Cumulative risk of diabetic foot complications in risk groups of type 1 and type 2 diabetes: Real‐world evidence from a 22‐year follow‐up study

1. BACKGROUND

The International Working Group on the Diabetic Foot (IWGDF) recommends annual foot screening for low‐risk individuals of diabetic foot ulcers (DFUs), with screening frequency increasing with an increased risk. ¹ These screening intervals are largely based on expert opinion, owing to limited supporting evidence. ¹ Annual screening intervals for diabetic retinopathy, once similarly recommended, were revised, and extended for low‐risk individuals upon evaluation of progression risks. ²

It is relevant to consider if foot screening intervals for individuals at very low‐risk (IWGDF risk category = 0) ¹ of diabetic foot complications (DFCs) may be extended beyond a year. A prerequisite for this guideline revision would be ascertaining the risk of DFCs. Hence, we aimed to estimate the cumulative risk (CR) of DFCs in risk groups of type 1 (T1D) and type 2 diabetes (T2D), using a simple risk‐stratification rule. Based on the IWGDF 4‐level risk‐stratification system, ¹ our study combines categories 1–2 into a single ‘high‐risk’ group and uses two clinically‐relevant risk factors.

2. METHODS

• Data sources: In this register‐based cohort study, data was accessed from Steno Diabetes Center Copenhagen (SDCC), a diabetes outpatient clinic in the Capital Region of Denmark; the Danish National Patient Register (DNPR) ³ ; the Danish Civil Registration System (CRS) ⁴ ; and the Danish Diabetes Register (DDR) constructed at SDCC. ⁵ The SDCC database, comprising electronic patient health records, was the primary data source. Data on DFCs was supplemented from the DNPR; diabetes type and diagnosis were sourced from the DDR; migration status and mortality were extracted from the CRS.

• Study population: Individuals aged 18 years and above, diagnosed with either T1D or T2D, with at least two foot screenings at SDCC between 1 January 1998 and 30 April 2020, and without prevalent DFCs at the first screening were included. SDCC has a representative sample of the T1D population in Denmark, whereas its T2D population represents 10% of individuals with the most complicated T2D in the region, since the majority are followed in general practice. Individuals were stratified into low and high risk using two risk‐stratification methods, which relied on clinically accessible risk factors of DFCs, selected by established knowledge and expert opinion. The first method used loss of protective sensation (LOPS), assessed by vibration perception threshold (VPT) using a biothesiometer, and peripheral arterial disease (PAD), assessed by dorsal foot pulse. A biothesiometer is a non‐invasive, hand‐held device which has readings from 0 to 50 V and detects the threshold at which individuals lose vibration perception. ‘Low risk’ was defined as VPT <25 V and present foot pulse (IWGDF category = 0), ¹ and ‘high risk’ as VPT ≥25 V and/or absent foot pulse. The second risk‐stratification method used LOPS and age, where ‘low risk’ was defined as VPT <25 V and age <50 years for T1D and age <60 years for T2D, and ‘high risk’ as VPT ≥25 V or age ≥50/60 years for T1D and T2D, respectively.

• Follow‐up: Baseline was defined as date of the first foot screening at SDCC with either a registered VPT or foot pulse. Date of exit from the study was date of the first DFC, death, end of contact at SDCC or data extraction (30 April 2020), whichever came first. Date of end of contact was calculated as 2 years after the last foot screening at SDCC.

• Variables and outcomes: Variables used for analysis were LOPS (VPT), foot pulse, age, sex, diabetes duration and calendar time. DFC was defined as the first event of DFU or lower‐extremity amputation (LEA), assessed using ICD‐10 and procedure codes (Table S1).

• Statistical analyses: CR of DFCs was calculated based on two risk‐stratification methods (LOPS/PAD or LOPS/age). Follow‐up time was split in 6‐month intervals. A multistate model was set up enabling transition from low to high risk, and from the two risk groups to DFCs or death during follow‐up. After accounting for competing risks of death and moving from low to high risk, Poisson regression models were used for estimating CR, unadjusted incidence rates (IRs) of DFCs, and IR adjusted for calendar time, age, sex, diabetes duration and the interaction between calendar time and risk status (from risk‐stratification method 1).

3. RESULTS

A total of 14 609 individuals (6279 with type 1 diabetes and 8330 with type 2 diabetes) were included in the study with a total follow‐up time of 135 143 person‐years (from risk‐stratification method 1). At baseline, 56.4% of the T1D population was at low risk, as compared to 27.3% of the T2D population. The median age was 34.9 years (Q1, Q3: 25.0, 45.1) and 50.9 years (Q1, Q3: 37.7, 61.5) in the T1D low‐ and high‐risk groups, respectively, while it was 52.4 years (Q1, Q3: 43.8, 60.1) and 63.6 years (Q1, Q3: 55.5, 71.6) in the T2D low‐ and high‐risk groups, respectively. Tables S2 and S3 show detailed baseline population characteristics.

During the follow‐up period, a total of 613 events (478 DFUs and 135 LEAs) and 961 events (763 DFUs and 198 LEAs) were recorded in the T1D and T2D populations, respectively. From 2000 to 2020, there was a decline in the incidence of DFCs in high‐risk men and women with T1D and T2D, whereas for the T2D low‐risk groups, we find a threefold increase from 2000 to 2011, which slightly decreases after 2017 (Figures [Link], [Link] and Tables S4 and S5). Based on risk‐stratification method 1, the 3‐year CR of DFCs was 0.20% (95% CI 0.13%–0.30%) and 3.92% (95% CI 3.62%–4.27%) in the T1D low‐ and high‐risk groups, respectively. Similarly, the 3‐year CR of DFCs in the T2D low‐risk group was 0.39% (95% CI 0.26%–0.58%), while in T2D high‐risk group, it was 4.31% (95% CI 4.01%–4.62%). Based on risk‐stratification methods, Tables 1 and S6 present yearly CR of DFCs, while Figures 1 and S3 illustrate unadjusted IR of DFCs in T1D and T2D.

TABLE 1.

Cumulative risk of developing diabetic foot complications every year after baseline, based on risk‐stratification method 1 (LOPS/PAD).

	Yearly cumulative risk a
	Year 1	Year 2	Year 3	Year 4	Year 5
Type 1 diabetes
High risk	1.36% (1.26%–1.49%)	2.67% (2.46%–2.91%)	3.92% (3.62%–4.27%)	5.12% (4.73%–5.58%)	6.28% (5.79%–6.84%)
Low risk	0.07% (0.05%–0.1%)	0.13% (0.09%–0.2%)	0.2% (0.13%–0.3%)	0.25% (0.17%–0.39%)	0.31% (0.21%–0.48%)
Type 2 diabetes
High risk	1.54% (1.43%–1.65%)	2.98% (2.77%–3.19%)	4.31% (4.01%–4.62%)	5.56% (5.17%–5.95%)	6.72% (6.26%–7.19%)
Low risk	0.14% (0.1%–0.21%)	0.27% (0.18%–0.41%)	0.39% (0.26%–0.58%)	0.5% (0.33%–0.74%)	0.6% (0.4%–0.89%)

^a Cumulative risk has been expressed as a percentage with 95% confidence intervals (CIs).

FIGURE 1.

Unadjusted incidence rate of diabetic foot complications (DFCs) in type 1 and type 2 diabetes, based on risk‐stratification method 1 (LOPS and PAD). Shaded areas represent 95% confidence intervals.

4. DISCUSSION

Our main finding is that for many people living with diabetes, the risk for DFC is very low and thus screening intervals could be prolonged. A strength of this study is the detailed registration of LOPS and foot pulse in a large cohort of individuals with an extensive follow‐up period. Some study limitations are as follows: Firstly, since DFC was included as a composite outcome, specific risk estimates for individual outcomes of DFUs and LEAs were unavailable. This was done due to the low no. of LEAs (<3) in the T1D low‐risk group, which could not have been reported. Therefore, outcomes in the low‐risk group were predominantly DFUs, while outcomes in the high‐risk group were a combination of DFUs and LEAs. Hence, it may be inferred that the T1D low‐risk group is not at significant risk for LEA. Secondly, since DFUs are poorly reported in the DNPR, the possibility that some individuals at baseline may have prevalent DFUs cannot be entirely rejected. ⁶ However, the under‐reporting of DFUs in the DNPR was overcome by using detailed DFU data from the SDCC electronic health records, where all individuals screened in the foot clinic are given an applicable diagnosis code. Some individuals at SDCC with minor DFUs might have been underreported if they were not screened in the foot clinic, but these numbers are assumed to be negligible since all major DFUs are screened and reported.

Conducting foot screenings in individuals with diabetes is important to prevent DFCs through early interventions, education, and proper footwear. The IWGDF 2023 guidelines‐recommended yearly foot screenings for low‐risk groups may be relevant in some clinical settings. However, the large difference observed in CR of DFCs between risk groups of T1D and T2D populations in our study serves as an initial guide to propose that foot screening intervals for individuals at low risk of DFCs are extended beyond 1 year in Denmark and comparable healthcare settings.

Previous studies conducted in Danish populations ⁷ , ⁸ found that the IR of LEA increase with age and DM duration, which is consistent with our findings. This supports the generalizability of our results, as we observe the same trends as seen in previous studies with different populations, including less complicated T2D. Additionally, it provides an argument for using age as a factor in risk stratification.

The risk of DFCs is expected to be the same for the T1D population across Denmark; while it is expected to be the same or lower for the T2D population followed in general practice, where a higher proportion of low‐risk individuals are treated. There was no notable difference between the CR estimated by risk‐stratification methods 1 and 2. We speculate that the increase in IR from 2000 until 2011 observed in T2D low‐risk groups might have been driven by the increased registration of DFCs in these groups at SDCC. These results may not be generalizable to populations in dissimilar healthcare settings, but this concept may be applied broadly to evaluate risk and consider local screening intervals.

5. CONCLUSION

These results would hence, support that foot screening intervals for low‐risk groups of DFCs are extended beyond 1 year in comparable healthcare settings. Additionally, comprehensive risk prediction models like the Steno T1 Risk Engine ⁹ incorporating additional predictors, such as smoking, hypertension, and blood glucose, could provide more precise risk profiles for DFCs in individuals with T1D and T2D.

AUTHOR CONTRIBUTIONS

A.A. contributed to the study design, formal analysis, writing the initial draft and editing the manuscript. P.F.R. contributed to study conceptualisation, study design, formal analysis, supervision of the work and reviewed and edited the manuscript. A.H. contributed to study conceptualisation, study design, formal analysis and reviewed and edited the manuscript. C.S.H. contributed to study conceptualisation, study design, and reviewed and edited the manuscript. V.K. contributed to the study design and reviewed and edited the manuscript. F.P. and A.R. contributed to study conceptualisation, and reviewed and edited the manuscript. P.R. and T.S.A. contributed to study conceptualisation, supervision of the work and reviewed and edited the manuscript. A.A. and T.S.A. are the guarantors of this study and take responsibility for the integrity of the data and the accuracy of the data analysis.

FUNDING INFORMATION

This project has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska‐Curie grant agreement no. 101073533 (DIALECT: Diabetes Lower Extremity Complications Research and Training Network in Foot Ulcer and Amputation Prevention). A.H. is supported by a Data Science Emerging Investigator grant (no. NNF22OC0076725) by the Novo Nordisk Foundation. T.S.A. was supported by the Novo Nordisk Foundation Grant (NNF23OC0084081; Map D‐Foot). P.F.R. was supported by a research grant from the Danish Diabetes Academy, which is funded by the Novo Nordisk Foundation, grant number NNF17SA0031406. A.A. was supported by a travel grant from the European Association for the Study of Diabetes (EASD) for attending their 60th Annual Meeting where some results from the current manuscript were presented as a short oral presentation.

CONFLICT OF INTEREST STATEMENT

P.R. has served as consultant on advisory boards for Astra Zeneca, Abbott, Bayer, Novartis, Boehringer Ingelheim, Gilead and Sanofi (honoraria to his institution); has received unrestricted grants (to his institution) from Astra Zeneca, Bayer, and Novo Nordisk; and grants from Bayer, Novo Nordisk and Lexicon pharmaceuticals (study medication to investigator initiated study). T.S.A. and P.F.R. own stocks with Novo Nordisk A/s. No other potential conflicts of interest were reported by the other authors. The funding bodies played no role in study design, data management, analysis, interpretation or writing of the manuscript.

PEER REVIEW

The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1111/dom.16200.

Supporting information

Supplementary Figure S1. Incidence rates of diabetic foot complications (DFCs) over calendar time in risk groups of men with type 1 and type 2 diabetes, using a logarithmic scale. Reference age is 50 years and reference diabetes duration is 20 years. Shaded areas represent 95% confidence intervals.

Supplementary Figure S2. Incidence rates of diabetic foot complications (DFCs) over calendar time in risk groups of women with type 1 and type 2 diabetes, using a logarithmic scale. Reference age is 50 years and reference diabetes duration is 20 years. Shaded areas represent 95% confidence intervals.

Supplementary Figure S3. Incidence rate of diabetic foot complications (DFCs) in type 1 and type 2 diabetes, based on risk‐stratification method 2 (LOPS and age). Shaded areas represent 95% confidence intervals.

Data S1. Supporting Tables.

Akerkar A, Rønn PF, Kosjerina V, et al. Cumulative risk of diabetic foot complications in risk groups of type 1 and type 2 diabetes: Real‐world evidence from a 22‐year follow‐up study. Diabetes Obes Metab. 2025;27(4):2284‐2287. doi: 10.1111/dom.16200

DATA AVAILABILITY STATEMENT

The data analysed in this study is available on Statistics Denmark. However, the data accessed was under licensing agreements for this study and is therefore not available for public use.