Risk of hepatitis C virus infection from tattooing and other skin piercing services

T HE EPI DE MI OL OGY OF HEPA TI TIS C IN CANADA IS NOT WELL de scribed. Re ported cases of newly di ag nosed hepa ti tis C vi rus (HCV) in fec tions con tinue to rise (1321 cases in 1992, 1634 in 1993 and 2588 in 1994 [pro vi sional]) (1, per sonal com mu ni ca tion), and there are good in di ca tions that the ma jor risk fac tors of preva lent in fec tions are in jec tion drug use fol lowed dis tantly by past blood trans fu sion (2). The pro por tion of acute HCV in fec tions at trib ut able to un ex plained ex po sures is of ten quoted as be ing in the or der of 40% (3). This fig ure ap pears to be an over es ti mate for Can ada. How ever, in re view ing the risk of trans mis sion of other bloodborne in fec tions, in par ticu lar hepa ti tis B, it is clear that pro ce dures in volv ing the pierc ing of skin, eg, tat too ing, must be con sid ered as a pos si ble means of in fec tion. The Na tional Meet ing on the Pre ven tion and Con trol of Hepa ti tis C, held in De cem ber 1994 and co or di nated by the Labo ra tory Cen tre for Dis ease Con trol (LCDC), rec om mended that “since hepa ti tis C and other bloodborne in fec tions are trans mis si ble through per sonal serv ices such as tat too ing, body pierc ing and elec troly sis, na tional guide lines and a na tional strat egy should be de vel oped for the pre ven tion of bloodborne in fec tions in these set tings” (4). To aid the pro cess of de vel op ing in fec tion con trol guide lines for ‘pe rsonal serv ices’ and to re view ex ist ing guide lines, the medi cal lit era ture on the epi de mi ol ogy of hepa ti tis C was searched and as sessed. Tat too ing is em pha sized in this re view be cause an abun dance of in for ma tion on that sub ject was avail able. How ever, lit era ture search sum ma ries of other serv ices in volv ing skin pierc ing, ie, acu punc ture, elec troly sis and body/ ear pierc ing are in cluded, be cause con clu sions from the tat too ing stud ies may pos si bly be ex trapo lated to these serv ices.

, and there are good in di ca tions that the major risk fac tors of preva lent in fec tions are in jec tion drug use fol lowed dis tantly by past blood trans fu sion (2). The pro portion of acute HCV in fec tions at trib ut able to un ex plained ex posures is of ten quoted as be ing in the or der of 40% (3). This fig ure ap pears to be an over es ti mate for Can ada. How ever, in re view ing the risk of trans mis sion of other blood-borne in fections, in par ticu lar hepa ti tis B, it is clear that pro ce dures involv ing the pierc ing of skin, eg, tat too ing, must be con sid ered as a pos si ble means of in fec tion.
The Na tional Meet ing on the Pre ven tion and Con trol of Hepa ti tis C, held in De cem ber 1994 and co or di nated by the Labo ra tory Cen tre for Dis ease Con trol (LCDC), rec om mended that "since hepa ti tis C and other blood-borne in fec tions are trans mis si ble through per sonal serv ices such as tat too ing, body pierc ing and elec troly sis, na tional guide lines and a national strat egy should be de vel oped for the pre ven tion of blood-borne in fec tions in these set tings" (4).
To aid the pro cess of de vel op ing in fec tion con trol guidelines for 'pe rsonal serv ices' and to re view ex ist ing guide lines, the medi cal lit era ture on the epi de mi ol ogy of hepa ti tis C was searched and as sessed. Tat too ing is em pha sized in this review be cause an abun dance of in for ma tion on that sub ject was avail able. How ever, lit era ture search sum ma ries of other serv ices in volv ing skin pierc ing, ie, acu punc ture, elec troly sis and body/ ear pierc ing are in cluded, be cause con clu sions from the tat too ing stud ies may pos si bly be ex trapo lated to these serv ices.

WHAT IS TATTOOING?
Tat too ing is the pro cess of punc tur ing the skin with a cluster of fine nee dles con tain ing in deli ble dyes to achieve a perma nent de sign or mark. The nee dles are sol dered on to a move able shaft called the 'ne edle bar', in serted into stain less steel tubes and ster il ized. The bar, which may con tain anywhere from one to 14 nee dles, is dipped in dif fer ent pig ments to out line a de sign and to fill in the ar eas with col our. The needle, bar and tube as sem bly are mounted onto a DC-po wered ma chine that vi brates the nee dle bar, caus ing the nee dle to pro trude out of the base of the tube (per sonal com mu ni cation). Other skin pierc ing serv ices are re lated to tat too ing in that they in volve nee dles pierc ing the skin.

EVIDENCE OF HCV TRANSMISSION
In 1991, Abild gaard and Pe terslund (5) re ported an in cident of HCV trans mis sion by a tat too ing nee dle. The pa tient had no other risk fac tor, such as his tory of blood trans fu sion, in jec tion use or mul ti ple sex ual part ners. How ever, two months be fore on set of symp toms, he had been tat tooed by a wan der ing mer chant. Though he had no pre vi ous medi cal his tory, he was re peat edly posi tive for an ti bod ies to HCV, and nega tive for hepa ti tis B vi rus, hepa ti tis A vi rus, and hu man immu no de fi ciency vi rus.

EPIDEMIOLOGICAL STUDIES
Sev eral HCV stud ies have in cluded ques tions about tat tooing. How ever, in the ma jor ity of these stud ies tat too ing played a mi nor role, and con se quently their de signs do not adequately ad dress tat too ing as a po tential risk fac tor. The three stud ies dis cussed be low rep re sent those stud ies that best ad dressed the as so cia tion be tween tat too ing and HCV.
In 1992, Ko et al (6) con ducted a cross-sectional study of Tai wan ese mili tary re cruits sam pled on tat too status. Non tattooed per sons were matched to tat tooed per sons by sex, age, edu ca tion, oc cu pa tion and geo graphic ori gin of their par ents. LCDC RE PORT Con found ing was ad di tion ally con trolled by ex clu sion of partici pants with mul ti ple sex ual part ners and in jec tion drug use.
All eli gi ble tat tooed per sons and their matched con trols who had suf fi cient amount of se rum sam ple for anti-HCV testing (87 tat tooed, 126 non tat tooed) were in cluded in the study. An odds ra tio (OR) of 5.9 (95% CI 1.6 to 22) was found among tat tooed per sons com pared with non tat tooed per sons. χ 2 tests for trend were per formed for vari ables that were ranked, and these in cluded vari ables fur ther de scrib ing tat too ing prac tices: per sons with multiple-site tat toos were at greater risk of in fec tion than those with single-site tat toos or their nontat tooed con trols (χ 2 test for trend P=0.002); and per sons with tat toos done by non pro fes sion als were at a higher risk of infec tion than per sons with pro fes sional tat toos or their non tattooed con trols (χ 2 test for trend P=0.002).
One limi ta tion of the cross-sectional study de sign is that tem po ral ity can not be es tab lished. How ever, due to the nature of the ex po sure in ques tion, the cross-sectional de sign has its mer its. Hav ing a tat too is an ex po sure that is memo rable and that does not vary over time, thereby avoid ing any bias stem ming from cit ing cur rent ex po sure prac tices rather than the more etio logi cally im por tant his tori cal ones.
Mis clas si fi ca tion bias of dis ease and con found ing fac tors can not be avoided or quan ti fied, how ever. Spe cifi cally, current in jec tion drug use or mul ti ple sex ual part ners may be under re ported. Since this mis clas si fi ca tion is non dif fer en tialmean ing that the rates of mis clas si fi ca tion be tween ex posed and non ex posed are com pa ra ble -the bias re duces the relative risk es ti mate (7). In con trast, mis clas si fi ca tion of dis ease caused by rela tively non spe cific labo ra tory tests only (du plicate en zyme im mu no sorbent as say) re sulted in an over es timate.
An other is sue is the ex ter nal va lid ity of the re sults. Tat tooing prac tices in Tai wan may not mir ror Ca na dian tat too ing prac tices, but they proba bly do over lap. Tat too ing is le gal in Can ada, and there is no rea son to think that every tat too done in Can ada is done un der hy gi enic con di tions. Con se quently, the tat tooed men in this Tai wan ese study popu la tion are proba bly com pa ra ble with many Ca na di ans who got tat tooed in vari ous situa tions, rang ing from the lo cal tat too par lour to the tat too par lour in a for eign coun try while on mili tary serv ice, or to the fel low prison in mate who of fered his or her tat too ing serv ices. Dif fer ences in preva lence rates of in fec tion, however, may be an im por tant dif fer ence be tween the two popu lations, since base line preva lence rates af fect the rela tive risk of tat too ing and HCV in fec tion.
Though the re sults should rep re sent a good es ti mate of the true ef fect, since the study de sign and analy sis mini mized po ten tial bi ases, the wide con fi dence in ter vals of the re sul tant rela tive risks limit their credi bil ity. Nev er the less, be cause the as so cia tion be tween tat too ing and HCV in fec tion was sig nificant, and be cause the lower bounda ries of the con fi dence inter vals are ap pre cia bly above one, these re sults show on a quali ta tive level that tat too ing is a risk fac tor for HCV in fec tion.
In the sec ond study, Kal dor et al (8) car ried out a casecontrol study of Aus tra lian blood do nors. Po ten tial cases com prised anti-HCV posi tive per sons who gave blood between Feb ru ary 1990 and April 1991. Pro spec tive con trols were all HCV-neg ative do nors iden ti fied in the last four months of the study pe riod (Janu ary to April 1991). Risk fac tor in forma tion was ob tained by oral ques tion naire; anti-HCV status was de ter mined by ELISA and con firmed by re com bi nant immunoblot as say (RIBA) -1. Two-hundred and twenty cases and 210 con trols were re cruited, with 74% and 67% re sponse rates, re spec tively.
Over all risk of HCV in fec tion and among spe cific sub groups were ana lyzed. Here, only the analy sis ex clud ing po ten tial con foun ders is of in ter est: when the analy sis was re stricted to non in jec tion drug us ers and/or non blood trans fu sion re cipients and strati fied by age and sex, a sig nifi cant as so cia tion be tween HCV in fec tion and pres ence of a tat too was found (Mantel-Haenszel ad justed OR=27; 95% CI 8.4 to 87; P<0.00001).
The fol low ing study de sign is sues should be noted. First, con found ing was ade quately con trolled for by re stric tion. Second, there may have been a se lec tion bias: par tici pa tion rates were low -74% for the cases and 67% for the con trolswhich may have com pro mised the in ter nal va lid ity of the study. In fact, the authors found that more of the non re sponders were male and that they were younger than the par ticipants. Third, re call bias is a con cern in case-control stud ies. How ever, since in this study the ques tion naire did not in quire about the de tails of the ex po sure (which would nor mally be the area where the cases would give bet ter in for ma tion than the con trols), the cases had lit tle op por tu nity to give more accu rate in for ma tion than the con trols. Fourth, ex po sure misclas si fi ca tion bias is a much greater con cern here, since the cases and con trols could eas ily give mis lead ing an swers about their in jec tion drug use with out the study co or di na tors be ing able to con firm or deny it. Fifth, the de gree of dis ease mis clas si fi ca tion de pends on the type of labo ra tory test used to de ter mine HCV in fec tion. In this case, be cause they used a screen ing test as well as a sup ple men tary test, dis ease misclas si fi ca tion was proba bly mini mal. The sixth con cern is exter nal va lid ity: ex po sures of blood do nors may not re flect those of the gen eral popu la tion. How ever, con duct ing a study from blood do nor sam ples is cost ef fec tive and gen er ates infor ma tion some what in dica tive of the low risk seg ment of the gen eral popu la tion.
The re sults show that tat too ing is a sig nifi cant risk fac tor for HCV in fec tion. Since the uni vari ate analy sis ex clud ing injec tion drug us ers and blood trans fu sion re cipi ents is sig nificant, and be cause the study was not se verely bi ased, it is safe to say that in this study HCV-i nfected per sons were more likely to be tat tooed than con trols.
The fi nal study is a matched case-control study of Eng lish blood do nors by Good rick et al (9). Cases were re cruited in a non ran dom, con ven ient man ner, and con trols were the first or sec ond do nor that matched the cases' age, sex and time of blood do na tion. Anti-HCV status was de ter mined by du pli cate ELISA, sup ple mented RIBA-2, and where re sults were in de termi nate, HCV RNA po lymerase chain re ac tion was used.
The re sults were as fol lows: when analy sis was re -stricted to non in jec tion drug us ers, HCV-i nfected per sons had a sev en fold greater like li hood of hav ing a tat too than non infected per sons (OR=7, 95% CI 0.9 to 100). This was not sig nificant at the 0.05 level. This study was de signed to ex am ine the as so cia tion between in jec tion drug use and HCV in fec tion rather than tat tooing. While a sam ple size of 50 matched pairs was suf fi cient to ad dress in jec tion drug use as a risk fac tor, it was not suf fi cient to ad dress tat too ing in non in jec tion drug us ers. None the less, though an OR of 7 (95% CI 0.9 to 100) of in fected per sons having a tat too was not sig nifi cant, it proba bly would have been if the sam ple size had been larger.

SUMMARY OF TATTOOING AND HCV STUDIES
Two of the three stud ies dem on strated a sta tis ti cally signifi cant as so cia tion, with the non sig nifi cant study likely to be sig nifi cant given a larger sam ple size. The pri mary strength of the stud ies is that they con trolled for in jec tion drug use.
The im por tant limi ta tions of these stud ies are that, first, only in one study was tat too ing the fo cus of the study; sec ond, only self-reporting of tat too ing, in jec tion drug use and other risk fac tor in for ma tion was avail able; third, none of the stud ies were con ducted on Ca na dian popu la tions; and fourth, in none of the stud ies could tem po ral ity be es tab lished, ie, that the tattoo came be fore the in fec tion. How ever, the so lu tion to that prob lem, con duct ing a co hort study, is not lo gis ti cally fea si ble for this re search ques tion.

OTHER SKIN PIERCING SERVICES
Ear pierc ing: Evi dence for ear pierc ing as a risk fac tor is at pres ent in suf fi cient (9)(10)(11)(12)(13). Three of five epi de mi ol ogi cal studies showed no as so cia tion be tween ear pierc ing and HCV infec tion. Two of the stud ies that showed an as so cia tion were not con vinc ing, ei ther be cause of small sam ple size or poor study de sign in re gard to ear pierc ing. None of the stud ies specifi cally ex am ined ear pierc ing as a pri mary ex po sure vari able, with two stud ies com bin ing ear pierc ing with other per cu ta neous ex po sures, di lut ing any po ten tially sig nifi cant out comes. Body pierc ing: Up to now, no epi de mi ol ogi cal stud ies have ex am ined body pierc ing. How ever, due to the na ture of body pierc ing, it is strongly sus pected to be a rare, but pos si ble, mode of trans mis sion of HCV. Body pierc ing is likely to be carried out in the same situa tions as tat too ing, but the pro ce dures are sim pler. It may be safe to as sume that risks of body piercing could be simi lar to those of tat too ing; how ever, this remains to be seen. Acu punc ture: Evi dence for acu punc ture as a risk fac tor for HCV in fec tion is in suf fi cient (10,11,(13)(14)(15). Four of five stud ies found no as so cia tion be tween acu punc ture and HCV in fec tion. The only study with posi tive re sults was based in a hepa ti tis C-endemic area in Ja pan, where acu punc ture was also a highly preva lent prac tice. Thus, it is un clear whether the results of this study can be at trib ut able to the prac tice of acupunc ture it self or to a high base line of HCV in fec tion and acu punc ture in the study popu la tion. In ad di tion, the study popu la tion of one of the no-association stud ies com prised Japa nese acu punc tur ists -per sons who should be at high risk if acu punc ture is risky -and their HCV preva lence was not differ ent from that of con trols. Elec troly sis: No evi dence is avail able of trans mis sion of HCV by elec troly sis (10,11). Of two stud ies that ad dressed electroly sis, one looked at elec troly sis spe cifi cally; the other study com bined all per cu ta ne ous ex po sures to gether. Nei ther study showed a sta tis ti cally sig nifi cant as so cia tion.

CONCLUSIONS
Based on avail able epi de mi ol ogi cal in for ma tion, tat too ing is the only skin pierc ing serv ice with a docu mented risk of HCV in fec tion, though the spe cific level of risk is un de fined. Accord ing to stud ies, ear pierc ing, acu punc ture and elec troly sis pose lit tle or no risk for in fec tion, al though pub lished work does not ade quately ex am ine the risk. Not yet evalu ated, body pierc ing may be a greater risk for in fec tion than ear pierc ing.
Ex ist ing guide lines and regu la tions re lat ing to per sonal serv ices have been col lected with the co op era tion of provinces and ter ri to ries. LCDC will be pro pos ing to ex am ine these docu ments in the near fu ture in or der to es tab lish na tional guide lines on per sonal serv ices that pierce the skin.