Study setting and period
This nationwide assessment was conducted at 24 health care facilities in Ethiopia. Recently, Ethiopia has 273,601 heath workers employed in public health facilities; of them, 181,872 (66.5%) were health professionals, and the remaining 91,723 (33.5%) were administrative staff .
Among health professionals, the highest three professional categories were nurses, health extension workers and midwifery, which accounted for 59,063 (21.59%), 41,826 (15%) and 18,336 (6.71%), respectively. The private health sector provides work opportunities for approximately 60,000 human resources for health in Ethiopia. There were 17,162, 3678 and 314 functional health posts, health centers and all types of hospitals, respectively, across all regions of the country that provide health services to the community; the hospitals were 22 tertiary hospitals, 73 general hospitals and 245 primary hospitals . Although the Ethiopian health system shows improvement in quality health services, CEP services are limited. The study was conducted from February to April 2021. All methods were performed in accordance with relevant guidelines and regulations.
An institution-based cross-sectional quantitative study triangulated with a qualitative approach (phenomenology design) was conducted . Quantitative data were gathered through a structured questionnaire administered to the HCWs. The facility status towards CEP was assessed using a checklist.
Additionally, qualitative data were collected from KIIs (medical directors, chief executive officers (CEOs), quality focal persons, CRC focal persons, and quality directors) through an in-depth interview and FGDs (HCWs selected from each department).
The source population for this study was all HCWs who worked at primary, secondary and tertiary public hospitals in Ethiopia, whereas the study population was all HCWs who worked at the selected public hospitals in Ethiopia.
Participant health facilities
For this study, 24 health care facilities were selected from 9 regions and 2 city administrations. Six tertiary, eleven secondary and seven primary hospitals were selected from all regions in Ethiopia. Two health care facilities were selected from each (Addis Ababa, Sidama, Afar, Benshangul Gumuz, Gambella), three health care facilities from Amhara, Oromia, Somali, SNNP and one health care facility selected from Harari and Dire Dawa. HCWs who had been working for more than six months in 24 health care facilities in Ethiopia were included. HCWs who had been on annual leave and transferred from other health care facilities that served less than six months were excluded.
For the qualitative study, the study participants were recruited from 24 health care facilities depending on their position and clinical experience using a purposive sampling technique. The KII participants were CEOs, medical directors, CRC focal persons, and quality focal persons. We invited them to participate in FGDs and in-depth interviews through invitation letters.
Sample size determination and sampling procedure
The sample size was determined by using Epi-Info version 3.5.1 software  by a single population proportion formula with the assumptions of a 95% confidence level and 5% precision, taking a 50% proportion due to the lack of a previous study. A sample size of 435 was obtained after adding a 13% nonresponse rate. The sample size was proportionally allocated based on the number of HCWs per facility.
The sampling frame was prepared for each hospital by obtaining lists of HCWs from the directors and the human resource management office. Finally, 435 study subjects were selected from participating hospitals using a simple random sampling technique.
Purposive sampling was used to collect qualitative data, with 47 study participants interviewed. The trustworthiness of qualitative data was determined by considering plausible information gathered from participants through face-to-face interviews.
For the qualitative study, 36 participants were selected from health care facilities for FGDs (four FGDs contained 7 participants, and one FGD contained 8 participants). For KIIs, eleven KII participants (CEO, medical director, CRC focal person, and quality focal person) were selected from five healthcare facilities.
Data collection tools, procedures and quality assurance
Data were collected using a standardized and pretested questionnaire. The questionnaire was adapted from different works in the literature and/or developed from other similar studies [34, 35]. The questionnaire contains sociodemographic characteristics, clinical ethical practice, knowledge and attitude of ethics, professional satisfaction, availability of clinical ethical committees (CECs), previous educational curriculum, and facilities auditing (an observational checklist) and was employed to observe and check the level of CEP (Additional file 1).
The tool's reliability test revealed a good estimate of the measurement: knowledge of CEP measured with 13 items and scale reliability coefficient (Cronbach's alpha) of 0.90, CEP measured with 12 items and scale reliability coefficient (Cronbach's alpha) of 0.83, and attitude towards CEP measured with 12 items and scale reliability coefficient (Cronbach's alpha) was 0.69. The overall scale reliability coefficient (Cronbach’s alpha) of all 37 items was 0.82. The questionnaire was prepared in the English language, translated into Amharic, Afan Oromo, and Somali languages and back-translated to English language-by-language experts. Data were collected using a self-administration questionnaire for HCWs.
Data were collected by 34 data collectors and supervised by 17 supervisors. Data collectors and supervisors were recruited based on their ability to speak Amharic language and fluent in each specific region language because they were recruited for data collection and supervision in regions with previous data collection experience.
The interviewers were health professionals with BSc and above and fluent speakers of Amharic language trained on the tools. The networks of the technical working group (TWG) were organized to enhance the data collection process. Three days of training was given for the data collectors and supervisors to adhere to the objective of the study. Additionally, COVID-19 prevention protocol training was given. The tools were also pretested (5% of the sample) in the health care facilities (non-selected facility). Based on the pretest findings, the tools were validated. During data collection, the tools were also checked by the investigators for completeness and consistency. In addition, the tools were coded during data collection and before entering the computer.
The qualitative data were collected through guiding questions in the Amharic language. A semi-structured interview guide was developed to assess CEP, awareness and factors associated with CEP. During KIIs and focus group discussions (FGDs), one modulator and reporter for each KII and FGD were recruited. The voices of the interviewed participants were recorded during the discussion and probing.
The outcome variable of CEP was measured using 12 items. Good practice (participants who scored > 75% and poor practice (participants who scored < 75% on practice-based questions) .
Participants who scored > 75% were labeled as having a favorable attitude and participants who scored < 75% were labeled as having an unfavorable attitude.
The level of knowledge and attitude were measured using 13 items each. Participants who scored > 75% were labeled knowledgeable, and those who scored < 75% were not knowledgeable .
Facility level CEP audit: measured with 15 items and reported as good score greater than 75% and poor if it is less than 75%.
Data processing and analysis
Quantitative data were cleaned and entered into EPIDATA version 3 and STATA version 14 to obtain summary figures and percentages. Pearson’s chi-square and chi square tests of association were applied to look for differences. Multiple logistic regressions were performed to assess the strengths of the association. The analytic part was analyzed with their 95% confidence interval, and a two-tailed p value was calculated to identify the presence and strength of association. The categorical variables were dichotomized based on the cut-off point of the mean for binary logistic regression. Variables with a p value ≤ 0.2 in the binary analysis were included in a multivariable logistic regression analysis to control the confounding effect among the variables. Statistical significance was declared if the p value < 0.05. The satisfaction of health professionals with their current profession, the availability of a functional clinical ethics committee, compassionate leaders, and pre-service clinical ethics education were all covariates in the multiple logistic regression model.
The qualitative data were transcribed and translated from local languages to English. The qualitative analysis was facilitated by Atlas.ti version 7. The verbatim transcripts were separately verified against the interviewer's audio-recorded and analysis team members. The analysis team further guaranteed the consistency of each focus group and interview session. The teams also analyzed the interviewer's adherence to each transcript's probing questions and provided feedback and correction.
They read through each transcript again to reduce differences between individual codes and marked out the transcript that was non-specified to CEP through the agreement. Rigors were further assured by the study team, who was not being analyzed in the interviews or focus groups, independently auditing the coding process. Finally, the narrative was organized and integrated according to emerging themes and concepts set in the research objectives. The result was triangulated with quantitative findings at the interpretation level. A thematic analysis method was employed to report the qualitative study.