For evaluation of timeliness, all incident cases reported to the NBCR in 2013 were included (N = 8654), and the difference in time between the earliest date of diagnosis and the reporting date in the registry was calculated.
Completeness was assessed by comparing the cases in the NBCR with registrations in the Swedish Cancer Registry (SCR) [9 ], to which reporting is mandatory according to the National Board of Health and Welfare’s regulations (SOSFS2006:15). Data from the time period 2010–2014 was used. The completeness of the SCR is secured as any diagnosed cancer case is reported by the clinician and from the pathology laboratory after verification of morphological examinations i e biopsies and autopsy. Two publications describe in detail the process [10 (link), 11 (link)].
Comparability refers to the recording and coding practices and should be clear, nationally uniform and follow international guidelines to enable comparisons between regions and countries. Inclusion criteria are: location (primary breast cancer); sex (women and men); age (all ages); morphology (invasive breast cancer and carcinoma in situ); basis for diagnosis (all cases except diagnosis at autopsy).
Two control functions secure comparability. Firstly, the manual and the report form are unique documents. Secondly, monitoring is performed at the regional cancer centers whereby adherence to inclusion criteria and or any erroneously reported data and or ambiguity will be corrected.
Comparability concerning the workflow was assessed by a questionnaire addressing how different breast units handled reporting routines, involved staff, time allotted, and management support [12 ].
To assess validity, re-abstracted data from medical records was compared to the reported data via an independent review process. Eight hundred recorded cases between September 2013 and January 2014, were randomly selected using a two-stage cluster sampling plan.
Two hospitals offering breast cancer services (ranked according to size) from each health care region were selected. Within each region (cluster), a subsample of all breast cancer patient records in the 12 selected hospitals were drawn with a probability proportional to the size of region and hospital. The sampling plan was chosen to ensure national representation as well as participation from both large and small breast cancer units.
Re-abstraction of medical records took place in the second part of 2014 and was performed by three specialist nurses with previous experience in register validation and monitoring, henceforth referred to as validators. The re-abstracted information was entered into a specially designed module and subsequently merged with the originally recorded data to calculate exact data agreement. Exact agreement corresponds to the proportion of women for whom the data recorded in the NBCR is the same as in the validation data set. Missing observations were also included in the calculations of exact agreement to account for the plausible situations when 1) data had been reported to the NBCR but could not be found in the medical records, 2) the information was available in the medical records but had not been reported to the NBCR. Strength of agreement was measured by Cohen’s Kappa (κ) scores for categorical variables, including 95% confidence intervals (CI), and Pearson correlation coefficients (r) for numerical variables.
Completeness was assessed by comparing the cases in the NBCR with registrations in the Swedish Cancer Registry (SCR) [9 ], to which reporting is mandatory according to the National Board of Health and Welfare’s regulations (SOSFS2006:15). Data from the time period 2010–2014 was used. The completeness of the SCR is secured as any diagnosed cancer case is reported by the clinician and from the pathology laboratory after verification of morphological examinations i e biopsies and autopsy. Two publications describe in detail the process [10 (link), 11 (link)].
Comparability refers to the recording and coding practices and should be clear, nationally uniform and follow international guidelines to enable comparisons between regions and countries. Inclusion criteria are: location (primary breast cancer); sex (women and men); age (all ages); morphology (invasive breast cancer and carcinoma in situ); basis for diagnosis (all cases except diagnosis at autopsy).
Two control functions secure comparability. Firstly, the manual and the report form are unique documents. Secondly, monitoring is performed at the regional cancer centers whereby adherence to inclusion criteria and or any erroneously reported data and or ambiguity will be corrected.
Comparability concerning the workflow was assessed by a questionnaire addressing how different breast units handled reporting routines, involved staff, time allotted, and management support [12 ].
To assess validity, re-abstracted data from medical records was compared to the reported data via an independent review process. Eight hundred recorded cases between September 2013 and January 2014, were randomly selected using a two-stage cluster sampling plan.
Two hospitals offering breast cancer services (ranked according to size) from each health care region were selected. Within each region (cluster), a subsample of all breast cancer patient records in the 12 selected hospitals were drawn with a probability proportional to the size of region and hospital. The sampling plan was chosen to ensure national representation as well as participation from both large and small breast cancer units.
Re-abstraction of medical records took place in the second part of 2014 and was performed by three specialist nurses with previous experience in register validation and monitoring, henceforth referred to as validators. The re-abstracted information was entered into a specially designed module and subsequently merged with the originally recorded data to calculate exact data agreement. Exact agreement corresponds to the proportion of women for whom the data recorded in the NBCR is the same as in the validation data set. Missing observations were also included in the calculations of exact agreement to account for the plausible situations when 1) data had been reported to the NBCR but could not be found in the medical records, 2) the information was available in the medical records but had not been reported to the NBCR. Strength of agreement was measured by Cohen’s Kappa (κ) scores for categorical variables, including 95% confidence intervals (CI), and Pearson correlation coefficients (r) for numerical variables.
Free full text: Click here
