Mental health is one of the vital components of health [1,2]. There is evidence that mental and physical illnesses may accompany, follow, or precede one another. There is also evidence which indicated that mental disorders increase the risk of physical illness and vice versa [3]. Persons with mental illness receive a wide range of responses across cultures. In the developing world, they are subjected to severe stigma and mistreatment, but in some cases are helped by community support structures. Traditional beliefs that attribute psychiatric disorders to moral transgression and misconstrue the dangerousness of patients lead to feelings of shame and fear of persons with mental illness. Such community values and beliefs influence treatment seeking behavior, treatment outcomes, and even determine the way mental health is practiced [4,5]. Beliefs regarding the causes of mental illnesses hover between the natural and the supernatural. They vary according to an individual's level of education and socioeconomic class. In less educated areas of the countryside, there exist a number of supernatural explanations of mental illness which include possession by spirit, black magic, or astrological misalignment [4,6,7]. Over 450 million people are estimated to be suffering from mental disorders in the world today. Only a small proportion of these people receive any form of modern treatment, and most untreated cases are found in low income countries [8]. In Ethiopia, for example, less than 10% of persons with severe mental illness had contact with modern psychiatric services. Fifteen to twenty percent of people who attend general medical clinics do so because of mental disorders, although their mental health problems are often not recognized [9,10]. Modern psychiatric services are very scarce, inaccessible, and relatively expensive for the majority of the population in Ethiopia. Therefore, patients usually resort to modern mental health-care services only after they have failed to recover after receiving traditional treatments. It is also a common practice in Ethiopia for family members to care for and support persons with mental illness at home. A study also showed that less than half of mental illness patients directly contacted a mental hospital, and the median delay between onset of illness and arrival at the psychiatric hospital was 38 weeks [11]. In pluralistic medical settings, laypeople choose what to do first, second, third, and fourth from a variety of treatment options [11,12]. A systematic analysis of the sequence of treatment options sought provides insights into patients' patterns of resort and suggests a tentative theory for how laypeople make medical choices. The strength of pathway models is that they depict health seeking as a dynamic process. Factors are sequentially organized, according to the different key steps (i.e. recognition of symptoms, decision making, medical encounter, evaluation of outcomes, and re-interpretation of illness) which determine the course of the therapy path [13,14]. Hence this study has investigated the patterns of health seeking behavior for mental illnesses in JUSH.

We conducted a quantitative, hospital-based cross sectional study to explore the patterns of health seeking behavior and related factors for mental illness from March to April, 2010, at JUSH, Ethiopia. The hospital is located in the city of Jimma, a town in southwestern Ethiopia 345 km from the capital city of Addis Ababa. The psychiatric facility at the hospital is staffed by one psychiatrist and three psychiatric nurses, and it has 26 inpatient beds. We approached 384 consecutive new patients attending the psychiatric facility of JUSH during the study period. The sample size was determined with a single population proportion formula by assuming that 50% of the patients will come early for mental illness treatment at a psychiatric facility (to obtain maximum sample size) with 95% confidence interval. Consecutive patients attending outpatient department of psychiatry with a new episode of illness during the study period were included in the study. Caregivers were interviewed whenever the person with mental illness could not respond due to the illness, had expression and/or hearing problems, or whenever the patient was younger than eighteen years. Patients were enrolled until the required sample size was obtained. Registration records were reviewed each day to select study subjects who were eligible for the study. Data was collected using a pretested questionnaire which was administered using the face-to-face interview method. Information regarding psychiatric diagnoses of participants was obtained from their medical card. Data was collected by trained psychiatric nurses who were fluent in Afaan Oromo and Amharic languages (local languages). The questionnaire was adopted from World health organization encounter form for Pathways to care [15] and items that assess the perception of patients on mental illness was developed from the Good's pathway model [14]. For the purpose of this study, a person with mental illness was defined as any patient who received any psychiatric diagnosis after being evaluated by mental health professionals. Mental health remedies sought for mental illness were categorized as religious (rituals/practices, herbalists and other traditional healings), biomedical (government and private 'modern' health institutions), and self care (home remedies). Perceived causes of mental illnesses were assessed based on the perception of the respondents and were categorized as a "traditional explanation" (i.e. spirit possession) or "modern explanation" (i.e. pathogens) Perceived susceptibility to mental illness was assessed by asking the open question 'who do you think mental illness affects?' Perceived severity was measured by asking 'how do you rate the severity of mental illnesses generally?' with possible responses being 'highly severe,' 'moderately severe,' 'less severe,' and 'not at all severe'. To assess the belief of remedies for mental illness, subjects were asked, 'can mental illnesses be cured?' To categorize significant delay for seeking mental health treatment, we took 38 weeks as a reference point from a study conducted at Amanuel mental specialized hospital in Addis Ababa, Ethiopia [11]. Hence we categorized individuals who sought mental health treatment from the psychiatric hospital above 38 weeks as delayed and below 38 weeks as not delayed. Data was checked for completeness and consistency and analyzed using SPSS16 statistical software. Descriptive statistics were presented using summary tables and graphs. Cross-tabulations and multivariate logistic regression was done to identify the most important predictor variables of mental health treatment seeking behavior. Ethical approval was obtained from ethical review board of Jimma University. Written informed consent was obtained from the participants. For literate people, they themselves read the already prepared consent. But for those who could not read, the data collectors read for them and obtained their signature or finger print to affirm their consent.

Most of the respondents came to treatment after significant delay. Only 35% of the patients with mental illness came directly to modern psychiatric treatment. There was a paradox between their belief of the causes of mental illnesses and the type of treatment sector, as a large proportion of subjects felt that mental illness was caused by supernatural means but was curable by biomedical treatment. Individuals who had the sign and symptoms of abdominal pain and headache were more likely to seek care early. Being within the age of 31-40 years was associated with seeking psychiatric help much earlier than other age groups. Interventions targeted at improving public awareness about the causes and treatment of mental illness could reduce delay in treatment and thus improve treatment outcomes.

The authors declare that they have no competing interests.

We would like to forward our gratitude to Jimma University for funding the study. We thank JUSH psychiatry clinic, the data collectors, supervisors and the respondents for their contribution. We would like to thank Joseph Lippi for editing this manuscript.