Table 5.
Estimated coefficients and standard errors for hazards models of contraceptive failure in 2011. The results for a standard model and a multiprocess model, modelling method choice and discontinuation conjointly, are compared. In the multiprocess model, IUD/Implants are not considered due to the limitations of the aML software and only the results of the hazard model part are shown
| Failure 2011 | Standard model | Multiprocess model | ||
|---|---|---|---|---|
| Variable | Coefficient | SE | Coefficient | SE |
| Constant | 9.82 | −4.10 | ||
| Duration (month) | ||||
| 0–12 | −0.156 | −0.16 | ||
| 12–24 | 0.228 | 0.228 | ||
| 24+ | −0.009 | −0.009 | ||
| Method | ||||
| Pill | 0 | 0 | ||
| IUD/Implants | 1.834*** | 0.365 | – | – |
| Injectables | −2.959*** | 0.532 | −1.763*** | 0.144 |
| Condoms | −1.532*** | 0.152 | −1.302*** | 0.369 |
| Education | ||||
| No Education | 0 | 0 | ||
| Primary | −0.971** | 0.347 | −0.168 | 0.141 |
| Secondary+ | −0.111 | 0.146 | −0.357 | 0.179 |
| Age at start | ||||
| < 25 years | 0 | 0 | ||
| 25–34 years | −0.254 | 0.188 | −0.626** | 0.211 |
| 35–49 years | −0.553** | 0.205 | −1.420*** | 0.188 |
| Wealth | ||||
| Low | 0 | 0 | ||
| Medium | −1.326*** | 0.216 | 0.169 | 0.219 |
| High | 0.258 | 0.261 | −0.301 | 0.200 |
| Area | ||||
| Rural | 0 | 0 | ||
| Urban | −0.160 | 0.266 | 0.201 | 0.164 |
| Religion | ||||
| Orthodox | 0 | 0 | ||
| Muslim | 0.277 | 0.177 | 0.408*** | 0.136 |
| Other | 0.477*** | 0.135 | −0.116 | 3.181 |
| Women level effect | ||||
| Sigma | 0.43 | 0.62 | 0.489*** | 0.133 |
| Rho (failure/injectable) | – | – | 0.142 | 0.110 |
| Rho (failure/condom) | – | – | 0.170 | 0.155 |
| Rho (injectable/condom) | – | – | −0.097*** | 0.035 |
Italics depict fixed parameters, estimated previously using a simple model with no variables. Bold depicts significance: *P < 0.05, **P < 0.01, ***P < 0.001