Defining modifiable barriers to uptake of dietary recommendations in Chinese immigrants with type 2 diabetes: a qualitative study

The study was approved by the University of Alberta Health Research Ethics Board (Pro00023449). Participants were recruited through advertisements in hospitals, Chinese supermarkets, cultural centres, and newspapers in Edmonton, Alberta. Prospective participants were screened against the inclusion criteria via telephone: Chinese descent, ≥35 years of age, diagnosed with T2D or prediabetes, living in North America ≥1 year, read and write English or Chinese. Those meeting the criteria (n = 14) provided written informed consent. Consented participants received a $25 gift card for a local grocery store. For the purposes of this analysis, persons with prediabetes (n = 1) were omitted because of potential confounding by differences in treatment intensity compared with persons with diabetes, thus responses from n = 13 participants were included.

A qualitative study employing one-on-one interviews was conducted. The bilingual (Mandarin, English) study coordinator held two meetings at the University of Alberta with each participant. At the first meeting, hemoglobin A1C (a marker of blood glucose control) and anthropometric measures were taken (Asaad et al. 2016). Demographic information was collected through a questionnaire. Perceived dietary adherence was evaluated by a questionnaire asking how often in the previous week (days out of 7 d) people had followed each of nine recommendations from Diabetes Canada Nutrition Therapy Guidelines, with a maximum score of 63 for high adherence (Asaad et al. 2015). The actual dietary adherence was reflected by participants’ Healthy Eating Index (HEI) scores. The HEI scoring criteria were adapted from the American Healthy Eating Index scoring criteria, and followed the recommendations from Canada’s Food Guide (Garriguet 2009). The total HEI score was calculated and could range from 0 (low diet quality) to 100 (high diet quality). Physical activity was assessed by the Godin-Shephard Leisure-Time Physical Activity Questionnaire, which acquired the frequency of strenuous, moderate, and mild physical activities. Scores for total physical activity were calculated (Godin 2011). Acculturation was assessed by the Suinn-Lew Asian Self-identity Acculturation Scale, which is a validated scale for the assessment of acculturation among Asian–Americans (Suinn et al. 1992). The adaptation was that the word Asian was replaced with Chinese, so that it was more specific to Chinese participants. It consisted of 21 questions inquiring about various aspects of culture including food preferences. A score was obtained through adding up the answers for all 21 questions and then dividing the total value by 21. Scores for individual items could range from 1 (low acculturation) to 5 (high acculturation). All questionnaires were available in both English and Chinese with accuracy verified by back-translation to English. Descriptive data were expressed as means ± SD or proportions.

During the second meeting, a one-on-one, 40-minute, semi-structured interview was conducted, focusing on facilitators and barriers to following current nutritional guidelines, health beliefs, factors influencing food choices, and opinions on the development of a culturally relevant menu plan for Chinese diabetes patients. The questions comprising the interview (Supplementary Material S1) proceeded in English (n = 3) or Mandarin (n = 11) by one researcher/author (FD) and were recorded using a digital voice recorder.

An audit trail protocol was developed to ensure transferability, i.e., the extent to which the findings from this study could be generalized to similar situations (Merriam 2002). Interviews were transcribed verbatim by the same researcher (FD). The protocol included the researcher (FD) reading the transcripts multiple times to ensure a comprehensive understanding prior to coding. Transcriptions were coded manually in Microsoft Word using an iterative approach where the researcher went back repeatedly to add or modify the codes until the information was properly captured. A research assistant (AZ) who was also fluent in Mandarin double-checked the codes against the transcripts. The researcher summarized the interview and confirmed the main points using a process called member checking to allow the participant to add anything missed or correct any misinterpretation. The preliminary analysis including codes was communicated back to the participants via email or phone call to ask for their views and evaluation of veracity. The goal of this step was to reduce bias introduced by the researcher (Creswell 2007) and increase credibility of the findings (Marshall and Rossman 2006). After member checking, codes generated from all transcripts were translated into English, pooled and organized in categories aligned with the research questions. Themes were generated from each category to answer the research questions, which was done in a similar iterative way as the coding process (FD); this was verified (CBC) with discrepancies discussed and resolved.

Table 1 summarizes participant characteristics. Average income fell below the low-income cut-off for a census metropolitan area (Income Statistics Division 2015). The majority had normal weight, none was obese, but 64.3% had central obesity. Acculturation was low (<2 out of 5). Diet quality measured by the Healthy Eating Index (Garriguet 2009) (mean = 66 out of 100, categorized as “acceptable”) is similar (mean = 69) to a mainly Caucasian cohort (Asaad et al. 2016). Adherence to 2013 Diabetes Canada Nutrition Therapy Guidelines measured by the Perceived Dietary Adherence Questionnaire (mean = 40 out of 63) was somewhat higher than two mainly Caucasian groups (both mean = 32) (Asaad et al. 2016; Raj et al. 2018). The leisure-time physical activity score of 25 is interpreted as “active” on the Godin Scale (Godin 2011).

The objective was to explore factors that influenced food choice, diet quality, and barriers to acquiring nutrition knowledge to identify specific problems as the first step in quality intervention development (Wight et al. 2016), i.e., defining and understanding the problem and its causes, identifying which causal or contextual factors are modifiable, and deciding on the mechanisms of change. This objective was addressed by analyzing participant responses to questions based on general concepts presented in the Conceptual Model of Diabetes Self-management among Chinese Immigrants (Zeng et al. 2014). The model is similar to more generalized discussions of social factors that have an impact on diabetes self-management, including demographic, social, psychological, and economic factors in immigrants (Gonzalez-Zacarias et al. 2016). The discussion points and quotations upon which the interpretations are based are summarized in Table 2 and categorized as barriers or facilitators to dietary adherence, also illustrated in Fig. 1. Individual participants are identified by a study code for Dietary needs Assessment of Chinese with type 2 diabetes [DAC#].

A main barrier or problem identified was a lack of culturally relevant, practical information for Chinese immigrants with diabetes, thereby lowering the utility of current dietary recommendations. This aligned with cultural characteristics in the conceptual model (Zeng et al. 2014) but also touched on social support because the respondents addressed who they received information from, including health care professionals and community as well as family and friends. When immigrants categorize their patient–provider relationship as poor, they are 16% less likely to participate in self-management practices (Hyman et al. 2017). Outside of the formal healthcare system [which was cited by participant DAC08], sources of dietary information included the Chinese Community Center [DAC13, DAC09], family and friends [DAC01], and newspapers and the internet [DAC05, DAC07]. Activities at the Chinese community centre including seminars given by a Chinese dietician or by non-Chinese health professionals (translated by a bilingual staff member) were the only sources of culturally relevant dietary information for many participants. Some participants were unaware of existing resources such as the Diabetes Canada website, where there is diabetes management information written in Chinese. Eating Well with Canada’s Food Guide (CFG) in Chinese was the main resource cited but only used by a few. Participants with low computer skills noted they could not access it easily from the Diabetes Canada website. Some participants had only received general recommendations from their family doctors, such as to reduce sugar and fat intake and to lose weight without further guidance [DAC13], and one took action based on “common sense” [DAC12]. Another helpful recommendation was about portion control [DAC04] but others found portion information confusing [DAC13].

Language was a barrier, especially among the elderly who tended to have lower English-language proficiency. The participants who mentioned language ranged in age from 68 to 75 years. Only one participant >70 years of age did not mention language. Moreover, for these participants, the language proficiency scale within the acculturation instrument averaged 1.33 versus 2.00 for the whole cohort. One participant did not attend the diabetes education program because she could not speak English and her English-speaking daughter was unavailable to accompany her. Another unilingual participant did not read any dietary information because it was mostly in English. These findings are consistent with the conceptual framework, which identified linguistic barriers as a main contributing factor to diabetes self-management uptake (Zeng et al. 2014). Low language proficiency has been identified as an independent risk factor for poor glycemic control. They also demonstrate that age (a socio-demographic factor identified in the conceptual model) is intertwined with factors such as literacy and overall acculturation. Low literacy may also include low health literacy as reported in a study of Irish immigrants (Thabit et al. 2009).

Participants repeatedly indicated that the current dietary recommendations lacked consideration of Chinese culture and the dietary habits of Chinese immigrants. Some participants indicated that current recommendations were too general and lacked detailed instructions on meal planning whereas others said that CFG was too complicated and confusing, making it too much trouble to follow. Moreover, the literally translated CFG did not take into account the cultural background and dietary patterns of Chinese immigrants [DAC04]. As a result, many participants thought the information was not helpful and became frustrated when they had to give up some traditional foods. However, rather than a disadvantage of the content of recommendations, this seems more related to their delivery. Misunderstanding of the recommendations could be due to the literacy level of the client (Anderson et al. 1995) as well as how the education was delivered. The conceptual framework cited traditional Chinese health beliefs, such as yin–yang balance and traditional Chinese medicine in immigrant populations but participants in our interviews rarely mentioned these as drivers of health behaviours, as evidenced by a lack of quotations (Table 2). Those who did comment said they paid attention to dietary balance in general but did not seem to consider this as one of the determining factors in their food choices, and others said they did not put much thought into yin–yang balance.

Personal taste preference was an important factor influencing food choices among many participants [DAC04, DAC08]. Taste preferences develop early in an individual’s life and thus the cultural background and traditional cuisine are the best predictors (Contento 2011). Participants were torn between giving up familiar foods or continuing to eat them while worrying about their effects on glycemic control and fear of complications. Eating out, such as eating at parties and family get-togethers and eating during traveling was another consistently mentioned barrier [DAC10]. Overeating and eating unhealthy food were common when eating out. This is consistent with previous literature on Chinese immigrants as well as other populations (Lv and Cason 2004; Cao et al. 2014; Kwon and Ju 2014; Moreira et al. 2015). Family members could be supportive in this regard, reminding participants to be careful of their diet [DAC05] or preparing healthful meals [DAC02, DAC08], illustrating the importance of social support.

Food price was another important factor as explained by participant [DAC09]; most participants in this study were low income, a sociodemographic factor in the conceptual model (Zeng et al. 2014). Financial stress is a strong predictor of poor diabetes self-management practices in recent immigrants from Asian and South Asian regions (Hyman et al. 2017). Even for one participant with higher income, price was an important concern because he was used to being thrifty [DAC14]. Nutrition and health concerns also played a role in determining food choices; participants suggested that traditional Chinese recipes be modified to contain less fat and salt so that they could enjoy traditional food while being healthy [DAC08, DAC13].

Other barriers were common across cultural boundaries. Some participants, especially those who were working or had family to care for, lacked time to spend on translating guidelines into practice. One participant [DAC02] was hoping for the invention of diabetic foods that could be easily stored and reheated to save time. Another [DAC05] prioritized other activities more highly than following dietary recommendations. Other barriers included lack of willpower and adequate dietary knowledge and cravings. Additional factors mentioned by participants included food quality and variety and convenience related to food shopping.

From the common barriers identified by participants, a problem list was developed as summarized in Table 3. From this, modifiable factors were extracted that could be summarized as providing culturally relevant resources for both healthcare professionals and Chinese immigrants with diabetes that would target specific nutritional issues while addressing food acceptability, which includes the types, taste, and price of foods.

Mechanisms to address barriers to consuming recommended diets for diabetes in other populations have included such resources as menu plans, grocery lists, and recipes (Cunningham et al. 2006; Bader et al. 2013; Soria-Contreras et al. 2014). Such strategies could illustrate nutritionally appropriate and culturally relevant foods, provide healthier versions of traditional recipes, and might also provide support for food procurement and time management for food preparation and budgeting (Ross and Geil 2010).

Like in other populations (Schlundt et al. 1994; Marcy et al. 2011; Faramand et al. 2015), cravings and time constraints were also mentioned by the participants as barriers to adhering to recommendations. Providing patients with tips on how to eat healthier meals away from home could be helpful.

Given our research group’s past experience with menu planning as part of nutrition education for people with T2D (Asaad et al. 2016), we queried whether a Chinese menu plan would benefit participants. Participants showed support and excitement for this idea: “If you could provide Chinese immigrants with a menu plan, I’m in urgent need. I need it very much! I hope you guys can provide this kind of information, guiding us on meal planning” [DAC12] and “…I hope to see the Chinese menu plan. Especially for Chinese people, it will be very useful, because most of us don’t eat western food very often. Of course sometimes we do. Like, when we go to banquets, we would also eat western food…” [DAC14]. When asked about the format, participants tended to prefer a more flexible rather than a fixed menu plan: “…The (nutrition) breakdown in details. Once I have that in details, then I can follow…Like, I can incorporate the information I have and then can prepare it with my own menu, right? I really need to know, one cup of cooked white rice, jasmine (rice) for example, one cup of the cooked rice, what is the percentage of carbohydrates and sugar? And then because I know myself, I’m only allowed how many units of carbohydrate. That’s why I’ve started counting since the morning…” [DAC04] and “…you can provide several options. What to eat for breakfast and what to eat for lunch and dinner. You can provide a few options, because if there’s only one option and we don’t have the ingredients for it, we won’t be able to follow it …” [DAC11], providing multiple options for each meal to accommodate food availability. Other suggestions for a menu plan included: simple and easy to follow, consider the whole family and not just the client with diabetes, inclusive of different types of food including dessert and drinks, based on research evidence, financially accessible, use spices to improve taste, appropriate for both treatment and prevention of T2D, and modify traditional recipes to reduce unhealthy ingredients.

Although acculturation was not purposefully evaluated in the interviews, some elements were reflected therein. The average acculturation was relatively low in the study sample, which was consistent with the following preferences found in interviews: speaking Chinese, reading information written in Chinese, eating Chinese food primarily, and attending the Chinese Community Centre health seminars. Interestingly, while the participants interviewed were concerned about their diet and nutrition, their diet quality was similar and their adherence to nutrition recommendations higher than the mainly Caucasian cohorts previously reported by our group (Asaad et al. 2016; Raj et al. 2018).

A strength of the study is that the sample, although small, had a wide range of age and immigration duration, which makes it more representative of the Chinese immigrant population. Another strength is that the study was designed around a conceptual model (Zeng et al. 2014) that had been developed to explain factors influencing Chinese immigrants’ health behaviours. Concurrent data collection and analysis could help determine data saturation, which determines an ideal sample size. However, according to a detailed analysis of qualitative research studies, a sample size of 10–20 is among the most common sample sizes (Mason 2010). Hence, a sample size of 14 is acceptable and many of the conclusions were substantiated by the literature. In qualitative research, documentation of processes to increase validity, reliability, and generalizability is important (Leung 2015). While we followed protocols to increase generalizability and validity, the reliability would have been increased by having another researcher independently code the transcripts. This was not possible without translating all of the transcripts into English, which in itself would have increased the potential for misinterpretation. Other limitations included the relatively good health behaviours of the participants (both diet and physical activity), which may have precluded identification of additional barriers.

This study largely validated the Conceptual Model of Diabetes Self-management among Chinese Immigrants (Zeng et al. 2014) although barriers related to traditional health beliefs such as traditional Chinese medicine and yin–yang were not articulated by this group of participants. Culturally relevant education programs could enhance immigrants’ knowledge on healthy eating practice and food choices because the staple foods remained traditional among immigrants and most participants showed support to the idea of developing a Chinese menu plan. Lack of culturally relevant information written in Chinese and lack of detailed instructions on meal planning are perceived disadvantages of current dietary recommendations and prevented Chinese immigrants from following recommendations properly. These limitations should be considered in future nutrition programming. Flexible, detailed examples of meal planning was requested by these Chinese immigrants. It is important to find the factors that determine food acceptability, and take these factors into consideration when developing dietary interventions.