Patient perspectives on interpersonal aspects of healthcare and patient-centeredness at primary health facilities: A mixed methods study in rural Eastern Uganda

Introduction: Patient-centered care (PCC) is an approach to involve patients in health care delivery, to contribute to quality of care, and to strengthen health systems responsiveness. This article aims to highlight patient perspectives by showcasing their perceptions of their experience of PCC at primary health facilities in two districts in Uganda. Methods: A mixed methods cross-sectional study was conducted in three public and two private primary health care facilities in rural eastern Uganda. In total, 300 patient exit survey questionnaires, 31 semi-structured Interviews (SSIs), 5 Focus Group Discussions (FGDs) and 5 feedback meetings were conducted. Data analysis was guided by a conceptual framework focusing on (1) understanding patients’ health needs, preferences and expectations, (2) describing patients perceptions of their care experience according to five distinct PCC dimensions, and (3) reporting patient reported outcomes and their recommendations on how to improve quality of care. Results: Patient expectations were shaped by their access to the facility, costs incurred and perceived quality of care. Patients using public facilities reported doing so because of their proximity (78.3% in public PHCs versus 23.3% in private PHCs) and because of the free services availed. On the other hand, patients attending private facilities did so because of their perception of better quality of care (84.2% in private PHCs versus 21.7% in public PHCs). Patients expectations of quality care were expressed as the availability of medication, shorter waiting times, flexible facility opening hours and courteous health workers. Analysis of the 300 responses from patients interviewed on their perception of the care they received, pointed to higher normalized scores for two out of the five PCC dimensions considered: namely, exploration of the patient’s health and illness experience, and the quality of the relationship between patient and health worker (range 62.1-78.4 out of 100). The qualitative analysis indicated that patients felt that communication with health workers was enhanced where there was trust and in case of positive past experiences. Patients however felt uncomfortable discussing psychological or family matters with health workers and found it difficult to make decisions when they did not fully understand the care provided. In terms of outcomes, our findings suggest that patient enablement was more sensitive than patient satisfaction in measuring the effect of interpersonal patient experience on patient reported outcomes. Discussion and conclusion: Our findings show that Ugandan patients have some understanding of PCC related concepts and express a demand for it. The results offer a starting point for small scale PCC interventions. However, we need to be cognizant of the challenges PCC implementation faces in resource constrained settings. Patients’ expectations in terms of quality health care are still largely driven by biomedical and technical aspects. In addition, patients are largely unaware of their right to participate in the evaluation of health care. To mitigate these challenges, targeted health education focusing on patients’ responsibilities and patient’s rights are essential. Last but not least, all stakeholders must be involved in developing and validating methods to measure PCC. This is a mixed methods cross-sectional study using patient exit survey questionnaires, semi-structured Interviews (SSIs), Focus Group Discussions (FGDs) and feedback meetings. This study was conducted between late 2017 and 2018 in the eastern Ugandan districts of Iganga and Mayuge, within the Iganga Mayuge Health and Demographic Surveillance Site (IMHDSS). It has a population of about 67,000 people in about 13,000 households. The IMHDSS is located on the boundary between the districts of Iganga and Mayuge, about 115 km from the capital Kampala. The area is predominantly rural with only about 10% living in a peri-urban environment. The majority of people are of Busoga culture and speak of local language of Lusoga. The Ugandan health system is organized into six levels of health care: level I comprises of village health teams (VHTs) and community medicine distributors; Health Centre II (HC II) led by an enrolled nurse; Health Centre III (HC III) led by a senior clinical officer; Health Centre IV (HC IV) and/or district hospital led by a senior medical officer; regional referral hospitals, and finally the National Referral and Teaching Hospital. Our study was conducted in the IMHDSS catchment area that has one district hospital, four government HC IIIs, three non-government HC IIIs, five government HC IIs and three Non-Governmental Organization (NGO) HC IIs. We selected level III facilities to enable the analysis of perceptions from patients visiting the facility for different forms of curative care. We specifically focused on people who came “once-off” at the clinic (routine care); women/parents coming as part of a pregnancy/child follow-up with planned re-visits whereby interpersonal relations with staff are bound to develop (maternal and child health care); and patients with chronic illnesses coming for follow-up in specialized clinics where repeated visits and external evaluations result in different relationships to the health practitioners, health system and different expectations in terms of quality of care (specialized clinics). A HC III has, on average, about 18 staff, led by a senior clinical officer, with a general outpatient clinic, a maternity ward and a laboratory. We selected five HC III facilities according to the following criteria: inclusion of both governmental as well as private health facilities in both urban and rural settings; different demographic and epidemiological characteristics; different experiences with community strategy and PCC approaches, if any. We selected three public HC IIIs and three private-for-profit HC IIIs (see Table 2). Private health facilities are mostly located in semi-urban areas and have more clinical staff. Despite these differences, utilization rates, calculated as number of contacts per inhabitant (based on 2014 census) per year, are roughly similar in public and private facilities (ranging from 0.30 to 0.42 contacts per inhabitant per year) with the exception of facility 3 (utilization rate of 0.56). The utilization of routine all-round care services is many times higher than the utilization of maternal and child health care, and of more specialized care. Notable is that facility 6 is rather atypical: it has a substantially lower catchment population with fewer patients (in absolute numbers) using its services than is the case in the other five facilities, while focusing on care for pregnant women (see Table 2 in the results section). We decided therefore to exclude it from our analysis. Further detailed descriptions of the stakeholders involved in the provision of PCC at primary health care level in Uganda can be found in a paper by Waweru et al. [39]. *for facility 4, we obtained an overall annual utilization rate, but have not the data to disaggregate utilization per type of service used Using the framework described, we developed three data collection tools: patient exit questionnaires, semi-structured interview guides and focus group discussion topic guides (see S1 Appendix). Validated tools and questions were also added [40, 43] and a tool used to measure quality of care at health centre level in Zimbabwe [44]. Further detail on how literature and existing validated tools contributed to the design of tools to measure PCC in Uganda can be found in Supplementary file S1 Table—a table that compiles a detailed list of the dimensions (components of PCC), and the contribution (adopted questions) of each validated instrument to each dimension. The patient exit questionnaire included questions about why they chose to visit the facility, who they interacted with, and how long they had to wait at each health point. Questions about the patient’s perception of the care they received that day included a Likert scale score structured according to the 5 dimensions of PCC (exploring perceptions on health and the illness experience, understanding the whole person, finding common ground, enhancing the patient doctor relationship, and health promotion). Semi-structured interviews (SSIs) were conducted to solicit patients’ perceptions on health-seeking behavior, who they thought was responsible for their health, their experiences at health facilities, their relationships with health workers and VHTs, their membership in support groups, their awareness of their rights and responsibilities as a patient, as well as how all these factors contribute to their perceptions of the quality of primary health care available to them (see sample questions in supporting file S1 Appendix). A follow-up Focus Group Discussion (FGD) was held with patients from each facility–the same as those who had participated in the SSIs–to validate and clarify the key messages (see S1 Appendix). As explained above, we did not collect data on adherence as the measurement techniques vary for different services and illnesses. For the quantitative patient survey, we estimated that a sample size of 240 patients would allow us to detect differences in average patient perceptions on their experience of PCC using frequencies between groups (public versus private) and services received (routine care, maternal and child health care or specialized care for patients with chronic illness) [45]. 60 patients were recruited per facility (20 patients receiving routine care, 20 receiving care at specialized clinics and 20 receiving curative care at the maternal and child health clinic). Our final sample of 300 patients, after the exclusion of patient responses from facility 6 was sufficient to describe the difference in mean scores for each dimension, and detect a mean difference of 0.3, with a significance level of 5% (p<0.05), at 90% power after accounting for 30% attrition, as outlined in a review on self-management interventions for people living with HIV/AIDS in Africa by Aantjes et al. [46]. Four field assistants were selected based on their experience with both quantitative and qualitative research, three of them had also worked with the IMHDSS teams previously. They were trained for 2 weeks on the concept of PCC, how to administer the patient exit questionnaires and how to moderate a focus group discussion. At the end of the training period the field assistants (under the supervision of the Principal Investigator (PI)) piloted the consent forms and tools with patients. From the pilot testing, we edited some of the questions, for example, the questions on how many nurses, clinical officers or lab technicians was generalized to ‘how many health workers did you interact with today’ as we discovered patients could not distinguish cadres of staff (including ourselves). Some words like patient rights and responsibilities (idemberio) were not easily understood and sometimes had to be explained from the starting point of a child’s right to be fed, protected etc.; and the mother’s responsibilities towards the child. Additionally, we also carried a summary of the patient rights charter to list the rights where the respondent did not understand completely. Empathy was also a concept that was understood as sympathy or taking pity and we had to train the field assistants to ask the question in order for the patient to understand it as ‘the health worker putting themselves in your position (wearing your shoes) and sharing your feelings’. Of the patients visiting the five HC IIIs, 300 patients were recruited for exit surveys and 31 patients were purposively recruited for semi-structured interviews which were conducted at their homes. At each facility, the principal investigator (PI; first author EW) or a field assistant provided study information during the morning health talks and gained initial consent from patients interested to participate. Only patients receiving curative care at the out-patient department of the five HCs were included in the study. Furthermore, an effort was made to ensure that interviewees were representative of the three categories of patients receiving (1) routine care, (2) curative maternal and child health care, or (3) attending a specialized clinic (people living with diabetes or HIV/AIDS). If the patient had provided initial consent, EW would request to sit in during their consultation with the health worker. After patients received curative care, a sequential sample was taken where every 5th patient for the routine care department; every 2nd patient at the MCH or specialized clinic were recruited for an exit questionnaire interview. Only two patients refused to be interviewed citing lack of time and they were replaced. An interviewer-administered questionnaire was filled in for each selected patient / caregiver (in either the local language of Lusoga or English) (see S1 Appendix). At each facility 60 exit interviews were conducted. After the questionnaire, a request was made to visit the patient at their home or a convenient location and date for them, for an SSI, including 2 patients from each of the 3 service areas. We also tried to keep a balance between male and female participants. In total, 31 SSIs were conducted (one facility had an extra patient interviewed). A follow-up FGD was held with patients from each facility–the same as those who had participated in the SSIs–to validate and clarify the key messages. In total five FGDs were conducted with 30 patients (5–7 patients in each FGD). In addition, 5 feedback talks were organized per facility with all patients present at the facility during morning health talks on the day of the scheduled feedback meeting (see Table 1). Data was collected between February and August 2018. Data analysis was conducted according to the conceptual framework i.e. categorized into three major areas of exploration: (1) understand the patients’ health needs, preferences and expectations, and factors that influence their health seeking behavior, (2) describe patients perceptions of their care experience, and (3) articulate patient reported outcomes and their recommendations of how to improve the quality of care that they receive (see Fig 2). Data from the survey questionnaire were entered into excel worksheets at the end of each day of data collection. Random checks were conducted by the PI to ensure completeness of data at the field. The compiled data was then imported to STATA version 14 where it was cleaned and data from one of the facilities was excluded as explained above (see data collection section). Although the ordinal variables have a rank order but cannot be conceived as having an underlying measurable standard (e.g. the interval difference between a strongly agree and agree response is not standardized), data from patient reported measures have been coded in this way to provide feedback that managers and policy makers can easily understand. For each question, strongly agree was recoded as 5 and strongly disagree as 1 so that higher values could reflect more patient-centeredness. Subsequently, for each of the PCC dimensions above, patients’ responses were then summed and normalized to fit on a scale of 1–100 i.e. (the sum of the scores / (number of questions*5))*100. This normalized score enables a comparison of patient experiences: across the 5 PCC dimensions, depending on the type of facility (public or private PHC), and according to the service that patients received (routine care, maternal and child health care and specialized clinic care for HIV/AIDS or Diabetes). The results were summarized in terms of the frequency of patients who gave positive or negative perceptions of PCC in public and private facilities, and normalized scores were calculated. For this article, chi-squared tests have been used to determine whether there are significant differences reported between patients attending public and private facilities (with a null hypothesis that there was no difference in patient-centeredness between public and private facilities). Within each category (public/private facilities) further descriptive analysis was conducted to describe differences in the perception of PCC in patients receiving routine care, maternal and child health care or specialized care. Determinants of the outcomes of interest (patient satisfaction and patient enablement) were investigated by performing univariate and multivariate regressions with patient perceptions of PCC, interaction effects, facility type, type of service received, age, gender and literacy as independent variables. Data on adherence was not presented because responses were too varied as they depended on the type of illness, whether the patient had to buy additional drugs, support at home and other factors that could not be comparable across patients or facilities. Qualitative data was converted into digital text format for analysis. Audio recordings of SSIs and FGDs were transcribed, and notes from feedback sessions were typed. Any text in Lusoga was translated into English. The written text was imported into NVivo 11, a qualitative data analysis software, for organization and qualitative thematic framework analysis as described by Gale et al. [47]. Thematic framework analysis of SSIs and FGDs was begun in the field by the first author. Following the first phase of data collection using patient exit questionnaires, qualitative questions were developed by the research team to clarify patient responses in the patient exit survey interviews. Emerging themes were added on while conducting the SSIs, and during de-brief meetings with translators. The themes were further explored and the information validated during follow-up FGDs and feedback meetings with patients. Post data collection, the research team developed a coding framework for use in NVivo 11 guided by the conceptual framework for understanding patients’ perception of PCC, and collected data. Data was coded according to the framework in NVivo 11 and organized into framework matrices or charts. Interpretation of data was conducted by the research team including the PI, research assistants who spoke the local language and supervisors. The interpretation of data was also validated through feedback meetings with patients, health workers and district health managers. Ethical approval for this study was obtained from the Institute of Tropical Medicine PhD committee (IRB/AB/ac/081), the Institute of Tropical Medicine Institutional Ethics Review Board (1166/17); University of Antwerp Ethics Review Board (17/24/278); the Makerere University School of Public Health Institutional Review Board (500); and the Ugandan National Council for Science and Technology. Prior to participation, written informed consent was obtained from stakeholders at national, district and facility level. In cases where patients could not write or provide a signature, verbal informed consent was obtained from patients and recorded in the study’s copy of the informed consent form.