is each sibling's death maternal, by DHS coding?

As is_preg_related_dhs, but stopping at mm9 = 5 — the 42-day window — and excluding deaths reported as due to violence or an accident (mm16). Only computable when sib.died.accident is present, which is DHS phase 7 and later, and in practice only 5 of the 43 surveys with a sibling roster carry it.

Usage

is_maternal_dhs(sib_df, na.action = NULL)

Arguments

sib_df

the prepped sibling dataset

na.action

retained for symmetry and ignored. It used to decide how a missing mm12 was treated; mm12 is no longer consulted

Value

a logical vector, one entry per row of sib_df

Details

The rule is the reference implementation's, verbatim (DHS-Indicators-Stata, Chap16_AM/AM_rates.do:728):

mm9 >= 2 & mm9 <= 5 & mm16 != 1 & mm16 != 2

Two details that look like edge cases but are not:

• A missing mm16 counts as "not an accident". mm16 is not asked when the death occurred during delivery, so for mm9 = 3 it is missing by design — in The Gambia 2019-20 all 40 such in-window deaths have mm16 missing and none of the mm9 = 2 or mm9 = 5 deaths do. Requiring mm16 == 0 would silently drop every delivery death.

• mm9 = 4 is included even though it is never assigned in practice. That matches the reference, which tests a range rather than a set.

Changed in this version

This function previously applied the accident exclusion to codes 2 and 5 only, took code 3 unconditionally, and additionally required mm12 to fall in the band 100–141. Given the skip pattern above, the first two are equivalent to the reference wherever a survey respects it — The Gambia 2019-20 reproduces its published Table 14.3 exactly either way.

They part company where a survey does not. South Africa 2016 has 3 deaths coded mm9 = 3 with mm16 reported as violence or an accident, which the old rule counted as maternal and the reference does not. The mm12 condition is dropped for the same reason it was dropped from is_preg_related_dhs: the reference does not use it.