Risk Factors for Hand Wound Infections in People with Diabetes: A Case-control Study

Hand infection has been associated with increased morbidity in people with diabetes mellitus. PURPOSE: This study was conducted to determine risk factors for hand wound infection in patients with diabetes mellitus. METHODS: A 1:3 matched prospective case-control study was conducted from December 2006 to December 2016. All study patients were consecutively identified through the inpatient records upon admission to the University of the Philippines Manila, Philippine General Hospital (Manila, Philippines), for a hand wound infection necessitating surgical treatment and were followed until hospital discharge, wound healing, or death. Adults (≥18 years old) with diabetes mellitus for at least 6 months and with (study group) or without (control group) a hand wound infection were eligible to participate. Persons with a history of amputation or who were in a chronic debilitated state were excluded. Infection was defined as the presence of inflammation and purulent discharge. Eligible control patients were consecutively recruited from the outpatient clinics and were matched to the study patients by age (± 5 years) and gender. Demographic (eg, age, gender, education, occupation, tobacco use) and clinical data (body mass index [BMI], duration of diabetes, HbA1c levels, wound location and duration, delay in treatment, neuropathy, surgical procedures, length of hospital stay, and presence of arteriovenous [AV] fistula) were collected from patient records and entered into Excel spreadsheets for analysis. Regression analysis was performed and reported as odds ratio (OR) with 95% confidence intervals (CI). Level of significance was set P <.05. RESULTS: Participants included 30 study and 90 control patients. No significant differences between study and control patients were noted in terms of BMI, duration of diabetes, presence of peripheral neuropathy, occupation, or education. Significantly more study patients had elevated HbA1c (86 vs. 30; P = .0001), used tobacco (17 vs. 8; P = .0001), and had an AV fistula (3 vs. 0; P = .015). After multivariate analysis, HbA1c ≥48 mmoL/moL (OR = 18.8; 95% CI: 2.3-153.8; P = .006) and tobacco use (OR = 10.7; 95% CI: 3.5-32.7; P = .0001) were identified as independent risk factors for hand/upper extremity infection. CONCLUSION: Patients with diabetes who smoked or exhibited elevated HbA1c levels were at higher risk of having a hand infection. Further research and efforts to help people with diabetes stop smoking and maintain good glycemic control may help decrease the burden of hand infection. 

The prevalence of diabetes mellitus (diabetes) in all age groups among persons in India, China, Sub-Saharan Africa, Latin America, the Caribbean, and the Middle East was estimated to be 2.8% in the year 2000 and is expected to double by 2030.¹ Retrospective cohort studies^2-4 report infections of the upper extremities and the hand, but they are not commonly reported in Western countries. In 2 cohort studies that investigated the outcome of hand infections in patients with diabetes, Mann and Peacock² reported an amputation rate of 35% (7/20), and Connor et al³ (N = 50) showed an amputation rate of 14% that increased to 17.5% in hand infections with abscess formation. A 2017 prospective cohort study in North America by Sharma et al⁴ investigated the effect of diabetes in patients with hand and upper extremity infections. Some form of trauma was present as the most common mechanism of infection both in patients with (27/76, 36%) and without diabetes (65/246, 26%). The study showed that compared to persons without diabetes, patients with diabetes are more likely to undergo repeat drainage to control infection (relative risk [RR] = 2.08; 95% confidence interval [CI]: 1.52-2.86) and are more likely to undergo amputation (RR = 3.64; 95% CI: 1.46-9.11). 

Severe hand infections in persons with diabetes have been described as tropical diabetic hand syndrome.^5-7 In 2001, Abbas et al⁶ published a case-control study of 127 Tanzanian patients with diabetes (1:3 ratio) that found a body mass index (BMI) <20, peripheral neuropathy, and type 1 diabetes were risk factors for tropical diabetic hand syndrome. The clinical course usually is progressive and in certain case series was shown to have the potential to progress to gangrene of the hand or upper extremity.^6-11 Cohort studies^12-14 have shown complications such as chronic peripheral neuropathy, end-stage renal disease (ESRD), and vascular disease contributed to poor wound healing and poor prognosis in patients with hand infections. Although upper extremity infection in patients with diabetes is less common than lower extremity infection,¹¹ it can be serious, and most related studies^2,13-15 have detailed the poor outcome of hand infection in patients with diabetes. However, a recent retrospective series¹⁶ reported better outcomes, especially in persons with nonspecific infections that were confined to soft tissues and that responded to broad-spectrum, parenteral antibiotic therapy. 

A few case-control or cohort studies have attempted to identify risk factors for hand infection or factors that contribute to poor outcome.^6,14,17 Ince et al¹⁴ reported ESRD, foot infections, smoking, and peripheral neuropathy were among the factors for poor prognosis in 34 diabetic hand infections. A clinical study¹⁸ that investigated hand ulcers among persons with diabetes who were admitted for a foot infection found 10 of 17 hand ulcers were infected. Neuropathy, long-standing diabetes, poorly controlled blood sugar, minor trauma, and delayed treatment were found to be factors for hand infection in this population with diabetes and foot infections. 

In their retrospective cohort study conducted among 41 patients with severe soft tissue infections, Francel et al¹⁹ found amputation rates may be as high as 63% to control serious hand infections. More research to ascertain risk factors would facilitate prevention and early detection efforts, which may decrease the morbidity associated with the disease. 

The purpose of this study was to determine the risk factors associated with the development of hand infection in patients with diabetes mellitus.

Patient population. A case-control study design was used to identify the risk factors for developing hand and upper extremity infection in patients with diabetes mellitus. Patients were eligible for participation if they were of either gender, at least 18 years old, diagnosed with diabetes mellitus at least 6 months before the study, and had an infection of the hand or upper extremity (study group) or no infection (control). Infection was diagnosed clinically based on evidence of inflammation, redness, swelling, tenderness, erythema, and purulent discharge. All patients were identified initially through the database records of the outpatient and inpatient clinics. Patients recruited with hand infections were matched by age and gender with control patients without hand infection. An initial sample size calculation was performed to detect an odds ratio (OR) of 2 with an alpha of 0.05 and beta of 0.80 and then subjected to a case-control ratio of 1:3 to decrease the number of cases. 

Ethical considerations. The Research Ethics Board of the authors’ institution (University of the Philippines Manila, Manila, Philippines) approved the study. The investigators retained all clinical investigation records required by applicable regulations. All participants were made aware their personal data were collected and processed in accordance with data protection legislation; each participant was allocated a unique identifier to maintain confidentiality. Participant anonymity was guaranteed at all times, and all documentation relating to a participant was kept confidential. This research was conducted in accordance with the principles of good clinical practices of the World Health Organization and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Study procedure. Data were collected prospectively from December 2006 to December 2016. All study patients were consecutively identified through the inpatient records upon admission for a hand wound infection necessitating surgical treatment and were followed until hospital discharge, wound healing, or death. Control patients who fulfilled the inclusion and exclusion criteria were consecutively recruited from the outpatient clinics and were matched to the study patients by age (± 5 years) and gender. Data were extracted from the patients’ health records and included age, gender, HbA1c (mmoL/moL), occupation, education (primary, secondary), BMI, duration of diabetes (years), treatment regimen (oral hypoglycemic, insulin, none), tobacco use, peripheral neuropathy of the hand, and presence of an arteriovenous fistula (AV). The variables of interest were BMI in kg/m², blood sugar control, occupation, tobacco use, duration of diabetes in years, and presence of peripheral neuropathy using static 2-point discrimination (2PD) of the contralateral hand upon entry to the study. 

Description of variables. The following variables were dichotomized: BMI < or ≥23 kg/m², HbA1c < or ≥48 mmoL/moL, occupation (manual [eg, mechanics and construction workers] vs. office work [eg, secretaries, bank tellers, and supervisors]), tobacco use (yes or no), and peripheral neuropathy of < or ≥10 mm on static 2PD. For occupation, manual laborers were considered persons whose occupations require them to use their hands for most of their work. 

Patient outcomes. All study patients had an infected wound and were evaluated in terms of wound location, history of trauma, delay in treatment (number of days since appearance of wound before hospital admission), previous treatment, number and type of surgeries, wound culture results, time to wound healing, length of hospitalization, and outcome of treatment. 

Data analysis. Data were encoded using Microsoft Excel 2007 for Windows (Microsoft Corp, Redmond, WA) and analyzed using STATA™ version 10.0 (College Station, TX). Descriptive data for cases and controls were presented as means with their corresponding standard deviations (SD). Categorical data were presented as frequencies and were analyzed using chi-square tests and Fisher’s exact test; P was set at <.05. A univariate conditional logistic regression analysis with P ≤.1 was initially performed. Identified risk factors then were subjected to multivariate analyses; P ≤.05 was set as the level of significance using the backward elimination method. OR with a 95% CI was estimated using logistic regression analysis for the significant factors of interest. 

Patient characteristics. The 30 patients recruited with hand infections were matched by age and gender with 90 control patients without hand infection. More than 50% of the study and control patients were men. No significant differences between study and control patients were noted in terms of BMI, duration of diabetes, presence of peripheral neuropathy, occupation, and education. Significantly more study patients had elevated HbA1c (86 vs. 30; P = .0001), use tobacco (17 vs. 8; P = .0001), and had an AV fistula (3 vs. 0; P = .015). More patients in the control group were on an insulin regimen (49/90) compared with case patients (15/30), but the difference was not significant (P = .06). A summary of the patients is shown in Table 1.

Study patients. All study patients (average age 51.6 ± 8.9 [range 26–67] years) presented with an open, infected wound; 18 (60%) had a history of some form of trauma, and 26 (87%) were manual laborers. The mean duration of diabetes was 6.4 ± 7.1 (range 0–30) years; 15 (50%) were maintained on oral hypoglycemic agents. The dominant hand was involved in 15 patients, with the middle finger the most commonly involved digit (9/30 fingers) followed by index, thumb, and small finger. Infection was localized to a single digit in 23 patients, with involvement of the flexor sheath in 4 middle fingers and in 1 small finger; 3 patients had 2 or more fingers involved — 3 with whole hand involvement and 1 with involvement of up to the elbow. Sixteen (16) patients had osteomyelitis of the hand. The mean delay to treatment was 21.5 ± 15.7 (range 4–60) days. 

Culture studies. All but 1 intraoperative specimen taken for culture studies showed bacterial growth. Staphylococcus aureus was isolated in 12 specimens, 3 showed methicillin-resistant S aureus (MRSA), 8 showed gram-negative bacteria, 2 had methicillin-resistant S epidermidis, and 1 was positive for beta hemolytic strep. In total, 8 patients had polymicrobial growth. Of the 26 cultures performed, 1 had no growth. S aureus was the most common isolate (12 in 26), 3 of which were MRSA; gram-negative bacteria were isolated in 8 of the 26 culture studies. 

Study patient treatment outcomes. All study patients (ie, with infection) had to undergo surgical debridement at the time of presentation to manage the infection. An average of 2.1 surgeries were performed. Nine (9) patients had 1, 12 had 2 , 7 had 3, and 2 patients had 4 surgical procedures. All wounds were followed until healing, closure or coverage with skin graft, or patient death or discharge from the hospital. The mean time to wound healing or coverage with skin graft with healing and without infection was 7.4 ± 6.1 (range 3–24) days. One (1) patient was discharged with an open wound to be cared for by a local health center. The mean length of hospitalization was 17.8 ± 9.7 (range 5–30) days. Nineteen (19) patients had amputation of the involved digit. Of the 16 patients with osteomyelitis, 14 had amputations to control the infection. 

Amputation was performed in 14 of 30 patients on the first surgery, an additional 5 amputations in the second surgery, and a repeat amputation in 2 patients to control infection (above elbow and ray amputation) for a total of 19 amputations in 30 patients (63%).

Risk factors for hand infection. Univariate analysis identified HbA1c ≥48 mmoL/moL and tobacco use as factors for hand infection (see Table 2). After multivariate analysis, only tobacco use and an elevated HbA1c were associated with the occurrence of hand infection in patients with diabetes mellitus (see Table 3). People who smoked tobacco were 11 times more likely to have a hand infection compared with those who did not smoke (95% CI: 3.5-32.7; P = .0001). Patients with an elevated HbA1c were 19 times more likely to have a hand infection (95% CI: 2.3, 153.8; P = .006). 

In this case-control study, patients with diabetes mellitus were at higher risk of developing hand infections if they were smokers or had an elevated HbA1c ≥48 mmoL/moL. A retrospective study²³ comparing HbA1c levels in patients with ketoacidosis in known and newly diagnosed diabetes showed that poor glycemic control has been associated with diabetic ketoacidosis in new onset and known patients with diabetes. A cohort study²⁴ of 233 admitted patients with uncontrolled diabetes showed poor glycemic control through the years can serve as a surrogate marker for the predisposition to develop ketoacidosis, which was postulated to impair phagocytosis and alter white cell chemotactic ability, impairing beta cell reserve and insulin receptors through the mechanism of glucose toxicity and subsequently creating an immunocompromised state prone to infection. In turn, this will make patients with diabetes who have poor glycemic control prone to infections. This has been investigated^25,26 among 2 large databases in the United Kingdom with more than 100 000 cases where patients with diabetes were admitted for a variety of infections, including soft tissue infections of the extremities. In 2018, Critchley et al²⁵ reported persons with poorly controlled diabetes had a 4.7 (95% CI: 4.24-5.21) incidence RR of being admitted because of an infection. In 2017, Hine et al²⁶ reported a similar finding among persons with poor versus moderate glycemic control admitted to the hospital for infections of the skin and soft tissues (OR 1.78; 95% CI: 169-1.86). A large database study in Denmark of 69 318 patients by Mor et al²⁷ also showed a higher risk of hospital treated infections in patients with poor glycemic control.

Another risk factor for hand wound infection in this study was smoking. Tobacco use or smoking has been identified in an experimental study,²⁸ may be seen as a cardiovascular risk factor, and also can induce dysfunction of the endothelial lining of blood vessels. In addition to vascular endothelial cell injury, a systematic review²⁹ found smoking was associated with decreased blood flow, impaired inflammatory response, increased connective tissue degradation and decreased fibroblast wound migration. In short, smoking has been identified as risk factor for hand infections in the literature.

In the current study, most study patients presented late for (delayed) treatment and with deep soft tissue infection or involvement of bone. This delay was longer than previously published data, including a prospective study.^2,22 All study patients required surgery before the infection was controlled. This was more than previously reported^2,17,21 rates of 16% to 50%, perhaps owing to the fact patients all presented with soft tissue or bone infections at the time of consult. The average number of procedures per patient was comparable to the literature.²¹ Amputation rates reported in the literature^{2,6,16,17,19,21,22} range from 4.1% to 63%, underscoring that infections with severe soft tissue involvement may be difficult to treat and the prognosis is usually poor. The long treatment delay and the involvement of the bone in most cases during presentation were probably the reasons for the high amputation rate in the current series of patients. 

Polymicrobial growth in this study had relatively lower rate compared with the 42.9% to 55% reported in literature.^2,3,21 

Mean time from first surgery to wound healing was similar to previous studies^13,21 showing slow resolution of diabetic hand infection (8 days or longer) with an average hospital stay of 18 days. Documented factors associated with prolonged wound resolution or healing were treatment delay, diabetes,^6,13 deep space infection,^6,21 and polymicrobial infection.²¹ Several authors^2,13,18 noted delay in seeking treatment also was a factor for poor outcome. Infections that were deep, anaerobic, or polymicrobial were associated with an increased risk of amputation and prolonged hospitalization because of repeated surgeries.²¹ 

The findings of Wang et al¹⁸ were similar to the current study data; a history of minor trauma was among the factors that predisposed patients to hand infection, along with poorly controlled blood sugar, neuropathy, long duration of diabetes, and delayed treatment. 

The present study included a small population of patients with diabetes and upper extremity wound infections. In addition to the small number of participants included in the study, a primary limitation of the study was the design. To establish a strong causal relationship between the factors identified and the occurrence of the infection, prospective cohort studies need to be conducted. The sample is also biased because it only includes patients with hand infections who presented to the hospital for treatment. Future long-term, prospective studies involving a cohort of patients with diabetes who developed hand infections are needed to further illuminate this problem before definitive conclusions can be made.

The purpose of this case-control study was to determine risk factors associated with the development of hand infection in patients with diabetes mellitus. Compared to control group patients (n = 90), study group patients (n = 30) had statistically significant elevated HbA1c and more tobacco users. Smoking and an elevated HbA1c (>48 mmoL/moL) were independent risk factors for hand infection. Diabetic hand infection was associated with significant morbidity requiring more than 1 surgical procedure in 21 of 30 patients (70%). The results of this and other studies suggest clinicians must emphasize the importance of smoking cessation and maintaining good glycemic control. Patients with diabetes and known risk factors also may benefit from close follow-up and additional education to help decrease the incidence of these infections.

Dr. Estrella is a Clinical Associate Professor, Microsurgery Unit, Section of Hand and Reconstructive Microsurgery, Department of Orthopedics, University of the Philippines Manila, Philippine General Hospital; and a Research Associate Professor, ASTRO Study Group, Institute of Clinical Epidemiology, National Institutes of Health, University of the Philippines Manila. Dr. Lee is an Associate Consultant, Department of Hand and Reconstructive Microsurgery, National University Health System, Singapore; and an affiliate member, ASTRO Study Group, Institute of Clinical Epidemiology, National Institutes of Health, University of the Philippines.

Please address correspondence to: Emmanuel P. Estrella, MD, MSc, Microsurgery Unit, Section of Hand and Reconstructive Microsurgery, Department of Orthopedics, University of the Philippines Manila, 1000, Philippine General Hospital, Manila, Philippines; email: epestrella@up.edu.ph.