Table 3.
Domains of exposures associated with 30-day and post-discharge mortality
|
30-day mortality |
Post-discharge mortality |
||||
|---|---|---|---|---|---|
| Adjusted HR (95% CI)* | p value | Adjusted HR (95% CI)* | p value | ||
| Base variables | |||||
| Nutritional stratum at hospital admission | |||||
| No wasting | 1 (ref) | .. | 1 (ref) | .. | |
| Moderate wasting | 2·03 (1·40–2·94) | <0·0001 | 2·58 (1·47–4·52) | 0·0012 | |
| Severe wasting or kwashiorkor | 4·51 (2·81–7·23) | <0·0001 | 4·36 (2·27–8·37) | <0·0001 | |
| Age, log months | 0·77 (0·65–0·91) | 0·0031 | 0·72 (0·55–0·94) | 0·017 | |
| Female | 1·19 (0·87–1·65) | 0·28 | 1·33 (0·86–2·06) | 0·19 | |
| Leaving against medical advice | .. | .. | 2·14 (1·33–3·43) | 0·0021 | |
| Admission duration, log days | .. | .. | 1·51 (0·95–2·39) | 0·083 | |
| Change in anthropometry at hospital discharge | |||||
| No change | .. | .. | 1 (ref) | .. | |
| Improved | .. | .. | 0·73 (0·57–0·93) | 0·010 | |
| Worsened | .. | .. | 0·81 (0·15–4·40) | 0·80 | |
| Exposure domains | |||||
| Signs of illness severity at admission | |||||
| Low | 1 (ref) | .. | 1 (ref) | .. | |
| Medium | 1·29 (0·76–2·19) | 0·34 | 1·36 (0·83–2·22) | 0·22 | |
| High | 3·82 (2·50–5·86) | <0·0001 | 1·76 (0·96–3·24) | 0·067 | |
| Signs of illness severity at discharge | |||||
| Low | .. | .. | 1 (ref) | .. | |
| Medium | .. | .. | 1·29 (0·91–1·84) | 0·15 | |
| High | .. | .. | 4·00 (1·66–9·64) | 0·0020 | |
| HIV status | |||||
| Negative | 1 (ref) | .. | 1 (ref) | .. | |
| Untested | 0·78 (0·37–1·65) | 0·52 | 1·88 (0·78–4·55) | 0·16 | |
| Exposed | 1·68 (1·29–2·18) | <0·0001 | 1·82 (1·19–2·80) | 0·0064 | |
| Infected | 2·45 (1·24–4·81) | 0·010 | 1·75 (0·91–3·37) | 0·094 | |
| Underlying medical conditions | |||||
| Low | 1 (ref) | .. | 1 (ref) | .. | |
| Medium | 2·45 (1·22–4·91) | 0·012 | 0·90 (0·61–1·33) | 0·60 | |
| High | 2·08 (1·25–3·45) | 0·0054 | 1·40 (0·99–1·97) | 0·058 | |
| Child-level nutritional risk exposures | |||||
| Low | 1 (ref) | .. | 1 (ref) | .. | |
| Medium | 0·59 (0·36–0·97) | 0·036 | 0·83 (0·41–1·67) | 0·61 | |
| High | 1·15 (0·78–1·69) | 0·47 | 1·20 (0·75–1·90) | 0·45 | |
| Caregiver characteristics | |||||
| Least adverse | 1 (ref) | .. | 1 (ref) | .. | |
| Moderately adverse | 2·03 (1·36–3·04) | 0·0010 | 1·37 (0·82–2·30) | 0·22 | |
| Most adverse | 1·53 (1·01–2·32) | 0·045 | 2·11 (1·18–3·77) | 0·012 | |
| Household-level exposures | |||||
| Least adverse | 1 (ref) | .. | 1 (ref) | .. | |
| Moderately adverse | 2·02 (1·35–3·03) | 0·0011 | 1·04 (0·78–1·38) | 0·80 | |
| Most adverse | 2·08 (1·38–3·13) | <0·0001 | 1·01 (0·68–1·50) | 0·95 | |
| Access to health care | |||||
| Least adverse | 1 (ref) | .. | 1 (ref) | .. | |
| Moderately adverse | 1·10 (0·74–1·64) | 0·64 | 1·09 (0·68–1·77) | 0·71 | |
| Most adverse | 1·77 (1·14–2·75) | 0·011 | 1·09 (0·64–1·85) | 0·74 | |
| Site variance (95% CI) | 0·05 (0·003–0·93) | .. | 0·02 (0·001–0·75) | .. | |
| Bootstrapped AUC (95% CI) | 0·80 (0·78–0·83) | .. | 0·81 (0·78–0·84) | .. | |
HR=hazard ratio. AUC=area under receiver operating characteristic curve.
*
Adjusted HR for all predictors in the multivariable model. All model results were weighted using sampling and lost to follow-up weights. HRs from a multi-level multivariable parametric (Weibull distribution) survival model with site as a random effect.