REASONS WHY PATIENTS BYPASS LOWER LEVEL PUBLIC FACILITIES: A CROSS SECTIONAL STUDY FROM UGANDA

The quality of services in Uganda at higher level health facilities are usually affected negatively by congestion when patients bypass care from their primary care health facilities (PCHF). The reason behind this bypass phenomenon in Uganda is limited. This study was conducted among patients receiving care at Lira Regional Referral Hospital in northern Uganda to identify reasons why patients bypass their PCHF. We performed a descriptive cross-sectional study between 29/12/2014 & 30/1/2015, 484 respondents attending the outpatient department were recruited by systematic random sampling. Four focus group discussions (FGDs) each involving 10 participants were conducted. Quantitative data was collected using a validated questionnaire, entered, analysed by Epidata Entry 3.1 and SPSS 18 versions respectively. Descriptive statistics and chi square test for differences in the study population were used. For qualitative data, thematic analysis of transcripts was done. Codes and categories were developed and interrogated following an iterative process based on grounded theory.

No information was available about the situation on bypass in Lira RRH. Therefore, the aim of this study was to answer the questions as to the frequency patients bypass lower public health facilities to seek care at the Lira RRH and the underlying reasons why in case this occurs.
The findings and evidence generated will inform policy making and implementation of an effective health referral system in order to reverse this situation.

Study design
We performed both a quantitative and a qualitative cross-sectional descriptive study. The quantitative part was to determine the frequency of bypassing lower level public health facilities and to describe the factors associated with this bypass .The quantitative part was complemented by a qualitative study with focus group discussions to identify the reasons for patients' bypass. The sample size of 484 for the quantitative part was determined by use of Kramer Greenhouse formula [19].

Study setting
The study was carried out in Lira RRH (LRRH) in northern Uganda about 370 km away from the Capital City Kampala. We performed a descriptive cross-sectional study between 29

Definition of Bypass
A patient was deemed to have engaged in bypass, if s/he was not living within a 5 km radius of LRRH catchment area and yet chose to seek health services directly from LRRH. More than 80% of Ugandan population lives within 5 km from a nearby health care facility. Any public health facility below the level of a RRH was regarded as lower level health facility Quantitative data collection Data was collected using a validated structured questionnaire after obtaining a written informed consent of the participants [14].
An interviewer-administered questionnaire was administered to the respondents under the guidance of a research assistant who first explained the purpose of the study and answered any question(s) raised. Three research assistants conducted the interviews. Health providers were not included in the interviews.
Interviews were conducted as soon as the respondents were registered before being seen by the health care providers available. Data was collected from adult patients 18 years and above seeking healthcare at LRRH OPD. We included all patients coming from more than a 5 km radius of LRRH catchment area and were aware of any available public health care facility within the LRRH catchment area. Patients with unsound mind and those who refused to consent were excluded.
A daily adult patients' attendance register in the OPD was used. A total of 484 adults attending the adult OPD were sampled using systematic random sampling with a sampling interval of 9. A random start on each day was got by giving 9 pieces of paper numbered 1 to 9 to a person who was not part of the study to choose. Each day every 9 th patient attending the OPD was included until the daily required number of 22 was reached. The process was repeated daily for 22 days till the sample size of 484 was obtained. Patients who participated in the study had stickers put on their files to avoid interviewing a person more than once. Data collected was coded, checked, edited for completeness and entered into Epidata 3.1 computer software, then exported to SPPS for statistical analysis.

Focus group discussion (FGDs)
FGD participants were recruited on the same day by the nurse-in-charge of the OPD with the primary aim of identifying the reasons for patients' bypass. The inclusion criteria was that the participants were adult patients attending LRRH and coming from more than 5 km radius of LRRH and had no referral notes. A total of 20 females and 20 males within the age range of 18-45 years participated in the FGDs. The FGDs was carried out in the boardroom of LRRH.
Four FGDs were held according to the age and gender in order to have groups of people with similar characteristics. The FGD participants were not part of the quantitative survey. FGDs were conducted in the local language 'Lango' with the help of a FGD guide. An open-ended question approach was used to investigate reasons for patients' bypass. Each FGD lasted about 45 minutes in the doctors' boardroom. Other information collected were socio demographic characteristics such as age, gender, occupation, marital status, level of education, distance from health facility, status of health facility functioning at night and type of illness. FGD proceedings were audio taped with a digital voicerecorder and the proceedings were transcribed verbatim. The transcripts were coded.

Independent variables
The variables included social demographic characteristics such as age, sex, gender, occupation, religion, level of education and distance from health facility, state of lower level facilities and complexity of illness.

Data analysis
Data was summarized into means, median and frequency. The results were presented in tables.
Logistic regression was performed to examine the strength of the associations between bypassing the lower health facility and potential predictors. For measuring the relationship between variables, the chi square test (χ 2 ) and the odds ratios at 95% confidence intervals were computed. Associations between categorical variables were tested using the chi-squared test with reports of the corresponding p-values set at less than or equal 0.05. Percentages were computed. The odds ratio and the corresponding 95% confidence intervals (95% CI) used to summarize the strength of association between logistic regression test. The outcome measure was the bypass of lower level public health care facility.
For qualitative data, thematic analysis of transcripts was done for emerging themes in the local language "Lango". Codes and categories were developed and interrogated following an iterative process. The findings were then translated into English and then finally written in a narrative form. The socio demographic characteristics of the 40 participants in the FGDs were similar to the quantitative survey except the age distribution of 18-45 years. Table 2 shows factors associated with patients' bypassing lower level facilities to seek care at LRRH where marital status and distance travelled of greater than 10 km were found to be statistically significant with p-value of 0.036 and 0.001 respectively. Table 3 on Logistic regression analysis of the odds of bypassing lower health public facilities was significant for those participants who attained tertiary education with p -value of 0.036.

Bypass of lower public health facilities
In the quantitative study, 387 (80.4%) participants bypassed lower public health facilities and 94(19.6%) did not bypass i.e. they were officially referred.

Quality of care at lower public health facilities
According to the qualitative interviews, the availability and quality of care were lack of trust in the health care providers was the main reason for bypassing lower facilities (30.9%). The second most common reason was lack of drugs (20.2%) followed by lack of diagnostic equipment (16.9%).

Lack of trust in the health care providers
Mistrust was a significant finding in this study with lots of implications for the person seeking care and the health care provider. Mistrust and low numbers of health care providers in the lower level public health facilities who at the same time lack empathy was mentioned by many participants in the discussion as the number one reason for their decision to bypass lower level facilities.

Lack of drugs
The other common reason for bypassing lower level facilities among FGDs participants was lack of drugs even for the common problems like Malaria: "I arrived at the health center, I was seen by the nurse and she told me the drugs are out of stocks so you can go and buy from the drug shop"(23 year old female) Even simple medicines such as 'Paracetamol' are commonly out of stock as reported by a participant who went to her nearest health facility with fever but was told to go and buy.

Discussion
According to our study 80.4% of the participants visiting the OPD of the LRRH bypassed the lower public health facilities. The main characteristics of these patients were marital status, attained tertiary education and had to travel > 10 km. Reasons to bypass the lower health care facilities were lack of trust in the health care workers, lack of drugs and lack of diagnostic equipment and tests.
Younger female participants accounted for 56.0% of the by-passers. This implied that most of the respondents were energetic and therefore able to move longer distances to seek health care services and more likely to bypass the lower level facilities. Similar results were described by Bronstein in Tanzania who found that self-referral to tertiary hospitals was common among females because of the perception of quality maternal and newborn health services offered particularly during delivery[16, 20, 21]. This finding among participants who travelled a distance of >10 km compared to those who travelled a distance between 5-10 km may be attributed to by-passers of young age who are energetic, mobile and more likely to lead to bypass. The findings were similar to a study in Pretoria,

South Africa[14].
Attainment of tertiary education compared to those with no formal education was a significant factor associated with bypass among the participants. These findings were in agreement with other study findings in Chad [13]. However, similar studies in Nigeria found that patients' educational status had no influence on whether they were referred or not as both the educated and uneducated bypassed lower levels of health care[17, 22, 23].
Another factor influencing bypassing was occupation. Most of the by-passers were peasant farmers (42%) of low socio economic class. They may not be adequately aware of the referral system and its significance as evidenced that only 19.6% having some form of referral note. The similar findings were reported in other studies[10, 13, 24-26].
Marital status was found to be a significant factor in bypassing. Married respondents were more likely to bypass compared to single respondents. This may probably be explained by the possible social and financial support that the partner may offer which subsequently facilitates bypass. Similar findings described in USA[7].
In the FGDs conducted, participants cited lack of trust and empathy as an important reason for bypassing lower public health facilities. Trust is a fundamentally important aspect of medical treatment relationships yet a vulnerable and fragile commodity. Studies have established that patient trust predicts use of preventive services, adherence and continuity of care as well as satisfaction. In health care, the absence or presence of trust in patient-provider relations can have life changing consequences. A patient who trusts a provider is more likely to seek care, to comply with treatment recommendations and to return for follow up care than a person who has little trust in a specific provider or health care system [13,18,27,28]. Mistrust was a very significant finding in this study with lots of implications for the person seeking care and the health care provider. Another reason cited for bypassing lower level facilities was lack of drugs for common problems like Malaria. This leads to disconnect in the continuity of care and patients' care becomes expensive. Nearly in all the FGDs it was clear that inadequate diagnostic equipment was a serious reason impeding the respondents from seeking care at lower level health facilities.
Bypassing and the reasons why are not only a problem in low and middle-income countries.
In addition to recent studies in Mozambique and Tanzania, Ghana and India also in Florida and Switzerland bypassing was described. The reasons why patients seeking care at higher level facilities were in all studies better quality of service, better access to advanced technology and competence of the staff, despite the distance they have to travel[9, 15, [28][29][30][31][32][33].
Bypassing lower health facilities creates congestion at higher-level facilities because many patients present with health problems that could be effectively handled at lower levels[9, 10, 13, 15, 29, 31].
Consequently, the higher level facilities performing the roles of the lower level facilities cause resource wastages, longer waiting times and unnecessary long queues and delays to access health care, deserving illnesses worsening and increased cost of health [5,6,10,20,[34][35][36].

Strength and limitations of the study
The study was carried out in a general OPD at LRRH; therefore the findings could not be generalized for the whole of Uganda. In addition, patients attending specialized units such as Ophthalmology, Orthopedic and Ear Nose and Throat were not included in the study. The participants were interviewed by nurses in uniform that could have led to bias in the responses.
LRRH comprised eight districts in the Lango sub region and received referrals from these districts. We performed quantitative and qualitative study with very limited drop out giving strength to the validity of the study for the Lira area. The findings of our study were in accordance to recent data obtained from several studies performed both in Western countries (Canada and Switzerland and USA) and African countries (Mozambique, Ghana, Kenya, Chad and Tanzania)

Conclusions
The majority of respondents seeking health care services from Lira RRH bypassed lower public health facilities. Our study indicated that utilization of services at these facilities is limited due to poor infrastructure, inadequate medicines and health supplies, shortage and low motivation of human resource. Improving the quality of care and health outcomes is incumbent upon competent and skilled    Table 3 Logistic Regression analysis of the odds of bypassing