IMIC Growth outcome summaries
Andrew Mertens
2024-06-11
Notes for integrated analysis of growth outcomes:
- For the main integrated paper, only analyses the primary and
secondary outcomes to start
- Primary outcomes: LAZ and WHZ at 3 months and 6 months
- Secondary outcomes:
- Continuous: WAZ at 3 months and 6 months, HCAZ at 3 and 6 months, and length and weight velocity Z-scores from birth to 6 months and from 3-6 months
- Binary: Growth faltering from birth to 6 months (this variable is a classification based on both birth size and faltering from that birth size)
- Subgroup analysis: examine WHZ and HAZ at 6 months among small vulnerable newborn (variable “svn”==1)
- Binary outcomes to be classified from continuous outcomes post-hoc:
- Stunting: LAZ < -2
- Wasting: WLZ < -2
- Underweight: WAZ < -2
- Low head circumference: HCAZ < -2
- Severe stunting, wasting, underweight, head circumference: < -3 Z-score for each
- Make sure the breastmilk collection dates are prior to the growth measurements, especially for the 3 month measures and 3-6 month velocity measures. For the birth to 6 month secondary measures, you can use all milk measurements as we do not have temporality regardless.
- The window around ages to 1 month in either direction (so 6 month age is the measurement closest to 6 months that is between 5-7 months) based on variability in measurement timing. Birth measurements are within the first 14 days of life.
- Excluding WHO outlier Z-scores. Also excluding extreme MUAZ and HCAZ outlier, but can’t find WHO standards for these two. Currently using -5 to 5 for both to be conservative.
- Sensitivity analysis to excluding measures that lead to implausible length or head circumference decreased beyond the technical error of measurement. Excluding the measure with the greatest HAZ deviation of the before vs after measurements. For example, Subjido 1029 has implausible length decreases. and HAZ drops from ~4 to ~-1.5, so the measurement of 4 is dropped. Not currently dropping other non-length measurements from the same visit. But not large change in growth distribution with exclusion, so keeping all measures.
Age-specific means and prevalences by study of growth outcomes
Outcomes are listed and ordered by priority, and the outcome name and R variable name is listed. For binary variables, prevalence and (n/N) is listed, and for continuous variables the mean and (SD, N) is listed.
| Outcome | variable name | CHILD | ELICIT | MISAME-3 | VITAL-Lactation | Combined | Age | Outcome priority |
|---|---|---|---|---|---|---|---|---|
| Length-for-Age Z | haz_3mo | 0.41 (1.3, N=279) | -0.93 (1, N=198) | -0.66 (1.3, N=286) | -1.1 (1.1, N=116) | -0.44 (1.3, N=879) | 3mo | Primary |
| Length-for-Age Z | haz_6mo | -0.53 (1.8, N=51) | -1.12 (1.1, N=199) | -0.53 (1.2, N=278) | -0.91 (1.1, N=126) | -0.78 (1.2, N=654) | 6mo | Primary |
| Weight-for-Length Z | whz_3mo | -0.38 (1.2, N=278) | 0.3 (1.3, N=195) | 0.51 (1.2, N=286) | -0.38 (1.2, N=116) | 0.06 (1.3, N=875) | 3mo | Primary |
| Weight-for-Length Z | whz_6mo | -0.35 (1.4, N=51) | 0.32 (1.2, N=199) | -0.13 (1.1, N=278) | -0.72 (1.1, N=126) | -0.12 (1.2, N=654) | 6mo | Primary |
| Growth Faltering | growth_faltering_birth_6mo | 25.25% (50/198) | 19.06% (53/278) | 34.01% (50/147) | 24.56% (153/623) | 6mo | Secondary | |
| Head circumference Z-score | hcaz_3mo | 0.28 (1.3, N=277) | -0.23 (1.1, N=197) | -0.42 (1.1, N=285) | -1.14 (1, N=116) | -0.25 (1.2, N=875) | 3mo | Secondary |
| Head circumference Z-score | hcaz_6mo | -0.47 (1.4, N=52) | -0.3 (0.9, N=198) | -0.62 (1.1, N=278) | -1.09 (1, N=126) | -0.6 (1.1, N=654) | 6mo | Secondary |
| Length velocity Z-score 3-6 months | len_vel_z_3mo_6mo | -0.45 (2, N=195) | 0.15 (1.3, N=276) | 0.17 (1.2, N=100) | -0.05 (1.6, N=571) | 6mo | Secondary | |
| Length velocity Z-score birth-6 months | len_vel_z_birth_6mo | -1.44 (2.4, N=50) | -0.93 (1.7, N=197) | -0.01 (1.3, N=278) | -0.15 (1, N=94) | -0.44 (1.6, N=619) | 6mo | Secondary |
| Weight-for-Age Z | waz_3mo | -0.06 (1.1, N=279) | -0.56 (1, N=197) | -0.21 (1.1, N=286) | -1.22 (1.2, N=148) | -0.4 (1.2, N=910) | 3mo | Secondary |
| Weight-for-Age Z | waz_6mo | -0.88 (1.4, N=52) | -0.52 (1, N=200) | -0.48 (1.1, N=277) | -1.19 (1.2, N=145) | -0.67 (1.1, N=674) | 6mo | Secondary |
| Weight velocity Z-score 3-6 months | wt_vel_z_3mo_6mo | -0.21 (1.6, N=196) | -0.8 (1.2, N=277) | -0.37 (1.2, N=146) | -0.51 (1.3, N=619) | 6mo | Secondary | |
| Weight velocity Z-score birth-6 months | wt_vel_z_birth_6mo | -1 (1.6, N=52) | -0.34 (1.2, N=197) | -0.18 (1.2, N=278) | -0.62 (1.3, N=146) | -0.38 (1.2, N=673) | 6mo | Secondary |
Density plots of continious outcomes
Other growth variables (not used as outcomes)
Note! Tertiary/quarternary outcomes are listed for potential importance for other papers but aren’t going to be analyzed in the primary growth paper. The binary outcomes are going to be analyzed, but based on categorizing output from models of continious Z-scores.
Tertiary outcomes: ‘low_hc’,‘stunt’,‘wast’,‘uwt’, ‘muaz’,‘thriving’,‘growth_faltering’,‘recovery’ at 3 months, secondary outcomes at other ages.
Quarternaty outcomes: All measures at age 18 and age 24 months measures (each time has one study), tertiary outcomes at other ages.
Outlier exclusions
| studyid | laz_outlier | wlz_outlier | waz_outlier | hcz_outlier | muaz_outlier |
|---|---|---|---|---|---|
| CHILD | 7 | 7 | 0 | 2 | 0 |
| ELICIT | 0 | 3 | 0 | 0 | 0 |
| MISAME-3 | 0 | 0 | 1 | 3 | 4 |
| VITAL-Lactation | 0 | 5 | 16 | 1 | 6 |
Implausible length or head circumference losses
Based on the Multicentre Growth Reference Study, WHO recommends
maximum allowable differences between measurements of 0.5 cm for
circumferences and 0.7 cm for length/height. The table below shows
measurements with length or head circumference losses as well as
measures before or after to identify the outlier measures.
drop_flag denontes which measure to drop (the measure
before the length loss, or the measure with length loss) identified as
the larger absolute deviation from the longitudinal mean Z-score.
Note! Misame is missing HCAZ, so doesn’t have flagged deviations.
Sensitivity analysis: excluding flagged measures
## Printing: Min. 1st Qu. Median Mean 3rd Qu. Max. NA's## LAZ without exclusions: -5.95 -1.52 -0.59 -0.52547 0.385 5.83 1047## LAZ with exclusions: -5.95 -1.52 -0.59 -0.5233197 0.38375 5.83 1047## WLZ without exclusions: -4.88 -1.02 -0.1825 -0.2256606 0.61 4.78 1088## WLZ with exclusions: -4.88 -1.0175 -0.18 -0.2232161 0.61 4.78 1087## HCAZ without exclusions: -5 -1.145 -0.36 -0.3525539 0.46 4.46 1168## HCAZ with exclusions: -5 -1.13 -0.36 -0.3492613 0.46 3.57 1168
Coding Notes
- Growth faltering at 3, 6, 12, 18, 24 * a fall across 1 or more weight centile spaces, if birthweight was below the 9th centile * a fall across 2 or more weight centile spaces, if birthweight was between the 9th and 91st centiles * a fall across 3 or more weight centile spaces, if birthweight was above the 91st centile * when current weight is below the 2nd centile for age, whatever the birthweight.
- “Thriving” Moving 2 centile spaces in a positive direction in the previous period regardless of anthropometric status
- Recovery: moving 2 centile spaces in a positive direction out of anthropometric deficit without relapse
- Child 2-year measurement have height, rather than length. Combine in
this tabulation
- Dropped rare age strata in specific studies (<25 measures at an age). Drops child at 6 (N=24) and 18 months (n=3) and vital (n=1)
- Based on measurement timing, the ‘12 month’ measurements are 10 months for Misame, 12 months for others.
- Not including BMI due to overlap with WLZ, and recommendation to use WLZ in children under 2 (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6324504/)
- QA checked Z-scores with the Zscorer package. Slight differences but may be due to package differences- check the original studies’ Z-scores.