Timothy P. Hofer

When Do Patients and Their Physicians Agree on Diabetes Treatment Goals and Strategies, and What Difference Does It Make?

BACKGROUND: For patients with chronic illnesses, it is hypothesized that effective patient-provider collaboration contributes to improved patient self-care by promoting greater agreement on patient-specific treatment goals and strategies. However, this hypothesis has not been tested in actual encounters of patients with their own physicians.OBJECTIVE: To assess the extent to which patients with type 2 diabetes agree with their primary care providers (PCPs) on diabetes treatment goals and strategies, the factors that predict agreement, and whether greater agreement is associated with better patient self-management of diabetes.DESIGN: One hundred twenty-seven pairs of patients and their PCPs in two health systems were surveyed about their top 3 diabetes treatment goals (desired outcomes) and strategies to meet those goals. Using several measures to evaluate agreement, we explored whether patient characteristics, such as education and attitudes toward treatment, and patient-provider interaction styles, such as shared decision making, were associated with greater agreement on treatment goals and strategies. We then examined whether agreement was associated with higher patient assessments of their diabetes care self-efficacy and self-management.RESULTS: Overall, agreement on top treatment goals and strategies was low (all κ were less than 0.40). In multivariable analyses, however, patients with more education, greater belief in the efficacy of their diabetes treatment, and who shared in treatment decision making with their providers were more likely to agree with their providers on treatment goals or strategies. Similarly, physician reports of having discussed more content areas of diabetes self-care were associated with greater agreement on treatment strategies. In turn, greater agreement on treatment goals and strategies was associated both with higher patient diabetes care self-efficacy and assessments of their diabetes self-management.CONCLUSION: Although patients and their PCPs in general had poor agreement on goals and strategies for managing diabetes, agreement was associated with higher patient self-efficacy and assessments of their diabetes self-management. This supports the hypothesis that enhancing patient-provider agreement on both overall treatment goals and specific strategies to meet these goals may lead to improved patient outcomes.

Estimated Benefits of Glycemic Control in Microvascular Complications in Type 2 Diabetes

Blindness and renal failure are two of the most feared complications of diabetes. These devastating outcomes frequently result from insidious progression of abnormalities in small blood vessels (microvascular disease) over many years. Although microvascular disease is associated with the level of glycemic control [1-3], the Diabetes Control and Complications Trial (DCCT) was the first study to convincingly show that in patients with type 1 diabetes, improved glycemic control can reduce the risk for early microvascular complications [4]. The DCCT settled many issues, but it also raised difficult questions about the management of diabetes. There is now nearly uniform agreement that intensive glycemic control should be attempted for most patients with type 1 diabetes. However, the implications of the DCCT for the treatment of type 2 diabetes remain controversial [5-8]. Cohort studies suggest that early microvascular disease is related to hemoglobin A1c level in both type 1 and type 2 diabetes [3, 9], but the incidence of end-stage complications is much lower in type 2 diabetes [10, 11], presumably because of the older age at onset and competing risks for death. Thus, the benefits of intensive therapy in type 2 diabetes seem less compelling. The perceived difficulty of treating patients with type 2 diabetes, the potential harms related to macrovascular complications, and concern about hypoglycemia further fuel the controversy. These concerns are important because roughly 90% of diabetic patients have type 2 disease. Although long-term benefit may result from improved glycemic control, economic costs may be increased [12]. For many patients, insulin injections, frequent laboratory monitoring, increased office visits, more restrictive diets, and intensive at-home monitoring of glucose levels are required. Thus, it is critical that the possible long-term benefits of aggressive glycemic control in type 2 diabetes be better quantified [7, 13]. Such information may facilitate the counseling of patients and could help health care systems prioritize and focus costly clinical interventions. We therefore created a model to calculate the risks for developing blindness and end-stage renal disease for patients at different ages of diabetes onset and levels of glycemic control. The base-case analysis used data from the DCCT for rates of early disease [4, 14] and used cohort data from patients with type 2 diabetes for rates of subsequent progression to end-stage disease [10, 11, 15, 16]. Methods Markov Model A Markov model was constructed to analyze the risk for retinopathy and nephropathy in patients with type 2 diabetes. The structure of the model is shown in the (Figure 1). Estimates for the two complications were modeled separately so that they were assumed not to interact. The simulated patients progressed sequentially through increasingly severe disease states; death could occur in any disease state (Figure 1). Incidence and progression of retinopathy were defined as in the DCCT [4, 17]. Blindness was defined as corrected visual acuity of 20/200 or worse and was restricted to blindness caused by retinopathy or its sequelae (for example, macular edema). Microalbuminuria was defined as an albumin level of 30 to 300 mg/g of creatinine, proteinuria was defined as a protein level greater than 300 mg/g of creatinine, and end-stage renal disease was defined as renal disease that required dialysis or transplantation. Figure 1. Structure of the model. Amputation was not modeled because of a lack of evidence on the relative contributions of microvascular and macrovascular disease to the risk for amputation. Moreover, the same patients who were identified in our analyses as being at high risk for retinal and renal complications (and who therefore will benefit more from intensified glycemic control) will have a similarly high risk for neuropathy and its associated outcomes [18]. Our study addressed clinical risks and benefits associated with glycemic control and did not directly evaluate costs. The costs of decreasing hemoglobin A1c levels are not well defined for patients with type 2 diabetes because glucose control can be improved by many methods. Each method has different implications for the costs to the patient, payers, and society. However, we present results in the form of expectation functions so that if the costs and effectiveness of a specific intervention are known, estimates of the cost per complication prevented can be calculated from our tables. Assumptions The construction of the model was based on the states defined in the DCCT, which provided estimates for the rates of early microvascular disease. Our review of the literature and input from diabetes experts consistently identified the Rochester, Minnesota, cohort study of diabetic patients and the Wisconsin Epidemiologic Study of Diabetic Retinopathy [3, 9-11, 15] as the best studies with which to provide estimates of the rates of progression to end-stage outcomes; thus, these studies were used in the base-case analysis. Further literature was identified by searching the MEDLINE database search with the keyword diabetes, cross-referenced with retinopathy, blindness, visual loss, nephropathy, kidney disease and failure, renal disease and failure, blood glucose, glycemic control, mortality, and hemoglobin A1c. We reviewed the abstracts of the identified articles and obtained the articles that were relevant to the model structure. Additional literature was identified by review of references in these articles and through discussion with national experts in the field. Patients were assumed to have no clinically detectable microvascular complications at the time of diagnosis of diabetes. Although up to 20% of patients have retinopathy at the time of diagnosis [19], we excluded this subgroup because data on the distribution of these patients across hemoglobin A1c levels are lacking. In addition, patients who present with complications have already declared themselves to be at high risk and thus should be considered for intensive control. We assumed that incidence and early progression of retinopathy and development of microalbuminuria were related to level of glycemic control at the rates of progression shown in the DCCT [4, 14]. Each 10% increase in hemoglobin A1c level is accompanied by a 20% increase in the rate of developing microalbuminuria, a 56% increase in the rate of developing retinopathy, and a 64% increase in the rate of progression of retinopathy [14]. Further progression (to renal disease beyond microalbuminuria and from retinopathy to blindness) was assumed not to be related to level of glycemic control [20-25]. Data for sensitivity analyses were derived from the 95% CIs in the DCCT and the range of estimates available in the literature (Table 1). Table 1. Model Components for the Base-Case and Sensitivity Analyses: Annual Rates of Transition between Disease States* Mortality estimates were based on U.S. data from the Department of Vital Statistics [40]. These estimates were modified to reflect the higher mortality rates in patients with type 2 diabetes [41-47]. Mortality was not adjusted for level of glycemic control. Some cohort studies show a relation between level of glycemic control and mortality, but it is not clear whether a causal relation exists [48]. Nephropathy, even at early stages, has been reported to be associated with increased rates of death, particularly death from cardiovascular disease [37, 49-52]. Much of this association may be due to confounding, but the effect persists after adjustment according to available severity and comorbidity measures [50]. The base-case model assumes that early nephropathy does not increase mortality rates. However, to evaluate the potential importance of this factor, we compared the mortality benefits of improved glycemic control by using half of or the full observed association between early renal disease and increased mortality rates. Model Structure and Implementation The model was run independently for each age of diabetes onset and level of glycemic control and was implemented by creating matrices of the probabilities of going from one health state to the next during a 1-year period (Figure 1). The percentage of patients developing each level of complication was tabulated annually, continuing through 90 years of age. Mortality rates were updated at 5-year intervals. This technique represents a standard implementation of a nonstationary Markov process [53, 54]. Average life expectancy was calculated by multiplying the proportion of patients dying during each year by the total number of years survived. Average time spent in each disease state was determined by dividing the total time spent in a state by the number of patients who develop that state. All calculations and modeling were performed by using Stata for Windows (Stata Corp., College Station, Texas). Sensitivity analyses were conducted by using varying estimates at all transition points (Table 1) [54]. We included the full range of estimates in the literature in one-way sensitivity analyses. However, because some studies had extreme values for the transition probabilities (most likely because of small sample sizes or atypical patient populations), three-way sensitivity analyses were conducted by using the midpoint between the base-case and extreme estimates. We also tested the validity of the model by modifying the simulated patients to fit the characteristics of the populations of various actual studies of natural history (for example, the mean values for age, hemoglobin A1c level, and initial prevalence of disease found in these studies were used to generate a limited set of results) and by comparing the predictions of the model (within one-way sensitivity analysis range) with the observed values from these studies. Using the model predictions and the characteristics of a population of patients with type 2 diabetes at a large staff-model health ma

Which colon cancer screening test? A comparison of costs, effectiveness, and compliance

PURPOSE Recent media reports have advocated the use of colonoscopy for colorectal cancer screening. However, colonoscopy is expensive compared with other screening modalities, such as fecal occult blood testing and flexible sigmoidoscopy. We sought to determine the cost effectiveness of different screening strategies for colorectal cancer at levels of compliance likely to be achieved in clinical practice. METHODS A Markov decision model was used to examine screening strategies, including fecal occult blood testing alone, fecal occult blood testing combined with flexible sigmoidoscopy, flexible sigmoidoscopy alone, and colonoscopy. The timing and frequency of screening was varied to assess optimal screening intervals. Sensitivity analyses were conducted to assess the factors that have the greatest effect on the cost effectiveness of screening. RESULTS All strategies are cost effective versus no screening, at less than

Who bounces back? Physiologic and other predictors of intensive care unit readmission.

20,000 per life-year saved. Direct comparison suggests that the most effective strategies are twice-lifetime colonoscopy and flexible sigmoidoscopy combined with fecal occult blood testing. Assuming perfect compliance, flexible sigmoidoscopy combined with fecal occult blood testing is slightly more effective than twice-lifetime colonoscopy (at ages 50 and 60 years) but is substantially more expensive, with an incremental cost effectiveness of

An epistemology of patient safety research: a framework for study design and interpretation. Part 2. Study design

390,000 per additional life-year saved. However, compliance with primary screening tests and colonoscopic follow-up for polyps affect screening decisions. Colonoscopy at ages 50 and 60 years is the preferred test regardless of compliance with the primary screening test. However, if follow-up colonoscopy for polyps is less than 75%, then even once-lifetime colonoscopy is preferred over most combinations of flexible sigmoidoscopy and fecal occult blood testing. Costs of colonoscopy and proportion of cancer arising from polyps also affect cost effectiveness. CONCLUSIONS Colonoscopic screening for colorectal cancer appears preferable to current screening recommendations. Screening recommendations should be tailored to the compliance levels achievable in different practice settings.

Preventing Hospital-Acquired Urinary Tract Infection in the United States: A National Study

ObjectiveTo determine the influence of changes in acute physiology scores (APS) and other patient characteristics on predicting intensive care unit (ICU) readmission. DesignSecondary analysis of a prospective cohort study. SettingSingle large university medical intensive care unit. PatientsA total of 4,684 consecutive admissions from Janu-ary 1, 1994, to April 1, 1998, to the medical ICU. InterventionsNone. Measurements and Main Results The independent influence of patient characteristics, including daily APS, admission diagnosis, treatment status, and admission location, on ICU readmission was evaluated using logistic regression. After accounting for first ICU admission deaths, 3,310 patients were “at-risk” for ICU readmission and 317 were readmitted (9.6%). Hospital mortality was five times higher (43% vs. 8%;p < .0001), and length of stay was two times longer (16 ± 16 vs. 32 ± 28 days;p < .001) in readmitted patients. Mean discharge APS was significantly higher in the readmitted group compared with the not readmitted group (43 ± 19 vs. 34 ± 18;p > .01). Significant independent predictors of ICU readmission included discharge APS >40 (odds ratio [OR] 2.1; 95% confidence interval [CI] 1.6–2.7;p < .0001), admission to the ICU from a general medicine ward (Floor) (OR 1.9; 95% CI 1.4–2.6;p < .0001), and transfer to the ICU from other hospital (Transfer) (OR 1.7; 95% CI 1.3–2.3;p < .01). The overall model calibration and discrimination were (H-L &khgr;2 = 3.8, df = 8;p = .85) and (receiver operating characteristic 0.67), respectively. ConclusionsPatients readmitted to medical ICUs have significantly higher hospital lengths of stay and mortality. ICU readmissions may be more common among patients who respond poorly to treatment as measured by increased severity of illness at first ICU discharge and failure of prior therapy at another hospital or on a general medicine unit. Tertiary care ICUs may have higher than expected readmission rates and mortalities, even when accounting for severity of illness, if they care for significant numbers of transferred patients.

When More Is Not Better Treatment Intensification Among Hypertensive Patients With Poor Medication Adherence

This is the second in a four-part series of articles detailing the epistemology of patient safety research. This article concentrates on issues of study design. It first considers the range of designs that may be used in the evaluation of patient safety interventions, highlighting the circumstances in which each is appropriate. The paper then provides details about an innovative study design, the stepped wedge, which may be particularly appropriate in the context of patient safety interventions, since these are expected to do more good than harm. The unit of allocation in patient safety research is also considered, since many interventions need to be delivered at cluster or service level. The paper also discusses the need to ensure the masking of patients, caregivers, observers and analysts wherever possible to minimise information biases and the Hawthorne effect. The difficulties associated with masking in patient safety research are described and suggestions given on how these can be ameliorated. The paper finally considers the role of study design in increasing confidence in the generalisability of study results over time and place. The extent to which findings can be generalised over time and place should be considered as part of an evaluation, for example by undertaking qualitative or quantitative measures of fidelity, attitudes or subgroup effects.

Projections of demand and capacity for colonoscopy related to increasing rates of colorectal cancer screening in the United States.

BACKGROUND Although urinary tract infection (UTI) is the most common hospital-acquired infection in the United States, to our knowledge, no national data exist describing what hospitals in the United States are doing to prevent this patient safety problem. We conducted a national study to examine the current practices used by hospitals to prevent hospital-acquired UTI. METHODS We mailed written surveys to infection control coordinators at a national random sample of nonfederal US hospitals with an intensive care unit and >or=50 hospital beds (n=600) and to all Veterans Affairs (VA) hospitals (n=119). The survey asked about practices to prevent hospital-acquired UTI and other device-associated infections. RESULTS The response rate was 72%. Overall, 56% of hospitals did not have a system for monitoring which patients had urinary catheters placed, and 74% did not monitor catheter duration. Thirty percent of hospitals reported regularly using antimicrobial urinary catheters and portable bladder scanners; 14% used condom catheters, and 9% used catheter reminders. VA hospitals were more likely than non-VA hospitals to use portable bladder scanners (49% vs. 29%; P=.001), condom catheters (46% vs. 12%; P=.001), and suprapubic catheters (22% vs. 9%; P=.001); non-VA hospitals were more likely to use antimicrobial urinary catheters (30% vs. 14%; P=.001). CONCLUSIONS Despite the strong link between urinary catheters and subsequent UTI, we found no strategy that appeared to be widely used to prevent hospital-acquired UTI. The most commonly used practices--bladder ultrasound and antimicrobial catheters--were each used in fewer than one-third of hospitals, and urinary catheter reminders, which have proven benefits, were used in <10% of US hospitals.

The Role of Clinical Uncertainty in Treatment Decisions for Diabetic Patients with Uncontrolled Blood Pressure

Background— Hypertension may be poorly controlled because patients do not take their medications (poor adherence) or because providers do not increase medication when appropriate (lack of medication intensification, or “clinical inertia”). We examined the prevalence of and relationship between patient adherence and provider treatment intensification. Methods and Results— We used a retrospective cohort study of hypertensive patients who had filled prescriptions for 1 or more blood pressure (BP) medications at Veterans’ Affairs (VA) healthcare facilities in a Midwestern VA administrative region. Our sample included all patients who received at least 2 outpatient BP medication refills during 2004 and had 1 or more outpatient primary care visits with an elevated systolic BP >140 but <200 mm Hg or diastolic BP >90 mm Hg during 2005 (n=38 327). For each episode of elevated BP during 2005 (68 610 events), we used electronic pharmacy refill data to examine patients’ BP medication adherence over the prior 12 months and whether providers increased doses or added BP medications (“intensification”). Multivariate analyses accounted for the clustering of elevated BP events within patients and adjusted for patient age, comorbidities, number of BP medications, encounter systolic BP, and average systolic BP over the prior year. Providers intensified medications in 30% of the 68 610 elevated BP events, with almost no variation in intensification regardless of whether patients had good or poor BP medication adherence. After adjustment, intensification rates were 31% among patients who had “gaps” of <20% (days on which patients should have had medication but no medication was available because medications had not been refilled), 34% among patients with refill gaps of 20% to 59%, and 32% among patients with gaps of 60% or more. Conclusions— Intensification of medications occurred in fewer than one third of visits in which patients had an elevated BP. Patients’ prior medication adherence had little impact on providers’ decisions about intensifying medications, even at very high levels of poor adherence. Addressing both patient adherence and provider intensification simultaneously would most likely result in better BP control.

BRIEF REPORT: The Burden of Diabetes Therapy: Implications for the Design of Effective Patient-centered Treatment Regimens

Background : There is debate about the optimal colorectal cancer screening test, partly because of concerns about colonoscopy demand.