4L
ORAL ABSTRACTS: CHRONIC DISEASE MANAGEMENT
Method: Participants were recruited from 73 primary care practices throughout Western Pennsylvania between 2010-2011. Patients completed questionnaires regarding socio-demographic data, race and ethnicity, and income, health literacy (HL), medical conditions, and BSM (47 questions). All BSM scores (overall BSM-score and 11 BSM-domain scores: overall health, physical functioning, social activities, depression, patient- physician communication, cost of care, inconvenience, medication adherence, knowledge about conditions, multi-morbidity, self-efficacy) were normalized to a 0–100 scale and standardized so that higher values represent less barriers. We compared all BSM-domains scores (mean, SD) in patients with diabetes to those without diabetes using t-tests and the overall BSM score using ANCOVA with all main effects considered including level of education, marital status, race, age, and income.
Result: 1,169 patients were included in our analysis; 428 (36.6%) had diabetes. Patients with and without diabetes were similar with respect to sex (P=0.758) and age (P=0.234). Participants with diabetes were less likely to be married or cohabiting (P=0.008), more likely to be African American (P=0.003), and more likely to have income <$34,999 (P<0.001). Participants with diabetes also had slightly lower health literacy (HL-score 5.1 vs 5.0; P=0.001).
The mean values of all but two BSM-domains scores (exceptions being “social activities” and “patient- physician-communication”) were significantly lower for patients with diabetes compared to patients without diabetes, indicating higher barriers to self-management. This was particularly notable in the barriers of “overall health” (47.0 vs 59.3; P<0.001), “self-efficacy” (71.9 vs 77.2; P<0.001) and “feeling overwhelmed” (62.4 vs 67.8; P<0.001). The ANCOVA showed, that having diabetes [F (1, 1169) =17.39; P< 0.001], higher age [F (1, 1169) = 17.13; P< 0.001] and lower income [F (2, 1169) = 31.78; P< 0.001] was significantly associated with worse overall BSM-score.
Conclusion: Patients with diabetes reported increased self-reported barriers to self-management. Those with lower income may be particularly at risk. Interventions should strengthen self-efficacy capabilities, knowledge about the disease and skills to handle overwhelming situations.
Method: We retrospectively identified adult managed-care beneficiaries in the electronic health records (EHR) of a healthcare system between 2010 and 2014. We required patients to have: (i) completed a Press-Ganey Patient Experience Survey; (ii) a prescription for a chronic disease medication ±14 days of the office encounter associated with the survey; (iii) corresponding pharmacy fills for the medications of interest; and (iv) ≥12 months of EHR activity prior to the date of the prescription. Patients’ rating of a “provider’s efforts to include you in decisions about your treatment” is answered on a scale from 1 (very poor) to 5 (very good), and was dichotomized as “very good” vs. not “very good”. We measured adherence as the proportion of days covered (PDC), calculated as the total days’ supply of fills divided by the length of the prescription (start to end date). A PDC ≥80% was considered adherent. We used logistic regression to test the association between adherence (outcome) and rating of treatment decision-making (main predictor), with and without controlling for other important variables, including sex, age, race/ethnicity, comorbidities, and average copayment. Odds ratios (ORs) and 95% confidence intervals (CIs) were generated.
Result: We identified 14,800 patients meeting study eligibility criteria. The most commonly prescribed medications in this cohort included antihypertensive (38.7%), antihyperlipidemic (17.0%), psychiatric (10.8%), antidepressant (10.2%), thyroid (7.2%), and antidiabetic (6.2%) agents. The overall adherence rate was 56%. Univariate analyses showed that a higher rating of treatment decision-making was significantly associated with greater odds of adherence (OR: 1.09; 95% CI: 1.00-1.18; P=0.042). This relationship persisted after controlling for other covariates (adjusted OR: 1.10; 95% CI: 1.01-1.20; P=0.025). The strongest association between rating of treatment decision-making and adherence was for antidiabetic agents (adjusted OR: 1.48; 95% CI: 1.05-2.09; P=0.025).
Conclusion: In this retrospective study of managed-care beneficiaries within a large healthcare delivery system, patients’ rating of their involvement in treatment decision-making was positively associated with statistically significant improvements in adherence to chronic disease medications. Therapeutic class effects were observed. Health systems and healthcare providers should be encouraged to involve patients in shared decision-making whenever appropriate to maximize adherence.
Purpose: Persons living with HIV (PLWH) who receive regular medical care have greater survival and higher likelihood of viral suppression compared to individuals inconsistently engaged in care. However, most recent national estimates indicate that only 53.8% of PLWHs are fully retained in care. One challenge to improving retention is identifying which patients are truly out of care, as healthcare providers are rarely informed when a patient receives care elsewhere, relocates, or dies. Some of this information is captured by state HIV surveillance data, but is not routinely shared with providers. The goal of this analysis is to estimate the value of real-time integration of HIV surveillance data with clinic electronic medical records (EMRs) to better target re-engagement efforts.
Methods: EMR data of HIV-infected patients ≥18 years-old seen at a public, hospital-based clinic in Minneapolis, MN from 2008-2014 were merged with state surveillance data on HIV-related laboratory tests, out-of-state relocation, and mortality. We then estimated the resources required to operate a hypothetical re-engagement program in this patient population. Under the program, we assumed that clinic staff would initiate case investigations for patients exceeding a given time threshold since their last clinical encounter, unless the clinic had been informed that the patient had moved, changed providers, or died. We varied the time threshold for being out of care from 8-24 months. For each threshold, we calculated the total number of required case investigations over the study period with and without surveillance data integration and the number of case investigations averted due to data integration.
Results: For a patient population of 2,194, the number of required case investigations in 2008-2014 using clinic data alone ranged from 1,945 to 592, using 8- and 24-month thresholds, respectively (Figure). For a 12-month threshold, a common definition of retention, surveillance data averted 29.4% of the 1,124 case investigations that would have been conducted with clinic data alone. The proportion of investigations averted with surveillance data increased with the time threshold for initiating a case investigation, reaching a maximum of 44.3% investigations averted for a 24-month threshold.
Conclusions: Integrating clinic and surveillance data greatly improved the efficiency of efforts to re-engage HIV patients back into primary care by avoiding unnecessary case investigations for patients who appeared out of care, but were not.
Methods: MA-CORD was a two-year, multi-sector intervention in two Massachusetts communities. The clinical intervention was implemented at two federally qualified health centers (FQHCs) – MA-CORD 1 and MA-CORD 2 – serving low-income children in the communities. In each FQHC, we conducted a variety of obesity-related quality improvement initiatives including implementation of EHR-based CDS tools to guide obesity management, clinician training, orientation, and coaching in obesity-specific quality improvement, and family resources to support health behavior change. Parents assessed their children’s obesity-related healthcare via eight questions adapted from the validated Patient Assessment of Chronic Illness Care (PACIC). We converted responses to a scale ranging 1 to 5. Given Cronbach’s alpha values (0.83-0.85) suggesting high internal consistency, we combined the eight questions into a mean healthcare satisfaction score. Using multivariable difference in differences linear regression models, we examined 1- and 2-year change in healthcare satisfaction among parents of children seen at the two intervention FQHCs compared to a non-randomized comparison FQHC.
Results: Among 419 children in the study, baseline mean (SD) obesity-related healthcare satisfaction among parents of children at MA-CORD 1 was 2.76 (1.19), MA-CORD 2 was 2.44 (1.15), and the comparison site was 2.50 (1.03). We observed 1-year improvement in parent-reported healthcare satisfaction, adjusted for child, parent and household characteristics, at both MA-CORD intervention FQHCs (MA-CORD 1 = 0.54 [95% confidence interval: 0.19, 0.89] and MA-CORD 2 =0.59 [0.12, 1.06]) relative to the comparison site. At one site, we observed a continued significant difference versus the comparator site at 2-year follow-up (MA-CORD 1 = 0.50 [0.15, 0.85]).
Conclusions: Parent-reported healthcare satisfaction improved following the implementation of a clinical childhood obesity intervention including EHR-based CDS, clinician training and family resources in pediatric primary care. Improved satisfaction may translate to enhanced family engagement in healthcare and yield improved child health outcomes.
Purpose: Alzheimer's disease (AD) is a continuum: patients progress from normal cognition to mild cognitive impairment (MCI) due to AD, followed by increasing severity of AD dementia. Therapies that delay the onset of MCI due to AD could have significant implications on the trajectory of AD-related outcomes, including a reduction in the number of patients who require institutionalization. However, the impact of a delay in MCI due to AD has not been established, nor has progression through the entire disease continuum been fully characterized using transition probabilities. The objectives of this study were to: 1) estimate progression from normal cognition to MCI due to AD to mild/moderate/severe AD dementia, including the age-specific likelihood of institutionalization and death from each health state using a well-defined US population; and 2) develop a decision analysis model, using the study progression rates, to estimate the expected effect of delayed onset of MCI due to AD on AD-related outcomes for a US cohort.
Method : Longitudinal data from the National Alzheimer's Coordinating Center (NACC) Uniform Data Set, which contain annual clinical evaluations for subjects across the AD continuum, were utilized. The NACC database is funded by NIA/NIH Grant U01 AG016976. At each visit, patients were classified by health state based on clinical diagnosis, global Clinical Dementia Rating score, and primary etiologic diagnosis of cognitive impairment (CI) (Figure 1). Age-specific transitional probabilities of progression to each health state were estimated using a multivariate, ordered probit model, controlling for the patient's prior health state and current age. A separate multivariate regression model was used to assess the likelihood of institutionalization. Age-specific risk of death from each health state was derived using published estimates and life tables for the US population.
Result: When the observed transition probabilities are applied to a cohort of 100 normal-cognition patients at age 65 years, a 2-year delay in onset of MCI due to AD is, over a lifetime, predicted to avoid 4.4 cases of AD dementia, delay average time to AD onset (1.4 years), increase survival (0.26 years), increase time with normal cognition (0.75 years), and reduce time spent institutionalized (14%).
Conclusion: Therapies that delay the onset of MCI due to AD could have significant implications for rates of AD dementia and AD-associated institutionalization and death.
Method: Through rigorous engagement of a national advisory board of lung cancer survivors, a discrete-choice experiment (DCE) was developed, pretested and piloted. The DCE was administered to lung cancer survivors and caregivers that attended a national summit. Respondents at the meeting completed 13 paired-comparison choice tasks described across six attributes: progression-free survival (PFS), short-term side effects, and physical, emotional, cognitive, and functional long-term side effects. A continuous preference model was estimated using mixed logit and, using PFS as the numeraire, the preference for avoiding the side-effects were estimated using their time equivalents by using maximum simulated likelihood.
Result: Of 114 survey participants, 102 (89.4%) completed all choice tasks - although no difference was identified between those who did not complete the task (p>0.05 for all observed characteristics). All attributes were statistically different form the null (p<0.001). Respondents valued a one-unit decrease in functioning the most, valuing it equivalent to extending PFS by 3.67 months. Changes in physical (2.34) and cognitive (2.29) were valued more than a composite of all short-term side effects (1.83). Heterogeneity analyses indicated that avoiding long-term side effects could be valued even more highly for some respondents. For example, accounting for preference heterogeneity, the 95% confidence interval for time equivalence of functional long-term side effects ranged from 0.62 up to 13.31 months.
Conclusion: While this small, retrospective study that focuses only on lung cancer has many limitations, the results indicate that avoiding the long-term side effects could have significant value, especially as many patients experience moderate long-term side effects across multiple domains.