Demographic characteristics
A total of 353 women completed all PCAC scale items; their demographic characteristics are presented in Table 2, stratified by abortion procedure type. Among both surgical abortion and medication abortion participants, most women were aged 20–29 years (63 and 68%, respectively) and not married, partnered, or cohabitating (73 and 81%, respectively). About half of women in both samples had a college or university degree (52% in the surgical abortion sample and 49% in the medication abortion sample). Slightly more women were employed for pay among the surgical abortion sample (62%) compared to the medication abortion sample (55%). A higher proportion of women among the medication abortion sample (54%) reported this to be their first pregnancy than in the surgical abortion sample (40%).
Exploratory factor analysis
Among the medication abortion sample, nearly 50% of women responded “not applicable” to the “pain medication given” item. Further, only one woman responded “yes” to the “bribe” item among the surgical abortion sample and all the women in the medication abortion sample responded “no, never.” As a result, the “bribe” item was removed from both samples and the “pain medication given” item was removed among the medication abortion sample. This left 25 scale items among the surgical abortion sample and 24 items among the medication abortion sample at the start of conducting factor analyses. All scale items among both samples had a KMO measure of sampling adequacy greater than 0.50, with an overall KMO value of 0.74 among the surgical abortion sample and 0.80 among the medication abortion sample, providing evidence that the items were sufficiently correlated for conducting principal components analysis.
Surgical abortion sample
Among the surgical abortion sample, the initial EFA yielded a 3-factor solution with eigenvalues of greater than 1, accounting for about 50% of the variance. All 25 items loaded onto at least one of the three factors at the 0.10 factor loading cutoff, with cross loading on 11 items. We therefore categorized the items that cross loaded to factors based on the factor they loaded higher on and theoretical reasoning. Of the 11 items that loaded positively onto more than one factor, eight were retained in the factor they loaded highest on. Two items, “explain exams” and “ask questions,” which had slightly higher factor loadings on Factor 1 than Factor 2, were categorized into Factor 2 for conceptual reasons. Similarly, the item “pain medication given” had a higher loading on Factor 2 but was ultimately categorized with Factor 1. Also, despite the item “language understand” having a positive loading of greater than 0.10, upon further review it was felt that the wording of the item was too ambiguous, and as a result, the item was removed. Thus, this process resulted in 12 items being categorized to Factor 1, 9 items being categorized to Factor 2, and only three items (“treated differently,” “verbal abuse,” and “physical abuse”) remaining in Factor 3. Upon further discussion, we decided three items were insufficient for a sub-scale, and given that these items were conceptually related to those included in Factor 1, they were categorized with the items in Factor 1.
We then performed EFA with oblique rotation again on the two factors, or sub-scales, separately (Table 3). For the first factor, a standardized alpha of 0.78 was obtained suggesting acceptable reliability. Eleven of the 15 items had factor loadings of at least 0.30 and two items had factor loadings of at least 0.10. The remaining two items, “privacy” and “physical abuse,” had factor loadings less than 0.10 but were retained due to their theoretical importance. For the second factor, a standardized alpha of 0.69 was obtained. With the exception of “called by name” and “involvement in care,” the remaining seven items had factor loadings greater than 0.30. The standardized alpha for the nine items was 0.72 suggesting acceptable reliability.
A final factor analysis was conducted on the remaining 24 items restricted to a single factor (Table 4). Twenty-one of the 24 items loaded onto the single dominant factor at the 0.10 cutoff, and in fact, 17 of these had factor loadings greater than 0.30. The item “called by name” had a factor loading less than 0.10; however, it was ultimately retained because of its acceptable factor loading on the sub-scale (Factor 2). “Privacy” and “physical abuse” items also had factor loadings less than 0.10, as in the two-factor solution, but again, were retained due to their theoretical importance – lack of patient privacy and physical abuse are considered central components of poor person-centered care, and thus, would be important to measure. The standardized alpha of the 24-item scale was 0.82 (mean score = 61.84; SD = 7.69; Range = 25–71). A summary of standardized alphas and associated means, standard deviations (SD), and the range of scores for the full scale and each sub-scale are provided in Table 5.
Medication abortion sample
Among the medication abortion sample, the initial EFA using principal factors with oblique rotation also yielded a 3-factor solution with eigenvalues of greater than 1, accounting for about 53% of the variance. All 24 items loaded onto at least one of the three factors at the 0.10 factor loading cutoff, with cross loading on 12 items. We categorized the items that cross loaded to factors based on the factor they loaded higher on and theoretical reasoning. Ten of the 12 items that loaded positively onto more than one factor were categorized according to the higher factor loading. Two items, “friendly” and “privacy”, had similar loadings on two factors but were categorized with Factor 1 because these items were deemed to be more conceptually related to the items in that factor. Despite “explain medicines” loading to Factor 1, it was ultimately recategorized to Factor 3 for conceptual reasons. As was done in the surgical abortion sample, the “language understand” item was removed after it was concluded that the item was too ambiguous. This process resulted in 12 items being categorized to Factor 1, three items (“treated differently,” “verbal abuse,” and “physical abuse”) being categorized to Factor 2, and 9 items being categorized to Factor 3. For consistency, the three items in Factor 2 were ultimately categorized with Factor 1 to yield a final 2-factor solution, like that proposed among the surgical abortion sample, with 23 items retained at this time.
Performing EFA with oblique rotation on the two factors, or sub-scales, separately, we found that the first factor had a standardized alpha of 0.82 suggesting acceptable reliability. All 14 items had factor loadings greater than 0.10, and 11 of the 14 items had factor loadings greater than 0.30 (Table 3). For the second factor, a standardized alpha of 0.65 was obtained. All items in Factor 2 had factor loadings greater than the 0.10 cutoff; six of the nine items had factor loadings greater than 0.30.
A final factor analysis was conducted on the remaining 23 items restricted to a single factor (Table 4), with all items loading onto the single factor at the 0.10 cutoff. The standardized alpha of the final 23 item scale was 0.82 (mean score = 56.98; SD = 7.44; Range = 13–69; Table 5).
To name the factors, we assessed the specific items and mapped the items out on the original domains that the authors conceptualized for person-centered care. For the first factor, items came from the domains of respectful and supportive care (RSC sub-scale). Items included in the second factor all related to communication and women’s ability to be involved in care (CA sub-scale). Therefore, again, guided by the original conceptualized domains of person-centered care, we named the second factor “Communication and Autonomy.”
Criterion validity
The results of the bivariate linear regressions assessing the association between the two PCAC sub-scales (RSC sub-scale and CA sub-scale) and the full PCAC scale and the receipt of adequate information regarding the abortion procedure, respectively, are provided in Table 6. Among both samples, we found that women who reported receiving adequate information regarding their abortion procedure had significantly higher PCAC scores (for each sub-scale, as well as the full scale) than women who did not receive adequate information. These results confirm our hypothesis and suggest that higher PCAC scores are associated with receiving adequate information.
