Program Impact: Facility-Based Birth

Quantitative Results

The proportion of FBB increased from 54.8% (1567/2859) in the Kalomo district

before the intervention to 64.6% (2336/3618) during the intervention, an absolute

difference of 9.8 percentage points (95% CI: 7.4, 12.2, p<0.01). In the comparison area

there was a slight, non-significant increase of 0.2 percentage points (95% CI: -1.4,1.7).

This resulted in a 9.6 net percentage point increase in the intervention area versus the

comparison area (Table 13).

Table 13. Differences-in-differences Analysis of Facility-based Birth Before and During SMGL between SMGL and Non-SMGL Areas

Time Period Facility-based birth

Pre- SMGL During SMGL Absolute Difference (95% CI) Kalomo N (%) 54.8 (1567/2859) 64.6** (2336/3618) +9.8 (7.4, 12.2) Comparison N (%) 64.6 (4881/7559) 64.7 (4949/7644) +0.2 (-1.4, 1.7) Net difference +9.6 **p<0.01

In the Kalomo intervention district, controlling for mother’s age, education, parity

and distance to the health facility, the adjusted odds of FBB during the intervention was

60% greater (OR: 1.60, 95% CI: 1.28,1.99) compared to the rate before the intervention

period (Table 14). In the comparison districts, there was no difference in the adjusted

odds of FBB during the intervention compared to before the intervention (OR: 1.05, 95%

Table 14. Likelihood of Facility-based Birth Before and During SMGL Implementation, Intervention vs. Comparison Districts*

Number of Births Facility-based Births No (%) Odds ratio (95% CI) Adjusted OR (95% CI) Kalomo District Before† 2859 1567 (54.8) 1.00 1.00 During 3618 2336 (64.6) 1.50 (1.21, 1.87) 1.60 (1.28,1.99) Comparison Before† 7559 4881 (64.6) 1.00 1.00 During 7644 4949 (64.7) 1.01 (0.85, 1.19) 1.05 (0.87,1.25) †reference group

*Controlling for mother’s age, education, parity, distance to health facility, children under five, and asset quartile

In the model, there was a significant interaction between area of intervention and

the time-period (p = 0.005) (Table 15). There was 49% relative increase in the odds of

FBB in the intervention district after SMGL compared to the comparison area (OR 1.49,

95% CI: 1.21-1.77), adjusting for household size, maternal age, any maternal education,

facility more than two hours away, four or more ANC visits, and parity. Another model

Table 15. Facility-based Birth Before and During the SMGL Program Implementation Time Period in Kalomo District and Comparison Districts

Adjusted OR (95% CI)*

Model 1** Model 2† Model 3‡

Intervention Period x Intervention Area

Before, Intervention versus Comparison

1 1 1

During, Intervention versus Comparison 1.54 (1.25, 1.83) 1.53 (1.26, 1.82) 1.49(1.21, 1.77) Respondent’s Distance to Facility <2 hours 1 1 >2 hours 0.51 (0.44, 0.60) 0.51 (0.44, 0.60) Antenatal Care <4 visits 1 >4 visits 1.54 (1.35, 1.75)

*Consecutive adjustment of covariates in the model and changes of odds ratios of facility-based birth before and during, intervention district versus comparison district.

** Adjusted for household size, mother’s age, mother’s education (any), parity, marital status, and asset quartile

† Adjusted for covariates in model 1 + respondent’s distance to facility ‡ Adjusted for covariates in model 2 + number of antenatal care visits

Results of the propensity score analysis (PSA) also demonstrated the positive

change in Kalomo. Using multivariate regression with the sample created using PSA, the

OR for the difference in FBB between SMGL and non-SMGL groups, before and during

SMGL—controlling for household size, maternal age, any maternal education, facility

more than hours away, four or more ANC visits, and parity–was 1.43 (95% CI: 1.14,

1.71). I also observed significant interaction between area of intervention and the time

Time Series Results

A time series analysis revealed that the increase in FBB in Kalomo did not occur

until the final 4–6 months of the study period, starting in early 2013 (Figure 6).

Following the three-phased start of the SMGL intervention which lasted from February to

June 2012, there was a positive trend in FBB, from 56% in May to 66% in November.

This was not statistically significant. The month of September saw a decrease to 63%

from 64% in August and therefore were not seven consecutive months of increase, as

required to reach statistical significance. However, starting in January 2013, there was a

statistically significant shift, with six points above the median (December 2012 through

June 2013). In December of both 2012 and 2013, there was a decrease in FBB,

indicating a possible seasonal decrease regardless of the overall trend. For the

comparison group, the run charts indicate neither positive trends nor statistically

Figure 6. Run Chart of Facility-based Births in Kalomo District, April 2011– June 2013

†Circled area indicates a statistically significant shift Pre-

SMGL

Launch During-

Figure 7. Run Chart of Facility-based Births in Kalomo District and Comparison Districts, April 2011– June 2013

Qualitative Results

When asked to describe their perception of SMGL’s impact, the majority of

respondents (65%, n=17/26) cited an increase in facility deliveries. This was the most

frequently mentioned outcome cited by every type of respondent. In the interviews, the

SMAG and health facility staff had the highest number of references to facility delivery. Pre-

SMGL

Launch During-

“Then at the same time this information is given to us, the same information also we are giving the SMAGs in the community, so we find that now all those mothers who are having a negative attitude, they are coming to deliver at the facility, they are now coming”

-Rural midwife

One of the respondents, in the donor/lead implementing partner category, had a

less-than-favorable response, indicating that facility delivery had in fact increased in

Kalomo, but to a lesser degree than the other three SMGL districts in Zambia.

Summary

Both the quantitative and qualitative data provide strong evidence of a change in

FBB in Kalomo during the implementation of the SMGL program, in comparison to the

three comparison districts where SMGL was not implemented.