From: Bjarte Aagnes
Sent: torsdag 10. mars 2022 16:08
To: Marnar Fríðheim Kristiansen marfr@ls.fo
Cc: Siri Larønningen sla@kreftregisteret.no
Subject: Regarding 10 year survival calendar periods
Hi Marnar,
You are right remembering a difference between criteria between the 5- and 10-year calendar periods. This was most likely presented in some presentation and/or discussion about how to get more estimates for smaller groups
Hi, the point 5 in inclusion criteria should be the same. I will read your response carefully and
review the code next week to assure the documentation reflect the code and vice versa.
I read it more thouroughly this morning, and understood it better. The main difference i guess, is both that the 10 year calendar period gives room for inclusion of more patients, but also that the two youngest age-groups and two oldest age-groups are combined leading to more cancers with small numbers to have enough patients for inclusion.
I thought I remembered that you had mentioned different inclusion criteria for the 5-year calendar period and 10-year. Specifically I thought that point 5 in the inclusion criteria i copied below differed for the two analyses. But i might just have misremembered this?
Inclusion
Set of predefined groups of entities
Patients are included with first case of each entity group to be analysed
DCO cases are not included
Patients 90 years or older are excluded
Groups are analysed if minimum 30 patients alive at start with minimum 3 in any one of age-groups used for weighting
So I guess my question is: Is point 5 in inclusion criteria defined here, used for both 5- and 10 year calendar periods?
Thanks for asking. It is not obvious to me what is missing. Thus, I give a short description of the current documentation:
The difference between the 5- and 10-year calendar periods is in the definition of follow-up. The definitions are described for both in the wiki and the follow-up definition for 10-year periods is described as follows:
Ten-year calendar periods:
5-year survival: cohort approach for all but last ten-year period for which the complete approach was used.
10-year survival: cohort approach for all but last ten-year period for which the period approach was used.
The above can be compared to the description of definition of follow-up for the “five-year calendar periods”.
The “rules of inclusion” is the same for 5- and 10-year calendar periods, and is describe as follows in the wiki:
Inclusion
Set of predefined groups of entities
Patients are included with first case of each entity group to be analysed
DCO cases are not included
Patients 90 years or older are excluded
Groups are analysed if minimum 30 patients alive at start with minimum 3 in any one of age-groups used for weighting
Maybe you could elaborate what you think is missing or unclear?
From: Bjarte Aagnes Sent: torsdag 10. mars 2022 16:08 To: Marnar Fríðheim Kristiansen marfr@ls.fo Cc: Siri Larønningen sla@kreftregisteret.no Subject: Regarding 10 year survival calendar periods
Hi Marnar,
You are right remembering a difference between criteria between the 5- and 10-year calendar periods. This was most likely presented in some presentation and/or discussion about how to get more estimates for smaller groups
However, the current implementation has no difference. The same criterium is (3) used for both 5 – and 10-year aggregation. This is the same rule used for Cancer In Norway. (internal ref: https://gitlab.kreftregisteret.no/CIN/2020/report_statistics/-/blob/master/statistics/progdef/krg_ns_check_conditions.ado)
The relevant part of the code is:
https://github.com/CancerRegistryOfNorway/nordcansurvival/blob/master/inst/stata/ado/extract_define_survival_data.ado
local noobs 3 // lower limit for n in age stratum local tot 30 // lower limit for N in group to be analysed
mark_small_n_strata , /// idvar(spid) /// groups(entity sex period_5) ///
agegr(agegroup_ICSS_5) /// nobs(
noobs') /// tot(tot')mark_small_n_strata , /// idvar(spid) /// groups(entity sex period_10) ///
agegr(agegroup_ICSS_3) /// nobs(
noobs') /// tot(tot')And described in the documentation: https://github.com/CancerRegistryOfNorway/NORDCAN/wiki/nordcansurvival as follows:
Thus, the documentation is consistent with the implementation.
Best wishes, Bjarte
From: Marnar Fríðheim Kristiansen [marfr@ls.fo](mailto:marfr@ls.fo) Sent: lørdag 5. mars 2022 16:22 To: Bjarte Aagnes [baa@kreftregisteret.no](mailto:baa@kreftregisteret.no) Subject: Sv: Regarding 10 year survival calendar periods
Thank you for your reply.
It is probably me that misremembered. But I remembered it, as if the criteria was the following for 5 year calendar period:
and for 10 year calendar period i remembered it as:
But this is probably just me misremembering.
Marnar
Fra: Bjarte Aagnes [baa@kreftregisteret.no](mailto:baa@kreftregisteret.no) Sendt: 5. marts 2022 14:28:41 Til: Marnar Fríðheim Kristiansen Emne: Regarding 10 year survival calendar periods
Hi, the point 5 in inclusion criteria should be the same. I will read your response carefully and review the code next week to assure the documentation reflect the code and vice versa.
Thanks again for your feedback.
-- Bjarte
From: Marnar Fríðheim Kristiansen [marfr@ls.fo](mailto:marfr@ls.fo) Sent: lørdag 5. mars 2022 12:20 To: Bjarte Aagnes [baa@kreftregisteret.no](mailto:baa@kreftregisteret.no) Subject: Sv: Regarding 10 year survival calendar periods
Hi Bjarte
Thank you for your quick reply.
I read it more thouroughly this morning, and understood it better. The main difference i guess, is both that the 10 year calendar period gives room for inclusion of more patients, but also that the two youngest age-groups and two oldest age-groups are combined leading to more cancers with small numbers to have enough patients for inclusion.
I thought I remembered that you had mentioned different inclusion criteria for the 5-year calendar period and 10-year. Specifically I thought that point 5 in the inclusion criteria i copied below differed for the two analyses. But i might just have misremembered this?
Inclusion
So I guess my question is: Is point 5 in inclusion criteria defined here, used for both 5- and 10 year calendar periods?
Sincerely, Marnar
Fra: Bjarte Aagnes [baa@kreftregisteret.no](mailto:baa@kreftregisteret.no) Sendt: 5. marts 2022 11:03:05 Til: Marnar Fríðheim Kristiansen Emne: Regarding 10 year survival calendar periods
Hi Marnar,
Thanks for asking. It is not obvious to me what is missing. Thus, I give a short description of the current documentation:
The difference between the 5- and 10-year calendar periods is in the definition of follow-up. The definitions are described for both in the wiki and the follow-up definition for 10-year periods is described as follows:
Ten-year calendar periods:
The above can be compared to the description of definition of follow-up for the “five-year calendar periods”.
The “rules of inclusion” is the same for 5- and 10-year calendar periods, and is describe as follows in the wiki:
Inclusion
Maybe you could elaborate what you think is missing or unclear?
Best wishes Bjarte
From: Marnar Fríðheim Kristiansen [marfr@ls.fo](mailto:marfr@ls.fo) Sent: fredag 4. mars 2022 21:22 To: Bjarte Aagnes [baa@kreftregisteret.no](mailto:baa@kreftregisteret.no) Subject: Regarding 10 year survival calendar periods
Dear Bjarte
I have read about the used method for calculating relative survival on NORDCAN here: https://github.com/CancerRegistryOfNorway/NORDCAN/wiki/nordcansurvival
However, if I remember correctly, the inclusion criteria was different for the 10 year calendar periods. Is that inclusion method described somewhere?
Bw Marnar