Role of Insulin Use and Social Determinants of Health on Non-melanoma Skin Cancer: Results From the Behavioral Risk Factor Surveillance System

Abstract

Background

Non-melanoma skin cancer (NMSC) is a frequent type of malignancy with a steadily increasing incidence rate worldwide. Although NMSC was shown to be associated with diabetes, no studies have addressed the extent to which insulin use influences the risk of NMSC in light of social determinants of health (SDOH). We conducted a quantitative study that examined the interplay between insulin use, SDOH, additional covariates, and NMSC among individuals with diabetes.

Methods

We based our analysis on the 2020 Behavioral Risk Factor Surveillance System (BRFSS), a national survey conducted yearly in the US. We performed weighted chi-squared test, logistic regression, and survival analyses on 8685 eligible participants with diabetes enrolled in the BRFSS.

Results

Kaplan Meier survival curves showed higher probability of NMSC event-free survival for participants with diabetes using insulin compared to participants with diabetes not using insulin (log-rank test P < .001). Significant associations were detected between insulin use and reduced odds of NMSC (OR .56; 95% CI: .38-.82), and decreased hazard (HR .36; 95% CI: .21-.62), along with indices of SDOH.

Conclusions

Our findings suggest that socioeconomic differences related to the healthcare system and behavioral patterns are linked to discrepancies in the use of insulin and the development of NMSC.

Keywords

behavioral risk factor surveillance system diabetes health inequity insulin use non-melanoma skin cancer social determinants of health

Background

Non-melanoma skin cancer (NMSC) is the most common human malignancy¹ that was marked over the past decades by a substantial worldwide increase in incidence and morbidity, as well as associated cost of treatment.^2-4 This type of skin cancer manifests in the form of a lump or discolored patch that develops slowly in the upper layers of the skin and presents mostly on sun-exposed skin.⁵ According to the World Health Organization (WHO), approximately 2 to 3 million new cases of NMSC are diagnosed each year globally,⁶ and in North America, this disease is characterized as being one of the most prevalent types of cancer^7,8 particularly among the aging population of White males.^8-10 The observed steady increase in its prevalence¹¹ and incidence rate⁹ over the last 3 decades marked NMSC as an important public health problem that is of global concern.

The primary risk factors for NMSC include fair skin, a history of sun damage, and extensive exposure to Ultraviolet (UV) radiation, particularly Ultraviolet B (UVB) radiation.⁷ Moreover, according to the Global Burden of Disease Study (1990-2017), with increasing age, the Disability-Adjusted Life Years (DALYs) for NMSC increase exponentially.⁹ The incidence of NMSC is also generally found higher in men than in women.¹⁰ Furthermore, results from meta-analyses suggest an association between BMI and NMSC, and between alcohol consumption and increased risk of NMSC.^12,13 Finally, while smoking has been consistently identified as a risk factor for several types of cancer,^14,15 there is evidence that current smoking is associated with a decreased risk of NSMC.^16,17

Diabetes is a chronic health condition marked by elevated levels of glucose in the blood, and over time, it can lead to severe systemic complications that cause considerable morbidity and mortality.¹⁸ It has also become a burden on the socioeconomic system,¹¹ thus emerging as a public health threat. Few studies have shown an association between diabetes and the development of certain cancers,¹⁹ including NMSC.^20,21 A retrospective cohort study in Taiwan found that participants with diabetes aged 60 or older had an increased risk of NMSC,²⁰ while record linkage studies in England found that hospitalization for diabetes was associated with an increased risk of several types of cancer, including NMSC.²¹ A number of observational randomized trials were performed to determine if insulin therapy could be associated with increased risk of cancer in individuals with diabetes.^22-24 The outcome of the results were inconsistent, some reporting that insulin therapy is associated with increased risk of cancer while others showed no relevance to cancer risk.^22-24 The conflicting results of existing studies could reflect the biological diversity of cancer. In general, the data from epidemiological studies negate the effects of insulin therapy on cancer risk in individuals with type 2 diabetes. The only trial that showed a link between insulin therapy and increased cancer mortality is the DIGAMI study, whereas all the other studies including the ORIGIN trial did not observe any increase in cancer risk in specific types that include lung, breast, prostate, colorectal or melanoma.^25,26 In addition, there is limited information regarding the association between anti-diabetic drugs or therapies and NMSC. A reduced risk of developing NMSC among individuals with type 2 diabetes using anti-diabetic medications, such as sodium-glucose co-transporter-2 (SGLT-2) inhibitors, Rosiglitazone, and Metformin was reported,^27-29 but there is a paucity of data specifically addressing the association between insulin use therapy and NMSC. In this regard, analyses of the Regenstrief Medical Record System found a lower incidence of NMSC among individuals with diabetes who use insulin compared to those who do not use insulin [Standardized Incidence Ratio (SIR) .53; 95% CI: .31-.84].³⁰ However, confounding factors such as smoking, level of education, income level, or employment status were not accounted for in these analyses.³⁰

Social determinants of health (SDOH) are a cluster of factors that are known to have a significant impact on the health and well-being of individuals. These factors include socioeconomic status, education, employment, access to healthcare, and other environmental factors. Studies have shown that SDOH contribute to the prevalence of non-communicable diseases (NCDs), including NMSC.^31-33 One example of how SDOH can affect NMSC prevalence is through racial disparities in skin cancer rates. In the United States (US), skin cancer is more prevalent among White individuals, with 20%-30% of all cases occurring among this demographic. By contrast, skin cancer rates are significantly lower among Asians (2%-4%) and Black individuals (1%-2%).³⁴ Furthermore, education has been identified as a significant predictor of skin cancer knowledge and awareness. Individuals with higher levels of education are more likely to be aware of the risks associated with exposure to UV radiation and may take precautionary steps to protect themselves from the sun. On the other hand, people with lower levels of education may not be aware of the risks or may have limited access to preventive measures, such as sunscreen or protective clothing.³⁵

Consequently, the extent to which insulin use influences the risk of NMSC in light of SDOH is yet to be determined. To address this gap in knowledge, we conducted a quantitative study that comprehensively examined the interplay between insulin use, SDOH, and additional covariates (including age, sex, BMI, smoking, and heavy alcohol consumption) on the occurrence of NMSC. Our study focused on a population of individuals with diabetes who participated in the 2020 Behavioral Risk Factor Surveillance System (BRFSS).

Methods

Study Population and Sampling

Our study population was based on the BRFSS conducted by the Centers for Disease Control and Prevention (CDC) in the year 2020 which includes information on US residents concerning their health-related risk behaviors, chronic health conditions, and use of preventative services. Data was collected through self-reported surveys from adult participants (aged 18 or above) from all over the US.³⁶

The BRFSS followed a multistage cluster design in which each participant is chosen randomly for the interview. Question about whether or not the participant had any skin cancer was asked in the “Chronic Health Condition” module that was administered in all 50 states. However, the question about the most recent type of cancer, through which those with skin cancer distinguished between whether the skin cancer was NMSC or other type of skin cancer was asked in a different module which is the “Cancer Survivorship”. In addition, the age of diagnosis with the first cancer type was also asked in this module. As for diabetes, age of diagnosis with diabetes, was included as a question in the “Chronic Health conditions” module, whereas inslin use was included in the “Diabetes” module. The Cancer Survivorship and Diabetes modules were not administered in all 50 states, as a result, we only included respondents from the following 8 states that administered the modules that had questions on type of most recent cancer, age of diagnosis with the first type of cancer, age of diagnosis with diabetes, and insulin use. These included the states of Delaware, Georgia, Indiana, Mississippi, Missouri, South Dakota, Virginia, and Wisconsin. Participants were excluded if they reported having no diabetes, pre-diabetes or borderline diabetes, gestational diabetes, or had missing data on their diabetes status. Accordingly, a total sample of 8685 participants with diabetes was considered in this study (Figure 1).

Figure 1.

Flowchart depicting how the final sample size was reached.

Concepts and Measures

The dependent variable used in our logistic regression was NMSC dichotomized into having this disease or not, while the outcome employed in our survival analysis was time to develop NMSC. The NMSC was determined by the question on type of most recently diagnosed cancer, included in the Cancer Survivorship module administered in some of the states as was indicated earlier. The time point of when the participant developed NMSC was determined by age of diagnosis of NMSC which is self-reported. Similarly, diabetes was assessed by the question on whether or not the participant was diagnosed with this disease. The time point when the participant developed diabetes was determined by age of diagnosis with diabetes which is also self-reported. The time-to-event which is time-to-NMSC was determined as the time period that elapsed from the age participant was diagnosed with diabetes till the age of developing NMSC. So the time-to-event reflects the difference in years between the age of diagnosis with NMSC and age of diagnosis with diabetes. However, it is important to note that for participants with more than 1 cancer type, the age of diagnosis with cancer was reported for the first type of cancer, which could be NMSC or other types of cancer. Thereofore, to take this point into consideration we carried out additional survival analysis whereby we included participants who either had only one type of cancer, or did not have NMSC. This sub-analysis ensured that for participants who had NMSC as the only type of cancer, the age of diagnosis was purely pertinent to NMSC, and time-to-event was the elapsed period from the age of diagnosis with diabetes till the age of diagnosis with NMSC.

The independent variables were insulin use and SDOH that included homeownership, marital status, healthcare coverage, employment status, urban/rural county, education level, income level, and race. The insulin use was assessed through a question asking the diabetic participants to self-report if they are currently using insulin or not. We also considered other covariates as potential associated factors with NMSC which included age, sex, body-mass index (BMI), smoking status, and heavy alcohol consumption. Insulin use, SDOH and the other covariates are all self-reported as time stationary variables and not time varying, and reflect the current time when the survey was filled in. The classifications of the different levels of the variables were detailed in Table 1.

Table 1.

Characteristics of the Study Sample.

Variables	N^a	Weighted %
NMSC^b
Yes	487	4.8
No	8110	95.2
Insulin use
Yes	2803	33
No	5637	67
Among those with NMSC
Insulin use yes	130	22.2
Insulin use No	357	77.8
Social determinants of health
Home ownership
Own	6402	77.2
Rent	1864	19.7
Other Arrangement	353	3.1
Marital status
Married or coupled	4295	56.6
Divorced or separated	1689	18.3
Widowed	1689	13.8
Never married	962	11.3
Health care coverage
Yes	8189	92.8
No	455	7.2
Employment status
Employed or self-employed	2398	34
Out of work/Unable to work	1685	21.6
Homemaker/Student	309	4.6
Retired	4177	39.8
Urban/Rural county
Urban	6925	85.5
Rural	1760	14.5
Education level
Graduated high school	2794	31.8
Did not graduate high school	979	16.5
Attended college or technical school	2499	31.5
Graduated college or technical school	2373	20.2
Income level
Less than $15,000	980	14
$15,000 to less than $25,000	1547	21.7
$25,000 to less than $35,000	870	11.7
$35,000 to less than $50,000	1014	14.1
More than $50,000	2479	38.5
Race
White only	5998	69.1
Black or African American only	1856	23.9
Asian only	59	2.2
Other race only^c	464	3.2
Multiracial	164	1.6
Other covariates
Age
18 to 44	597	12.3
45 to 54	985	14.8
55 to 64	2076	27.4
65 or older	4895	45.5
Sex
Male	3923	49.3
Female	4762	50.7
Body-Mass index (BMI)^d
Normal weight	1034	13.6
Underweight	56	0.7
Overweight	2442	29.8
Obese	4369	55.9
Smoking status
Never smoked	4320	51.3
Current smoker	1125	15.3
Former smoker	2868	33.4
Alcohol consumption^e
Not a heavy drinker	7988	96.9
Heavy drinker	209	3.1

^aUnweighted counts.

^bNon-Melanoma Skin Cancer.

^cOther race included American Indian, Alaskan Native, Native Hawaiian/other Pacific Islander, or those who did not identify with any of the aforementioned race categories.

^dAccording to the Centers for Disease Control and Prevention (CDC) classification, BMI was classified into 4 categories: underweight (less than 18.5 kg/m²), normal weight (between 18.5 kg/m² and 25.0 kg/m²), overweight (between 25.0 kg/m² and 30.0 kg/m²), or obese (more than 30.0 kg/m²).

^eHeavy drinkers are defined in the Behavioral Risk Factor Surveillance System (BRFSS) as a man who consumes more than 14 drinks per week or a woman who consumes more than 7 drinks per week.

Statistical Analysis

To account for the BRFSS survey’s complex sampling design, we conducted weighted analyses using sampling and cluster weights. Frequency analyses were conducted, and respective summary statistics including counts and weighted percentages were reported for all the variables. Descriptive analysis along with weighted chi-squared tests and weighted simple logistic regressions were performed to determine the distribution of insulin use, the main exposure in this study, among the levels of each variable, and the extent of its correlation with the social and other factors considered in our analysis. Counts, weighted percentages, unadjusted odds ratios (ORs), 95% confidence intervals (CIs), and P-values were reported. Following that, frequency analysis showing the distribution of NMSC among the levels of each variable was conducted, in addition to performing the crude analysis using weighted simple logistic regression to assess the relationship between insulin use, SDOH, the other covariates, and NMSC (outcome of interest). Counts and weighted percentages for NMSC were generated, and unadjusted ORs and 95% CIs were calculated. The cutoff for eligibility for multiple logistic regression was raised to .2, as the significance level of .05 has been shown to fail in effectively selecting covariates associated with the outcome.^37,38 Therefore, variables with P-values of .2 or less were considered eligible to be included in the adjusted analyses. Weighted multiple logistic regression was carried out to generate adjusted measures of association between insulin use, SDOH, and the other covariates as predictors and NMSC as the outcome.

Survival analysis was also conducted to assess the effect of insulin use on the time-to-develop NMSC. For uncensored participants who developed NMSC, the time-to-event endpoint was taken as the time from the diagnosis of diabetes to the time of diagnosis with NMSC. Participants who did not develop NMSC at the time of the study were considered censored, and censoring time was defined as the duration of diabetes that passed without experiencing the event of NMSC. To test if the proportional-hazards assumption was met in all our survival analysis models, we conducted a test of proportionality of hazards that is based on Schoenfeld residuals. A P-value for the global test of porportinality of hazards that is greater than .05 indicated that this assumption was met.

Kaplan-Meier survival analysis and log-rank test were performed to compare the probability of survival with NMSC event-free among participants who use and do not use insulin, and were also carried out for indices of SDOH that were shown to be significantly associated with time-to-NMSC in the multivariable survival model. The time-to-event depicted in the Kaplan Meier curve was in years which represents the difference between age of diagnosis in NMSC and diabetes. Specifically, it respresents the time period that elapsed between diagnosis with diabetes till diagnosis with NMSC. The stratification into insulin use or not is done based on whether or not currently the participant is taking insulin, but it cannot be determined from the data when exactly the participant started taking insulin. Participants who survive long enough to take insulin are those who are classified as insulin-users. However, those who die before they have the chance to take insulin are considered non-insulin users, but this group of participants did not participate in the study since information is self-reported by the participants themselves. Data on duration of insulin use is not collected in the BRFSS data so analysis cannot be adjusted for the effect of duration of insulin intake.

Weighted unadjusted and adjusted survival analyses using Cox proportional hazards were conducted to model the time-to-NMSC as a function of insulin use, SDOH, and the other covariates. Unadjusted and adjusted hazard ratios (HR) and their corresponding 95% CIs were generated and reported in our results.

Results

Characteristics of the Study Sample

The characteristics of the participants included in this study are displayed in Table 1. Summary statistics are presented in terms of frequency (N) and weighted percentages to account for the sampling design. Out of the 8685 participants, 4.8% (n = 487) had NMSC, and 33% (n = 2803) were using insulin therapy. More specifically, among participants who have NMSC (n = 487), 22.2% were using insulin therapy (n = 130), while the remaining majority (77.8%, n = 357) were not using insulin. Majority of the participants owned their home (77.2%, n = 6402), were married or coupled (56.6%, n = 4295), had healthcare coverage (92.8%, n = 8189), and lived in an urban county (85.5%, n = 6925). Moreover, 39.8% (n = 4177) were retired and 34% (n = 2398) were employed or self-employed. With respect to educational level, around 31% graduated high school (n = 2794) or attended college or technical school (n = 2499). The biggest proportion of the participants had an income of more than $50,000 (38.5%, n = 2479), and reported their race as White (69.1%, n = 5998). Almost half of the participants were females (50.7%, n = 4762), and had an age of 65 years or above (45.5%, n = 4895). More than half of the participants were obese (55.9%, n = 4369), never smoked (51.3%, n = 4320), and most of them were not heavy drinkers (96.9%, n = 7988).

Frequency Distibution of the SDOH and the Other Covariates in the Insulin Use and NMSC

The frequency distribution of the SDOH and the other covariates among insulin users and non-users, and among participants with and without NMSC are presented in terms of weighted column percent in Table 2, as well as Tables 3 and Supplemental Table A1 respectively.

Table 2.

Frequency Distribution of Insulin Use in Each Category of the Different Variables, and Frequency Distribution of Each Category of the Variables Between Insulin Users and Non Users Along With the Association Between Social Determinants of Health, Other Covariates, and Insulin Use (Outcome).

Variables	Insulin Use N (Weighted Row %)^a [Weighted Column %]^b	No Insulin Use N (Weighted Row %)^c [Weighted Column %]^d	p-value^e	Weighted Unadjusted OR (95%CI)^f	P-value^f
Social determinants of health
Home ownership
Own	1938 (31.7) [74.9]	4310 (68.3) [79.3]	.005*	Ref
Rent	726 (37.8) [21.9]	1063 (62.2) [17.7]		1.31 (1.10-1.56)	.002*
Other Arrangement	116 (34.1) [3.2]	222 (65.9) [3.0]		1.11 (.79-1.56)	.54
Marital status
Married or coupled	1294 (31.8) [54.8]	2871 (68.2) [57.9]	.14	Ref
Divorced or separated	592 (34.5) [18.9]	1051 (65.5) [17.6]		1.13 (.93-1.38)	.20
Widowed	547 (32.4) [13.7]	1106 (67.6) [14.0]		1.03 (.85-1.25)	.77
Never married	352 (37.3) [12.5]	577 (62.7) [10.3]		1.28 (1.01-1.61)	.04*
Health care coverage
No	147 (37.1) [8.1]	293 (62.9) [6.8]	.24	Ref
Yes	2644 (32.6) [[91.9]	5316 (67.4) [93.2]		.82 (.58-1.15)	.24
Employment status
Employed or self-employed	635 (28.8) [29.7]	1676 (71.2) [36.2]	<.001*	Ref
Out of work/Unable to work	722 (42.4) [27.4]	912 (57.6) [18.4]		1.82 (1.47-2.24)	<.001*
Homemaker/Student	84 (28.0) [3.8]	218 (72.0) [4.9]		.96 (.64-1.44)	.85
Retired	1331 (32.3) [39.1]	2750 (67.7) [40.5]		1.18 (.99-1.41)	.07
Urban/Rural county
Urban	2160 (32.6) [84.5]	4555 (67.4) [85.8]	.24	Ref
Rural	643 (35.0) [15.5]	1082 (65.0) [14.2]		1.11 (.93-1.32)	.24
Education level
Graduated high school	940 (33.9) [33.0]	1796 (66.1) [31.7]	.03*	Ref
Did not graduate high school	366 (37.4) [19.1]	594 (62.6) [15.7]		1.17 (.93-1.46)	.18
Attended college or technical school	805 (31.7) [30.2]	1628 (68.3) [31.9]		.91 (.76-1.09)	.30
Graduated college or technical school	678 (29.7) [17.7]	1595 (70.3) [20.7]		.82 (.68-.98)	.05*
Income level
Less than $15,000	377 (38.9) [16.3]	572 (61.1) [12.6]	<.001*	Ref
$15,000 to less than $25,000	583 (37.8) [25.0]	939 (62.2) [20.3]		.95 (.73-1.24)	.71
$25,000 to less than $35,000	289 (34.4) [12.2]	561 (65.6) [11.5]		.82 (.60-1.12)	.21
$35,000 to less than $50,000	298 (28.6) [12.2]	690 (71.4) [15.1]		.63 (.46-.85)	.002*
More than $50,000	666 (29.5) [34.3]	1716 (70.5) [40.5]		.66 (.51-.84)	.001*
Race
White only	1847 (31.4) [66.4]	3982 (68.6) [71.0]	.12	Ref
Black or African American only	642 (36.7) [26.6]	1171 (63.3) [22.4]		1.26 (1.06-1.51)	.01*
Asian only	12 (26.6) [1.7]	42 (73.4) [2.4]		.79 (.33-1.88)	.59
Other race only	213 (41.4) [3.9]	237 (58.6) [2.7]		1.54 (1.01-2.35)	.05*
Multiracial	45 (30.3) [1.4]	112 (69.7) [1.5]		.95 (.44-2.05)	.89
Other covariates
Age
18 to 44	215 (37.5) [13.8]	349 (62.5) [11.3]	.21	Ref
45 to 54	321 (33.9) [15.0]	623 (66.1) [14.3]		.85 (.62-1.17)	.33
55 to 64	707 (32.6) [27.2]	1314 (67.4) [27.7]		.80 (.61-1.07)	.13
65 or older	1526 (31.7) [44.0]	3258 (68.3) [46.7]		.77 (.59-.98)	.05*
Sex
Male	1251 (32.6) [49.0]	2552 (67.4) [49.6]	.66	Ref
Female	1552 (33.3) [51.0]	3085 (66.7) [50.4]		1.03 (.89-1.19)	.66
Body-Mass index (BMI)
Normal weight	310 (33.7) [13.8]	699 (66.3) [13.5]	.49	Ref
Underweight	21 (39.5) [.8]	34 (60.5) [.6]		1.28 (.57-2.90)	.55
Overweight	729 (31.4) [28.1]	1645 (68.6) [30.6]		.90 (.69-1.17)	.44
Obese	1510 (34.1) [57.3]	2735 (65.9) [55.3]		1.02 (.79-1.30)	.89
Smoking status
Never smoked	1344 (31.9) [49.5]	2860 (68.1) [52.3]	.30	Ref
Current smoker	378 (33.6) [15.4]	710 (66.4) [15.1]		1.08 (.87-1.34)	.50
Former smoker	962 (34.8) [35.1]	1827 (65.2) [32.6]		1.14 (.97-1.33)	.12
Alcohol consumption
Not a heavy drinker	2605 (33.5) [97.6]	5156 (66.5) [96.5]	.15	Ref
Heavy drinker	48 (25.5) [2.4]	155 (74.5) [3.5]		.68 (.40-1.16)	.15

^aFrequency distribution (unweighted counts and weighted percentages) of insulin use in each category of the variables.

^bWeighted percentages of each category of the variables among insulin users.

^cFrequency distribution (unweighted counts and weighted percentages) of no insulin use in each category of the variables.

^dWeighted percentages of each category of the variables among non-insulin users.

^eWeighted Chi-square test showing the unadjusted association between each variable and insulin use.

^fWeighted simple logistic regression showing the unadjusted associations between each level of the variable in comparison with the reference category and insulin use.

*P-value < .05 indicating significant results.

Table 3.

Frequency Distribution of NMSC in Each Category of the Different Variables, and Frequency Distribution of Each Category of the Variables for Participants With and Without NMSC, Along With the Unadjusted and Adjusted Associations Between Insulin Use, Social Determinants of Health, Other Covariates, and NMSC^a (Outcome).

NMSC^a	NMSC (Weighted Row %)^b [Weighted Column %]^c	No NMSC (Weighted Row %)^d [Weighted Column %]^e	Weighted Unadjusted OR (95%CI)^f	Weighted Adjusted OR (95%CI)^g	P-Value^g
Insulin use
No	357 (5.8) [77.8]	5226 (94.2) [66.5]	Ref	Ref
Yes	130 (3.4) [22.2]	2640 (96.6) [33.5]	.57 (.41-.77)	.56 (.38-.82)	.003*
Social determinants of health
Home ownership
Own	425 (5.6) [90.6]	5904 (94.4) [76.4]	Ref	Ref
Rent	46 (1.9) [7.9]	1806 (98.1) [20.4]	.33 (.21-.51)	.75 (.41-1.40)	.37
Other Arrangement	12 (2.2) [1.5]	339 (97.8) [3.2]	.38 (.14-1.07)	.64 (.15-2.64)	.53
Marital status
Married or coupled	290 (5.9) [69.2]	3956 (94.1) [56.0]	Ref	Ref
Divorced or separated	54 (2.8) [10.5]	1623 (97.2) [18.6]	.45 (.28-.73)	.74 (.42-1.32)	.31
Widowed	117 (5.8) [16.4]	1548 (94.2) [13.5]	.98 (.69-1.40)	.98 (.65-1.49)	.94
Never married	23 (1.6) [3.9]	936 (98.4) [11.8]	.27 (.14-.50)	.74 (.36-1.55)	.43
Employment status
Employed or self-employed	84 (2.7) [18.7]	2300 (97.3) [34.9]	Ref	Ref
Out of work/Unable to work	45 (3.9) [17.5]	1631 (96.1) [21.8]	1.49 (.80-2.75)	3.40 (1.47-7.86)	.004*
Homemaker/Student	15 (5.2) [4.9]	287 (94.8) [4.6]	2.00 (.64-6.27)	2.39 (.70-8.15)	.16
Retired	339 (7.2) [58.9]	3781 (92.8) [38.7]	2.84 (2.05-3.93)	1.88 (1.20-2.93)	.01*
Education level graduated high school	122 (4.0) [26.5]	2639 (96.0) [32.0]	Ref	Ref
Did not graduate high school	29 (2.9) [10.0]	936 (97.1) [16.8]	.73 (.42-1.27)	1.01 (.51-1.99)	.98
Attended college or technical school	161 (6.2) [40.9]	2313 (93.8) [31.1]	1.59 (1.08-2.36)	1.62 (1.04-2.51)	.03*
Graduated college or technical school	173 (5.4) [22.6]	2184 (94.6) [20.1]	1.36 (.97-1.92)	1.30 (.86-1.97)	.21
Income level
Less than $15,000	30 (2.4) [6.9]	944 (97.6) [14.4]	Ref	Ref
$15,000 to less than $25,000	52 (2.8) [12.1]	1472 (97.2) [22.0]	1.16 (.62-2.17)	.91 (.45-1.83)	.79
$25,000 to less than $35,000	47 (4.9) [11.5]	812 (95.1) [11.8]	2.05 (1.04-4.01)	1.38 (.65-2.95)	.40
$35,000 to less than $50,000	70 (6.5) [18.3]	938 (93.5) [13.8]	2.77 (1.27-6.01)	1.80 (.71-4.62)	.22
More than $50,000	192 (6.6) [51.2]	2269 (93.4) [38.0]	2.82 (1.63-4.88)	2.10 (1.02-4.33)	.04*
Race
White only	460 (6.9) [97.0]	5458 (93.1) [68.0]	Ref	Ref
Black or African American only	4 (.2) [.8]	1848 (99.8) [26.0]	.02 (.01-.06)	.03 (.01-.11)	<.001*
Other race only	8 (2.4) [1.6]	453 (97.6) [4.0]	.33 (.15-.76)	.55 (.21-1.44)	.22
Multiracial	9 (1.7) [.6]	155 (98.3) [2.0]	.23 (.09-.57)	.28 (.09-.90)	.03*
Other covariates
Age
18 to 44	5 (.6) [1.6]	591 (99.4) [13.0]	Ref	Ref
45 to 54	12 (2.1) [6.5]	966 (97.9) [15.3]	3.35 (.71-15.78)	2.17 (.52-8.99)	.28
55 to 64	76 (4.2) [23.9]	1986 (95.8) [27.6]	6.88 (2.34-20.22)	3.43 (1.08-10.94)	.04*
65 or older	387 (7.3) [68.0]	4444 (92.7) [44.1]	12.24 (4.40-34.04)	5.71 (1.90-17.11)	.002*
Smoking status
Never smoked	264 (5.4) [55.3]	4011 (94.6) [51.1]	Ref	Ref
Current smoker	35 (1.9) [5.9]	1080 (98.1) [15.8]	.34 (.20-.58)	.50 (.28-.90)	.02*
Former smoker	186 (5.9) [38.8]	2650 (94.1) [33.1]	1.08 (.79-1.49)	.94 (.67-1.32)	.71

^aNon-Melanoma Skin Cancer.

^bFrequency Distribution (unweighted counts and weighted percentages) of NMSC in each category of the variables.

^c Weighted percentagesof each category of the variables for participants with NMSC.

^dFrequency Distribution (unweighted counts and weighted percentages) of no NMSC in each category of the variables.

^eWeighted percentages of each category of the variables for particpants without NMSC.

^fWeighted simple logistic regression showing the unadjusted associations between each level of the variable in comparison with the reference category and NMSC.

^gWeighted multiple logistic regression showing the adjusted associations between each level of the variable in comparison with the reference category and NMSC.

*P-value < .05 indicating significant results.

Our results showed that for the variables that had significant association with insulin, the highest percentages of insulin use were among homeowners (74.9%), married or coupled (54.8%), retired (39.1%), graduated high school (33%), had an income that is more than $50,000 (34.3%), white race (66.4%), and age 65 or older (44%) (Table 2).

As for the frequency distribution of the SDOH and other covariates among participants with NMSC which had significant adjusted association with NMSC, our results showed that 22.2% of NMSC participants were using insulin, 58.9% were retired, 40.9% attended college or technical school, 51.2% had an income of more than $50,000, 97% were of white race, 68% had an age of 65 years or older, and 5.9% were current smokers (Table 3).

Frequency Distribution of Insulin Use in Each Category of the Different Variables and its Association with SDOH and Other Covariates

The frequency distribution of insulin use among the SDOH and the other covariates expressed in terms of weighted row percentages, along with the weighted unadjusted ORs and the 95% CIs are presented in Table 2. Our sub-analysis indicated that there is a significant association between insulin use and the different indices of SDOH that included categories of home ownership, marital status, employment status, education level, income level, race, and age.

In this regard, our results showed that there was an increase in the weighted odds and percentages of insulin use among participants who rented their home (OR 1.31; 95% CI: 1.10-1.56; 37.8%), have never been married (OR 1.28; 95% CI: 1.01-1.61; 37.3%), are out of work/unable to work (OR 1.82; 95% CI: 1.47-2.24; 42.4%), and who are Black (OR 1.26; 95% CI: 1.06-1.51; 36.7%), or of other race (OR 1.54; 95% CI: 1.01-2.35; 41.4%) compared to their respective reference categories shown in Table 2. On the other hand, our results also indicated that there was a decrease in the odds and percentages of insulin use among participants who reported their level of education as graduated college or technical school (OR .82; 95% CI: .68-.98; 29.7%), have an income between $35,000 and $50,000 (OR .63; 95% CI: .46-.85; 28.6%), or more than $50,000 (OR .66; 95% CI: .51-.84; 29.5%), and are of older age of 65 years and more (OR .77; 95% CI 0.59-.99; 31.7%), in comparison with their respective reference groups presented in Table 2.

Frequency Distribution of NMSC in Each Category of the Different Variables and its Association with Insulin Use, Social Determinants of Health, and Other Covariates

Supplemental Table A1 and Table 3 display the frequency distribution of NMSC in each category of the different variables, along with the unadjusted and adjusted associations expressed in terms of the weighted ORs, 95% CI, and P-values between insulin use, SDOH, the other covariates, and NMSC (outcome) respectively.

Our adjusted weighted logistic regression analysis showed that participants using insulin, who are Black, multiracial, and currently smoking had lower weighted percentages of NMSC and decreased weighted odds of NMSC compared to their respective reference categories (Table 3). In this respect, our results indicated that 3.4% of participants using insulin had NMSC, while 5.8% of participants who do not use insulin had NMSC; in addition, insulin use was significantly associated with decreased odds of NMSC by .56 (reflecting a reduction of 43% in the odds) compared to no insulin use (95%CI: .38-.82). Our results also showed that NMSC was less prevalent among Black individuals (.2%) and multiracial individuals (1.7%) who also had decreased odds of NMSC of .03 (95% CI: .01-.11), and .28 (95% CI: .09-.9), respectively, when compared to White individuals (6.9%). Moreover, our results also indicated that current smokers had lower weighted percentages of NMSC (1.9%), and reduced odds of NMSC of .50 (95% CI: .28-.90) compared to participants who never smoked (5.4%).

On the other hand, NMSC was more prevalent among participants who were out of work/unable to work (3.9%), retired (7.2%), attended college/technical school (6.2%), had income of more than $50,000 (6.6%), and were aged between 55 and 64 years (4.2%), or more than 65 years (7.3%) compared to their respective reference levels shown in Table 3. These categories were also associated with increased odds of NMSC, with out of work/unable to work having an OR of 3.40 (95% CI: 1.47-7.86), retired (OR 1.88; 95% CI: 1.20-2.93), attended college or technical school (OR 1.62; 95% CI: 1.04-2.51), an income of more than $50,000 (OR 2.10; 95% CI: 1.02-4.33), age between 55 and 64 years (OR 3.43; 95% CI: 1.08-10.94), and more than 65 years (OR 5.71; 95% CI: 1.90-17.11).

Association Between Insulin Use, Social Determinants of Health, Other Covariates, and the Time-To-NMSC

Figure 2 displays the Kaplan-Meier survival estimates for the time-to-diagnosis with NMSC pertaining to participants who use insulin compared to those who do not use insulin. The probability of event-free survival was significantly higher for participants who use insulin compared to those who do not use insulin (P < .001 by the log-rank test). Supplemental Table A2 and Table 4 respectively display the results of the simple and multiple Cox proportional hazards analysis with outcome time-to-NMSC expressed as a function of insulin use, SDOH, and other variables. The unadjusted and adjusted weighted hazard ratios (HRs) for NMSC were presented in Supplemental Table A2 and Table 4 respectively. The assumption of proportionality of hazard was tested for and confirmed for the survival multivariable model presented in Table 4 with a global P = .48.

Figure 2.

Kaplan-Meier survival curves and Log Rank test for time-to-diagnosis with NMSC among participants who use insulin compared to those who do not use insulin.

Table 4.

Unadjusted and Adjusted Hazard Ratios (HR) for Time-To-NMSC.^a

Variables	Weighted Unadjusted HR (95% CI)^b	Weighted Adjusted HR (95% CI)^c	P-Value^c
Insulin use
No	Ref	Ref
Yes	.40 (.26-.62)	.36 (.21-.62)	<.001*
Social determinants of health
Home ownership
Own	Ref	Ref
Rent	.38 (.19-.75)	.71 (.30-1.68)	.43
Other Arrangement	.62 (.16-2.41)	1.15 (.21-6.33)	.87
Marital status
Married or coupled	Ref	Ref
Divorced or separated	.54 (.26-1.10)	.80 (.35-1.79)	.58
Widowed	.81 (.50-1.31)	1.05 (.58-1.89)	.87
Never married	.27 (.12-.64)	.48 (.17-1.38)	.17
Employment status
Employed or self-employed	Ref	Ref
Out of work/Unable to work	1.22 (.47-3.14)	3.94 (1.18-13.23)	.03*
Homemaker/Student	.81 (.30-2.19)	1.21 (.38-3.82)	.75
Retired	1.65 (1.05-2.60)	1.90 (1.11-3.24)	.02*
Education level
Graduated high school	Ref	Ref
Did not graduate high school	.57 (.27-1.22)	.75 (.32-1.74)	.50
Attended college or technical school	2.07 (1.21-3.54)	1.88 (1.05-3.36)	.03*
Graduated college or technical school	1.60 (.99-2.60)	1.42 (.82-2.46)	.21
Income level
Less than $15,000	Ref	Ref
$15,000 to less than $25,000	.70 (.30-1.66)	.62 (.25-1.55)	.31
$25,000 to less than $35,000	1.65 (.66-4.13)	1.21 (.47-3.14)	.69
$35,000 to less than $50,000	3.08 (.99-9.56)	2.18 (.55-8.64)	.27
More than $50,000	2.57 (1.19-5.51)	2.01 (.78-5.17)	.15
Race
White only	Ref	Ref
Black or African American only	.02 (.005-.09)	.03 (.01-.13)	<.001*
Other race only	.29 (.003-.21)	.05 (.01-.39)	.004*
Multiracial	.20 (.06-.74)	.27 (.07-1.05)	.06
Other covariates
Smoking status
Never smoked	Ref	Ref
Current smoker	.36 (.16-.83)	.48 (.20-1.13)	.09
Former smoker	1.03 (.66-1.62)	.90 (.55-1.45)	.66

^aNon-Melanoma Skin Cancer.

^bWeighted simple survival analysis using Cox proportional hazards showing the unadjusted associations between each level of the variable in comparison with the reference category and Time-to-NMSC.

^cWeighted multiple survival analysis using Cox proportional hazards showing the adjusted associations between each level of the variable in comparison with the reference category and Time-to-NMSC.

P-value < .05 indicating significant results.

In the adjusted survival analysis (Table 4), a significant association was detected between insulin use and NMSC indicating that participants with diabetes who use insulin had a 64% reduction in the hazard of developing NMSC (HR .36; 95% CI: .21-.62) compared to participants with diabetes who do not use insulin. Furthermore, our analysis of SDOH indicated that participants who are out of work/unable to work, or retired had higher hazard of 3.94 (95% CI: 1.18-13.23) and 1.90 (95% CI: 1.11-3.24) respectively to develop NMSC, compared to participants who are employed or self-employed. Kaplan-Meier survival curves shown in Figure 3(a) also showed a significant association between employment status and time-to-NMSC with log-rank test P < .001, with a lower survival curve pertinent to retired compared to employed or self-employed.

Figure 3.

Kaplan-Meier survival curves (a-c), and Log-Rank tests for selected indices of Social Determinants of Health and time-to-NMSC.

In addition, study participants who attended college or technical school had a higher hazard of NMSC of 1.88 (95% CI: 1.05-3.36) compared to participants who only graduated high school, and lower survival curve as illustrated in Figure 3(b) (log-rank test P < .001).

With respect to race, Black/African American participants (HR .03; 95% CI: .01-.13) and participants who reported being from other races (HR .05; 95% CI: .01-.39) had significantly lower hazards of developing NMSC compared to White participants, and higher survival curves as shown in Figure 3(c) (log-rank test P < .001).

To address the issue of multiple cancers, we carried out additional multivariable sub-group analysis using the multiple Cox proportional hazards with outcome time-to-NMSC on participants who either had 1 type of cancer, or did not have NMSC (controls), excluding those with more than 1 type of cancer. A total of 117 participants with NMSC and multiple types of cancer were excluded from this sub-analysis out of a total of 487 NMSC cases, leaving a total of 370 participants with NMSC. This sub-analysis ensured that for the 370 participants who had NMSC, the age of diagnosis was purely pertinent to NMSC. Results of this sub-analysis, not presented in a table but discussed here, were in concordance with the aforementioned survival analysis that had all 487 NMSC cases included. Once more, insulin use for diabetic participants was shown to be associated with decreased hazard for NMSC with HR = .32; 95% CI: .17-.58; P < .001, (68% reduction in hazard). In addition, some indices of SDOH also showed significant associations with time-to-NMSC. These included, employment status comparing out of work/unable to Work to employed or self-employed (HR = 3.82; 95% CI: 1.10-13.24; P = .03); education level comparing attended college or technical school to graduated high school (HR = 1.89; 95% CI: 1.01-3.55; P = .04); and race comparing Black or African American (HR = .03 95% CI HR = .01-.14 P < .001); Other race (HR = .05 95% CI HR = .01-.41 P = .005); and Multiracial (HR = .16 95% CI HR = .03-.89 P = .03) to White. The propotional-hazards assumption for this sub-analysis was also assessed and confirmed with a global P = .61.

Discussion

Our study underscored the complex relationship between insulin use and NMSC by demonstrating that socioeconomic disparities are related to differences in insulin use and the development of NMSC. Our findings emphasize the importance of considering various factors, including insulin use and SDOH, in the prevention and management of NMSC among individuals with diabetes.

In this regard, our results showed that individuals with diabetes using insulin had 44% lower odds of developing NMSC compared to those who do not use insulin therapy. Furthermore, Kaplan-Meier survival curves showed that the likelihood of being NMSC-free was higher for participants with diabetes who were using insulin compared to those who were not. In addition, our Cox-proportional weighted analysis revealed a 64% lower hazard of developing NMSC among participants who use insulin. In line with our findings, insulin therapy in diabetic patients was shown to lower the risk of developing NMSC and the protective effect of insulin use becomes more distinct with increasing age.³⁰ Moreover, insulin use improved the 5-year progression-free survival in diabetic patients with coexisting melanoma.³⁹ The underlying molecular and mechanistic pathways through which insulin therapy reduces the risk to develop NMSC in individuals with diabetes are still not clearly defined. It is conceivable that the link between insulin use and reduction in NMSC risk is not direct but rather multifactorial. Management of diabetes through insulin therapy will influence the metabolic state of the diabetic patients by controlling hyperglycemia, improving insulin resistance and attenuate the inflammatory and oxidative stress signals, factors that are associated with increased risk of developing NMSC in individuals with diabetes.^40,41

Moreover, our analysis highlighted the association between the SDOH, other factors, and NMSC among individuals with diabetes. Socioeconomic status, access to medical care, culture and social setting, race, inequities in living and working environments have direct influence on biological and behavioral outcomes associated with insulin use and management of diabetes.⁴² In this regard, our results showed that higher income levels and older age were significantly associated with a lower percentage of insulin use and increased odds of NMSC. On the other hand, Black participants showed a higher prevalence of insulin use and a decreased hazard of NMSC compared to White participants. The observed increased hazards of NMSC among unemployed and retired participants compared to employed or self-employed participants could be attributed to the difference in sun exposure, as those who do not have a job may spend more time outdoors during their free time, leading to increased exposure to UVB radiation in contrast to those who work during the day.⁴³ Moreover, retired individuals are typically older in age, which is another contributing factor to the higher prevalence of NMSC among this age group, as age is known to be a risk factor for cancer in general,^30,44,45 including NMSC.⁹

Our analyses also revealed that higher levels of education are associated with an increased hazard of developing NMSC. This discrepancy in NMSC risk among the different levels of education may be attributed to better access to health information and awareness of prevention strategies that people with higher education typically have.⁴⁶ Improved knowledge and awareness can lead to early recognition of signs and symptoms of NMSC and more frequent access to healthcare coverage and medical screening.^46,47 Conversely, lower levels of education have been associated with decreased perceptions of the risks of skin cancer,^48,49 lack of awareness of its early signs and symptoms, and missed opportunities for skin examinations.³⁷ These factors, combined, may explain the observed association between lower educational levels and advanced stages of skin cancer at diagnosis.⁴⁸ Moreover, the level of education influences the communication between healthcare professionals and patients about their health and diseases. In this respect, it was found that healthcare providers tend to share more detailed information on the risks of skin cancer and the importance of self-performed and medical skin examinations with patients who have a college education than with patients that have only a high school education.⁴⁸ Consequently, individuals with higher levels of education are more likely to have a better awareness of NMSC, and its diagnosis and prognosis.^46,49

Our analysis further showed that individuals with higher income levels were associated with increased odds of NMSC. This result is in line with findings from the literature which reported that cases of skin cancer are more prevalent among people of higher socioeconomic status.^46,50 Nevertheless, a high socioeconomic status does not necessarily imply a greater risk of NMSC; rather, it is more probable that it reflects an increased likelihood of detecting and reporting skin cancer. In fact, despite the higher reported prevalence of NMSC in populations with higher socioeconomic status, studies have shown that advanced stages of NMSC, thicker tumors, and a 5-year reduction in survival rates were more prevalent among underprivileged communities.^46,50-52 These observations indicate that early recognition of signs and symptoms of skin cancer coupled with early diagnosis are less likely to be received for underprivileged individuals. A probable explanation for the aforementioned discrepancies in the detection of skin cancer can be referred to the inequity in accessing cancer screening and preventative healthcare facilities across the different socioeconomic classes. The privatized healthcare system in the US contributes to this disproportionality since thus far it has not succeeded in securing equitable healthcare coverage for all communities,⁵³ a drawback that needs to be accounted for when reforms are to be considered in the restructuring of the healthcare system.

Our study also revealed that White individuals were associated with a higher hazard of developing NMSC compared to the other races. This result is consistent with previously reported findings that underscored the disproportionate prevalence of skin cancer among different races in the US, with White individuals having the highest incidence rates.³⁴ Age-adjusted incidence rates in the US showed that White individuals are 70 times more likely to develop skin cancer compared to Black individuals, a discrepancy in skin cancer risk that may be attributed to biologic and genetic differences.³⁴ Nevertheless, despite having a lower risk of NMSC, Black individuals were found to present with a poorer prognosis and a higher mortality rate when diagnosed with this disease compared to White individuals.^34,54 This racial disparity in survival outcomes may be related to the perception among Black individuals that they are less susceptible to skin cancer,⁴⁹ their likelihood of not seeking immediate medical consultancy for suspicious skin lesions or injury,⁵⁵ and subsequently their diagnosis with this disease at advanced stages⁵⁴ compared to White individuals.

It is important to point out here, that the prevalence and incidence of diabetes on risk and outcomes is vastly influenced by SDOH factors. People with lower socioeconomic status (SES) and lower educational levels are more likely to develop diabetes and are more susceptible to sub-optimal heath care management.^42,56 In addition, disparities in racial and ethnic populations are associated with significantly higher prevalence and incidence of diabetes and access to quality care are more likely to experience greater levels of morbidity and early mortality.^42,56 Financial hardship as well as psychosocial and neighborhood influences, such as social cohesion and religiosity, have been linked to poor glycemic control in older adults with diabetes resulting in adverse diabetes outcomes over time.⁵⁷ Accounting for these disparities in SDOH in the management of diabetes and to expand the utilization of technologies such as continuous glucose monitoring to underprivileged socioeconomic individuals will result in better glycemic control and ultimately lessen the adverse clinical outcomes of diabetes.⁵⁸

Strength and Limitations

The study is novel since it is the first to incorporate the social factors into the association with insulin use and NMSC, and triangulate these relationships. Moreover, basing our quantitative study on a large-scale surveillance system such as the CDC-BRFSS makes the generalizability of our inferences and results feasible since this data encompasses a national representation of the US population.

Despite the novelty in the research question and the strength in the selection of the BRFSS as our sample population, there are some limitations in this study. The BRFSS data is collected through telephone surveys and therefore the diagnosis of cancer and diabetes was self-reported. However, there is evidence that self-reported diagnoses are concordant enough with medical records.⁵⁹ Moreover, there is no distinction in our data between type I and type II diabetes. However, given that the age of diagnosis with type I diabetes is approximately below the age of 30 to 40 years and this forms only 8% to 24% of our sample, and given that the diagnosis with type II diabetes is typically above the age of 40 and this age group forms 76% to 92% of our data, then type II diabetes is expected to be much more prevalent in our sample. Nonetheless, given that there is no direct question on the type of diabetes, then this cannot be fully confirmed. Even though our study focused on diabetes mellitous (DM), selection bias should not be of concern since the frequency distributions of insulin use and SDOH, and their link to NMSC are considered among diabetic participants only. Accordingly, study population characteristics are compared between different variables within the sample of diabetic participants. The BRFSS data did not contain data on the concurrent medication the DM participant is taking other than insulin, accordingly the medication use such as statin, sulfonyl urea and drug-related to cancer incidence could not be adjusted for in the multivariable model. Morevoer, the BRFSS asked participants to report their most recent cancer type. Thus, participants with more than 1 type of cancer, could have had NMSC as an earlier type of cancer, but was not recorded since it is not the most recent type. This particular group of particiapnts are recorded as not having NMSC as their most recent cancer type, but could have had it in the past. In addition, participants with multiple types of cancer were asked to report only the age of diagnosis with their first type of cancer, not the age of diagnosis with their most recent cancer. To tackle this issue and to confirm our results, we carried out a sub-group analysis on participants who either reported having 1 type of cancer, or did not have NMSC cancer (controls). This sub-analysis ensured that for NMSC cases, the reported age of diagnosis was purely pertinent to NMSC and not for any other type of cancer. Results of this sub-analysis were in concordance with the survival model that encompassed all participants who did not have cancer, had 1 type of cancer or multiple cancers; whereby, similar conclusions were drawn in both survival analyses concerning the role of insulin use and SDOH on NMSC.

Conclusion

In conclusion, our study revealed a significant association between insulin use and reduced risk of NMSC among individuals with diabetes and underscored the critical role of SDOH in this relationship. Our findings suggest that insulin use may have a protective effect against NMSC and that indices of SDOH such as income, level of education, employment status, and race can significantly influence the likelihood of developing this disease, indicating potential disparities in exposure to risk factors and healthcare access across different socioeconomic groups. To ensure equal opportunities for skin cancer prevention, early detection, and improved prognosis, it is essential to address healthcare access inequities. Healthcare providers should consider several factors, including insulin use and SDOH when developing prevention and management strategies for NMSC in people with diabetes. While our results highlight the potential of insulin as a protective agent against NMSC, this knowledge can only have a meaningful impact on public health when it is accompanied by efforts to address health inequity in the early diagnosis of skin cancer and the associated treatment modalities, promote awareness of its risk factors, and advocate for access to healthcare across all social groups.

Supplemental Material

Supplemental Material - Role of Insulin Use and Social Determinants of Health on Non-melanoma Skin Cancer: Results From the Behavioral Risk Factor Surveillance System

Supplemental Material for Role of Insulin Use and Social Determinants of Health on Non-melanoma Skin Cancer: Results From the Behavioral Risk Factor Surveillance System by Nour Massouh, Ayad A. Jaffa, and Miran A. Jaffa in Cancer Control

Footnotes

Authors’ Contributions

MAJ identified the BRFSS dataset. MAJ conceptualized the research ideas. MAJ and NM conceived the study concept and design. AAJ refined and finalized the conceptual research ideas. NM and MAJ carried out the data analysis. MAJ oversaw the data analysis. MAJ and AAJ supervised the impelementation of the project. NM drafted the manuscript. AAJ and MAJ extensively revised and edited the manuscript. NM, AAJ, and MAJ all approved the final version of the manuscript.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Ethical Statement

ORCID iDs

Nour Massouh

Miran A. Jaffa

Data Availability Statement

The dataset supporting the conclusions of this article is available on the CDC website, https://www.cdc.gov/brfss/annual_data/annual_2020.html.

Supplemental Material

Supplemental material for this article is available online

Note

Appendix

References

Ciążyńska

Kamińska-Winciorek

Lange

, et al. The incidence and clinical analysis of non-melanoma skin cancer. Sci Rep. 2021;11(1):4337.

Faur

Moldovan

Văleanu

Rotar

Filip

Roman

. The prevalence and treatment costs of non-melanoma skin cancer in cluj-napoca maxillofacial center. 2023;59(2):220.

Sung

Ferlay

Siegel

, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71(3):209-249.

Eisemann

Waldmann

Geller

, et al. Non-melanoma skin cancer incidence and impact of skin cancer screening on incidence. J Invest Dermatol. 2014;134(1):43-50.

NHS . Overview skin cancer (non-melanoma). 2020; https://www.nhs.uk/conditions/non-melanoma-skin-cancer/

WHO . Radiation: ultraviolet (UV) radiation and skin cancer. 2017; https://www.who.int/news-room/questions-and-answers/item/radiation-ultraviolet-(uv)-radiation-and-skin-cancer

Yazdi

Jourabchi

. Skin cancer, non-melanoma. 2015:1064-1066.

Gandhi

Kampp

. Skin cancer epidemiology, detection, and management. Med Clin North Am. 2015;99(6):1323-1335.

Urban

Mehrmal

Uppal

Giesey

Delost

. The global burden of skin cancer: a longitudinal analysis from the Global Burden of Disease Study, 1990-2017. JAAD Int. 2021;2:98-108.

10.

Fitzmaurice

Allen

Barber

, et al. Global, regional, and national cancer incidence, mortality, years of life lost, years lived with disability, and disability-adjusted life-years for 32 cancer groups, 1990 to 2015: a systematic analysis for the global burden of disease study. JAMA Oncol. 2017;3(4):524-548.

11.

Dobrică

Banciu

Kipkorir

, et al. Diabetes and skin cancers: risk factors, molecular mechanisms and impact on prognosis. World J Clin Cases. 2022;10(31):11214-11225.

12.

Zhou

Luo

. Body mass index and risk of non-melanoma skin cancer: cumulative evidence from prospective studies. Sci Rep. 2016;6:37691.

13.

Yen

Dhana

Okhovat

Qureshi

Keum

Cho

. Alcohol intake and risk of nonmelanoma skin cancer: a systematic review and dose-response meta-analysis. Br J Dermatol. 2017;177(3):696-707.

14.

Alberg

Shopland

Cummings

. The 2014 surgeon general’s report: commemorating the 50th anniversary of the 1964 report of the advisory committee to the US surgeon general and updating the evidence on the health consequences of cigarette smoking. Am J Epidemiol. 2014;179(4):403-412.

15.

Yeh

Golozar

Brancati

. Cancer and diabetes. In: Cowie

Casagrande

Menke

, et al., eds. Diabetes in America. Bethesda (MD): National Institute of Diabetes and Digestive and Kidney Diseases (US); 2018.

16.

Arafa

Mostafa

Navarini

Dong

J-Y

. The association between smoking and risk of skin cancer: a meta-analysis of cohort studies. Cancer Causes Control. 2020;31(8):787-794.

17.

Rollison

Iannacone

Messina

, et al. Case-control study of smoking and non-melanoma skin cancer. Cancer Causes Control. 2012;23(2):245-254.

18.

WHO . Diabetes. 2023. https://www.who.int/news-room/fact-sheets/detail/diabetes

19.

Vigneri

Frasca

Sciacca

Pandini

Vigneri

. Diabetes and cancer. Endocr Relat Cancer. 2009;16(4):1103-1123.

20.

Tseng

H-W

Shiue

Y-L

Tsai

K-W

Huang

W-C

Tang

P-L

Lam

H-C

. Risk of skin cancer in patients with diabetes mellitus: a nationwide retrospective cohort study in Taiwan. Medicine (Baltimore). 2016;95(26):e4070.

21.

Wotton

Yeates

DGR

Goldacre

. Cancer in patients admitted to hospital with diabetes mellitus aged 30 years and over: record linkage studies. Diabetologia. 2010;54(3):527-534.

22.

Leitner

Siebel

Akingbesote

Zhang

Perry

. Insulin and cancer: a tangled web. Biochem J. 2022;479:583-607.

23.

Home

. Insulin therapy and cancer. Diabetes Care. 2013;36(Suppl 2):S240-S244.

24.

Vicentini

Ballotari

Venturelli

, et al. Impact of insulin therapies on cancer incidence in type 1 and type 2 diabetes: a population-based cohort study in Reggio Emilia, Italy. Cancers. 2022;14:2719.

25.

Malmberg

Ryden

Wedel

, et al. FASTTRACK ´ intense metabolic control by means of insulin in patients with diabetes mellitus and acute myocardial infarction (DIGAMI 2): effects on mortality and morbidity. Eur Heart J. 2005;26:650-661.

26.

ORIGIN Trial Investigators Gerstein

Bosch

, et al. Basal insulin and cardiovascular and other outcomes in dysglycemia. N Engl J Med. 2012;367:319-328.

27.

Tang

Yang

Song

Han

. Meta‐analysis of the association between sodium‐glucose co‐transporter‐2 inhibitors and risk of skin cancer among patients with type 2 diabetes. Diabetes Obes Metab. 2018;20(12):2919-2924.

28.

Tseng

C-H

. Rosiglitazone may reduce non-melanoma skin cancer risk in Taiwanese. BMC Cancer. 2015;15(1):41.

29.

Tseng

C-H

. Metformin is associated with decreased skin cancer risk in Taiwanese patients with type 2 diabetes. J Am Acad Dermatol. 2018;78(4):694-700.

30.

Chuang

Lewis

Spandau

. Decreased incidence of nonmelanoma skin cancer in patients with type 2 diabetes mellitus using insulin: a pilot study. Br J Dermatol. 2005;153(3):552-557.

31.

Braveman

Cubbin

Egerter

Williams

Pamuk

. Socioeconomic disparities in health in the United States: what the patterns tell us. Am J Public Health. 2010;100(Suppl 1):S186-196.

32.

Braveman

Gottlieb

. The social determinants of health: it’s time to consider the causes of the causes. Public Health Rep. 2014;129(Suppl 2):19-31.

33.

Marmot

Bell

. Social determinants and non-communicable diseases: time for integrated action. BMJ. 2019;364:l251.

34.

Gloster

Neal

. Skin cancer in skin of color. J Am Acad Dermatol. 2006;55(5):741-760.

35.

Padovese

Franco

Valenzano

Pecoraro

Cammilli

Petrelli

. Skin cancer risk assessment in dark skinned immigrants: the role of social determinants and ethnicity. Ethn Health. 2018;23(6):649-658.

36.

CDC . 2020 BRFSS survey data and documentation. 2021; https://www.cdc.gov/brfss/annual_data/annual_2020.html. Accessed April 11, 2022.

37.

Bursac

Gauss

Williams

Hosmer

. Purposeful selection of variables in logistic regression. Source Code Biol Med. 2008;3(1):17.

38.

Mickey

Greenland

. The impact of confounder selection criteria on effect estimation. Am J Epidemiol. 1989;129(1):125-137.

39.

Karlin

Mangold

Amin

, et al. Survival and glycemic control in patients with coexisting melanoma and diabetes mellitus. Future Sci OA. 2019;5(3):FSO368.

40.

Zhao

Qingbo

, et al. Insulin-attenuated inflammatory response of dendritic cells in diabetes by regulating RAGE-PKCβ1-IRS1-NF-κB signal pathway: a study on the anti-inflammatory mechanism of insulin in diabetes. J Diabetes Res. 2020;2020:1.

41.

Long

Martin

Pipkin

Herfarth

Sandler

Kappelman

. Risk of melanoma and nonmelanoma skin cancer among patients with inflammatory bowel disease. Gastroenterology. 2012;143:390-399.

42.

Hill-Briggs

Adler

Berkowitz

, et al. Social determinants of health and diabetes: a scientific review. Diabetes Care. 2021;44:258-279.

43.

Ahn

Jimeno

Ugidos

. ‘Mondays in the sun:’ unemployment, time use, and consumption patterns in Spain. Contrib Econ Anal. 2004;271:237-259.

44.

Xia

Dong

, et al. Cancer statistics in China and United States, 2022: profiles, trends, and determinants. Chin Med J (Engl). 2022;135(5):584-590.

45.

Deshpande

Harris-Hayes

Schootman

. Epidemiology of diabetes and diabetes-related complications. Phys Ther. 2008;88(11):1254-1264.

46.

Harvey

Patel

Sandhu

Wallington

Hinds

. Social determinants of racial and ethnic disparities in cutaneous melanoma outcomes. Cancer Control. 2014;21(4):343-349.

47.

McDaniel

Nuhu

Ruiz

Alorbi

. Social determinants of cancer incidence and mortality around the world: an ecological study. Global Health Promot. 2019;26(1):41-49.

48.

Pollitt

Swetter

Johnson

Patil

Geller

. Examining the pathways linking lower socioeconomic status and advanced melanoma. Cancer. 2012;118(16):4004-4013.

49.

Buster

You

Fouad

Elmets

. Skin cancer risk perceptions: a comparison across ethnicity, age, education, gender, and income. J Am Acad Dermatol. 2012;66(5):771-779.

50.

Singh

Jemal

. Socioeconomic and racial/ethnic disparities in cancer mortality, incidence, and survival in the United States, 1950-2014: over six decades of changing patterns and widening inequalities. J Environ Public Health. 2017;2017:2819372.

51.

Reyes-Ortiz

Goodwin

Zhang

Freeman

. Socioeconomic status and chemotherapy use for melanoma in older people. Can J Aging. 2011;30(1):143-153.

52.

Reyes-Ortiz

Goodwin

Freeman

Kuo

. Socioeconomic status and survival in older patients with melanoma. J Am Geriatr Soc. 2006;54(11):1758-1764.

53.

Walsh

Silles

O'Neill

. The role of private medical insurance in socio-economic inequalities in cancer screening uptake in Ireland. Health Econ. 2012;21(10):1250-1256.

54.

Cormier

Xing

Ding

, et al. Ethnic differences among patients with cutaneous melanoma. Arch Intern Med. 2006;166(17):1907-1914.

55.

Friedman

Bruce

Weinberg

Cooper

Yen

Hill

. Early detection of skin cancer: racial/ethnic differences in behaviors and attitudes. J Cancer Educ. 1994;9(2):105-110.

56.

Hill-Briggs

Fitzpatrick

. Overview of social determinants of health in the development of diabetes. Diabetes Care. 2023;46:1590-1598.

57.

Walker

Garacci

Palatnik

Ozieh

Egede

. The longitudinal influence of social determinants of health on glycemic control in elderly adults with diabetes. Diabetes Care. 2020;43:759-766.

58.

Isaacs

Bellini

Biba

Cai

. Health care disparities in use of continuous glucose monitoring. Diabetes Technol Ther. 2021;23(S3):S81-S87.

59.

Tisnado

Adams

Liu

, et al.

What is the concordance between the medical record and patient self-report as data sources for ambulatory care?

Med Care. 2006;44(2):132-140.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.16 MB