Table 4.
Correlational validity by Quality of Contraceptive Counseling (QCC) Scale domain: Results from logistic regression analysis, Mexico City and San Luis Potosía
| External variable | n(%) | Information Score OR (95% CI) p‐value | Relationship Score OR (95% CI) p‐value | Disrespect Scoreb OR (95% CI) p‐value | Total Score OR (95% CI) p‐value |
|---|---|---|---|---|---|
| Convergent validity variables | |||||
| Highest rating of overall experience with provider (n=496)c | 250(50.4) | 5.7(3.6–8.8) p<0.0001 |
10.8(6.3–18.5) p<0.0001 |
1.4(1.3–1.7) p<0.0001 |
22.6(10.9–46.8) p<0.0001 |
| Intention to use method selected at baseline (n=395)d | 361(91.4) | 2.6(1.5–4.4) p=0.001 |
2.6(1.5–4.6) p=0.001 |
1.3(1.0–1.6) p=0.03 |
4.2(2.2–7.9) <p=0.0001 |
| Predictive validity variables | |||||
| Use of contraception at 1–3 months follow‐up (n=169)e | 135(79.9) | 1.8(0.8–4.0) p=0.1 |
3.2(1.1–9.4) p=0.03 |
1.0(0.7–1.4) p=0.9 |
2.6(0.9–7.5) p=0.07 |
| Informational needs met at 1–3 months follow‐up (n=205)c | 148(72.2) | 2.6(1.4–4.9) p=0.003 |
3.9(2.1–7.4) p<0.0001 |
1.1(0.9–1.3) p=0.5 |
4.7(2.0–10.8) p<0.0001 |
| Divergent validity variable | |||||
| Highest rating of waiting room (n=494)c | 392(79.4) | 0.9(0.6–1.4) p=0.7 |
0.8(0.5–1.5) p=0.6 |
0.9(0.8–1.1) p=0.4 |
0.9(0.5–1.6) p=0.6 |
Odds ratios are from bivariate logistic regression models estimating the odds of each dichotomous external variable associated with a one‐unit increase in QCC Scale Scores, accounting for clustering by provider through use of robust standard errors in complete case analysis. We conducted sensitivity analyses without use of robust standard errors to allow for inclusion of 41 additional cases where the provider was not known; these analyses revealed similar findings, with the exception of the relationship between the disrespect score and intention to use the selected method which had a similar OR but was no longer significant (p=0.08) (data not shown). In models for predictive validity variables, we ran sensitivity analyses controlling for amount of time since follow‐up and results were unchanged (data not shown).
Disrespect and Abuse score dichotomized into highest score (higher=better quality) versus all else, due to high skew.
Missing data ranged from 2–5 cases for these variables.
This variable was assessed only among participants reporting they selected a method (94 participants indicated not having selected a method at baseline); an additional 10 cases were missing.
Seven participants who reported not wanting to prevent pregnancy and 30 who reported being pregnant at baseline were dropped from this analysis; one additional case was missing data.