Pre-ESRD changes in body weight and survival in nursing home residents starting dialysis

Time-series deep survival prediction for hemodialysis patients using an attention-based Bi-GRU network

BACKGROUND AND OBJECTIVE

The number of end-stage renal disease (ESRD) patients treated with hemodialysis (HD) has significantly increased, but the prognosis remains poor. Time-series features have been included in only a few studies to predict HD patient survival, and how to utilize such features effectively remains unclear. This article aims to develop a more accurate, interpretable, and clinically practical personalized survival prediction model for HD patients.

METHODS

This study proposed and evaluated an attention-based Bi-GRU network using time-series features for survival prediction. A distance-based loss function was proposed to improve performance. We used data from 1232 ESRD patients who received regular hemodialysis treatment for ≥ 3 months from 2007 to 2016 at the First Affiliated Hospital of Zhejiang University. The proposed model was compared with representative sequence modeling deep learning architectures and existing survival analysis methods in terms of the C-index and IBS value. Post hoc tests were used to test statistical significance. The attention map was used to assess feature importance over time. The impact of time-series changes on survival was investigated after controlling initial values (using BMI as an example).

RESULTS

The proposed method outperformed other sequence modeling architectures and the state-of-the-art survival analysis approaches in terms of the C-index and the integrated Brier score (IBS) value. Our method achieved a C-index of 0.7680 (95% confidence intervals [CI]: 0.7645, 0.7716) and an IBS of 0.1302 (95% confidence intervals [CI]: 0.1292, 0.1313), showing an improvement of up to 5.4% in terms of the C-index and a decrease of 3.2% in terms of the IBS value. The addition of the distance-based loss function improved the performance. The predicted risk and actual risk levels closely agreed. This study also found that even after controlling the initial body mass index (BMI) values, different 3-month BMI trends could produce different survival outcomes.

CONCLUSIONS

This study proposed a more effective and interpretable method to use time-series information in survival analysis. The proposed method may help promote personalized medicine and improve patient prognosis.

Incidence of Fractures Before and After Dialysis Initiation

Fractures are common in dialysis patients, but little is known about the trajectory of incidence rates of different types of fractures before and after dialysis initiation. To address this, we investigated the incidence of major fractures before and after dialysis initiation. We performed a retrospective statistical analysis using the Swedish Renal Registry of 9041 incident dialysis patients (median age 67 years, 67% men) starting dialysis 2005 through 2015 to identify major fractures (hip, spine, humerus, and forearm) occurring during the dialysis transition period from 1 year before until 1 year after dialysis initiation. Using flexible parametric hazard models and the Fine-Gray model, we estimated adjusted fracture incidence rates and predictors of major fractures. We identified 361 cases with primary diagnosis of major fracture, of which 196 (54%) were hip fractures. The crude incidence rate of major fractures before dialysis initiation was 17 per 1000 patient-years (n = 157) and after dialysis initiation it was 24 per 1000 patient-years (n = 204). The adjusted incidence rate of major fractures began to increase 6 months before dialysis initiation, and then stabilized at a higher rate after 1 year. The adjusted incidence rate of hip fractures started to increase sharply 3 months before dialysis initiation, peaked at initiation, and declined thereafter. In contrast, the adjusted incidence rate of non-hip fractures was stable during the transition period and gradually increased over time. Higher age, female sex, and history of previous major fractures were associated with increased fracture incidence both before and after dialysis initiation. We conclude that the incidence of major fractures, especially hip fractures, start to rise 6 months before initiation of dialysis therapy, indicating that heightened surveillance with implementation of preventive measures to avoid fractures is warranted during the transition period to dialysis. © 2020 The Authors. Journal of Bone and Mineral Research published by Wiley Periodicals LLC on behalf of American Society for Bone and Mineral Research (ASBMR).

Association Between Weight Loss Before Deceased Donor Kidney Transplantation and Posttransplantation Outcomes

RATIONALE & OBJECTIVE

There is debate on whether weight loss, a hallmark of frailty, signals higher risk for adverse outcomes among recipients of deceased donor kidney transplantation (DDKT).

STUDY DESIGN

Retrospective cohort study.

SETTING & PARTICIPANTS

Using national Organ Procurement and Transplantation Network data, we included all DDKT recipients in the United States between December 4, 2004, and December 3, 2014, who were adults (aged ≥ 18 years) when listed for DDKT.

EXPOSURES

Relative pre-DDKT weight change as a continuous predictor and categorized as <5% weight change from listing to DDKT, ≥5% to <10% weight loss, ≥10% weight loss, ≥5% to <10% weight gain, and ≥10% weight gain.

OUTCOMES

We examined 3 post-DDKT outcomes: (1) transplant hospitalization length of stay (LOS) in days, (2) all-cause graft failure, and (3) mortality.

ANALYTIC APPROACH

Unadjusted fractional polynomial methods, multivariable log-gamma models, and multivariable Cox proportional hazards models.

RESULTS

Among 94,465 recipients of DDKT, median pre-DDKT weight change was 0 (interquartile range, -3.5 to +3.9) kg. There were nonlinear unadjusted associations between relative pre-DDKT weight loss and longer transplant hospitalization LOS, higher all-cause graft loss, and higher mortality. Compared with recipients with <5% pre-DDKT weight change (n = 49,366; 52%), recipients who lost ≥10% of their listing weight (n = 10,614; 11%) had 0.66 (95% CI, 0.23-1.09) days longer average transplant hospitalization LOS (P = 0.003), 1.11-fold higher graft loss (adjusted HR [aHR], 1.11; 95% CI, 1.06-1.17; P < 0.001), and 1.18-fold higher mortality (aHR, 1.18; 95% CI, 1.11-1.25; P < 0.001) independent of recipient, donor, and transplant factors. Pre-DDKT dialysis exposure, listing body mass index category, and waiting time modified the association of pre-DDKT weight change with hospital LOS (interaction P < 0.10), but not with all-cause graft loss and mortality.

LIMITATIONS

Unmeasured confounders and inability to identify volitional weight change. Also, the higher significance level set to increase the power of detecting interactions with the fixed sample size may have resulted in increased risk for type 1 error.

CONCLUSIONS

DDKT recipients with ≥10% pre-DDKT weight loss are at increased risk for adverse outcomes and may benefit from augmented support post-DDKT.

Pro: The rationale for dietary therapy for patients with advanced chronic kidney disease

Dietary treatment offers many benefits to patients with advanced chronic kidney disease (CKD) who are approaching the need for renal replacement therapy. A large number of these benefits are independent of whether diets slow the rate of progression of CKD. These diets are low in protein and many minerals, and provide adequate energy for the CKD patient. The diets can reduce accumulation of potentially toxic metabolic products derived from protein and amino acid degradation, maintain a healthier balance of body water, sodium, potassium, phosphorus, calcium and other minerals, and prevent or improve protein-energy wasting. Such diets may enable patients to safely delay the onset of chronic dialysis therapy or kidney transplantation. Dietary therapy may also augment the effectiveness of infrequent or incremental dialysis by maintaining healthier metabolic and clinical status and may enable some end-stage renal disease patients to avoid the need for temporary placement of hemodialysis catheters while their arterial venous fistulae or grafts mature. The anxiety that many advanced CKD patients commonly experience with regard to starting dialysis may incentivize them to accept and adhere to dietary therapy.

Self-reported Intentional Weight Loss and Risk of Death in Moderate Chronic Kidney Disease in the United States

OBJECTIVE

We hypothesized that intentional weight loss is associated with lower mortality risk, whereas unintentional weight loss is associated with higher mortality risk in chronic kidney disease (CKD).

DESIGN AND METHODS

We examined this hypothesis in 872 participants with age >20 years, body mass index ≥ 25 kg/m² and CKD from 1999-2004 National Health and Nutrition Examination Survey who reported their 1 year prior and current weights and the intent to lose weight. We examined the association of self-reported intentional versus unintentional weight loss with all-cause mortality. Participants with no intent to lose weight and no change in weight were the reference group. A multivariable Cox regression model was used to relate mortality with intentional and unintentional weight losses after adjustment for demographics and comorbidity.

RESULTS

There were 446 deaths over 6271 years of follow-up. Compared to the reference group, intentional weight loss of 5% to <10% (hazard ratio (HR) 1.22, 95% confidence interval (CI): 0.74-1.99), intentional weight loss of ≥10% (HR 1.53, 95% CI: 0.75-3.12), and unintentional weight loss of 5% to <10% (HR 1.11, 95% CI: 0.71-1.75) were not associated with mortality; however, unintentional weight loss of ≥10% (HR 1.66, 95% CI: 1.06-2.58) was significantly associated with higher risk of mortality. Retrospective design and self-reported weight loss were the limitations.

CONCLUSIONS

Intentional weight loss in CKD participants was not associated with lower mortality risk. This might reflect residual confounding. Mechanistic and interventional studies are warranted to determine the effects of intentional weight loss in CKD.

Longitudinal Weight Change During CKD Progression and Its Association With Subsequent Mortality

BACKGROUND

Few studies have investigated the changes in weight that may occur over time among adults with the progression of chronic kidney disease (CKD). Whether such weight changes are independently associated with death after the onset of end-stage renal disease has also not been rigorously examined.

STUDY DESIGN

Prospective cohort study.

SETTING & PARTICIPANTS

We studied 3,933 participants of the Chronic Renal Insufficiency Cohort (CRIC) Study, a longitudinal cohort of patients with CKD. We also performed similar analyses among 1,067 participants of the African American Study of Kidney Disease and Hypertension (AASK).

PREDICTORS

Estimated glomerular filtration rate (eGFR) and weight change during CKD.

OUTCOME

Weight and all-cause mortality after dialysis therapy initiation.

RESULTS

During a median follow-up of 5.7 years in CRIC, weight change was not linear. Weight was stable until cystatin C-based eGFR (eGFRcys) decreased to <35mL/min/1.73m2; thereafter, weight declined at a mean rate of 1.45 kg (95% CI, 1.19-1.70) for every 10 mL/min/1.73m2 decline in eGFRcys. Among the 770 CRIC participants who began hemodialysis or peritoneal dialysis therapy during follow-up, a >5% annualized weight loss after eGFR decreased to <35mL/min/1.73m2 was associated with a 54% higher risk for death after dialysis therapy initiation (95% CI, 1.17-2.03) compared with those with more stable weight (annualized weight changes within 5% of baseline) in adjusted analysis. Similar findings were observed in the AASK.

LIMITATIONS

Inclusion of research participants only; inability to distinguish intentional versus unintentional weight loss.

CONCLUSIONS

Significant weight loss began relatively early during the course of CKD and was associated with a substantially higher risk for death after dialysis therapy initiation. Further studies are needed to determine whether interventions to optimize weight and nutritional status before the initiation of dialysis therapy will improve outcomes after end-stage renal disease.

Current Uses of Dietary Therapy for Patients with Far-Advanced CKD

For several decades, inquiry concerning dietary therapy for nondialyzed patients with CKD has focused mainly on its capability to retard progression of CKD. However, several studies published in recent years indicate that, independent of whether diet can delay progression of CKD, well designed low-protein diets may provide a number of benefits for people with advanced CKD who are close to requiring or actually in need of RRT. Dietary therapy may both maintain good nutritional status and safely delay the need for chronic dialysis in such patients, offering the possibility of improving quality of life and reducing health care costs. With the growing interest in incremental dialysis, dietary therapy may enable lower doses of dialysis to be safely and effectively used, even as GFR continues to decrease. Such combinations of dietary and incremental dialysis therapy might slow the rate of loss of residual GFR, possibly reduce mortality in patients with advanced CKD, improve quality of life, and also, reduce health care costs. The amount of evidence that supports these possibilities is limited, and more well designed, randomized clinical trials are clearly indicated.

Early changes in body weight and blood pressure are associated with mortality in incident dialysis patients

BACKGROUND

While much research is devoted to identifying novel biomarkers, addressing the prognostic value of routinely measured clinical parameters is of great interest. We studied early blood pressure (BP) and body weight (BW) trajectories in incident haemodialysis patients and their association with all-cause mortality.

METHODS

In a cohort of 357 incident patients, we obtained all records of BP and BW during the first 90 days on dialysis (over 12 800 observations) and analysed trajectories using penalized B-splines and mixed linear regression models. Baseline comorbidities and all-cause mortality (median follow-up: 2.2 years) were obtained from the French Renal Epidemiology and Information Network (REIN) registry, and the association with mortality was assessed by Cox models adjusting for baseline comorbidities.

RESULTS

During the initial 90 days on dialysis, there were non-linear decreases in BP and BW, with milder slopes after 15 days [systolic BP (SBP)] or 30 days [diastolic BP (DBP) and BW]. SBP or DBP levels at dialysis initiation and changes in BW occurring in the first month or during the following 2 months were significantly associated with survival. In multivariate models adjusting for baseline comorbidities and prescriptions, higher SBP value and BW slopes were independently associated with a lower risk of mortality. Hazard ratios of mortality and 95% confidence intervals were 0.92 (0.85-0.99) for a 10 mmHg higher SBP and 0.76 (0.66-0.88) for a 1 kg/month higher BW change on Days 30-90.

CONCLUSIONS

BW loss in the first weeks on dialysis is a strong and independent predictor of mortality. Low BP is also associated with mortality and is probably the consequence of underlying cardiovascular diseases. These early markers appear to be valuable prognostic factors.

Impact of BMI Variations on Survival in Elderly Hemodialysis Patients

OBJECTIVES

In elderly hemodialysis patients, protein-energy wasting is associated with poor outcome; however, the association between body mass index (BMI) changes over time, and survival has been seldom studied in this particularly frail population.

DESIGN AND METHODS

This prospective study recruited 502 hemodialysis patients aged ≥75 years from the French cohort ARNOS and followed them from 2005 to 2009. BMI changes over time were modeled by individual linear regression models. Survival analyses used frailty Cox models.

RESULTS

The population average age was 80.9 years. Forty-one percent of the patients died during follow-up. A 1 kg/m(2) lower baseline BMI was associated with a 4% increase in the risk of death over the study period (hazard ratio [HR] 1.04, 95% confidence interval [1.01-1.08], P = .02). A 5% BMI loss per year was associated with a 52% increase in the risk of death (HR 1.52, 95% confidence interval [1.32-1.75], P < .001). In patients who lost weight (>5% BMI loss per year), the lower was the baseline BMI, the higher was the HR for death. There was a similar trend in the patients with stable weight (5% BMI loss-5% BMI gain per year). In patients who gained weight, the HR was unexpectedly higher than in those with stable weight.

CONCLUSIONS

In elderly hemodialysis patients, the impact of the BMI percent change on survival was stronger than that of the baseline BMI. Patients with stable weight had longer survivals than patients who lost or gained weight. Thus, in this population, BMI changes should be regularly assessed. Further studies should assess the safety of weight gain strategies.