Overestimation of the probability of death on peritoneal dialysis by the Kaplan-Meier method: advantages of a competing risks approach

New horizons in the analysis of hospital readmissions of older adults

This short communication highlights analytical methods that can be usefully applied to the problem of hospital readmissions of older adults. The limitations of the models currently used in studies of hospital readmissions are described. In summary, analyses of hospital readmissions face two important methodological and statistical problems not accounted for by these currently used statistical models: the potential recurrence of readmissions, and death, a terminal event which absorbs the readmission process. Not addressing the issue raised by recurrent events and terminal event generates biased estimates. We discuss an approach for the analysis of hospital readmission risk and death in the same framework. Understanding the features of this kind of approaches is essential at a time when high-quality data on hospital readmission in older patients are becoming available to a large number of researchers. Models adapted for the analysis of recurrent and terminal events are presented, and their application to studies of hospital readmission are explained, with reference to two cohorts of several thousand older individuals.

Impact of assistance on peritonitis due to breach in aseptic procedure in diabetic patients: A cohort study with the RDPLF data

BACKGROUND

Diabetic patients often have physical impairment that could lead to manipulation errors in peritoneal dialysis (PD) and touch contamination. Nurse assistance in diabetic PD patients is known to help prevent peritonitis. We made the hypothesis that this lower risk of peritonitis was observed thanks to prevention of breach in aseptic procedure. We evaluated the impact of nurse-assisted PD on specific causes of peritonitis, especially on peritonitis due to a breach in aseptic procedure.

METHODS

This was a retrospective observational study of the data from the French Language Peritoneal Dialysis Registry. All diabetic patients older than age 18 years starting PD in France between 1 January 2012 and 31 December 2015 were included in the study. The event of interest was the first peritonitis event due to a breach in aseptic procedure. Death, kidney transplantation and peritonitis due to another mechanism were considered as competing events. We examined the association of the covariates with all the possible outcomes using a subdistribution hazard model developed for survival analysis in the presence of competing risks.

RESULTS

Four thousand one hundred one diabetic patients incident in PD were included in the study. At least one peritonitis event occurred in 1611 patients over the study period. A breach in aseptic procedure was reported in 441/1611 cases (27.3%): 209/575 (36.3%) in the self-care PD group, 56/217 (25.8%) in the family-assisted PD group and 176/819 (21.5%) in the nurse-assisted PD group. Both nurse and family assistance were associated with a lower risk of peritonitis due to breach in aseptic procedure in bivariate analysis. After adjustment on age, modified Charlson index, sex and diabetic nephropathy, patients treated by nurse-assisted PD (subdistribution hazard ratio (sd-HR) 0.52, 95% confidence interval (CI) 0.40-0.67) and those treated by family-assisted PD (sd-HR 0.70, 95% CI 0.51-0.95) had a lower likelihood of peritonitis due to a connection error compared to self-care PD in multivariate analysis. The modality of assistance was not associated with other causes of peritonitis in the multivariate analysis.

CONCLUSION

While both nurse-assisted PD and family-assisted PD were associated with lower risk of peritonitis due to a breach in aseptic procedure compared to self-care PD in our study, the protective effect was greater with nurse assistance.

Risk factors associated with outcomes of peritoneal dialysis in Taiwan: An analysis using a competing risk model

Peritoneal dialysis (PD) is one option for renal replacement therapy in patients with end-stage renal disease (ESRD). Maintenance of the PD catheter is an important issue for patient outcomes and quality of life. The aim of this retrospective cohort study is to clarify the risk factors of technique failure and outcomes at a single institute in Taiwan.The study enrolled ESRD patients who had received PD catheters in a tertiary hospital in northern Taiwan. Using a competing risks regression model, we reviewed clinical data and analyzed them in terms of the time to technical failure and clinical outcomes, including PD-related peritonitis and mortality.A total of 514 patients receiving PD between 2001 and 2013 were enrolled in the study. According to the multivariate analysis model, we found that diabetes mellitus was a risk factor for PD-related peritonitis (subdistribution hazard ratio [SHR] 1.47, 95% confidence interval [CI] 1.06-2.04, P = .021). Female gender and higher serum albumin levels were associated with lower risks of technique failure (SHR 0.67, 95% CI 0.48-0.94, P = .02; SHR 0.75, 95% CI 0.58-0.96, P = .023, respectively), but Gram-negative and polymicrobial infection increased the technique failure rate (SHR 1.68, 95% CI 1.08-2.61, P = .021; SHR 1.93, 95% CI 1.11-3.36, P = .02, respectively). Female gender was a risk factor associated with overall mortality (SHR 6.4, 95% CI 1.42-28.81, P = .016). Higher weekly urea clearance (Kt/V) and weekly creatinine clearance (WCCr) were associated with a lower risk of mortality (SHR 0.1, 95% CI 0.01-0.89, P = .04; SHR 0.97, 95% CI 0.96-0.99, P = .004, respectively).Diabetes mellitus is a risk factor contributing to PD-related peritonitis. Male patients and lower serum albumin levels were associated with higher rates of technique failure. Female gender, lower Kt/V, and WCCr are risk factors for overall mortality in PD patients.

Technique failure in Korean incident peritoneal dialysis patients: a national population-based study

Background

Technique failure is an important issue for peritoneal dialysis (PD) patients. In this study, we aimed to analyze technique failure rate in detail and to determine the predictors for technique failure in Korea.

Methods

We identified all patients who had started dialysis between January 1, 2005, and December 31, 2008, in Korea, using the Korean Health Insurance Review and Assessment Service database. A total of 7,614 PD patients were included, and the median follow-up was 24.9 months.

Results

The crude incidence rates of technique failure in PD patients were 54.1 per 1,000 patient-years. The cumulative 1-, 2-, and 3-year technique failure rates of PD patients were 4.9%, 10.3%, and 15.6%, respectively. However, those technique failure rates by Kaplan–Meier analysis were overestimated compared with the values by competing risks analysis, and the differences increased with the follow-up period. In multivariate analyses, diabetes mellitus and Medical Aid as a crude reflection of low socioeconomic status were independent risk factors in both the Cox proportional hazard model and Fine and Gray subdistribution model. In addition, cancer was independently associated with a lower risk of technique failure in the Fine and Gray model.

Conclusion

Technique failure was a major concern in patients initiating PD in Korea, especially in diabetic patients and Medical Aid beneficiaries. The results of our study offer a basis for risk stratification for technique failure.

[Comparison of peritoneal dialysis and hemodialysis survival in Provence-Alpes-Côte d'Azur]

PURPOSE

To analyze and compare survival of patients initially treated with peritoneal dialysis (PD) or hemodialysis (HD).

METHODS

We used data from the French REIN registry. We included all patients aged 18 years or more who started dialysis between 1st January 2004 and 12 December 2012 in Provence-Alpes-Côte d'Azur Region (PACA). These patients were followed up until 30 June 2014. Survival curves were generated using the Kaplan-Meier technique and tested using the log-rank test. Variables predictive of all-cause mortality were determined using Cox regression models. The propensity score was used.

MAIN RESULTS

Survival was similar between initial dialysis modalities: PD and HD, even after adjusting for the propensity score. But, when we exclude the patients who had switched from one technique of dialysis to another, survival was better in HD patients. According to the multivariate analysis, advanced age and the lack of walking autonomy appear to be associated with an increase in mortality in dialysis patients. But, the presence of hypertension improve the survival in this cohort.

CONCLUSION

The survival is similar between hemodialysis and peritoneal dialysis.

Competing risk bias in Kaplan-Meier risk estimates can be corrected

BACKGROUND

Kaplan-Meier (KM) analyses are frequently used to measure outcome risk over time. These analyses overestimate risk whenever competing events are present. Many published KM analyses are susceptible to such competing risk bias. This study derived and validated a model that predicted true outcome risk based on the biased KM risk.

METHODS

We simulated survival data sets having a broad range of 1-year true outcome and competing event risk. Unbiased true outcome risk estimates were calculated using the cumulative incidence function (CIF). Multiple linear regression was used to determine the independent association of CIF-based true outcome risk with the biased KM risk and the proportion of all outcomes that were competing events.

RESULTS

The final model found that both the biased KM-based risk and the proportion of all outcomes that were competing events were strongly associated with CIF-based risk. In validation populations that used a variety of distinct survival hazard functions, the model accurately predicted the CIF (R(2) = 1).

CONCLUSIONS

True outcome risk can be accurately predicted from KM estimates susceptible to competing risk bias.

Clinical-pathological characteristics and prognosis of a cohort of oesophageal cancer patients: a competing risks survival analysis

BACKGROUND

To determine the clinical course, follow-up strategies, and survival of oesophageal cancer patients using a competing risks survival analysis.

METHODS

We conducted a retrospective and prospective follow-up study. The study included 180 patients with a pathological diagnosis of oesophageal cancer in A Coruña, Spain, between 2003 and 2008. The Kaplan-Meier methodology and competing risks survival analysis were used to calculate the specific survival rate. The study was approved by the Ethics Review Board (code 2011/372, CEIC Galicia).

RESULTS

The specific survival rate at the first, third, and fifth years was 40.2%, 18.1%, and 12.4%, respectively. Using the Kaplan-Meier methodology, the survival rate was slightly higher after the third year of follow-up. In the multivariate analysis, poor prognosis factors were female sex (hazard ratio [HR] 1.94; 95% confidence interval [CI], 1.24-3.03), Charlson's comorbidity index (HR 1.17; 95% CI, 1.02-1.33), and stage IV tumours (HR 1.70; 95% CI, 1.11-2.59). The probability of dying decreased with surgical and oncological treatment (chemotherapy and/or radiotherapy) (HR 0.23; 95% CI, 0.12-0.45). The number of hospital consultations per year during the follow-up period, from diagnosis to the appearance of a new event (local recurrences, newly appeared metastasis, and newly appeared neoplasias) did not affect the probability of survival (HR 1.03; 95% CI, 0.92-1.15).

CONCLUSIONS

The Kaplan-Meier methodology overestimates the survival rate in comparison to competing risks analysis. The variables associated with a poor prognosis are female sex, Charlson's comorbidity score and extensive tumour invasion. Type of follow-up strategy employed after diagnosis does not affect the prognosis of the disease.

Registration on the renal transplantation waiting list and mortality on dialysis: an analysis of the French REIN registry using a multi-state model

BACKGROUND

Access to the renal transplantation (RT) waiting list depends on factors related to lower mortality rates and often occurs after dialysis initiation. The aim of the study was to use a flexible regression model to determine if registration on the RT waiting list is associated with mortality on dialysis, independent of the comorbidities associated with such registration.

METHODS

Data from the French REIN registry on 7138 incident hemodialysis (HD) patients were analyzed. A multi-state model including four states ('HD, not wait-listed', 'HD, wait-listed', 'death', and 'RT') was used to estimate the effect of being wait-listed on the probability of death.

RESULTS

During the study, 1392 (19.5%) patients were wait-listed. Of the 2954 deaths observed in the entire cohort during follow-up, 2921 (98.9%) were observed in the not wait-listed group compared with only 33 (1.1%) in the wait-listed group. In the multivariable analysis, the adjusted hazard ratio for death associated with non-registration on the waiting list was 3.52 (95% CI, 1.70-7.30). The risk factors for death identified for not wait-listed patients were not found to be significant risk factors for wait-listed patients, with the exception of age.

CONCLUSIONS

The use of a multi-state model allowed a flexible analysis of mortality on dialysis. Patients who were not wait-listed had a much higher risk of death, regardless of co-morbidities associated with being wait-listed, and did not share the same risk factors of death as wait-listed patients. Registration on the waiting list should therefore be taken into account in survival analysis of patients on dialysis.

Risk factors of CVD mortality among the elderly in Beijing, 1992 - 2009: an 18-year cohort study

Few researchers have examined the effects of multiple risk factors of cardiovascular disease (CVD) mortality simultaneously. This study was to determine the associations of combined lifestyle and other factors with CVD mortality among the elderly (n = 3,257), in Beijing, China, through data mining of the Beijing Longitudinal Study of Aging (BLSA). BLSA is a representative cohort study from 1992 to 2009, hosted by Xuan Wu Hospital. Competing risk survival analysis was conducted to explore the association between risk factors and CVD mortality. The factors focused mainly on lifestyle, physical condition, and the model was adjusted for age and gender. There were 273 of the 1,068 recorded deaths caused by CVD among the 2010 participants. Living in a suburban area (HR = 0.614, 95% CI: 0.410-0.921) was associated with lower CVD mortality. Increasing age (66-75: HR = 1.511, 95% CI: 1.111-2.055; ≥ 76: HR = 1.847, 95% CI: 1.256-2.717), high blood pressure (HR = 1.407, 95% CI: 1.031-1.920), frequent consumption of meat (HR = 1.559, 95% CI: 1.079-2.254) and physical inactivity (p = 0.046) were associated with higher CVD mortality. The study provides an instructional foundation for the control and prevention of CVD in Beijing, China.

Risk factors for cerebrovascular disease mortality among the elderly in Beijing: a competing risk analysis

Objective

To examine the associations of combined lifestyle factors and physical conditions with cerebrovascular diseases (CBVD) mortality, after accounting for competing risk events, including death from cardiovascular diseases, cancers and other diseases.

Methods

Data on 2010 subjects aged over 55 years were finally analyzed using competing risk models. All the subjects were interviewed by the Beijing Longitudinal Study of Aging (BLSA), in China, between 1 January 1992 and 30 August 2009.

Results

Elderly females were at a lower risk of death from CBVD than elderly males (HR = 0.639, 95% CI = 0.457–0.895). Increasing age (HR = 1.543, 95% CI = 1.013–2.349), poor self-rated health (HR = 1.652, 95% CI = 1.198–2.277), hypertension (HR = 2.201, 95% CI = 1.524–3.178) and overweight (HR = 1.473, 95% CI = 1.013–2.142) or obesity (HR = 1.711, 95% CI = 1.1754–2.490) was associated with higher CBVD mortality risk. Normal cognition function (HR = 0.650, 95% CI = 0.434–0.973) and living in urban (HR = 0.456, 95% CI = 0.286–0.727) was associated with lower CBVD mortality risk. Gray’s test also confirmed the cumulative incidence (CIF) of CBVD was lower in the ‘married’ group than those without spouse, and the mortality was lowest in the ‘nutrition sufficient’ group among the ‘frequent consumption of meat group’ and the ‘medial type group’ (P value<0.001).

Conclusions

CBVD mortality was associated with gender, age, blood pressure, residence, BMI, cognitive function, nutrition and the result of self-rated health assessment in the elderly in Beijing, China.

Survival prognosis after the start of a renal replacement therapy in the Netherlands: a retrospective cohort study

Background

There is no single model available to predict the long term survival for patients starting renal replacement therapy (RRT). The available models either predict survival on dialysis until transplantation, survival on the transplant waiting list, or survival after transplantation. The aim of this study was to develop a model that includes dialysis survival and survival after an eventual transplantation.

Methods

From the Dutch renal replacement registry, patients of 16 years of age or older were included if they started RRT between 1995 and 2005, still underwent RRT at baseline (90 days after the start of RRT) and were not registered at a non-renal organ transplant waiting list (N = 13868). A prediction model of 10-year patient survival after baseline was developed through multivariate Cox regression analysis, in one half of the research group. Age at start, sex, primary renal disease (PRD) and therapy at baseline were included as possible predictors. A sensitivity analysis has been performed to determine whether listing on the transplant waiting list should be added. The predictive performance of the model was internally validated. Calibration and discrimination were computed in the other half of the research group. Another sensitivity analysis was to assess whether the outcomes differed if the model was developed and tested in two geographical regions, which were less similar than the original development and validation group. No external validation has been performed.

Results

Survival probabilities were influenced by age, sex, PRD and therapy at baseline (p < 0.001). The calibration and discrimination both showed very reasonable results for the prediction model (C-index = 0.720 and calibration slope for the prognostic index = 1.025, for the 10 year survival). Adding registration on the waiting list for renal transplantation as a predictor did not improve the discriminative power of the model and was therefore not included in the model.

Conclusions

With the presented prediction model, it is possible to give a reasonably accurate estimation on the survival chances of patients who start with RRT, using a limited set of easily available data.

Early failure in patients starting peritoneal dialysis: a competing risks approach

BACKGROUND

Technical failure is more likely to occur in the first 6 months of peritoneal dialysis (PD). This study was carried out to identify risk factors for early transfer from PD to haemodialysis (HD) in a country where assisted PD is available.

METHODS

All patients from the French Language Peritoneal Dialysis Registry (RDPLF) who started PD between 1 January 2002 and 31 December 2010 were included. Time to transfer, death and transplantation during the first 6 months on PD were analysed by the multivariate Cox proportional hazard model. The Fine and Gray model was used to examine the occurrence of technical failure by considering death and transplantation as competing events.

RESULTS

Of 9675 patients included, 615 (6.3%) moved to HD during the first 6 months of PD. Cumulative incidence of transfer to HD was 6.6% at 6 months. On multivariate analysis by both the Cox model and the Fine and Gray model, HD prior to PD, allograft failure and early peritonitis were associated with a higher risk of early technical failure, whereas being dialysed in a centre treating more than 20 new patients per year was associated with a lower risk of early transfer to HD.

CONCLUSIONS

Patients treated by HD before PD and failed transplant patients had a higher risk of early PD failure when competing events were considered.

Would increasing centre volumes improve patient outcomes in peritoneal dialysis? A registry-based cohort and Monte Carlo simulation study

OBJECTIVE

To estimate the association between centre volume and patient outcomes in peritoneal dialysis, explore robustness to residual confounding and predict the impact of policies to increase centre volumes.

DESIGN

Registry-based cohort study with probabilistic sensitivity analysis and Monte Carlo simulation of (hypothetical) intervention effects.

SETTING

112 secondary-care centres in France.

PARTICIPANTS

9602 adult patients initiating peritoneal dialysis.

MAIN OUTCOME MEASURES

Technique failure (ie, permanent transfer to haemodialysis), renal transplantation and death while on peritoneal dialysis within 5 years of initiating treatment. Associations with underlying risk measured by cause-specific HRs (cs-HRs) and with cumulative incidence by subdistribution HRs (sd-HRs). Intervention effects measured by predicted mean change in cumulative incidences.

RESULTS

Higher volume centres had more patients with diabetes and were more frequently academic centres or associative groupings of private physicians. Patients in higher volume centres had a reduced risk of technique failure (>60 patients vs 0-10 patients: adjusted cs-HR 0.46; 95% CI 0.43 to 0.69), with no changed risk of death or transplantation. Sensitivity analyses mitigated the cs-HRs without changing the findings. In higher volume centres, the cumulative incidence was reduced for technique failure (>60 patients vs 0-10 patients: adjusted sd-HR 0.49; 95% CI 0.29 to 0.85) but was increased for transplantation and death (>60 patients vs 0-10 patients: transplantation-adjusted sd-HR 1.53; 95% CI 1.04 to 2.24; death-adjusted sd-HR 1.28; 95% CI 1.00 to 1.63). The predicted reduction in cumulative incidence of technique failure was largest under a scenario of shifting all patients to the two highest volume centre groups (0.091 reduction) but lower for three more realistic interventions (around 0.06 reduction).

CONCLUSIONS

Patients initiating peritoneal dialysis in high-volume centres had a considerably reduced risk of technique failure but simulations of interventions to increase exposure to high-volume centres yielded only modest improvements.

Modelling competing risks in nephrology research: an example in peritoneal dialysis

BACKGROUND

Modelling competing risks is an essential issue in Nephrology Research. In peritoneal dialysis studies, sometimes inappropriate methods (i.e. Kaplan-Meier method) have been used to estimate probabilities for an event of interest in the presence of competing risks. In this situation a competing risk analysis should be preferable. The objectives of this study are to describe the bias resulting from the application of standard survival analysis to estimate peritonitis-free patient survival and to provide alternative statistical approaches taking competing risks into account.

METHODS

The sample comprises patients included in a university hospital peritoneal dialysis program between October 1985 and June 2011 (n = 449). Cumulative incidence function and competing risk regression models based on cause-specific and subdistribution hazards were discussed.

RESULTS

The probability of occurrence of the first peritonitis is wrongly overestimated using Kaplan-Meier method. The cause-specific hazard model showed that factors associated with shorter time to first peritonitis were age (≥55 years) and previous treatment (haemodialysis). Taking competing risks into account in the subdistribution hazard model, age remained significant while gender (female) but not previous treatment was identified as a factor associated with a higher probability of first peritonitis event.

CONCLUSIONS

In the presence of competing risks outcomes, Kaplan-Meier estimates are biased as they overestimated the probability of the occurrence of an event of interest. Methods which take competing risks into account provide unbiased estimates of cumulative incidence for each specific outcome experienced by patients. Multivariable regression models such as those based on cause-specific hazard and on subdistribution hazard should be used in this competing risk setting.