We conducted a facility-based cross-sectional study from May 16 to June 17, 2022, at purposively selected three public hospitals in eastern Ethiopia, namely: Hiwot Fana Comprehensive Specialized University Hospital (HFCSUH), Haramaya General Hospital (HGH), and Dil Chora Referral Hospital (DCRH). HFSCUH is a comprehensive teaching hospital affiliated with Haramaya University College of Health and Medical Sciences, which is located in Harar town at a distance of 526 km to the east of Addis Ababa, the capital city of Ethiopia. It serves around 5.8 million people of the surrounding population (21). The Haramaya General Hospital is located in Haramaya town, at a distance of 507 km to the east of Addis Ababa. It serves about 1,143,909 Haramaya district residents and the neighboring population. DCRH is located in the Dire Dawa administration in eastern Ethiopia, located 515 km away from Addis Ababa and it annually serves around 193,485 people from nearby populations (1).

Women who gave birth at the selected public hospitals and were within the six-week postpartum period were eligible to participate in this study. However, we excluded those who were unable to provide information due to severe physical or mental health conditions.

Since there had been no previous similar study that had used double population mean formula, the sample size was calculated by an online sample size calculator (https://www.surveysystem.com/sscalc.htm). So, it was calculated using a double population mean formula with a (95%) confidence level, (2%) margin of error, (80%), power, (β = 4.69, 95% CI: 2.63, 6.76) confidence interval from a previous study (22), and an average of 1,190 women who gave birth each month in the selected public hospitals as a source population. Then 10% of the calculated sample size was added to account for potential non-responses. And the final calculated sample size was 420.

We had proportionally allocated the calculated sample size (420) to each of the three selected hospitals based on their estimated number of monthly deliveries. The study participants were selected by a systematic random sampling technique. The sampling interval (k value) was calculated by dividing the total number of estimated monthly delivery reports by the required sample size, which is 1,190/420 ≈ 3. Finally, every third woman was recruited using her respective delivery registration number, and the first participant was selected by the lottery method.

The data collection tool was adapted by reviewing different literature (19, 22–24). This tool comprised sociodemographic characteristics, obstetrical factors, facility-related factors, and PCMC experiences of the participants. The data on PCMC items were collected using a validated person-centered maternity care scale adopted from previous studies (19, 22–24). The study participants' obstetrical data were extracted from the participants' medical records. The questionnaire was prepared first in English language and translated into local languages (Afan Oromo, Amharic, and Af-Somali) and back into English language by different language experts to ensure consistency.

Six BSc Midwives, who do not work in the study area, collected the data under the supervision of the principal investigator and four MSc Midwives. They conducted interviewer-administered face-to-face exit interviews in the postnatal unit, using each participant's respective medical record.

The questionnaire was pretested on 5% of the total sample size outside the study area to identify any ambiguity, check for consistency and acceptability, as well as to make necessary corrections before the actual data collection period. The questionnaire had a Cronbach's alpha of 0.89, indicating good internal reliability. Data collectors and supervisors were trained for one day by the principal investigator regarding the objectives of the study, data collection procedures, and the maintenance of confidentiality. The data were checked daily for completeness and consistency and, corrective measures were taken in a timely manner.

Person-centered maternity care (PCMC), which is composed of three domains, namely dignity and respect, communication and autonomy, and supportive care, was the outcome variable. Furthermore, sociodemographic characteristics (age, religion, marital status, residence, educational status, employment status, and average monthly family income), obstetrical factors (parity, ANC, number of institutional deliveries, mode of delivery, time of delivery, childbirth complications, and newborn outcome), and facility-related factors (sex of the main birth attendant and length of hospital stay) were the independent variables of this study.

PCMC is measured by the PCMC scale. The PCMC scale is a validated scale that comprises three domains (i.e., dignity and respect, communication and autonomy, and supportive care) (23, 24). There are a total of 30 items with each item having four-response options: i.e., 0- no, never, 1- yes, a few times, 2- yes, most of the time, and 3- yes, all the time. Negative items such as verbal abuse, physical abuse, and crowdedness were reversely coded to reflect a scale of 0 as the lowest level and 3 as the highest level. The total PCMC score is a summative score from the responses to the individual items, which ranges from 0 to 90 (19, 24). To enable easy comparison across the PCMC domains, the scores were rescaled and standardized to range from 0 to 100. Furthermore, verbal abuse is when a woman feels that the health providers shouted, scolded, insulted, threatened, or have spoken to her rudely. Physical abuse is when a woman feels that she is being treated roughly, such as being pushed, beaten, slapped, pinched, or physically restrained (24).

Our data was entered into EpiData version 4.6 software and then exported to SPSS version 26 for further cleaning, coding, and analysis. Descriptive statistics was carried out to compute frequencies, proportions, means, and standard deviations. The normality assumption was assessed using a P-P plot and histogram. Linearity was checked using a scatter plot, and multicollinearity was checked using the Variance Inflation Factor (VIF). After creating dummy variables, simple and multiple linear regression analyses was used to determine the factors associated with PCMC. Those variables with p-value ≤0.25 in the simple linear regression were qualified to the final model. Unstandardized β coefficient, along with a 95% CI was used to report the strength of association and statistical significance was declared at a P-value of <0.05.

Out of a total of 420 expected respondents, 412 provided complete responses, resulting in a response rate of 98.1%. More than half 242 (58.7%) of the study participants were from urban residences. The participants were in the age range of 16–42 years with a mean age of 25.83(±5.7) years. Nearly half 195 (47.3%) of these study participants had attended formal education (Table 1).

Above three-fourths 316 (76.7%) of the participants were multiparous. The average parity of the participants was three. More than a quarter 123 (29.9%) of them had a history of abortion. There were 297 (72.1%) vaginal deliveries among the 412 participants. Of these, 65.3% were spontaneous vaginal deliveries, and 6.8% were assisted by operative vaginal delivery methods. Almost six-in-ten 254 (61.7%) of the overall participants had faced childbirth complications (Table 2).

The participants mean percentage score of PCMC was 68.1 (CI: 59.94, 62.66) with a standard deviation of ±14.1. From the subscales, the rescaled mean score of dignity and respect was 80.6%, communication and autonomy 61.1%, and 67.3% for supportive care (Figure 1).

The percentage mean score of the study participants was 80.6 (±2.4). About two-thirds (66.6%, n = 272) of the total study participants felt that they were treated with respect all the time, and 240 (58.3%) of them reported that they were treated in a friendly manner all the time during their stay in the hospital. On the other hand, 61(14.8%) and 25 (6.1%) of women reported that they experienced verbal and physical abuse at least once during their stay at the hospital respectively (Table 3).

The percentage mean score of the participants was 61.1 (±6.2). More than three-fourths of the study participants 321 (77.9%) reported that providers never introduced themselves when they came to see them for the first time. More than half of the study participants 244 (59.2%) reported that providers called them by their names throughout their stay in the hospital. Slightly more than a quarter 108 (26.2%) women reported that providers had never explained the purpose of the examinations or procedures to them. Furthermore, one-third 134 (32.5%) of the participants reported that they were never asked for permission or consent during examinations (Appendix Table A1).

The percentage mean score of the supportive care subscale of the participants was 67.3 (±7.3). Nearly half of the participants, 201 (48.8%) reported that they were not allowed to be with someone they wanted during labor and the majority of them 369 (89.6%) were without a companion during delivery. In addition, 219 (53.2%) of the women reported that providers supported their anxiety and fears all the time during the childbirth process and 254 (61%) of them felt providers took the best care for them all the time (Table 3).

From the overall variables entered into the simple linear regression analysis, residence, ANC follow-up, time of delivery, childbirth complications, newborn outcome, and length of hospital stay were eligible for the multiple linear regression analysis based on a P-value of less than 0.25. Ultimately, women who had ANC follow-up, faced childbirth complications, gave birth to a live newborn, and more than two days length of hospital stay remained significantly associated with person-centered maternity care (PCMC).

Women who'd had ANC follow-up had increased PCMC score by about 6 units compared women who had no ANC follow-up (β = 5.66, 95% CI: 2.79, 8.53). Women who faced childbirth complications had a lower PCMC score than those who didn`t face childbirth complication by 7 units (β = −7.01, 95% CI: −9.88, −4.13). Additionally, a live newborn outcome increased PCMC by about 8 units as compared to dead fetus newborn outcome (β = 7.59, 95% CI: 3.97, 11.20). By keeping all other variables constant, PCMC score was decreased by 4 units on those who had more than two days of hospital stay (β = −4.08, 95% CI: −6.79, −1.38) (Table 3).

As the first study conducted in this area, it uncovered the status of PCMC and its associated factors. However, a limitation of this study is that it was conducted only in public hospitals, which means it did not address the status of PCMC in private facilities within the study area. Moreover, the data related to PCMC were collected solely from participants' self-reports, making them prone to social desirability and recall bias. To help participants recall information, the time frame was limited to the six weeks postpartum period. Also, participants were informed and assured that their responses will remain anonymous and confidential, to encourage more honest responses. Furthermore, data collectors were trained to avoid leading questions in order to reduce the pressure to respond in socially desirable ways.