THE MEDICAL ACADEMY’S FACULTY OF MEDICINE OF THE LITHUANIAN UNIVERSITY OF HEALTH SCIENCES
Alan Bareiss
Quality of Patient Education for Patients with Diabetes Mellitus Type I and their Parents in Lithuania and the Multidirectional Impact of Successful Patient Education
Master Thesis
Department of Health Psychology and Endocrinology Supervised by Jolanta Žilinskienė
TABLE OF CONTENTS
TABLE OF CONTENTS 1
SUMMARY 2
ACKNOWLEDGMENTS 2
CONFLICTS OF INTEREST 3
PERMISSION ISSUED BY THE ETHICS COMMITTEE 3
ABBREVIATIONS 3
TERMS 3
INTRODUCTION 3
AIM AND OBJECTIVES 5
LITERATURE REVIEW 6
RESEARCH METHODOLOGY AND METHODS 13
RESULTS 16
DISCUSSION OF THE RESULTS 37
CONCLUSIONS 39
PRACTICAL RECOMMENDATIONS 39
SUMMARY
Author(s). Alan Bareiss and Jolanta Žilinskienė examined the Quality of Patient Education for Patients with Diabetes Mellitus Type I and their Parents in Lithuania and the Multidirectional Impact of Successful Patient Education. We were aiming to make the situation of quality of patient education in this particular field measurable, to elaborate specific challenges patients and their parents face and relate them to different factors in order to gain a better insight into what predicts treatment adherence, to compare the outcome of the survey to other countries and to make practical suggestions what is to improve.
Methodology. 55 Participants comprising each a pair of one parent and one child suffering from type 1 diabetes mellitus. The Coping-Self-Efficacy-Scale has been used to determine the child's ability to cope. The WE-CARE-questionnaire has been used to elaborate challenges of the parents. In addition objective data has been collected about duration of suffering, last measured HbA1c level as well as patient and parent age.
Results. There were surprisingly little correlations of nearly all items of both questionnaires and the influence on HbA1c level. Yet the extracted correlations suggest the best best intervention point in patient education to improve coping of the patients is enhancing problem solving skills and their ability to make a plan of action and follow it when confronted with a problem and the ability to keep from feeling sad. Training healthcare providers in life coaching basics and sensitizing them to the importance of coping skills can improve the patients experience. Another important finding is that a good HbA1c does not mean good coping skills. When looking at the ability to seek emotional help quite the contrary is case. This finding suggests that psychological assistance should be offered regardless of glycemic control.
ACKNOWLEDGMENTS
I would like to thank my supervisor Jolanta Žilinskienė who dedicated substantial time to inspire the research, to contribute with constructive comments and to overcome stepping stones along the way of creating this thesis. I will also remain forever grateful for the support by loved ones who were supportive more than just during stressful periods of conducting this research.
CONFLICTS OF INTEREST
The author reports no conflicts of interest.PERMISSION ISSUED BY THE ETHICS COMMITTEE
The permission issued by the ethics committee was granted on the 19th November 2018 with the document number BEC-MF90. The permission was signed by Eimantas Peičius.
ABBREVIATIONS
T1DM - Type 1 Diabetes MellitusCSII - continuous subcutaneous insulin infusion FOH - fear of hypoglycemia
HFS - hypoglycemia fear survey
FIIT - flexible intensive insulin therapy / basal bolus therapy regimen
TERMS
HbA1c - glycosylated hemoglobin in percent / can be viewed as an indicator for the glycemia level of the last three months
INTRODUCTION
Type 1 Diabetes mellitus (T1DM) is a chronic condition based on inadequate production of insulin because of autoimmune destruction of β-cells. Its onset is usually in childhood or adolescence which gave the condition its earlier name of juvenile diabetes mellitus. From a current standpoint of research a combination of genetical susceptibility and an environmental trigger builds the etiological background. For young people and their caregivers a diagnosis with T1DM and its life-long treatment along with all the complications and lifestyle changes presents a real challenge. Serious complications and long term consequences arise from inadequate management of the patients blood sugar levels. Reaching blood glucose concentration targets is inevitably linked to treatment adherence and health behaviour. This in turn can be crucially influenced by well structured and well executed patient education, which consists of far more than an initial lecture and an occasional visit at the family doctor to adjust the insulin regimen. Successful treatment of diabetes mellitus type 1 consists of many aspects
of which each individually should be kept up to date with newest insights and adjusted to the patients individual need as well as tailored to patients and parents ability to understand. On the other hand it is crucial to make all elements interplay for long term success.
Successful T1DM treatment is patient and parent centered while the adolescent shift in responsibility from the parent to the patient has to be kept in mind and ideally monitored by a trained psychologist. Patients diagnosed with T1DM enter a lifelong journey on which collaboration with the multidisciplinary team of clinical staff is key. The goals of such a collaborative treatment concept are (i) prevent diabetic ketoacidosis; (ii) prevent severe hypoglycemia; (iii) maintain normal growth and development; and (iv) prevent long-term diabetic complications [1]. A special challenge in establishing a working therapy relationship is the shift of parents taking care of their children's diseases without much involvement of the patients themself towards an adolescent patient who takes more and more responsibilities by him or herself. Simply going by age does not adequately reflect the stage of independence and competence in respect to individual development and family dynamics.
Another reason why patient education is key for reaching the aforementioned goals is that the subjective experience of the patient who is going through a variable degree of blood sugar oscillations does not always correlate well with simple logical thinking without medical background knowledge. An example of this would be the negative downward spiral of “sensing” hypoglycemias. In scientific terms this would be the hypoglycemia-associated autonomic failure where in the first step periods of insulin excess (therapeutic or relative) without the counterbalancing acting of glucagon hypoglycemia is established. These episodes then cause downregulation of the autonomic response including adrenomedullary. Ultimately this leads to reduced symptoms of hypoglycemia and and even less effective counterbalancing system [2]. Educating the patient appropriately can drastically interfere which such feedback loops.
In an acute setting hypoglycemia can be lethal. In a long term setting the fear of hypoglycemia is especially in young patients a limiting factor to follow the insulin scheme and achieve physiological levels of blood sugar. This is yet another point of intervention of a multidisciplinary team to work on with the patients [3].
not in tightly supervised clinical settings the most important part is to ensure the patents fully understands the handling of his or her diabetes including factors such as how to correctly apply insulin, when to do it, what influences blood sugar level to what degree and how to prevent the dangerous complication of hypoglycemia or ketoacidosis. Hence this work focuses on the patient education that adolescents in the age of 12-17 years and their parents receive in the hospital of LSMU in Kaunas.
AIM AND OBJECTIVES
● To make the situation of quality of patient education in this particular field measurable.
● To elaborate specific challenges patients and their parents face and relate them to different factors in order to gain a better insight into what predicts treatment adherence.
● To compare the outcome of the survey to other countries. ● To make practical suggestions what is to improve.
LITERATURE REVIEW
At first I would like to focus on the research that has been conducted in Lithuania in order to outline the current situation in the country.
2016 Lašaitė et al. from the institute of endocrinology at LSMU Kaunas were looking at the Diabetes distress in adult type 1 diabetes mellitus men and women with disease onset in childhood and in adulthood [4]. They contacted 700 randomly selected patients from the diabetes registry and got 214 replies. The level of distress was evaluated according to the emotional burden, physician-related distress, regimen-related distress and interpersonal distress. They then compared the levels of distress of the group that had their first manifestation of T1DM at adult age to the group which had it before adulthood and found that adult childhood-onset T1DM women have higher regimen-related distress than adulthood-onset women. They did not find the same to be true for men and explained this with a difference in coping strategies and willingness to admit to distress. Nonetheless this study points out the relevance of a well-established multidisciplinary team that supports the patient in coping with the T1DM diagnosis from early on. The same author conducted a research about Diabetes distress in males and females with type 1 diabetes in adolescence and emerging adulthood [5]. They were looking at 255 adolescents and 283 emerging adults with T1DM using the Problem Areas in Diabetes scale. Their results show the importance of addressing diabetes distress in clinical care and the necessity of wider picture beyond the physical manifestation of diabetes to be taken into consideration. To be more precise their result of the survey was that high diabetes distress level was found in 22.8% of participants. Lack of confidence in self-care, negative emotional consequences and overall score were higher in adult than in adolescent males and females. In the discussion of these outcomes the author mentions that in other studies, analyzing a wider age range, significantly higher diabetes distress was observed in younger than in older adults such as Fisher et al. conducted in 2015 [5].
Another interesting investigation relevant to the topic of this thesis has been performed by Jolanta Žilinskienė who looked at a population of 260 adolescents (100 healthy, 160 with T1DM) to examine the characteristics of depressiveness among adolescents with diabetes mellitus in 2007 [6]. Their results confirmed that depressiveness of 14- and 16-year-old girls with diabetes and that of their mothers were related. The relationship between the depressiveness of mothers of 14-year-old boys with diabetes mellitus and bad glycemic control of their sons was determined. Furthermore they concluded that 14-year-old girls had more negative attitude toward themselves; 16-year-old girls were very sensitive and had depressed mood; 16-year-old boys complained about lowered daily activity.
This points out the individuality of challenges faced by young patients and their parents when confronted with having to deal with T1DM in their life.
2.Quality of Patient Education in Lithuania
In Lithuania the quality of diabetes care at the largest outpatient clinics in Vilnius has been assessed in 2016 by Visockienė et al. [7]. They performed a retrospective study using data from the largest 5 outpatient clinics in Vilnius from 2012 to 2013 of both T1DM and T2DM and compared the finding with the guidelines for monitoring of DM complications. The analysis revealed good glycaemic control in T2DM, but insufficient in T1DM. Continuous monitoring of diabetes complications and cardiovascular risk factors did not meet the local Diabetes Care Guidelines. However the study is only partially comparable to what we are trying to evaluate as is looks also at T2DM as well as it focuses on hard facts such as annual screening for diabetic foot, retinopathy, nephropathy, renal function and lipids. Nonetheless it found that the prevalence of nephropathy, polyneuropathy, retinopathy, and angiopathy was higher in T1DM. Underlying this finding is a generally higher HbA1c in T2DM patients than in T1DM. The difference becomes very obvious when looking at the treatment modality. T2DM can be treated with consistent improvement of diet and lifestyle, which has reached the lowest HbA1c in the study. When treated with oral antidiabetics the outcome was already worse. T1DM patients rely on complex insulin regimens as well as dietary changes, tracking habits and health behaviour that has to be learned and followed consequently leading to a more difficult treatment and a worse HbA1c
1.Incidence of type 1 diabetes mellitus in Lithuania
R. Ostrauskas from the Institute of Endocrinology at the Lithuanian University of Health Sciences has collected and published data about the prevalence of T1DM in Lithuania in 2015 [8]. In particular he was looking at the prevalence of type 1 diabetes mellitus among 15 to 34 year old Lithuania inhabitants from 1991 to 2010. After statistical work-up he concluded that the prevalence of type 1 diabetes in the given interval of age had a tendency to increase in this period. To put these numbers in a bit more of a perspective it has to be said that Lithuania is a low-incidence country with an incidence of 14.2 / 100.000 from 2004-2008 [9] increasing at a yearly estimated rate of increase of incidence of 5.5% since 1989. This allows the conclusion that the topic has been increasing in relevance and probably will continue to do so. It also leads to a question that should not be underestimated. What are the consequences of this low incidence compared to a high incidence country such as Sweden with 35.1 / 100.000 from 2004-2008 [9]? Sadauskaite-Kuehne from the Laboratory of Paediatric
Endocrinology together among others with Žilvinas Padaiga examined the severity at onset of childhood type 1 diabetes in countries with high and low incidence of the condition in 2002 and compared the south-east of Sweden with the country of Lithuania. They were looking at new cases of T1DM of children from 0 to 15 years old from 1996 to 1999. They found that Lithuanian children were diagnosed in a more severe condition, mean pH 7.30 and more Lithuanian than Swedish children were diagnosed in ketoacidosis (pH < or = 7.2, hyperglycemia and ketonuria). 21.3 versus 7.3% HbA1c 11.5% compared with mean pH 7.36 and HbA1c 9.7% in Swedish children [10].
Shalitin claims that hypoglycemia is one of the most common and acute complications of insulin therapy. It can lead to uncomfortable counter-regulatory symptoms including headaches, shakiness, nervousness, sweating, irritability, confusion, sleepiness and fatigue, weakness, dizziness, and dangerous neuroglycopenia [11]. In the most extreme cases, seizures, loss of consciousness, and death may occur. For many individuals, acute complications including fear of hypoglycemia preclude them from optimal diabetes management. While some degree of fear is considered appropriate and adaptive given the potential danger of hypoglycemia, for some individuals it may become more extreme and problematic. For these individuals, fear of hypoglycaemia (FOH) may result in increased anxiety about diabetes management, obsessive self-monitoring, deliberately keeping blood glucose levels too high, dependence on others, feelings of guilt and frustration, a sense of loss of control, embarrassment, relationship stress and avoidant behavior [11]. This puts another aspect on the table. A key to reaching the treatment goals of DMT1 is addressing stepping stones like FOH early on and continuously. The FOH in the field of research is most commonly evaluated with the hypoglycemia fear survey (HFS) [37]. Interestingly Gonder-Frederick, Fisher, et al. found in 2006 that adolescent FOH and parent FOH not correlated. For adolescents, higher trait anxiety and more frequent episodes of severe hypoglycemia predicted higher FOH; for parents, whether adolescent carried rescue carbohydrates predicted lower parent FOH. However it has to be mentioned that a review by Driscoll et al. in 2016 of 16 studies from different countries, which have examined the association between FOH and glycemic control in children, shows controversial evidence [38]. They have found that the majority of these studies demonstrate no association. However, a small number of studies have found a positive association between parents’ FOH and children’s glycemic control. An example would be Haugstvedt, Wentzel-Larsen, et al. (2010) who found that higher HFS score of parents worry related to higher A1C and higher frequency of hypoglycemic events [12]. This in turn substantializes the impact of good patient education for both the patient and the parents characterized by ability to address the individual issues at the heart of their concerns.
Psychological Aspects
Psychologic distress is not just intuitively linked to worse treatment outcomes. It has been examined thoroughly how distress is linked to worse glycemic control and thereby to premature mortality and morbidity of T1DM patients. Hislop et al. from the School of Medicine and Pharmacology in Western Australia [13] who looked at ninety -two participants with type 1 diabetes of early adult age in 2008 found that approximately one-third of young adults with type 1 diabetes experience psychological distress, which is associated with poorer glycaemic control. On a side note it has to be mentioned that significant greater distress in this study has been reported by users of a CSII. A similar finding was reported just this year in February by Stahl.Pehe et al who looked at cross -sectional relationships between diabetes distress and health-related variables, and prospective associations between diabetes distress and future glycaemic control (HbA1c) and health status among young adults with early-onset Type 1 diabetes. Their results suggest that diabetes distress impairs health-related outcomes in young adults with early-onset diabetes [14]. Both of the researches listed here underline the importance of attempting to reduce the distress of patients, which can be achieved by patient centered collaborative design of treatment regimen as well as competent psychological support.
Especially in teenagers there are plenty aspects that complicate a good treatment adherence. Not just that the treatment is complex but in addition heightened concerns about social context and peers, premature shift in responsibility for management from parents to teens, developmental inclination towards risk taking, incomplete knowledge and understanding of treatment regimens and future health risks, fatigue from care of a chronic illness (‘diabetes burnout’), and physiological changes that lead to greater insulin resistance during puberty. This is how Borus and Laffel sum up the factors in their article about treatment adherence in adolescent patients with T1DM from 2010 [15]. Special attention in this context deserves the shift of responsibility from the caregiver to the patient as he or she ages and develops. Vesco et al have performed a cross sectional study in 2010 examining what way is the best to perform this process [16]. There is plenty of constellations of responsibility sharing that are somewhat less than optimal. For example a parent (“caregiver”) who takes full responsibility for treatment adherence until late adolescence. Onset of puberty and upcoming conflicts can then make a sudden shift of responsibility necessary which then inevitably leads to poorer outcome as the young patient is overwhelmed. If patients provoke the shift too early on the other hand, poor glycemic control is also recorded. Vesco et al suggest what they call a “healthy sharing of responsibilities” and point out that a crucial factor in succeeding with the shift is that the adolescent at any time knows clearly what is his or her responsibility and what is the one of the caregiver. For a better understanding: Tasks include
glucose monitoring, journaling, interpreting results, arranging medical appointments, getting supplies,carrying supplies and many more.
Medical Aspects
Another phenomenon of interest raising evidence to support the relevance of establishing good glycemic control especially in the early pediatric years of T1DM manifestation has become known as “metabolic memory” which can be described briefly by diabetic vascular stresses persisting after glucose normalization. It may be part of the reason why mortality of T1DM is still high even when euglycemia is reached [17]. Its effect is summarized well by Ceriello et al. in their second clinical review about this topic in 2009 [18]: The emergence of this metabolic memory suggests the need for early aggressive treatment aiming to "normalize" metabolic control together perhaps with the addition of agents which reduce cellular reactive species and glycation in order to minimize long-term diabetic complications. Combined with what we have established so far as necessary elements for effective glycemic control this also supports the importance of a well-working healthcare system, including patient education.
Structured education programmes (SEPs) are in use all around the globe for children, teenagers / adolescents and adults with T1DM. The National Institute of Clinical Excellence describes the following characteristics of such a SEP as “comprehensive in scope, flexible in content, responsive to an individual's clinical and psychological needs and adaptable to his or her educational and cultural background” [19]. It has to be mentioned that this old guideline from 2003 has been replaced by a 2015 version from which the following information is extracted [20]. A model of patient education for T1DM patients that has proven effective and is used most widely is the DAFNE model. It stands for Dose Adjusting For Normal Eating. DAFNE is a SEP for people with Type 1 diabetes delivered over 5 consecutive days by a multidisciplinary team. The program covers all aspects of living with diabetes but places a strong emphasis on blood glucose testing, carbohydrate counting and matching quick-acting insulin to this. Walker et al. among others have demonstrated the positive outcome of DAFNE in 2018 in Scotland [21]. The results they got confirmed that DAFNE participation improves glycaemic control in T1DM with benefits being sustained for 5 years. Interestingly enough this study claims the first to demonstrate reduced HbA1c variability after completion of structured education. However different literature states that the good result after initial improving of psychological markers as well as HbA1c decreases over time. So does the research by Heller from the Sheffield Teaching Hospitals NHS Foundation Trust, Sheffield, UK et al. in 2014[39]. They conclude that glycaemic
outcomes are not always improved or sustained when the DAFNE programme is delivered routinely, although improvements in psychosocial outcomes are maintained. As practical conclusion on how to counteract the decrease in effect of this SEP after some time they suggest that continuous DAFNE courses and follow-up support are needed to help participants instil and habituate key self-management practices such as regular diary/record keeping. Some research has been trying to look behind the reasons for the decline in effectiveness over time. Rankin et al. from the University of Edinburgh published an article about the experiences of using blood glucose targets when following an intensive insulin regimen [22]. They found that patients are motivated to stick to their regimen if it leads to reaching the blood glucose targets. Over time however patients developed a tendency to simplify or adapt targets over time probably in order to not be demotivated by not reaching the target. Another reason they named is to minimize the feeling of failure by not reaching the target. They recommend that blood glucose targets should be regularly revisited during clinical reviews and revised targets agreed to accommodate patients’ concerns and difficulties.
Just last year in 2018 the issue of decreasing effectiveness of self-management has been addressed by a comprehensive systematic review written by Campbell et al. [23] under the name of the FUSED model (Follow-Up Support for Effective type 1 Diabetes self-management). They looked at 18 papers from six studies and abstracted ten recommendations from the included papers to provide a logic model for a programme of individualised and responsive follow-up support. The challenges the patients face have been summarized. For example the complexity of life can get in the way of sticking to the learned skills of self-management as it has been taught in the classroom. Eating out spontaneously can be named as an example of that [24]. Disconnection between effort and reward is another challenge that patients face, which I have described earlier. Some patients and parents of them face a lack of confidence in their own judgement. Murphy et al. described that many participants are not sure about their ability to correctly interpret blood glucose changes over time to a degree that it allows them to adjust meal sizes or adjustments in insulin dosing [25]. Reluctance and hesitation to reach out to a physician or a SEP-educator is yet another challenge faced by patients. Rankin et al. have found that often patients expressed a reluctance to initiate contact with clinicians/educators in-between scheduled appointments: “While all patients had access to course educators’ telephone and email contact details, some found that calls or requests for advice left on answer-machines were not returned. This, as a participant explained, was off-putting and led to his decision to only contact educators with “really important issues” because “I don’t want to bother them unless I feel that I need to” More often, patients expressed a reluctance to initiate contact with clinicians/educators in-between scheduled appointments (e.g. clinical reviews), despite having questions and concerns.” [26]. In the same
investigation by Rankin et al. another issue is addressed: Patients (or parent respectively if they are mainly responsible for glycemia control) shift their blood glucose targets upward consciously or unconsciously making them more easily achievable. FOH as described earlier is another motivation to shift blood glucose levels upward in order to avoid the traumatizing results of a potential hypoglycemia. Lawton et al. noted a lack in motivation to take time for documentation of blood glucose variation in the long term [40]. Some patients after initial motivation find it burdensome to keep keen track of their reedings. Another effect related to this burden of documentation and described by the same study is that patients overly rely on corrective factors instead of adjusting the background insulin upon reflexion about the development of blood sugar level ultimately making them fail to achieve their targets. The FUSE-model based upon their findings of challenges suggests (i) modelling collaboration and empowerment - for example not just reviewing the past glucose monitorings and making adjustments accordingly but rather set up a collaborative environment for problem-solving on an individual level to fit T1DM management into the complexity of day-to-day life. (ii) Preparing for and addressing motivational issues - this might look like giving qualified motivational support for burdensome tasks such as keeping a record of blood glucose monitoring. Cognitive behavioural therapy (CBT) is also mentioned in this context. (iii) exploring and facilitating social support - staying sensitive to the patients preferences. (iv) supporting the use of technology - this can reduce the burden of keeping a record drastically. (v) educating mainstream health professionals - this is to make sure that non-physicians are well informed about the FIIT (flexible intensive insulin therapy / basal bolus therapy regimen) scheme and are able to give specific advice. (vi) building knowledge and skill over time - this suggests to use tie for a follow-up meeting not just to treat the documented values but rather examine skill-gaps and individual challenges. (vii) considering and revising routines and life circumstances - this is to get feedback on how the life changed after establishing a new treatment routine of T1DM. (viii) reviewing and revising blood glucose monitoring and treatment practices - again in a collaborative way in close communication with the patient or caregiver, rather then just changing basal insulin dose, correction doses and then releasing to patient to go on by himself. (ix) reviewing and revising hypoglycemia management - the FUSE model emphasises that this should be done for all participants regardless of if they show signs of hypoglycemia unawareness or not. (x) providing dietary advice- follow-up sessions need to explore what has changes in the diet and if the participant is restricting certain foods. They should also identify difficulties in carbohydrate counting. Upon this base the authors have developed what is called DAFNE plus which is being piloted for implementation in three sites which are not further specified. Limitations of this program are that the experiences reviewed only come from the UK and Southern Ireland and might not necessarily be as
applicable the same degree to other countries. Which is ultimately why this research will shed light on the situation in Lithuania.
While DAFNE and DAFNEplus are being implemented or piloted with good results, there are also approaches that are especially adapted for children. The Kids In Control OF Food (KICk-OFF) course for example is based on DAFNE principles and aims to provide young people with self-management skills and strategies to help overcome some of the barriers to effective self-management associated with an intensive insulin regimen. Price et al developed and piloted this model in an age group from 11-16 because as they claim “although structured education courses have been delivered to children in Germany for many years, there have been no randomised controlled trials of a DAFNE-type intervention in children” [27]. Their result of the teenager-adapted SEP did not have a significantly better HbA1c in most patients compared to a standard DAFNE SEP but they were able to work out factors that need special attention when education teens. These are the involvement of parents and parent–child communication, support of friends without diabetes, creating a feeling of being like everyone else and social support from other young people with diabetes.
RESEARCH METHODOLOGY AND METHODS
Prior to starting the investigations the authors of the questionnaires used were contacted and granted permission for use in the context of this final master thesis. The two questionnaires that were used are the “Coping Self Efficacy Scale” (CSE) and the WEll-being and Satisfaction of CAREgivers of Children with Diabetes Questionnaire (WE-CARE) by Quattrin. The CSE has been chosen not just because of its proven validity but also because its items cover emotion-focused coping as well as problem-focused coping. The WE-CARE questionnaire has been chosen because it covers psychosocial well-being and treatment satisfaction of parents who have a child with type 1 diabetes and therefore is a perfect fit for the planned investigation. The questionnaires have then been professionally translated into Lithuanian by an office in Vilnius and confirmed by back translation with several native speakers.
The design of the study is innovative as two validated and commonly used questionnaires were handed to one child or adolescent (CSE) and one corresponding parent (WE-CARE). In addition to the questionnaires themselve compact objective data has been collected, namely the age of the child, the age of the parent filling the questionnaire, the last measured HbA1c acting as indicator for treatment effectiveness and the period since diagnose. The CSE is a 26-item likert scale type questionnaire with values from 0 - “can not do at all” to 10 - “ Certain can do”. The WE-CARE questionnaire is a
originally 77-item likert scale as well as dichotomous answers type questionnaire, of which only the first 37 items were applicable as the latter 40 questions concerned or involved inhaledinsulin, which is not in use in the Lithuanian health care service.
All items have been run through principal component analysis for extraction of factors. Then the Varimax rotation method with Kaiser Normalization clustered the items accordingly. They were given a descriptive term to assist in analysis. Table 1.1 presents the factors and included items for the CSE scale, Table 1.2 the ones for the WE-CARE-questionnaire.
Table 1
Rotated Component Matrix - Factors of the CSE-Scale
Factor Items
Mental_Internal_Coping_Strate gies
Ability to make unpleasant thoughts go away
Ability to do something positive for yourself when you are feeling discouraged
Ability to keep yourself from feeling lonely.
Ability to take your mind off unpleasant thoughts / Stop unpleasant thoughts.
Ability to visualize a pleasant activity or place
Ability to try other solutions to your problems if your first solutions don’t work
Ability to develop new hobbies or recreations
Ability to stop yourself from being upset by unpleasant thoughts Ability to sort out what can be changed, and what can not be changed Ability to think about one part of the problem at a time.
Seeking_Emotional_Support
Ability to get emotional support from friends and family
Ability to see things from the other person's point of view during a heated argument
Ability to get friends to help you with the things you need Ability to leave options open when things get stressful
Problem_solving_capacity
Ability to find solutions to your most difficult problems Ability to look for something good in a negative situation Ability to keep from getting down in the dumps
Ability to break an upsetting problem down into smaller parts
Emotional_strength
Ability to talk positively to yourself.
Ability to make a plan of action and follow it when confronted with a problem.
Ability to keep from feeling sad
Ability_to_seek_Support_fro m _community_and spirituality
Ability to get emotional support from community organizations or resources
Ability to pray or meditate Get emotional support from community organizations or resources
Table 1.2
Rotated Component Matrix - Factors of the WE-CARE questionnaire
Factor Items
Burden_on_Family
The impact of your child's T1DM on your sexual life
The impact of your child's T1DM on your marriage/partnership The burden of my child's T1DM on my marriage/partnership
The impact of my child's T1DM on my relationship with your children The burden of storing insulin
Social/work_Burden
The care for my child made you spend less time at work The care for my child interrupted your work
The care for my child interrupted your social activities
The care for my child had an impact on my leisure time activities
The care for my child had an impact on my work (job) situation Made you spend less time with your other children or family members
Satisfaction_with_Insulin
I want my child to continue using the current insulin regime I would recommend the current insulin regimen to others My child is compliant with the current insulin regime I find the time it takes for each dosing acceptable
Self_Administration_of_Insuli n
How much of a burden is administering insulin prior to meals? How much of a burden is administering insulin in public places? How much of a burden is administering insulin at home?
Overall how satisfied are you with the insulin Tx?
How much of a burden is having to administer insulin to your child?
Burden_of_care
In the last 4 weeks I felt depressed.
In the last 4 weeks the burden of care was overwhelming. In the last 4 weeks I worried about the complications of T1DM. Preparing insulin for administration is a burden for me .
In the last 4 weeks I got frustrated a lot.
Insulin_Adminstration Difficulties
Difficulty to use insulin
Difficulty to prepare insulin dose Difficulty to carry insulin
Difficulty to carry supplies
Insulin_Flexibilities
I am flexible in planning social activities.
I am flexible in my daily activities due to insulin regime. The mealtimes are flexible due to insulin regime.
The data was collected in the outpatient clinic as well as the inpatient clinic at Kaunas University Hospital.
There were 55 participants in this sample. Participants first provided informed consent about participation in this research, confirming this with date and signature on a seperate sheet, keeping the
data collection anonymous. They were furthermore provided with the contact information of the author of this research.
The sample was heterogeneous with respect to age, gender of both the parents and children, HbA1c levels and duration of suffering. This is due to the natural randomization effect of asking every possible patient who matched the inclusion criteria: (i) the patient is suffering from type 1 diabetes mellitus (ii) the patient is 12-17 years old (iii) the patient arrives at the location of research conduction with at least one of his or her parents. The interval of age for the inclusion criteria has been decided upon the requirement of being able to understand the CSE items and has most likely tried to apply most of the items.
RESULTS
Descriptive StatisticsIntroduction. Summary statistics were calculated for each interval and ratio variable. Frequencies and percentages were calculated for each nominal variable.
Frequencies and Percentages. The most frequently observed categories of HbA1c were in target (<7.5%), above target (7.6%-8.9%) and critically elevated (≥ 9%), each with an observed frequency of 17 (33%). Frequencies and percentages are presented in Table 2.
Table 2
Frequency Table for Nominal Variables
Variable n % Cumulative %
Last HbA1c
above target (7.6%-8.9%) 17 33.33 33.33 critically elevated (≥ 9%) 17 33.33 66.67
in target (<7.5%) 17 33.33 100
Note. Due to rounding errors, percentages may not equal 100%.
Summary Statistics. The observations for HbA1c had an average of 8.43 (SD = 2.01). The observations for Parent_Age had an average of 41.78 ( SD = 5.73). The observations for Patient_Age had an average of 14.61 (SD = 1.61). The summary statistics can be found in Table 3.
Table 3
Summary Statistics Table for Interval and Ratio Variables
Variable M SD n
HbA1c 8.43 2.01 51
Parent_Age 41.78 5.73 54
Patient_Age 14.61 1.61 54
Note. '-' denotes the sample size is too small to calculate statistic.
Pearson Correlation Analysis
Introduction. A Pearson correlation analysis was conducted between HbA1c and the overall score on the Coping Self Efficacy Scale (CSE). Cohen's standard was used to evaluate the strength of the relationship, where coefficients between .10 and .29 represent a small effect size, coefficients between .30 and .49 represent a moderate effect size, and coefficients above .50 indicate a large effect size [28].
Results. The correlations were examined based on an alpha value of 0.01. There were no significant correlations between any pairs of variables. Table 4 presents the results of the correlation.
Table 4
Pearson Correlation Results Between HbA1c and CSE
Combination rp p
HbA1c-CSE -0.16 .329
Note. The confidence intervals were computed using α = 0.01; n = 40
Spearman Correlation Analysis
Introduction. A Spearman correlation analysis was conducted between HbA1c and Burden_for_Parent. Cohen's standard was used to evaluate the strength of the relationship, where coefficients between .10 and .29 represent a small effect size, coefficients between .30 and .49 represent a moderate effect size, and coefficients above .50 indicate a large effect size [28].
Assumptions.
Monotonic Relationship. A Spearman correlation requires that the relationship between each pair of variables does not change direction [29]. This assumption is violated if the points on the scatterplot between any pair of variables appear to shift from a positive to negative or negative to positive relationship. Figure 1 presents the scatterplot of the correlation. A regression line has been added to assist the interpretation.
Figure 1. Scatterplots between each variable with the regression line added.
Results. The correlations were examined based on an alpha value of 0.05. There were no significant correlations between any pairs of variables. Table 5 presents the results of the correlation.
Table 5
Spearman Correlation Results Between HbA1c and Burden_for_Parent
Combination rs p
HbA1c-Burden_for_Parent -0.12 .393
Note. The confidence intervals were computed using α = 0.05; n = 51
Pearson Correlation Analysis
Introduction. A Pearson correlation analysis was conducted between Diagnosed_Since and HbA1c. Cohen's standard was used to evaluate the strength of the relationship, where coefficients between .10 and .29 represent a small effect size, coefficients between .30 and .49 represent a moderate effect size, and coefficients above .50 indicate a large effect size [28].
Results. The correlations were examined based on an alpha value of 0.01. There were no significant correlations between any pairs of variables. Table 7 presents the results of the correlation.
Table 7
Pearson Correlation Results Between Diagnosed_Since and HbA1c
Combination rp p
Diagnosed_Since-HbA1c 0.05 .713
Note. The confidence intervals were computed using α = 0.01; n = 51
Pearson Correlation Analysis
Introduction. A Pearson correlation analysis was conducted among Patient_Age, Parent_Age, and the total CSE. Cohen's standard was used to evaluate the strength of the relationships, where coefficients between .10 and .29 represent a small effect size, coefficients between .30 and .49 represent a moderate effect size, and coefficients above .50 indicate a large effect size [28].
Results. The correlations were examined using Holm corrections to adjust for multiple comparisons based on an alpha value of 0.01. There were no significant correlations between any pairs of variables. Table 8 presents the results of the correlations.
Table 8
Pearson Correlation Results Among Patient_Age, Parent_Age, and CSE
Combination rp p
Patient_Age-Parent_Age 0.29 .063
Patient_Age-CSE 0.19 .234
Parent_Age-CSE -0.32 .037
Note. The confidence intervals were computed using α = 0.01; n = 42; Holm corrections used to adjust
p-values.
ANOVA
Introduction. An analysis of variance (ANOVA) was conducted to determine whether there were significant differences in Emotional_Support_Seeking by HbA1c level.
Normality. The assumption of normality was assessed by plotting the quantiles of the model residuals against the quantiles of a Chi-square distribution, also called a Q-Q scatterplot [30]. For the assumption of normality to be met, the quantiles of the residuals must not strongly deviate from the theoretical quantiles. Strong deviations could indicate that the parameter estimates are unreliable. Figure 1 presents a Q-Q scatterplot of model residuals.
Figure 1. Q-Q scatterplot for normality of the residuals for the regression model.
Homoscedasticity. Homoscedasticity was evaluated by plotting the residuals against the predicted values (34; 35; 36)[31;32;33]. The assumption of homoscedasticity is met if the points appear randomly distributed with a mean of zero and no apparent curvature. Figure 2 presents a scatterplot of predicted values and model residuals.
Figure 2. Residuals scatterplot testing homoscedasticity.
Outliers. To identify influential points, Studentized residuals were calculated and the absolute values were plotted against the observation numbers [32;34]. Studentized residuals are calculated by dividing the model residuals by the estimated residual standard deviation. An observation with a Studentized residual greater than 3.31 in absolute value, the 0.999 quartile of a t distribution with 39 degrees of freedom, was considered to have significant influence on the results of the model. Figure 3 presents the Studentized residuals plot of the observations. Observation numbers are specified next to each point with a Studentized residual greater than 3.31.
Figure 3. Studentized residuals plot for outlier detection.
Results. The ANOVA was examined based on an alpha value of 0.05. The results of the ANOVA were significant, F(2, 37) = 4.92, p = .013, indicating there were significant differences in Emotional_Support_Seeking among the levels of HbA1c (Table 9). The eta squared was 0.209936774044914 indicating Level of HbA1c explains approximately 21% of the variance in Emotional_Support_Seeking. The means and standard deviations are presented in Table 10.
Table 9
Analysis of Variance Table for Emotional_Support_Seeking by HbA1c_Level
Term SS df F p ηp2
HbA1c_Level 7.72 2 4.92 .013 0.21
Figure 4. Emotional_Support_Seeking Means by factors levels of HbA1c
Table 10
Mean, Standard Deviation, and Sample Size for Emotional_Support_Seeking by HbA1c
Combination M SD n
above target (7.6%-8.9%) 0.24 0.83 14
critically elevated (≥ 9%) 0.39 0.47 11
in target (<7.5%) -0.59 1.13 15
Note. - indicate sample size was too small to calculate statistic.
Post-hoc. Paired t-tests were calculated between each pair of measurements to further examine the differences among the variables. Tukey pairwise comparisons were conducted for all significant effects based on a an alpha of 0.05. For the main effect of HbA1c, the mean of Emotional_Support_Seeking for ‘HbA1c level above target’ (M = 0.24, SD = 0.83) was significantly larger than for ‘HbA1c level in target’ (M = -0.59, SD = 1.13), p = .040. For the main effect of HbA1c, the mean of Emotional_Support_Seeking for ‘HbA1c level critically elevated’ (M = 0.39, SD = 0.47) was significantly larger than for ‘HbA1c in target’ ( M = -0.59, SD = 1.13), p = .022. No other significant effects were found.
Introduction. A Spearman correlation analysis was conducted among problem_solving_capacity, Emotional_Strength_Capacity, Community_and_Spiritual_Support_Seeking, Ability_to_make_Friends, Burden_on_Family, Societal_and_work_burden, and HbA1c. Cohen's standard was used to evaluate the strength of the relationships, where coefficients between .10 and .29 represent a small effect size, coefficients between .30 and .49 represent a moderate effect size, and coefficients above .50 indicate a large effect size [28].
Assumptions.
Monotonic Relationship. A Spearman correlation requires that the relationship between each pair of variables does not change direction [29]. This assumption is violated if the points on the scatterplot between any pair of variables appear to shift from a positive to negative or negative to positive relationship. Figure 5-Figure 11 presents the scatterplots of the correlations. A regression line has been added to assist the interpretation.
Figure 5. Scatterplots between each variable with the regression line added.
Figure 7. Scatterplots between each variable with the regression line added.
Figure 8. Scatterplots between each variable with the regression line added.
Figure 10. Scatterplots between each variable with the regression line added.
Figure 11. Scatterplots between each variable with the regression line added.
Results. The correlations were examined using Holm corrections to adjust for multiple comparisons based on an alpha value of 0.05. A significant positive correlation was observed between problem_solving_capacity and Emotional_Strength_Capacity (rs = 0.58, p < .001). The correlation
coefficient between problem_solving_capacity and Emotional_Strength_Capacity was 0.58, indicating a large effect size. This correlation indicates that as problem_solving_capacity increases, Emotional_Strength_Capacity tends to increase. No other significant correlations were found. Table 11 presents the results of the correlations.
Table 11
Spearman Correlation Results Among problem_solving_capacity, Emotional_Strength_Capacity, Community_and_Spiritual_Support_Seeking, Ability_to_make_Friends, Burden_on_Family, Societal_and_work_burden, and HbA1c
Combination rs Lower Upper p
problem_solving_capacity-Emotional_Strength_Capacity 0.58 0.28 0.78 < .001 problem_solving_capacity-Community_and_Spiritual_Support_ Seeking 0.19 -0.18 0.52 .307 problem_solving_capacity-Ability_to_make_Friends 0.13 -0.24 0.47 .492 problem_solving_capacity-Burden_on_Family -0.1 1 -0.45 0.27 .580 problem_solving_capacity-Societal_and_work_burden 0.24 -0.13 0.55 .197 problem_solving_capacity-HbA1c -0.1 9 -0.52 0.18 .305 Emotional_Strength_Capacity-Community_and_Spiritual_Supp ort_Seeking 0.01 -0.35 0.37 .969 Emotional_Strength_Capacity-Ability_to_make_Friends 0.18 -0.19 0.51 .335
Emotional_Strength_Capacity-Burden_on_Family -0.1 1 -0.45 0.26 .564 Emotional_Strength_Capacity-Societal_and_work_burden 0.14 -0.23 0.48 .464 Emotional_Strength_Capacity-HbA1c 0.02 -0.34 0.38 .904 Community_and_Spiritual_Support_Seeking-Ability_to_make_ Friends 0.02 -0.34 0.38 .904 Community_and_Spiritual_Support_Seeking-Burden_on_Famil y -0.3 1 -0.60 0.06 .100 Community_and_Spiritual_Support_Seeking-Societal_and_wor k_burden 0.11 -0.26 0.45 .569 Community_and_Spiritual_Support_Seeking-HbA1c -0.0 4 -0.39 0.33 .853 Ability_to_make_Friends-Burden_on_Family -0.0 9 -0.44 0.28 .629 Ability_to_make_Friends-Societal_and_work_burden -0.1 3 -0.47 0.24 .499 Ability_to_make_Friends-HbA1c 0.15 -0.22 0.49 .421 Burden_on_Family-Societal_and_work_burden -0.1 5 -0.48 0.23 .438 Burden_on_Family-HbA1c 0.16 -0.21 0.49 .393 Societal_and_work_burden-HbA1c -0.1 9 -0.51 0.19 .321
Note. The confidence intervals were computed using α = 0.05; n = 30; Holm corrections used to adjust
p-values.
Spearman Correlation Analysis
Introduction. A Spearman correlation analysis was conducted among Satisfied_with_Insulin, Self_Administration_of_Insulin, Burden_of_Care, Difficulties_In_Admin_of_Insulin, Routine_Flexibilities, Difficulty_using_insuline_away_from_home, Physical_Stress_and_Pain, and HbA1c. Cohen's standard was used to evaluate the strength of the relationships, where coefficients between .10 and .29 represent a small effect size, coefficients between .30 and .49 represent a moderate effect size, and coefficients above .50 indicate a large effect size [28].
Assumptions.
Monotonic Relationship. A Spearman correlation requires that the relationship between each pair of variables does not change direction [29]. This assumption is violated if the points on the scatterplot between any pair of variables appear to shift from a positive to negative or negative to positive
relationship. Figure 12-Figure 21 presents the scatterplots of the correlations. A regression line has been added to assist the interpretation.
Figure 12. Scatterplots between each variable with the regression line added.
Figure 13. Scatterplots between each variable with the regression line added.
Figure 15. Scatterplots between each variable with the regression line added.
Figure 16. Scatterplots between each variable with the regression line added.
Figure 18. Scatterplots between each variable with the regression line added.
Figure 19. Scatterplots between each variable with the regression line added.
Figure 20. Scatterplots between each variable with the regression line added.
Figure 21. Scatterplots between each variable with the regression line added.
Results. The correlations were examined using Holm corrections to adjust for multiple comparisons based on an alpha value of 0.05. There were no significant correlations between any pairs of variables. Table 12 presents the results of the correlations.
Spearman Correlation Results Among Satisfied_with_Insulin, Self_Administration_of_Insulin, Burden_of_Care, Difficulties_In_Admin_of_Insulin, Routine_Flexibilities, Difficulty_using_insuline_away_from_home, Physical_Stress_and_Pain, and HbA1c
Combination rs Lower Upper p
Satisfied_with_Insulin-Self_Administration_of_Insulin 0.11 -0.23 0.43 .524 Satisfied_with_Insulin-Burden_of_Care 0.03 -0.32 0.36 .885 Satisfied_with_Insulin-Difficulties_In_Admin_of_Insulin 0.12 -0.23 0.44 .515 Satisfied_with_Insulin-Routine_Flexibilities 0.11 -0.24 0.43 .542 Satisfied_with_Insulin-Difficulty_using_insuline_away_from_ho me 0.04 -0.30 0.38 .811 Satisfied_with_Insulin-Physical_Stress_and_Pain -0.0 6 -0.39 0.28 .728 Satisfied_with_Insulin-HbA1c -0.0 3 -0.37 0.31 .844 Self_Administration_of_Insulin-Burden_of_Care 0.18 -0.17 0.49 .318 Self_Administration_of_Insulin-Difficulties_In_Admin_of_Insuli n 0.13 -0.22 0.45 .454 Self_Administration_of_Insulin-Routine_Flexibilities -0.0 1 -0.35 0.33 .956 Self_Administration_of_Insulin-Difficulty_using_insuline_away_ from_home 0.07 -0.28 0.40 .708 Self_Administration_of_Insulin-Physical_Stress_and_Pain 0.04 -0.30 0.38 .812 Self_Administration_of_Insulin-HbA1c 0.09 -0.26 0.42 .609 Burden_of_Care-Difficulties_In_Admin_of_Insulin -0.1 7 -0.48 0.18 .335 Burden_of_Care-Routine_Flexibilities 0.07 -0.28 0.40 .701 Burden_of_Care-Difficulty_using_insuline_away_from_home -0.0 3 -0.37 0.31 .859 Burden_of_Care-Physical_Stress_and_Pain -0.0 9 -0.41 0.26 .630 Burden_of_Care-HbA1c 0.16 -0.19 0.47 .360 Difficulties_In_Admin_of_Insulin-Routine_Flexibilities 0.08 -0.26 0.41 .633 Difficulties_In_Admin_of_Insulin-Difficulty_using_insuline_awa y_from_home -0.0 6 -0.39 0.29 .744 Difficulties_In_Admin_of_Insulin-Physical_Stress_and_Pain 0.07 -0.27 0.40 .687 Difficulties_In_Admin_of_Insulin-HbA1c 0.30 -0.04 0.58 .080 Routine_Flexibilities-Difficulty_using_insuline_away_from_hom e 0.06 -0.28 0.39 .717 Routine_Flexibilities-Physical_Stress_and_Pain 0.01 -0.33 0.35 .962 Routine_Flexibilities-HbA1c 0.13 -0.21 0.45 .452 Difficulty_using_insuline_away_from_home-Physical_Stress_and _Pain 0.06 -0.28 0.39 .731 Difficulty_using_insuline_away_from_home-HbA1c 0.15 -0.19 0.47 .382 Physical_Stress_and_Pain-HbA1c 0.06 -0.29 0.39 .751
Note. The confidence intervals were computed using α = 0.05; n = 34; Holm corrections used to adjust
p-values.
Hierarchical Linear Regression
Introduction. A two-step hierarchical linear regression was conducted with HbA1c as the dependent variable. For Step 1, Mental_And_Internal_Coping_Strategy, Emotional_Support_Seeking,
problem_solving_capacity, Emotional_Strength_Capacity,
Community_and_Spiritual_Support_Seeking, and Ability_to_make_Friends were entered as predictor variables into the null model. Burden_on_Family, Societal_and_work_burden, Satisfied_with_Insulin, Self_Administration_of_Insulin, Burden_of_Care, Difficulties_In_Admin_of_Insulin, Routine_Flexibilities, Difficulty_using_insuline_away_from_home, and Physical_Stress_and_Pain were added as predictor variables into the model at Step 2.
Assumptions.
Normality. Normality was evaluated for each model using a Q-Q scatterplot. The Q-Q scatter plot compares the distribution of the residuals (the differences between observed and predicted values) with a normal distribution (a theoretical distribution which follows a bell curve). In the Q-Q scatterplot, the solid line represents the theoretical quantiles of a normal distribution. Normality can be assumed if the points form a relatively straight line. The Q-Q scatterplots for normality are presented in Figure 22.
Figure 22. Q-Q scatterplot for normality for models predicting HbA1c.
Homoscedasticity. Homoscedasticity was evaluated for each model by plotting the model residuals against the predicted model values [33]. The assumption is met if the points appear randomly
distributed with a mean of zero and no apparent curvature. Figure 23 presents a scatterplot of predicted values and model residuals.
Figure 23. Residuals scatterplot for homoscedasticity for models predicting HbA1c.
Multicollinearity. Variance Inflation Factors (VIFs) were calculated to detect the presence of multicollinearity between predictors for each regression model. Multicollinearity occurs when a predictor variable is highly correlated with one or more other predictor variables. If a variable exhibits multicollinearity then the regression coefficient for that variable can be unreliable and difficult to interpret. Multicollinearity also causes the regression model to have a loss in statistical power [35]. High VIFs indicate increased effects of multicollinearity in the model. Variance Inflation Factors greater than 5 are cause for concern, whereas VIFs of 10 should be considered the maximum upper limit (39)[36]. For Step 1, all predictors in the regression model have VIFs less than 10. For Step 2, all predictors in the regression model have VIFs less than 10. Table 13 presents the VIF for each predictor in the model.
Table 13
Variance Inflation Factors for Each Step
Variable VIF Step 1 Mental_And_Internal_Coping_Strategy 1.24 Emotional_Support_Seeking 1.51 problem_solving_capacity 1.48 Emotional_Strength_Capacity 1.23 Community_and_Spiritual_Support_Seeking 1.09 Ability_to_make_Friends 1.30 Step 2 Mental_And_Internal_Coping_Strategy 1.92
Emotional_Support_Seeking 2.24 problem_solving_capacity 2.02 Emotional_Strength_Capacity 1.77 Community_and_Spiritual_Support_Seeking 1.73 Ability_to_make_Friends 2.35 Burden_on_Family 1.57 Societal_and_work_burden 1.62 Satisfied_with_Insulin 1.21 Self_Administration_of_Insulin 1.93 Burden_of_Care 1.48 Difficulties_In_Admin_of_Insulin 2.38 Routine_Flexibilities 1.89 Difficulty_using_insuline_away_from_home 1.73 Physical_Stress_and_Pain 1.93
Note. - indicates that VIFs were not calculated as there were less than two predictors for the model step.
Outliers. To identify influential points, Studentized residuals were calculated and the absolute values were plotted against the observation numbers. An observation with a Studentized residual greater than 3.40 in absolute value, the 0.999 quartile of a t distribution with 29 degrees of freedom, was considered to have significant influence on the results of the model. Figure 24 presents a Studentized residuals plot of the observations. Observation numbers are specified next to each point with a Studentized residual greater than 3.40.
Figure 24. Studentized residuals plot for outlier detection for models predicting HbA1c.
Results. The hierarchical regression analysis results consist of model comparisons and a model interpretation based on an alpha of 0.05. Each step in the hierarchical regression was compared to the previous step using F-tests. The coefficients of the model in the final step were interpreted.
Comparing Models. The F-test for Step 1 was significant, F (6, 23) = 5.28, p = .002, ΔR 2 = 0.58. This
model indicates that adding Mental_And_Internal_Coping_Strategy, Emotional_Support_Seeking,
problem_solving_capacity, Emotional_Strength_Capacity,
Community_and_Spiritual_Support_Seeking, and Ability_to_make_Friends explained an additional 57.92% of the variation in HbA1c. The F-test for Step 2 was not significant, F (9, 14) = 0.49, p = .859, ΔR2 = 0.10. This model indicates that adding Burden_on_Family, Societal_and_work_burden,
Satisfied_with_Insulin, Self_Administration_of_Insulin, Burden_of_Care,
Difficulties_In_Admin_of_Insulin, Routine_Flexibilities,
Difficulty_using_insuline_away_from_home, and Physical_Stress_and_Pain did not account for a significant amount of additional variation in HbA1c. The results for the model comparisons are in Table 14.
Table 14
Model Comparisons for Variables predicting HbA1c
Model R2 df
mod dfres F p ΔR2
Step 1 0.58 6 23 5.28 .002 0.58
Step 2 0.68 9 14 0.49 .859 0.10
Note. Each Step was compared to the previous model in the hierarchical regression analysis.
Model Interpretation. Mental_And_Internal_Coping_Strategy did not significantly predict HbA1c, B = -0.04, t(14) = -0.10, p = .921. Based on this sample, a one-unit increase in Mental_And_Internal_Coping_Strategy does not have a significant effect on HbA1c. Emotional_Support_Seeking did not significantly predict HbA1c, B = -0.00, t(14) = -0.01, p = .994. Based on this sample, a one-unit increase in Emotional_Support_Seeking does not have a significant effect on HbA1c. problem_solving_capacity did not significantly predict HbA1c, B = 0.72, t(14) = 1.70, p = .111. Based on this sample, a one-unit increase in problem_solving_capacity does not have a significant effect on HbA1c. Emotional_Strength_Capacity significantly predicted HbA1c, B = -1.24,
t(14) = -3.62, p = .003. This indicates that on average, a one-unit increase of
Emotional_Strength_Capacity will decrease the value of HbA1c by 1.24 units. Community_and_Spiritual_Support_Seeking did not significantly predict HbA1c, B = -0.64, t(14) =
-1.81, p = .092. Based on this sample, a one-unit increase in
Ability_to_make_Friends did not significantly predict HbA1c, B = -0.80, t(14) = -1.78, p = .097. Based on this sample, a one-unit increase in Ability_to_make_Friends does not have a significant effect on HbA1c. Burden_on_Family did not significantly predict HbA1c, B = -0.04, t(14) = -0.11, p = .913. Based on this sample, a one-unit increase in Burden_on_Family does not have a significant effect on HbA1c. Societal_and_work_burden did not significantly predict HbA1c, B = -0.09, t(14) = -0.26, p = .801. Based on this sample, a one-unit increase in Societal_and_work_burden does not have a significant effect on HbA1c. Satisfied_with_Insulin did not significantly predict HbA1c, B = 0.07,
t(14) = 0.24, p = .810. Based on this sample, a one-unit increase in Satisfied_with_Insulin does not have a significant effect on HbA1c. Self_Administration_of_Insulin did not significantly predict HbA1c, B = 0.25, t(14) = 0.60, p = .561. Based on this sample, a one-unit increase in Self_Administration_of_Insulin does not have a significant effect on HbA1c. Burden_of_Care did not significantly predict HbA1c, B = -0.19, t(14) = -0.56, p = .583. Based on this sample, a one-unit increase in Burden_of_Care does not have a significant effect on HbA1c. Difficulties_In_Admin_of_Insulin did not significantly predict HbA1c, B = -0.77, t(14) = -1.64, p = .122. Based on this sample, a one-unit increase in Difficulties_In_Admin_of_Insulin does not have a significant effect on HbA1c. Routine_Flexibilities did not significantly predict HbA1c, B = 0.05, t(14) = 0.13, p = .896. Based on this sample, a one-unit increase in Routine_Flexibilities does not have a significant effect on HbA1c. Difficulty_using_insuline_away_from_home did not significantly predict HbA1c, B = 0.08, t(14) = 0.22, p = .826. Based on this sample, a one-unit increase in Difficulty_using_insuline_away_from_home does not have a significant effect on HbA1c. Physical_Stress_and_Pain did not significantly predict HbA1c, B = 0.20, t(14) = 0.45, p = .661. Based on this sample, a one-unit increase in Physical_Stress_and_Pain does not have a significant effect on HbA1c. The results for each regression are shown in Table 15.
Table 15
Summary of Hierarchical Regression Analysis for Variables Predicting HbA1c
Variable B SE β t p Step 1 (Intercept) 8.33 0.25 0.00 32.97 < .001 Mental_And_Internal_Coping_Strategy 0.18 0.26 0.10 0.69 .495 Emotional_Support_Seeking 0.26 0.34 0.13 0.76 .453 problem_solving_capacity 0.61 0.32 0.31 1.87 .075
Emotional_Strength_Capacity -1.30 0.26 -0.76 -5.06 < .001 Community_and_Spiritual_Support_Seeking -0.48 0.25 -0.27 -1.92 .067 Ability_to_make_Friends -0.85 0.30 -0.43 -2.82 .010 Step 2 (Intercept) 8.24 0.35 0.00 23.55 < .001 Mental_And_Internal_Coping_Strategy -0.04 0.37 -0.02 -0.10 .921 Emotional_Support_Seeking -0.00 0.47 -0.00 -0.01 .994 problem_solving_capacity 0.72 0.42 0.37 1.70 .111 Emotional_Strength_Capacity -1.24 0.34 -0.73 -3.62 .003 Community_and_Spiritual_Support_Seeking -0.64 0.35 -0.36 -1.81 .092 Ability_to_make_Friends -0.80 0.45 -0.41 -1.78 .097 Burden_on_Family -0.04 0.34 -0.02 -0.11 .913 Societal_and_work_burden -0.09 0.33 -0.05 -0.26 .801 Satisfied_with_Insulin 0.07 0.27 0.04 0.24 .810 Self_Administration_of_Insulin 0.25 0.42 0.13 0.60 .561 Burden_of_Care -0.19 0.34 -0.10 -0.56 .583 Difficulties_In_Admin_of_Insulin -0.77 0.47 -0.38 -1.64 .122 Routine_Flexibilities 0.05 0.40 0.03 0.13 .896 Difficulty_using_insuline_away_from_home 0.08 0.35 0.04 0.22 .826 Physical_Stress_and_Pain 0.20 0.44 0.09 0.45 .661
Note. Confidence intervals (CI) for B are based on an alpha of 0.05.
DISCUSSION OF THE RESULTS
Findings. An interesting finding is that the level of HbA1c explains approximately 21% of the variance in what is summarized as Emotional_Support_Seeking in the study. This factor arises from a rotated component matrix of the CSE and contains the items “Ability to get emotional support from friends and family”, “Ability to see things from the other person's point of view during a heated argument”, “Ability to get friends to help you with the things you need” and “Ability to leave options open when things get stressful”. For the main effect of HbA1c level, the mean of Emotional_Support_Seeking for ‘HbA1c levels above target’ (M = 0.24, SD = 0.83) was significantly larger than for ‘HbA1c levels in target’(M = -0.59, SD = 1.13), p = .040. For the main effect of HbA1c level, the mean of Emotional_Support_Seeking for ‘HbA1c levels critically elevated’ (M = 0.39, SD = 0.47) was significantly larger than for ‘HbA1c levels in target’ (M = -0.59, SD = 1.13), p = .022. In other words the worse the HbA1c the higher the capacity to cope in the areas written above. This can potentially be explained by higher necessity of seeking emotional support in patients with worse blood glucose control.
