CHANGING PATIENT
-DOCTOR RELATIONSHIP
Demographic features of a patient playing a role?
Word count: 17,232
Terhi Kangas
Student number: 01714224
Supervisor: Prof. Piet Bracke
A dissertation submitted to Ghent University in partial fulfillment of the requirements for the degree of Master of Science in Sociology
Deze pagina is niet beschikbaar omdat ze persoonsgegevens bevat.
Universiteitsbibliotheek Gent, 2021.
This page is not available because it contains personal information.
Ghent University, Library, 2021.
Abstract:
Modern technologies are increasing their availability and significance in healthcare. Type 1 diabetics are using Continuous Glucose Monitoring to monitor their blood sugar level around the clock, to ease the burden of the disease, and to obtain better results in diabetes management. Diabetics work closely with their diabetes physicians to reach the best possible blood glucose levels. This paper scrutinizes the patient – doctor relationship of type 1 diabetics by looking at the demographic features gender, age, race and educational level of a patient, and if these features can help to predict their satisfaction towards their diabetes physician, at the time when they have started to use Continuous Glucose Monitoring.
This paper draws on a quantitative study from the Jaeb Center for Health Research (USA) from their research “A Randomized Clinical Trial to Assess the Efficacy of Real-Time Continuous Glucose Monitoring in the Management of Type 1 Diabetes.” In the study, participants answered multiple diabetes-related questionnaires when starting to use Continuous Glucose Monitoring. Logistic regression analysis was conducted to see if demographic features of a patient can help to predict the outcome in their doctor satisfaction.
Other than expected, the results suggest that age, educational level or gender of a patient do not help to predict the outcome in patient satisfaction. The variable race was omitted from the analysis since it lacked variance, 96% of the sample being white. Also, the results show how participants across the sample were satisfied with their physician, making the dependent variable skewed and making it difficult to find
significant differences. Yet, this case offers a way to shift the focus onto a more moderate critique of medical technologies, since the patient satisfaction stays good throughout different patient groups after a year of using Continuous Glucose Monitoring.
Keywords: digital health; type 1 diabetes; continuous glucose monitoring; patient-doctor relationship;
1. Introduction
On the 17th of April 2019, the World Health Organisation released for the first time ever guidelines on how countries can use digital health innovations to improve the health of their citizens. They stress the point of how digital health technologies give better opportunities for the people in more vulnerable positions in society, therefore enabling countries to achieve better overall health coverage by using said technologies. The main concerns are data privacy and the equality of health care. It is acknowledged, that the possibility of remote doctor consultations cannot fully replace face-to-face consultations. Rather, new digital health consultation possibilities should complement existing traditional possibilities (WHO, 2019).
Nevertheless, in digital health, new social movements and self-help possibilities are gaining foothold. This causes a shift in a patient – doctor relationship, as the agency is shifting from the specialist to the client. The prominence of the professions’ claim-makers is decreasing, making the patient satisfaction increasingly important in the patient – doctor relationship. This new health care scenario has established itself as an essential change for the whole field, therefore enabling a market shift (Conrad, 2005; Macionis & Plummer, 2012). However, digital health should not be seen as a silver bullet to change the whole field of health care, but rather as an added value to the existing possibilities (WHO, 2019).
In recent literature, medicalization is seen to be more market-driven than before, changing the physician– patient interaction and underlining the importance of satisfied clients. New technologies bringing many new health care devices on the markets, prompts a change in the profession of medicine. The change is rooted in the alteration of the power-relation between physician and client, through the agency-shift facilitated by new health care devices. Physicians are still the gatekeepers for the treatment, but their role is changing. The market-driven medicalization, more prominent in the United States, has created a situation in which the physicians have to cater to the needs of their patients, otherwise, they might lose their clients. Therefore, the patient has gained more agency in this relationship and the very fundamental setting as a doctor being the authority and the patient being the inferior listener, changes when a patient holds the information that before was only in the hands of a physician. Thus, the shift in sociological research on medicalization has to also adapt to these changes in the current medicalization process. Nevertheless, in this conversation, geographical location is a differentiating factor – health care systems are very different in different nation states. The United States is a textbook example of the liberal welfare state with minor state-control, making the society more market-driven than the European counterparts, where state control over health care systems is more prominent (Cohen, Mccubbin, & Collin, 2015; Conrad, 2007; Wendt, Frisina, & Rothgang, 2009).
One specific case wherein the development of health care technologies has been prominent is type 1 diabetes treatment. Type 1 diabetes is a result of a situation where one’s insulin-producing beta cells in the
pancreas are destroyed and therefore, insulin has to be injected. The function of insulin is to balance blood glucose level. Complications stemming from a blood sugar level that’s either too low or too high are not only constantly worrisome but also life threatening, thus maintaining a stable glucose level is a necessity for type 1 diabetics. This causes an emotional and physiological burden for diabetics, and that is what Continuous Glucose Monitoring aims to ease (Vaddiraju, Burgess, Tomazos, Jain, & Papadimitrakopoulos, 2010). Continuous Glucose Monitoring allows the glucose level to be monitored around the clock. This gives the patient the possibility to maintain his/her blood glucose levels very close to those of non-diabetics, which supports the general well-being of diabetics and prevents additional complications. It also allows the patient to strengthen and exert more agency over the diabetes physician when the data about diabetes management is in the hands of a patient (Clarke & Foster, 2012).
The aim of this paper is to contribute to the dialogue in the field of medical sociology about new health technologies and the way they are changing the patient – doctor relationship in the context of type 1 diabetics and the situation in which they are adapting to new health care technologies.
The inspiration for the study came from my Bachelor’s thesis, where I studied the effects of Continuous Glucose Monitoring in the everyday life of a type 1 diabetic. The most interesting result was, how the patients did not seem to be aware of the changing nature of the patient – doctor relationship, regardless of the agency-shift from the physician to the patient. To continue scrutinizing that aspect of the patient – doctor relationship further, this paper asks the question “why”? Can certain demographic features - age, race, educational level or gender - of a type 1 diabetic, explain the satisfaction towards their physician?
Therefore, the focus point in this paper is type 1 diabetics and the demographic features of the patients. I will be researching whether variables age, race, educational level or gender can help to predict the outcome of the dependent variable ‘unsatisfaction towards the diabetes physician’ when type 1 diabetics have taken a new health technology, Continuous Glucose Monitoring, into use. I expect to find differences in patient’s satisfaction towards the physician, when patients are at different ages, I expect to find women to be more satisfied than men, highly educated subjects to be more satisfied and lastly, I expect there to be differences between races.
The results of this paper in a wider context are aiming to fill the gap in literature between the users of Continuous Glucose Monitoring and the patient satisfaction while taking into account the patient’s increased agency in the patient – doctor relationship through technology-led self-care. By performing a logistic regression analysis, it is possible to see the separate effects of different demographic features of a patient to the patient satisfaction towards their diabetes physician, and by building a model of these independent variables it is possible to scrutinize their effect on the dependent variable about patient satisfaction, if the patient is or is not satisfied with their diabetes physician.
The research question is studied by using secondary data. Jaeb Center for Health Research (JCHR) - Diabetes Research Studies from the United States provides data from their study “A Randomized Clinical Trial to Assess the Efficacy of Real-Time Glucose Monitoring in the Management of Type 1 Diabetes”. In the used data sets, they have 223 respondents for questionnaires related to different life-areas for diabetics. Logistic regression analysis is carried out to address the research question, and the results are discussed accordingly.
In the literature review, I’m scrutinizing the patient – doctor relationship and its changing nature, focusing especially on patient satisfaction and the effect of the demographic features. I address the health inequalities creating differences in health outcomes between different demographic groups. Then medicalization as a phenomena is discussed; its developments, and the complexity of the concept itself. Afterwards, I give an introduction to type 1 diabetes, most importantly to Continuous Glucose Monitoring. Finally, before discussing the methods and the analysis, I aim to build an understanding about the sociological side of the disease by addressing the burden of diabetes.
2. Literature review
2.1 Satisfaction in a patient – doctor -relationship
Patient satisfaction is one of the variables, and often portrayed as the most important one, to measure the quality of health care. Satisfaction for the physician is a great indicator when health care services are evaluated. The discussion about the patient-doctor satisfaction in the academic literature is situated in an interdisciplinary field of medical sociology, containing also features from communication studies. Patient satisfaction consists of various features, for example, verbal and non-verbal communication, technical competence of the physician, possible interruptions of the consultations, and so on. Also, patient satisfaction is playing an extremely important part when digital health is becoming a more general way of handling health care, and remote consultations are increasing their importance (Govere & Govere, 2016; Kruse, Krowski, Rodriguez, Tran, & Vela, 2017; Sebo, Herrmann, & Haller, 2016).
Jalil, Zakar, Zakar & Fisher (2017) conducted a study about patient satisfaction within diabetics and their diabetes physicians in Pakistan. They considered the aspects of technical expertise, interpersonal aspects, communication, consultation time, and access and availability. The biggest sectors in need of development to improve the patient satisfaction were recognized to be interpersonal skills of physicians and their clinical skills.
The study also addressed social-demographic differences within diabetics playing a role in their satisfaction. They showed that low educated subjects had sometimes issues with communication with their diabetes physician since they are not educated about for example new technical developments. Therefore, they can be said to have a cultural health capital –gap (discussed in a section 2.4) between a doctor and a patient, making the communication harder. Besides, they found it difficult to rate themselves as dissatisfied with their diabetes physician, since for them the physician is an authority, a person to be respected, and the patient-doctor communication is not seen as a relationship based on mutual respect from both ways. They also showed that women were seen as more likely to be satisfied with their physician, than men. The likelihood to be more satisfied increased with the technical expertise of the doctor being better, and when the interpersonal aspects were taken into consideration. Another important remark of the study was that three-quarters of diabetics reported themselves having a high level of satisfaction, indicating that only very few subjects in the sample were not satisfied. This result supports the idea of dissatisfaction being hard to report, causing the overall satisfaction level to be very high (Jalil et al., 2017; Shim, 2010).
Some studies take into consideration the gender of both, the gender of a patient and a doctor. A study conducted in the United States in the 90s showed how patients who chose a doctor of the opposite gender,
tended to be more satisfied with their practitioner that the patients who had chosen a same-gender physician. To explain this, it was suggested that it can be partly due to different expectations what men and women patients are are holding against the physician – female patients seemed to have higher expectations towards their physicians than men. Another study that took into consideration the age and gender of a patient, concluded that above 36 years old women were significantly more satisfied with their practitioner that their male counterparts. For younger women, 35 or below, they were found to be the most satisfied when the time spent with the doctor was scrutinized, but then slightly contradictory results gave an indicator that they were also the most dissatisfied when general satisfaction was measured. Therefore, the results when satisfaction and gender are compared, are not very clear-cut, but support the hypothesis expecting women being more satisfied than men (Samohyl et al., 2018; Sebo et al., 2016).
Laveist and Nuru-Jeter (2002) conducted a study on the relationship between race and doctor satisfaction. In their study, they found that throughout all race groups that were studied, all were more satisfied with their doctor if they had a similar racial background. The sample contained white, African American, Hispanic, and Asian American respondents. Also, the study showed that from all these groups, white people had the highest likelihood to choose their physician in line with their race compared to other groups. Nevertheless, all groups studied, if they had the freedom to choose the physician, were more likely to be race concordant. Govere & Govere (2016) in their review of studies about cultural competence training for health care providers show, that the physicians who had received cultural competence training, had better results in patient satisfaction. Cultural competence training means that the physicians are educated to give health care in culturally sensitive way by taking into consideration individual patient and their cultural background. Moreover, the satisfaction towards the physician was seen increasing when a patient comes from a minority group.
2.2 Patient – doctor communication
Electronic correspondence is becoming a more general way of changing information about one’s health. It has a goal to make face-to-face consultations more efficient by changing the information about collected data already before the consultation, saving time from the consultation to focus fully in the matter-at-hand. On the other hand, new devices and for example new computer programs bring more work for health care professionals who have to learn how to use them to get the advantages that they offer. Also, new forms of consultations and information gathering moves the authority and the responsibility from one’s health care from the physician to the patient, challenging the prevailing idea about health professional having better knowledge than the patient themselves. However, the backbone of a patient – doctor relationship is still in personal communication and understanding the needs of an individual patient (Wehbe, Curcio, Gajjar, & Yadlapati, 2015).
In the study by Aryoseto, Tamtomo & Murti (2016), they studied how younger doctors, as “early adopters”, take over digital platforms easier than their older colleagues. It was proven that younger doctors are more likely to have satisfied patients than older doctors. This was an interesting observation since older doctors have had more career to gain their clinical and technical expertise and learn more about the interaction with the patients. However, the older physicians were found to have significantly lower job satisfaction than their younger colleagues. This emphasized the importance of work satisfaction from the physician’s side to have a good patient – doctor relationship.
Communication between the patient and the doctor can be divided into two categories: Instrumental communication, which includes providing directions or instructions, educating patients and discussing test results. An inevitable part of communication, which is now threatened by the development of health apps and new medical devices, is affective communication. Affective communication includes providing reassurance and verbal support, also showing empathy, compassion, and concern (Kashgary, Alsolaimani, & Mosli, 2017). WHO also takes a stand about this by addressing health services that are provided through digital services, for example iPads or smartphones. They stress the importance of equal possibilities to have consultations from specialists, but that it cannot replace human contact and actual appointments completely. Importance to ensure equal possibilities for everyone to have access to personal health care consultations is stressed, also for people in vulnerable positions in society. Also, the question of who is the actual care provider, was raised. The assistance regarding one’s health care that is offered through chat-window, has to be provided by an educated health-care professional (WHO, 2019).
Yellowlees, Chan & Burke (2015) in their study acknowledge the fact, that new, younger generation of physicians and patients are already living in an information-driven networked world, and new ways of communicating between health care professionals and patients is made ordinary. For some patient groups in telepsychiatry, it has been clear for a while that using ‘virtual space’ rather than in-person consultations is a more efficient way of reaching patients, which improves patient satisfaction in this group. For the future, they see especially for psychiatrists, that hybrid-model will become a more general way of working, both in-person and online, to improve patient care.
2.3 Health inequalities
In the field of medical sociology, health inequalities are widely discussed. The aim of public health care systems is to make health care equal and as accessible to everyone, albeit that is not the only goal. The definition of the concept is not fully agreed about, but mostly these conflicting views agree about health inequalities describing differences between better- and worse-off groups between different demographic groups. In the United States, the concept used most often is ”health disparity”, which as a core is the same as health inequality, but has an implication of injustice, as well as inequality. In the United States health
care injustice more often also includes the differences in how different race groups or ethnicities are treated in the field of health care. The definition that is often used for inequality is “lack of equality as of opportunity, treatment or status” (p.472) and this is prominent also in the field of health care with better- and worse-off groups (Carter-Pokras, 2002).
The inequalities in health care have been a research topic since the 1960s, dating back to the same time when medicalization was starting to be discussed and the concept formulated. Already in a study conducted in the 1970s, it was concluded how social class makes a difference in the use of health care services. The middle class was seen to make more use of preventive services, and that they receive better care than people from other social classes. They were also seen having a more satisfactory relationship with their doctors, while being more critical towards the care they are given. They had a better relationship with their practitioners, and the practitioners knew more about their domestic situations. Middle-class patients were seen to be more prone to ask questions, compared to working-class patients. One suggestion as an explanation was that the patients and doctors feel more in the ease with each other since doctors themselves belong to the middle class. Nevertheless, working-class patients were very likely to be satisfied with their consultation visit. Another more recent study from 2008 confirms, that the issues that were in place already decades ago still exist in a modern society; massive health inequalities were found by social class and by race and ethnicity. The general health is shown to be worse based on a race; health is consistently worse for black people compared to white people. Also, people with less economic capital are in the worse-off -group. The study was conducted in the United States, where also geographic location and time played a role in forming health inequalities (Adler & Rehkopf, 2008; Cartwright, Brien, Cartwright, & Brien, 1968; Conrad, 2005)
When it comes to gender, in several studies men are shown to be more noncompliant or less adherent to treatments than women. They can be seen more defensive towards authorities, and less conciliatory to follow advices given. On the other hand, several studies have not found significant differences between genders, and some have shown women to be less adherent than men, for example in a case of highly active antiretroviral therapy. There the research showed, that men are more likely to be adherent towards the treatment than women (Berg et al., 2004; Peyman, Louis, & Karel, 2016).
Age can also be seen causing differences in health outcomes when it comes to being compliant towards treatments. Patients less than 60 years old are observed to be less compliant than patients who were above 60 years old, in a study where people suffering from osteoporosis in Germany were studied. The importance of age in adherence to medication has been in research for decades, in a study conducted already in the 1980s showed that patients suffering from hypertension, patients from the age 55 to 65 had significantly lower adherence than patients younger than 55 years old or older than 65 years (Peyman et al., 2016; Weingarten & Cannon, 1988).
Health differences can be observed from different perspectives, also through Bourdieu’s forms of capital. Bourdieu’s theory of capital as its core is to try to situate people in social space, by using social, economic, and cultural capital. These forms are strongly intertwined, and they can be converted within each other, and used to gain more of others. This theory is applied to study health inequalities when looking at the relation of the amount of capital as one has, and health outcomes. It is shown, that each form of capital has positive effects on the health of the subjects in a study conducted in Belgium. The study took into consideration both, mental and physical health of the subjects. Cultural capital items were seen to slightly differ from other forms of the capital. It only had a positive effect when looked at physical health, not with mental health (Bourdieu, 1984; Pinxten & Lievens, 2014). The concept of cultural capital, and cultural health capital, is elaborated further in the following section.
2.4 Cultural health capital -theory
Cultural factors play a fundamental role in building the structures of society. Cultural capital is one of three capital’s in Bourdieu’s theory of situating people in social space, and also applicable to public health. The theory of cultural capital is strongly theoretical, making it possible to apply it to practical health promotion. Also, Bourdieu’s approach includes different levels of analysis, so when applied to the field of health, different groups, families, peer groups, communities, can be observed and theory can be applied. In the cultural capital, Bourdieu’s theory takes into consideration people’s values, norms, and behaviours, which can directly be applied to health behaviour; eating habits, physical activities, etc. Cultural processes are seen as part of social differentiation and unequal life chances; which can help to create a broader approach for health inequalities (Abel, 2004; Bourdieu, 1984).
Nevertheless, the cultural capital theory also makes the differentiation between other forms of capital by stating how the amount of cultural capital one has, does not determine the socio-economic status of oneself. It is widely acknowledged, that lower socio-economic status has an association with worse health outcomes, but cultural capital can bring another perspective to contribute to understanding health inequalities. Education is a strong indicator of the amount of cultural capital one has, but not the only one, also when predicting the health outcomes based on their culture-based resources (Abel, 2008; Gagné, Frohlich, & Abel, 2015).
A concept of cultural health capital was proposed by Shim (2010) to build up on Bourdieu’s work on cultural capital. Cultural health capital describes the differences in situations when patients and professionals are communicating with each other. The concept can be applied to any situation when these interactions are taking place, for example when patients are being advised by a health care professional to take over new treatments. Cultural health capital sheds the light on those communication situations and the ways
disparities can be produced. She describes cultural health capital ”as the repertoire of cultural skills, verbal and nonverbal competencies, attitudes and behaviours, and interactional styles, cultivated by patients and clinicians alike, that, when deployed, may result in more optimal health care relationships” (Shim, 2010, p.1).
She points out, how often people with low socioeconomic status have low cultural health capital and vice versa. Other things are playing a part as well, for example, cultural resources: ways people who are interacting are dressed, the way they talk to each other, and their general manners. However, the amount of cultural health capital that an individual is holding is not always predictable. People with higher cultural health capital are more likely to take over new devices and treatments, and their communication with a health care professional seems to be better and their health care satisfaction is higher. On the other way around, people with lower cultural health capital are more likely to not to get along with their practitioner so well, and they are often not so happy with their treatment (Shim, 2010).
Missinne, Neels, and Bracke (2014) also discuss the importance of accumulating cultural health capital from the early years on in children’s life. They suggest that building up cultural health capital starts already from a young age, and it plays a part in participating in preventive health-care services, their study focusing on mammography screenings. Their results indicate, that women coming from a family with higher cultural health capital are more likely to go to mammography screenings later in life. Cultural health capital was measured with many indicators, some related to the health services and some to the cultural or economic capital, for example by examining if a woman as a child was going regularly to dentist check-ups, how many books were there when they were children, etcetera.
People with higher cultural health capital have a higher likelihood to take new health care applications to use. They can, in fact, be ahead of health care professionals when it comes to using new digital devices to access health information (Reis, Visser, & Frankel, 2013). Doctor – patient communication is still the very cornerstone of medical practice. Communication between a patient and a doctor is still a very important factor for patient satisfaction, medication adherence, and general health outcomes. This personal communication between patient and health practitioner is a very important factor in the consultation appointment, how a patient feels that they are heard and encountered at this moment when they are in need of help (Kashgary et al., 2017).
2.5 Modern health care technologies
Smartphones allow us to download numerous self-care apps to promote our well-being without the need to contact health care professionals. Many health-apps are focusing on reducing anxiety, meditation, yoga, or mindfulness. A study made for elite university students in the U.S. showed that with using these apps, the
students found the apps as a useful tool, as a ”therapy”, to cope with stress caused by the university. This is an example of how new developments in a health care sector are producing positive medicalization (Maturo & Moretti, 2018).
Modern information technologies are increasing their significance in healthcare. Patients can inform themselves through websites dedicated to diseases, where medical details and case studies give patients the possibility to be more active in their self-care (Macionis & Plummer, 2012). Tablets and smartphones are allowing patients to have full consultation at home, and the need to go for a doctor’s appointment to hospital or clinic has changed. At the same time, the patient is gaining more agency in their self-care in relation to health care professionals, who used to be seen as the ones holding more knowledge regarding patients’ health (Wehbe et al., 2015). Maturo and Moretti (2018) discuss, how only 10 years ago all diagnostic tools were controlled by medical practitioners, and now they can be found from smartphones. Self-monitoring enables us to see changes in our health, how many hours we sleep, how many steps we take per day, or do we exercise enough. These possibilities of self-monitoring can cause technostress in individuals, who may feel overwhelmed with these new information and communication possibilities. The concept of technostress refers to a situation when the individual cannot handle the information flow from all communication technologies without unnecessary mental weight. It is recognised, that in the health care sphere, this technostress often relates to confidentiality and to concerns about how well the data the devices are collecting is protected (Ash, Berg, & Coiera, 2003; Real et al., 2018). Therefore, managing the knowledge and the data is currently a very important issue, in health care as well as other spheres where information technology is gaining importance. In the health care sector ’information paradox’ is an upcoming issue. Information paradox means that health care professionals have so much new relevant research to read and to be used in everyday work, that they are overwhelmed with the amount of information. Then, when it comes to finding particular information when it is needed, that can turn out to be a challenge (Nicolini, Powell, Conville, & Martinez-Solano, 2008).
When new devices are developed, they always come with intended, but also with unintended consequences. This was discussed in Robert Merton’s landmark publication “Social Theory and Social Structure” where he introduced manifest and latent functions. With manifest function, we mean all intended consequences, but latent functions are the ones that occur when specific development happens but causes consequences that were not intended (Merton, 1957). Ash, Berg, & Coiera (2003) studied what kind of unintended consequences come with information technology of health care. One category of the unintended consequences is the errors when entering and retrieving information, and another category is problems with the coordination and communication process. The first category means problems when physicians have only a limited amount of time for one patient, and human errors occur when working at a high pace. A physician might give wrong advice for the wrong patient because they have the file of the wrong patient, or they can by accident add some wrong information to the patient’s file. Typing down wrong
information causes flaws, other health care professionals, phone calls, etcetera, often interrupt computer work time. That troubles focus from the task at hand. The second category of errors is about communication errors, for example when data is misrepresented and miscommunicated, and based on that, wrong outcomes (advices for patients) can occur. Health care practitioners’ work can sometimes be presented as clear-cut and linear, meaning that the doctor is the one giving the orders, nurses execute them and then the doctor evaluates the outcome. In reality, this is often not the case, and the possibility for flaws increases. The study shows plenty of grass-root level examples, how developments in information technology of health care come with (negative) unintended consequences, on top of intended positive consequences.
2.6 The development of medicalization
Going back in history, to the 1970s, people did not identify diseases such as attention-deficit/hyperactivity disorder (ADHD), anorexia, or panic disorder to even exist. During the last decades, the number of medical diagnoses has increased enormously, as its part enhancing the creation of a phenomenon called ‘medicalization’. Sociologists have studied medicalization since the late 1960s, starting from the medicalization of deviance. In the 1970s, medicalization was seen to be applicable also for a wider range of human problems, which were making an entrance into the medical sphere (Conrad, 2007).
Michel Foucault in “Madness and Civilization” (1961) opened a discussion about medicalization. He does that by discussing the case of what we consider now being “mentally ill” people. Before, they were living alongside the normal, sane, citizens. They were considered to be different, instead of ill or crazy. Now, they are hospitalized, medicated, and looked after by health professionals. This is an attitude he wanted to be demolished (and at the same time, demedicalize mental illnesses) because he considered that for those people, their lives were better during the renaissance when they were allowed to be free and live among everyone else. In his next publication “The Birth of the Clinic” (1963) he looked at medicalization from a broader perspective, and systematically criticised how medicine and healthcare had not become more human throughout history. He acknowledged how medicines and treatments have developed, but gave a strong critique for health care professionals who have started to see their patients as a “broken organ” instead of a person. This is what he called “the medical gaze”.
The definition of medicalization is “the process by which events and experiences are given medical meaning and turned into medical problems” (Macionis & Plummer, 2012, p. 743). In other words, it is a process where some aspects of human life, which were not considered as pathological, are now considered as medical problems (Maturo & Moretti, 2018). The literal meaning of “medicalize” is “to make medical” (Conrad, 2007, p.14). Medicalization has turned many problems that were previously seen as deviant behaviour, into part of the medical work. Some issues that were seen as moral issues before,
become medical ones, for example, alcoholism is now in the field of medicine, when before it was not. However, the field of medical sociology has been struggling to find a solid definition for medicalization. Recently, growing complexity in the market of health care and changed power relations in the medical profession in relation to their patients has resulted in the different layers of medicalization. (Christiaens & Teijlingen, 2017; Conrad, 2005; Macionis & Plummer, 2012)
The ’medicalization thesis’ by Conrad (2007) focuses on two dimensions, both associated with Foucault’s work. First, how initiators of medicalization develop, and how it is applied to already existing medical categories. Secondly, it focuses on how people have internalized medical perspectives and take medical solutions to new medical issues for granted. In other words, “medicalization of all sorts of life problems, is now a common part of our professional, consumer and market culture” (p.14). Here the context is important to keep in mind – Conrad is an American medical sociologist, therefore, his medicalization thesis can be more applicable to the United States –context where the medical sector is more market-driven. The claim about growing spending in health expenditure is however backed up by statistics: Worldwide average of health expenditure has grown from 8.561 (% of GDP) 10.021 in 2016. In the United States, the growth has been even stronger, from 12.502 in 2000, to 17.073 in 2016 (Wendt et al., 2009; “World Health Organization Global Health Expenditure database,” 2020).
Medicalization is often treated in the literature as something that either exists or does not, as a state, rather than a continuum that can increase and decrease over time. It causes problems, since then one has to determine a certain threshold that has to be filled before the issue can be said being medicalized or demedicalized. Also, then increases and decreases of medicalization are disregarded, if the change is not seen as so significant that it would produce a drastic change that issue could be said completely ”demedicalized”, even tough difference would be significant in other ways. Lastly, it ignores the fact that medicalization and demedicalization can happen simultaneously, as it often does (Halfmann, 2012). One example of a problem is when medicalization is seen as a state rather than a continuum, occurs with the case of childbirth. Conrad (2007) discusses, how on the other hand childbirth often nowadays happens without the physician being present, also often without episiotomy or medicines, it is still not considered as demedicalization, since midwife is often present in labour to help. He sees that in order to call childbirth demedicalized, it should happen without hospitals, medicines, and health care professionals playing any part in it. Thus, even though change is drastic compared to childbirth in a hospital with all medical help available, if medicalization is not seen as a continuum, this important progress is ignored. Additionally, it is important to observe how we understand ‘medicalization’. It can be seen as a continuum, or as a binary positioning, as in something that exists or does not exist. If we consider medicalization as something that only happens or does not, we are disregarding important developments in the medical sphere (Halfmann, 2012; Van den Bogaert, Ayala, & Bracke, 2017)
Van den Boagert, Ayala, and Bracke (2017) in their work examine medicalization and demedicalization as a continuum, and distinguish mechanisms how this can be seen happening in current society. Driving forces of medicalization are for example increasing competition between health insurance companies, who have a high interest in making people invest more money in their commodities they are selling. Also, as discussed before, the GDP share of the pharmaceutical industry has increased enormously. Drivers for demedicalization are also discussed, how over-consumption of medicines is recognized, and therefore new campaigns from health care policymakers are focusing rather on self-care than the use of medicines and money consumption on the medical industry. States are also starting to step by creating new policies, in order to decrease medicalization and to decrease the expenses that the medical industry is causing.
Conrad (2005) in his other work also discusses drivers of medicalization and how they are changing in the 21st century. He claims that the importance of professional claim-makers is decreasing when commercial and market interests are increasing their importance. This has given medical practitioners a role as a sort of gatekeeper between markets and patients. Medicalization as its core stays the same, but the availability of new pharmaceutical and potentially genetic treatments are driving medicalization, the role of practitioners is emphasized when they are the ones giving recommendations and they can prefer some companies over others who are offering similar possibilities. Medical devices can already work as a part of the body, imitating an organic function better than traditional treatments, for example, insulin pump replacing insulin injections. There, the development of medicalization can be scrutinized through the development of medical treatments and tools in diabetes management, first, the function of the pancreas was replaced by manual injections, and then the next step in medical technology, insulin pump, is taking over the task and delivers the treatment half-automatically, aiming to ease the burden of diabetes and work as an artificial pancreas (Bradby, 2009).
New medical equipments, which allow people to self-monitor themselves without the practitioner’s guidance, for example health –apps, are one of the driving forces of medicalization. The apps make the detection of potential abnormalities already in very low quantities possible for everyone. This gives agency of one’s health care for people’s own hands instead of a health practitioner, who has traditionally seen as having the ”better knowledge” of health, based on their education. This development can be seen in two different lights; firstly, it can be seen leading to over-diagnosing illnesses when individuals can question their own health sooner than before. Before patients would have booked a consultation only noticeable symptoms would have occurred and therefore more time would have passed, but also secondly it can enable diagnostics to be done earlier, so potential illnesses can be prevented before the situation develops further (Maturo & Moretti, 2018). When the authority of health-care moves from health care professionals to individuals, self-monitoring becomes easier and the process of medicalization grows stronger (Macionis & Plummer, 2012). New arising problems regarding medical devices come with the division of people into two categories. Firstly to the ones who are willing to take over new devices and treatments, “early adopters”,
and secondly to those who are not interested in new, modern, possibilities, and because of that, end up being ‘left out’ from new treatments that could be better than traditional ones (Kerr, Axelrod, Hoppe, & Klonoff, 2018).
2.7 Type 1 diabetes
In this chapter, the type 1 diabetes is discussed to explain why it is an important research topic in the field of medical sociology. Type 1 diabetes is a chronic disease that cannot be cured. Therefore, living with the disease is aimed to make as easily managed as possible, to ease the burden what the disease causes. For that, developing health care technologies are offering help. These themes are elaborated further in the following section.
Globally in 2014, an estimated 422 million people were living with diabetes. This number has increased enormously, compared to 108 million diagnosed diabetics in the 1980s. Considering the adult population, nearly 10% of the whole carries diabetes. An important distinction to make is the difference between type 1 and type 2 diabetes. The majority of people with diabetes (approximately 90%-95% (CDCP, 2017)) are type 2 diabetics, people whose body cannot properly use the insulin it produces. A smaller proportion of diabetics are type 1 diabetics, whose body starts to shut down the insulin production, and eventually, the body does not produce insulin at all, and continuous insulin injections throughout their lifetime are required to survive (WHO, 2016).
Type 1 diabetes is a result of a situation where ones’ insulin-producing beta cells in the pancreas are destroyed, and therefore insulin has to be injected. The function of insulin is to balance an individual’s blood sugar levels. Complications from too low or too high blood sugar are life-threatening, thus maintaining a stable glucose level is a necessity for type 1 diabetics. To do that, it is recommended that type 1 diabetics should check their blood glucose levels at least four times per day. This causes an emotional and physiological burden for diabetics, and it is something that new devices in diabetes care are trying to mitigate, for example through Continuous Glucose Monitoring, which has a goal to release a diabetic from continuous blood sugar level measurements, and therefore ease the burden that the disease is causing (Vaddiraju et al., 2010).
New diabetes care devices can give more accurate and rich data to help patients in their self-care and make their everyday-life more stress-free. Data that is produced with these new devices bring more possibilities for self-measurement, which allows patients to be more autonomous and gain the agency in their own self-care. It is no longer needed to consult a health care practitioner regarding for example laboratory examinations to get average blood glucose level information – patients can have large amounts of data for themselves. At the same time, the patients’ self-care is growing more dependency on
technologies, computer-based programs, and algorithms, instead of manual treatments (Liberman, Buckingham, & Phillip, 2014; Maturo & Moretti, 2018).
The development in diabetes technologies, more specifically in blood glucose measuring, has been fast during the last decades. The first steps in measuring blood glucose were taken already at the beginning of the 20th century, the first home-tests were developed right before the 1960s, when reagent strip reacted to urine, and from the colour the test stripe turned to, blood glucose level could be defined. In the 1970s after hbA1c (hemoglobin a1c, an indicator of how well diabetes is managed), was introduced to the public and made diagnosing diabetes a lot easier, interest in developing diabetes treatment grew and development started to be increasingly faster. In 1974, the first blood glucose meter that required only a small amount of blood from fingertips was introduced. During the 1980s and 1990s blood glucose meters developed fast; smaller in size, smaller blood volume required, capillary-fill stripes were examples of the developments made at a fast pace during a couple of decades. In 1997, data from the blood glucose meter was possible to download to one’s own computer. Continuous Glucose Monitoring systems were again a big leap forward, although, in most of the systems, Continuous Glucose Monitoring still requires traditional blood glucose measuring next to it, to calibrate the system (Clarke & Foster, 2012).
2.7.1 Diabetes treatments
Type 1 diabetes treatments can be generally divided into three categories. The first option is simple traditional systems, such as multiple daily injections with a blood glucose meter. Secondly, more modern insulin pump therapy, sometimes paired with continuous glucose monitoring. Lastly, techniques integrating these devices, as in sensor-augmented pump (SAP) therapies (Naranjo, Tanenbaum, Iturralde, & Hood, 2016). The newest addition to these technologies came with MiniMed 670G, which with its automated insulin delivery is the first artificial pancreas in commercial use (Barnard & Breton, 2018).
These three categories have an increasing amount of new features, the newest devices having for example bolus calculators and low glucose suspend, which require highly developed algorithms to predict where one’s blood sugar levels are going and how to react to these changes. The first category, multiple daily injections, represents the more traditional way of diabetes health care, where blood sugar level is measured from the fingertip and based on that, the patient decides what would be the right amount of insulin to inject, without new technologies assisting with the decision (Barnard & Breton, 2018).
The use of insulin pump therapies has been increasing since it was launched in the late 1970s. The greatest prevalence of pump use is in the U.S. (between 40%-62% of adults with T1D), while in Europe numbers are still significantly lower, although increasing (from less than 5% to greater than 15%). It is
proven, that for adults, insulin pump therapy provides a significant reduction of hbA1c lower frequency of hypoglycaemia episodes, and reduced insulin dosages (Naranjo et al., 2016). insulin pump therapy is complex and demanding – patients with it report better cooperation with health care professionals compared to other treatments. This indicates, that people using insulin pump therapy are holding higher cultural health capital (Franklin, 2016).
The “human factor” in diabetes care is important to take into consideration. Satisfaction to insulin pump therapy and Continuous Glucose Monitoring is largely dependent on whether the device is user-friendly or not, or if use is seen as convenient or inconvenient. Another important factor is if technologies are helping effectively the everyday management of diabetes (Liberman et al., 2014). Adults who are starting insulin pump therapy, are endorsing improved glycemic control, they want to gain more flexibility in their lifestyle and reduce fluctuations in blood glucose levels. These demands are often met with long-lasting pump therapy. Benefits are not so evident with new users participating in a trial, but for existing pump users benefits can be shown. This speaks for the fact, that it takes time to get used to new devices and adjust them to be part of everyday treatment. Often the problem with new diabetes devices is that they are used inconsistently or patients are not satisfied with them directly from the start and therefore they do not want to continue using them (Naranjo et al., 2016).
insulin pump therapy can offer flexibility and possibly a more normal lifestyle, but it comes with a constant reminder of diabetes since it has to be worn at all times every day. A fact that one has to be constantly connected to a medical device can create a psychosocial burden. It can affect ones’ body-image, which is a concern that has been more prominent amongst female users (Franklin, 2016).
2.7.2 Continuous Glucose Monitoring
Typically, Continuous Glucose Monitor consists of three different parts. The first one is the sensor itself, connected to the patient’s body, and it measures physiological glucose levels continuously. The second part is an electronic processing unit, which can be connected to the sensor either with wire or wirelessly. The third part is a data display unit. This data is used to determine the need for insulin to balance the glucose level (Vaddiraju et al., 2010).
Continuous Glucose Monitoring came to the commercial markets only in 1999 when Medtronic MiniMed Continuous Glucose Monitoring System was approved (Naranjo et al., 2016). Liberman et al. (2014) in their article collected together studies about Continuous Glucose Monitoring and its effects on diabetes treatment. A study by Peyrot and Rubin proved that if patients see their devices as convenient and they decrease the amount of diabetes-related daily activities, it leads to good treatment satisfaction.
What is most often seen in positive user feedback for Continuous Glucose Monitoring, are trends in blood glucose levels, which make it easier to detect hypoglycemia and correct out-of-range numbers (Naranjo et al., 2016). Use of Continuous Glucose Monitoring is increasing continuously, but clear guidelines on how patients could make the best out of the data are not available. The optimal way to adjust insulin based on Continuous Glucose Monitoring data is through complex processes, where patients have to take into consideration numerous amounts of variables, for example their last meal and its nutritional values, physical activities and the last bolus. Large amounts of data are collected, and the best use of it can be achieved through user education and good cooperation between patient and health care professionals (Castle & Jacobs, 2016).
Sometimes attitudes towards using new devices, like Continuous Glucose Monitoring, can be negative. Barriers to Continuous Glucose Monitoring use can be divided into different categories. First, structural barriers, which include lack of time to educate users for technologies, and costs associated with devices and insurance systems. Secondly, psychological and demographic barriers, which include depression, infrequent monitoring of blood glucose and female gender, coming from a single-parent household and older age when diagnosed with diabetes. Also, some factors are associated with the likelihood of discontinuing use of the device, for example having severe hypoglycemic events after starting insulin pump therapy, the physical discomfort of wearing Continuous Glucose Monitoring, general difficulties with functioning of devices, frequency of alarms, skin reactions to tapes in sensors and interference with physical activities. These problems are also likely to cause technostress and make patients stop using Continuous Glucose Monitoring (Naranjo et al., 2016).
Multiple pieces of research suggest, that more committed a patient is to use complex diabetes treatments, the higher their uptake of Continuous Glucose Monitoring. For youth, lower hbA1c, two-parent families and higher quality of life are variables that can be seen determining interest in starting Continuous Glucose Monitoring. Adult Continuous Glucose Monitoring users are more likely to have private insurance and have higher household income and higher education, compared to non-users (Naranjo et al., 2016).
2.7.3 The burden of diabetes
In the U.S. 76% of people with diabetes reported that they feel stigmatized because of diabetes. 72% of respondents felt that diabetes was seen as a “failure or personal responsibility”, although it is worth noting that in this category type 1 and 2 diabetics were not differentiated from each other (Barnard & Breton, 2018).
A study by Barnard & Breton (2018) shows that the burden of diabetes extends larger than just physiological conditions. Psychobehavioral problems are acknowledged, yet not fully understood or taken into consideration. This is the case most often for diabetic teenagers, who perceive more unwanted attention from health care professionals than before, because of larger amounts of data and better possibilities to reach better outcomes in treatment. They can feel that they are seen more as a “broken pancreas” than a person, linking to Foucault’s (1963) writing about “medical gaze” how health care professionals might forget to look at their patient as a whole entity and focus only on this one specific part of a human that they have the expertise to make better.
The emotional distress of diabetes is associated with many variables. For example, younger age and another chronic illness and high hbA1c were seen as factors causing distress. The largest difference in emotional burden level was associated with low diabetes-specific support, low generic quality of life, and diabetes empowerment (Joensen, Almdal, & Willaing, 2016). Results of another research indicated women and participants with higher hbA1c levels having significantly higher emotional distress (Graue et al., 2012).
New devices can be used to make emotional distress smaller. This can be done for example by remote physiological monitoring of diabetics. Data can be sent to the doctor via computer, and with a short look, they can give advice to which direction patients insulin doses should be changed, without need to go to the doctor’s office for a regular check-up. This decreases the need to schedule doctor appointments and travel to the hospital or clinic, and check-ups could be done more frequently and in that way make hbA1c of a patient better (Ubl, 2007).
3. Methods
Research strategy
The research strategy for this paper is to conduct a quantitative research by using secondary data. The analysis is conducted with binary logistic regression. The original data set is a public data set from Jaeb Center for Health Research (JCHR) - Diabetes Research Studies, from the United States where they conducted research “A Randomized Clinical Trial to Assess the Efficacy of Real-Time Glucose Monitoring in the Management of Type 1 Diabetes”. The aim of the study was to determine whether Continuous Glucose Monitoring improves the quality of life and glycemic control in children and adults who have type 1 diabetes. The data was collected by using questionnaires, between January 2007 and January 2009. This dataset was deemed appropriate for my research because it included information about demographic features of the subjects, and it measured many aspects of type 1 diabetic’s life at the time when they started using Continuous Glucose Monitoring, moreover the diabetics satisfaction towards the physician. The datasets, codebook, and the case report forms are public and freely available on the Research Center’s website (Beck & Ruedy, 2008). This paper draws from the original study by taking into consideration aspects the original study did not discuss. Regardless of the data sets being a decade old, the original data set has not been used for an analysis of this kind.
Binary logistic regression was chosen as the analysis tool because of the skewed nature of the dependent variable, making OLS regression not possible to use because the assumption tests were not passed. The binary logistic regression does not have assumptions of that kind and the analysis can be carried out. The dependent variable was recoded into binary variable “satisfied – non satisfied” to balance the variable and to create more variance to the variable which as in its original form was extremely skewed and did not have much variance.
Sample and design
Participants in the randomized clinical trial had to be clinically diagnosed with type 1 diabetes, be more than 8 years old and have as a treatment method either insulin pump therapy or multiple daily injections. In the analysis, I take into consideration responses from the second phase of the original study, so the questionnaires conducted after 52 weeks in Continuous Glucose Monitoring use, because that describes the situation of the satisfaction towards the physician when the new treatment method has been in use. The sample, 223 respondents, are individuals who have filled the questionnaire after using a Continuous Glucose Monitoring, or in the case of young children being the diabetic, their primary caregiver (most often father or mother) can have filled the questionnaire for them. Data sets used in this paper are Summary dataset, CGM Satisfaction Scale, and Problem Areas with Diabetes (adult subject version).
Before the start of the research, informed consent was gathered, and for younger participants another form, Child Assent Form, was given to read or to read with the child. The study included children, and that was taken into consideration when study procedures were discussed. All information was given for the subjects themselves, for the children on a level in which they can understand, and also to their primary caregivers (Beck & Ruedy, 2008). In this paper, the variable educational level of the primary caregiver in most of the cases is the diabetic him/herself, but in case of a child, the person filling the form (eg the information about the satisfaction towards the diabetes physician) can be for example mother or father. The variables age and gender are referring to the age and gender of the diabetic.
Next I explain how the data was collected. The original study by Jaeb Research Center divided subjects into two groups in the beginning of the study; the first group had Continuous Glucose Monitor in use the whole 52 weeks, and the control group used traditional Home Glucose Meter for the first 6 months (26 weeks) and for another 6 months they used the Continuous Glucose Monitoring. Jaeb Center for Health Research provides the original questionnaires, which participants were asked to fill out in different phases to gather the data. The subjects had planned visits to the clinical center, and phone contacts with researchers in between protocol visits. Groups had follow-up visits on weeks 1, 4, 8, 13, 19, and 26. Between every visit, all participants had one phone contact. The original study was conducted in two phases; in the first phase, which lasted 6-months, half of the sample used home glucose meter and another half used Continuous Glucose Monitoring. Both of the groups have the same number of contacts; visits and phone contacts. A second phase, also lasting 6 months, both of the groups were using Continuous Glucose Monitoring. Subjects of the study were randomly assigned to one of these groups. All participants did not answer all questionnaires, for example, the control group using Home Glucose Meter did not answer the questionnaire about using Continuous Glucose Meter after the phase 1 since they did not have it in use. Hence, different data sets have different sample sizes. Also, the subjects were divided into two groups based on their HbA1C level, the primary study being subjects whose HbA1c level was less than 7.0%. The secondary study included subjects whose HbA1c level was higher than 7.0%. In the final database, the data collected from both of these HbA1C clusters are included.
Variables
(Table 1: Summary of the variables used in the analysis)
Variable Classification Frequency (%/absolute
number)
Mean Median Standard deviation
Size of the sample
Feeling unsatisfied with your diabetes physician Satisfied Not satisfied 214 9 223 0 Educational level of the primary caregiver Bachelors Masters Professional Associates Other 44.4/99 21.5/48 8.1/18 13.0/29 13.0/29 223 0 Gender Men Women 42.6/95 57.4/128 223 0 Age 18 thru 72 36.96 37.00 14.125 223 0 Race White
More than one race Black/African American Asian American Indian/Alaskan native 96.0/214 1.8/4 1.3/3 0.4/1 0.4/1 223 0 White
More than one race Black/African American Asian American Indian/Alaskan native 96.0/214 1.8/4 1.3/3 0.4/1 0.4/1 223 0
The dependent variable of the model is ‘Feeling unsatisfied with your diabetes physician’. Because of the very skewed nature of the variable, it was recoded into binary “satisfied” and “non-satisfied” clusters. Because of the very skewed nature of the variable, firstly the variable was recoded in a way that neutral answers were missings, but to create more variation to the variable, also neutral answers were counted in and the clusters were named ‘satisfied’ and ‘not satisfied’, legitimizing the decision to count also neutral answers as not satisfied. The original variable before decoding had a mean of 0.18 and a median of 0, which indicates, that after using Continuous Glucose Monitoring, most people are not unsatisfied with their diabetes physician. Most of the subjects describe themselves as strongly disagreeing or disagreeing. Only
9 people say to agree or strongly agree, or have a neutral opinion. It is possible to conclude that the variable that measures the level of unsatisfaction with your diabetes physician is strongly asymmetrical, because most of the subjects chose the lower values of this scale. The standard deviance of this variable is 0,606. This also makes the decision of analysis strategy to point to the direction of logistic regression analysis, when the dependent variable is strongly skewed and recoded into binary variable – satisfied or not satisfied.
The first independent variable, ‘Educational level of the primary caregiver’ shows that majority of the subjects (44.4%) have a Bachelor’s degree. Caregiver in this context means the person who is in charge of the treatment of diabetes, in most of the cases diabetic him/herself, but in some cases (children), also their (often) mother or father can be the primary person in charge of the diabetes management - to be in charge of controlling the blood glucose level and injecting the correct doses of insulin. Therefore, in the respondents, there is one answer per diabetic, mostly from themselves but also in case of children, the person’s who is in charge of the diabetes management. Master’s degrees have obtained 21.5% of the respondents. This variable was recoded into dummies, and ‘Bachelors’ is used as the reference category, because that is a group with the most respondents.
The variable ‘gender’ has a nominal measurement level and consists of two categories: men and women. In the analysis, ‘men’ is used as a reference category. In the sample, both sexes are almost equally represented, 57.4% of the sample being women and 42.6% being men. All subjects in the sample answered the question with the binary understanding of ‘gender’. Gender refers to the gender of the diabetic, regardless of if the primary caregiver would be someone else that the subject and filling the questionnaire for the diabetic.
The last independent variable in the analysis is age. The youngest respondent is 18 years and the oldest is 72 years old. The mean of this variable is 36.96 and the median age, the age of the person ‘in the middle of the line’ is 37. The standard deviance is 14.125. Age refers to the age of the diabetic, regardless if the primary caregiver would be someone else and filling the questionnaire for the diabetic.
A variable ‘race’ was part of the original conceptual model of the analysis, and would have been an interesting variable to add to the model, but after scrutinizing the frequencies from the data set, it became clear that the variable race lacked variation, 96% of the respondents being white. Therefore, the variable race was omitted from the analysis.
(Original SPSS output tables can be found from the annex) Analysis strategy
The analysis was conducted by using the IBM SPSS Statistics -program, version 26. A binary logistic regression analysis was conducted to scrutinize variables age, gender, and educational level of a primary caregiver. First, ordinal least regression analysis was run, but because of the very skewed nature of the dependent variable, the assumptions were not fulfilled in a satisfactory manner and the analysis strategy had to be changed. Also, at the same time, the variable ‘race’ was omitted from the analysis since the variation was almost non-existent when 96% of the respondents were white.
In the analysis, the demographic features are scrutinized, if these features of a patient can help to predict if they feel satisfied or not with their diabetes physician, after using the new diabetes management device. For this type of research question, logistic regression analysis is the right tool to scrutinize the statistical significance of the model, because it scrutinized the relationship between in this case three predictor variables and a binary dependent variable. I’m addressing the question of what is the probability that a given case falls into one of two categories on the dependent variable, given the predictors in the model.
Because I wanted to examine the result after the Continuous Glucose Monitoring has been in use, in the logistic regression analysis I used the “phase 2” answers, therefore, the questionnaires filled after the subjects have used the Continuous Glucose Monitoring either 52 weeks or 26 weeks, depending on were they in the beginning in the study in a control group or in the examination group.
For logistic regression analysis, the assumption tests are run and checked if they are violated. It can be asked, why the logistic regression analysis is used, and not ordinary least squares regression. In the ordinary least squares regression, a linear relationship between the independent variables and the dependent variables is assumed, also the residuals should be normally distributes, and the residuals should show constant variance. In the case of very skewed dependent variable, these assumptions are violated. Binary logistic regression takes into account that probabilities are bounded to 0 and 1, and binary logistic regression analysis does not assume normally distributed residuals, neither that they would exhibit constant variance (Pituch & Stevens, 2016).
Logistic regression is often tought of as having no assumptions, since it does not make any assumptions about the distributions of the independent variables, but in order to conduct an adequate binary logistic regression analysis, there are assumptions that have to be tested and passed. Firstly, the dependent variable has to be measured on a dichotomous scale, which in this case it is, since the dependent variable shows if the subject is satisfied or unsatisfied with the diabetes physician. The second assumption is that there has to be one or more independent variables, continuous or catecorigal ones. In this care, these independent variables are gender, age and an educational level of the caregiver, therefore the assumption is fulfilled. Third assumption is independence of observations, meaning that each subject is represented
