An institutional based cross-sectional study was conducted from April to June 2019 at Jimma University Medical Center (JUMC), which is found in Jimma. Jimma town is found 352 km from southwest of Addis Ababa, the capital city of Ethiopia, providing specialized clinical services to about 15 million people in the catchment. Currently, on average 518 schizophrenic patients are having follow-up at the Psychiatry Outpatient Department (OPD) monthly. At Jimma University Medical Center, a psychiatric clinic was established in 1988 and has been serving more than 10,000 psychiatry patients annually. Currently, the clinic has 26 beds for in-patient service and 04 outpatients department with 2 psychiatrists, 10 psychiatric nurses, 2 clinical psychologists, and 9 M.Sc., 1 Ph.D., and 1 Ph.D. fellow mental health professionals serving about 15 million people in Southwest Ethiopia.

Source population: all caregivers of patients with schizophrenia, visiting the Psychiatry Outpatient Department at Jimma University Medical Center.

The sample size was determined using the single population proportion formula by taking the result done in Nigeria; the result of high expressed emotions of caregiver was 50.0%. To get the possible sample at 95% CI that is Z–the value of 1.96 and marginal error of 5% is calculated as follows:

n = (Z α/2)2p (1−p)/d2, where n = number of sample size. Z = desired 95% confidence, Z = 1.96. p = population proportion. q = 1− p = 1− 0.5 = 0.5, d = is the margin of sampling error tolerated (5%).

$$n = \frac \right)^ \left( \right)\left( \right)}} \right)^ }},n\;}\, = \,}$$

By considering 10% (10/100*384 = 38) non-response rate, the final sample size was 422.

Study participants were recruited by consecutive sampling techniques.

Caregivers who were  ≥ 18 years of age and were taking care of patients with schizophrenia were included.

A structured questionnaire was developed after reviewing related literature. It was used to collect data about caregivers and patient socio-demographic variables. The last psychiatric diagnosis was taken from medical records of patients. The caregivers Expressed Emotion status was measured by Family Questionnaire (FQ) which is developed by Wiedemann, Rayki, Feinstein, and Hahlweg in 2002, with 20—items which included two domains—Critical Comments CC (10 items—2, 4, 6, 8, 10, 12, 14, 16, 18, 20), and Emotional Over Involvement [EOI] (10 items—1, 3, 5, 7, 9, 11, 13, 15, 17, 19), with maximum value of 40 and the cut-off point for high EE relatives being  > 23 for critical comments and  > 27 for emotional over-involvement [4, 15].

Low EE relatives are those with cutoff point of  ≤ 23 for critical comments and  ≤ 27 for emotional over-involvement. Possible responses of participants is never or very rarely, rarely, frequently, and very frequently, with values ranging from one to four for each response, respectively. The FQ had better agreement with the gold standard questioners of CFI (Camberwell Family Interview) on CC and EOI than did other short EE questionnaires, because both have sensitivity 80%, specificity 70%. Criticism (α = 0.86, n = 257) and emotional over-involvement (α = 0.80, n = 256) subscales showed strong internal consistency (22,9, (18) 33,19,34).

The dependent variable of the study was Expressed Emotion (EE) of caregivers (Yes/No).

Independent variables included caregivers socio-demographic factors (age, gender, ethnicity, occupation, marital status, educational status, family size, relationship with the patient), distance from hospital in km, and average household monthly income,

Socio-demographic factors of patients: age, gender, marital status, educational status, employment status, the impact of illness on occupation.

Clinical variables: duration of illness, duration of taking care of the patient, time spent with the patient per day, number of the episode, number of previous hospitalizations, number of a family member with psychiatric illnesses, and co-morbid disorder of the patients.

High expressed emotion: high EE relatives are with cutoff point of  > 23 for critical comments and  > 27 for emotional over-involvement (22,9,33,19,34).

Low expressed emotion: low EE relatives are with cutoff point of  ≤ 23 of critical comments and  ≤ 27 for emotional over-involvement (22,9,33,19,34).

The data were checked for consistency and completeness throughout the time of data collection. Data coded and entered twice into double EPI-DATA version 4.41 and then exported to SPSS version 25 for analysis. Before performing binary regression the scores were checked for assumption and whether the model fits or not via Hosmer–Lemeshow. Bivariate regression was computed for each independent variables separately from the dependent variables. Finally, those variables having a p < 0.25 were taken to multiple logistic regressions model once, and those with a p < 0.05 were considered as having statistically significant association with the dependent variable.

Ethical clearance was obtained from the Institutional Review Board of JU after approval of the proposal. Official permission was collected from Jimma University Medical Center’s psychiatry clinic. The purpose of the study was communicated with study participants and data were collected after written consent is obtained. Caregivers with HEE were consulted with mental health professionals and psychologists working in the unit.