We compared representative samples of the KP and DHS diabetes populations and found significant differences in received SMS. People with diabetes in KP reported that they received substantially more SMS, compared to the Danish population. However, received SMS was not optimal in either system. One third of the KP respondents received the level of SMS recommended in international guidelines, compared to only one in 10 DHS respondents.
In addition, more KP patients used SMS tools than did those in the DHS; three out of four respondents had used SMS tools within the last 12 months (e.g., patient education, support groups or websites, or written information about health and diseases). There were more men in the Danish sample population (62 %) than in the sample population from KP (48 %). Males are known not to use self-management tools to the same degree as women . Thus, the level of use of SMS tools in the DHS population could have been higher if the sample had included more women. In KP, received SMS did not vary between different educational groups. This was not the case in the DHS, where a greater proportion of people with the highest amount of education received SMS and discussed their medical test results with their doctor, compared to the group with the least amount of education. Conversely, more people in the DHS with the least amount of education discussed the importance of medication adherence with their doctor. An explanation for the variation in SMS provided to different educational groups in the DHS and lack of difference between SMS provided to different educational groups in KP could be the significantly lower educational attainment of the DHS population compared to the KP population. Another potential explanation for the variation in SMS provided to different educational groups in the DHS may be that there is less focus on tailoring care to specific groups in a welfare system like Denmark’s, which has relatively low social disparities compared to the US. However, Kaiser members also tend to come from middle to mid-lower socioeconomic group because wealthier families mostly opt for more flexible and more expensive healthcare options. Thus, the KP population also has relatively low social disparities compare to the general US population. Prior studies found that lower education is associated with poorer diabetes-related health behaviors [23–25]. Karter et al. (2007) concluded that, during the course of life, the cumulative effect of reduced practice of multiple self-care behaviors among less educated patients may play an important part in shaping the social health gradient.
With respect to self-management behavior, less than half of the respondents in both systems reported that they exercised as recommended in national guidelines and adhered to prescribed diabetes medication. The cross-sectional study design prevents us from concluding whether self-management behavior in the two populations are linked to HbA1c levels and self-management support, although prior research shows that education and physical activity improve blood glucose regulation [26, 27]. Furthermore, medication (and, by obvious implication, medication adherence) has a significant effect on regulation of blood glucose levels. The populations’ mean HbA1c level may also be influenced by treatment intensity, disease severity, and measurement/monitoring practices. However, in order to investigate these relationships, longitudinal experimental studies or quasi-experimental follow-up studies are needed.
When conducting comparative analysis it is important to be aware that the specific configurations of any healthcare system depend on the historical and cultural context of health and healthcare that varies across and within countries when conducting comparative research [28–30]. When engaging in a cross-sectional, comparative study there are therefore potential lessons to be learned but also methodological challenges and results should therefore be interpreted with care. Strengths of our study include two relatively large samples of individuals with diabetes randomly selected from two different well-described health care systems. Furthermore, the data was collected with a questionnaire developed specifically to capture dimensions of SMS and behavior in a diabetes population.
A limitation to our study is that the inclusion criteria for the two study populations differed, as both people with both type 1 and type 2 diabetes were included in the KP survey and only those with type 2 diabetes were included in the Danish survey. However, 90 % of the diabetes population in KP has type 2 diabetes, and the recommendations regarding SMS are comparable for type 1 and type 2 diabetes. Therefore, we can assume that the responses regarding SMS do not differ significantly between people with type 1 and type 2 diabetes and, consequently, do not explain the differences in SMS between the two systems.
A more relevant difference between the two systems is that KP screens members at risk for diabetes, meaning that people with undiagnosed diabetes will be identified earlier and that diabetes care, including SMS, can be initiated earlier than is the case in the DHS. To some extent, this may explain the different levels in the prevalence of diagnosed type 2 diabetes: 7.7 % in KP compared to 4.2 % in the DHS. The level of SMS support may vary according to duration of disease, as people newly diagnosed with diabetes may receive more education and support than people who have lived with diabetes for many years. It was not possible to obtain information about duration of disease from KP, but we can assume that the survey sample in KP has, on average, been diagnosed and treated for diabetes longer than the Danish survey sample. We surmise that, if it had been possible to adjust our analysis for disease duration, we would have found a greater difference between the systems in the level of SMS provided.
Another limitation of the study is the questionnaire used. The questionnaire was developed for use in a sample population from a US integrated healthcare plan and was not constructed to be used in a comparative analysis of health care systems. Thus, some of the questions may have described more accurate the way SMS is provided in KP and less accurate how SMS is provided in the DHS. Most likely this means that the questionnaire has captured SMS provided in KP to a greater extend than SMS provided in the DHS. It can also be questioned whether the differences in reported SMS between the two health care systems can be explained by differences in interpretation of SMS. However, we asked about specific activities related to SMS to make it more likely that the understanding of the questions would be consistent in the two cultures. For example, the participants were asked how often their regular doctor helped them with a specific plan for what they could do to improve their own health. For both systems, the questionnaire was subjected to extensive cognitive testing to ensure that the questions were easy to understand and understood as intended after being translated from English to Danish using a two-stage process. The surveys were designed for an eighth-grade reading level.
Another potential limitation to our study is self-selection bias. The non-respondent rate was substantially higher in the KP survey sample. Non-respondents had more co-morbid conditions than did respondents, and it is possible that individuals with multiple co-morbidities receive more SMS support. The level of SMS reported as received might have been lower in KP if non-respondents had also participated in the survey.
Despite the minor differences between the two populations and the limitations of the study design, we believe that the results reflect real variations in the level of SMS provided in the two health systems. Furthermore, we believe that the differences in extent and distribution of SMS between the systems can be attributed by differences in the organization of care delivery. This includes the systematic approach to diabetes care including SMS in KP which comprises use of clinical guidelines, stratification of patients according to need for care and support, use of the integrated HIT system allowing for systematical follow-up on patients, panel management and an overview of available SMS services within the healthcare system. Further research is needed to examine how such approaches will influence the delivery of chronic care in public financed healthcare systems.