Table 1.
Association Between CD4 Count at Baseline and Trends in Labor Force Participation and Household Asset Ownership
| (1) | (2) | |
|---|---|---|
| Labor Force Participation | Asset Score | |
| Baseline CD4 count | ||
| <200 | Ref | Ref |
| >200 | 0 13*** (0.057, 0.21) | 0.0049 (−0.47, 0.48) |
| Time on ART (years) | ||
| Baseline | Ref | Ref |
| 1 year | 0.12*** (0.051, 0.18) | 0.16 (−0.23, 0.56) |
| 2 years | 0.12*** (0.057, 0.19) | 0.39* (−0.016, 0.80) |
| 3 years | 0.16*** (0.10, 0.22) | 0.27 (−0.12, 0.66) |
| 4 years | 0.18*** (0.12, 0.24) | 0.70*** (0.30,1.10) |
| 5 years | 0.19*** (0.13, 0.25) | 0.95*** (0.52,1.38) |
| 6 years | 0.20*** (0.13, 0.28) | 1.17*** (0.65, 1.70) |
| Time on ART*CD4 Group Interactions | ||
| 1 year*CD4≥200 | −0.12* (−0.24, 0.013) | 0.21 (−0.48, 0.90) |
| 2 years*CD4≥200 | −0.16** (−0.31, −0.001) | −0.47 (−1.24, 0.30) |
| 3 year*CD4≥200 | −0.18** (−0.33, −0.025) | −0.2 (−0.93, 0.52) |
| 4 years*CD4≥200 | −0.13 (−0.29, 0.026) | −0.31 (−1.11, 0.49) |
| 5 years*CD4≥200 | −0.07 (−0.26, 0.12) | −0.11 (−0.93, 0.71) |
| 6 years*CD4≥200 | −0.20* (−0.44, 0.038) | 0.33 (−0.81, 1.47) |
| Number of Participants | 505 | 505 |
| Person Years | 2,349 | 2,349 |
Notes: Regression coefficients are reported with 95% confidence intervals in parentheses. Each column represents a separate regression.
- p<0.10
p<0.05
p<0.01. Confidence intervals computed using heteroskedasticity correct standard errors.
Models for Labor Force Participation were estimated using probit regression, with the dependent variable =1 if the individual reported participation in any formal or informal income generating activity at the time of interview. All models included controls for baseline age, age squared, a binary variable for gender, interactions between gender and baseline age, and binary indicators for marital status, completing some secondary schooling, and interview during the rainy season (see Table S2 for coefficient estimates on covariates). Reported coefficients are marginal effects. These can be interpreted as follows: for a continuous variable, the marginal effect coefficient reflects the percentage point increase in the probability of observing the dependent variable for a 1 unit change in the explanatory variable; for binary variables, it reflects a similar change in the dependent variable associated with a change from 0 to 1 on the explanatory variable of interest. Models for Asset Scores (integer ranging from 0-16, representing the count of the number of assets owned from the following: iron, gas or electric stove, refrigerator, telephone, motorbike, bicycle, car, clock, television, radio, bed, sofa, lantern, cupboard, and mattress) were estimated using ordinary least squares regressions.
For all models, the main explanatory variables were CD4 count at baseline and their interactions with the Time on ART dummy variables.
The “Number of Participants” refers to the number of unique individuals in the estimation sample. “Person Years” refers to the total number of Person-Year observations.