Tattoo practices and risk of hepatitis B and hepatitis C infection in the French Constances study

Highlights

• • •
    The impact of tattoo circumstances on hepatitis B/hepatitis C virus (HBV/HCV) transmission is assessed.
  • •
    One in four tattooed individuals received at least one tattoo outside of a studio.
  • •
    One in five received a tattoo in a country without strict hygiene regulations.
  • •
    Unsafe tattooing practices were related to an increased prevalence of HBV/HCV.
  • •
    Safe tattooing choices were not related to increased HCV or HBV transmissions.

Abstract

Objectives

Reassessing the impact of tattooing circumstances on hepatitis B (HBV) and hepatitis C (HCV) virus transmission to inform the World Health Organization’s hepatitis strategy, which aims to reduce transmission by 90% by 2030.

Methods

Using data from the Cancer Risk Associated with the Body Art of Tattooing (CRABAT) study embedded in the French cohort Constances, we examined the associations of tattooing circumstances with history of HBV and/or HCV infections (self-reported and/or hospital records). Cross-sectional multivariate logistic regression models were fitted in the population aged 45 years and older, as well as retrospective cohort analyses restricted to a subsample of participants with confirmed first tattoo and infection date, all adjusted for known hepatitis risk factors.

Results

Of 77,826 questionnaire respondents, 7.4% (n = 5766) were tattooed, and 3330 (4.9%) answered a complementary tattoo exposure questionnaire in 2023. Tattooing was associated with an increased prevalence of any hepatitis in the multivariate (prevalence odds ratio: 1.49 [95% confidence interval: 1.16-1.91]) and Cox models (hazard ratio [HR]: 1.6 [1.22-2.08]). The strongest risks were found for HCV with tattooing outside studios (HR: 4.14 [2.33-7.35]) and for HBV with tattooing outside countries with regulations (HR: 3.22 [1.39-7.44]).

Conclusions

Unsafe tattooing practices as preventable risk factor for hepatitis transmissions could be underestimated.

Introduction

It is estimated that, worldwide, 254 million people are living with hepatitis B virus (HBV) infection and 50 million with hepatitis C virus (HCV) infection. As a result of these infections, annually, more than 1.3 million people are dying from liver cirrhosis and primary liver cancer [1]. The World Health Organization’s global hepatitis strategy, endorsed by all member states, aims to reduce new infections by 90% and related deaths by 65% between 2016 and 2030 [2]. Major known transmission routes are needle sharing in individuals with injection drug use (IDU) for HCV and vertical mother-child transmission for HBV; other transmission routes account for smaller numbers (e.g. men who have sex with men [MSM], unscreened blood transfusions, sharing of personal hygiene products) [[2], [3], [4]]. However, taken together, these transmission routes explain only about 15% of HBV and 35% of HCV infections in Europe, underlining the need for the identification of further risk factors for transmission [5,6].

Because tattooing has become increasingly common recently, it could be a known yet widely underestimated risk factor of HBV/HCV infections and even lead to an increase in those infections [3,7,8]. Tattoo-associated transmission of various blood-borne diseases has been occasionally reported since the 1950s [9,10]. However, because hepatitis prevention is traditionally focused on IDU and MSM, monitoring tattooing-related hepatitis infections has not become priority. A recent systematic review found a more than 200% and more than 50% elevated risk for HCV and HBV infections among tattooed individuals, respectively [8]. The large majority of the 121 included studies were purely descriptive and not adjusted for known transmission risk factors, equally correlated with tattooing; thus, those risk estimates remain elusive [11].

To better address this question, we used data from the Cancer Risk Associated with the Body Art of Tattooing (CRABAT) study to assess the risk of HBV and HCV transmission after tattooing, adjusting for other transmission risk factors and stratified by different tattooing practices [12]. We further extrapolate the number of HCV transmissions associated with unsafe tattooing in France, illustrating the immediate preventive potential.

Methods

CRABAT is a prospective study of tattoo-associated long-term cancer risk nested in the French national cohort study Constances [12,13]. Constances consists of over 200,000 voluntary participants from the French metropolitan areas, recruited to reflect the countries’ age, sex, and socioeconomic structure. It collects a wealth of self-reported data on sociodemographic and lifestyle factors as well as various medical and paramedical outcomes, starting from the baseline investigations between 2012 and 2019 and subsequent annual follow-up questionnaires. These data are enriched by record linkage to national health databases tracking all medical acts received and (fully or partially) reimbursed by the national public health insurance in which all French citizens are automatically enrolled.

In 2020, the screening question (“Are you tattooed?”) was included in the Constances annual follow-up questionnaire to identify the tattooed study population within Constances for the CRABAT study (Figure 1). From July to December 2023, detailed tattoo exposure data were collected from the tattooed subgroup using a validated tattoo exposure questionnaire [14]. In February 2024, the CRABAT database was established, merging the collected tattoo exposure data with sociodemographic, lifestyle, and medical data for all participants having answered the 2020 follow-up questionnaire. The remaining untattooed Constances participants served as the unexposed comparison group.

Figure 1.

Flow chart of determining the study sample used for the present analyses.

Abbreviation: EpiTAT: Epidemiological Tattoo Assessment Tool (Tattoo exposure questionnaire).

Outcome data

Individual level data retrieved for the period January 1, 2007 to December 31, 2020 from the national health database included information on hospitalization due to HBV/HCV infection, as well as all conducted HBV/HCV tests during the respective years but not the test results. We also retrieved self-reported data on infection status from baseline (“Have you ever been diagnosed with Hepatitis B?/Hepatitis C?”). For the main outcome analysis, we considered participants to have ever been infected with HBV or HCV if they either were hospitalized for the infection (International Classification of Diseases codes B16, B17.0, B17.1, B18.0, B18.1, and B18.2) or if they reported the infection in the baseline questionnaire. To avoid outcome misclassification, self-reported infections were confirmed by at least two tests for the respective infection recorded in the national health databases in the previous 2 years until 2 years after the reported infection date. As this information was only available from January 1, 2007 onward; for earlier self-reported infection dates, we considered that the tattoo preceded the infection because tattoos are usually acquired in early adulthood. Dates of infection were set to the earliest reported date, either recorded by national health data or self-reported infection date.

Tattoo exposure data

Detailed data on visual and contextual factors of tattooing were collected using the validated exposure assessment tool EpiTAT [14]. As exposure variables for the present study, we extracted data on being tattooed, the tattoo setting (five multiple choice categories combined into “Only in tattoo studio,” “At least one tattoo outside a studio”), and the country where tattoos were received dichotomized into having at least one tattoo done outside countries with national tattoo hygiene regulations for at least 10 years (subsequently called “countries with regulation;” countries in Supplementary List 1 [15]). To be used in the retrospective cohort analysis, we also estimated the tattoo date from a question on the time period of tattooing (five multiple choice categories: last year, >1 to ≤5 years ago, >5 to ≤10 years ago, >10 to ≤15 years ago, and 15 years ago and longer). We defined the first tattoo date as the category midpoint of the earliest indicated category. For those who indicated to be tattooed “>15 years and longer,” the first tattoo date was set to 20 before the date of filling in the EpiTAT questionnaire. Therefore, tattoo dates are left-censored, with the earliest date on June 21, 2003.

Sociodemographic and lifestyle covariates

We extracted the following information from the Constances database: age, sex, highest educational level (according to the French education system), birth region (categorized into France metropolitan area; Other European country; Africa, Asia, and French oversea departments; other regions), alcohol consumption (assessed using the Alcohol Use Disorders Identification Test), lifetime number of sexual partners, current condom use, sex of preferred sexual partner (male; female; prefer not to answer) to create the new variable for MSM, and “Ever worked in a medical profession” extracted from occupational history data. Because IDU was not assessed in Constances, we used opioid substitution therapy (either with butaprenorphine or with methadone) at any time since 2007, identified via national health data, as a proxy variable [16]. Furthermore, participants who were hospitalized with substance use disorders or intoxication of any illicit drug were identified via the respective International Classification of Diseases codes (F11-F18; T40.1-T40.7) from the national health database. Participants with HIV, a population with higher HCV/HBV incidence, were equally identified from these data [4,17,18].

Exclusion of data points and sample restrictions

Participants with known first tattoo date after the diagnosis date were excluded from analysis, which led to exclusion of only one observation. Consequently, we assumed that the tattoo date preceded the infection if timeliness of both could not be verified, which is in line first tattoos being usually acquired in early adulthood. Participants with hepatitis infections in childhood were excluded because we assumed that the minimal age to get a first tattoo was 18 years. All analyses were restricted to persons who were aged 45 years or older in 2020 (i.e. the year of the initial tattoo assessment) to prevent screening bias. Although the percentage of participants tested for HBV/HCV was higher in those aged <45 years group, the sample was strongly biased because disproportionally more women (>80%) than men (∼40%) were tested in younger age groups because of routine HBV/HCV screening during pregnancy. Moreover, only few HBV and HCV cases could be identified in younger age groups that is in line with generally decreasing HBV/HCV rates in these age groups but might also partly be due to long periods between infection and symptom development [2]. We also excluded participants who received opioid replacement therapy, participants hospitalized due to illicit drug use, and individuals positive for HIV from the main analyses given their small numbers but strong association with the exposure and outcome. Finally, we excluded participants who were first tattooed after December 31, 2016 because we considered a minimum lag time of 5 years between tattoo-acquired infection and hospitalization.

Statistical analyses

Characteristics of tattoo exposures, sociodemographic, and risk-profile variables were described overall and for tattooed vs non-tattooed participants and by HBV and HCV status. Potential differences across groups were assessed using the chi-square test for categorical variables and two-sided Student’s t-test for continuous variables. We used two different statistical approaches to analyze the relationship in between the dichotomous hepatitis infection status outcomes (HBV; HCV; HBV or HCV) and each of the tattoo exposure variables being tattooed (yes/no), tattoo setting (not tattooed; tattooed in studio; at least one tattoo outside a studio), and tattoo country (not tattooed; tattooed in countries with regulations; at least one tattoo outside those countries).

• 1.Using the whole eligible study sample, multivariate logistic regression models were fitted with two different levels of adjustment: a minimally adjusted model including sex and age and a fully adjusted model that additionally included educational level, alcohol use, number of sexual partners, condom use, men having sex with men, birth region, and ever working in a medical profession. The resulting prevalence odds ratio (POR) and 95% confidence intervals (CIs) display the strength of association and related statistical uncertainty. To assess response bias due to ∼40% of participants with only minimal exposure data, hepatitis risks for binary tattoo exposure were also calculated for responder vs non-responder of our exposure questionnaire. Restricted to the tattooed population, we further explored the interaction in between tattoo setting and country by including an interaction term to the multivariate logistic model. Two more sensitivity analyses were fitted by (i) biological sex and by (ii) level of education (basic [no diploma, general education certificate, and vocational training certificate], regular [high school or equivalent education, and 2-3 years of higher education], higher [4+ years of higher education]).
• 2.A retrospective cohort approach was chosen to assess the exposure-outcome relationship in a time-dependent manner. We used the same covariates for adjustment, except for a categorical variable “recent pregnancy” (yes; no; non-applicable [male sex]) that was added to the fully adjusted models. This analysis was further restricted to study participants with confirmed outcome (either through hospital records or confirmed by two recorded hepatitis tests as described earlier) and/or recorded hepatitis testing to reduce outcome misclassification. Cox proportional hazard models, yielding hazard ratios (HRs) and 95% CIs, were fitted on an age time scale using two different analysis entry points: a first model was fitted, in which participants became at risk at date of birth (= study entry) treating tattoo exposure variables as fixed time-independent exposure (subsequently referred to as dob model). This first model was chosen because the majority of recorded diagnoses occurred in the late 90s, whereas tattoo dates are left-censored at 20 years before second exposure assessment (i.e. from June 21, 2003 to June 31, 2003). In the second model, the study entry date was set to the left-censored tattoo date for the tattooed population and a randomly assigned entry date within this period for the untattooed population (subsequently referred to as 2003 model). Tattoo exposure was treated as a time-varying exposure variable leading to the exclusion of participants without a known tattoo date (i.e. those who did not respond to the EpiTAT exposure questionnaire) even in binary exposure models. The STATA command stsplit was used to split the follow-up time of tattooed records after the date of first tattoo. As the advantage of the 2003 model, the respective participants will contribute person-years analysis time to the non-exposed (untattooed) group until that date.

Post hoc analyses

To get a rough estimate of the public health impact of our results, we estimated the number of HCV infections due to unsafe tattoo practices (i.e. tattooing outside a tattoo studio) in the French population for ages 45 to 75 years. To do so, we (i) multiplied the French population size (United Nations data), stratified by 10-year age bands by the (age-stratified) population proportion of people undergoing unsafe tattoo practices determined by our study [19]; (ii) multiplied the resulting sums with respective age-stratified HCV population prevalence from Constances to estimate the number of HCV cases in the French population per age group; (iii) multiplied these sums with the OR for HCV when being tattooed outside a studio. The resulting numbers reflect the preventable HCV infections in the respective populations of 45-75 years in 2022. We did not estimate attributable cases for HBV because the evidence for an association between tattooing and HBV prevalence was less convincing than for HCV, particularly, for infections acquired outside studios.

Results

After exclusion of non-eligible participants (Figure 1), 7.4% (n = 5766) reported to be tattooed (Table 1). Of the tattooed participants with full exposure data, one-fourth was tattooed outside a studio and one in 10 reported tattoo(s) from outside countries with regulations (Table 1). Tattooed individuals had lower educational levels, showed more often alcohol abuse, had a higher number of sexual partners, and, among men, more often had homosexual intercourse (Table 2).

Table 1.

Distributions of (HBV) and HCV infection status amongst tattoo exposure categories in the CRABAT study population over 45, excluding participants with HIV, with opioid replacement therapy, and with recorded illicit drug use other than cannabis.

Factor	Overall	HBV	HCV	Co-infection HBV/HCV	No infection	P-value
N	77,826	606	321	36	76,863
Biological sex						0.097
Male	35,989 (46.2%)	275 (45.5%)	152 (47.4%)	21 (58.3%)	35,541 (46.2%)
Female	41,837 (53.8%)	330 (54.5%)	169 (52.6%)	15 (41.7%)	41,323 (53.8%)
Age at first tattoo assessment in 2020						<0.001
>45-≤55 years	24,150 (31.0%)	59 (9.7%)	41 (12.8%)	1 (2.8%)	24,049 (31.3%)
>55-≤65 years	25,217 (32.4%)	193 (31.9%)	134 (41.7%)	21 (58.3%)	24,869 (32.4%)
>65 years	28,459 (36.6%)	354 (58.4%)	146 (45.5%)	14 (38.9%)	27,945 (36.4%)
Tattoo status in 2020						<0.001
Not tattooed	72,060 (92.6%)	566 (93.4%)	277 (86.3%)	29 (80.6%)	71,188 (92.6%)
Tattooed	5766 (7.4%)	40 (6.6%)	44 (13.7%)	7 (19.4%)	5675 (7.4%)
Response status to EpiTAT questionnaire						0.076
Responders	3330 (57.8%)	21 (52.5%)	18 (40.9%)	5 (71.4%)	3286 (57.9%)
Non-responders	2436 (42.2%)	19 (47.5%)	26 (59.1%)	2 (28.6%)	2389 (42.1%)
Tattoo context	Among EpiTAT responders					<0.001
Inside a studio	2507 (76.2%)	15 (71.4%)	6 (33.3%)	1 (20.0%)	2485 (76.6%)
Outside a studioa	781 (23.8%)	6 (28.6%)	12 (66.7%)	4 (80.0%)	759 (23.4%)
Tattooed country						0.006
Countries with regulationb	2944 (89.2%)	15 (71.4%)	15 (83.3%)	3 (60.0%)	2911 (89.4%)
No or recent regulation	358 (10.8%)	6 (28.6%)	3 (16.7%)	2 (40.0%)	347 (10.6%)

HBV, hepatitis B virus; HCV, hepatitis C virus.

^aDefined as at least one tattoo outside a professional tattoo studio.

^bDefined as at least one tattoo outside the countries listed in Supplementary List 1.

Table 2.

Demographic characteristics and risk profile, overall and by tattoo status, in the eligible CRABAT study population over age 45 years.

Factora	Overall	Not tattooed	Tattooed	P-value
N	77826	72060	5766
Biological sex				0.84
Male	35989 (46.2%)	33316 (46.2%)	2673 (46.4%)
Female	41837 (53.8%)	38744 (53.8%)	3093 (53.6%)
Age				<0.001
>45-≤55 years	24150 (31.0%)	21031 (29.2%)	3119 (54.1%)
>55-≤65 years	25217 (32.4%)	23318 (32.4%)	1899 (32.9%)
>65 years	28459 (36.6%)	27711 (38.5%)	748 (13.0%)
Geographical region of origin				0.084
French metropolitan area	70740 (90.9%)	65476 (90.9%)	5264 (91.3%)
Europe, other	2877 (3.7%)	2663 (3.7%)	214 (3.7%)
Asia, Africa, French overseas departments	2824 (3.6%)	2647 (3.7%)	177 (3.1%)
Other	1385 (1.8%)	1274 (1.8%)	111 (1.9%)
Highest education				<0.001
No diploma, vocational training	20623 (27.0%)	18345 (25.9%)	2278 (40.2%)
High school diploma	11909 (15.6%)	10837 (15.3%)	1072 (18.9%)
College bachelor diploma	27064 (35.4%)	25323 (35.8%)	1741 (30.7%)
College master diploma and higher	16666 (21.8%)	16102 (22.7%)	564 (9.9%)
Missing or unclassifiable	197 (0.3%)	181 (0.3%)	16 (0.3%)
Alcohol use				<0.001
Abstinent	13330 (17.2%)	12439 (17.3%)	891 (15.5%)
Abuse	5823 (7.5%)	5281 (7.3%)	542 (9.4%)
Dependent	3914 (5.0%)	3602 (5.0%)	312 (5.4%)
Neither abuse nor dependence	53618 (69.0%)	49707 (69.1%)	3911 (68.0%)
Missing	1039 (1.3%)	940 (1.3%)	99 (1.7%)
Ever cannabis use	22989 (29.5%)	20186 (28.0%)	2803 (48.6%)
Number of sexual partners				<0.001
1 partner	13023 (16.7%)	12595 (17.5%)	428 (7.4%)
2-5 partners	19898 (25.6%)	18593 (25.8%)	1305 (22.6%)
6-10 partners	10285 (13.2%)	9253 (12.8%)	1032 (17.9%)
>10 partners	6858 (8.8%)	5985 (8.3%)	873 (15.1%)
Prefer not to answer	27762 (35.7%)	25634 (35.6%)	2128 (36.9%)
Men who have sex with menb				<0.001
Never	33261 (92.4%)	30851 (92.6%)	2410 (90.1%)
Ever	1718 (4.8%)	1526 (4.6%)	192 (7.2%)
Prefer not to answer	1010 (2.8%)	939 (2.8%)	71 (2.7%)
Current condom use				<0.001
Yes	12131 (15.6%)	11046 (15.3%)	1085 (18.8%)
No	47829 (61.5%)	44180 (61.3%)	3649 (63.3%)
Not currently in a relationship	5880 (7.6%)	5433 (7.5%)	447 (7.8%)
Missing	11986 (15.4%)	11401 (15.8%)	585 (10.1%)
Ever working in medical profession	11973 (15.4%)	11078 (15.4%)	895 (15.5%)	0.78
Recent pregnancyc	4599 (11.7%)	4011 (5.6%)	588 (10.2%)	<0.001
Tested for hepatitis B virus	20762 (26.7%)	18779 (26.1%)	1983 (34.4%)	<0.001
Tested for hepatitis C virus	22956 (29.5%)	20763 (28.8%)	2193 (38.0%)	<0.001

^aBecause participants with HIV (n = 284), injection drug use, or other illicit drug use (n = 205) were not eligible for analyses, the respective summary statistics and group differences are not displayed here. For both variables, percentages were significantly higher in the tattooed (1.1% participants with HIV, 0.6% with opioid replacement therapy, 0.4% with other illicit drug use) compared with the non-tattooed (0.3%, 0.1%, and 0.1%, respectively).

^bDenominator for % is only the male population.

^cDenominator for % is only the female population.

According to our criteria, 606 (0.9%) participants had ever been infected with HBV and 321 (0.4%) with HCV alone, and 36 (0.1%) individuals were co-infected with HBV/HCV. HCV infections were more common in tattooed (54 of 5766, 0.9%) than in non-tattooed participants (328 of 72,061, 0.5%) and more participants with HCV and HBV got tattooed outside studios and outside countries with regulations (Table 1). Individuals with hepatitis infections showed a similar sociodemographic profile with the tattooed subgroup (Supplementary Table 1 vs Table 2). Approximately one-third of tattooed participants were tested for HBV or HCV against one-fourth of the participants without tattoos (data not shown). However, tattooing outside a studio was particularly common in untested male participants compared with tested men and women in general (Supplementary Table 2).

Tattooed participants had a 1.49-fold (CI: 1.16-1.91) elevated POR of either hepatitis infections after full adjustment (Table 3). When analyzed by tattoo context, the increase was confined to those who had at least one tattoo outside of tattoo studios and/or outside countries with regulations, culminating in an over seven-fold higher prevalence for those tattooed outside studios and outside countries with regulations.

Table 3.

PORs and 95% CIs of the multivariate logistic regression models on risk of HBV or HCV infection and tattoo exposure variables for two different levels of adjustment in the CRABAT study population over age 45 years. All models, excluding participants with HIV and with opioid replacement therapy.

	HBV or HCV				HBV				HCV
	Model 1a		Model 2b		Model 1a		Model 2b		Model 1a		Model 2b
	POR	95% CI	POR	95% CI	POR	95% CI	POR	95% CI	POR	95% CI	POR	95% CI
Tattooed	1.59	(1.28; 1.98)	1.49	(1.16; 1.91)	1.20	(0.90; 1.61)	1.12	(0.79; 1.59)	2.64	(1.95; 3.58)	2.13	(1.54; 2.95)
Response to EpiTAT	1.12	(0.8; 1.55)	1.14	(0.80; 1.62)	0.88	(0.57; 1.36)	0.92	(0.56; 1.50)	1.94	(1.25; 3.01)	1.70	(1.09; 2.67)
No response to EpiTAT	2.24	(1.70; 2.96)	1.98	(1.44; 2.73)	1.64	(1.13; 2.40)	1.41	(0.88; 2.25)	3.60	(2.45; 5.29)	2.71	(1.78; 4.13)
Tattoo context
Not tattooed	ref		ref		ref		ref		Ref		ref
Only in a studio	0.60	(0.36; 1.01)	0.56	(0.31; 0.99)	0.58	(0.31; 1.08)	0.52	(0.24; 1.10)	0.72	(0.32; 1.63)	0.65	(0.29; 1.48)
Outside a studioc	2.65	(1.74; 4.05)	2.91	(1.87; 4.52)	1.78	(0.97; 3.25)	2.11	(1.11; 4.01)	5.49	(3.29; 9.16)	4.72	(2.79; 7.99)
Tattoo country
Not tattooed	ref		ref		ref		ref		Ref		ref
Countries with regulationd	1.00	(0.69; 1.45)	1.06	(0.71; 1.57)	0.71	(0.43; 1.20)	0.75	(0.42; 1.35)	1.80	(1.11; 2.92)	1.63	(1.00; 2.67)
No or recent regulation	2.15	(1.06; 4.35)	1.85	(0.86; 3.98)	2.29	(1.01; 5.16)	2.12	(0.86; 5.23)	3.21	(1.19; 8.68)	2.47	(0.90; 6.74)
Interaction amongst the tattooed
In a studio in regulating country	ref		ref		ref		ref		ref
Outside a studio in regulating country	4.41	(2.05; 9.48)	5.33	(2.31; 12.29)	2.40	(0.83; 6.89)	3.06	(0.92; 10.13)	7.99	(2.73; 23.36)	8.07	(2.71; 24.00)
Inside a studio outside in regulating country	2.31	(0.64; 8.28)	1.49	(0.30; 7.32)	2.27	(0.48; 10.78)	1.04	(0.1; 10.33)	1.87	(0.22; 16.16)	1.56	(0.18; 13.78)
Outside a parlor studio in regulating country	6.68	(2.26; 19.74)	7.20	(2.29; 22.65)	7.77	(2.24; 26.97)	9.02	(2.27; 35.82)	9.6	(2.21; 41.77)	8.59	(1.93; 38.35)

CI, confidence interval; HBV, hepatitis B virus; HCV, hepatitis C virus; POR, prevalence odds ratio; ref, reference.

^aAdjusted for sex and age at initial tattoo assessment 2020.

^bAdjusted for sex, age at initial tattoo assessment, highest educational level, birth region, alcohol consumption (Alcohol Use Disorders Identification Test score), ever cannabis consumption, number of sexual partners, condom use, men that have sex with men, and ever working in medical profession.

^cDefined as at least one tattoo outside a professional tattoo studio.

^dDefined as at least one tattoo outside the countries listed in Supplementary List 1.

Looking at hepatitis subtypes separately, prevalence was not elevated for HBV overall but was around two times higher if tattooed outside a studio and outside countries with regulations. Although the numbers were small, the interaction of tattoo context and tattoo country showed a strong association, with a nine-fold increased POR for participants tattooed outside studios and outside countries with regulations.

For HCV, prevalence in tattooed individuals was two times higher than in non-tattooed participants after full adjustment. The adjusted POR was almost five times higher in individuals tattooed outside a studio than in those without tattoos and, among those tattooed, >8.5-fold elevated if tattooed outside a studio and outside countries with regulation compared with those tattooed in studio inside regulating countries. For all outcomes, PORs were higher in EpiTAT non-responders than in responders, pointing toward participation bias. Compared with other covariates, tattoo context showed the strongest effect for HCV, whereas for HBV, homosexual intercourse among men was strongest associated with infection (Supplementary Figures 1 and 2). Sensitivity analyses stratified by biological sex or socioeconomic categories confirmed results, although the case numbers were very small (Supplementary Table 3 and 4).

The results pattern seen in retrospective cohort analyses was similar, although hampered by small case numbers, particularly, in the analyses for context-specific or time-dependent exposure variables, resulting in more imprecise estimates in these models (Supplementary Tables 5 and 6). In the fully adjusted models, being tattooed was associated with a ∼60% higher HR of either hepatitis infection in the dob model, with a slightly higher estimate in the 2003 model. HRs were highest in the 2003 model for tattoos outside studios, such as 4.5-fold elevated and almost three-fold for tattoos outside regulating countries, although the latter did not reach statistical significance.

When analyzing by hepatitis type, compared with the untattooed participants, HBV risk was ∼42% higher in individuals with tattoos in the dob model (Figure 2). HRs were over two times elevated for participants with tattoos outside studios and over three times for those tattooed outside regulating countries; for the latter, the HR was almost six-fold elevated in the 2003 model but based on only three cases in exposure this category.

Figure 2.

Hazard ratios and 95% CIs for tattoo associated hepatitis B infection (a) and hepatitis C infection (b) by tattoo setting for different analysis entry dates in the eligible CRABAT study population over the age of 45 years, excluding participants living with HIV and those who received opium replacement therapy as proxy for injection drug use. Models are adjusted for sex, age at initial tattoo assessment, highest educational level, birth region, alcohol consumption (Alcohol Use Disorders Identification Test score), ever cannabis consumption, number of sexual partners, condom use, men that have sex with men, ever working in medical profession, and recent pregnancy.

Abbreviation: CI, confidence interval.

For HCV, risk was ∼90% elevated in the population with vs without tattoos in both models, driven by a higher risk in participants tattooed outside studios. HCV infection risks in this group were around four- and six-fold elevated in the dob and 2003 models, respectively, compared with untattooed participants.

In the exploratory analyses adjusted for sex, age, being tattooed and either HIV status or opioid replacement therapy, ever having had an opioid replacement therapy (with an POR of 6.95 [CI: 3.03-15.97] for HBV and of 34.33 [CI: 19.61-60.12] for HCV), as well as being HIV-positive (with a POR of 16.22 [CI: 10.90-24.16] for HBV and of 21.60 [CI: 14.12-33.03] for HCV) were strongly related to hepatitis infections, confirming our rationale to exclude them from the main analysis (data not shown).

Extrapolating the prevalence rates of HCV and unsafe tattooing practices from our study to the French population aged 45-75 years in France revealed 11,592 HCV transmissions attributable to tattooing outside a studio (Supplementary Table 7). A large majority of these cases were in the age category from >55 to ≤65 years.

Discussion

In this analysis, we found an over two-fold increased adjusted POR of HCV infection in tattooed vs non-tattooed individuals in a cross-sectional analysis, whereas the prevalence for HBV infection alone was not increased. Although tattoos done in tattoo studios were safe, we found an elevated prevalence odds ratios for HCV and, to a lesser extent, HBV associated with tattooing outside tattoo studios and also outside countries with regulations. The same pattern was seen in a retrospective cohort analysis when analyzing a subset of individuals with recorded hepatitis tests and/or hospital-confirmed diagnoses. Context-specific risk estimates were particularly elevated when treating tattooing as time-varying exposure (e.g. six-fold for HCV when tattooed outside a studio), but small case numbers led to imprecise estimates.

To put these results into context, we estimated ∼11,600 avoidable prevalent HCV infections in France based on HCV prevalence rates in Constances, our logistic regression results, and United Nations population estimates. However, because this estimation assumes causality and absence of bias in any of the estimated parameters, it only intends to provide a rough estimate of the public health impact and needs to be interpreted cautiously. Nevertheless, the results underscore unsafe tattooing as a potentially underestimated risk factor for HCV/HBV transmission and, from our data, would make unsafe tattooing the second most important avoidable risk factor after IDU for HCV.

To the best of our knowledge, this is the first epidemiologic study to estimate the risk of HCV and HBV infection associated with different tattooing practices since over 15 years, which is a public health priority given the already increasing popularity of tattooing worldwide and the recent rebound of new HCV infections in many parts of the world [20], [21], [22]. (Notably, in many European countries, HCV incidence seem to decline; however, in many countries, this estimation is based on the prevalence, treatment, and prevention programs in people with IDU alone [23,24]).Since the 1990s, industrialized countries implemented hygiene standards for tattoo studios in which tattooists today practice under sterile conditions [15]. However, given that tattooing outside studios is common and tattooing is a global trend (with a tattoo prevalence of 22% in Brazil and 12% in China and Russia) and many countries lack hygiene regulations, infection risk from poor unsafe tattooing is unlikely eradicated [25]. Furthermore, because, compared with other risk groups, the tattooed population is less often screened, the number of tattoo-related viral infections might be underestimated [26,27].

Strengths of our study are the rich data of Constances and the validated detailed exposure assessment [14]. Results for known transmission pathways are as expected, confirming the generalizability of our results. However, there are also limitations. Although we could assure timeliness of exposure and outcome in a subset of participants, our main analyses were cross-sectional. Because most people get tattooed in early adulthood, it seems likely that tattooing would precede diagnosis, but retracing viral transmissions to an infectious event is generally error-prone. Also, outcome ascertainment relied partly on self-reported data that was for a subgroup diagnosed after 2007, confirmed by testing recorded in national insurance records. To account for timeliness of exposure and outcome, we identified a subset of participants in which outcome misclassification was less likely to complement the cross-sectional results with a retrospective cohort analysis. Although the results of these analyses generally confirm our results, even in this subset, some outcome and/or exposure misclassification is likely because of recall bias and left-censoring of the first tattoo date and missing test but without confirmation of serology status.

Moreover, the analyses were restricted to the age group >45 years. Tattoos associated HBV/HCV risk might be attenuated in younger age groups, assuming decreasing hepatitis prevalence in the population and that today’s tattooing conditions outside studios benefit from improvements of hygiene standards inside studios. However, low HBV/HCV rates in younger age groups in Constances might also be due to under-presentation of routinely screened high-risk groups in a lifetime cohort with voluntary participation (healthy cohort effect).

Finally, opioid replacement therapy is an imperfect proxy for IDU because individuals with former IDU might never have needed replacement therapy or have stopped before 2007. However, Constances is a population-based cohort with voluntary participation and, as such, healthier than the general population, lowering the likelihood of a high number of participants that do or ever injected drugs. Moreover, the adjusted prevalence ratios were the highest for tattooing outside a studio in the higher education group, which is the least likely to engage in IDU; thus, the bias seems small [10], [28], [29].

Conclusion

Given that unsafe tattooing was among the strongest risk factors in our study, medical personnels should be aware of this risk and recommend testing to participants with unsafe tattoos to prevent a global increase of HBV/HCV infections. Finally, the results call for future research. Seroprevalence studies on tattooing-associated viral infections on a global scale are needed to address associated risk of blood-borne virus infections, particularly in endemic areas.

Funding

The CRABAT study was supported by the French National Cancer Institute (INCa; grant No. 2021-137). None of these funding sources had any role in the design of the study, collection and analysis of data or decision to publish. The Constances cohort study was supported and funded by the French national health insurance fund (“Caisse nationale d’assurance maladie,” Cnam). Constances is a national infrastructure for biology and health (“Infrastructure nationale en biologie et santé”) and benefits from a grant from the French national agency for research (ANR-11-INBS-0002). Constances is also partly funded to a small extent by industrial companies, notably in the healthcare sector, within the framework of Public-Private Partnerships.

Ethical approval

Reviewed and approved by Western Institutional Review Board; approval #472920.

Author contributions

Milena Foerster: Conceptualization, formal analysis, data curation, visualization, funding acquisition, investigation, methodology, project administration, writing—original draft, writing— review & editing. Marie Zins: Conceptualization, investigation, project administration, resources, writing—review and editing. Marcel Goldberg: Conceptualization, investigation, project administration, resources, writing—review and editing. Celine Ribet: Investigation, project administration, resources. Sofiane Kab: Investigation, project administration, resources, software. Rachel D McCarty: Writing— review & editing. Valerie McCormack: Methodology, writing— review & editing, validation. Khaled Ezzedine: Conceptualization, investigation, project administration, funding acquisition, resources, writing—review and editing. Joachim Schüz: Conceptualization, funding acquisition, investigation, methodology, validation, supervision, writing—review and editing.

Ethics statement

The CRABAT study received approval by the IARC Ethics Committee (IEC 22-02) and was authorized by the CNIL (#22015584). The Constances study was approved by the Institutional review board of the French Institute of Health (Inserm) (Opinion n°01-011, then n°21-842) and authorized by the by the French Data Protection Authority (“Commission Nationale de l’Informatique et des Libertés”, CNIL) (Authorization #910486).

Data availability statement

Due to European data protection regulations, the CRABAT data cannot be made available to the public. In accordance with the Constances Charter, de-identified participant data from the Constances cohort are available to researchers who meet the legal and ethical requirements set by the French National Commission governing data privacy laws. International researchers can access the data set at https://www.constances.fr/en/scientific-area/access-to-constances-2/. In addition, all study materials, including the study protocol and data dictionary of the Constances cohort, are freely accessible. For more information on data usage and interest in research collaboration, please contact Milena Foerster (foersterm@iarc.who.int).

Declaration of use of generative AI and AI-assisted technologies

No generative AI and AI-assisted technologies were used in the manuscript preparation process.

Disclaimer

Where authors are identified as personnel of the International Agency for Research on Cancer/World Health Organization, the authors alone are responsible for the views expressed in this article and the views do not necessarily represent the decisions, policy, or views of the International Agency for Research on Cancer/ or the World Health Organization.

Declaration of competing interest

The authors have no competing interests to declare.