1
|
Yang W, Jepson C, Xie D, Roy JA, Shou H, Hsu JY, Anderson AH, Landis JR, He J, Feldman HI. Statistical Methods for Recurrent Event Analysis in Cohort Studies of CKD. Clin J Am Soc Nephrol 2017; 12:2066-2073. [PMID: 28716856 PMCID: PMC5718286 DOI: 10.2215/cjn.12841216] [Citation(s) in RCA: 28] [Impact Index Per Article: 4.0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/23/2022]
Abstract
Cardiovascular events, such as hospitalizations because of congestive heart failure, often occur repeatedly in patients with CKD. Many studies focus on analyses of the first occurrence of these events, and discard subsequent information. In this article, we review a number of statistical methods for analyzing ordered recurrent events of the same type, including Poisson regression and three commonly used survival models that are extensions of Cox proportional hazards regression. We illustrate the models by analyzing data from the Chronic Renal Insufficiency Cohort Study to identify risk factors for congestive heart failure hospitalizations in patients with CKD. We show that recurrent event analyses provide additional insights about the data compared with a standard survival analysis of time to the first event.
Collapse
Affiliation(s)
- Wei Yang
- Department of Biostatistics, Epidemiology, and Informatics and
- Center for Clinical Epidemiology and Biostatistics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania; and
| | - Christopher Jepson
- Center for Clinical Epidemiology and Biostatistics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania; and
| | - Dawei Xie
- Department of Biostatistics, Epidemiology, and Informatics and
- Center for Clinical Epidemiology and Biostatistics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania; and
| | - Jason A. Roy
- Department of Biostatistics, Epidemiology, and Informatics and
- Center for Clinical Epidemiology and Biostatistics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania; and
| | - Haochang Shou
- Department of Biostatistics, Epidemiology, and Informatics and
- Center for Clinical Epidemiology and Biostatistics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania; and
| | - Jesse Yenchih Hsu
- Department of Biostatistics, Epidemiology, and Informatics and
- Center for Clinical Epidemiology and Biostatistics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania; and
| | - Amanda Hyre Anderson
- Department of Biostatistics, Epidemiology, and Informatics and
- Center for Clinical Epidemiology and Biostatistics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania; and
| | - J. Richard Landis
- Department of Biostatistics, Epidemiology, and Informatics and
- Center for Clinical Epidemiology and Biostatistics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania; and
| | - Jiang He
- Department of Epidemiology, Tulane University, New Orleans, Louisiana
| | - Harold I. Feldman
- Department of Biostatistics, Epidemiology, and Informatics and
- Center for Clinical Epidemiology and Biostatistics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania; and
| | | |
Collapse
|
2
|
Williamson T, Ravani P. Marginal structural models in clinical research: when and how to use them? Nephrol Dial Transplant 2017; 32:ii84-ii90. [DOI: 10.1093/ndt/gfw341] [Citation(s) in RCA: 51] [Impact Index Per Article: 7.3] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/26/2016] [Accepted: 08/18/2016] [Indexed: 12/11/2022] Open
|
3
|
Ravani P, Barrett BJ, Parfrey PS. Longitudinal studies 3: Data modeling using standard regression models and extensions. Methods Mol Biol 2015; 1281:93-131. [PMID: 25694306 DOI: 10.1007/978-1-4939-2428-8_6] [Citation(s) in RCA: 2] [Impact Index Per Article: 0.2] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 01/28/2023]
Abstract
In longitudinal studies the relationship between exposure and disease can be measured once or multiple times while participants are monitored over time. Traditional regression techniques are used to model outcome data when each epidemiological unit is observed once. These models include generalized linear models for quantitative continuous, discrete, or qualitative outcome responses, and models for time-to-event data. When data come from the same subjects or group of subjects, observations are not independent and the underlying correlation needs to be addressed in the analysis. In these circumstances extended models are necessary to handle complexities related to clustered data, and repeated measurements of time-varying predictors and/or outcomes.
Collapse
Affiliation(s)
- Pietro Ravani
- Division of Nephrology, Department of Medicine, University of Calgary, 1403, 29th St NW (Foothills Medical Centre), Calgary, AB, Canada, T2N 2T9,
| | | | | |
Collapse
|
4
|
Leffondre K, Boucquemont J, Tripepi G, Stel VS, Heinze G, Dunkler D. Analysis of risk factors associated with renal function trajectory over time: a comparison of different statistical approaches. Nephrol Dial Transplant 2014; 30:1237-43. [PMID: 25326471 DOI: 10.1093/ndt/gfu320] [Citation(s) in RCA: 79] [Impact Index Per Article: 7.9] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/07/2014] [Accepted: 08/30/2014] [Indexed: 11/12/2022] Open
Abstract
BACKGROUND The most commonly used methods to investigate risk factors associated with renal function trajectory over time include linear regression on individual glomerular filtration rate (GFR) slopes, linear mixed models and generalized estimating equations (GEEs). The objective of this study was to explain the principles of these three methods and to discuss their advantages and limitations in particular when renal function trajectories are not completely observable due to dropout. METHODS We generated data from a hypothetical cohort of 200 patients with chronic kidney disease at inclusion and seven subsequent annual measurements of GFR. The data were generated such that both baseline level and slope of GFR over time were associated with baseline albuminuria status. In a second version of the dataset, we assumed that patients systematically dropped out after a GFR measurement of <15 mL/min/1.73 m(2). Each dataset was analysed with the three methods. RESULTS The estimated effects of baseline albuminuria status on GFR slope were similar among the three methods when no patient dropped out. When 32.7% dropped out, standard GEE provided biased estimates of the mean GFR slope in normo-, micro- and macroalbuminuric patients. Linear regression on individual slopes and linear mixed models provided slope estimates of the same magnitude, likely because most patients had at least three GFR measurements. However, the linear mixed model was the only method to provide effect estimates on both slope and baseline level of GFR unaffected by dropout. CONCLUSION This study illustrates that the linear mixed model is the preferred method to investigate risk factors associated with renal function trajectories in studies, where patients may dropout during the study period because of initiation of renal replacement therapy.
Collapse
Affiliation(s)
- Karen Leffondre
- University of Bordeaux, ISPED, Centre INSERM U897-Epidemiology-Biostatistics, Bordeaux, France
| | - Julie Boucquemont
- University of Bordeaux, ISPED, Centre INSERM U897-Epidemiology-Biostatistics, Bordeaux, France
| | - Giovanni Tripepi
- CNR-IBIM/IFC, Clinical Epidemiology and Physiopathology of Renal Diseases and Hypertension of Reggio Calabria, Calabria, Italy
| | - Vianda S Stel
- Department of Medical Informatics, Academic Medical Center, ERA-EDTA Registry, Amsterdam, Netherlands
| | - Georg Heinze
- Medical University of Vienna, Center for Medical Statistics, Informatics and Intelligent Systems, Vienna, Austria
| | - Daniela Dunkler
- Medical University of Vienna, Center for Medical Statistics, Informatics and Intelligent Systems, Vienna, Austria
| |
Collapse
|
5
|
Ravani P, Maas R, Malberti F, Pecchini P, Mieth M, Quinn R, Tripepi G, Mallamaci F, Zoccali C. Homoarginine and mortality in pre-dialysis chronic kidney disease (CKD) patients. PLoS One 2013; 8:e72694. [PMID: 24023762 PMCID: PMC3762798 DOI: 10.1371/journal.pone.0072694] [Citation(s) in RCA: 32] [Impact Index Per Article: 2.9] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/15/2013] [Accepted: 07/11/2013] [Indexed: 11/18/2022] Open
Abstract
Background and Aims Homoarginine, a precursor of nitric oxide, is an inverse predictor of death in dialysis patients and in subjects with cardiovascular disease and normal kidney function but its relationship with clinical outcomes in chronic kidney disease (CKD) patients not yet on dialysis is unknown. Design, setting, participants and measurements We enrolled 168 consecutive predialysis CKD patients (Age: 70±11 yrs; 26% Diabetics; eGFR 34±18 ml/min/1.73 m2) referred to a tertiary care centre and measured laboratory data on kidney function and cardiovascular risk factors. We modeled progression to dialysis or death as a function of homoarginine, using Cox’s regression, accounting for clinical characteristics, baseline levels of kidney function, and markers of inflammation. Results On crude and adjusted analyses homoarginine was directly associated with the eGFR and patients with more compromised renal function exhibited lower homoarginine levels. Furthermore homoarginine was also independently related to L-arginine, serum albumin and body mass index, and inversely related to proteinuria, C-reactive protein and age. During the study (follow up median time 4 years, inter-quartile range 1.7 to 7.0 years) 56 patients started dialysis and 103 died and homoarginine was a strong inverse predictor of the incidence rate of both outcomes (P = 0.002 and P = 0.017). Conclusions Homoarginine declines with advancing renal disease and is inversely related to progression to dialysis and mortality. The nature of the link between homoarginine and clinical outcomes is amenable to testing in clinical trials.
Collapse
Affiliation(s)
- Pietro Ravani
- Department of Medicine and Faculty of Medicine, University of Calgary, Calgary, Alberta, Canada
| | - Renke Maas
- Institute of Clinical and Experimental Pharmacology and Toxicology, Friedrich-Alexander Universität Erlangen-Nürnberg, Nuremberg, Germany
| | - Fabio Malberti
- Divisione di Nefrologia, Azienda Istituti Ospitalieri di Cremona, Cremona, Italy
| | - Paola Pecchini
- Divisione di Nefrologia, Azienda Istituti Ospitalieri di Cremona, Cremona, Italy
| | - Maren Mieth
- Institute of Clinical and Experimental Pharmacology and Toxicology, Friedrich-Alexander Universität Erlangen-Nürnberg, Nuremberg, Germany
| | - Robert Quinn
- Department of Medicine and Faculty of Medicine, University of Calgary, Calgary, Alberta, Canada
| | - Giovanni Tripepi
- Divisione di Nefrologia, Ospedali Riuniti and CNR IBIM, Reggio Calabria, Italy
| | - Francesca Mallamaci
- Divisione di Nefrologia, Ospedali Riuniti and CNR IBIM, Reggio Calabria, Italy
| | - Carmine Zoccali
- Divisione di Nefrologia, Ospedali Riuniti and CNR IBIM, Reggio Calabria, Italy
- * E-mail:
| |
Collapse
|
6
|
Ravani P, Gillespie BW, Quinn RR, MacRae J, Manns B, Mendelssohn D, Tonelli M, Hemmelgarn B, James M, Pannu N, Robinson BM, Zhang X, Pisoni R. Temporal risk profile for infectious and noninfectious complications of hemodialysis access. J Am Soc Nephrol 2013; 24:1668-77. [PMID: 23847278 DOI: 10.1681/asn.2012121234] [Citation(s) in RCA: 79] [Impact Index Per Article: 7.2] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/03/2022] Open
Abstract
Vascular access complications are a major cause of morbidity in patients undergoing hemodialysis, and determining how the risks of different complications vary over the life of an access may benefit the design of prevention strategies. We used data from the Dialysis Outcomes and Practice Patterns Study (DOPPS) to assess the temporal profiles of risks for infectious and noninfectious complications of fistulas, grafts, and tunneled catheters in incident hemodialysis patients. We used longitudinal data to model time from access placement or successful treatment of a previous complication to subsequent complication and considered multiple accesses per patient and repeated access complications using baseline and time-varying covariates to obtain adjusted estimates. Of the 7769 incident patients identified, 7140 received at least one permanent access. During a median follow-up of 14 months (interquartile range, 7-22 months), 10,452 noninfectious and 1131 infectious events (including 551 hospitalizations for sepsis) occurred in 112,085 patient-months. The hazards for both complication types declined over time in all access types: They were 5-10 times greater in the first 3-6 months than in later periods after access placement or a remedial access-related procedure. The hazards declined more quickly with fistulas than with grafts and catheters (P<0.001; Weibull regression). These data indicate that risks for noninfectious and infectious complications of the hemodialysis access decline over time with all access types and suggest that prevention strategies should target the first 6 months after access placement or a remedial access-related procedure.
Collapse
|
7
|
Ravani P, Parfrey P, MacRae J, James M, Quinn R, Malberti F, Brunori G, Mandolfo S, Tonelli M, Hemmelgarn B, Manns B, Barrett B. Modeling survival of arteriovenous accesses for hemodialysis: semiparametric versus parametric methods. Clin J Am Soc Nephrol 2010; 5:1243-8. [PMID: 20413435 DOI: 10.2215/cjn.06190809] [Citation(s) in RCA: 11] [Impact Index Per Article: 0.8] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 01/29/2023]
Abstract
BACKGROUND AND OBJECTIVES Comparing outcomes of arteriovenous grafts and fistulas is challenging because the pathophysiology of access dysfunction and failure rate profiles differ by access type. Studying how risks vary over time may be important. DESIGN, SETTING, PARTICIPANTS, & MEASUREMENTS Longitudinal data from 535 incident hemodialysis patients were used to study the relationship between access type and access survival, without (semiparametric Cox modeling) and with specification of the underlying hazard function (parametric Weibull modeling). RESULTS The hazard for failure of fistulas and grafts declined over time, becoming proportional only after 3 months from surgery, with a graft versus fistula hazard ratio of 3.2 (95% confidence interval 1.9 to 5.3; Cox and Weibull estimation) and time ratio of 0.11 (i.e., the estimated access survival time was approximately one tenth shorter in grafts; 95% confidence interval 0.04 to 0.28; Weibull estimation only). Considering the entire observation period, grafts had slower hazard decline (P<0.001) with shorter median survival times than fistulas (8.4 versus 38.3 months; Weibull regression only). CONCLUSIONS Parametric models of arteriovenous access survival may provide relevant information about temporal risk profiles and predicted survival times.
Collapse
Affiliation(s)
- Pietro Ravani
- Department of Medicine, Division of Nephrology, University of Calgary, Foothills Medical Centre, 1403 29th St NW, Calgary, AB, T2N2T9, Canada.
| | | | | | | | | | | | | | | | | | | | | | | |
Collapse
|
8
|
|