1 INTRODUCTION

Plantar foot ulcer development in people with diabetes and peripheral neuropathy is strongly associated with abnormal biomechanics while bearing weight.^{1, 2} These abnormal biomechanics are reflected by higher, mostly repetitive mechanical stresses, suggested to lead to callus formation or pre-ulcerative lesions, and eventually plantar foot ulceration.^{1, 2} This pathway is the most commonly reported mechanism of diabetes-related plantar foot ulcers.^{1, 2} Despite 18.6 million people worldwide living with a diabetes-related foot ulcer and 130 million at risk,³ our understanding of the precise role of mechanical stress in plantar foot ulceration remains limited. To help prevent plantar foot ulcers and reduce their impact on people and society, a better understanding of their aetiology is needed.⁴

Peak plantar pressure is most commonly used to assess foot biomechanics in people with diabetes. While high barefoot peak plantar pressure is a proven risk factor for plantar foot ulcers,⁵ it is only a moderate predictor.⁶ Other foot-loading factors, including in-shoe plantar pressure, shear stress, weight-bearing activity and adherence to wearing footwear, have limited or no predictive value in foot ulcers.^{4, 5} However, the total mechanical stress that the foot endures is the interplay of these foot-loading factors.^{4, 7, 8} For example, if someone with high barefoot peak plantar pressures uses their pressure-reducing footwear at all times, total mechanical stress is lower than a person with moderate barefoot peak plantar pressure who does not adhere to wearing their footwear. Such differences are captured in the ‘cumulative plantar tissues stress’ (CPTS), a concept in which all foot-loading factors are taken into account. CPTS has therefore been suggested as a better predictor for plantar foot ulceration than any individual foot-loading factor.^{4, 8} CPTS can be estimated by integrating multiple foot-loading factors in a biomechanical model. Waaijman and colleagues⁸ found a non-significant trend towards higher CPTS in people who developed a recurrent plantar foot ulcer compared to those who did not. However, in their CPTS model, they only included walking, while people with diabetes and peripheral neuropathy stand longer than they walk, affecting CPTS.⁹ However, studies to date have shown limited or no predictive value for CPTS in plantar foot ulceration, meaning that further investigation is needed.⁵ We recently developed a more comprehensive CPTS model that includes both walking and standing, better reflecting the mechanical stress on the foot.¹⁰ The mechanical pathway of plantar foot ulceration suggests that CPTS may be indirectly linked to ulceration via callus formation or pre-ulcerative lesions,^{2, 11, 12} known predictors of plantar foot ulcers.^{2, 8, 11-14} However, the relationship between CPTS, callus formation or pre-ulcerative lesions, and plantar foot ulceration remains unclear.

We primarily aimed to investigate the predictive value of CPTS alongside socio-demographic and clinical characteristics in plantar foot ulceration in high-risk people with diabetes, and hypothesised that CPTS is an independent predictor of plantar foot ulceration.^{5, 8} Secondarily, we explored the association between CPTS and abundant calluses and pre-ulcerative lesions, to for the first time experimentally test the mechanical pathway of plantar foot ulceration.^{1, 2}

2 METHODS

2.3 Cumulative plantar tissue stress

To calculate CPTS, we integrated the measures of plantar pressure at baseline, weight-bearing activity and footwear adherence over seven consecutive days directly after baseline. We used a previously published and validated CPTS model that included the pressure–time integrals of each walking stride and period of standing while barefoot and shod, as shown in equation 1 and Figure 1.¹⁰ We calculated the mean daily CPTS using custom Matlab scripts for the following five forefoot regions: metatarsal 1, metatarsal 2–3, metatarsal 4–5, hallux and digits 2–5.

$$$$ {\mathrm{CPTS}}_{\mathrm{region}}=\sum \limits_{i=1}^{\mathrm{activity}}\sum \limits_{j=1}^{\mathrm{condition}}{\left({\mathrm{PTI}}_{\mathrm{region}}\times \mathrm{activity}\ \mathrm{level}\right)}_{i,j} $$$$(1)

2.3.1 Note equation 1

‘CPTSregion’ concerns the cumulative plantar tissue stress level in the forefoot region with the highest barefoot peak plantar pressure (expressed as GPa^.s/day); ‘activity’ concerns the specific activity, either walking or standing; ‘condition’ concerns the specific plantar pressure measurement, either barefoot or in-shoe; ‘PTIregion’ concerns the pressure–time integral level in the forefoot region with the highest barefoot peak plantar pressure (expressed as kPa^.s); and ‘activity level’ concerns the activity level of walking (expressed as number of strides) and of standing (expressed as the total standing duration in seconds).

3 RESULTS

3.2 Prediction of plantar foot ulceration by CPTS

CPTS level was not significantly different between the plantar ulcer and plantar-ulcer-free groups (1.35 [IQR 0.74; 1.95] vs. 1.36 [IQR 0.89; 2.17] GPa^.s/day, respectively, p = 0.560, Figure 1). Furthermore, no significant group differences were found for the level of barefoot and in-shoe plantar pressure, weight-bearing activity and footwear adherence (all p > 0.05, Table 2).

TABLE 2. CPTS and its foot-loading factors of all participants and those who did or did not develop a plantar foot ulcer during 12-month follow-up.

	All participants (n = 60)	Participants with plantar foot ulcer (n = 22)	Participants plantar-ulcer-free (n = 38)	p-value
CPTS (GPa.s/day)	1.36 [IQR 0.86; 2.12]	1.35 [IQR 0.74; 1.95]	1.36 [IQR 0.89; 2.17]	0.560
Barefoot plantar pressure
Walking PTI (kPa.s)	444 [IQR 251; 669]	537 [IQR 244; 799]	384 [IQR 239; 644]	0.125
Standing PTI (kPa.s)	186 [IQR 141; 300]	185 [IQR 139; 335]	189 [IQR 138; 278]	0.866
Walking PPP (kPa)	919 [IQR 611; 1207]	1060 [IQR 709; 1260]	889 [IQR 594; 1179]	0.249
Standing PPP (kPa)	186 [IQR 141; 300]	185 [IQR 139; 335]	189 [IQR 138; 278]	0.866
In-shoe plantar pressure
Walking PTI (kPa.s)	79 [IQR 66; 97]	82 [IQR 70; 114]	77 [IQR 59; 92]	0.177
Standing PTI (kPa.s)	70 [IQR 48; 100]	71 [IQR 51; 101]	70 [IQR 47; 100]	0.613
Walking PPP (kPa)	192 [IQR 151; 251]	188 [IQR 169; 232]	194 [IQR 138; 259]	0.890
Standing PPP (kPa)	79 [IQR 58; 111]	87 [IQR 58; 108]	78 [IQR 56; 112]	0.707
Weight-bearing activity (per day)
Number of steps	5087 [IQR 2699; 7916]	4508 [IQR 1812; 6915]	5745 [IQR 3389; 8942]	0.149
Day-to-day variation in stepsa	1674 [IQR 911; 2717]	1783 [IQR 862; 2529]	1669 [IQR 913; 2953]	0.895
Standing duration (min)	119 [IQR 82; 140]	113 [IQR 71; 141]	121 [IQR 83; 145]	0.500
Adherence to wearing footwear (% per day)
Walking	71 [IQR 54; 81]	71 [IQR 59; 75]	71 [IQR 53; 82]	0.759
Standing	65 [IQR 51; 78]	63 [IQR 53; 76]	65 [IQR 51; 79]	0.976

• Note: Continuous data are median [interquartile range (IQR) 25th interquartile; 75th interquartile].
• Abbreviations: CPTS, cumulative plantar tissue stress; PTI, pressure–time integral; PPP, peak plantar pressure.
• ^a Calculated as the standard deviation of the number of steps per day using data from all valid days.

Together with the level of CPTS, ‘time since last ulcer healed’, ‘presence of pre-ulcer’ and ‘self-selected walking speed in the laboratory’ were included in the regression analysis (based on significant group differences found, Table 1, and sufficiently low correlations and variance inflation factors, Appendix S6). The baseline characteristic ‘use of walking aid’ was excluded due to high correlation with walking speed (r = −0.57). The interaction terms were not included in the analysis (all p > 0.05).

CPTS (in GPa^.s/day) was not a significant predictor of plantar foot ulceration (odds ratio (OR) = 0.90 (95% confidence interval (CI): 0.50–1.59), Beta coefficient (β) = −0.11 (standard error (SE) = 0.29), p = 0.705). Significant predictors of plantar foot ulceration were pre-ulcer presence (OR = 9.97 (95%CI: 1.41–70.65), β = 2.30 (SE = 1.00), p = 0.021) and walking speed (in m/s) (OR = 0.01 (95%CI: 0.00–0.32), β = −4.43 (SE = 1.67), p = 0.008). The AUC was 0.82 (SE = 0.06), indicating good discriminative ability (Figure 2).

The regression analysis performed using CPTS in the forefoot region, selected based on clinical history and showed comparable results (Appendix S2).

4 DISCUSSION

In 12 months, 22 of the 60 participants developed a plantar foot ulcer, a recurrence rate of 37% that is consistent with previous studies.¹ Contrary to our hypothesis, the level of CPTS at baseline did not predict plantar foot ulceration during the 12-month follow-up. This shows that ulcers can develop in high-risk individuals with low CPTS, while others remain ulcer-free despite high CPTS. Significant predictors of plantar foot ulceration were pre-ulcer presence and a lower walking speed—both assessed at baseline—the latter being a novel finding. Additionally, at baseline, CPTS was significantly associated with pre-ulcer presence in the forefoot region. Our findings increase our understanding of plantar foot ulcer development in high-risk people with diabetes, provide new evidence for the mechanical pathway of plantar ulceration and show two easy-to-obtain strong parameters for predicting plantar foot ulceration.

CPTS is more representative of the mechanical load on the foot than the individual foot-loading factors that determine CPTS because it combines plantar pressure, weight-bearing activity and footwear adherence to represent cumulative mechanical load.^{4, 7, 8} However, its use did not improve ulcer prediction. Our findings align with Waaijman and colleagues,⁸ who found no significant difference in CPTS between people who developed a plantar foot ulcer and those who did not. Possible explanations for the lack of predictive value of CPTS may include limitations in CPTS-model accuracy. Our CPTS model did not include shear stress because validated measurement equipment does not exist, despite its suggested role in plantar foot ulceration.⁴ Secondly, we modelled CPTS based on 1-week measurements. The lack of continuous measurement of CPTS and dependency on models may affect representation of the true and current mechanical stress on the foot.⁴ While smart systems for continuous plantar pressure monitoring exist,²³ their validity and reliability are still unknown; spatial resolution is low, and pressure is not measured directly at the skin.²⁴ Thirdly, we did not measure load-bearing capacity, that is the ability of the skin and underlying soft tissue to tolerate mechanical stress without breaking down.²⁵ The load-bearing capacity differs among individuals and may affect the association between CPTS and ulceration.^{26, 27} While regional plantar pressures, pre-ulcers (i.e. tissue damage) and lower walking speed (reflecting lower general health) may act as surrogates of load-bearing capacity, direct assessment will likely improve ulcer prediction. Lastly, latent aspects within the foot-loading factors may influence CPTS or ulceration, which traditional statistical techniques cannot identify; machine learning techniques may help to identify such associations.

Pre-ulcer presence and lower walking speed at baseline were strong predictors of plantar foot ulceration (ORs of 10 and 0.01, respectively). The AUC was 0.82, which is high for just two parameters. In comparison, previous models using five biomechanical and clinical predictors and seven easy-to-obtain clinical predictors found AUCs of 0.68 and 0.66, respectively.^{13, 14} Pre-ulcer presence was a predictor in these models.^{13, 14} A lower walking speed was found in people with diabetes-related foot ulcers compared to people without ulcers or diabetes.²⁸ However, walking speed was not considered in previous prediction models, making it a novel factor strongly increasing predictive power. Lower walking speed reflects a lower general health or pre-frailty status in older adults,²⁹ suggesting that people with diabetes at high ulcer risk who walk slower may be physically more deconditioned and frailer, increasing the vulnerability of their plantar soft tissue to break down.²⁶ An increase in weight-bearing activity during follow-up may potentially increase the risk of ulceration due to reduced tolerance of plantar tissue to mechanical loading; however, the relationship between weight-bearing activity and foot ulceration is not straightforward and requires further investigation, including assessment of plantar tissue tolerance, as we have discussed in a recent review.⁵ Our findings highlight pre-ulcers and a lower walking speed as important clinical markers for identifying those at highest risk for plantar foot ulceration. Clinically, walking speed and pre-ulcer presence should be assessed, with pre-ulcers closely monitored and treated to improve prediction and prevention of plantar foot ulcers in high-risk people.

Higher CPTS levels were significantly associated with pre-ulcer presence in forefoot regions at baseline. This was expected, as pre-ulcers have been reported to develop as a result of mechanical stress; however, this had not yet been experimentally proven.^{1, 2} Our combined findings of an association between CPTS levels and pre-ulcers, and pre-ulcers being a strong predictor of plantar foot ulceration, support the mechanical pathway of diabetes-related foot ulceration.^{1, 2} While providing valuable insights, these results do not yet imply causation, as CPTS and pre-ulcers were assessed cross-sectionally. Future studies should include pre-ulcers as outcomes over time, to further support the mechanical pathway of plantar foot ulceration.

A strength of our study was the objective measurement of all foot-loading factors using validated equipment and the most comprehensive CPTS model available, given the available equipment.¹⁰ However, measuring foot-loading parameters and calculating CPTS is challenging and labour-intensive because it requires multiple sensors, complex algorithms and technically skilled people, limiting its clinical application.³⁰ By contrast, the strong predictors—walking speed and pre-ulcer presence—are both easily accessible in any foot care practice worldwide. Lastly, this study focused on recurrent plantar foot ulcers in high-risk people with diabetes, where known predictors like neuropathy and ulcer history were not included, potentially affecting CPTS's predictive power. Larger studies are needed to assess CPTS's predictive power in first-ever plantar foot ulceration and should be initiated in the field.

In conclusion, our study showed that baseline CPTS did not predict plantar foot ulceration during 12-month follow-up in high-risk people with diabetes, implying a lack of direct association. The presence of pre-ulcers and a lower walking speed at baseline were, however, strong predictors of plantar foot ulceration, the latter being a novel finding. In addition, CPTS was associated with pre-ulcer presence, providing support for the mechanical pathway of plantar foot ulceration via pre-ulcers, although causality requires further investigation. Our results emphasise the importance of assessing walking speed and early identification and treatment of pre-ulcers to better predict and prevent plantar foot ulcers in people with diabetes at high ulcer risk.

AUTHOR CONTRIBUTIONS

According to the CRediT statement: C.M.H., J.J.v.N., M.P. and S.A.B. contributed to the conceptualisation; C.M.H., J.J.v.N., M.P. and S.A.B. contributed to the methodology; C.M.H. contributed to the software; C.M.H., J.J.v.N., M.P. and S.A.B. contributed to the validation; C.M.H., J.J.v.N., M.P. and S.A.B. contributed to the formal analysis; C.M.H. and J.J.v.N. contributed to the investigation; C.M.H., J.J.v.N., T.E.B.W., L.W.E.S. and E.J.G.P. contributed to the resources; C.M.H. and J.J.v.N. contributed to the data curation; C.M.H. contributed to writing—original draft; J.J.v.N., T.E.B.W., L.W.E.S., E.J.G.P., M.P. and S.A.B. contributed to writing—review and editing; C.M.H. contributed to visualisation; J.J.v.N., M.P. and S.A.B. contributed to the supervision; C.M.H. and J.J.v.N. contributed to the project administration; J.J.v.N., T.E.B.W., L.W.E.S., E.J.G.P., M.P. and S.A.B. contributed to the funding acquisition. All authors approved the final version of the manuscript. J.J.v.N. 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.

FUNDING INFORMATION

This study was supported by research grants from the Amsterdam Movement Sciences research institute and ZGT Wetenschapsfonds.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

ARTIFICIAL INTELLIGENCE-ASSISTED TECHNOLOGIES

The authors declare that they have not used any type of generative artificial intelligence for the writing of this manuscript, nor for the creation of images, graphics, tables or their corresponding captions.