Diabetes Distress and Associations With Demographic and Clinical Variables: A Nationwide Population-Based Registry Study of 10,186 Adults With Type 1 Diabetes in Norway Free

Diabetes distress reflects the emotional responses to the challenges of living with diabetes (1,2) and impacts on self-management (3). Consequently, routine diabetes distress monitoring is recommended in clinical guidelines (e.g., using the 20-item Problem Areas in Diabetes [PAID-20]) (4–6). However, we currently lack data from population-based studies regarding the prevalence of clinically significant diabetes distress that may require additional support, because existing estimates are derived from samples obtained in clinical studies (1,7). Therefore, in this nationwide population-based registry study, we aimed to 1) calculate PAID-20 scores and the proportion of respondents reporting high diabetes distress, 2) describe the distribution of PAID-20 item scores to identify the most reported problem areas, and 3) examine associations of demographic and clinical variables with PAID-20 scores.

We conducted a nationwide population-based study using data from the Norwegian Diabetes Register for Adults (NDR-A), which consists of data about adults (age ≥18 years) with diabetes who attend outpatient clinics. The Norwegian Regional Committee for Medical and Health Research Ethics (REK Vest 171685) and data protection officials approved the study.

In 2020, 21,484 adults with type 1 diabetes (from 51 of 52 diabetes clinics) were registered in the NDR-A. We included the 17,828 (83.0%) who attended consultations between October 2019 and December 2020 (Fig. 1). In May 2021, NDR-A administrators sent a digital questionnaire to 16,255 individuals reachable on national digital platforms, and 10,391 (63.9%) responded. Diabetes distress was assessed with the PAID-20, measuring common diabetes-related problems scored from 0 (not a problem) to 4 (serious problem) (8). Scores are transformed to a 0–100 scale, where ≥40 is the established cut point for clinically significant high distress (1,7). Recently, moderate distress (score 17–39) was suggested (9). Furthermore, we extracted demographic variables (sex, age, ethnic origin, educational level, work, and cohabitation status), clinical process variables (diabetes duration, insulin regimen, continuous glucose monitoring [CGM], BMI, and smoking habits), and clinical outcome variables (hemoglobin A_1c [HbA_1c] and acute and long-term complications) from the NDR-A database.

We conducted all statistical analyses using Stata SE (version 17.0). We calculated the PAID-20 total score, the proportion reporting high distress, and the distribution of the PAID-20 response options for all items. We identified the five items with the highest endorsement of “somewhat serious problem” or “serious problem” (score 3 or 4). Among PAID-20 completers (<50% missing items), we used person-mean substitution for missing item scores in 240 individuals (2.4%) (Supplementary Table 1).

We used linear regression models to examine associations of demographic and clinical variables with distress scores. We imputed missing data using a chained equations algorithm and applied Rubin combinations rules in pooling β and B coefficients with 95% CIs across 100 imputed data sets. We report regression coefficients (B) with 95% CIs, signifying the change in PAID-20 score associated with a one-unit increase in the exposure variables, and standardized coefficients (β), indicating the strength and relative importance of each coefficient. We performed regression analyses, crude and with adjustment in a hierarchical manner, mutually adjusting the demographic variables for one another (model 1), the clinical process variables for one another and the demographics (model 2), and clinical outcome variables for one another, the demographic and clinical process variables (model 3). We interpreted the coefficients in models 1 to 3 block diagonally. Using Spearman, tetrachoric, and polychoric correlations and variance inflation factors, we found no collinearity.

Norwegian legislation prevents patient-level data sharing in public repositories. Data requestors must submit a proposal to the NDR-A ([email protected]), sign a data access agreement, and obtain necessary approvals to gain access.

We analyzed PAID-20 data from 10,186 participants (Fig. 1), of whom 54.0% were male (Table 1); mean age was 46.7 years (SD 15.3), mean diabetes duration was 22.9 years (SD 14.4), 35.6% received pump treatment, 69.7% used CGM, and mean was HbA_1c 59 mmol/mol (SD 12.7). Furthermore, 14.7% had treated retinopathy, 12.9% had an elevated urine albumin-to-creatinine ratio, and 10.9% had reduced foot sensitivity.

The mean PAID-20 score was 25.4 (SD 18.4), and 21.7% reported high distress (Table 1). The 20-item score distribution is shown in Fig. 2 and Supplementary Table 2. The five items most endorsed as problematic were “worrying about the future and the possibility of serious complications” (23.0%), “feeling that diabetes is taking up too much of your mental and physical energy every day” (19.7%), “feeling burnt out by the constant effort needed to manage diabetes” (18.7%), “feelings of guilt or anxiety when you get off track with your diabetes management” (17.4%), and “worrying about low blood sugar reactions” (13.0%). In total, 42.2% reported that at least one item was a somewhat serious or serious problem (Supplementary Table 3). Among participants with a PAID-20 score <40, this proportion was 20.9%.

In the fully adjusted regression (model 3), female sex, CGM use, daily smoking, symptomatic hypoglycemia, HbA_1c, treated retinopathy, and reduced foot sensitivity were all associated with higher PAID-20 scores (Table 2), whereas older age, longer diabetes duration, European origin, university/college education, and working/studying were associated with lower PAID-20 scores. In examining these associations, sex displayed the largest effect (β) on PAID-20 scores, followed by age and HbA_1c. The associations observed for the demographic and clinical process variables remained consistent across the models, except for university/college education, which was attenuated in model 3.

In this large population-based cohort of adults with type 1 diabetes and PAID-20 data, we found that 21.7% of respondents reported high distress. The most common distress sources were concerns about complications, self-management burdens, and feelings of guilt or anxiety when self-management efforts fell short. Additionally, we identified associations between higher diabetes distress and sex, age, non-European origin, primary education only, not working/studying, shorter diabetes duration, CGM use, more symptomatic hypoglycemia, higher HbA_1c, retinopathy, reduced foot sensitivity, and smoking.

We found that one in five participants reported high distress, confirming results from smaller studies reporting 17% to 24% (7,9,10). The most endorsed problem areas related to worries about complications and diabetes self-management are consistent with previous research in health care settings (9,10). The associations of higher distress with female sex and younger age also confirm previous studies (7,9,11). Surprisingly, CGM use was associated with a higher PAID-20 score. One could assume that some choose CGM because of distress and that using CGM would positively affect and lower distress. However, in a recent review, only two of nine randomized controlled trials reported distress reduction after initiation of CGM (12), possibly explained by devices serving as constant reminders of disease or contributing to information overload. Prospective studies are needed to examine the CGM–distress mechanisms.

Our results support the relationship between poorer HbA_1c and higher distress (9,11,13). The association could be bidirectional, since previous studies have found that high distress is associated with higher HbA_1c (14,15). Not reaching the recommended glycemic target can be a potential source of diabetes distress, and distress can contribute to less effective self-management, typically increasing HbA_1c (3,16). However, individuals reaching target HbA_1c might also experience high distress, and distress sources may vary by HbA_1c level. Consequently, the link between HbA_1c and distress requires further investigation.

Diabetes distress is an expected part of living with diabetes that can arise from many sources (1,2,10). Elevated diabetes distress should be recognized as clinically important, since research indicates associations with psychological and somatic variables (1,2,9,12,14). If unaddressed, diabetes distress tends to increase or persist (17,18). Therefore, clinicians should be able to address the multifaceted nature of living with diabetes in routine care (1,4,5). Individuals with diabetes typically want to talk to clinicians about their challenges (10,19). Discussing distress can reduce its impacts on living with diabetes, thereby improving self-management and glycemic control (2,3,7,10,18). Clinicians should pay particular attention to those who report feeling guilty or anxious about not succeeding with self-management, worrying about complications, and consequently feeling exhausted from everyday demands. Addressing specific problem areas with elevated scores is probably more relevant than examining total scores. However, many clinicians report lack of training, confidence, and resources to address distress (20). Health care services clearly need to allocate resources for this purpose. Diabetes teams should have access to assistance from behavioral health specialists and specialist follow-up of patients with very high distress.

Our population-based study contributes to an increased understanding of diabetes distress among adults with type 1 diabetes. Major strengths are the real-world setting, sample size (the largest reported to date), and high response rate. The national population-based data add weight to smaller studies in health care settings that may have been biased by participant selection. Furthermore, we used the PAID-20, recommended for research and clinical settings (1,6). However, timing and technology literacy may have affected the response rate. Therefore, diabetes distress rates might have been underestimated. Also, the study design prohibits causal inference.

To conclude, diabetes distress is common among adults with type 1 diabetes in Norway and associated with clinically relevant factors, warranting greater attention in regular diabetes care, which should include efforts to identify and address diabetes distress. By doing so, we may help improve health outcomes for individuals with type 1 diabetes.

Funding. This study was funded by the Western Norway University of Applied Sciences, the Norwegian Nursing Association, and grant F-12610 from the Western Norway Regional Health Authority (Helse Vest).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. I.H. wrote the initial draft of the manuscript, with assistance from J.G.C., R.M.N., T.C.S., R.B.S., M.M.I., M.G., K.F.L., T.V.M., S.S.L., D.A.R., G.Å.U., and A.H. I.H. and A.H. designed the study, with input from R.M.N., T.C.S., R.B.S., M.M.I., and M.G. I.H. and T.E. conducted the statistical analyses, under the supervision of R.M.N. All authors contributed revisions to the paper and approved the final version. I.H. and A.H. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. This study was presented in part in poster form at the International Diabetes Federation World Diabetes Congress, Lisbon, Portugal, 5–8 December 2022.