Matching methods to create paired survival data based on an exposure occurring over time: a simulation study with application to breast cancer

BMC Medical Research Methodology

Yann De Rycke¹  Pascale Tubert-Bitter³  Caroline Giard²  Alexia Savignoni³ 

[1] Institut Curie, Public Health Team, Paris, France;Institut Curie, Pharmacological Unit, Saint-Cloud, France;Univ Paris-Sud, UMRS1018, F-94807 Villejuif, France

关键词: Simulation study;    Breast cancer;    Pregnancy;    Marginal Cox model;    Stratified Cox model;    Event occurring over time;    Correlated survival data;    Matched time-to-event data;    Matching on time-dependent covariates;   

Others  :  1091451
DOI  :  10.1186/1471-2288-14-83

 received in 2013-11-08, accepted in 2014-06-11,  发布年份 2014

【 摘 要 】

Background

Paired survival data are often used in clinical research to assess the prognostic effect of an exposure. Matching generates correlated censored data expecting that the paired subjects just differ from the exposure. Creating pairs when the exposure is an event occurring over time could be tricky. We applied a commonly used method, Method 1, which creates pairs a posteriori and propose an alternative method, Method 2, which creates pairs in “real-time”. We used two semi-parametric models devoted to correlated censored data to estimate the average effect of the exposure View MathML">: the Holt and Prentice (HP), and the Lee Wei and Amato (LWA) models. Contrary to the HP, the LWA allowed adjustment for the matching covariates (LWA_a) and for an interaction (LWA_i) between exposure and covariates (assimilated to prognostic profiles). The aim of our study was to compare the performances of each model according to the two matching methods.

Methods

Extensive simulations were conducted. We simulated cohort data sets on which we applied the two matching methods, the HP and the LWA. We used our conclusions to assess the prognostic effect of subsequent pregnancy after treatment for breast cancer in a female cohort treated and followed up in eight french hospitals.

Results

In terms of bias and RMSE, Method 2 performed better than Method 1 in designing the pairs, and LWA_a was the best model for all the situations except when there was an interaction between exposure and covariates, for which LWA_i was more appropriate. On our real data set, we found opposite effects of pregnancy according to the six prognostic profiles, but none were statistically significant. We probably lacked statistical power or reached the limits of our approach. The pairs’ censoring options chosen for combination Method 2 - LWA had to be compared with others.

Conclusions

Correlated censored data designing by Method 2 seemed to be the most pertinent method to create pairs, when the criterion, which characterized the pair, was an exposure occurring over time. In such a setting, the LWA was the most appropriate model.

【 授权许可】

   
2014 Savignoni et al.; licensee BioMed Central Ltd.

【 图 表 】

Figure 1.

Figure 2.

Figure 3.

Figure 4.

Figure 5.

【 参考文献 】

• [1]Kelsey J, Berkowitz GS: Breast cancer epidemiology. Cancer Res 1998, 48:5615-5623.
• [2]Kroman N, Jensen M, Melbye M, Wohlfahrt J, Ejlertsen B: Pregnancy after treatment of breast cancer - a population-based study on behalf of danish breast cancer cooperative group. Acta Oncologica 2008, 47:545-549.
• [3]Azim H, Kroman N, Paesmans M, Gelber S, Rotmensz N, Ameye L, De Mattos-Arruda L, Pistilli B, Pinto A, Jensen M, Cordoba O, de Azambuja E, Goldhirsch A, Piccart M, Peccatori F: Prognostic impact of pregnancy after breast cancer according to estrogen receptor status: a multicenter retrospective study. J Clin Oncol 2013, 31:73-79.
• [4]Velentgas P, Daling J, Malone K, Weiss N, Williams M, Self S, Mueller BA: Pregnancy after breast carcinoma: Outcomes and influence on mortality. Cancer 1999, 85(11):2424-2432.
• [5]Kranick J, Schaefer C, Rowell S, Desai M, Petrek J, Hiatt R, Senie R: Is pregnancy after breast cancer safe? Breast J 2010, 16(4):404-411.
• [6]Lawrenz B, Banys M, Henes M, Neunhoeffer E, Grischke E, Fehm T: Pregnancy after breast cancer: case report and review of literature. Arch Gynecol Obstet 2011, 315:851-855.
• [7]Sankila R, Heinavaara S, Hakulinen T: Survival of breast cancer patients after subsequent term pregnancy: healthy mother effect. Am J Obstet Gynecol 1994, 170:818-823.
• [8]Ives A, Saunders C, Bulsara M, Semmens J: Pregnancy after breast cancer: population based study. BMJ 2007, 334:194-198.
• [9]Largillier R, Savignoni A, Gligorov J, Chollet P, Guilhaume M, Spielmann M, Luporsi E, Asselain B, Coudert B, Namer M: Prognostic role of pregnancy occurring before or after treatment of early breast cancer patients aged <35 years: a get(n)a working group analysis. Cancer 2009, 115(22):5155-5165.
• [10]Mueller B, Simon M, Deapen D, Kamineni A, Malone K, Daling J: Childbearing and survival after breast carcinoma in young women. Cancer 2003, 98(6):1131-1140.
• [11]Cox D: Regression models and life-tables. J Royal Stat Soc (B) 1972, 34:187-220.
• [12]Andersen P, Keiding N: Multi-state models for event history analysis. Stat Methods Med Res 2002, 11:91-115.
• [13]de Bock G, Putter H, Bonnema J, van der Hage J, Bartelink H, van de Velde C: The impact of loco-regional recurrences on metastatic progression in early-stage breast cancer: a multistate model. Breast Cancer Res Treat 2009, 117:401-408.
• [14]Putter H, Fiocco M, Geskus R: Tutorial in biostatistics: Competing risks and multi-state models. Stat Med 2006, 26:2389-2430.
• [15]Savignoni A, Hajage D, Tubert-Bitter P, De Rycke Y: Effect of an event occurring over time and confounded by health status: estimation and interpretation. a study based on survival data simulations with application on breast cancer. Stat Med 2012, 31:4444-4455.
• [16]Holt JD, Prentice RL: Survival analyses in twin studies and matched pair experiments. Biometrika 1974, 61(1):17-25.
• [17]Vaupel JW, Manton KG, Stallard E: The impact of the heterogeneity in individual frailty on the dynamics of mortality. Demography 1979, 16(1):439-454.
• [18]Sarpel U, Hefti M, Wisnievsky J, Roayaie S, Schwartz M, Labow D: Outcome for patients treated with laparoscopic versus open resection of hepatocellular carcinoma: case-matched analysis. Ann Surg Oncol 2009, 16:1572-1577.
• [19]Wei L, Lin D, Weissfeld L: Regression analysis of multivariate incomplete failure time data by modelling marginal distributions. J Am Stat Assoc 1989, 84:1065-1073.
• [20]Lee E, Wei L, Amato D: Cox-type regression analysis for large numbers of small groups of correlated failure time observations. In Survival Analysis: State of the Arts. Edited by Klein JP, Goel PK. Netherlands: Springer; 1992:237-247.
• [21]Lin DY: Cox regression analysis of multivariate failure time data: the marginal approach. Stat Med 1994, 13:2233-2247.
• [22]Lu S, Shih J: Case-cohort designs and analysis for clustered failure time data. Biometrics 2006, 62:1138-1148.
• [23]Woolson R, O’Gorman T: A comparison of several tests for censored paired data. Stat Med 1992, 11:193-208.
• [24]O’Brien P, Fleming T: A paired prentice wilcoxon test for censored paired data. Biometrics 1987, 43:169-180.
• [25]Manatunga A, Oakes D: Parametric analysis for matched pair survival data. Lifetime Data Anal 1999, 5:371-387.
• [26]Mehrotra D, Su S, Li X: An efficient alternative to the stratified cox model analysis. Stat Med 2012, 31:1849-1856.
• [27]Katsahian S, Latouche A, Mary J, Chevret S, Porcher R: Practical methodology of meta-analysis of individual patient data using a survival outcome. Contemp Clin Trials 2008, 29(2):220-230.
• [28]R Core Team: R: A Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical Computing; 2013. [http://www.R-project.org webcite]