Table 4.
Impacts of the Team-Based Goals and Incentives (TBGI) intervention on home visits from front-line workers, as reported by maternal respondents, Begusarai, Bihar, 2012–2014*
| Modelled indicators | Baseline (%)† (May to June 2012) n=1680 |
Postimplementation (%)† (November to December 2014) n=1690 |
Difference in difference‡ n=3370 | ||||||||
| No. in model | Control (n=849) | Intervention (n=831) | P value | No. in model | Control (n=831) | Intervention (n=859) |
P value | No. in model | Per cent difference attributable to TBGI | P value | |
| At least two antenatal home visits in final trimester | 1664 | 38.2 | 33.3 | 0.33 | 1690 | 55.1 | 64.7 | 0.01 | 3354 | 14.5 | 0.027 |
| At least one home visit within 24 hours of delivery, among women who had a home delivery | – | – | – | – | 276 | 21.9 | 21.3 | 0.71 | – | – | – |
| At least one home visit within 1 week of delivery | 1679 | 11.2 | 9.8 | 0.80 | 1678 | 46.5 | 52.7 | 0.19 | 3357 | 7.6 | 0.42 |
| Complementary feeding home visit for women with infant 6–11 months old | 796 | 1.1 | 2.2 | 0.39 | 748 | 23.8 | 35.2 | 0.05 | 1544 | 10.6 | 0.86 |
| Postpartum family planning home visits for women with infant <6 months old | 820 | 11.4 | 20.7 | 0.07 | 896 | 9.8 | 16.9 | 0.04 | 1716 | −2.2 | 0.87 |
| Postpartum family planning home visits for women with infant 6–11 months old | 820 | 15.1 | 13.5 | 0.80 | 748 | 17.0 | 18.8 | 0.75 | 1568 | 3.3 | 0.53 |
*Survey-weighted percentages and counts are reported to account for the survey design. Survey logistic regression models were performed that accounted for village as the primary sampling unit and subcentre as the primary stratum within the sampling unit, with proportional sampling weights at the maternal respondent/household level; each model also included maternal age (as a continuous variable) and caste (non-Hindu, Hindu scheduled caste/scheduled tribe (SC/ST), Hindu not SC/ST).
†Separate regression models were conducted for each outcome, at baseline and postimplementation; p values reflect comparisons of the intervention and control groups at each time point.
‡In order to estimate the effect of the TBGI intervention on a particular indicator, we conducted a separate regression model for each outcome that included all maternal respondents. These models each contained a term representing time (baseline or postimplementation), a term representing treatment (intervention or control) and an interaction of these two terms, which is represented by the difference-in-difference (DID) estimator and its p value. The DID reflects treatment effects (positive values reflect the amount of improvement attributable to the intervention).