Patient-reported quality of primary care and facility endorsement in Addis Ababa

Abstract

Background

High-quality primary healthcare is essential for improving health outcomes and patient satisfaction. Facility endorsement—patients’ willingness to recommend a health facility—reflects trust and perceived service quality and can provide actionable feedback for quality improvement. This study assessed perceived quality of care and its association with facility endorsement among primary care patients in Addis Ababa.

Methods

A mobile-based interactive voice response (IVR) survey was administered to 1,584 patients receiving primary care services from public and private health facilities in Addis Ababa. Perceived quality of care was assessed across three dimensions: structural quality, provider competence, and patient-centered care. Linear and ordered logit regression models were used to examine associations between quality dimensions and endorsement, and a Chow test assessed structural differences across facility types.

Results

Overall perceptions of care quality were moderate, corresponding to average scores of around 50–60% across all quality dimensions, indicating room for improvement. Average scores across facility types were 55.9% for structural quality, 58.0% for provider competence, and 51.2% for patient-centered care. Private facilities—particularly clinics—consistently outperformed public hospitals and health centers, with 60–70% of patients in private facilities rating quality indicators as “good” or “very good,” compared with 40–50% in public facilities. The Chow test confirmed significant differences in patient responses by facility type (χ² = 96.7, p < 0.001). Provider competence was a strong and consistent predictor of facility endorsement across all settings; for example, in private hospitals, a one-unit increase in perceived competence more than doubled the odds of endorsement (OR = 2.37, 95% CI: 1.42–3.95). Patient-centered care was also positively associated with endorsement across all facility types. Structural quality influenced endorsement only in public hospitals.

Conclusion

Perceived quality of primary care in Addis Ababa remains moderate, with clear disparities between public and private facilities. Provider competence and patient-centered care are the strongest drivers of patient endorsement, highlighting the importance of clinical skills, communication, and respectful care as key factors associated with patient endorsement. In public hospitals, addressing structural gaps through targeted investments in infrastructure, medicines, and diagnostics remains important for improving patient experience and may contribute to broader system performance.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12913-026-14750-5.

Introduction

Ethiopia continues to face high levels of preventable morbidity, disability, and premature mortality, much of which is attributable to gaps in the quality of health care delivery [1, 2]. Evidence consistently shows that poor-quality care is widespread across the Ethiopian health system, particularly within primary care services, where deficits in infrastructure, provider competence, and patient-centered care persist [2–5]. These challenges have been further exacerbated by recent system shocks, including the COVID-19 pandemic and prolonged armed conflict in parts of the country [6, 7]. Despite these pressures, little is known about how perceived quality of care varies across different facility types—public health centers, public hospitals, and private providers—or how these perceptions influence patients’ confidence in and endorsement of health facilities.

Global health system frameworks emphasize that delivering high-quality, people-centered care is essential for improving health outcomes and strengthening trust in health systems. The Lancet Global Commission on High-Quality Health Systems highlights effectiveness, patient-centeredness, and equity as core attributes of quality care [2]. Similarly, the Institute of Medicine defines six key dimensions of quality, including safety, effectiveness, timeliness, efficiency, equity, and patient-centeredness [8], while the WHO Health System Framework emphasizes the role of service delivery, health workforce, essential inputs, and information systems in ensuring quality [9]. Together, these frameworks underscore the importance of measuring not only structural readiness and provider competence but also patients’ experiences of care as essential components of health system performance.

In Ethiopia, improving quality of care has been a central pillar of national health policy. The Federal Ministry of Health has articulated this commitment through the National Quality Strategy and successive Health Sector Transformation Plans [10–12]. However, major gaps in service readiness and performance remain. The 2021 Service Provision Assessment reported that only one in five health facilities provides all basic services, with persistent shortages of health workers, medicines, diagnostic capacity, and basic infrastructure [13]. Other studies similarly report low adherence to clinical guidelines in primary care settings, with quality scores ranging widely across facilities and averaging well below optimal levels [4]. These systemic shortcomings continue to undermine the delivery of effective and equitable primary healthcare services.

Beyond structural and clinical deficits, patients’ perceptions of care quality play a critical role in shaping healthcare-seeking behavior, future utilization, and trust in the health system [14, 15]. In Ethiopia, willingness to recommend a facility—a proxy for patient satisfaction and confidence—has important implications for service uptake, treatment adherence, and community trust [16–19]. Poor perceived quality discourages repeat visits and referrals within social networks, eroding confidence in public facilities and encouraging bypassing toward private providers [20, 21]. Such patterns can increase out-of-pocket spending, reduce public-sector revenue, and exacerbate inequities in access to care, particularly when lower-income populations are left with under-resourced public services.

Despite the importance of patient-reported quality and endorsement, evidence in Ethiopia remains limited. Routine health information systems primarily capture service volume and clinical indicators, offering little insight into patient experience or perceived quality [22, 23]. Moreover, few studies have examined how different dimensions of perceived quality—structural quality, provider competence, and patient-centered care—relate to facility endorsement across public and private providers. Addressing this gap is essential for identifying targeted quality improvement priorities and strengthening trust in primary care services.

Guided by the Lancet High-Quality Health Systems framework and the WHO Health System Framework, this study assesses perceived quality of primary care and its association with patients’ willingness to recommend health facilities in Addis Ababa. Specifically, it examines how multiple dimensions of perceived quality relate to facility endorsement and compares these relationships across public and private providers. To achieve these objectives, the study uses data from an Interactive Voice Response (IVR) survey, an innovative and underutilized approach for collecting patient-reported health system data in Ethiopia [24]. By linking patient perceptions of quality to endorsement across facility types, this study generates policy-relevant evidence to inform quality improvement efforts and strengthen people-centered primary healthcare in urban Ethiopia.

Methods

Data source

We conducted an Interactive Voice Response (IVR) patient satisfaction survey among patients seeking outpatient primary care services across four types of health facilities: public health centers, public general hospitals, private clinics, and private hospitals. IVR is an automated phone-based survey method where participants respond to pre-recorded questions using their keypad. It was well-suited for our study in Addis Ababa due to high mobile phone access, reduced social desirability bias, and its efficiency in reaching diverse patients across facility types. Details of the IVR survey we employed are described in our earlier paper [25].

The patient satisfaction survey was developed based on the Lancet Global Commission on High Quality Health Systems framework [2]. The survey included a total of 15 questions including 12 questions grouped under three dimensions of quality, namely structural quality, provider competence, and patient-centered care. Participants were also asked to report their monthly household income, the level of affordability of the services received and how likely they were to recommend the health facility to a family member or friend. The survey included mostly 5-point Likert scale questions. Respondents were instructed to use their mobile phone touch pad to answer. Participants were permitted to repeat or skip the questions. A message was played to notify them if they pressed an invalid key. If respondents either pressed an invalid key or did not answer within 20 s, the survey question was repeated. The call would be disconnected if they pressed an invalid key three times or failed to respond after three repetitions of the question. The survey methodology and response rates are described in more detail elsewhere [25]. The survey instrument is provided in the Supplementary File.

Enrollment

Using facility lists from the DHIS2, we randomly selected four facilities from each of four categories: public health center, public hospital, private clinic, and private hospital. Using facility lists from DHIS2, we randomly selected four facilities from each of four categories: public health centers, public hospitals, private clinics, and private hospitals. An equal number of facilities was selected from each facility type to enable meaningful and appropriate comparisons. However, the number of sampled collected from each facility was determined based on patient load data obtained from DHIS2. Accordingly, samples were allocated to each facility proportional to its patient load. Within each selected facility, eligible patients were contacted and recruited consecutively until the requited number of samples from each facility was obtained.

Individuals were eligible for the survey if they were aged 18 and above, they came to receive outpatient care for themselves, they received at least one of the 11 services stated above, they had a functioning mobile phone with them at the time of recruitment, and they spoke or understood the IVR survey languages (Amharic or Afaan Oromo). Inpatients were excluded from the study. First, consent and phone numbers were obtained in person from primary care patients at health facilities. Patients’ age, gender, education level, preferred time for the IVR survey (morning, midday, afternoon, evening or night), and type of primary care service received were collected during in-person the recruitment. To ensure comparability in patient groups across facility types, the study was limited to 11 types of primary care services including care for: chronic diseases (hypertension, diabetes, or kidney disease), an injury, infectious disease testing or treatment (tuberculosis, sexually transmitted infections, HIV), a gastrointestinal condition, or a respiratory disease. Women receiving prenatal care or family planning were also eligible. The recruitment took 14 days. Before conducting the survey, cognitive testing was conducted to check whether participants understood the IVR survey. During recruitment, study participants were also introduced to the IVR system by listening to a sample survey question to familiarize themselves with its functionality. Participants were advised to answer to the questions honestly. A 50 Ethiopian Birr mobile phone airtime credit (approximately US$ 0.90 at the time of the study (July 2024)) was included as incentive for those who completed the entire survey. As the survey was phone-based, a limitation of this approach is that it may have excluded individuals who did not own or could not operate a mobile phone [26]. In the current study, among the 3,167 eligible patients approached, 182 individuals (5.7%) were excluded due to not having a functioning mobile phone during the recruitment period.

Measurement of outcomes

The outcome variables for this study included facility endorsement and 12 indicators of patient-perceived quality. Facility endorsement was used as a proxy for patient satisfaction because it reflects patients’ confidence in and satisfaction with the health system, and indicates whether care is timely, effective, and appropriate [27, 28]. It was measured by asking respondents how likely they were to recommend the healthcare facility to family or friends, with responses captured on a 4-point Likert scale: very likely, somewhat likely, somewhat unlikely, and not at all likely. This variable serves as a key proxy for assessing patients’ overall experience and their propensity to advocate for the facility within their social networks.

The 12 patient-perceived quality indicators were captured on 5-point Likert scales from very poor to very good. Structural quality included four questions asking the patient to rate the facility cleanliness, the availability of essential equipment, medicines, and diagnostic services. Provider competence included four questions rating the provider skills, effectiveness of care in solving the patient’s problems, service coordination, and the clarity of the provider’s explanations. Patient-centered care focused on respect for patients, auditory and visual privacy, convenience of operating hours, and waiting times. By treating endorsement as an ordinal outcome variable, the study sought to provide a more nuanced understanding of how different aspects of perceived care quality influence patients’ likelihood of recommending healthcare facilities. The 12 perceived quality indicators were analyzed both individually and as composite scores at the dimension level and overall service quality level. First, a composite score was created for each of the three quality dimensions by calculating the mean of the four indicators within each dimension after confirming internal consistency. Subsequently, an overall (grand) composite quality score was developed by averaging the three dimension-specific composite scores, as strong internal consistency was confirmed (Cronbach’s α = 0.86).

Statistical analysis

Descriptive and inferential statistics were employed. Inferential statistics included non-parametric tests such as chi-square test and Kruskal Wallis test. To identify factors associated with perceived quality of care, a linear regression model was applied, adjusting for age, education level, income, cost of services, and facility type. The outcome variable, perceived quality of care, was treated as a continuous measure by converting the 12 perceived quality indicators into three variables (dimensions) and calculating average perceived quality. An ordered logit model was used to examine the association between perceived quality of care and patients’ willingness to endorse a facility, with endorsement measured on a 4-point Likert scale ranging from “not at all likely” to “very likely.”, controlling for covariates such as age, education level, income and perception of affordability of services. Prior to interpreting results, a likelihood-ratio Chow test was conducted to assess structural stability in coefficients. This test is important in understanding whether different subgroups respond differently to different variables [29]. In this study, the Chow test was specifically used to assess whether the effect of different variables on willingness to endorse a facility varied across the pooled model and the independent facility-specific models. In addition, model selection criteria—specifically the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC) were used.

Results

Profile of the participants

A total of 1,584 patients were enrolled in the study across four healthcare facility types in Addis Ababa. As reported in Table 1, more than half of the participants were female (57.3%), with the highest proportion of female respondents in public hospitals (65.5%). The age distribution showed that younger adults (18–30 years) constituted the largest group (39.3%), particularly in health centers (45.4%), while older adults (aged ≥ 50 years) were more represented in public hospitals (20.4%). Amharic was the dominant language among the majority (82.7%) of participants, with 17.3% speaking Afaan Oromo, most of whom utilized health centers (22.0%) and private clinics (20.4%), which tend to serve individuals from rural areas surrounding Addis Ababa.

Table 1.

Characteristics of the patients enrolled in the IVR survey, N = 1,584

Variable	Health centers		Public hospitals		Private hospitals		Private clinic		Total
	N	%	N	%	N	%	N	%	N	%
Gender
Female	239	53.95	269	65.45	177	53.15	223	56.17	908	57.32
Male	204	46.05	142	34.55	156	46.85	174	43.83	676	42.68
Age group
18–30	201	45.37	142	34.55	136	40.84	144	36.27	623	39.33
30–39	141	31.83	130	31.63	120	36.04	125	31.49	516	32.58
40–49	62	14.00	55	13.38	43	12.91	59	14.86	219	13.83
>=50	39	8.80	84	20.44	34	10.21	69	17.38	226	14.27
Education
No education at all	27	6.09	17	4.14	2	0.60	21	5.29	67	4.23
Primary education	140	31.60	106	25.79	11	3.30	69	17.38	326	20.58
Secondary education	146	32.96	128	31.14	45	13.51	117	29.47	436	27.53
Higher education	130	29.35	160	38.93	275	82.58	190	47.86	755	47.66
Care type
Maternal or reproductive	110	24.89	138	33.91	88	26.51	38	9.60	374	23.72
Communicable disease	46	10.41	36	8.85	13	3.92	11	2.78	106	6.72
Non-communicable disease	145	32.81	183	44.96	132	39.76	201	50.76	661	41.92
Diarrhea and respiratory	141	31.90	50	12.29	99	29.82	146	36.87	436	27.65
Language
Afan Oromo	101	22.80	47	11.44	45	13.51	81	20.40	274	17.30
Amharic	342	77.20	364	88.56	288	86.49	316	79.60	1310	82.70

Note: The table above shows the characteristics of those study participants who answered at least one IVR survey question

Educational attainment varied considerably across healthcare facilities. Private hospitals had the highest proportion of participants with higher education (82.6%), while individuals with no formal education were more prevalent in health centers (6.1%) and private clinics (5.3%). Overall, nearly half of the participants (47.7%) had attained higher education, while 20.6% had completed only primary education.

Non-communicable diseases (NCDs) were the primary reason for seeking healthcare services, accounting for 41.9% of healthcare visits. NCD-related care was most frequently sought in private clinics (50.8%) and public hospitals (45.0%). Maternal and reproductive health services were predominantly utilized in public hospitals (33.9%), while care for communicable diseases was the least common (6.7%) across all facility types.

Perceived quality and affordability across facility type

As shown in Fig. 1 and supplemental Table 1, private facilities, including both private hospitals and clinics, consistently outperformed public facilities across all 12 patient-perceived quality indicators. Public health centers and public hospitals performed similarly overall, each ranking lowest in 6 of the 12 indicators. Private clinics performed best overall, ranking highest in 10 of the 12 quality indicators, while private hospitals ranked highest in the remaining 2 indicators. A Chi-square test confirmed a significant association between facility type and quality indicators (p < 0.05), demonstrating that private facilities, particularly clinics, are rated more highly by patients.

Fig. 1.

Proportion of patients rating each quality item as “good or very good” by facility type

Quality dimension

Overall, respondents reported average perceptions of care quality, with clear gaps across all three dimensions. Averaging the item-level scores across facility types, structural quality was rated at 55.9%, provider competency at 58.0%, and patient-centered care at 51.2%. This indicates that a substantial portion of patients viewed care as inadequate—for example, nearly half of the patient-centered care was perceived as unmet, and over 40% of structural and competency expectations were rated below “good.”

The differences between public and private facilities were particularly pronounced. As shown in Fig. 2, private clinics and hospitals consistently scored higher than public hospitals and health centers, with structural quality at 70.1% and 67.5% versus 44.1% and 45.4%, respectively, reflecting gaps of 25–26% points. Similar disparities were seen in provider competency (68.7% vs. 49%) and patient-centered care (63.2% vs. 40.7%), with public facilities recording the lowest proportions of patients rating care as “good” or “very good.” ANOVA results confirmed these differences were statistically significant (p < 0.001), highlighting that overall perceived quality remains modest and that substantial gaps persist between public and private facilities.

Fig. 2.

Perceived quality of care by facility type

Endorsement of health facilities

The pattern of endorsement, as depicted in Fig. 3, mirrors the variations in perceived quality. More than 82.1% of patients were “somewhat” or “very likely” to recommend a facility to family or friends. Private clinics had the highest endorsement rate (87.4%). In contrast, health centers recorded the lowest percentage of endorsement, with only 78.2% of respondents likely to recommend the facility. Higher endorsement rates for private facilities reflect their relatively better performance across quality indicators. A Kruskal–Wallis test confirmed a significant association between facility type and endorsement (χ²(3) = 13.302, p = 0.004), indicating that endorsement varies meaningfully across facility types.

Fig. 3.

Percentage of primary care patients likely to recommend their facility to family or friends, by facility type

Factors associated with perceived quality of primary health care services

We assessed factors predicting positive quality perceptions using a linear regression model. In this analysis, the outcome, a continuous variable from 1 to 5, was derived as the average perceived quality indicators. Table 2 summarizes results from the linear regression model.

Table 2.

Results of a linear regression model for the factors associated with perceived primary health care service quality

Variables	Quality of care
Sex (ref=Male)	-0.0618
	[-0.153,0.0294]
Age category (ref = < 30)
30–39	-0.108 *
	[-0.206,-0.00962]
40–49	-0.123
	[-0.282,0.0364]
50–59	-0.213 *
	[-0.416,-0.0113]
>=60	-0.219
	[-0.448,0.0110]
Education level (ref = No education)
Primary	-0.00690
	[-0.293,0.279]
Secondary	-0.209
	[-0.478,0.0596]
Higher	-0.148
	[-0.418,0.122]
Income level (ref < 10,000)
10,000–20,000	0.202 **
	[0.0576,0.347]
> 20,000	0.0470
	[-0.119,0.213]
Cost of service (ref=very expensive)
Somewhat expensive	-0.151 *
	[-0.293,-0.00906]
Somewhat affordable	-0.00169
	[-0.149,0.146]
Very affordable	-0.152
	[-0.351,0.0468]
Free	-0.0936
	[-0.284,0.0972]
Facility type (ref=health center)
Public hospital	0.0547
	[-0.0739,0.183]
Private hospital	0.506 ***
	[0.356,0.656]
Private clinic	0.595 ***
	[0.462,0.728]
Constant	3.584***
	[3.256,3.911]
N	1014
R 2	0.155

Note: 95% confidence intervals in brackets; ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001

Overall, perceived quality declined with increasing age, with patients aged 30–39 and 50–59 reporting 0.108 (95% CI: -0.416 – -0.0113) and 0.213 (95% CI: -0.416 – -0.011) lower perceived quality, respectively, compared to those under 30. Income and cost of service were also associated with perceived quality. Patients earning 10,000–20,000 birr per month reported 0.202 higher perceived quality (95% CI: 0.058–0.347) compared to those earning less than 10,000 birr, while higher income above 20,000 birr showed no significant difference. Participants who rated the cost of services as “somewhat expensive” reported 0.151 lower perceived quality (95% CI: -0.293, − 0.009) compared to those who perceived the cost as “very expensive,” with no significant differences observed for other cost categories.

Type of facility visited was also a significant determinant of perceived quality. Patients who received care at private hospitals and clinics reported higher perceived quality on average by 0.506 (95% CI: 0.356 to 0.656) and 0.595 (95% CI: 0.462–0.728), respectively, compared to those who visited health centers. Public hospitals, however, showed no significant difference in perceived quality compared to health centers.

Factors associated with facility endorsement among primary care patients

An ordered logit model was used to examine the association between perceived quality of care and patients’ willingness to endorse a facility. Model diagnostics, including the Chow structural stability test (χ² = 96.7, p = 0.001) and model selection criteria (AIC and BIC), indicated that separate regressions were preferable to a pooled model. Accordingly, results are presented and interpreted separately for health centers, public hospitals, private hospitals, and private clinics.

Perceived provider competence significantly predicted the likelihood of endorsement across all models, controlling for sociodemographic variables (Table 3). Higher perceived competence was strongly associated with higher odds of endorsing the facility. Overall, a one-unit increase in perceived provider competence was associated with a 87.6% higher likelihood of choosing a higher endorsement category (OR = 1.876, 95% CI: 1.490–2.361). Conversely, the effect of structural quality on endorsement varied by facility type, showing a significant positive association only in public hospitals. The result also revealed patient-centered care influenced facility endorsement regardless of the facility type, with statistically significant effects observed in health centers, public hospitals, private hospitals, and private clinics.

Table 3.

Results from ordered logistic regression for the association between perceived quality of care and willingness to endorse a facility

Variables	(Pooled)	(Public health centers)	(Public hospitals)	(Private hospitals)	(Private clinics)
	AOR(95% CI)	AOR(95% CI)	AOR(95% CI)	AOR(95% CI)	AOR(95% CI)
Quality dimension
Structural quality	1.200	0.943	1.688 *	1.299	1.232
	[0.968,1.489]	[0.614,1.449]	[1.088,2.621]	[0.774,2.180]	[0.731,2.078]
Provider competence	1.876 ***	1.734 *	1.646 *	1.747*	2.369 ***
	[1.490,2.361]	[1.100,2.734]	[1.009,2.686]	[1.062,2.875]	[1.421,3.949]
Patient-centered care	2.008 ***	1.839 **	2.387 ***	2.670***	1.704 *
	[1.591,2.533]	[1.208,2.798]	[1.450,3.929]	[1.540,4.631]	[1.002,2.899]
Patient characteristics
Female (ref=male)	0.944	0.753	1.177	1.468	0.601
	[0.732,1.218]	[0.460,1.231]	[0.622,2.230]	[0.861,2.504]	[0.351,1.030]
Age category (ref = 18–30)
30–39	1.025	1.028	1.256	1.024	0.897
	[0.770,1.365]	[0.590,1.790]	[0.670,2.356]	[0.563,1.864]	[0.481,1.673]
40–49	1.310	1.879	2.239	1.194	0.987
	[0.875,1.961]	[0.856,4.124]	[0.885,5.663]	[0.513,2.778]	[0.428,2.273]
50–59	1.165	1.418	1.607	0.858	1.788
	[0.687,1.977]	[0.483,4.165]	[0.555,4.656]	[0.254,2.899]	[0.544,5.876]
>=60	1.080	5.224	0.491	0.743	1.516
	[0.541,2.156]	[0.903,30.21]	[0.130,1.861]	[0.125,4.412]	[0.387,5.943]
Education level (ref = no education)
Primary	1.854	4.097 *	0.133	0.995	2.409
	[0.777,4.424]	[1.155,14.53]	[0.0136,1.306]	[0.143,6.908]	[0.488,11.90]
Secondary	1.385	3.652 *	0.0884 *	1.072	1.731
	[0.595,3.227]	[1.067,12.50]	[0.00895,0.873]	[0.452,2.543]	[0.370,8.095]
Higher education	1.162	2.813	0.0939*	1	1.590
	[0.506,2.671]	[0.827,9.567]	[0.00957,0.922]	[1,1]	[0.335,7.538]
Household income (ref < 10,000 Ethiopian birr per month)
10,000–20,000	0.970	0.631	0.912	1.335	0.784
	[0.649,1.450]	[0.206,1.930]	[0.247,3.364]	[0.705,2.528]	[0.346,1.778]
> 20,000	0.999	0.850	1.011	1.945	0.544
	[0.639,1.563]	[0.112,6.435]	[0.0766,13.33]	[0.973,3.889]	[0.230,1.288]
Perceived cost of service (ref=Very expensive)
Somewhat expensive	0.980	0.124 ***	0.113 **	1.880*	1.357
	[0.661,1.453]	[0.0396,0.389]	[0.0232,0.547]	[1.022,3.457]	[0.649,2.837]
Somewhat affordable	1.444	0.240 **	0.195 *	5.980***	1.481
	[0.984,2.119]	[0.0897,0.641]	[0.0499,0.762]	[2.721,13.14]	[0.691,3.174]
Very affordable	1.184	0.199 **	0.223 *	1.935	0.896
	[0.726,1.930]	[0.0670,0.591]	[0.0537,0.927]	[0.386,9.699]	[0.301,2.665]
Free	1.575	0.175 **	0.208 *	0.985	4.548
	[0.979,2.533]	[0.0537,0.567]	[0.0528,0.819]	[0.216,4.496]	[0.588,35.16]
cut1	9.249***	1.130	0.282	44.66***	11.16*
	[2.901,29.49]	[0.145,8.812]	[0.0154,5.174]	[8.367,238.4]	[1.018,122.4]
cut2	44.14***	5.338	1.578	307.6***	40.08**
	[13.87,140.5]	[0.688,41.42]	[0.0871,28.58]	[56.04,1689.0]	[3.768,426.2]
cut3	1092.7***	119.7***	47.24*	9180.9***	2051.2***
	[320.4,3726.7]	[14.12,1014.2]	[2.484,898.5]	[1309.2,64380.1]	[157.3,26746.3]
N	1012	273	235	255	249
Pseudo R2	0.138	0.129	0.200	0.204	0.168
AIC	1940	578	460	481	442
BIC	2038	650	529	548	513

Note: Exponentiated coefficients; 95% confidence intervals in brackets: ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001

Discussion

This study assessed patient-reported quality of primary care and its association with facility endorsement across four types of health facilities in Addis Ababa, Ethiopia. Overall, perceived quality was average, with structural quality, provider competence, and patient-centered care averaging 55.9%, 58.0%, and 51.2%, respectively, indicating that a substantial portion of patients perceived care as below optimal standards. Private facilities consistently outperformed public facilities across all dimensions, reflecting differences in resources, infrastructure, and patient engagement [4, 30, 31]. These disparities are consistent with prior studies in Addis Ababa, which found higher patient satisfaction and perceived quality in private facilities due to better infrastructure, resource availability, and staffing [30]. Among public providers, health centers and hospitals were rated lower on structural and patient-centered measures, highlighting persistent gaps in infrastructure, essential equipment, and timely service delivery [12, 32–34]. Similar findings have been reported in other low- and middle-income countries, including Bangladesh and Sri Lanka, where private facilities also scored higher on patient-reported quality indicators [35, 36].

Provider competence and patient-centered care emerged as the strongest determinants of patients’ willingness to recommend a facility. Higher perceived competence—including clinical skills, communication, and care coordination—was associated with higher endorsement across all facility types. Consistent with Ethiopia’s Health Sector Transformation Plan II and the WHO framework on integrated care, people-centered health services, patient-centered care elements such as respect, privacy, convenience, and reasonable wait times also strongly influenced patients’ willingness to recommend facilities [10, 37]. Structural quality had a significant effect mainly in public hospitals, where deficits in resources and infrastructure are most pronounced. These results echo previous research showing that clinical competence and patient-centeredness are key drivers of patient satisfaction and facility trust in Ethiopia and other LMICs [19, 38]. Together, these findings suggest that continuous professional development, pre-service training, and strategies to enhance both technical and interpersonal competencies of public sector health workers are likely important for improving care quality and endorsement.

Structural quality also influenced patient perceptions, particularly in public hospitals. Public hospitals faced deficits in essential equipment, medicines, diagnostic services, and basic infrastructure, compounded by overcrowding and dual roles as both primary and referral care centers [19, 38–40]. Prior Ethiopian studies have documented similar gaps, including limited availability of essential drugs, inadequate laboratory services, and insufficient facility capacity [39, 40]. Private facilities, by contrast, benefit from better resource availability and staffing, which may contribute to higher perceived quality and endorsement. Strengthening structural capacity in public hospitals and improving routine primary care capacity in health centers and private clinics could help alleviate congestion, enhance patient experience, and potentially support more equitable access to quality care.

Patient characteristics, including age, income, and education, were associated with perceived quality. Older patients and those with lower income reported lower quality scores, consistent with previous research showing that vulnerable populations experience greater barriers to high-quality care [22, 41]. Affordability also influenced perceptions: patients who rated services as less expensive reported slightly higher perceived quality, suggesting that cost remains an important determinant of patient experience.

The positive association between perceived quality and willingness to recommend facilities has practical implications for healthcare utilization. Patient endorsement can influence healthcare-seeking behaviors, particularly in communities with low health literacy, and strengthen trust in the health system [42, 43]. This finding aligns with prior evidence that improved patient-perceived quality increases repeat visits and engagement with health services [17, 18].

Improving perceived quality across structural, clinical, and interpersonal dimensions may enhance both patient satisfaction and endorsement. Policy efforts may consider prioritizing strengthening provider competence, enhancing patient-centered practices, and addressing structural gaps in public facilities. These efforts align with Ethiopia’s Health Sector Transformation Plan II and global initiatives such as the WHO framework on integrated, people-centered health services, which emphasize competent, respectful, and equitable care [2, 10, 37, 44].

Using an Interactive Voice Response (IVR) survey allowed efficient and standardized collection of patient-reported data while minimizing social desirability bias [25, 45, 46]. IVR surveys provide a scalable and resource-efficient method for routine monitoring, enabling policymakers to identify gaps in near real-time and implement targeted quality improvement interventions [46–49].

This study has several limitations. First, the cross-sectional design limits the ability to draw causal inferences between perceived quality of care and facility endorsement. Second, all measures were based on patient self-reports, which may be influenced by individual perceptions and expectations rather than objective assessments of care quality. Third, as the study was conducted in Addis Ababa, the findings may not be generalizable to rural areas or other regions of Ethiopia with different healthcare contexts. Additionally, the use of IVR surveys may have excluded individuals with limited digital literacy and may introduce some response bias. Finally, while facility endorsement is a useful proxy for patient satisfaction and trust, it may not fully capture all dimensions of patient experience. Nevertheless, the large sample size and inclusion of multiple facility types strengthen the validity of the findings and provide valuable insights into perceived primary care quality and its association with patient confidence and service utilization in urban Ethiopia.

Conclusion

This study shows that overall perceived primary care quality in Addis Ababa was average, with substantial gaps across structural quality, provider competence, and patient-centered care. Nearly half of patient-centered care standards were perceived as unmet, and over 40% of expected standards in structural quality and provider competency fell below “good.” Private facilities generally scored higher than public ones, but the findings highlight that many patients experience insufficient care across all facility types. Provider competence and patient-centered care were the strongest drivers of patients’ willingness to recommend a facility, suggesting the importance of clinical skills, clear communication, and respectful interactions in building trust and satisfaction. Continuous training to strengthen both clinical expertise and interpersonal care is therefore important.

Structural deficiencies in public hospitals—including shortages of essential equipment, medicines, and diagnostic services—remain a key barrier to quality. Targeted investments in infrastructure, alongside expanding the capacity of health centers and clinics to provide routine primary care, could improve service delivery, reduce hospital overcrowding, and enhance patient confidence in the health system. Additionally, this study shows that Interactive Voice Response surveys provide a cost-effective method to monitor health service quality. Integrating and scaling up IVR within routine health information systems can provide timely insights on patient-centered care and guide evidence-based policy decisions, which is particularly valuable in resource-limited settings like Ethiopia. Addressing these gaps may help improve healthcare utilization, patient satisfaction, and potentially broader health outcomes.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

IVR

Interactive Voice Response

SD

Standard Deviation

OR

Odds Ratio

CI

Confidence Interval

FMoH

Federal Ministry of Health

SPA

Service Provision Assessment

DHIS2

District Health Information System, version 2

Author contributions

ADM and DS contributed to the survey design, data cleaning, analysis, and report writing. CA was also involved in survey design, data preparation and analysis, and critical review of the manuscript. All authors read and approved the final version.

Funding

This work was supported by a Research Innovation Award from the Office of Research Excellence, Milken Institute School of Public Health, The George Washington University, United States (grant 02RIA022023).

Data availability

The datasets analyzed and generated in the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study adhered to the ethical principles specified in the Declaration of Helsinki. The study received ethical approval from the Institutional Review Board of the College of Health Sciences at Addis Ababa University (Approval No. 057/23/SPH) and from the Addis Ababa City Health Bureau. Prior to conducting the survey, informed consent was obtained from all participants, and only those who consented were interviewed. All data were deidentified to protect participant privacy and maintain confidentiality. A copy of the consent form is provided as Supplemental Materials.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.