Appendix 6.
NCDS cases where one or more outcomes had year entered as ‘9999’,
so that chronological order could not be easily ascertained.
We looked at dates of other outcomes: where there were at least two other proper
dates entered, we could guess whether the whole lot were put in the wrong order or
were wrongly bunched. Sometimes it was possible to get information from the type
of outcome. Summary of types:-
(a) No bunching, order seems correct based on at least two other dates entered (or
where one other date/type of outcome makes it clear, e.g. ‘still pregnant’). Leave as
is.
(b) No bunching, and no way of telling if order correct. Leave as is.
(c) Bunched in same pregnancy as another, but no way of telling if bunching is correct
(otherwise chronological order seen to be correct from at least two other dates).
Leave as is.
(d) Bunched in same pregnancy as another, but no way of telling if bunching is
correct (also no way of telling if chronological order is correct). Leave as is.
(e) Only one baby ever carried, so no question about bunching or order. Leave as is.
(f) Miscellaneous complex cases where no choice but to leave as is. See comment.
(g) No bunching, but order incorrect on at least two other dates (or where one other
date is so late as to leave no room). Reverse order.
(h) Bunching, with 2 or more others which should clearly not be bunched because >9
months apart (but correct chronol. order). Spread to slots 1,6,11, etc.
(j) Bunching, with 2 or more others which should clearly not be bunched because >9
months apart (and chronol. order wrong as well). Reverse order and spread to slots
1,6,11, etc.
(k) Bunching, with one other specific date, both live births. Assume order correct but
should be two different pregnancies. Spread to slots 1,6.
(l) Miscellaneous complex cases needing alteration.
nserial type changed? Comments
041034A j changed
081043C b
082020V e
084017T a
088006K a
093273P b changed Other pre-March 1991 birth event deleted.