Interdialytic weight gain, systolic blood pressure, serum albumin, and C-reactive protein levels change in chronic dialysis patients prior to death.

Reports from a United States cohort of chronic hemodialysis patients suggested that weight loss, a decline in pre-dialysis systolic blood pressure, and decreased serum albumin may precede death. However, no comparative studies have been reported in such patients from other countries. Here we analyzed dynamic changes in these parameters in hemodialysis patients and included 3593 individuals from 5 Asian countries; 35,146 from 18 European countries; 8649 from Argentina; and 4742 from the United States. In surviving prevalent patients, these variables appeared to have notably different dynamics than in patients who died. While in all populations the interdialytic weight gain, systolic blood pressure, and serum albumin levels were stable in surviving patients, these indicators declined starting more than a year ahead in those who died with the dynamics similar irrespective of gender and geographic region. In European patients, C-reactive protein levels were available on a routine basis and indicated that levels of this acute-phase protein were low and stable in surviving patients but rose sharply before death. Thus, relevant fundamental biological processes start many months before death in the majority of chronic hemodialysis patients. Longitudinal monitoring of these dynamics may help to identify patients at risk and aid the development of an alert system to initiate timely interventions to improve outcomes.

Monitoring dialysis outcomes across the world--the MONDO Global Database Consortium.

BACKGROUND/AIMS: Dialysis providers frequently collect detailed longitudinal and standardized patient data, providing valuable registries of routine care. However, even large organizations are restricted to certain regions, limiting their ability to separate effects of local practice from the pathophysiology shared by most dialysis patients. To overcome this limitation, the MONDO (MONitoring Dialysis Outcomes) research consortium has created a platform for the joint analysis of data from almost 200,000 dialysis patients worldwide. METHODS: We examined design and operation of MONDO as well as its methodology with respect to patient inclusion, descriptive data and other study parameters. RESULTS: MONDO partners contribute primary databases of anonymized patient data and collaboratively analyze populations across national and regional boundaries. To that end, datasets from different electronic health record systems are converted into a uniform structure. Patients are enrolled without systematic exclusions into open cohorts representing the diversity of patients. A large number of patient level treatment and outcome data is recorded frequently and can be analyzed with little delay. Detailed variable definitions are used to determine if a parameter can be studied in a subset or all databases. CONCLUSION: MONDO has created a large repository of validated dialysis data, expanding the opportunities for outcome studies in dialysis patients. The density of longitudinal information facilitates in particular trend analysis. Limitations include the paucity of uniform definitions and standards regarding descriptive information (e.g. comorbidities), which limits the identification of patient subsets. Through its global outreach, depth, breadth and size, MONDO advances the observational study of dialysis patients and care.

Practical Implementation and Challenges of Artificial Intelligence-Driven Electronic Health Record Evaluation: Protected Health Information.

Detecting protected health information in electronic health record systems is often an early step in health care analytics, and it is a nontrivial problem. Specific challenges include finding clinician names and diseases, which lack a fixed format and are often context-dependent. The general problem of finding entities, termed named-entity recognition, has received a substantial amount of attention in the natural language processing and deep learning communities. This paper begins by outlining recent methods for finding protected health information, and it then introduces a hybrid system which combines regular expressions with a natural language processing framework called FLAIR. FLAIR is open-source, it includes state-of-the-art deep learning models, and it supports straightforward development of new models for language tasks including named-entity recognition. Finally, there is a discussion of how to apply the system to structured text in a database table as well as unstructured text in clinical notes.

Temporal evolution of clinical parameters before death in dialysis patients: a new concept.

BACKGROUND: Chronic hemodialysis (HD) patients experience an appallingly high mortality in the range of 20% per year. Little is known on the dynamics of key clinical and laboratory variables in the weeks and months preceding death. In order to gain more insight into events preceding death, we embarked on a novel methodological approach which encompasses data analysis from the date of death backwards in time. METHODS: The current study investigates the dynamics of postdialytic weight and serum albumin levels in the 24 months preceding death. We performed a retrospective analysis of 2,462 maintenance HD patients who died between July 1, 2005 and April 30, 2008. Patients' monthly serum albumin levels were extracted for the 24 months preceding the date of death. Similarly, the median weekly postdialysis weight was extracted for the 104 weeks prior to death. Data were analyzed with linear mixed models. RESULTS: Both albumin levels and postdialytic body weight showed a significant decrease irrespective of gender and race in the 3 final months of life. The most pronounced decreases in postdialytic weight and albumin levels were observed in patients with infection as cause of death, the smallest changes occurred in subjects with cerebrovascular events. CONCLUSIONS: In their final 3 months of life, HD patients experience a marked decrease in body weight and serum albumin levels. A better understanding of dynamic patterns of key variables before death may be useful in developing processes which alert medical caregivers to patients at increased risk, in order to institute timely diagnostic and therapeutic interventions.

Association between genotype and phenotype in uromodulin-associated kidney disease.

BACKGROUND AND OBJECTIVES: Uromodulin-associated kidney disease (UAKD) is an autosomal dominant disease caused by uromodulin (UMOD) gene mutations. This study explored genotype-phenotype correlations by examining the relationship between the type of UMOD mutation and the age at onset of ESRD. DESIGN, SETTING, PARTICIPANTS & MEASUREMENTS: Extensive bibliographic research was used to ascertain patient-level data of all patients with UAKD published up to October 2011. Data included sex; ages at onset of hyperuricemia, gout, and ESRD; and UMOD genotype. Kaplan-Meier analysis and Cox proportional hazards models fitted with shared gamma frailty terms to adjust for within-family correlations were used to model time to event. RESULTS: Thirty-one peer-reviewed publications reporting on 202 patients from 74 families with 59 different UMOD mutations were included. Median ages at onset of hyperuricemia, gout, and ESRD were 24, 40, and 56 years, respectively. Men developed gout and ESRD significantly earlier than did women (age at ESRD was 50 years for men and 60 for women; P=0.04, shared frailty model). Median ages at ESRD development were lowest with Cys77Tyr (37.5 years) and highest with Gln316Pro (65.5 years) UMOD mutations. Onset of ESRD was significantly earlier with UMOD mutations located within the epidermal growth factor domains 2 and 3 (range, 45-52 years; P<0.01 and 0.04, respectively) compared with the cysteine-rich domains (range, 60-65 years; by shared frailty model). CONCLUSIONS: The UMOD genotype is related to the clinical phenotype of UAKD. This finding may assist in counseling of patients.

Mixture modeling of microarray gene expression data.

About 28% of genes appear to have an expression pattern that follows a mixture distribution. We use first- and second-order partial correlation coefficients to identify trios and quartets of non-sex-linked genes that are highly associated and that are also mixtures. We identified 18 trio and 35 quartet mixtures and evaluated their mixture distribution concordance. Concordance was defined as the proportion of observations that simultaneously fall in the component with the higher mean or simultaneously in the component with the lower mean based on their Bayesian posterior probabilities. These trios and quartets have a concordance rate greater than 80%. There are 33 genes involved in these trios and quartets. A factor analysis with varimax rotation identifies three gene groups based on their factor loadings. One group of 18 genes has a concordance rate of 56.7%, another group of 8 genes has a concordance rate of 60.8%, and a third group of 7 genes has a concordance rate of 69.6%. Each of these rates is highly significant, suggesting that there may be strong biological underpinnings for the mixture mechanisms of these genes. Bayesian factor screening confirms this hypothesis by identifying six single-nucleotide polymorphisms that are significantly associated with the expression phenotypes of the five most concordant genes in the first group.