Estimated effect of maternal anemia on birth weight by model specification and estimator.
| . | |$\widehat{DR}$| . | ES-SE . | ES-95% CI . | IF-SE . | IF-95% CI . | NB-SE . | NB-95% CI . |
|---|---|---|---|---|---|---|---|
| Full covariate set | |||||||
| Classic AIPW | −37 | 56 | (−147, 73) | 58 | (−151, 76) | 63 | (−161, 86) |
| Wtd regression AIPW | −41 | 56 | (−151, 69) | 56 | (−151, 68) | 61 | (−160, 78) |
| TMLE | −37 | 56 | (−147, 73) | 58 | (−151, 76) | 63 | (−160, 85) |
| Naive outcome | |||||||
| Classic AIPW | −36 | 57 | (−148, 76) | 61 | (−156, 84) | 64 | (−162, 90) |
| Wtd regression AIPW | −36 | 57 | (−148, 77) | 61 | (−156, 84) | 64 | (−161, 90) |
| TMLE | −36 | 57 | (−148, 77) | 61 | (−156, 84) | 63 | (−160, 89) |
| Naive propensity | |||||||
| Classic AIPW | −41 | 58 | (−154, 73) | 56 | (−150, 69) | 58 | (−155, 74) |
| Wtd regression AIPW | −47 | 57 | (−159, 65) | 54 | (−153, 59) | 59 | (−162, 68) |
| TMLE | −41 | 58 | (−154, 73) | 56 | (−150, 69) | 59 | (−155, 74) |
| . | |$\widehat{DR}$| . | ES-SE . | ES-95% CI . | IF-SE . | IF-95% CI . | NB-SE . | NB-95% CI . |
|---|---|---|---|---|---|---|---|
| Full covariate set | |||||||
| Classic AIPW | −37 | 56 | (−147, 73) | 58 | (−151, 76) | 63 | (−161, 86) |
| Wtd regression AIPW | −41 | 56 | (−151, 69) | 56 | (−151, 68) | 61 | (−160, 78) |
| TMLE | −37 | 56 | (−147, 73) | 58 | (−151, 76) | 63 | (−160, 85) |
| Naive outcome | |||||||
| Classic AIPW | −36 | 57 | (−148, 76) | 61 | (−156, 84) | 64 | (−162, 90) |
| Wtd regression AIPW | −36 | 57 | (−148, 77) | 61 | (−156, 84) | 64 | (−161, 90) |
| TMLE | −36 | 57 | (−148, 77) | 61 | (−156, 84) | 63 | (−160, 89) |
| Naive propensity | |||||||
| Classic AIPW | −41 | 58 | (−154, 73) | 56 | (−150, 69) | 58 | (−155, 74) |
| Wtd regression AIPW | −47 | 57 | (−159, 65) | 54 | (−153, 59) | 59 | (−162, 68) |
| TMLE | −41 | 58 | (−154, 73) | 56 | (−150, 69) | 59 | (−155, 74) |
Abbreviations:CI = confidence interval, ES = empirical sandwich, IF = influence function, NB = nonparametric bootstrap, SE = standard error, Wtd = weighted. The naive outcome model included only the exposure, and the naive propensity model included only an intercept. The nonparametric bootstrap was based on 5000 resamples.
Estimated effect of maternal anemia on birth weight by model specification and estimator.
| . | |$\widehat{DR}$| . | ES-SE . | ES-95% CI . | IF-SE . | IF-95% CI . | NB-SE . | NB-95% CI . |
|---|---|---|---|---|---|---|---|
| Full covariate set | |||||||
| Classic AIPW | −37 | 56 | (−147, 73) | 58 | (−151, 76) | 63 | (−161, 86) |
| Wtd regression AIPW | −41 | 56 | (−151, 69) | 56 | (−151, 68) | 61 | (−160, 78) |
| TMLE | −37 | 56 | (−147, 73) | 58 | (−151, 76) | 63 | (−160, 85) |
| Naive outcome | |||||||
| Classic AIPW | −36 | 57 | (−148, 76) | 61 | (−156, 84) | 64 | (−162, 90) |
| Wtd regression AIPW | −36 | 57 | (−148, 77) | 61 | (−156, 84) | 64 | (−161, 90) |
| TMLE | −36 | 57 | (−148, 77) | 61 | (−156, 84) | 63 | (−160, 89) |
| Naive propensity | |||||||
| Classic AIPW | −41 | 58 | (−154, 73) | 56 | (−150, 69) | 58 | (−155, 74) |
| Wtd regression AIPW | −47 | 57 | (−159, 65) | 54 | (−153, 59) | 59 | (−162, 68) |
| TMLE | −41 | 58 | (−154, 73) | 56 | (−150, 69) | 59 | (−155, 74) |
| . | |$\widehat{DR}$| . | ES-SE . | ES-95% CI . | IF-SE . | IF-95% CI . | NB-SE . | NB-95% CI . |
|---|---|---|---|---|---|---|---|
| Full covariate set | |||||||
| Classic AIPW | −37 | 56 | (−147, 73) | 58 | (−151, 76) | 63 | (−161, 86) |
| Wtd regression AIPW | −41 | 56 | (−151, 69) | 56 | (−151, 68) | 61 | (−160, 78) |
| TMLE | −37 | 56 | (−147, 73) | 58 | (−151, 76) | 63 | (−160, 85) |
| Naive outcome | |||||||
| Classic AIPW | −36 | 57 | (−148, 76) | 61 | (−156, 84) | 64 | (−162, 90) |
| Wtd regression AIPW | −36 | 57 | (−148, 77) | 61 | (−156, 84) | 64 | (−161, 90) |
| TMLE | −36 | 57 | (−148, 77) | 61 | (−156, 84) | 63 | (−160, 89) |
| Naive propensity | |||||||
| Classic AIPW | −41 | 58 | (−154, 73) | 56 | (−150, 69) | 58 | (−155, 74) |
| Wtd regression AIPW | −47 | 57 | (−159, 65) | 54 | (−153, 59) | 59 | (−162, 68) |
| TMLE | −41 | 58 | (−154, 73) | 56 | (−150, 69) | 59 | (−155, 74) |
Abbreviations:CI = confidence interval, ES = empirical sandwich, IF = influence function, NB = nonparametric bootstrap, SE = standard error, Wtd = weighted. The naive outcome model included only the exposure, and the naive propensity model included only an intercept. The nonparametric bootstrap was based on 5000 resamples.
This PDF is available to Subscribers Only
View Article Abstract & Purchase OptionsFor full access to this pdf, sign in to an existing account, or purchase an annual subscription.
