Research Design
The present descriptive analytical cross-sectional study was conducted in Golestan province in northern Iran in the calendar year started on 21 March 2019. The population under study included all patients who were referred by a family physician at level 1 to a specialist at level 2 within the framework of electronic health referral system, but they did not visit the designated specialist to receive services. The study was conducted in the cities of Bandar-e-Turkman, AqQala and Aliabad-e-Katoul, located in the west of Golestan province in northern Iran,wherein the electronic health referral system for outpatient services in these three cities was launched as a national pilot plan.
Recruitment
Of the 15,053 individuals who had been referred to the second level of services in the first six months of 1397 AH (21 March 2018 -22 September 2018), but had not followed up the process after referral; 5,103,5,895 and 4,355 persons were from Bandar-e-Turkman, AqQalaand Ali Abadaktol, respectively. By using Morgan's table and taking into account the 10% drop in the samples, 431 people were calculated as sample size. The samples were selected by stratified random sampling method with allocation proportionate to the number of referred patients. Therefore, the number of samples considered for Bandar-e-Turkman city, AqQala and Aliabad was 146, 169,116, respectively.
Data Collection
Records of the patients referred to hospitals in the cities of Bandar-e-Turkman city, AqQala and Aliabad were first reviewed. It came out that some patients who had been referred by a family physician in rural health centers to a specialist in the district hospital in the past month, had not visited the appointed specialist to receive outpatient services. Since the receipt of appointment from the family physician, i.e. in maximum 30 days, the referred patients could visit a specialist. The sampling was done in two phases: 1) proportionate stratified sampling method, and 2) systematic random sampling. By using a proportionate stratified sampling method in the first phase, 429 patients were selected. These patients were selected according to the referral code registered in the health information software system. In this way, from the list of patients' names, first the starting point for this regular process was selected and then a suitable interval was selected and the names were selected at equal intervals in the list.
A self-made questionnaire was used to collect the data. To design the questionnaire, first of all the questions were extracted from scientific sources and previously-conducted studies [20-24] and the views of the scientific experts and executives were taken into account. The questionnaire focused on five dimensions and consisted of 34 items, of which 11 items dealt with the designated specialist, 6 items dealt with the admission and queuing system, 10 items related to the clinic conditions, 2 items related to the recommendation and suggestions of others, and 5 items related to the conditions and side expenditures. The responses were scored as a five point Likert scale: "I Strongly disagree" with a score of 1, "I disagree" with a score of 2, "I have no idea" with a score of 3, "I agree" with a score of 4 and "I strongly agree" with a score of 5. At the end of the questionnaire, in an open-ended question, patients were asked to state other reasons for not visiting a specialist, if there were any.
Qualitative and quantitative methods were used to determine the content validity. To assess the qualitative validity, 15 experts and elites in the field of the subject were asked to review the questions and to apply their corrective opinion on grammar and sentence structure; and put phrases in appropriate order. To quantitatively validate content validity, Content Validity Ratio (CVR) was used, and a team of experts was asked to review each question based on a three-part spectrum of "essential, useful but not essential, not necessary." Since the acceptable range depends on the number of experts, based on the judgment of 11 experts in this study, the number was considered to be 0.59, which was finally confirmed based on the answers of the content validity for all questions. Cronbach's alpha coefficient was also used to determine the internal consistency and reliability of the questionnaire. Given the total sample size of the study required for calculating Cronbach's alpha (more than 10% of the total sample size), the minimum sample size was 50. In this study, the overall value of Cronbach's alpha coefficient was calculated to be 0.9 and for the dimensions of the questionnaire were as the following:the designated specialist (a = 0.81), the admission and queuing system(a = 0.74), the clinic conditions(a = 0.80), the recommendation and suggestions of others(a = 0.70), the conditions and side expenditures (a = 0.75); which indicates the high internal consistency of the questionnaire items. The questionnaire was filled in through telephone conversations and by a trained interviewer, and the information on 429 people was collected completely (response rate of 99%).
Ethics
The research project was approved by the Research Committee on Ethics in Golestan University of Medical Sciences (IR.GOUMS.REC.1398.048). At the time of data collection, respondents were assured that their information would remain confidential; nevertheless, the questionnaire was filled in anonymously.
Data Analysis
Descriptive statistics methods (tables, frequency, percentage for qualitative data and mean, standard deviation (SD) for quantitative data) were used and given the abnormality of data distribution according to Clemogrov Smirnov's test (p < 0.001),Mann-Whitney and Kruskal–Wallis tests were used to analyze the data. Data were analyzed using SPSS 16. The significance level of all tests was considered to be 0.05.