A cluster analysis for stratifying diabetic foot ulcers at baseline: Are those associated with ulcer healing and patient outcome?

Naemi, Roozbeh; Abbas, Zulfiqarali

Abstract

Background and aims: Systematic reviews of the classification systems of DFUs, investigated their association with DFU-related clinical outcomes. However, these are investigated using conventional statistics where the DFU future incident was treated as the outcome measure. It would be of value to investigate if DFUs can be grouped based on a variety of baseline measures in the first place and to investigate the association with healing outcome afterwards. This study aimed to investigate if diabetic foot ulcers can be grouped into clusters with similar baseline characteristics and if these clusters are associated with ulcer healing and clinical outcomes during follow-up.

Materials and methods: During the study period December-2021 - August-2022, 1,099 diabetic patients with 1,250 active foot ulcers were recruited from a specialist diabetic service in Dar es Salaam, Tanzania. At baseline demographics, presentation delay, ulcer location, details of ulcer, amputation history, peripheral neuropathy, peripheral arterial diseases (PAD), ulcer classification and wound size were taken. A Hierarchical Cluster Analysis was utilised to identify homogeneous groups of DFUs based on all baseline measures.

Results: Hierarchical Cluster Analysis identified that the 1,250 ulcers were distributed across three distinct clusters 1, 2 and 3 containing 449(35.9%), 269(21.5%), and 532(42.6%) ulcers respectively. The mean SINBAD score were 4.2, 5.3 and 4.5 while the median ulcer area was 4 cm2, 9 cm2 and 12 cm2 in Clusters 1, 2 and 3 respectively. Theses proportion of ulcers with SINBAD score 3, were 72.2%, 5.6% and 53.9% in Clusters 1, 2 and 3 respectively. These proportions for SIMBAD score of 4 was 24.3%, 58.4% and 44.2% and, was 0.7%, 36.1% and 1.3% for SIMBAD score of 5 across Clusters 1, 2, and 3 respectively. The proportion of ulcers with stage 3 tissue loss level, were 74.4%, 60.6% and 63.3% in Clusters 1, 2 and 3 respectively, while this proportion was 25.2%, 39.4% and 36.5% across the three clusters for stage 4 ulcers. Most ulcers in Cluster-3 were first ulcers (99.4%), compared to 53.9% in Cluster 2, and no first ulcer in Cluster-1 (0.0%). PAD prevalence was 0.4%, 98.9%, and 0.6% in Cluster 1, 2 and 3 respectively. The history of amputation was 5.1%, 4.1% and 0.0% in Cluster 1, 2 and 3 respectively. The median delay in presentation was 7, 7 and 5 days in Cluster 1, 2 and 3 respectively while the median duration of presentation to follow-up across the three clusters was 73, 78 and 87 days. The median duration of treatment was 88, 93 and 96 days in Cluster 1, 2 and 3 respectively. During the 122.5±235.4 days follow-up and treatment, 818 ulcers (65.8%) healed, and 425 ulcers (34.2%) remained unhealed. In Clusters 1, 2 and 3: the proportion of healed ulcers were 82.7%, 57.3% and 55.8%; amputation rates were 1.6%, 5.2% and 0.0% and the mortality rates due to foot ulcer were 2.0%, 3.0% and 1.1% respectively, while the mortality rate due to other causes were 2.0%, 3.0% and 0.6% respectively.

Conclusion: Three distinct DFU clusters with similar baseline characteristics in each cluster were identified. The orders in which the three clusters ranked based on wound classification at baseline did not follow either the proportion of ulcers healed, amputation or mortality rate during follow-up. Stratification methods using only baseline clinical measures may not fully predict the wound healing or patient outcome.

• Presentation Conference Type: Conference Abstract
• Conference Name: 60th EASD Annual Meeting of the European Association for the Study of Diabetes
• Start Date: Sep 9, 2024
• End Date: Sep 13, 2024
• Acceptance Date: May 29, 2024
• Online Publication Date: Sep 12, 2024
• Publication Date: 2024-09
• Deposit Date: Apr 30, 2025
• Journal: Diabetologia
• Print ISSN: 0012-186X
• Electronic ISSN: 1432-0428
• Publisher: Springer Verlag
• Peer Reviewed: Peer Reviewed
• Volume: 67
• Issue: S1
• Pages: 420-421
• DOI: https://doi.org/10.1007/s00125-024-06226-0