Prenatal intentions
Seventy seven percent of participants reported that they were likely to exclusively breastfeed, 11.8% were not sure while 10.9% were unlikely going to exclusively breastfeed their babies for 24 weeks. The prenatal intended duration of exclusive breastfeeding ranged from 4-24 weeks. Mothers intended to EBF their infants for different reasons. These reasons included lack of money to purchase infant formula, belief that breast milk is the best, and to conceal their HIV status because choosing not to breastfeed was seen as self-revealing one’s HIV status.
Correlates of prenatal intentions of exclusive breastfeeding
Prenatal intended duration of EBF was significantly associated with maternal education (r = -.417, p < .001), parity (r = .213, p < .05), disclosure of HIV status (r = -.195, p < .05), PEBB (rho = -.198, p < .05), NorB (r = .194, p < .05) and EBCB (r = .286, p < .01). The likelihood of HIV-positive mothers to exclusively breastfeeding for six months was significantly associated with NorB [X2 = 8.553, phi = .279, p < .01], and EBCB [X2 = 16.589, phi = .388, p < .001).
The prenatal intended duration of EBF was then dichotomized as “Less than 24 weeks and Equal to 24 weeks’ and its relationship with the rest of the variables was re-assessed through Pearson’s chi-square test to compare the results with those obtained through Pearson’s product-moment correlation coefficient as a way of validating them. Statistically significant associations were observed between: (1) prenatal intended duration of EBF and maternal education [X2 = 21.686, phi = -.444, p < .01], PEBB [X2 = 4.089, phi = -.193, p < .05], NorB [X2 = 3.945, phi = .189, p < .05], and EBCB [X2 = 4.980, phi = .213, p < .05]. Although non-parametric analyses failed to demonstrate significant association between prenatal intended duration and parity (Fisher’s Exact Test at p = .09, phi = .169, p > .05) and disclosure of HIV status (Fisher’s Exact Test at p = .079, phi = -.168, p > .05), the direction of the associations and the rest of the findings were consistent with those obtained from parametric analyses.
Prenatal intentions were positively associated with EBF at 12 weeks postnatal (r = .106; rho = .251). However, the relationship was not statistically significant (p > .05). This may be due to the reduced statistical power of the sample at 12 weeks postnatal because of the high loss to follow up that this study suffered. Fifty-five out of 110 participants returned for follow up at 12 weeks postnatal. Analysis of attrition bias was conducted. The ‘drop-outs’ and the ‘non-drop-outs’ were not significantly different on all the baseline variables except for two (prenatal intentions and normative beliefs scores) variables. At baseline, 83.6% of the participants who came back for their 12 weeks postnatal follow-up reported that they intended to EBF their babies for six months compared to 70.9% of those who dropped out of the study [X2 = 11.217 (2, 55); Cramer’s value = .319; p = .004]. Furthermore, Mann Whitney U Test revealed a significant difference in the total NorB scores of those who returned for the 12 week follow up (Median = 80, n = 55) and those who did not come for the 12 week follow up (Median = 60, n = 55), U = 1190, z = -1.929, p = 0.05, r = -.184. Considering the significantly high prenatal intentions for EBF and high NorB scores indicating a high prenatal motivation for EBF and a high possible support for EBF from significant others respectively, the ‘non-drop outs’ group had high opportunity for EBF for longer periods. Social desirability may not fully explain these findings.
Predictors of prenatal intended duration of exclusive breastfeeding
Multiple linear regression analyses were run with number of weeks of intended EBF duration as an outcome variable. A sequential modeling approach was used. Maternal education, parity and disclosure of HIV status were entered together at the first step, and then behavioral beliefs (PEBB) at second step, normative beliefs (NorB) at third step and control beliefs (EBCB) at fourth step to assess their effects on number of weeks of intended EBF. Table
3 presents the results from multiple linear regressions. The model containing maternal education, parity, disclosure of HIV status, PEBB, NorB and EBCB explained about 35% of the variance in number of weeks of intended duration of EBF. In the order of their strength, starting with the strongest, EBCB, maternal education and PEBB were significant predictors of prenatal intended duration of EBF. Although addition of PEBB to the model did not significantly improve R
2, it became a significant predictor (p < .01) of prenatal intended duration of EBF after EBCB was added in the final model. NorB, that significantly improved R
2 when added to the model, was not a significant predictor of prenatal intended duration of EBF when EBCB was added to the equation. This suggests some possible mediation effect in the relationships. However, this was not explored further because it was beyond the realm of the current study.
Table 3
Multiple linear regression predicting intended duration of EBF in weeks at baseline (n = 110 across all models)
Constant | 26.596(2.747) | 30.152(15.570) | 33.285(15.308) | 61.246(15.737) |
*** | | * | *** |
Maternal Education | -2.295(.581) | -2.272(.592) | -2.262(.580) | -2.154(.540) |
-.390*** | -.386*** | -.385*** | -.366*** |
Parity | .221(.962) | .223(.966) | .259 (.946) | -.045(.883) |
.023 | .023 | .027 | -.005 |
Disclosure of HIV status to Spouse & family | -1.693(.967) | -1.669(.977) | -1.573(.957) | -1.113(.897) |
-.153 | -.151 | -.142 | -.101 |
Positive Beliefs (PEBB) | | -1.544(6.654) | -3.180(6.554) | -20.610(7.394) |
| -.021 | -.043 | -.278** |
Normative Beliefs (NorB) | | | 2.631(1.126) | 1.700(1.071) |
| | .201* | .130 |
Control Beliefs (EBCB) | | | | .328(.079) |
| | | .415*** |
R2
| .197 | .198 | .238 | .348 |
R2 change | .197*** | .00 | .040* | .110*** |
Adjusted R2
| .175 | .167 | .201 | .310 |
Model pvalue | P < .001 | P < .001 | P < .001 | P < .001 |
Logistic regression analyses were also run to predict prenatal intended duration of EBF to validate the findings. For this, prenatal intended duration was categorized into two (0 = Less than 24 weeks of EBF; 1 = Equal to 24 weeks of EBF). Block 1 and the model as a whole containing maternal education as a predictor, was statistically significant (X
2 = 23.141, p < .001). Addition of: (1) PEBB in Block 2 resulted in no significant contribution (X
2 = .866, p = .352) to the model; (2) NorB in Block 3 resulted in a significant contribution (X
2 = 5.409, p < .05) to the model; and (3) EBCB contributed significantly to the model [X
2 = 4.947, p < .05). The final model containing all the predictors was statistically significant, X
2 (4, n = 110) = 34.363, p < .001 indicating that the model was able to distinguish between HIV-positive mothers who reported and those who did not report that they would EBF for 24 weeks. The chi-square value for the Hosmer-Lemeshow Goodness of Fit Test was 5.850 with a significance level of p = .440; indicating support for the model. The model as a whole explained 26.8% (Cox and Snell R square) and 39.9% (Nagelkerke R square) of the variance in prenatal intended duration of EBF in weeks, and overall correctly classified 78.2% of HIV-positive mothers. The Wald statistics showed that only maternal education (p < .001) and EBCB (p < .05) made a significant contribution to the final model predicting prenatal intended duration of EBF. See Table
4.
Table 4
Logistic regression analysis predicting intended duration of EBF in weeks at baseline (n = 110)
Constant | 1.945 | .586 | 10.996 | 1 | .001 | 6.991 | - | - |
Maternal education | -2.395 | .626 | 14.614 | 1 | .000 | .091 | .027 | .311 |
PEBB | -1.109 | .617 | 3.233 | 1 | .072 | .330 | .099 | 1.105 |
NorB | 1.405 | .861 | 2.664 | 1 | .103 | 4.075 | .754 | 22.013 |
EBCB | 1.330 | .620 | 4.610 | 1 | .032 | 3.782 | 1.123 | 12.740 |
Although PEBB was a significant predictor of prenatal intended duration of EBF when data were analyzed through multiple linear regression, it had a borderline significance (Wald = 3.233, p = .072; OR = .330, 95% CI; .099 to 1.105) when logistic regression was used. The strongest predictor of prenatal intended duration was EBCB with an odds ratio (OR) of 3.782, CI; 1.123-12.740; Wald = 4.610, p < .05 indicating that HIV-positive mothers with high EBCB scores were about 4 times more likely to intend to EBF for 24 weeks compared to those who had low EBCB scores. Higher maternal education had OR of .091, 95% CI; .027 to .311 indicating that HIV-positive mothers with higher education (≥MSCE) were .909 (91%) less likely to intend to exclusively breastfeed for 24 weeks than those who had less than MSCE qualification.
Results from multiple linear regression analyses have demonstrated that maternal education, PEBB, and EBCB were significant predictors of prenatal intended duration of EBF. These results were consistent with those obtained from multiple logistic regression except that PEBB failed to reach significant level (p = .072).
Predictors of likelihood of HIV-positive mothers to EBF for 24 weeks
Sequential logistic regression analysis was used to predict likelihood of HIV-positive mothers to EBF for 24 weeks (assessed as 0 = unlikely and not sure; 1 = likely) with normative beliefs (NorB) and exclusive breastfeeding control beliefs (EBCB) as predictors. These two variables were the only possible predictors that showed a statistically significant association with likelihood of EBF for 24 weeks. The model containing NorB and EBCB as predictors of likelihood of EBF for 24 weeks was statistically significant, X2 = 23.528, p < .01. Hosmer and Lemeshow Test of Goodness of fit at significance level of P > .05. The model explained 19.3% (Cox and Snell R square) and 29.3% (Nagelkerke R square) of the variance in the likelihood of HIV-positive mothers to EBF for 24 weeks, and overall correctly classified 77.3% of the HIV-positive mothers. Only EBCB (Wald = 9.421, p < .01) made a significant contribution to the model predicting the likelihood of EBF for 24 weeks. The OR was 4.706 with 95% CI of 1.750 to 12.652 indicating that HIV-positive mothers with high EBCB scores were almost 5 times more likely to EBF for 24 weeks than their counterparts with lower EBCB scores.