Skip to main content

Table 4 Longitudinal risk of malaria and SMA infections and all-cause mortality

From: Genetic variation in CSF2 (5q31.1) is associated with longitudinal susceptibility to pediatric malaria, severe malarial anemia, and all-cause mortality in a high-burden malaria and HIV region of Kenya

Variable names

Event

Baseline

Coefficient

Std. Errora

z value

HR (95% CI)

HR inverse (95% CI)

P value

Malaria infections

 Age at first hospital visit

  

− 0.041

0.003

− 14.865

0.960 (0.956–0.964)

1.042 (1.036–1.047)

< 0.0001

 Cohort*

498

575

0.071

0.045

1.572

1.074 (1.014–1.137)

0.931 (0.853–1.018)

0.1159

 HIV 1 (+)

36

1037

− 0.196

0.143

− 1.369

0.822 (0.670–1.009)

1.217 (0.919–1.611)

0.1711

 Sex (female)

540

533

− 0.082

0.043

− 1.896

0.921 (0.873–0.972)

1.086 (0.997–1.182)

0.0580

 -α3.7/-α3.7

215

858

− 0.092

0.054

− 1.709

0.912 (0.850–0.978)

1.097 (0.987–1.219)

0.0874

 Hb SS

14

1059

− 0.726

0.174

− 4.163

0.484 (0.340–0.689)

2.066 (1.468–2.908)

< 0.0001

 Hb AS

159

914

− 0.242

0.063

− 3.858

0.785 (0.723–0.852)

1.274 (1.127–1.441)

0.0001

 CSF2 AC haplotype

30

1043

− 0.325

0.114

− 2.840

0.723 (0.606–0.863)

1.384 (1.106–1.731)

0.0045

 CSF2 GC/GT diplotype

66

1007

− 0.153

0.096

− 1.595

0.859 (0.756–0.975)

1.165 (0.966–1.405)

0.1108

Model fit

Test

  

Test stat

DF

P value

Number of total visits (N)

13,253

 

 LR test

  

587.8063

9

< 0.0001

Number of malaria events

5299

 

 Score test

  

539.2203

9

< 0.0001

AIC

68,758.3586

 

 Wald test

  

273.68

9

< 0.0001

Concordance (Std. Error)

0.6175 (0.0059)

 

SMA infections

 Age at first hospital visit

  

− 0.047

0.010

− 4.587

0.954 (0.936–0.973)

1.048 (1.027–1.070)

< 0.0001

 Cohort*

498

575

− 0.426

0.146

− 2.915

0.653 (0.494–0.863)

1.532 (1.150–2.040)

0.0036

 HIV 1 (+)

36

1037

1.050

0.253

4.147

2.858 (1.761–4.640)

0.350 (0.213–0.575)

< 0.0001

 Hb SS

14

1059

0.989

0.467

2.118

2.688 (1.322–5.465)

0.372 (0.149–0.929)

0.0342

 Hb AS

159

914

− 0.706

0.302

− 2.335

0.494 (0.309–0.789)

2.026 (1.120–3.665)

0.0196

 CSF2 AT haplotype

485

588

0.232

0.133

1.745

1.262 (0.982–1.621)

0.793 (0.611–1.029)

0.0809

Model fit

Test

  

Test stat

DF

P value

Number of total visits

13,022

 

 LR test

  

81.9523

6

< 0.0001

Number of SMA events

248

 

 Score test

  

83.5002

6

< 0.0001

AIC

3306.7994

 

 Wald test

  

66.77

6

< 0.0001

Concordance (Std. Error)

0.6814 (0.0171)

 

All-cause mortality

 Age at enrollment

  

− 0.070

0.022

− 3.173

0.932 (0.893–0.974)

1.0728 (1.027–1.120)

0.0015

 HIV 1 (+)

36

1038

2.801

0.320

8.763

16.452 (8.794–30.779)

0.061 (0.033–0.114)

< 0.0001

 Hb SS

14

1060

1.882

0.605

3.110

6.568 (2.005–21.509)

0.152 (0.047–0.499)

0.0019

 Hb AS

158

916

− 1.216

0.600

− 2.026

0.296 (0.091–0.961)

3.373 (1.041–10.937)

0.0427

 CSF2:g.-7032 GA genotype

436

638

0.632

0.294

2.150

1.881 (1.058–3.345)

0.532 (0.299–0.946)

0.0315

 CSF2 GC/GC diplotype

56

1018

0.945

0.501

1.887

2.574 (0.964–6.869)

0.389 (0.146–1.037)

0.0592

Model fit

Test

  

Test stat

DF

P value

   

 LR test

  

71.1327

6

< 0.0001

   

 Score test

  

150.5798

6

< 0.0001

AIC

671.609

 

 Wald test

  

87.77

6

< 0.0001

Concordance (Std. Error)

0.7598 (0.0334)

 
  1. Independent increments according to Anderson–Gill method were used for ordered multiple-outcome-per-subject Cox proportional hazard model fit to investigate on the time-to-event of covariates for malaria and SMA infections. Additionally, the COX proportional hazard model was used to predict all-cause mortality outcomes. A positive coefficient indicates a worse prognosis, whereas a negative coefficient indicates a better prognosis. Data are ranked per variables as follows; the metric variables (age at first hospital visit), followed by categorical variables (cohort, HIV 1), genetic variables (sex, -α3.7/-α3.7, Hb SS, Hb AS) and CSF2 genetic variants. aStandard error robust estimate. *Cohort presented in the table are patients recruited into the study in the 2007–2012 study period. The sample size (N = 13,253) consisted of all visits of N1 = 1073 (= event + baseline) patients retained for this analysis. Std. Error standard error, HR hazard ratio, HR inverse reciprocal of hazard ratio, 95% CI 95% confidence interval, HIV 1 human immunodeficiency virus 1, -α3.7/-α3.7 α+-thalassemia homozygous mutant, Hb SS sickle cell diseases, Hb AS sickle cell trait, CSF2 colony-stimulating factor 2, AIC Akaike information criterion, LR test likelihood ratio test
Back to article page

Tropical Medicine and Health

ISSN: 1349-4147

Contact us