Merel van Diepen

Multiple imputation: dealing with missing data

In many fields, including the field of nephrology, missing data are unfortunately an unavoidable problem in clinical/epidemiological research. The most common methods for dealing with missing data are complete case analysis-excluding patients with missing data--mean substitution--replacing missing values of a variable with the average of known values for that variable-and last observation carried forward. However, these methods have severe drawbacks potentially resulting in biased estimates and/or standard errors. In recent years, a new method has arisen for dealing with missing data called multiple imputation. This method predicts missing values based on other data present in the same patient. This procedure is repeated several times, resulting in multiple imputed data sets. Thereafter, estimates and standard errors are calculated in each imputation set and pooled into one overall estimate and standard error. The main advantage of this method is that missing data uncertainty is taken into account. Another advantage is that the method of multiple imputation gives unbiased results when data are missing at random, which is the most common type of missing data in clinical practice, whereas conventional methods do not. However, the method of multiple imputation has scarcely been used in medical literature. We, therefore, encourage authors to do so in the future when possible.

Speckle Tracking Echocardiography Detects Uremic Cardiomyopathy Early and Predicts Cardiovascular Mortality in ESRD

Cardiovascular mortality is high in ESRD, partly driven by sudden cardiac death and recurrent heart failure due to uremic cardiomyopathy. We investigated whether speckle-tracking echocardiography is superior to routine echocardiography in early detection of uremic cardiomyopathy in animal models and whether it predicts cardiovascular mortality in patients undergoing dialysis. Using speckle-tracking echocardiography in two rat models of uremic cardiomyopathy soon (4-6 weeks) after induction of kidney disease, we observed that global radial and circumferential strain parameters decreased significantly in both models compared with controls, whereas standard echocardiographic readouts, including fractional shortening and cardiac output, remained unchanged. Furthermore, strain parameters showed better correlations with histologic hallmarks of uremic cardiomyopathy. We then assessed echocardiographic and clinical characteristics in 171 dialysis patients. During the 2.5-year follow-up period, ejection fraction and various strain parameters were significant risk factors for cardiovascular mortality (primary end point) in a multivariate Cox model (ejection fraction hazard ratio [HR], 0.97 [95% confidence interval (95% CI), 0.95 to 0.99; P=0.012]; peak global longitudinal strain HR, 1.17 [95% CI, 1.07 to 1.28; P<0.001]; peak systolic and late diastolic longitudinal strain rates HRs, 4.7 [95% CI, 1.23 to 17.64; P=0.023] and 0.25 [95% CI, 0.08 to 0.79; P=0.02], respectively). Multivariate Cox regression analysis revealed circumferential early diastolic strain rate, among others, as an independent risk factor for all-cause mortality (secondary end point; HR, 0.43; 95% CI, 0.25 to 0.74; P=0.002). Together, these data support speckle tracking as a postprocessing echocardiographic technique to detect uremic cardiomyopathy and predict cardiovascular mortality in ESRD.

Risk of postoperative acute kidney injury in patients undergoing orthopaedic surgery--development and validation of a risk score and effect of acute kidney injury on survival: observational cohort study.

Study question What is the predicted risk of acute kidney injury after orthopaedic surgery and does it affect short term and long term survival? Methods The cohort comprised adults resident in the National Health Service Tayside region of Scotland who underwent orthopaedic surgery from 1 January 2005 to 31 December 2011. The model was developed in 6220 patients (two hospitals) and externally validated in 4395 patients from a third hospital. Several preoperative variables were selected for candidate predictors, based on literature, clinical expertise, and availability in the orthopaedic surgery setting. The main outcomes were the development of any severity of acute kidney injury (stages 1-3) within the first postoperative week, and 90 day, one year, and longer term survival. Study answer and limitations Using logistic regression analysis, independent predictors of acute kidney injury were older age, male sex, diabetes, number of prescribed drugs, lower estimated glomerular filtration rate, use of angiotensin converting enzyme inhibitors or angiotensin receptor blockers, and American Society of Anesthesiologists grade. The model’s predictive performance for discrimination was good (C statistic 0.74 in development cohort, 0.70 in validation cohort). Calibration was good in the development cohort and after recalibration in the validation cohort. Only the highest risks were over-predicted. Survival was worse in patients with acute kidney injury compared with those without (adjusted hazard ratio 1.53, 95% confidence interval 1.38 to 1.70). This was most noticeable in the short term (adjusted hazard ratio: 90 day 2.36, 1.94 to 2.87) and diminished over time (90 day-one year 1.40, 1.10 to 1.79; >1 year 1.28, 1.10 to 1.48). The model used routinely collected data in the orthopaedic surgery setting therefore some variables that could potentially improve predictive performance were not available. However, the readily available predictors make the model easily applicable. What this study adds A preoperative risk prediction model consisting of seven predictors for acute kidney injury was developed, with good predictive performance in patients undergoing orthopaedic surgery. Survival was significantly poorer in patients even with mild (stage 1) postoperative acute kidney injury. Funding, competing interests, data sharing SB received grants from Tenovus Tayside, Chief Scientist Office, and the Royal College of Physicians and Surgeons of Glasgow; PT receives grants from Novo Nordisk, GlaxoSmithKline, and the New Drugs Committee of the Scottish Medicines Consortium. No additional data are available.

Predicting Mortality in Patients with Diabetes Starting Dialysis

Background While some prediction models have been developed for diabetic populations, prediction rules for mortality in diabetic dialysis patients are still lacking. Therefore, the objective of this study was to identify predictors for 1-year mortality in diabetic dialysis patients and use these results to develop a prediction model. Methods Data were used from the Netherlands Cooperative Study on the Adequacy of Dialysis (NECOSAD), a multicenter, prospective cohort study in which incident patients with end stage renal disease (ESRD) were monitored until transplantation or death. For the present analysis, patients with DM at baseline were included. A prediction algorithm for 1-year all-cause mortality was developed through multivariate logistic regression. Candidate predictors were selected based on literature and clinical expertise. The final model was constructed through backward selection. The models predictive performance, measured by calibration and discrimination, was assessed and internally validated through bootstrapping. Results A total of 394 patients were available for statistical analysis; 82 (21%) patients died within one year after baseline (3 months after starting dialysis therapy). The final prediction model contained seven predictors; age, smoking, history of macrovascular complications, duration of diabetes mellitus, Karnofsky scale, serum albumin and hemoglobin level. Predictive performance was good, as shown by the c-statistic of 0.810. Internal validation showed a slightly lower, but still adequate performance. Sensitivity analyses showed stability of results. Conclusions A prediction model containing seven predictors has been identified in order to predict 1-year mortality for diabetic incident dialysis patients. Predictive performance of the model was good. Before implementing the model in clinical practice, for example for counseling patients regarding their prognosis, external validation is necessary.

Uric acid: association with rate of renal function decline and time until start of dialysis in incident pre-dialysis patients

BackgroundIn patients with chronic kidney disease (CKD) hyperuricemia is common. Evidence that hyperuricemia might also play a causal role in vascular disease, hypertension and progression of CKD is accumulating. Therefore, we studied the association between baseline uric acid (UA) levels and the rate of decline in renal function and time until start of dialysis in pre-dialysis patients.MethodsData from the PREPARE-2 study were used. The PREPARE-2 study is an observational prospective cohort study including incident pre-dialysis patients with CKD stages IV-V in the years between 2004 and 2011. Patients were followed for a median of 14.9 months until start of dialysis, kidney transplantation, death, or censoring. Main outcomes were the change in the rate of decline in renal function (measured as estimated glomerular filtration rate (eGFR)) estimated using linear mixed models, and time until start of dialysis estimated using Cox proportional hazards models.ResultsIn this analysis 131 patients were included with a baseline UA level (mean (standard deviation (SD)) of 8.0 (1.79) mg/dl) and a mean decline in renal function of -1.61 (95% confidence interval (CI), -2.01; -1.22) ml/min/1.73 m2/year. The change in decline in GFR associated with a unit increase in UA at baseline was -0.14 (95% CI -0.61;0.33, p = 0.55) ml/min/1.73 m2/year. Adjusted for demography, comorbidities, diet, body mass index (BMI), blood pressure, lipids, proteinuria, diuretic and/or allopurinol usage the change in decline in eGFR did not change. The hazard ratio (HR) for starting dialysis for each mg/dl increase in UA at baseline was 1.08 (95% CI, 0.94;1.24, p = 0.27). After adjustment for the same confounders the HR became significant at 1.26 (95% CI, 1.06;1.49, p = 0.01), indicating an earlier start of dialysis with higher levels of UA.ConclusionAlthough high UA levels are not associated with an accelerated decline in renal function, a high serum UA level in incident pre-dialysis patient is a risk factor for an earlier start of dialysis.

Uric acid is not associated with decline in renal function or time to renal replacement therapy initiation in a referred cohort of patients with Stage III, IV and V chronic kidney disease

BACKGROUND Although many studies have suggested an association between higher uric acid (UA) and both development of chronic kidney disease (CKD) and faster decline in renal function in Stage I and II CKD, it is not clear whether this effect is consistent throughout higher CKD stages. The aim of this study was to investigate the association between baseline UA and renal outcomes in patients with established CKD (Stages III-V). METHODS We analysed data in the Swedish Renal Registry-Chronic Kidney Disease (SRR-CKD), which is a nationwide registry of referred CKD patients. Patients with a visit between January 1(st), 2005 and December 31(st), 2011 were followed until initiation of renal replacement therapy (RRT), death, referral to primary care or end of follow-up. Decline in renal function was assessed with a linear mixed model using all estimated glomerular filtration rate (eGFR) assessments recorded during median 28 months of follow-up, adjusting for important confounders such as demographic factors, primary renal disease, age, sex, relevant medication, diet, blood pressure and body mass index. RESULTS There were 2466 patients with a baseline UA measurement {mean [standard deviation (SD)] of 7.81 [1.98] mg/dL}. The mean decline in renal function was -1.48 (95% CI -1.65; -1.31) mL/min/1.73 m(2) per year. The overall adjusted change in decline in renal function per unit increase in baseline UA was 0.08 (95% CI -0.01; 0.17) mL/min/1.73 m(2) per year indicating no association between higher UA levels and decline in renal function. In Stage III, IV and V CKD patients, the mean decline in renal function was -1.52 (95% CI -1.96; -1.08), -1.52 (95% CI -1.72; -1.32) and -1.19 (95% CI -1.75; -0.64) mL/min/1.73 m(2) per year, respectively. The adjusted change in the decline in renal function per unit increase in baseline UA was -0.09 (95% CI -0.30; 0.13) in Stage III CKD, 0.16 (95% CI 0.04; 0.28) in Stage IV CKD and 0.18 (95% CI -0.09; 0.45) in Stage V CKD. The overall adjusted hazard ratio for start of RRT was 0.97 (95% CI 0.93-1.02). For Stage III, IV and V CKD, it was 0.99 (95% CI 0.73-1.34), 0.97 (95% CI 0.91-1.03) and 0.99 (95% CI 0.91-1.07), respectively. CONCLUSION UA is not associated with the rate of decline in renal function or time to start of RRT in Stage III, IV and/or V CKD patients.

Prediction versus aetiology: common pitfalls and how to avoid them

Abstract Prediction research is a distinct field of epidemiologic research, which should be clearly separated from aetiological research. Both prediction and aetiology make use of multivariable modelling, but the underlying research aim and interpretation of results are very different. Aetiology aims at uncovering the causal effect of a specific risk factor on an outcome, adjusting for confounding factors that are selected based on pre‐existing knowledge of causal relations. In contrast, prediction aims at accurately predicting the risk of an outcome using multiple predictors collectively, where the final prediction model is usually based on statistically significant, but not necessarily causal, associations in the data at hand. In both scientific and clinical practice, however, the two are often confused, resulting in poor‐quality publications with limited interpretability and applicability. A major problem is the frequently encountered aetiological interpretation of prediction results, where individual variables in a prediction model are attributed causal meaning. This article stresses the differences in use and interpretation of aetiological and prediction studies, and gives examples of common pitfalls.

The tumor area occupied by Tbet+ cells in deeply invading cervical cancer predicts clinical outcome

BackgroundDeep invasion of the normal surrounding tissue by primary cervical cancers is a prognostic parameter for postoperative radiotherapy and relatively worse survival. However, patients with tumor-specific immunity in the blood at the time of surgery displayed a much better disease free survival. Here we analyzed if this was due to a more tumor-rejecting immune population in the tumor.MethodsTumor sections from a group of 58 patients with deep normal tissue-invading cervical tumors were stained for the presence of immune cells (CD45), IFNγ-producing cells (Tbet) and regulatory T cells (Foxp3) by immunohistochemistry. The slides were scanned and both the tumor area and the infiltration of the differently stained immune cells were objectively quantified using computer software.ResultsWe found that an increased percentage of tumor occupied by CD45+ cells was strongly associated with an enhanced tumor-infiltration by Tbet+ cells and Foxp3+ cells. Furthermore, the area occupied by CD45+ immune cells, Tbet+ cells but not Foxp3+ cells within the tumor were, in addition to the lymph node status of patients, associated with a longer disease free survival and disease specific survival. Moreover, interaction analyses between these immune parameters and lymph node status indicated an independent prognostic effect of tumor infiltrating Tbet+ cells. This was confirmed in a multivariate Cox analysis.ConclusionsThe area occupied by a preferentially type I oriented CD45+ cell infiltrate forms an independent prognostic factor for recurrence-free and disease-specific survival on top of the patient’s lymph node status.

Increased risk of graft failure and mortality in Dutch recipients receiving an expanded criteria donor kidney transplant

Survival of expanded criteria donor (ECD) kidneys and their recipients has not been thoroughly evaluated in Europe. Therefore, we compared the outcome of ECD and non‐ECD kidney transplantations in a Dutch cohort, stratifying by age and diabetes. In all first Dutch kidney transplants in recipients ≥18 years between 1995 and 2005, both relative risks (hazard ratios, HR) and adjusted absolute risk differences (RD) for ECD kidney transplantation were analysed. In 3062 transplantations [recipient age 49.0 (12.8) years; 20% ECD], ECD kidney transplantation was associated with graft failure including death [HR 1.62 (1.44–1.82)]. The adjusted HR was lower in recipients ≥60 years of age [1.32 (1.07–1.63)] than in recipients 40–59 years [1.71 (1.44–2.02) P = 0.12 for comparison with ≥60 years] and recipients 18–39 years [1.92 (1.42–2.62) P = 0.03 for comparison with ≥60 years]. RDs showed a similar pattern. In diabetics, the risks for graft failure and death were higher than in the nondiabetics. ECD kidney grafts have a poorer prognosis than non‐ECD grafts, especially in younger recipients (<60 years), and diabetic recipients. Further studies and ethical discussions should reveal whether ECD kidneys should preferentially be allocated to specific subgroups, such as elderly and nondiabetic individuals.

Pre-dialysis decline of measured glomerular filtration rate but not serum creatinine-based estimated glomerular filtration rate is a risk factor for mortality on dialysis

Background. Monitoring of renal function is important in patients with chronic kidney disease progressing towards end-stage renal failure, both for timing the start of renal replacement therapy and for determining the prognosis on dialysis. Thus far, studies on associations between estimated glomerular filtration rate (eGFR) measurements in the pre-dialysis stage and mortality on dialysis have shown no or even inverse relations, which may result from the poor validity of serum creatinine-based estimation equations for renal function in pre-dialysis patients. As decline in renal function may be better reflected by the mean of the measured creatinine and urea clearance based on 24-h urine collections (mGFR by CCr-U), we hypothesize that in patients with low kidney function, a fast mGFR decline is a risk factor for mortality on dialysis, in contrast to a fast eGFR decline. Methods. For 197 individuals, included from the multicentre NECOSAD cohort, pre-dialysis annual decline of mGFR and eGFR was estimated with linear regression, and classified according to KDOQI as fast (>4 mL/min/1.73 m2/year) or slow (⩽4 mL/min/1.73 m2/year). Cox regression was used to adjust for potential confounders. Results. Patients with a fast mGFR decline had an increased risk of mortality on dialysis: crude hazard ratio (HR) 1.84 (95% confidence interval: 1.13–2.98), adjusted HR 1.94 (1.11–3.36). In contrast, no association was found between a fast eGFR decline in the pre-dialysis phase and mortality on dialysis: crude HR 1.20 (0.75–1.89), adjusted HR 1.14 (0.67–1.94). Conclusions. This study demonstrates the importance of mGFR decline (by CCr-U) as opposed to eGFR decline in patients with low kidney function, and gives incentive for repeated mGFR measurements in patients on pre-dialysis care.