Factors associated with loneliness in patients with diabetes mellitus

Abstract Aim To (a) explore the prevalence of loneliness in patients with diabetes mellitus and (b) identify loneliness‐related factors in the group of hospitalized patients with diabetes mellitus. Design A cross‐sectional study. Methods The study included 248 patients with diabetes mellitus who were staying in six Polish hospitals. A questionnaire including sociodemographic and clinical data, and the Revised UCLA Loneliness Scale (R‐UCLA), was used for research data collection. Data were collected from March 2019 to June 2019. Results Patients with diabetes generally experience moderate loneliness, with almost one‐fifth (16%) of patients experiencing intense loneliness. The patients scored a mean 9.94 out of a possible 20 in belongings and affiliation category and 20.14 out of a possible 40 in the intimate others category. Lower education, being single and the presence of chronic complications of diabetes mellitus were risk factors for increased loneliness.


Factors
. Cardiometabolic factors potentially associated with loneliness may contribute to type 2 diabetes (T2D) (Foti et al., 2020;Hackett & Steptoe, 2017). The mean prevalence of type 2 diabetes mellitus was significantly higher in the lonely group compared with the non-lonely group in a study assessing the relationship between loneliness and vascular biomarkers in older individuals (O'Luanaigh et al., 2012). Diabetes mellitus and hypertension were important predictors of loneliness in a study assessing the prevalence of loneliness among older patients with non-infectious diseases (Grover et al., 2019). Research to identify biological mechanisms linking loneliness with poor health shows that the effects of loneliness on health are mediated through the neuroendocrine system (Cacioppo et al., 2015).
An analysis of the relationship between loneliness in the older and metabolic dysregulation showed a fivefold higher potential risk of high HbA1c and threefold higher risk of high BMI among those who experienced loneliness (Shiovitz-Ezra & Parag, 2019). Scientific evidence suggests a relationship between loneliness and elevated glycated haemoglobin (O'Luanaigh et al., 2012) and metabolic syndrome (Whisman, 2010), both of which are risk factors for T2D.
Diabetes can lead to reduced mobility of the patient, causing difficulties in establishing social contacts, increasing depression and reducing social cohesion among older patients with T2D compared with non-diabetic patients (Steptoe et al., 2014). It was shown in patients with type 2 diabetes and pre-diabetes that patients reported an increased sense of isolation and loneliness even in the presence of strong family and social support (McConatha et al., 2020). Lukaschek et al. (2017) showed that low satisfaction with social relationships is associated with an increased risk of diabetes; the prevalence of T2D was higher in socially isolated individuals (Brinkhues et al., 2017).

| Aim
To (a) explore the prevalence of loneliness in patients with diabetes mellitus and (b) identify loneliness-related factors in the group of hospitalized patients with diabetes mellitus.

| Research design
A cross-sectional study.

| Participants
The study included 250 hospitalized patients with diabetes mellitus who had no communication problems and who gave their informed consent to participate in the study. The included patients were diagnosed with diabetes at least a year earlier and received pharmacotherapy.

| Data collection
Data were collected from March 2019 to June 2019 in diabetes units of six Polish hospitals, during the first or second day of hospitalization. Sociodemographic and clinical data were collected using a questionnaire; the R-UCLA scale was used for the assessment of loneliness. A total of 248 correctly filled questionnaires were qualified for the analysis.

| Instruments
The following tools were used:

| A questionnaire
Including a set of sociodemographic data: age, sex, marital status, education level, employment status and the place of residence.
Disease-related variables included the following: the reason for hospital admission, duration and type of diabetes, treatment used and the presence of chronic complications.

| Polish version of the Revised UCLA Loneliness Scale (R-UCLA)
Polish version of the Revised UCLA Loneliness Scale (R-UCLA) validated by Kwiatkowska et al. (2017) was used in the study. The original version of the scale (UCLA LS) was designed by Russell et al., (1980). The scale consists of 20 statements ranked by patients on a 4-point scale (1 = I never feel this way, 4 = I often feel this way).
The maximum score was 80. The total score consists of the sum of scores from 3 subscales: belongings and affiliation, intimate others and social others. According to Perry's classification system (1990), four degrees of loneliness were defined: a score of 65-80 -a high degree; 50-64 -a moderately high degree; 35-49 -a moderate degree; and 20-34 -a low degree of loneliness.

| Data analysis
The normality of the distribution was verified using the Shapiro-Wilk test. Homogeneity of variance was assessed with Levene's test. The following tests were used for hypothesis verification: for a comparison between two groups: t test (Student's t test) for independent variables or Cochran-Cox test in the case of failure to meet the assumption of homogeneity of variance; for a comparison between more than two groups: one-way analysis of variance (along with post hoc RIR Tukey test), Welch's F test was used in the case of failure to meet the assumption of homogeneity of variance; nonparametric Kruskal-Wallis rank ANOVA test (along with the post hoc Dunn's test) was used to compare small samples and highly diverse sample sizes; analysis of correlations: Pearson's linear correlation coefficient. The results were considered statistically significant if the calculated p-value was <.05. Statistica 10.0 was used for statistical analysis.

| Ethical issues
The research was performed in accordance with the Declaration of Helsinki. It was voluntary for the patients to answer the questionnaire, and they had the right to withdraw their participation at any time. All the answers were strictly confidential, and the patients were guaranteed full anonymity. Oral informed consent to participate in the study was obtained from participants. The study was approved by the Bioethics Committee of the Medical University of Warsaw.

| Trustworthiness
To ensure the validity and reliability of this study, we employed valid and tested tool. In previous analyses, each item had a discriminating power higher than 0.40 and as in the previous and later version, the reliability was perfect (α = 0.94).

| Characteristics of the participants
The mean age of patients was 57.9 years (SD 17.4), with the youngest patient aged 18 years, the oldest patient aged 94 years and 52% of patients over the age of 50 years. Marriage/relationship was reported by 71% of respondents; 43% of respondents had secondary education and 15% of patients declared higher education. Retirees and pensioners accounted for 56% of respondents; 35% of respondents were professionally active. Mean duration of diabetes in the study group was 12.14 years (SD 9.54), with the longest duration of 42 years. Type 2 diabetes was reported for 70% of patients; 24% of patients received hypoglycaemic agents and insulin therapy.
Diabetic neuropathy and retinopathy were reported for 37% and 21% of respondents, respectively; a total of 54% of patients had chronic complications of diabetes. The reasons for admission included uncontrolled diabetes (high glucose levels) in 54.4%, heart diseases in 8.4%, hypertension in 6.4% and diabetic foot syndrome in 4.4% of respondents. Max. = 76). Low, moderate, high and very high degree of loneliness was reported for 47%, 36%, 13% and 3% of respondents, respectively. The patients obtained a mean score of 9.94 out of a possible 20 (SD = 3.05) in belongings and affiliation category and 20.14 out of a possible 40 (SD = 7.10) in intimate others category.

| Correlations between loneliness and sociodemographic/clinical factors
As presented in Table 2, higher loneliness was shown in single patients (mean = 41.05; p = .025). Significant differences (p = .015) were shown between loneliness and the level of education. Patients with primary/junior high school education (mean = 43.72) had higher degree of loneliness compared with those with higher education (mean = 33.71; p = .008). Table 3. Significantly higher loneliness was observed in patients with chronic complications of diabetes (retinopathy, nephropathy and neuropathy) and patients with two or more complications compared with those with one or no complications (p = .000). There were no differences in loneliness in the study group in relation to the type of diabetes and the treatment used.

| D ISCUSS I ON
To the best of our knowledge, this is the first study conducted in Poland to assess loneliness in patients with diabetes mellitus. This study also provides insight in the relationship between sociodemographic and clinical factors and the feeling of loneliness. High levels of loneliness may have a negative impact on diabetic outcomes, while the long-term and demanding treatment of the disease itself and its potential complications may affect patient's perception of treatment and increase the feeling of loneliness. Studies conducted in a group of patients with chronic diseases showed that the participants perceived their loneliness level as moderate and that their illness perception was negatively affected as their loneliness levels increased (Özkan Tuncay et al., 2018). Loneliness has an impact on self-management ability in chronically ill individuals (Theeke et al., 2019).
In our study, the average loneliness score obtained by the respondents was 38.22 out of possible 80 (SD 11.55). Similar results were obtained in a group of patients with type 2 diabetes (mean = 36.89; SD 11.83) in a study which hypothesized that loneliness, as a negative psychosocial stressor, may further  (Kouris, Platsidaki, et al., 2016) and tuberculosis (mean = 44.36) (Yilmaz & Dedeli, 2016).
The widespread experience of loneliness was confirmed by Hackett et al. (2019). The authors showed that 25% of patients with T2D scored 45 or higher out of a possible 80 on the ULCA scale. This score is higher compared with normative data for the scale: 37.06 for men and 36.06 for women (Russell et al., 1980). In our study, 16% of respondents scored 50 or more, with a mean score of loneliness higher than that reported by Hackett et al., (2019), whose results corresponded with normative values and other behavioural medicine findings (Steptoe et al., 2004).
In our study, low, moderate, moderate high and severe high loneliness was reported for 47%, 37%, 13% and 3% of respondents, respectively, as in accordance with Perry's classification (1990). In a study assessing loneliness in patients with T1D and the relationship between loneliness and perception of treatment, Kusaslan Avci (2018) showed that the percentage of patients with severe high loneliness was higher (4%) and the percentage of patients with moderate high (9.2%) and moderate (21.8%) loneliness was lower compared with our findings. When using total score in the UCLA LS scale for the assessment of the severity of loneliness in older patients with non-infectious diseases, half of respondents showed moderate (13.2%), moderate high (17.6%) and severe high (18.6%) loneliness. The presence of diabetes was one of the important predictors of loneliness in this study (Grover et al., 2019). In a group of patients with tuberculosis, moderate and high loneliness levels were reported for 49% and 31.2% of respondents, respectively (Yilmaz & Dedeli, 2016).
In this study, patients had a mean score of 9.94 in the belongings and affiliation subscale and 8.13 out of a possible 20 in the social others subscale, and 20.14 out of a possible 40 in the intimate others subscale. Literature review shows that data that would allow for comparison of these findings are missing. As pointed out by adult patients with type 1 diabetes, the disease is present in all aspects of their life and they are dissatisfied with social relationships with regard to diabetes (Joensen et al., 2016). Participants of this study said that they felt "alone" in their everyday life with diabetes, described feeling "excluded from society" and experienced the lack of understanding by friends, family, healthcare workers and society.
It seems that age is an important factor in loneliness. be noted that the study included patients at the end of life (Çıracı et al., 2016).
Women can be more sensitive to stress in social relationships than men (Cohen-Mansfield et al., 2016). However, no significant relationship was found between loneliness scores and sex in studies in patients with different chronic diseases (Akin et al., 2014;Russell et al., 1980), which corresponds to our findings and those presented by other authors investigating patients with diabetes (Hackett et al., 2019;Kusaslan Avci, 2018;Niemcryk et al., 1990). In our study, men scored slightly higher than women, which was not confirmed by Kusaslan Avci (2018), who showed that women scored higher than men; however, the difference was not significant. Significantly higher loneliness levels were observed in women with chronic leg ulceration compared with men .
Previous studies showed a protective effect of marriage on glucose level control, especially in men (Cohen, 2004;Umberson, 1992).
An increase in the negative or positive qualities of a marriage may also reduce the prevalence of DM in women and men (Liu et al., 2016;Whisman et al., 2014). Studies on loneliness in patients with chronic diseases also confirmed significantly lower scores in married versus unmarried individuals (Akin et al., 2014;Çıracı et al., 2016;Russell et al., 1980). We believe that a positive marriage reduces loneliness in patients with diabetes, who are likely to be more successful in the long-term treatment and diabetes self-control. Hackett et al. (2019) showed that loneliness was significantly correlated with marital status, with lower mean levels of loneliness in married versus single, divorced/separated or widowed individuals. Our study, which also classified patients as single and non-single, also showed that patients in a relationship experienced lower loneliness. Compared with cancer patients, of whom urban residents experienced significantly higher loneliness (Çıracı et al., 2016), no such a correlation was found in our study.
Several studies investigating loneliness in patients with chronic diseases showed no significant differences between education or employment status and loneliness scores (Akin et al., 2014;Demir, 1989;Russell et al., 1980). However, our study and Kusaslan   (2018) showed that significantly lower loneliness scores were obtained by patients with higher education and those professionally active. This probably results from the fact that professionally and socially active patients are less lonely and that higher education level makes it easier to find regular employment. In our study, 35% of patients were professionally active.
Also, there was no significant correlation between loneliness scores and the use of hypoglycaemic agents in both our study and Hackett et al., (2019). Kusaslan Avci (2018)

| LI M ITATI O N S
Small sample size and the fact that only hospitalized patients participated in our study are a limitation of our research. The next limitation is that we do not have a comparator non-diabetic group.
Considering the cross-sectional nature of data, it is impossible to determine the direction of key relationships. Prospective studies are needed to understand whether loneliness is a predictor of poor clinical outcomes and psychosocial functioning indicators in patients with diabetes mellitus.

| CON CLUS ION
Our study showed that patients with diabetes generally experience

ACK N OWLED G M ENTS
We thank all nurses for their contribution.

CO N FLI C T O F I NTE R E S T
No conflict of interest has been declared by the authors.

AUTH O R CO NTR I B UTI O N S
The study idea and study design were conceived by EK and AS. EK, BD and JK-P wrote the first draft. JK-P and AS collected the data.
EK and BD performed the statistical analyses. All authors have been involved in the interpretation of the results and made important contributions to the drafting of the manuscript. All authors read and approved the final manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The date sets used for the current study are available from the corresponding author on reasonable request.