Understanding the Impact of Diabetic Peripheral Neuropathy and Neuropathic Pain on Quality of Life and Mental Health in 6,960 People With Diabetes

Diabetes is the leading course of peripheral neuropathy, resulting in both small- and large-fiber impairment (1). Diabetic peripheral neuropathy (DPN) affects up to 50% of people with diabetes (1–3) and often leads to increased morbidity (e.g., neuropathic pain, foot ulcers, and lower-extremity amputations) (4–6).

The International Association for the Study of Pain (IASP) defines neuropathic pain as “pain caused by a lesion or disease of the somatosensory system” (7,8). Painful polyneuropathy is a cardinal symptom of small-fiber neuropathy (9,10), affecting up to 33% of people with DPN (11–13). Painful DPN can be an early manifestation of DPN but is also often seen in late stages with complete loss of protective sensation (14). As such, the severity of small-fiber neuropathy is not linearly related to the pain level, and painful neuropathy is also seen in people with alleged pure large-fiber neuropathy (9). Patient-reported questionnaires have been proposed as a method for classifying individuals as having neuropathic pain in large-scale studies (15). Furthermore, it can be used for stratifying painful neuropathy into sensory phenotypes, which might be related to underlying pathophysiological mechanisms (16–18), although this so far has not led to more efficient management.

While sensory phenotyping may aid the understanding of the pathophysiology, there is a need to understand the biopsychosocial aspects of painful and painless DPN (19,20). Quality of life (QoL) is of utmost importance when looking at diabetes complications, because these are chronic and might develop at a young age. Studies of painful DPN have shown a decreased QoL and an increase in symptoms of both anxiety and depression. Only few studies have investigated the impact of painless DPN (21–24). Overall, the investigations have been performed in small, selected study populations. Socioeconomic factors have been found to be associated with painful neuropathy along with poorer mental health, and painless neuropathy is less investigated (21,24,25). The different sensory phenotypes of painful DPN are sparsely investigated regarding their impact on QoL and mental health (26).

The current study aimed to 1) investigate the associations between painful and painless DPN and QoL, mental health, and socioeconomic factors in a large, unselected population with diabetes compared with a group without DPN, 2) evaluate the prevalence of different sensory pain descriptors, and 3) evaluate the association between sensory pain descriptors and QoL, mental health, and socioeconomic factors.

This observational, cross-sectional survey study was performed in the Northern Danish Region during November 2022. The data were acquired from a regional questionnaire survey distributed digitally to all adults diagnosed with diabetes residing in the Northern Danish Region at the time of the study. The survey was distributed through secure electronic mail. The Northern Danish Region is an area of ∼8,000 km² with ∼600,000 citizens. Identification of eligible participants within the region was facilitated through data sourced from the local Business Intelligence unit, which operates within the Danish National Patient Registry. This registry encompasses comprehensive medical data for all individuals residing in Denmark, intricately linked across various registers at an individual level through unique social security numbers. Identification of individuals living with diabetes was accomplished through the National Health Insurance Service Registry, ensuring inclusion of individuals receiving care at outpatient clinics within hospitals as well as those under the care of general practitioners.

According to Danish law, approval from the Regional Ethical Committee was not warranted, but approval was granted from the Regional Data Security Authority (ref. no. F2022-089; Aalborg, Denmark). The study was conducted in accordance with the Helsinki Declaration of 1975. Data on prevalence of painful and painless DPN in the Northern Danish Region have been published previously (27).

The full electronic survey consisted of multiple validated questionnaires. The current study included the Michigan Neuropathy Screening Instrument questionnaire (MNSIq), the modified Douleur Neuropathique en 4 Questions (DN4-interview), the 36-item Short Form Health Survey (SF-36), and the Hospital Anxiety and Depression Scale (HADS). An aggregate score (with a maximum of 100) was calculated for SF-36, while two aggregate scores (with a maximum of 21 each) were calculated for HADS, HADS anxiety subscale (HADS-A) and HADS depression subscale (HADS-D). An abnormal score of HADS-A and HADS-D was set to ≥8 (28). Questionnaires were chosen based on current recommendations, keeping burden of disease in mind. Neuropathic pain phenotyping by international consensus states that classifying people into possible painful neuropathy could be done using a simple symptom checklist, such as DN4, which has good performance in DPN (15,23). For the 7-item DN4-interview, the validated cutoff of ≥3 was used (29). Aiming for a similar screening tool for DPN, the well-known, symptom based MNSIq was chosen, with the validated cutoff of ≥4 (3,23,30). HADS was used because it covers both anxiety and depression, while focusing on the cognitive symptoms of affective distress to reduce the impact of somatic distress on results (24,25,28). SF-36 constitutes a practical and valid method for both QoL in physical and mental conditions and has previously been used in multiple studies on painful DPN (28,31).

Basic characteristics of the participants were self-reported and consisted of age, sex, height, weight, diabetes type, diabetes duration, and socioeconomic factors, including education level, employment status, and annual household income. Income was categorized into 10 intervals, ranging from <15,000 DKK/month (∼€2,000) to >100,000 DKK/month (∼€13,500) before taxes. Education level was graded in five different tiers based on the highest completed education.

Participants with complete information of MNSIq, SF-36, HADS, and socioeconomic factors were included (n = 6,960) and subsequently subgrouped at multiple levels. Firstly, the participants were grouped into people with possible DPN (MNSIq score ≥4) and no DPN (MNSIq score <4). Secondly, participants with DPN, who answered “yes” to having pain in both feet, also answered the DN4-interview and were grouped into possible painful DPN (MNSIq score ≥4 and DN4-interview score ≥3) and possible painless DPN (MNSIq score ≥4 and no bilateral pain or DN4-interview score <3). Participants without a DN4-interview answer (n = 208) were included in the analysis as part of the possible painless DPN group, because they answered “no” when asked about pain in their feet. Thirdly, the participants with possible painful DPN (MNSIq score ≥4 and DN4-interview score ≥3) were further subgrouped based on three pain descriptors (burning pain, cold pain, and electric pain) and four sensory descriptors (tingling, pins and needles, numbness, and itching). Each participant could be present in multiple groups if multiple answers were checked in the DN4-interview. Analyses on QoL and socioeconomic factors were performed at all levels.

Difference between groups (possible DPN/no DPN and possible painful DPN/possible painless DPN) were assessed based on the data types; nominal data (χ² test), ordinal data (Mann-Whitney U test), and ratio and interval data (independent Student t test or Mann-Whitney U test dependent on normality). Normality of the data was assessed using quantile–quantile plot and density plot. The level of statistical significance was set to 0.01.

Relationships between predictors and outcomes were assessed using multiple linear regression models to ensure adequate adjustment. Backward selection was performed to identify the best model for each outcome, using Akaike information criterion as the stopping criterion. Collinearity was assessed for all full models using variance inflation factor, with a cutoff value of 5. Interaction terms were included for BMI and height along with education, salary, and work status to address any collinearities. See details in Supplementary Material.

Regression analysis was performed on the total population. To facilitate interpretation, we show all associations expressed as unstandardized regression coefficients (β) and the corresponding 95% CI in Table 2. Multiple linear regression analyses were used to investigate the associations of determinants (i.e., possible neuropathy [yes/no] and possible painful neuropathy [yes/no]) with outcomes (i.e., SF-36 score, HADS-A score, and HADS-D score). Model 1 results are adjusted for age and sex. Model 2, in addition to the adjustment in model 1, was adjusted for BMI and salary. We chose these variables based on backward selection analysis, because these variables were important in four or more of the models performed. Subgroup analysis was performed stratified based on diabetes type and is available in the Supplementary Material.

The prevalence of sensory pain descriptors was calculated as the proportion with a given pain descriptor or sensory descriptor (e.g., cold pain) of the total group of participants with possible painful DPN (MNSIq ≥4 and DN4-interview ≥3).

Pearson correlation analysis was performed to evaluate correlation between pain/sensory descriptors and QoL. Here, both subscores of QoL and the aggregate score were analyzed individually. Pearson correlation analysis was performed to evaluate correlation between pain/sensory descriptors and mental health, based on the two aggregate scores HADS-A and HADS-D.

All analyses were performed with R x64 4.1.1 statistical software. The level of statistical significance was set to 0.01.

The data sets generated during and/or analyzed during the current study are not publicly available due to restrictions to availability by regional rules of data protection but are available from the corresponding author upon reasonable request. Data analysis is available from the authors upon reasonable request.

The survey was answered by 7,743 individuals with diabetes. Participants with incomplete socioeconomic and mental health factors (n = 783) were excluded (Fig. 1). The cohort was 60% male and had an average age of 65 years. There were 980 with type 1 diabetes, 5,965 with type 2 diabetes, and 15 with other types of diabetes. The median diabetes duration was 10 years. Differences between groups with and without possible DPN were minimal. See demographics in Supplementary Table 1.

Mental health differed between participants with possible DPN and without DPN, with a median HADS-A score of 5.00 (interquartile range [IQR] 2, 9) and 2.00 (IQR 1, 5) in the two groups (P value <0.001), respectively. HADS-D scores were 4.00 (IQR 1, 8) and 1.00 (IQR 0, 3) for possible DPN and without DPN, respectively. See Table 1.

QoL assessed with SF-36 differed significantly between participants with possible DPN and without DPN, with a median of 55.1 (IQR 36.7, 73.6) and 82.2 (IQR 63.6, 90.9), respectively (P < 0.001). Education level did not differ between the two groups, on the contrary a significant lower work status and salary was seen for the DPN group (P < 0.001), see Supplementary Tables 4–7.

Multiple regression analyses for model 1 and model 2 are seen in Table 2. All β-coefficients were significant after adjustment. For model 2, R² was 0.23, 0.16, and 0.15.

Regression analysis based on stepwise backward selection found that the presence of neuropathy was an important predictor of both SF-36 and anxiety, but not depression. SF-36 and HADS-A still differed significantly between participants with possible DPN and without DPN after adjustment for relevant variables, see Supplementary Material.

A total of 1,601 participants had possible DPN (MNSIq score ≥4). Of these, 1,085 had possible painful DPN (bilateral pain and DN4-interview score ≥3) and 516 had possible painless DPN (DN4-interview score <3). Basic demographic and clinical data for the two groups are seen in Supplementary Table 3.

Significantly lower QoL, HADS-A, and HADS-D scores were seen in the group with painful DPN compared with the group with possible painless DPN (P < 0.001) (Table 1). An abnormally high anxiety score was more frequent in people with possible painful DPN (420 [38.7%]) than in people with possible painless DPN (143 [27.7%]; P < 0.001). An abnormally high depression score was seen more frequently in people with possible painful DPN (298 [27.5%]) than in people with possible painless DPN (106 [20.5%]; P = 0.004). SF-36 median pain was 45.0 (IQR 33, 68) for possible painful DPN and 67.5 (IQR 45, 88) for possible painless DPN. See Supplementary Tables 8–10. Socioeconomic factors did not differ between the two groups. See Supplementary Table 11.

Multiple regression analyses for model 1 and model 2 are seen in Table 2. All β-coefficients were significant after adjustment. For model 2, the R² was 0.13, 0.15, and 0.14.

Regression analysis based on stepwise backward selection showed that the presence of neuropathic pain was an important predictor of all outcomes. SF-36, HADS-A, and HADS-D still differed significantly between participants with possible painful DPN and possible painless DPN after adjustment for relevant variables, see Supplementary Material.

Multiple regression analysis stratified by diabetes type 1 and type 2 showed a larger association between both DPN and painful DPN with SF-36 in diabetes type 1. A larger association was seen between DPN and HADS-A/HADS-D in diabetes type 2, but not between painful DPN and HADS-A/HADS-D. See Supplementary Tables 12 and 13.

The most prevalent pain descriptor was burning pain (73%), followed by electric pain (43%) and cold pain (31%). The most prevalent sensory descriptor was pins and needles (93%), followed by tingling (82%), numbness (55%), and itching (44%). See Supplementary Table 14.

The correlation analysis between pain/sensory descriptors and QoL, anxiety, and depression found weak correlations (Table 3). Weak positive correlations with anxiety were found for electric pain (ρ = 0.118, P < 0.001) and itch (ρ = 0.169, P < .001). Weak positive correlations with depression were found for cold pain (ρ = 0.125, P < 0.001), electric pain (ρ = 0.128, P < 0.001), and itch (ρ = 0.133, P < 0.001). Weak negative correlations with QoL were found for cold pain (ρ = −0.193, P < 0.001), electric pain (ρ = −0.114, P < 0.001), itch (ρ = −0.159, P < 0.001), and numbness (ρ = −0.121, P < 0.001).

The study showed diminished QoL and reduced mental health in participants with DPN compared with participants with diabetes without DPN. The addition of pain diminished QoL and mental health further. The findings were present before and after adjustment. Painful DPN was associated with lower education level, salary, and work status.

Painful DPN and painless DPN have shown decreased QoL and increased risk of anxiety and depression (18,21–23,28,32–35), as supported by the present large cohort study. Only a few studies have found no effect on mental health in painless DPN (36).

Gylfadottir et al. (23) found that DPN and painful DPN were independently and additively associated with lower QoL. The current study found that the DPN group experienced limitations due to physical health regardless of pain. Limitations due to mental health were not affected in the painless DPN group but were highly affected in the painful DPN group. Aslam and Singh (28) investigated QoL in painful DPN using SF-36 with two aggregate scores and found that QoL with regard to physical health was the most affected compared with mental health. The present study found that diabetes type 1 had larger associations compared with diabetes type 2. A wider CI was seen for all measures in diabetes type 1 due to a smaller, more heterogeneous group. While the strength of the associations varies slightly, the directionality and significance are similar between diabetes type 1 and type 2.

Gylfadottir et al. (23) found more symptoms of anxiety and depression for both DPN and painful DPN and that DPN had a greater effect on mental health than painful DPN. Similar findings were described by Kec et al. (24). They showed contributing factors for emotional distress to be the severity of pain, cognitive processing, and the severity of DPN. Bai et al. (32) found that depression was greatly associated with neuropathy independent of pain. The current study had similar results, with a 2.2-higher HADS-D score in the DPN group than the no-DPN group, after adjustment of age, sex, BMI, and yearly salary. Using backward selection, we found that DPN did not explain the variance in HADS-D, meaning other variables might explain the higher HADS-D score seen in DPN.

The current study found that salary and work status were negatively associated with DPN, while education, salary and work status were negatively associated with painful DPN. Additional analysis of DPN risk factors conducted in our previous study (27) aligns well with findings from other population-based studies (37). Barbosa et al. (21) found that education was negatively associated with painful DPN, while Kec et al. (24) found a weak association between lower education and higher anxiety. The features of emotional distress in relation to painful DPN and socioeconomic factors are complex, and while education level, salary, and work status are indicative of socioeconomic status, one cannot extrapolate these factors as being associated with poor socioeconomic demographics, because socioeconomic level was not evaluated using validated methods in either the present study or available literature (21,24,25).

The current study found that the most prevalent pain descriptor in painful DPN was burning pain. The most common sensory descriptor was numbness. Furthermore, itch was present in 44% of the participants. The prevalence of these pain and sensory descriptors in diabetes was investigated by Tesfaye et al. (38), who found that the most prevalent pain descriptor was burning (71%), while the most common sensory descriptors were tingling (70%) and numbness (63%). The main difference between the studies is the prevalence of the sensory descriptor “pins and needles,” which was only present in 58% compared with 91% in present study (38).

Few studies have investigated the prevalence of itch in relation to diabetic neuropathic pain. Tesfaye et al. (38) found that 38% experienced itch in relation to pain, which is comparable with the prevalence of 44% in present study. Chronic itch in diabetes is associated with a higher degree of anxiety and depression along with decreased QoL (39), as supported by the present large cohort study. Itch transmission through C nerve fibers, which are affected in neuropathy, provides a plausible pathophysiological basis for its occurrence in DPN. The DN4 questionnaire is not designed to diagnose neuropathic itch. However, the observed prevalence of itch in our cohort highlights the need for more in-depth investigations into neuropathic itch and its potential impact on individuals with DPN. With a high prevalence and an impact on overall QoL and mental health, neuropathic itch constitutes an underestimated focus.

In general, there was low correlation between the different pain and sensory descriptors and mental health and QoL. The associations that were present were weak and primarily present for cold pain, electric pain, and itch.

The phenotypes defined in the literature are primarily based on quantitative sensory testing (QST) or patient-reported outcomes (17). Baron et al. (26) performed cluster analysis using QST and ended up with three clusters, one with sensory loss, one with thermal hyperalgesia, and one with mechanical hyperalgesia. They did not evaluate the difference in psychosocial factors between these three groups. From their results, one can see that 47% of individuals with sensory loss experienced depression based on HADS and that this number was 31% and 22% for thermal and mechanical hyperalgesia, respectively (26). Unfortunately, QST has not led to a mechanisms-based treatment approach. Overall, the impact of sensory phenotypes on QoL and mental health has scarcely been investigated, especially in relation to patient-reported outcomes. Bouhassira et al. (16) performed a cluster analysis using data from the Pain Symptom Inventory but did not include psychosocial factors.

The impact of chronic pain on QoL and mental health is well known in DPN. By confirming these results in a large cohort, we underline the importance of an interdisciplinary approach for management of painful DPN, along with DPN and diabetes in general. We found that specific sensory descriptors might, to some degree, have a greater association with poor QoL and mental health. It may be considered including itch in the clinical assessment and management, as this had a high prevalence (44%) and impact on QoL, anxiety, and depression.

The study is a survey based cross-sectional study and therefore suffers the limitations thereof. Including the inability to address causality; this would especially be of interest concerning the possible bidirectional relationship between painful DPN and mental health disorders. The diagnoses of painful and painless DPN are based on MNSIq and DN4-interview, not including a clinical assessment. MNSIq is commonly used as a questionnaire-based screening tool for DPN (23,30,40), with a high specificity of 92%. However, with a sensitivity of 40%, it might underestimate the DPN prevalence (3,23,30). In the current study, the prevalence of DPN (23%) and painful DPN (15%) was lower than expected, potentially due in part to the online administration of the survey. We found that a large proportion of individuals with possible DPN had painful DPN, this might indicate that MNSIq is better at finding painful DPN, as pain is a more noticeable symptom. Gylfadottir et al. (23) found that all neuropathy groups reported general pain, also groups without painful DPN. We found lower SF-36 pain scores in all groups, using the SF-36 bodily pain z score, indicative of pain being present in all groups. Keeping this in mind, both MNSIq and DN4-interview results might be affected by pain in other locations than the feet.

Compromises were made concerning participant burden for this prospective survey study. One consideration is that the DN4-interview does not assess paroxysmal pain, evoked pain, or parasthesia, all symptoms involved in neuropathic pain (33). Secondly, although good for group-level analysis of QoL, SF-36 does not assess disease-specific health distress, sexual function, cognitive function, or sleep quality (31). Lastly, the current study did not collect information on glycemic control or pain medication use. Including these factors could have provided valuable insights into their potential impact on the observed associations.

The current study showed that DPN with concomitant neuropathic pain impacts QoL and mental health negatively and likewise for DPN without neuropathic pain. Further, it reveals differences between the seven DN4 pain descriptors in increased anxiety, increased depression, and reduced overall QoL. This substantiates the focus on interdisciplinary management of different phenotypes. An important finding is the high prevalence of itch (44%) in the cohort with painful DPN, as this descriptor was associated with lower QoL and poor mental health. Neuropathic itch has been an underestimated focus in diabetes.

Acknowledgments. The authors thank Maria Bitsch Poulsen, from Mech-Sense, Aalborg University Hospital, for her assistance in data management.

Funding. The Danish Diabetes Association has funded A.M.W.’s work.

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

Author Contributions. M.K.B. wrote the first draft of the manuscript. M.K.B., A.-M.W., A.N., C.B., P.V., and J.R. were involved in the conception, design, and conduct of the study. M.K.B., C.D.M., L.A.-N., N.E., and J.R. were involved in the analysis and interpretation of the results. All authors edited, reviewed, and approved the final version of the manuscript. J.R. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented as an oral and poster presentation at the 34th Annual Meeting of the Diabetic Neuropathy Study Group (NEURODIAB), Rome, Italy, 5–8 September 2024, and as an oral presentation at the 60th Annual Meeting of the European Association for the Study of Diabetes (EASD) 2024, Madrid, Spain, 9–13 September 2024.

Handling Editors. The journal editors responsible for overseeing the review of the manuscript were John B. Buse and Rodica Pop-Busui.