Epidemiology - CKD 5D II

Abstract

Introduction and Aims: Adherence to treatment recommendations on diet, fluid and medication is important to maximize good clinical outcomes in Hemodialysis yet it remains suboptimal and not well-understood. This trial set out to examine the effect of the HED- SMART intervention, a four-session, group-delivered self-management intervention on treatment adherence indicators.Methods: Eligible HD patients were randomized to either usual care (N= 133) or HED-SMART intervention (n=102). Measures of self-report adherence, self-management skills and biochemical markers were collected at baseline, immediately and at 3 and 9 months post-intervention. The intervention was facilitated by renal healthcare professionals and involved problemsolving and goal-setting for fluid control, diet and medication.Results: A total of 235 participants were enrolled [mean age ± 53.46 (±10.41) years]. The study was completed by 74.8%. Significant differences between groups were found in change in interdialytic weight gains, potassium and phosphate levels during the intervention phase and the 3-month follow-up indicating improved dietary/fluid control and medication intake for the intervention participants (all p <0.001). In these cohorts, the AUCs of LVMI for all-cause death were 0.71±0.03 (CREED) and 0.67±0.03 (HM) and those for CV death 0.64±0.04 and 0.69±0.03 which were lower than those by age alone both for all-cause (CREED: 0.81±0.03; HM: 0.88±0.02) and CV mortality (CREED: 0.66±0.04; HM: 0.78±0.03]. All predictive models were well calibrated, i.e. there was no significant difference between observed and predicted outcomes. In the CREED cohort a predictive model including Framingham risk factors, anti-hypertensive treatment, CV comorbidities, heart rate and two major ESKD-related risk factors (Hb and albumin) produced an AUC of 0.89±0.02 for all-cause death and 0.76±0.03 for CV death. The corresponding figures in the HM cohort were 0.92±0.02 and 0.87±0.02, respectively. LVMI did not materially affect the discrimination power for all-cause (CREED 0.89 vs. 0.89; HM: 0. 93 vs. 0.92) and CV death (CREED:0.76 vs. 0.76; HM: 0.88 vs. 0.87). In an aggregate analysis of the two cohorts (n=524) the net reclassification index (NRI) by LVMI was low and not significant both for all-cause (NRI: 4.5%, P=0.11) and CV mortality (NRI: 3.4%, P=0.33). A re-classification analysis carried out by calculating the integrated discrimination improvement (IDI) provided similar results (all-cause mortality, IDI: P=0.89; CV death, IDI: P=0.88).Conclusions: LVMI is a strong CV risk factor in the ESKD population. However, the prognostic power of this biomarker is by far lower than that by age alone or combined with standard,easily available, risk factors. While LVH remains a fundamental treatment target in ESKD, measurement of LVMI solely for risk stratification is unwarranted in these patients because it does not provide any additional information as compared to standard risk factors.Introduction and Aims: The appropriate timing of dialysis initiation in outpatients with progressive chronic kidney disease remains controversial with concerns that initiation at a higher GFR is associated with an increase in mortality. The purpose of this study is to determine the variation in timing of dialysis initiation across dialysis facilities and geographic regions in Canada after accounting for patient level factors (case-mix).Methods: Data on 33, 263 dialysis patients, 63 dialysis facilities and 12 geographic regions from the Canadian Organ Replacement Registry (CORR) with an eGFR measure at dialysis initiation between Jan. 2001 and Dec. 2009 were included in the final analysis. eGFR was estimated by the MDRD equation. Multi-level models were used to evaluate the variation in timing of dialysis by eGFR at the patient-, facility- and geographic-level. Models were adjusted for patient and facility characteristics to determine the relative variability at each level.Results: The mean eGFR and proportion initiated with an eGFR > 10.5 mls/min/m2 varied considerably across geographic regions and over the study period. For instance, in patients with >3months of predialysis care, the proportion initiating dialysis with an eGFR>10.5mls/min was 37.3% varying from 20.2% to 60.2% across geographic regions. In unadjusted models, variation of 2.6, 8.2 and 89.2% were attributable to geography, facility and patient-level characteristics. After adjustment for case-mix and facility-level quality indicators, 95.3, 4.5 and 0.2% of the variability was attributable to patient, facility and geography. The adjusted odds ratio for initiating dialysis with an eGFR > 10.5 was similar across all geographic regions except one suggesting that the noted variation across facilities and geographic regions was due to patient differences. This was consistent when eGFR was examined as a continuous variable, categorized as > 12.0 mls/min/m2 or in an analysis limited to patients with > 3 months of pre-dialysis care.Conclusions: We observed significant variation in timing of dialysis initiation across geographic regions, which were predominantly explained by patient-level variation. These data suggest similar practice patterns across Canada, with the predominant factor impacting dialysis initiation being patient characteristics.⇓View larger version: In this page In a new window Download as PowerPoint SlideIntroduction and Aims: Cross-sectional measures of Health Related Quality of Life (HR-QOL) are associated with mortality and hospitalization among hemodialysis (HD) patients. Our aims were to describe within-patient changes in HR-QOL and estimate their effects on the rates of mortality and hospitalization.Methods: 13,786 patients had >1 measurement of HR-QOL from the Dialysis Outcomes and Practice Patterns Study (DOPPS) annual patient questionnaire (PQ). Changes in physical (PCS) and mental (MCS) component summary scores of the KDQOL-36TM were defined as the score from the second PQ (PQ2) minus score from the first PQ (PQ1). Median time from PQ1 to PQ2 was 12 months (IQR: 11, 14). Effects of change in HR-QOL (per 5 point decline) on both mortality and first hospitalization were estimated using Cox regression with time at risk (median: 11 months, IQR: 6, 18) beginning at PQ2, adjusting for potential confounders. In addition, effects of HR-QOL at PQ2 (3 categories) were estimated in separate Cox models by category of HR-QOL at PQ1.Results: Mean ± SD age was 61±4 years; 59% were male, 32% diabetic, and mean albumin was 3.8±0.5 g/dL. Median PCS and MCS from PQ1 were 37.5 (IQR: 29.4, 46.2) and 46.4 (IQR: 37.2, 54.9); mean changes in PCS and MCS from PQ1 to PQ2 were -0.2 (IQR: -5.5, +4.7) and -0.1 (IQR: -6.8, +5.9). A decline in PCS and MCS from PQ1 to PQ2 was associated with all-cause mortality (PCS, HR=1.10 per 5 points, 95% CI: 1.07-1.14; MCS, HR=1.06 per 5 points, 1.04-1.08) and hospitalization (PCS, HR=1.02 per 5 points, 1.01-1.04; MCS, HR=1.02 per 5 points, 1.01-1.04). Change in HR-QOL was associated with all-cause mortality across levels of HR-QOL scores at PQ1 (Table).Conclusions: Changes in HR-QOL in HD patients are common, and are associated with mortality and hospitalization. Monitoring changes in self-reported HR-QOL measures in HD patients may help to identify a subset of patients at high risk for adverse outcomes and allow for targeted interventions to improve HR-QOL and reduce these risks.⇓View larger version: In this page In a new window Download as PowerPoint SlideIntroduction and Aims: Dialysis providers often use the facility proportion of patients meeting clinical targets (“facility quality indicators”) as indicators of quality of care. Patients from dialysis units with better quality indicators may experience lower mortality. The combination of quality indicators associated with the best patient outcomes is not known. We assessed the distribution of quality indicators and their association with mortality, individually and in combination, in the international DOPPS cohort.Methods: 12,3