The purpose of this study was to characterize the recurrence dynamics

The purpose of this study was to characterize the recurrence dynamics in breast cancer patients after delayed reconstruction. is mandatory for all physicians in Norway and the latest published evaluation from 2007 showed a 99?% completeness of data [30]. For ER81 data quality purposes patient’s records were studied for validation of diagnosis patient and tumor characteristics adjuvant therapy reconstructive surgery (excluded from the control group) RS-127445 time of first recurrence and recurrent site in the same way as was done with the cases. Among the 1341 patients a total of 473 patients were excluded (see Fig.?1 for details) leaving 868 patients whose characteristics are shown in Table?1 which hereafter will be labeled “control group.” Table?1 Patient tumor and treatment characteristics Matching For each patient in the reconstruction group all patients in the control group with identical T and N stages age?±?2?years and follow-up without recurrence equal to or longer than the time to reconstruction of the respective matched reconstructed patient were considered. In this initial step each case could have a number of candidate controls of 0-X. A was calculated for each of the controls in these groups representing time from primary surgery for the control plus time from primary surgery until reconstruction for the matched case. Therefore time from primary surgery until could by calculated for cases and controls respectively. A paired control was randomly selected from this group. If this group was empty increased age interval up to 5?years was allowed as a first step and in a few cases when the age difference was considered clinically relevant (e.g. pre- vs postmenopausal) patients with similar but not identical T classification (e.g. T2 instead of T1) within the right age interval were considered. This group of 312 patients whose characteristics are shown in Table? 1 will be hereafter labeled “matched control group.” RS-127445 Follow-up Time to recurrence (TTR) was recorded as the time from primary surgery to recurrence. The endpoint of primary interest was the first evidence of recurrence: survival times were calculated as the time elapsed since primary surgery to recurrence or to the last documented follow-up with no evidence of disease. Both locoregional recurrence and distant metastasis were defined as the events of interest whereas all new primary tumors including contralateral breast cancers were considered competing events thus for these patients survival times were censored at the time of their occurrence. Adjuvant local and/or systemic treatment was given according to national guidelines at the given time period and was not affected by delayed breast reconstruction. Follow-up after curative breast cancer treatment in Norway does not include radiologic evaluation or blood samples other than upon clinical suspicion of distant metastases. Thus diagnosis of relapse is most commonly made after patients’ experience of symptoms. Even when adopting more meticulous follow-up regimens more than 85?% of recurrences are recognized following symptomatic alert RS-127445 and not at settings [31]. Oncological follow-up is not affected by reconstructive surgery. Statistical analysis The event dynamics were analyzed by estimating with the life-table method the risk rate for recurrence i.e. the conditional probability of manifesting recurrence given that the patient is definitely clinically free from any recurrence at the beginning of the interval. The probability of recurrence over time i.e. crude cumulative incidence (CCI) was estimated according to a proper nonparametric estimator modifying for the presence of competing events and compared from the Gray test [32]. A discretization of the time axis in six-month devices was applied and a Kernel-like smoothing process [33] was used. For multivariable regression analysis the piecewise exponential model was used. The piecewise exponential model provides a flexible semiparametric tool in the study of the risk function for survival data in the same fashion like a Cox regression model [34]. The log-hazard function was modeled as an additive function of the baseline log-hazard and the covariate effects. For estimation of the piecewise exponential model the follow-up time was split into 3-month disjoint intervals and the event rate was assumed to be constant within each interval. The model accounts for reconstruction like a time-dependent covariate (i.e. RS-127445 switching from 0 to 1 1 at the time it was performed). The model was prolonged to account for the new.