Effectiveness following World Trade Center Disaster (WTCD)

PI:Joseph Boscarino, PhD

The purpose of this study is to access to mental health care, delayed treatment seeking, and the effectiveness of brief mental health treatments following the WTCD attack in New York City. The goal of this study is to identify risk-prediction models for emergency mental interventions following traumatic event exposures. This study is based on a prospective cohort study of approximately 2,400 adults following exposure to the WTCD conducted by the PI from 2002-2006. The plan is to use these findings to develop new funding opportunities related to PTSD research. Funded by the PA Department of Health.

Chronic Hepatitis B & C Longitudinal Cohort Study

PI: Joseph Boscarino, PhD

This is a 5-year, multi-site longitudinal prospective study of chronic hepatitis B & C infection among patients seen within five sites of the National HMO Research Network. The purpose of this study is to study the epidemiology and health outcomes associated with chronic hepatitis infection. This study represents the largest the largest study ever done in this clinical area. This study is funded by the Centers for Disease Control Foundation.

Comparative Effectiveness of Health Care Delivery Models for Reducing the Burden of Prostate Cancer

Site PI: Porat Erlich, PhD, MSc

Prostate cancer screening is widely disseminated but there is currently a lack of population-based information on appropriate utilization by type of health delivery system. Geisinger Health Plan (GHP) insurance allows for managed care within the Geisinger Clinic (GC) or care through other physician systems for the GHP-insured. While GC ambulatory care is managed under defined expectations for prostate cancer screening, treatment and surveillance, care from non-GC providers is typically not centrally managed. In this study, we evaluate the comparative effectiveness of managed versus non-managed ambulatory care for controlling the burden of prostate cancer.

PREDICTING PROGRESSION OF CHRONIC KIDNEY DISEASE

Site PI: Walter (Buzz) Stewart, PhD, MPH

The primary objective of this proposed study is to describe the natural history of chronic kidney disease (CKD) progression and to develop a predictive model for CKD progression. The focus of the project is on Stage 3 CKD, and related transition rates from Stage 3 to more severe stages among primary care patients and to identify risk factors associated with each stage of transition. A retrospective cohort study describes the natural history of CKD progression. The nested case-control design within the cohort of primary care patients develops a predictive model for onset and progression of CKD.  We are using longitudinal electronic health record (EHR) data from the Geisinger Clinic for analysis.

Randomized Trial of Behavioral Economic Interventions to Reduce CVD Risk

Site PI: Walter (Buzz) Stewart, PhD, MPH

Comparative effectiveness research (CER) has identified many approaches to reduce the risk for many pervasive health conditions, including cardiovascular disease (CVD).  However, the uptake of practices known to be successful is significantly lower than what is needed to optimize the health of the American population.  This project proposes to test a novel strategy – financial incentives – to improve uptake of CER findings among physicians and patients to reduce the risk for cardiovascular disease