Information identified as archived is provided for reference, research or record keeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please contact us to request a format other than those available.
To: All users of the Survey on Living with Neurological Conditions in Canada—Canadian Community Health Survey linked file
Subject: Error in CCCDDIA logic
Survey on Living with Neurological Conditions in Canada—Canadian Community Health Survey linked file
Description of the problem(s):
The Diabetes Type derived variable (CCCDDIA) has a flaw in the logic that allowed respondents who should have been coded to gestational diabetes to be coded as Type 2 diabetes or ‘not stated’. Female respondents who answered CCC_10A=1 and CCC_10B=2 should have been classified as gestational diabetes. However, several cases also met the condition for Type 2 diabetes or for not stated. Because these other categories were derived first, these cases were coded as such, before the check for gestational diabetes could be run.
Additionally, all of the cases that were coded as ‘4’ could actually be correctly coded as type 2 diabetes. Because of the faulty logic, these cases did not meet any of the conditions and would not have been coded. This is likely the reason for the creation of the category 4 in the first place.
According to the Ng-Dasgupta-Johnson algorithm, upon which CCCDDIA is based, gestational diabetes should be derived first. In practice however, it is typical to list the DV code types in ascending order after the processing codes are derived.
To preserve the order of DV codes and fix the problems, the conditions for ‘not stated’ and type 2 have been rewritten with an extra condition that ensures there is no overlap with gestational diabetes. The new rule for type 2 diabetes now also covers all of the cases that would have otherwise been left uncoded and assigned a value of 4 – ‘unable to classify’. The new DV code now properly classifies all respondents without the need for this catch-all code. The corrected code for the derived variable is:
|6||CCC_101 > 1||Population exclusions||NA|
|9||(CCC_10A in (7,8,9) or
CCC_10B in (7,8,9) or
CCC_10C in (97,98,99) or
CCC_101 in (7,8,9) or
CCC_102 in (997,998,999) or
CCC_105 in (7,8,9) or
CCC_106 in (7,8,9)) and (CCC_10A in (2,6,7,8,9) or
CCC_10B in (1,6,7,8,9))
|At least one required question
was not answered (don’t know, refusal, not stated).
Also, gestational diabetes cannot be derived.
|1||(DHH_SEX = 1 and
CCC_101 = 1 and
CCC_105 = 1 and
CCC_106 = 2 and
(CCC_10C <=3 and (DHH_AGE < 30 or
CCC_102 < 30))) or
(DHH_sex = 2 and CCC_101 = 1 and CCC_10B in (1,6) and CCC_105 = 1 and CCC_106 = 2 and (CCC_10C <=3 and (DHH_AGE < 30 or
CCC_102 < 30)))
|Type 1 diabetes|
|2||CCC_101 = 1 and
(CCC_105 = 2 or
CCC_102 >= 30 or
CCC_10C > 3 or
CCC_106 = 1 or
(CCC_105 = 1 and (CCC_102 >= 30 or CCC_10C > 3))) and (CCC_10A in (2,6,7,8,9) or
CCC_10B in (1,6,7,8,9))
|Type 2 diabetes|
|3||CCC_101 = 1 and DHH_SEX = 2 and CCC_10A = 1 and CCC_10B = 2||Gestational diabetes|
“Patch files” have been created for all affected files. They contain all records with the unique record identifier(s), as well as the correctly derived CCCDDIA. These patches can be used to replace CCCDDIA from the affected files by merging with the record identifier(s).
Errata “patch” files are available and should be merged with the existing data files if CCCDDIA is being used in analysis.
We regret any inconvenience this may have caused you or your organization and thank you in advance for your understanding.
Should you have any questions, please do not hesitate to contact us at: Health Statistics Division
Electronic mail: firstname.lastname@example.org