Associations of Glycemic Control and Diabetes Duration With Dupuytren Disease in Men and Women With Type 1 and 2 Diabetes

Abstract

Background:

Dupuytren disease (DD) is a benign fibroproliferative disorder affecting the hand. Although diabetes mellitus is a known risk factor, the underlying mechanisms behind this association remain unclear. This study aimed to examine the relationship between glycemic control and DD in type 1 (T1D) and type 2 diabetes (T2D), and to identify other metabolic risk factors influencing DD risk.

Methods:

In this retrospective registry study, data from the Swedish National Diabetes Register and the Skåne Healthcare Register were cross-linked. In total, 96,039 individuals aged 18 years or older with T1D or T2D were included. Sex-stratified, multivariable logistic regression models calculated associations between HbA1c levels and DD risk. Interaction analyses evaluated whether diabetes duration modified the association between HbA1c levels and DD risk.

Results:

Longer diabetes duration consistently increased the risk of DD in both T1D and T2D groups. A trend toward increased DD risk with higher HbA1c levels was seen in T1D (P > 0.05). Higher body mass index was inversely associated with DD in men and women with T2D (P < 0.05). No interaction was observed between HbA1c levels and diabetes duration.

Conclusions:

Diabetes duration seems to be a strong and independent risk factor for DD in T1D and T2D. Although a trend toward higher DD risk with elevated HbA1c was observed in T1D, no interaction with diabetes duration was found. A higher body mass index was associated with a lower risk of DD in individuals with T2D.

Takeaways

Question: Do glycemic control and other metabolic factors influence the risk of developing Dupuytren disease (DD) in individuals with type 1 (T1D) or type 2 diabetes (T2D)?

Findings: In this large-scale, retrospective registry study, longer diabetes duration was consistently associated with a higher DD risk in T1D and T2D. Poor glycemic control increased the risk only in men with T1D.

Meaning: Diabetes duration, independent of glycemic control, seems to increase DD risk, highlighting the role of cumulative metabolic exposure in DD development among individuals with diabetes.

INTRODUCTION

Dupuytren disease (DD) is a benign fibroproliferative disease primarily affecting the palmar fascia. Over time, progressive collagen cord formation in the palmar fascia may lead to flexion deformities, most commonly affecting the ring and little fingers.^1,2 Although various treatment options exist, including nonsurgical and surgical methods, no cure is currently available.³

DD has a multifactorial etiology involving both genetic and environmental factors. A significant number of genes potentially involved in the development of the disease have been identified, and a recent Danish twin study suggests a heritability of 80%.^4–6 Risk factors include male gender, age, diabetes mellitus (DM), epilepsy, liver disease, cigarette smoking, alcohol consumption, manual labor, and hand trauma.^7,8 In contrast, individuals with a higher body mass index (BMI) and blood lipid levels have been shown to have a lower risk of developing DD.^9–11 Furthermore, genetic correlations have been observed between DD and BMI, type 2 diabetes (T2D), and triglycerides (TG), indicating a shared genetic etiology among these traits.¹²

DD is part of the “diabetic hand” syndrome together with other conditions more commonly affecting individuals with diabetes, including trigger finger and carpal tunnel syndrome.¹³ However, the impact of glycemic control on DD risk in individuals with type 1 diabetes (T1D) and T2D remains poorly studied. Therefore, the aim of this study was to investigate the relationship between glycemic control and DD risk, along with other metabolic factors, including diabetes duration, that may influence the development of DD in individuals with T1D and T2D.

METHOD

Study Design and Data Sources

This retrospective register study cross-linked data from the National Diabetes Register (NDR) and the Skåne Healthcare Register (SHR) collected during 2004–2019.

The NDR collects a wide range of data, including type of DM, diabetes duration, HbA1c levels, diabetic medications, blood lipid levels, BMI, smoking, and physical activity. Data are generally collected annually, in primary care for T2D and specialist care for T1D. As of 2019, when data for this study were gathered, the NDR covered approximately 87% of Sweden’s DM patients aged 18 years and older.¹⁴

SHR is an administrative healthcare register, covering the entire population in the region of Skåne in southern Sweden (1.4 million in 2019).¹⁵ Since 2004, the SHR has included records from a majority of all healthcare visits in both public and private care. Healthcare providers are required to record diagnoses using the International Classification of Diseases (ICD) codes. ICD codes are automatically transferred to the SHR from the health record systems. Registration has been mandatory for reimbursement since 2004, contributing to an increase in assigned diagnoses from that year onward.¹⁶

By linking individual personal identification numbers, cases of DD recorded in the SHR during 2004–2019 were identified using the ICD-10 code for DD (M72.0). These cases were then cross-linked with data from the NDR collected within the same time period. Only first-time diagnoses of DD, occurring after or during the same year as each individual’s first recorded DM diagnosis, were included (Fig. 1).

Fig. 1.

Study population selection flowchart.

Quantitative Data in the NDR

Quantitative data from the NDR were analyzed for the period of 2004–2019. HbA1c values (mmol/mol), triglyceride levels (mmol/L), and BMI (kg/m²) were calculated as the average of all recorded values across the years of registration in the NDR. Smoking status and participant age were recorded based on data from the participant’s last registry entry in the NDR. Diabetes duration was calculated by subtracting the year of the first recorded diabetes diagnosis from the year of the last registry in the NDR.

Statistical Methods

All analyses were stratified by sex and conducted separately for T1D and T2D. The analyses were performed with Stata, version 18. A P value of less than 0.05 was considered significant.

Baseline characteristics for individuals with T1D and T2D were stratified by sex and by the presence or absence of DD. Data were reported as mean ± SD for normally distributed data and as median ± interquartile range for skewed data. To evaluate associations between HbA1c exposure and DD, 2 sex-stratified multivariable binary logistic regression analyses were performed separately for patients with T1D and T2D. In accordance with the American Diabetes Association’s 2024 glycemic control guidelines, individuals were stratified by HbA1c into 3 groups: optimal glycemic control (<53 mmol/mol), intermediate glycemic control (53–64 mmol/mol), and poor glycemic control (>64 mmol/mol).¹⁷ These groups were then analyzed in the logistic regression models with the optimal glycemic control group serving as the reference category. Model 1 included adjustments for age (continuous) and HbA1c categories. Model 2 included additional adjustments for BMI (continuous), diabetes duration (continuous), triglyceride levels (continuous), and smoking status (binary).

To assess for potential effect modification (interaction) of the association between HbA1c categories and DD by diabetes duration categories, a hierarchical modeling approach was applied. A base multivariable logistic regression model was first specified, including HbA1c levels (categorized), diabetes duration (dichotomized), and age (continuous) as predictors. Diabetes duration was dichotomized using median values within the T1D and T2D populations, resulting in cutoff values of 37 years for T1D and 14 years for T2D. To assess whether the effect of HbA1c on DD was modified by diabetes duration, an interaction term was added to the model. We then compared the models to evaluate changes in discriminatory accuracy (DA). After regression analyses were conducted in each model, the predicted probability of DD was computed and used to obtain the receiver operating characteristics curve and the area under the receiver operating characteristics curve (AUC).¹⁸ The AUC is a measure of DA and indicates the ability of a model to correctly discriminate between individuals who develop DD and individuals who do not. The larger the AUC, the greater the DA. The AUC can take a value between 0.5 and 1, with 1 representing perfect discriminative accuracy and 0.5 indicating no discriminative accuracy at all. Using the criteria proposed by Hosmer,¹⁹ the DA was classified as absent or very small (0.50 ≤ AUC ≤ 0.60), small (0.60 < AUC ≤ 0.70), large (0.70 < AUC ≤ 0.80), or very large (AUC > 0.80). The different AUCs were then compared to evaluate the DA of the models.

Ethics Approval

This study was approved by the National Ethics Committee (DNR: 2019-02042) and conducted in accordance with the Declaration of Helsinki.

Informed Consent

Not applicable in this register study.

RESULTS

Study Selection

The study selection process is illustrated in Figure 1. The initial study population consisted of 99,901 patients registered in the NDR. Of these, 4464 patients were excluded due to having a diabetes type other than T1D or T2D, and 398 patients were excluded due to a baseline diagnosis of DD. An additional 20,396 individuals were excluded due to missing data. The final analysis included 8085 individuals with T1D and 66,558 individuals with T2D.

Baseline Characteristics

Baseline characteristics are presented in Table 1. Among the final study population, 91 women and 162 men with T1D, and 214 women and 747 men with T2D, were diagnosed with DD during the 16-year time period. In patients with T2D, mean HbA1c levels were slightly higher, and BMI values were lower in both men and women with DD compared with those without DD. Age and diabetes duration were consistently higher in both T1D and T2D patients with DD. Additionally, men with T1D and DD had higher triglyceride levels, whereas men with T2D and DD had lower triglyceride levels than those without DD.

Table 1.

Baseline Characteristics for Individuals With and Without DD, Stratified by Sex and Diabetes Type

Baseline Characteristics
	T1D		T2D
	No DD	DD	No DD	DD
Women	n = 3497	n = 91	n = 26,988	n = 214
HbA1c, mmol/mol	64 ± 12	64 ± 12	53 ± 11	55 ± 11
Age, y	52 ± 19	64 ± 12	71 ± 13	75 ± 10
BMI, kg/m2	26 ± 5	26 ± 4	31 ± 6	29 ± 5
Triglycerides, mmol/L	0.9 (0.5)	0.9 (0.5)	1.6 (1.0)	1.6 (1.0)
Smoker, n (%)	464 (13)	6 (7)	3600 (13)	15 (7)
Diabetes duration, y	22 (23)	39 (20)	9 (10)	14 (12)
Men	n = 4335	n = 162	n = 38,609	n = 747
HbA1c, mmol/mol	62 ± 13	64 ± 10	54 ± 12	55 ± 11
Age, y	50 ± 18	64 ± 11	70 ± 12	74 ± 9
BMI, kg/m2	26 ± 4	26 ± 3	30 ± 5	28 ± 4
Triglycerides, mmol/L	1.0 (0.6)	1.1 (0.5)	1.6 (1.1)	1.5 (1.0)
Smoker, n (%)	652 (15)	17 (10)	5800 (15)	94 (13)
Diabetes duration, y	20 (22)	37 (20)	9 (9)	15 (12)

Data are presented as mean ± SD, n (%), or median (interquartile range).

Logistic Regression Analyses

Logistic regression analyses adjusted for age are presented in Table 2, and analyses with additional adjustments are presented in Table 3. In patients with T1D, men with poor glycemic control had an increased risk of developing DD (P < 0.01), whereas no significant associations were observed between HbA1c levels and DD risk in women with T1D or in men with T2D. Regarding BMI and DD risk, a low but significant inverse relationship was observed in men (P < 0.001) and women (P < 0.05) with T2D but not in T1D. Furthermore, diabetes duration was associated with an increased risk of DD across all groups (P < 0.001).

Table 2.

Sex-stratified Binary Multivariable Logistic Regression for DD by Glycemic Control in T1D and T2D

Model 1*
	T1D		T2D
	OR (95% CI)	P	OR (95% CI)	P
Women
Optimal glycemic control	Reference	—	Reference	—
Intermediate glycemic control	1.21 (0.62–2.39)	0.57	1.60 (1.18–2.16)	<0.01
Poor glycemic control	1.45 (0.76–2.76)	0.26	1.68 (1.17–2.42)	<0.01
Age, y	1.04 (1.02–1.05)	<0.001	1.02 (1.01–1.03)	<0.001
AUC	0.69		0.59
Men
Optimal glycemic control	Reference	—	Reference	—
Intermediate glycemic control	1.84 (1.09–3.11)	<0.05	1.38 (1.17–1.62)	<0.001
Poor glycemic control	2.31 (1.39–3.84)	<0.01	1.20 (0.97–1.47)	0.09
Age, y	1.05 (1.04–1.06)	<0.001	1.04 (1.03–1.04)	<0.001
AUC	0.73		0.62

Data are presented as median (interquartile range).

^*Adjusted for age; boldface indicates P < 0.05.

CI, confidence interval; OR, odds ratio.

Table 3.

Sex-stratified, Fully Adjusted Logistic Regression Models for DD by Glycemic Control in T1D and T2D

	Model 2*
	T1D		T2D
	OR (95% CI)	P	OR (95% CI)	P
Women
Optimal glycemic control	Reference	—	Reference	—
Intermediate glycemic control	1.14 (0.58–2.25)	0.71	1.33 (0.97–1.82)	0.08
Poor glycemic control	1.45 (0.76–2.79)	0.26	1.29 (0.87–1.90)	0.21
Age, y	1.03 (1.01–1.04)	<0.001	1.01 (1.00–1.02)	0.06
BMI, kg/m2	1.00 (0.95–1.05)	0.89	0.97 (0.94–0.99)	0.02
Diabetes duration	2.89 (1.84–4.52)	<0.001	2.02 (1.50–2.71)	<0.001
Triglycerides	0.90 (0.61–1.33)	0.59	1.06 (0.93–1.20)	0.38
Smoker	0.52 (0.22–1.20)	0.13	0.51 (0.30–0.86)	0.01
AUC	0.74		0.64
Men
Optimal glycemic control	Reference	—	Reference	—
Intermediate glycemic control	1.65 (0.97–2.80)	0.06	1.09 (0.97–1.82)	0.30
Poor glycemic control	2.04 (1.22–3.41)	<0.01	0.88 (0.71–1.10)	0.27
Age, y	1.04 (1.03–1.05)	<0.001	1.02 (1.01–1.03)	<0.001
BMI, kg/m2	1.01 (0.97–1.06)	0.60	0.95 (0.93–0.96)	<0.001
Diabetes duration	2.28 (1.62–3.22)	<0.001	2.66 (2.27–3.12)	<0.001
Triglycerides	1.03 (0.84–1.26)	0.77	1.1 (0.94–1.09)	0.72
Smoker	0.65 (0.39–1.09)	0.10	0.81 (0.65–1.00)	0.06
AUC	0.76		0.69

Data are presented as median (interquartile range).

^*Adjusted for age, BMI, diabetes duration, triglycerides, and smoking status; boldface indicates P < 0.05.

CI, confidence interval; OR, odds ratio.

Finally, interaction terms between HbA1c and diabetes duration were included to test for effect modification in T1D and T2D, and no significant interactions were found. (See table, Supplemental Digital Content 1, which displays the sex-stratified, multivariable logistic regression models for DD in T1D by glycemic control categories and diabetes duration. Model 2 includes interaction terms between HbA1c and diabetes duration categories. All models are adjusted for age, and boldface indicates P < 0.05, https://links.lww.com/PRSGO/E616.) (See table, Supplemental Digital Content 2, which displays the sex-stratified, multivariable logistic regression models for DD in T2D by glycemic control categories and diabetes duration. Model 2 includes interaction terms between HbA1c and diabetes duration categories. All models are adjusted for age, and boldface indicates P < 0.05, https://links.lww.com/PRSGO/E617.)

DISCUSSION

The findings in this study highlight the cumulative exposure to duration-related mechanisms in individuals with diabetes as a risk factor for DD. Longer diabetes duration consistently increased the risk of DD in individuals with both T1D and T2D, regardless of age or glycemic control. These findings align with prior studies identifying diabetes as a risk factor for DD, including a meta-analysis of 21 studies that reported an odds ratio of 3.1 compared with individuals without diabetes.²⁰ However, this study added new insights by highlighting the associations of diabetes duration and glycemic control in the development of DD.

The mechanisms by which hyperglycemia and diabetes duration contribute to DD remain incompletely understood. One theory is that hyperglycemia in individuals with DM leads to increased levels of advanced glycation end products (AGEs). AGEs are believed to drive DD progression by inducing chronic inflammation, oxidative stress, and the transformation of fibroblasts into myofibroblasts, driving excessive ECM deposition in the palmar fascia of the hand.^21,22 In support of this, AGEs have been found at elevated levels in biopsies of the palmar fascia from patients with DD.²³ Moreover, AGE accumulation has been shown to result from both hyperglycemia and prolonged diabetes duration independent of glycemic control.^22,24 In the present study, hyperglycemia alone, independent of diabetes duration, was not consistently associated with increased DD risk across all subgroups. Although trends toward higher DD risk with increasing HbA1c levels were observed in both men and women with T1D, a significant association was found only in men with poor glycemic control in T1D, and this association remained significant even after full adjustments. In women with T2D, a significant association with increased DD risk was observed in the intermediate and poor glycemic control groups in the age-adjusted model, but this significance disappeared after adjusting for additional variables.

To evaluate whether diabetes duration modified the effect of glycemic control on DD risk, interaction terms between HbA1c and diabetes duration were included in the regression models. These analyses showed no significant interaction in either T1D or T2D, suggesting that the effect of glycemic control on DD risk did not vary by duration of diabetes. Instead, diabetes duration seemed to be a strong and independent risk factor across all subgroups, reinforcing its likely central role in DD development and progression.

In line with our findings, Arkkila et al²⁵ reported that longer diabetes duration rather than glycemic control increased the risk of developing DD. In contrast, other studies have reported associations between elevated HbA1c levels and increased prevalence or risk of DD.^10,26 However, these studies did not adjust for diabetes duration in their analyses, which may have influenced their results. This potential confounding factor is addressed in the current study, where the association between elevated HbA1c levels and increased DD risk disappeared in the logistic regression model after adjusting for diabetes duration.

Moreover, T1D and T2D have different underlying disease mechanisms, which potentially could influence DD risk differently. T1D is primarily characterized by autoimmune destruction of insulin-producing cells, leading to an absolute insulin deficiency.²⁷ In contrast, T2D is associated with insulin resistance and, often, hyperinsulinemia, which has been linked to pro-fibrotic effects that could contribute to DD progression differently in individuals with T2D.^27,28

In summary, these findings suggest that although glycemic control may influence DD risk, particularly in certain subgroups, diabetes duration seems to be a more consistent and stronger risk factor for the development of DD. This stronger association may reflect the cumulative effects of metabolic stress and collagen glycation over time.

Body Mass Index

This study found a small but inverse relationship between BMI and DD risk in individuals with T2D, where lower BMI was associated with a higher risk of developing DD. Although this relationship is not widely documented in individuals with DM, the findings align with previous studies indicating that higher BMI may lower the risk of developing DD in the general population.^9–11,29,30

One possible explanation for this association involves the role of sex hormones, given the higher prevalence of DD in men and the hormonal effects of adipose tissue. Increased BMI is causally linked to lower serum testosterone levels in men, which in turn has been associated with a reduced risk of DD.^11,31 In addition to its effect on testosterone, adipose tissue produces estrogen, which has demonstrated protective properties against various fibroproliferative conditions, further suggesting that hormonal mechanisms related to BMI may play a role in DD risk.³²

It is possible that the reduced risk of DD associated with higher BMI only becomes significant above a certain BMI value. Because individuals with T1D, including those with T1D in this study, generally have lower BMI compared with individuals with T2D, they may not reach this BMI value. As a result, this could have led to a type II error, limiting the ability to identify an association between BMI and DD in individuals with T1D, even if one exists.

Moreover, genetic correlations have been observed among DD, BMI, and T2D, suggesting a shared genetic etiology among these traits.¹² Together, these genetic associations, along with differences in metabolic profiles, could explain why an inverse relationship between BMI and DD was not observed in individuals with T1D.

Strengths and Limitations

This study has notable strengths, including its large, high-quality datasets from national and regional registers, providing extensive clinical data during a 16-year time period. The large total study population in this study increases the generalizability of findings within the Swedish population. However, the generalizability of findings to populations with different demographic characteristics may be limited, as data on race and ethnicity were not available in the SHR or NDR. The large number of included individuals further allowed for stratification by sex in addition to diabetes types, which has not been examined in previous research.

Despite the large total study population, certain subgroups, particularly those with DD and T1D, were relatively small. This may limit statistical power in subgroup analyses and increase the likelihood of type II errors. As a retrospective population-based cohort study, all eligible individuals were included, and therefore, a formal power analysis was not performed. Moreover, despite adjustments for multiple covariates, residual confounding may remain due to unmeasured factors that have previously been related to DD. Furthermore, data on blood glucose levels and the frequency of glucose monitoring were not available, and the use of mean annual HbA1c levels, without accounting for short-term fluctuations in glycemic control, might reduce the accuracy of associations between HbA1c and DD risk. Temporary variations in glycemic control that are not captured by long-term averages could mask or alter the relationship with DD. In addition, serum testosterone levels were not available in the NDR, limiting the ability to directly assess hormonal factors as potential contributors to DD risk.

The use of population-based data from the SHR minimizes selection bias, whereas high-quality exposure measures from the NDR reduce the risk of exposure misclassification. Outcome measures from the SHR included ICD-coded diagnoses from both primary and specialist care. As primary care doctors handle fewer DD cases, misclassification may occur, but rates are expected to be consistent across both groups of T1D and T2D.³³ Data on DD severity or treatment-specific information for DD were not available in the data extracted from the SHR, limiting the ability to differentiate between mild and severe forms of DD or to assess whether glycemic control is associated with more clinically significant outcomes. Nevertheless, as patients with DD in this study had sought medical care, it is likely that the identified cases represent clinically relevant disease. This assumption is further supported by previous SHR-based research showing that, in datasets including treatment codes, the majority of patients with DD underwent treatment during the study period.³⁴

Finally, the study’s retrospective, observational design limits the ability to evaluate causality, as it demonstrates correlations rather than causal relationships. To demonstrate causality, randomized controlled trials are the gold standard.³⁵ However, implementing a randomized controlled trial in the context of this study would present significant ethical and practical challenges. Alternatively, Mendelian randomization provides another way to assess causal relationships by using genetic variants linked to exposures.³⁶

In conclusion, this study has notable strengths, including its large, high-quality datasets from national and regional registers, providing extensive clinical data during a 16-year time period, and enabling separate analyses for both sex and diabetes types. However, limitations related to confounding, study design, and the inability to prove causal relationships should still be considered when interpreting results.

Areas for Future Research

Based on the findings from this study, future studies should explore the role of AGEs in the development of DD, particularly their accumulation over time and their contribution to DD development independent of glycemic control. Investigating fluctuations in glucose levels could also provide valuable insights into the relationship between glycemic variability and DD risk. Furthermore, studies with larger study populations and a broader distribution of BMI values may help clarify the observed trend toward higher DD risk with increased HbA1c levels in individuals with T1D and examine whether higher BMI is associated with a decreased risk of DD in individuals with T1D.

CONCLUSIONS

The findings from this study identify diabetes duration as a strong and consistent risk factor for DD in individuals with both T1D and T2D. These findings highlight the importance of diabetes duration in DD pathogenesis and suggest that metabolic and hormonal factors, such as BMI, may influence DD risk differently across diabetes types.

DISCLOSURES

The authors have no financial interest to declare in relation to the content of this article. This work was supported by the Regional Agreement on Medical Training and Clinical Research (ALF) between Region Skåne and Lund University Local (DNR: 2024-YF0007, RSID 163177) and funds from Skåne University Hospital and Lund University.