We developed a tool measuring cultural differences in healthcare perceived by foreign patients visiting South Korean hospitals. The study was divided into two phases: (1) tool development of perceived cultural differences in healthcare by foreign patients; and (2) measurement of perceived cultural differences in healthcare by foreign patients.
The first phase involved defining various domains of cultural differences in healthcare perceived by foreign patients; generating the items comprising the tool; evaluating content validity; performing two rounds of stakeholder feedback via face-to-face interviews and a pilot study based on the “Core Outcome Measures in Effectiveness Trials” (COMET) handbook, version 1.0 . In the second phase, we measured healthcare cultural differences perceived by foreign patients, and evaluated reliability and validity.
Phase 1: Development of a tool measuring cultural differences in healthcare perceived by foreign patients
Defining the scope of perceived cultural differences in healthcare
We first reviewed the literature to define perceived cultural differences in healthcare. Leininger and McFarland  stated that: “Culture is the values, beliefs, norms, and practices of a particular group that are learned and shared and that guide thinking, decision, and actions in a patterned way.” Based on this definition, cultural differences is defined as differences in group values, beliefs, norms, and practices that are learned and shared, and that guide thinking, decisions, and actions. As we were concerned with cultural differences in healthcare perceived by foreign patients, we designed a tool by which patients can compare their own culture with South Korean culture in a set of domains reflecting various aspects of healthcare.
Before generating the items of the tool, we identified the domains of cultural differences relevant to healthcare from literature review. Communication, food, and religion domains were extracted from the six phenomena comprising the Transcultural Assessment Model of Giger and Davidhizar  and the 12 cultural domains of the Purnell Model for Cultural Competence . The six phenomena of the Giger and Davidhizar include communication, space, social orientation, time, environmental control, and biological variation. The 12 domains of the Purnell model include overview/heritage, communication, family roles and organization, workforce issues, biocultural ecology, high-risk behavior, nutrition, pregnancy, death rituals, spirituality, healthcare practices, and healthcare providers. The communication domain was included in both models. The religion domain was included as the social orientation domain of Giger and Davidhizar and as the spirituality domain of the Purnell model. The food domain was included as the biological variation phenomenon of Giger and Davidhizar and as the nutrition domain of the Purnell model. The communication domain includes both verbal and non-verbal communication; the food domain includes the quality of food provided and the extent to which staff understand the patient’s food culture; the religion domain includes the available religious facilities and the extent to which staff understand the patient’s religion.
The following four domains were extracted from Flores  and Lynn and Deanna : healthcare facility, health beliefs, patient-caregiver relationship, and healthcare system. The healthcare facility domain includes environmental features, such as the layout of the hospital room; the health beliefs domain includes traditional beliefs shaped by specific cultural beliefs; the patient-caregiver relationship domain includes interactions between patients and healthcare providers; and the healthcare system domain includes the mode of referral and the number of nurses per patient.
Determining what to measure
We generated items based on the literature review reflecting the values, beliefs, norms, and practices of particular groups that are learned and shared, and guide thinking, decisions, and actions. The items for the food and patient-caregiver relationship domains were developed by reference to two earlier tools: the Customer Satisfaction Survey for Nutrition and Food Service and the Patient-Doctor Depth of Relationship scale [26, 27]. Items for the health beliefs domain were developed by reviewing the literature on health beliefs by cultural background . Items for the communication domain were developed by reviewing the literature on language barriers in healthcare [29, 30]. Items for other domains, including the healthcare system, the healthcare facility, and religion were developed by reference to Giger  and Purnell . In total, seven cultural domains comprising 37 items were developed: 3 items on religion, 6 items on communication, 4 items on the healthcare facility, 7 items on food, 4 items on health beliefs, 5 items on the patient-caregiver relationship, and 8 items on the healthcare system.
Stakeholder feedback round 1: Face-to-face interviews
Face-to-face interviews were conducted with four medical coordinators and six patients to validate the seven identified cultural domains and generate additional items. A convenience sample of four senior medical coordinators working at one of the study hospitals were recruited. They were a Chinese-speaking coordinator from South Korea, a Mongolian-speaking coordinator from Mongolia, a Russian-speaking coordinator from Kyrgyzstan, and an Arabic-speaking coordinator from South Korea. All had worked for more than 3 years as medical coordinators in one of our study hospitals. Six patients were recruited by the snowball technique via the English, Chinese, and Arabic language coordinators. The six patients included two Arabic-speaking patients from the United Arab Emirates (UAE), two English-speaking patients from the USA, and two Chinese-speaking patients from China. These six patients represent the English, Chinese, and Arabic language groups which native medical coordinator were not represented. The first author explained the purpose of the study and conducted semi-structured interviews with four medical coordinators and six patients, lasting 20–40 min. They were asked to review the seven identified cultural domains and state whether they had experienced any other cultural differences in healthcare between South Korea and their home countries.
Four bilingual professional translators translated the English tool into Arabic, Russian, Chinese, and Mongolian. Another four bilingual translators performed back-translation into English. Eight translators reviewed the four translated and back-translated tools. We chose these five languages because most foreign patients treated in South Korea came from China, Russia, the USA, Kazakhstan, the UAE, and Mongolia. According to Jeanrie and Bertrand , and Peña , linguistic and cultural equivalence were assured by back-translation involving two native translators bilingual in English and one of the four other languages who understand the different cultural backgrounds.
Content validity was evaluated by two physicians and eight nurses working in the international healthcare department at one of our study hospitals. The first author emailed 10 experts explaining the purpose of the study. All subjects were not involved either in validating the seven identified cultural domains or generating the items for the tool. The first author distributed questionnaires to those who agreed to participate. They were asked to rate the relevance of the 41 items of the tool to the seven domains of cultural differences using the 4-point scale: 1 = not relevant, 2 = somewhat relevant, 3 = quite relevant, and 4 = highly relevant. Completed questionnaires were placed in a dropbox located in the international healthcare department where the respondents worked.
Stakeholder feedback round 2: Pilot study
We performed a pilot study to evaluate whether all items were readily understood, and the time required to complete the questionnaire. We selected 20 foreign patients using a convenience sampling who visited one of the study hospitals. The first author explained the purpose of the pilot study to five medical coordinators working in one of our study hospital. These coordinators recruited six English-, five Arabic-, three Chinese-, three Russian-, and three Mongolian-speaking patients who visited the study hospital from May 29, 2016 to June 2, 2016. The first version of the questionnaires was distributed to the 20 foreign patients who agreed to participate. The medical coordinators recorded the time required to complete the questionnaire and asked the respondents how well they understood each item.
Phase 2: Measurement of cultural differences in healthcare perceived by foreign patients
We enrolled 256 foreign patients who visited three tertiary hospitals in Seoul, South Korea. The enrolment criteria were age over 19 years; the ability to read and understand the questionnaire; the ability to communicate in English, Arabic, Mongolian, Chinese, or Russian; an understanding of the purpose of the study; and agreement to participate. The minimum sample size required to assess validity and reliability was calculated to be 205, corresponding to five times the total number of items .
Data were collected over 3 months from August to October 2016 in three hospitals in Seoul that agreed to participate following institutional review board approval (IRB no. 1606–121-772). As the respondents were not South Koreans, the first author introduced the research to 15 medical coordinators (1 for each language at each hospital). The first author and the medical coordinators explained the purpose of the study to prospective respondents as they were leaving the hospital, and distributed questionnaires to those who agreed to participate. We asked the respondents to place completed questionnaires in dropboxes located in the international healthcare departments of the study hospitals. In total, 260 questionnaires were returned, of which 256 had been completed.
We evaluated the validity of the tool by measuring structural validity and testing hypotheses, and the reliability of the tool by measuring internal consistency. Based on the recommendation by Gorsuch , we used exploratory factor analysis (EFA) to explore structural validity because the tool was not based on a model or theory related to cultural differences in healthcare. Validity was further evaluated by testing two hypotheses. Foreign patients visiting South Korea for medical treatment will almost certainly perceive cultural differences. One useful validation procedure involves determining whether cultural differences perceived by foreign patients (compared to their native countries) differ significantly from 0, which corresponds to no difference. Hypothesis 1 was that the cultural difference in healthcare perceived by foreign patients is not equal to zero. Kramsch and Widdowson  state that culture is expressed, embodied, and symbolized by language. Therefore, the cultural differences in healthcare perceived by foreign patients will vary by their language. We thus also explored whether the cultural difference in healthcare perceived by foreign patients visiting South Korea differed by language. Hypothesis 2 was that cultural differences in healthcare perceived by foreign patients would differ by language group.
Data were analyzed using SPSS software (ver. 21.0; SPSS Inc., Chicago, IL, USA). We performed a descriptive analysis to identify general characteristics of the respondents. We evaluated the structural validity and internal consistency of all items with mean, standard deviation, skewness, kurtosis, and corrected item-total correlation values. Structural validity was analyzed via EFA using varimax rotation and Kaiser normalization. We considered that structural validity was evident when the Kaiser–Meyer–Olkin (KMO) parameter was ≥0.80 and the Bartlett test of sphericity yielded a p-value < 0.05. We determined the number of factors to retain using eigenvalue-greater-than-1 rule, and eliminated items exhibiting factor loadings of < 0.40 or > 0.95. We labeled all extracted factors based on the cultural domains into which they fell during development, and on literature reviews on healthcare factors influencing patient satisfaction [37, 38].
Hypothesis 1 (that the cultural difference in healthcare perceived by foreign patients is not equal to zero) was tested using the one-sample t-test. Hypothesis 2 (that cultural differences in healthcare perceived by foreign patients would differ by language group) was tested employing analysis of variance (ANOVA). We converted the mean scores for all factors to percentages, and compared these across all language groups. We used Cronbach’s alpha as a measure of internal consistency.