add pregnancy-related and maternal death info to a sibling dataset

Usage

add_maternal_deaths(
  sib_df,
  style = c("dhs", "mics6", "mics4"),
  na.action = NULL,
  preg.window = c("2months", "42days"),
  prmr.accident.recode = FALSE,
  keep_missing = FALSE,
  verbose = TRUE
)

Arguments

sib_df: the prepped sibling dataset (probably from prep_dhs_sib_histories or prep_mics_sib_histories)
style: which questionnaire the coding follows: "dhs" (the default), "mics6" for MICS6/MICS7, or "mics4" for MICS4/MICS5
na.action: how to treat a death that falls in the right window but whose timing detail is missing – "include" counts it, "exclude" does not. Required for the MICS styles; see Details
preg.window: width of the postpartum window used for the pregnancy-related column under the MICS styles: "2months" (the default, and what this package has always done) or "42days", which is what published MICS tables report. Ignored for style = "dhs", which already applies a 42-day cut. See Details
prmr.accident.recode: DHS style only: apply the 2016 PRMR redefinition, which drops a death during pregnancy (mm9 = 2) reported as violence or an accident. Default FALSE, matching the reference implementation, which documents the rule but does not execute it. See is_preg_related_dhs
keep_missing: not currently used
verbose: report detailed summaries?

Value

sib_df with columns sib.preg_related.death.date and sib.maternal.death.date added

Details

Two quantities are computed, and they are not the same thing:

pregnancy-related – died while pregnant, during childbirth, or within the postpartum window, whatever the cause
maternal – as above but within 42 days, and excluding deaths due to violence or an accident

sib.maternal.death.date is only computable when the data identify accidental deaths, which means DHS phase 7 and later, or MICS6 and later. Otherwise it is NA and only the pregnancy-related column is usable – the same limitation applies to DHS phases 2–6 and to MICS4/MICS5.

Siblings who did not die of the relevant cause get a death date of -1 rather than NA, so that they still contribute exposure. Both columns are set to NA for male siblings.

Note for MICS users: the tables published in MICS survey reports under the heading "Maternal mortality", with a column labelled "Maternal Deaths", do not report maternal deaths as defined above. UNICEF's own tabulation syntax counts MM22 = 1 | MM23 = 1 | (MM24 = 1 & MM25 < 42) and never reads the violence (MM26) or accident (MM27) items at all, despite the footnote in the reports saying those causes are excluded. So the published column is a pregnancy-related count on a 42-day window. To reproduce it, use sib.preg_related.death.date with preg.window = "42days".

Choosing `na.action`

na.action has no default for the MICS styles, because it is a substantive choice about the estimand rather than a coding detail, and it should be made deliberately and reported.

In MICS, the number of days after the end of a pregnancy (MM25) is asked only of sisters who died within two months (MM24 = 1). When that day count is missing, na.action decides whether she falls inside the 42-day maternal window. In three MICS6 surveys examined this affected 5 of 38 such deaths in Iraq 2018 and none at all in Zimbabwe 2019 or Madagascar 2018, so the choice is usually immaterial – but not always, and it only ever moves the maternal column, never the pregnancy-related one.

na.action no longer does anything under style = "dhs". It governed how a missing sib.time.delivery.death (mm12) was treated, and mm12 is no longer consulted by either DHS column — the reference implementation states that "mm12 is not needed" and drops it. The argument is kept so existing DHS call sites keep working, and still applies to both MICS columns.

Examples

  # TODO - write example code