Declining Incidence Rates of Distal Symmetric Polyneuropathy in People With Type 1 and Type 2 Diabetes in Denmark, With Indications of Distinct Patterns in Type 1 Diabetes

Diabetic neuropathy is a debilitating complication of diabetes, leading to a considerable burden on the health care system (1). Half of all people with diabetes are estimated to develop diabetic neuropathy during their lifetime (2). The most common form is distal symmetric sensorimotor polyneuropathy (DSPN), which constitutes up to 90% of cases (2). DSPN is a major risk factor for foot ulcerations and is associated with increased risk of amputations and death (3,4). Approximately 15–20% of all people with diabetes are estimated to develop painful DSPN (5), which is associated with decreased quality of life (6).

Declining trends in some diabetes-related complications have been described in past decades (7–10). Whether this applies to DSPN and other forms of DN is not evident due to the paucity of large-scale, contemporary, longitudinal studies investigating trends in the incidence of DSPN and DN. Only one large-scale Western study (Pittsburgh Epidemiology of Diabetes Complications Study) has investigated changes in incidence of DSPN in type 1 diabetes and found indications of declining incidence rates (11). Several studies have shown decreasing trends in amputations (12,13), which may indicate that the incidence of DSPN is declining and/or improvements in management of ulcers and revascularization.

In the present study, we examined trends in the incidence of DSPN in a large longitudinal cohort of individuals with type 1 and type 2 diabetes treated at a tertiary outpatient clinic in Denmark, between 1996 and 2018, using both >25 V and age-, sex-, and height-specific (hereafter, age-sex-height–specific) cutoff values for vibration perception threshold (VPT).

All individuals attending outpatient clinical care at Steno Diabetes Center Copenhagen (SDCC), in Herlev, Denmark, during the period from 1 January 1996 to 1 January 2018 were included in this cohort study (N = 19,342). All individuals were identified in the electronic patient records and followed from date of first VPT measurement until first diagnosis of DSPN, death, or the last update of the electronic medical record at SDCC (1 January 2018).

Recorded data included information on sex, date of birth, date of diabetes diagnosis, type of diabetes, height, and dates of foot examinations when VPT measurements were obtained. The VPT measurements were performed at the tip of the first toe bilaterally, according to standardized guidelines at SDCC (14) and were scheduled to be obtained at least annually. Two different cutoff values for VPT were applied to define DSPN: 1) >25 V, which is a cutoff voltage common to many diabetes clinics to categorize patients into high-risk or low-risk categories for foot ulcers (15); and 2) an age-sex-height–specific cutoff value an as individualized and more precise threshold (16,17).

DSPN was classified as bilateral abnormal VPT. Individuals with missing data on one foot due to unilateral amputation (n = 214) and VPT below the cutoff on the opposite foot were classified as having DSPN. Individuals with bilateral amputations were excluded from analysis (n = 30). Individuals with missing information on date or type of diabetes diagnosis and VPT measurements were excluded (n = 5,592). For analyses using age-sex-height–specific cutoff values, individuals with missing height measure (or height measure longer than 10 years before the first VPT examination) were also excluded (n = 300). For those who had several height measures, the closest measure to the VPT measurement was used. Finally, to investigate the incidence rates, all individuals with prevalent DSPN were excluded (n = 4,277 for >25 V as the cutoff and n = 7,558 for the age-sex-height–specific cutoff). Supplementary Fig. 1 presents a detailed flow of included individuals in the study.

To investigate changes in known risk factors associated with DSPN, mean annual values for HbA_1c, systolic blood pressure, LDL cholesterol, and triglycerides were calculated for type 1 and type 2 diabetes throughout the study period.

The follow-up time (risk time) of the population was split into 6-month intervals, each recording the age, calendar time, and diabetes duration for the interval. Incidence rates of DSPN were analyzed separately for type 1 and type 2 diabetes using Poisson models with log person-year (PY) as the offset. Age, calendar time, and diabetes duration were included as splines with four knots for age and diabetes duration and with three knots for calendar time. To allow for different age effects for men and women, an interaction between age and sex was included as well.

Analyses were repeated as a sensitivity analysis in which all participants with amputations were excluded. Mean annual values of HbA_1c, systolic blood pressure, LDL-cholesterol, and triglycerides were calculated as annual means for included calendar years. For repeated measurements, the mean of all measurements for the given calendar year was calculated and included in the annual mean. The means for HbA_1c, systolic blood pressure, LDL-cholesterol, and triglycerides are illustrated graphically in Q-Q plots. Data are presented graphically with 95% CIs. All analyses and graphs were generated with R, version 3.6.2 (R Project for Statistical Computing, Vienna, Austria).

According to Danish law, ethics approval and participant consent are not required for registry-based studies. Access and use of the described data were approved by the Danish Data Protection Agency (j-No. VD-2019-197) and the Danish Patient Safety Authority (j-No. 3-3013-2959/1).

The data that support the findings of this study are available from Statistics Denmark, but restrictions apply to the availability of these data, which were used under license for the present study and, therefore, are not publicly available.

In total, 19,342 individuals were identified in the electronic medical record system at SDCC from 1 January 1996 to 1 January 2018. After removal of ineligible individuals according to the exclusion criteria, data on 13,750 people remained for the analysis using >25 V as the cutoff (Supplementary Fig. 1). Another 4,277 individuals were excluded because they had DSPN at the first measurement of VPT, leaving data on 9,473 individuals (with type 1 diabetes, n = 4,761 [50.0%]; with type 2 diabetes, n = 4,712 [50.0%]) for analysis.

For the analysis using the age-sex-height–specific cutoff, 13,450 individuals were eligible (Supplementary Fig. 1), but 10,667 individuals had DSPN at the first measurement of VPT at SDCC and were excluded, leaving data on 2,783 individuals (with type 1 diabetes, n = 1,071 [38.5%]; with type 2 diabetes, n = 1,712 [61.5%]) for analysis.

The median (interquartile range) age at study entry was 35 (24–46) years for patients with type 1 diabetes and 56 (47–64) years for patients with type 2 diabetes. Median diabetes durations were 8.0 (2.8–15.0) and 3.1 (0.1–6.9) years, respectively, for type 1 and type 2 diabetes. A total of 2,492 incident cases of DSPN (type 1 diabetes: n = 1,128; type 2 diabetes: n = 1,364) for 78,597 PY were recorded in the period 1996–2018. These correspond to an incidence rate of 2.8/100 PY for type 1 diabetes and 6.5/100 PY for type 2 diabetes across the entire study period.

From 1996 to 2018, the incidence rate of DSPN decreased for both type 1 and type 2 diabetes (Figs. 1A, 1B, 2A, and 2B). The incidence rate (95% CI) decreased from 4.78 (3.60–6.33)/100 PY in 1996 to 1.15 (0.91–1.47)/100 PY in 2018 for a 40-year-old man with type 1 diabetes (Fig. 1B) and from 16.54 (11.80–23.18)/100 PY to 8.02 (6.63–9.69)/100 PY for a 60-year-old man with type 2 diabetes (Fig. 2B). The incidence rates decreased correspondingly for women (Figs. 1A and 2A).

Incidence rates increased with older age and with higher incidence across all ages for both type 1 and type 2 diabetes (Figs. 1C and 2C).

Incidence rates increased with longer diabetes duration for type 1 diabetes, although with a slower increase after a diabetes duration of 10 years (Fig. 1D). For type 2 diabetes, incidence rates increased with longer diabetes duration (≤10 years). Thereafter, incidence rates were constant. The incidence rates for men were significantly higher than for women (Fig. 2D).

For the subset of participants with data on the age-sex-height–specific cutoff for VPT, the median (interquartile range) age at study entry was 42 (27–53) years for type 1 diabetes and 59 (51–67) years for type 2 diabetes. The median diabetes duration (interquartile range) was 5.9 (1.2–13.2) years and 3.4 (0.2–7.4) years, respectively. A total of 1,146 (type 1 diabetes: n = 737; type 2 diabetes: n = 709) incident cases of DSPN for 18,677.5 PY were recorded (among 2,783 individuals) in the period 1996–2018. This corresponds to an incidence rate of 19.5/100 PY for type 1 diabetes and 15.6/100 PY for type 2 diabetes.

From 1996 to 2008, the incidence rate of DSPN increased marginally; thereafter, marked decreases for both type 1 and type 2 diabetes were observed. In the period 1996–2018, the incidence rate (95% CI) decreased from 17.31 (12.31–24.36)/100 PY to 16.21 (12.41–21.16)/100 PY for a 40-year-old man with type 1 diabetes (Fig. 3B) and from 15.67 (9.78–25.09)]/100 PY to 13.20 (9.59–18.16)/100 PY for a 60-year-old man with type 2 diabetes (Fig. Fig. 4B). The incidence rates decreased correspondingly for women (Figs. 3A and 4A).

For type 1 diabetes, there was a decreasing incidence rate with older age (Fig. 3C). In type 2 diabetes, the incidence rate was constant until the age of 60 years, after which the incidence rates increased (Fig. 4C). There were no sex differences in the incidence rates for either type 1 or type 2 diabetes (Figs. 3C and 4C).

The incidence rates increased with longer diabetes duration for type 1 diabetes (Fig. 3D), whereas for type 2 diabetes, incidence rates increased with longer diabetes duration ≤10 years; thereafter, they remained constant (Fig. 4D).

To test the assumption that an amputated leg could constitute an extremity without DSPN, we performed sensitivity analyses in which patients with unilateral amputations were excluded (n = 170 for >25 V as the cutoff; n = 40 for the age-sex-height–specific cutoff). Estimates in these sensitivity analyses did not change significantly and are presented in Supplementary Figs. 2–5.

For HbA_1c and systolic blood pressure, measurements were available from 1998 to 2017; for LDL-cholesterol and triglyceride levels, measurements were available from 1999 to 2017. Mean HbA_1c was 8.5% (69 mmol/mol) for type 1 diabetes and 8.6% (71 mmol/mol) for type 2 diabetes in 1998 and decreased to 7.9% (63 mmol/mol) and 7.9% (63 mmol/mol) in 2017 for type 1 and type 2 diabetes, respectively (Supplementary Fig. 6). Mean systolic blood pressure decreased from 131 to 129 mmHg for type 1 diabetes and from 142 to 132 mmHg for type 2 diabetes in the same period. The mean LDL-cholesterol level decreased from 2.8 mmol/L and 3.3 mmol/L in 1999 to 2.4 mmol/L and 1.9 mmol/L in 2017 for type 1 and type 2 diabetes, respectively. The mean of triglyceride levels increased for type 1 diabetes (from 1.2 mmol/L in 1999 to 1.3 mmol/L) and decreased for type 2 diabetes in the same period (from 2.4 mmol/L to 2.3 mmol/L) (Supplementary Fig. 6).

In this large population cohort of individuals with type 1 and type 2 diabetes who were followed at a tertiary outpatient clinic and receiving standard diabetes treatment, we demonstrated declining incidence rates for DSPN in the period 1996 to 2018. The declining incidence rates were observed for men and women from all age groups and were consistent using two different cutoffs to diagnose DSPN. To our knowledge, these results are the first to indicate that the incidence of DSPN follows the trends seen for other diabetes-related micro- and macrovascular complications (7,8,18).

The declining trends in the incidence of DSPN may be a consequence, as for other complications, of concomitant improvement in diabetes-related risk factors. The beneficials effects of glucose control to prevent DSPN in type 1 diabetes are described in several studies (19,20). For type 2 diabetes, some studies showed beneficial effects of tight glucose control (21–23), whereas others did not confirm these findings (22,24–26). These conflicting results for type 2 diabetes indicate that a glucocentric approach to preventing DSPN is not sufficient, and the control of other risk factors has a substantial impact. Risk factors associated with the development of DSPN include hypertension, dyslipidemia, higher BMI, and smoking (27,28). The reduction in incidence of DSPN may be attributed to improvement in risk factor management; we demonstrated a decline in mean levels of HbA_1c and systolic blood pressure from 1998 to 2017 and for LDL-cholesterol levels from 1999 to 2017, supporting this assumption. Similarly, a study of a Danish population-wide diabetes cohort (also including people treated at SDCC), showed a decrease in HbA_1c and in the proportion of individuals with glycemic dysregulation (HbA_1c ≥7.5% [58 mmol/mol]) in the period 2011–2017 for both type 1 and type 2 diabetes (12). In the same study, reduction in dyslipidemia, a decline in frequency of daily smokers, and decline in systolic and diastolic blood pressures were observed for both type 1 and type 2 diabetes (12). A Danish nationwide cohort study showed an increase in use of antihypertensive drugs from 28% to 42% among individuals with type 1 diabetes during the study period and an increase in use of lipid-lowering drugs from 5% in 2000 to 30% in 2010, with a plateau after 2010, which may have resulted in declining levels of risk factors (29). A similar pattern for use of lipid-lowering drugs (an increase from 9% in 2000 to 58% in 2012, with a subsequent decline) was observed for type 2 diabetes in Denmark (12). Thus, an improvement in risk factors for DSPN in Denmark over the past decades may be attributed to the observed decrease in the incidence rates of DSPN.

VPT is a commonly used measure to define DSPN in diabetes specialist centers (30). A cutoff of >25 V is commonly used to define DSPN in clinical settings because it is positively associated with developing foot ulcers (15), whereas the age-sex-height–specific cutoff gives a more exact estimate of DSPN (31). In this study, the incidence rate of DSPN was higher for both types of diabetes when using the age-sex-height–specific cutoff, compared with the voltage cutoff >25 V. One reason for this could be that the age-sex-height–specific cutoff identifies DSPN in earlier stages or, alternatively, that the >25 V cutoff underestimates DSPN because it does not consider changes in VPT with age (32–35) and greater height (36–38).

For both type 1 and type 2 diabetes, we observed higher incidence rates with increasing age and, when using VPT >25 V, higher incidence rates. This finding is in line with observations of other diabetes-related complications (e.g., chronic kidney disease, end-stage kidney disease, amputations) (12).

For type 1 diabetes, we observed a higher incidence rate of DSPN for younger compared with older individuals when using the age-sex-height–specific cutoff. The higher incidence rates for younger individuals could be due to 1) a higher sensitivity of finding DSPN at an early diabetes stage, which may be useful to identify potentially reversible DSPN. However, no studies have shown the reversibility of DSPN using this threshold; instead, it might rather lead to an overestimation of the incidence rates for younger individuals and/or an underestimation of the incidence rates for older individuals.

Although we observed increasing incidence rates with increasing age for type 2 diabetes for both cutoffs, we observed a discrepancy in incidence rates in type 1 diabetes with older age. Here, incidence rates increased with older age when using the >25 V cutoff but decreased using the age-sex-height–specific cutoff. This decrease in incidence with age for the age-sex-height–specific cutoff is surprising and does not fit with the general understanding of DSPN for which age and diabetes duration are known risk factors. Because these findings could be due to the nature of the cutoff algorithm possibly over diagnosing younger individuals, a healthy survivor bias may also play an influencing role. However, because similar trends are not seen for type 2 diabetes, this may be unlikely.

For type 1 diabetes, we observed an increasing incidence rate with longer diabetes duration, with similar patterns using both cutoffs. For type 2 diabetes, the incidence rate of DSPN increased until a diabetes duration of 10 years by both cutoff types. After 10 years’ diabetes duration, we observed a plateau. This ceiling effect may be due to healthy survivor bias and may partially be explained by improvement in diabetes-related risk factors for DSPN in Denmark during the study period (12).

This cohort study benefits from its large sample size and the prospective design that allowed us to assess temporal changes. The results will be representative for the Danish population with type 1 diabetes, because all individuals with type 1 diabetes are referred to specialist care, whereas the individuals with type 2 diabetes followed at SDCC compose a heavily selected group with generally dysregulated diabetes, often with multiple complications; less complicated cases are typically followed in primary care and compose approximately 80% of people with type 2 diabetes in Denmark. Therefore, incidence rates for DSPN may be overestimated compared with the overall population with type 2 diabetes in Denmark. Specialist care of diabetes has become increasingly focused on more severe diabetes during the study period but with unchanged referral criteria for individuals with type 2 diabetes from primary to specialist care. Despite this, we demonstrated a decline in incidence of DSPN with calendar time, suggesting a potentially greater decline in the overall population.

Our results for incidence rates by age and diabetes duration are inconsistent, especially for the specified cutoff, which may be due to survival bias. However, because the cohort is open, we do not consider that this will affect the decreasing incidence rates by calendar time.

Lead-time bias could have affected estimates. However, VPT measurements were from the routine controls, where methods were unchanged in the study period. Therefore, it is unlikely that lead-time bias has affected incidence rates in this study.

Though use of standardized guidelines for the VPT measurements at SDCC, the examinations were performed by different physicians and podiatrists, which can lead to interobservational variations. Furthermore, we stratified patients based on a single measurement of VPT. Repeated measurements of VPT could potentially increase sensitivity for DSPN. We used single measurements to diagnose DSPN in line with clinical practice and research to allow for comparability between studies.

Diagnosing DSPN by scoring systems allowing the inclusion of both symptoms and objective measurements would increase the sensitivity of DSPN diagnoses. However, such data are not available in this cohort. In addition, more sensitive measures of DSPN could have been applied to allow for more insights about various forms of neuropathy. However, these have not found their way into clinical practice in Denmark. Finally, because measurement of VPT is part of a larger annual diabetes examination at SDCC, participants absent from this control at one or more time points could have influenced results.

In this large cohort study, we demonstrated decreasing incidence rates for DSPN among individuals with type 1 and type 2 diabetes in the period of 1996 to 2018. The incidence rates varied considerably when using two different cutoffs for VPT. In addition, we found distinct age-related patterns for type 1 diabetes when using the age-sex-height–specific cutoff values of VPT and the >25 V cutoff.

Funding. The work leading to this article received funding from internal sources.

Duality of Interest. P.R. has received honoraria to his institution for consultancy and/or speaking fees from Astellas, Abbott, AstraZeneca, Bayer, Boehringer Ingelheim, Eli Lilly, Gilead, MSD, Mundipharma, Novo Nordisk, and Sanofi Aventis, and research grants from AstraZeneca and Novo Nordisk (to the institution). H.A. affiliated with Novo Nordisk after the completion of the formal analysis. D.V. has received research grants from Bayer A/S, Sanofi, Novo Nordisk A/S, and Boehringer Ingelheim, and holds shares in Novo Nordisk. No other potential conflicts of interest relevant to this article were reported.

The funders of the study had no role in study design, data collection, analysis, interpretation, or writing of the report.

Author Contributions. H.I.M. contributed to the formal analysis, investigation, and writing the original draft. T.W.H. and P.R. contributed to supervision of the work and reviewed and edited the manuscript. D.V. reviewed and edited the manuscript. V.R.C. contributed to the formal analysis and reviewed and edited the manuscript. H.A. contributed to the study design, formal analysis, and supervision of the work, and reviewed and edited the manuscript. C.S.H. contributed to conceptualization of the study and to study supervision and formal analysis, and reviewed and edited the manuscript. H.I.M. and C.S.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. Parts of the study findings were presented orally at the virtual 58th European Association for the Study of Diabetes Annual Meeting, 19–23 September 2022.