• 2019-07
  • 2019-08
  • 2019-09
  • 2019-10
  • 2019-11
  • 2020-03
  • 2020-07
  • 2020-08
  • 2021-03
  • Prostaglandin J2 Our results are consistent with


    Our results are consistent with results from simulation studies that pointed out that estimates are not influenced noticeably by the number of knots, as long as a sensible number of knots is selected [13,14]. In large datasets, the AIC and BIC will select models with high df when a lower value for the df provide a similar fit. In our analysis, the selection criteria chose more complicated models but the differences in the estimates between the models selected by the selection criteria and the reference model were negligible.
    Declarations of interest
    Funding This work was supported by Cancer Research UK [Grant number C1483/A18262].
    Authorship contribution
    Introduction As prostate cancer is biologically heterogeneous, it Prostaglandin J2 is important to differentiate clinically between indolent, low-risk disease that is localised to the prostate and disease that is highly aggressive and likely to metastasise [1]. Management options for non-metastatic disease differ significantly depending on the clinical stage and grade of disease, ranging from active surveillance to radical local treatment (with or without supplementary systemic therapy). Patients presenting with distant metastases are managed differently with systemic and palliative treatments taking precedence [2]. Cancer registry and administrative hospital data provide a vast source of cancer-specific information [3]. These data are used extensively in population-based studies to investigate nationwide trends in cancer diagnosis and management [[3], [4], [5]]. When prostate cancer-specific information is not known, or simply not recorded, a patient’s cancer stage often cannot be determined. This is potentially problematic when these data are being used for the purposes of research and health service evaluation, especially when staging greatly affects prognosis and clinical management [3,6,7]. In many countries completeness of cancer registry data is improving over time but there is still a reliance on historical data, which are important in studying long-term outcomes and survival trends in prostate cancer [8,9]. The necessary omission of patients with incomplete data from analyses may introduce significant selection bias and will lead to loss of statistical power. Statistical methods, such as multiple imputation, are often used when dealing with missing data. These techniques re-classify missing values based on the patterns present in other variables within the sample. [4,[10], [11], [12], [13], [14]]. Staging data is often missing in certain clinical situations. For example, in prostate cancer registration, negative results for nodal involvement (N0) or distant metastases (M0) may be more likely to be missing than positive results (N1 and M1, respectively). This is a consequence of two factors: first, the staging practice, as patients with low/intermediate-risk prostate cancer may not have further staging investigations and second, record keeping, as negative results are less likely to be recorded than positive results. Specific clinical assumptions may therefore be used to impute missing staging items in cancer registry data in a systematic way in order to risk stratify more men with prostate cancer. We explored to what extent staging data completeness can be improved by using the following four clinical assumptions:
    Material and methods All patients who were diagnosed with prostate cancer between January 1st 2010 and December 31st 2013 were identified in the English cancer registry using the ICD-10 code ‘C61’ [15]. Data collected by the eight regional English cancer registries have been combined into a national data set. This was then linked at patient-level to Hospital Episode Statistics (HES), an administrative database of all hospital admissions in the English National Health Service and data from the Office for National Statistics (ONS), giving 139,807 patients over Codominant alleles time period. Follow-up was available to 31st December 2014. Data collected from the English cancer registry included age, Gleason score, and T-, N- and M-stages (TNM). TNM data used preferentially clinical cancer registry items and then pathological cancer registry items, in line with the Union for International Cancer Control (UICC) TNM 7th edition, taking staging information that was updated as much as possible by cancer registry staff.