Outcomes of Routine Diabetes Screening for Patients with Hidradenitis Suppurativa

To the Editor,

Hidradenitis suppurativa (HS) is a chronic inflammatory skin disease characterized by nodules, abscesses, and scarring in intertriginous areas with an estimated prevalence of 1% in the United States (Vinkel and Thomsen 2018, Wieczorek and Walecka 2018). HS is known to be strongly associated with type II diabetes mellitus (DM); a 2017 systematic review found the prevalence of DM to be approximately three times higher in HS patients compared to unafflicted individuals (Bui et al. 2018).

An analysis of National Health and Nutrition Examination Survey data from 2011–2016 found that undiagnosed diabetes and pre-diabetes were present in 4.6% and 37.5% of individuals surveyed, respectively (Cheng et al. 2019). Given these findings and the increased rate of DM in the HS population, it is likely that many HS patients have undiagnosed diabetes and dysglycemia. However, while the American Diabetes Association (ADA) recommends screening for diabetes in patients with associated conditions such as polycystic ovarian syndrome, HS is not explicitly recognized as a risk factor (American Diabetes Association).

Early identification and referral have the potential to prevent or delay diabetes development and sequelae. Researchers have advocated for widespread diabetes screening of HS patients regardless of age or BMI, but no studies to date have investigated the results of implementing such screening (Bui et al. 2018, Ergun 2018). Routine screening is a cost-effective method of secondary disease prevention. In this current cross-sectional study, we investigate whether screening with hemoglobin A1c (HbA1c) in a dermatology clinic is useful for detecting undiagnosed dysglycemia and DM in HS patients.

All patients treated for HS at the University of North Carolina Department of Dermatology between January 2019 and July 22, 2019 were identified using a clinical registry. During this time routine HbA1c screening was performed for all HS patients without known history of diabetes. Study procedures were approved by the university’s institutional review board. Patients younger than 12 were excluded; no subjects were excluded on the basis of gender, race, or ethnicity. Data on patient demographics, prior diabetes diagnoses, and Hurley stage were collected using a standardized case report form. For patients without a prior diagnosis of diabetes who were not screened at our clinic, data from outside screening were imputed when available. Data from fasting and non-fasting glucose tests were substituted for HbA1c testing when HbA1c results were unavailable. Dysglycemic-range results were defined as 5.7–6.4 mg/dL for HbA1c and 100–125 mg/dL for fasting blood glucose. Diabetic-range results were defined as >6.4 mg/dL for HbA1c, >126 for fasting blood glucose, and >200 for non-fasting blood glucose.

Covariate selection was based on known associations of variables with HS and diabetes. Continuous variables were compared using the Mann-Whitney U Test; categorical variables were compared using Fisher’s Exact Test. For patients without known DM, logistic regression was used to obtain crude odds ratio (OR) estimates for each covariate using abnormal screening results as the response variable. All covariates were then included in a multivariate logistic regression model to obtain adjusted odds ratio estimates. All tests were two-sided and a p-value of less than 0.05 was considered statistically significant. Analyses were conducted using Stata version 15 (College Station, TX, StataCorp, 2017).

Sample demographics

256 consecutive patients seen during the enrollment period were identified. Of these, 6 nondiabetic patients were missing screening data, leaving an analyzable sample of 250 patients. Most patients were female, non-white, and obese, with Hurley Stage III disease. Sample demographics are presented in Table 1.

Table 1:

Descriptive statistics for 250 patients with HS

Sample demographics (%)		Prevalence of screening results and prior DM diagnoses (%)
		Normal screen	Dysglycemic screen	Diabetic screen	Prior DM diagnosis
Overall	250 (100.0)	151 (60.4)	42 (16.8)	7 (2.8)	50 (20.0)
Sex
Male	52 (20.8)	30 (57.7)	12 (23.1)	0 (0.0)	10 (19.2)
Female	198 (79.2)	121 (61.1)	30 (15.2)	7 (3.5)	40 (20.2)
Age group (n=250)
10–19	22 (8.8)	19 (86.4)	3 (13.6)	0 (0.0)	0 (0.0)
20–29	62 (24.8)	49 (79.0)	8 (12.9)	1 (1.6)	4 (6.5)
30–39	68 (27.2)	40 (58.8)	12 (17.7)	3 (4.4)	13 (19.1)
40–49	55 (22.0)	27 (49.1)	11 (20.0)	3 (5.5)	14 (25.5)
50–59	30 (12.0)	11 (36.7)	6 (20.0)	0 (0.0)	13 (43.3)
60+	13 (5.2)	5 (38.5)	2 (15.4)	0 (0.0)	6 (46.2)
Race (n=244)
White	90 (36.9)	60 (66.7)	10 (11.1)	4 (4.4)	16 (17.8)
Black/African American	137 (56.2)	79 (57.7)	26 (19.0)	2 (1.5)	30 (21.9)
Other	17 (7.0)	9 (52.9)	4 (23.5)	1 (5.9)	3 (17.7)
BMI category (n=249)
<25	17 (6.8)	13 (76.5)	3 (17.7)	0 (0.0)	1 (5.9)
25–29.9	53 (21.3)	36 (67.9)	6 (11.3)	2 (3.8)	9 (17.0)
30–39.9	94 (37.8)	54 (57.5)	18 (19.2)	2 (2.1)	20 (21.3)
40+	85 (34.1)	48 (56.5)	15 (17.7)	3 (3.5)	19 (22.4)
Hurley stage (n=247)
1	24 (9.7)	20 (83.3)	2 (8.3)	0 (0.0)	2 (8.3)
2	93 (37.7)	59 (63.4)	14 (15.1)	2 (2.2)	18 (19.4)
3	130 (52.6)	71 (54.6)	25 (19.2)	5 (3.9)	29 (22.3)
Averages for Continuous Variables
Age in years (SD)	37.1 (12.9)	33.5 (12.3)	39.2 (11.7)	36.9 (7.3)	46.4 (11.5)
BMI (SD)	36.0 (8.6)	34.9 (8.0)	37.9 (10.2)	39.0 (8.7)	37.6 (8.6)

Percentages shown in parentheses. For demographics calculations, columns total to 100%. For prevalence calculations, rows total to 100%.

Most patients with abnormal screens were female (37/49, 75.5%), in the 30–39.9 BMI subgroup (20/49, 40.8%), and non-white (33/47, 70.2%), with Hurley stage III disease (30/48, 62.5%). The highest proportion of patients with abnormal screening were aged 30–39 (15/49, 30.6%). All patients (n=7) who screened positive for diabetes were female.

Most established diabetics were female (40/50, 80.0%), non-white (33/49, 67.3%), and obese (39/49, 79.6%) with Hurley stage 3 disease (29/49, 59.2%).

Prevalence of abnormal screening

Of 200 patients screened, 21% (n=42) screened dysglycemic-range and 3.5% (7/200) screened diabetic-range. 20% (50/250) of patients were known diabetics (Table 1).

The age and BMI subgroups with the highest prevalence of abnormal screening were 50–59 years (6/17, 35.3%) and BMI>40 (18/66, 27.3%).

18.8% (3/16) of screened patients with BMI < 25 and 13.6% (3/22) of screened patients aged 10–19 were dysglycemic.

Mann-Whitney U testing found older age to be associated with both prior DM diagnosis (p<0.001) and abnormal screening (p=0.003).

Logistic regression

Multiple imputation was used to impute data for five patients with no recorded race and two patients with no recorded Hurley stage. No patterns of missingness were noted and it was assumed that data were missing at random. The imputed data were used to estimate odds ratios for associations with abnormal DM screening.

Univariate analysis of 200 screened patients found age (OR 1.20 for 5-year increase, 95% CI 1.05 – 1.37), BMI (OR 1.04 for 1-point increase, 95% CI 1.00 – 1.08), and Hurley stage (OR 1.81 for 1-point increase, 95% CI 1.06–3.09) to be associated with greater odds of an abnormal screen (Table 2). In the multivariate model, BMI (OR 1.05 for 1-point increase, 95% CI 1.01 – 1.10) and age (OR 1.19 for 5-year increase, 95% CI 1.03–1.37) were associated with abnormal screening.

Table 2:

Crude and adjusted odds ratios for associations with abnormal diabetes screening results. N=200; includes 49 patients with abnormal screening and 151 patients with normal screening. Multiple imputation was used to impute values for 7 patients with missing data.

Independent variables	Univariate analysis		Multivariate analysis
	Odds ratio (95% CI)	P-value	Odds ratio (95% CI)	P-value
Non-white race	1.64 (0.81 – 3.34)	0.17	1.95 (0.90 – 4.24)	0.09
5-year increase in age	1.20 (1.05 – 1.37)	0.01	1.19 (1.03 – 1.37)	0.02
1-point increase in BMI	1.04 (1.00 – 1.08)	0.02	1.05 (1.01 – 1.10)	0.01
1-point increase in Hurley stage	1.81 (1.06 – 3.09)	0.03	1.39 (0.78 – 2.47)	0.27
Female sex	0.76 (0.36 – 1.64)	0.49	0.75 (0.33 – 1.70)	0.49

In conclusion, in a population of 250 HS patients, 20% carried a prior diagnosis of diabetes; in the remaining 80%, HbA1c screening produced a meaningful number of new diagnoses of dysglycemia and DM across all age and BMI groups. Older age and higher BMI correlated with increased likelihood of an abnormal screen controlling for race, sex, and Hurley stage. However, 13.6% (95% CI 2.9% - 34.9%) of screened patients under the age of 20 and 18.8% (95% CI 4.1% - 45.7%) of patients with BMI<25 had abnormal screening, suggesting that screening based on specific age and BMI criteria may be inappropriate in this population. Study limitations include the cross-sectional design and the fact that enrolled patients were from a subspecialty clinic at a single institution, mostly seeing one provider. Our findings suggest that routine screening of HS patients for dysglycemia and DM, irrespective of age or BMI, may be warranted in dermatology clinics as it may lead to earlier diagnosis, referral and intervention of this common comorbidity.

ABBREVIATIONS USED:

HS

Hidradenitis suppurativa

DM

Type II Diabetes Mellitus

HbA1c

Hemoglobin A1c

ADA

American Diabetes Association

OR

Odds ratio