Screening and treatment of immigrants and refugees to Canada for tuberculosis : Implications of the experience of Canada and other industrialized countries

Elisabeth Bruyère Health Centre, University of Ottawa, and Quarantine Health Services, Laboratory Centre for Disease Control, Health Canada, Ottawa, Ontario Correspondence: Dr RE Thomas, Elisabeth Bruyère Health Centre, 75 Bruyère Street, Ottawa, Ontario K1N 5C8. Telephone 613-241-3344, fax 613-241-1971 Received for publication January 9, 1995. Accepted June 13, 1995 RE THO MAS, B GUSHU LAK. Screen ing and treat ment of im mi grants and refu gees to Can ada for tu ber cu lo sis: Im pli ca tions of the ex pe ri ence of Can ada and other in dus tri al ized coun tries. Can J In fect Dis 1995;6(5):246255.

T HE PUR POSES OF THIS PA PER ARE TO AS SESS, FIRST, THE IN CI - dence of ex po sure to tu ber cu lo sis and the pres ence of ac tive tu ber cu lo sis in im mi grants and refu gees in Can ada and, sec ond, in other in dus tri al ized coun tries; third, the ef fective ness of the screen ing pro cess for tu ber cu lo sis in im migrants and refu gees in Can ada; fourth, the in ci dence of drug re sis tance in im mi grants; fifth, rec om men da tions for pro phylaxis and ther apy and the com pli ca tions of ther apy; and fi nally, in di ca tions for and ef fec tive ness of di rectly ob served ther apy (DOT).
A MED LINE lit era ture search for 1985 to 1995 was conducted us ing 't ube rc ul osis' and the medi cal sub ject head ings 'Ca nada' (54 ti tles); 'i mm igr ation and emi gra tion' (162 ti tles); 'drug re sis tance' and 'drug re sis tance, mul ti ple' (66 ti tles); and the ti tle words 'd irectly ob served ther apy' (10 ti tles).

TU BER CU LO SIS IN IM MI GRANTS AND REFU GEES IN CAN ADA
Forty-two per cent of tu ber cu lo sis pa tients in Can ada from 1985 to 1987 were born out side Can ada (1).Im mi grants have rates of tu ber cu lo sis five times higher than those of Ca na dians who are nei ther im mi grants nor Abo rigi nal.From 1970 to 1990, rates fell in im mi grants of Euro pean ori gin but rose in those of Asian ori gin (2).In a study of 7028 im mi grants in Mont real, 15% of those from coun tries with low rates of tu bercu lo sis, 26% of those from in ter me di ate coun tries, and 35% of those from coun tries with high rates were Mantoux-positive (de fined as greater than 15 mm if they had been vac ci nated with Ba cille Calmette-Guérin, greater than 10 mm if not) (3).
Groups with sig nifi cant tu ber cu lo sis rates other than im migrants are the aged (29% of cases are in those over 65 years of age), Abo rigi nals (17.5% of cases) and the ur ban in di gent.Fifty per cent of pe di at ric tu ber cu lo sis cases in Can ada are Abo rigi nal chil dren (4).
There has been a steady de cline in rates in the Inuit (from 580/100,000 popu la tion in 1970-72, to 80/ 100,000 in 1983-89) and Status In dian popu la tions (from 150/100,000 in 1970-72, to 80/100,000 in 1983-89) (1).Cen sus di vi sions with rates over 20/100,000 are all in north ern Can ada (80% of which are in the Abo rigi nal popu la tion), and those with rates of 10 to 19/100,000 are ei ther in north ern re gions or in ma jor met ro poli tan ar eas (with a dis pro por tion ate number of cases among the im mi grant and Abo rigi nal popu la tions).
In Can ada, com pared with the United States, there are rela tively few pa tients who both have tu ber cu lo sis and are human im mu no de fi ciency vi rus (HIV) posi tive.In Brit ish Co lum -bia it is es ti mated that since the HIV epi demic be gan fewer than 2% of the ac tive cases of tu ber cu lo sis have HIV-pos itive status (5).How ever, the rela tive risk of tu ber cu lo sis in a person who is HIV-pos itive is 400 times greater than the risk in the gen eral Ca na dian popu la tion, and thus HIV status con sti tutes the most im por tant risk fac tor for tu ber cu lo sis (6).

TU BER CU LO SIS IN IM MI GRANTS AND REFU -GEES IN OTHER IN DUS TRI AL IZED COUN TRIES
In Can ada, the United States, Aus tra la sia, Europe and Israel, im mi grants have a con sid er able bur den of la tent tu bercu lo sis with few clini cal signs.The high est rates of clini cal tu ber cu lo sis oc cur in im mi grants from coun tries with high preva lence rates and in the first five years af ter im mi gra tion.How ever, higher rates per sist even 20 to 40 years later in immi grants from coun tries with high preva lence rates (7,8).
Aus tra lia has an im mi grant popu la tion of simi lar ori gin to that of Can ada.In a study of 2290 in ner city school chil dren in Syd ney, 27.5% of the foreign-born had a posi tive Man toux com pared with 2% of Australian-born chil dren (9).If the overseas-born ar rived in Aus tra lia af ter they were 10 years of age, they were twice as likely to have a posi tive tu ber cu lin test than if they were younger than 10 years on ar ri val.Im mi grants to Aus tra lia born in Viet nam, Cam bo dia and the Phil ip pines had tu ber cu lo sis rates be tween 100 and 400/100,000 compared with 1.5/100,000 for those born in Aus tra lia (10).The in ci dence of tu ber cu lo sis in those with AIDS was 1579/100,000.
In New Zea land, tu ber cu lo sis rates were 10 times higher in Pa cific Is land ers, 15 times higher in other im mi grant groups and five times higher in Mao ris than in those of Euro pean origin (11).The age of the im mi grants was a key risk fac tor: compared with those aged 0 to 14 years, the rates in those 15 to 24 years were four times higher; in those 25 to 44 years six times higher; in those 45 to 64 years 12 times higher; and in those over 65 years 16 times higher.
In Europe, tu ber cu lo sis no ti fi ca tions stead ily de clined un til 1985, but since then have in creased es pe cially in drug-users, in HIV-pos itive pa tients and in re cent im mi grants from sub-Saharan Af rica.In the Vaud can ton of Swit zer land 256 cases of tu ber cu lo sis were found by mass minia ture x-ray test ing among 50,784 im mi grants: 1.4% of those from Af rica, 1% from Tur key, 0.62% from the Mid dle East and 0.55% from Yugo sla via were di ag nosed as need ing pro phy laxis (12).In a Bel gian pe di at ric home care cen tre 80% of the tu ber cu lo sis re -«tu ber cu lose», «Can ada», «dépistage», «résis tance médica men teuse», «résis tance médica men teuse mul ti ple», «immi gra tion et émi gra tion», et les mots ti tres «traite ment di recte ment ob servé».SÉLEC TION DES ÉTUDES : Toutes les études fai sant état de données sur l'in ci dence, les résul tats de dépistage, la résistance médica men teuse, le traite ment et le traite ment di recte ment ob servé ont été re te nues.SYN THÈSE DES DONNÉES ET CON CLU SION : La syn thèse des données mène à la con clu sion que les im mi grants et les réfu giés qui ar rivent au Can ada souf frent cinq fois plus de tu ber cu lose que les non-immigrants et les non-aborigènes; que les pro grammes de dépistage dé cou vrent de nou veaux cas de tu ber cu lose à l'ar rivée et après, que jusqu'à 20 % des su jets at ten dus dans les clin iques de tu ber cu lose après leur ar rivée ne s'y pré sen tent pas, que la résis tance médica men teuse est répan due dans les pays d'où ar rivent les im mi grants pour le Can ada et que le traite ment di rectement ob servé de cas ac tifs peut gé né rer des taux de con ver sion de l'or dre de 100 %.
fer rals were im mi grants; 48% of the chil dren who were sympto matic had an iden ti fi able adult source (13).Tu ber cu lo sis rates among im mi grants to the United King dom in 1988 were 135/100,000 in In di ans and 101/100,000 in Paki stanis and Bang la deshis com pared with 4.7/100,000 in the popu la tion born in the United King dom (14).Hindu fe males who mi grate to the United King dom have rates of tu ber cu lo sis 5.5 times higher than Mus lim fe males, and Hindu males 3.7 higher than Mus lim males.One hy pothe sis to ex plain this is that the Hindu vege tar ian diet re duces im mu nity (15).Par ticu lar con cen trations of HIV-pos itive pa tients with tu ber cu lo sis have been noted: 12% of tu ber cu lo sis pa tients in Paris and 18% in a neigh bour ing Dépar te ment were HIV-pos itive; 10% of AIDS patients in Frank furt and It aly and 37% in Spain had tu ber cu lo sis (16).Dur ing the 1980s HIV emerged as the most ef fec tive promoter of tu ber cu lo sis, with rates of re ac ti va tion of 8% per year (17).
In the United States, tu ber cu lo sis is 13 times more prevalent among the im mi grant popu la tion than among the nativeborn (18).In a study of il le gal Mexi can im mi grants in Den ver, 42% had Man toux tests greater than 10 mm, and 17% of the chest x-rays ob tained showed old or cur rent tu ber cu lous infec tion (19).Im mi grants to Se at tle from Kam pu chea and Laos had a 10-fold greater risk of hav ing a posi tive spu tum cul ture than other im mi grants (20).Fifty-eight per cent of foreign-born stu dents reg is ter ing at the Uni ver sity of Cali for nia at Los Ange les in 1988 had posi tive Man toux re ac tions (21).The United States is a ma jor res er voir of 10 to 15 mil lion peo ple with tu ber cu lo sis (22), with high con cen tra tions among very dis ad van taged groups: 50% of the home less (23) and 18% of those liv ing in shel ters (24).These pa tients are a large highrisk group, but few can mi grate to Can ada from the United States, and im mi grants from other parts of the world will there fore have a greater im pact on Cana da's tu ber cu lo sis rates.
Two per cent of Ethio pian im mi grants to Is rael had ac tive tu ber cu lo sis (25).
Im mi grants tend to have high rates of ex trapul mon ary disease (26).In one study in the United King dom 30% of Asian di aly sis pa tients pre sented with ex trapul mon ary tu ber cu lo sis, al though they had lived in the United King dom, on av er age, for 15 years with out any clini cal signs of tu ber cu lo sis.Asian di aly sis pa tients now rou tinely re ceive tu ber cu lo sis pro phylaxis (27).

THE EF FEC TIVE NESS OF THE SCREEN ING PROC ESS FOR IM MI GRANTS AND REFU GEES
The World Health Or gani za tion (WHO) es ti mates that 1.7 bil lion per sons world wide are or have been in fected with tuber cu lo sis, with each year 8 mil lion new cases of ac tive tu bercu lo sis and 2.9 mil lion deaths (28), and 100 mil lion with drug-resistant strains (29).There are 3.1 mil lion peo ple who have tu ber cu lo sis and are HIV-pos itive (30).Tu ber cu lo sis control in Third World coun tries re ceives only two cents of every 10 dol lars spent on health (29).
Screen ing for ex po sure to tu ber cu lo sis and the find ing of ac tive cases in both im mi grants and refu gees and the rest of the popu la tion de pend on both an ex cel lently or gan ized national pro gram, and the per sis tent ap pli ca tion of history-taking and clini cal skills of in di vid ual phy si cians.The key ele ment in de tec tion is an ap pro pri ate level of sus pi cion that the pa tient could be as ymp to mati cally in fected (Ta ble 1).
The screen ing re quire ments of the Ca na dian im mi gra tion authori ties in clude a full his tory, physi cal ex ami na tion and labo ra tory in ves ti ga tions (Ta ble 2).The Ca na dian Tho racic So ci ety, the Ca na dian Lung As so cia tion, the Tu ber cu lo sis Direc tors of Can ada and Health Can ada sug gest in ad di tion a tu ber cu lin test and cul tures (Ta ble 2).
De tailed evi dence on how this sur veil lance and treat ment sys tem works is avail able for Mani toba, Brit ish Co lum bia and

Screen ing for ex po sure to tu ber cu lo sis
Prob lem: Whom to screen So lu tion: Pay me ticu lous at ten tion in history-taking to risk fac tors (coun try of ori gin, ex posed fam ily mem ber, al co hol ism, HIV status, prior tuber cu lo sis ther apy), have a high sus pi cion for in di vidu als with risk fac tors, take a care ful his tory of past an ti tu ber cu lous medi ca tion, ob tain a tu ber cu lin test (and the two-step in non re ac tors) and a chest x-ray in any in di vid ual with any risk fac tors

Case-finding of ac tive tu ber cu lo sis
Prob lem: Where to di rect case-finding ac tivi ties So lu tion: Have a high sus pi cion for any in di vid ual with any risk fac tor or symp tom, re mem ber ing that: • In fec tion at any pe riod of life con fers a 10% life time risk of tu ber cu lo sis • Al though 85% pres ent as pul mo nary tu ber cu lo sis, im mi grants are more likely to pres ent with non pul mon ary sites • Al though most in fants and chil dren pres ent with clas sic res pi ra tory symp toms, fe ver and weight loss, most cases in im mi grants are found by con tact trac ing • Al though 80% pres ent with the symp toms of an ac tive in fec tion (cough, fe ver, mal aise), 20% are as ymp to matic • The tu ber cu lin test may be falsely nega tive in 20%, es pe cially in in fants a few months of age, the eld erly, those with ad vanced pul mo nary dis ease, tu ber cu lous se rosi tis, those who are HIV-positive and other im mu no sup pressed pa tients • The chest x-ray may be nor mal in 4%, and 20% are atypi cal (es pe cially in the HIV-positive) • The spu tum cul ture may be nega tive in 15%  (31).Be fore they emi grated, 1173 (5.3%) had been judged to have in ac tive tu ber cu lo sis; on ar ri val in Can ada 14 (1.5% of the 932 who were ex am ined) were found to have ac tive tu ber cu lo sis, and a fur ther seven de vel oped ac tive tu ber cu lo sis over the next four years (0.33% per year, eight times the Brit ish Co lum bia rate).Ninety-four of the group ex am ined in their coun try of ori gin and thought to have in ac tive tu ber cu lo sis ei ther did not emi grate or landed in a prov ince other than Brit ish Co lum bia, and 147 who landed did not com ply with the man da tory screen ing pro gram af ter ar ri val.
Dur ing 1979 to 1982, 8692 In do chi nese refu gees emigrated to Brit ish Co lum bia, and the im mi gra tion serv ice was aware of 21 cases of ac tive tu ber cu lo sis be fore the im mi grants en tered Can ada (32).Seventyfive more cases were dis covered af ter en try: 28 in the first three months, four in the next three months, 16 in months seven to 12, and 16 in months 12 to 24 of resi dence.
Of the im mi grants ad mit ted to Mani toba be tween 1981 and 1985, 523 (2.4%) were placed un der sur veil lance for tu ber culo sis, and their sub se quent in ci dence of tu ber cu lo sis was 4.5 times higher than in those not placed un der sur veil lance (33).Of the 429 who were seen at least once in the tu ber cu lo sis clinic, eight had had a posi tive cul ture in their coun try of ori gin, a fur ther nine had a nega tive cul ture in their home land but a posi tive cul ture in Can ada, and one was not tested at home but was posi tive in Can ada, for a to tal of 19 cases (5%).Of the 483 im mi grants sched uled to be seen in Win ni peg, 94 never at tended, and 138 missed some clinic vis its.
The ex pe ri ence of Mani toba and Brit ish Co lum bia shows that screen ing pro grams need to be able to lo cate and encour age co op era tion in those who do not com ply with screening af ter ar ri val, and to fol low refu gees for at least five years.
The de tec tion of tu ber cu lo sis in im mi grant chil dren is es -pe cially im por tant, be cause their rates are higher than those in the native-born, they are likely to be as ymp to matic, and the high ini tial mo bil ity of im mi grant chil dren re sults in a large number of po ten tial con tacts with the at ten dant costs of follow ing up en tire schools.Child hood ex po sure to an ac tive case gives a life time risk of tu ber cu lo sis re ac ti va tion of 5 to 10% (34).In a study of two schools in Mont real, 9.4% of foreign-born chil dren, but only 0.4% of Canadian-born children, had a posi tive Man toux test (35).In Brit ish Co lum bia dur ing 1979 to 1988 there were 233 cases of tu ber cu lo sis in chil dren up to the age of 15 (36).Only 40% of the 197 cases of pri mary pul mo nary tu ber cu lo sis pre sented with symp toms (mainly cough and fe ver) and only 48 were di ag nosed because they pre sented with symp toms.The re main ing 154 were di ag nosed as a re sult of trac ing (mostly adult) con tacts.Bac te rio logi cal stud ies were ob tained in 83%, and were positive in 22%.Chest x-rays were avail able for 94%, and lym phade no pa thy was seen in 94% of these x-rays.The slow ness of re sponse to ther apy and the dili gence re quired in follow-up are dem on strated by the 125 pa tients with follow-up films: lym pha de no pa thy and con soli da tion wors ened in 33% despite three months of ther apy.Af ter six months of ther apy only two had clear chest x-rays, most cleared by two years, but 7% still had ab nor mal chest x-rays three years later.
The ma jor ity of im mi grants set tle in the 10 ma jor cit ies, and health care work ers in these cit ies need to be sen si tized to the num bers ex pected to be in fected with tu ber cu lo sis and the need for screen ing.
Within the first year each case of ac tive tu ber cu lo sis is es timated to trans mit tu ber cu lo sis to 2 to 4% of their house hold con tacts (37), but 37% of HIV-pos itive pa tients will de velop tuber cu lo sis within five months of ex po sure in their house hold (38)(39)(40).Each case of un treated ac tive tu ber cu lo sis over a life time is es ti mated to in fect be tween seven and 77 other indi vidu als.

TA BLE 2 Screen ing of im mi grants to Can ada for tu ber cu lo sis
Sec tion 11 of the Im mi gra tion Act* re quires that every im mi grant shall re ceive: • a com plete his tory and physi cal ex ami na tion • a uri naly sis • a VDRL (for those 15 years and over) • a 36x43 cm chest x-ray (for those over 11 years).If the chest x-ray is ab nor mal, the per son will be de clared in ac tive only • if tu ber cu lo sis has been ade quately treated or • there are two sta ble chest x-rays three months apart plus three nega tive spu tum cul tures taken more than 24 h apart (or one bron chial wash ing or one gas tric lavage or one la ryn geal swab) The costs of screen ing 7573 il le gal im mi grants to Den ver for tu ber cu lo sis were es ti mated at US$209,000 (50% for person nel costs and 41% for x-rays), yield ing a cost for each patient who com pleted isonia zid ther apy of $285 (19).
When spu tum sam ples (which cost $50 to 75) were also col lected in a study of 249 im mi grants in Se at tle with find ings on two chest x-rays con sis tent with pul mo nary tu ber cu lo sis, only 5% of spu tum cul tures were found to be posi tive, and the cost per sputum-confirmed case of tu ber cu lo sis was US$1,996 to 2,994 (20).
The costs of trac ing the con tacts of school chil dren can be very high, partly be cause many stu dents may be ex posed to im mi grants who tend to be mo bile on first ar ri val in a new coun try.Af ter a 10-year-old im mi grant to Scar bor ough, Ontario, was noted to have ac tive pul mo nary tu ber cu lo sis, nine other fam ily mem bers were found to have mark edly posi tive Man toux re ac tions and chest x-ray changes com pati ble with tu ber cu lo sis, and were treated as hav ing ac tive tu ber cu lo sis.In the two schools in volved, 722 chil dren were screened with Man toux tests: on first test ing 7.1% (des ig nated as 5 mm or greater if the child is in a class with an in dex case, or 10 mm or greater if not in such a class) and an ad di tional 4.4% on second test ing two to three months later had a posi tive Man toux (35).

DRUG RE SIS TANCE IN IM MI GRANTS AND REFU GEES
Drug re sis tance can be sys tem ati cally re duced if five require ments of ap pro pri ate ther apy are achieved: pre scrib ing an ade quate regi men for the ini tial epi sode; test ing for drug sen si tivi ties; en sur ing pa tient com pli ance; rec og niz ing treatment fail ure (due mainly to re sis tance to isonia zid); and prescrib ing to over come treat ment fail ure (41).
Al though HIV-pos itive pa tients are 20 times more likely to have other HIV-pos itive per sons in their house holds, and HIV-pos itive pa tients have an 8% an nual ac qui si tion of tu bercu lo sis, there does not yet ap pear to be an ex cess of drug resis tance in such pa tients stud ied in Haiti, Za ire or the United States (42).
Short-course thera pies of six months have be come an impor tant form of treat ment.It is there fore of ma jor con cern that these thera pies can have high rates of fail ure if of fered to patients in fected with My co bac te rium tu ber cu lo sis re sis tant to isonia zid and to ri fampin (43).For ex am ple, only half of the pa tients in Den ver with pul mo nary tu ber cu lo sis re sis tant to both isonia zid and ri fampin even tu ally had nega tive spu tum cul tures, even af ter pro longed in di vidu ally tai lored DOT with mul tidrug regi mens, and 46% of those with treat ment fail ure or re lapse died (44).
In Mani toba from 1980 to 1989, 1478 cases of ac tive tu bercu lo sis were di ag nosed, speci mens were cul tured for 1392, and 78% of these were posi tive (45).Seven per cent of these culture-positive cases were drug re sis tant.For im mi grants in whom tu ber cu lo sis de vel oped in their first year in Can ada, the odds ra tio of be ing re sis tant was 9.9 com pared with other Cana di ans with tu ber cu lo sis.Ninety-one per cent of the drug resis tance was to first line drugs (isonia zid and strep to my cin).
Treaty In di ans in Mani toba, how ever, had the same risk of resis tance as the other pa tients born in Can ada.In Mani toba 15% of culture-positive im mi grants were re sis tant to one or more drugs com pared with 4% of other pa tients (45).
In New York City dur ing April 1991, 41% of pa tients with posi tive tu ber cu lo sis cul tures were HIV-pos itive, and prior an titu ber cu lo sis ther apy was the strong est fac tor pre dict ing drug re sis tance to one or more drugs (odds ra tio 2.7) or to isonia zid plus ri fampin (odds ra tio 5.3) (46).In pa tients who had not pre vi ously been treated, from 1955 to 1991 re sis tance to isonia zid in creased from 3% to 15% and to strep to my cin from 1% to 9%; from 1978 to 1991 re sis tance to ri fampin in creased from 1% to 9% (46).By 1991, 25% of the pa tients could transmit isoniazid-resistant or gan isms, and 20% could trans mit organ isms re sis tant to both isonia zid and ri fampin.In a sur vey dur ing 1991-92 of pa tients re ported to the New York City Depart ment of Health as hav ing tu ber cu lo sis, 14% of both im migrant and American-born pa tients were re ported as show ing com bined re sis tance to isonia zid and ri fampin (47).In Santa Clara, Cali for nia, 41% of im mi grants were re sis tant to one or more drugs (48).
In Is rael 37% of im mi grants from Rus sia, 16% from Ethiopia and 12% of other pa tients were re sis tant to at least one drug (49).
There is con sid er able re sis tance to an ti tu ber cu lous drugs in coun tries from which Can ada draws im mi grants (Ta ble 3).Among un treated pa tients there was re sis tance to at least one drug in 12% in Ma lawi (50), 15% in Ko rea (51), 17% in a sample of 10 Latin Ameri can coun tries (52), 20% in both Gu jarat, In dia (53) and Haiti (54), 27% in Ghana (55) and 35% in Turkey (56).Among treated pa tients 32% were re sis tant to at least one drug in La hore, Paki stan (57), 41% in Haiti (54) and 47% in Ko rea (51), and 56% were re sis tant to isonia zid in Gujarat (53).In a small study of gene mu ta tions in M tu ber cu lo sis at the Pitié-Salpétrière Hos pi tal in Paris, of the 11 un treated non-European im mi grants 55% were re sis tant to ri fampin, 64% to strep to my cin and 100% to isonia zid, all by known gene muta tions (58), but at Mount Si nai Hos pi tal in New York City immi grants from the Third World were less likely to be multiple-drug re sis tant than other pa tients (59).In Rio de Janeiro 8% of the con tacts of multiple-drug re sis tant pa tients con tracted tu ber cu lo sis within a two-year pe riod, with 50% ac quir ing tu ber cu lo sis with the same mul ti ple re sis tance pattern, 10% a dif fer ent mul ti ple-drug re sis tant pat tern, and 40% with sus cep ti ble or gan isms, show ing that tu ber cu lo sis can be ac quired from mul ti ple sources, even in those liv ing with an index case (60).
Par ticu larly trou bling is the find ing that re sis tance in some de vel op ing coun tries is higher among young peo ple (who are more likely to mi grate).In Ko rea ini tial drug re sis tance was found in 32% of pa tients with tu ber cu lo sis un der 30 years of age and 14% in those over 59 (51).Simi larly in the United States, 22% of chil dren un der 15 and 9% of pa tients over 65 with tu ber cu lo sis were re sis tant (61).
From the 1960s to the 1980s rates of re sis tance world wide were high est in those Asian coun tries with in ter me di ate devel op ment lev els suf fi cient to per mit medi ca tions to be read ily avail able but with out health care sys tems well fi nanced enough to en sure regu larly su per vised ad mini stra tion (62).How ever, a poor coun try such as Tan za nia, through an ex cellent na tional pro gram, has main tained ini tial ri fampin re sistance be low 1% for the past 15 years, de spite the HIV epi demic; Ko rea and Al ge ria also have well-run na tional control pro grams (61).

REC OM MEN DA TIONS FOR PRO PHY LAXIS AND THER APY, AND COM PLI CA TIONS OF THER APY
The ef fec tive ness of isonia zid pro phy laxis has been confirmed by ran dom ized con trolled tri als, with maxi mum bene fit oc cur ring af ter 12 months, al though com pli ance is bet ter with six months' pro phy laxis (6).Pa tients with only an ab nor mal tuber cu lin test should be given pro phy laxis for six months, those with a sta ble ab nor mal ra dio graph con sis tent with previ ous tu ber cu lo sis for 12 months, and those with HIV for 24 months (63).Nine months' pro phy laxis is rec om mended for chil dren (64).
Af ter re view ing the world lit era ture on drug re sis tance from the 1960s on wards, Ise man and Sbar baro (62) con cluded that, for a pa tient with out ap par ent risk fac tors for re sis tance, a three-drug regi men of isonia zid, ri fampin and eth am bu tol is in di cated, and for pa tients from an area with a high prevalence of re sis tance, a four-drug regi men of isonia zid, ri fampin, pyrazi na mide and eth am bu tol is in di cated.For pa tients with a his tory of prior che mo ther apy or who are epi de mi ol ogi cally at risk for re sis tance, a five-drug regi men in clud ing three drugs to which the pa tient had not pre vi ously been ex posed should be used.
The Cen ters for Dis ease Con trol and Pre ven tion in Atlanta, Geor gia (65), based on ex pe ri ence with drug re sis tance in the United States (Ta bles 4,5), rec om mends a four-drug regi men (isonia zid, ri fampin, pyrazi na mide, and strep to my cin or eth am bu tol) for the first two months, un til drug sus cep ti bilities are known.This would en sure that 95% of pa tients would be treated with a regi men that in cludes two drugs to which they are sen si tive.Spu tum con ver sion is also ac com plished faster than with a three-drug regi men.A pa tient who de faults on the four-drug regi men is also more likely to be cured and not re lapse when treated for the same du ra tion as a pa tient re ceiv ing three-drug ther apy.
Care ful dis cus sion of po ten tial side ef fects and the benefits of ther apy is es sen tial to avoid dis con tinu ance of ther apy.

THE EF FEC TIVE NESS OF AND IN DI CA TIONS FOR DOT
Stud ies us ing DOT with a patient-centred ap proach, training of the health team to be sen si tive to pa tients' needs, so cial serv ice as sis tance and thor ough case man age ment of therapy have been shown to be 100% ef fec tive in treat ing tu bercu lo sis, and all other meth ods are in ef fec tive com pared with this gold stan dard (66).
A study in Cali for nia that did not use DOT did use sev eral patient-centred ap proaches: the edu ca tors spoke Ko rean and Span ish; the tak ing of medi ca tions was linked with other activi ties; and ways to re mem ber medi ca tions and fu ture appoint ments and test re sults were re corded in a book let (67).At each visit ques tions were an swered, mis un der stand ings were ex plained, and tai lored edu ca tion mes sages based on the ini tial in ter view and sub se quent as sess ments, edu cational re in force ment about tu ber cu lo sis, en list ment of sup port from fam ily and friends, and con tract ing were used.Among pa tients with ac tive tu ber cu lo sis ad her ence to the medi ca tion regi men im proved from 57% with usual care to 71%, and for pa tients re ceiv ing pro phy lac tic treat ment ad her ence improved from 14% with usual care to 42%, but these re sults are not as good as in stud ies that used DOT.The Ko rean Na tional Tu ber cu lo sis Pro gram in 1989 without us ing DOT achieved 96% ef fi cacy (con ver sion in those com plet ing isonia zid/ri fampin/eth am bu tol) and 80% ef ficiency (con ver sion in those start ing isonia zid/ ri fampin/etham bu tol) through ex cel lent or gani za tion, care ful moni tor ing of re sis tance and shorter treat ment regi mens (51).

TA BLE 3 Per cent age re sis tance to an ti tu ber cu lous drugs in se lected coun tries of emi gra tion to Can ada
A study of il le gal ali ens in Den ver who had been screened for tu ber cu lo sis linked ad her ence to medi ca tions with sub sequent ac qui si tion of le gal im mi grant status, and ob tained 70% ad her ence to ther apy (19).
Stud ies that used small fi nan cial in cen tives (68) or telephone re mind ers with per sonal con tact af ter a failed ap pointment showed mini mal ef fects (69).
The ef fects of in tro duc ing DOT were as sessed in a ret rospec tive re view of all pa tients with posi tive M tu ber cu lo sis cultures in Tar rant County Health De part ment, Texas (70).A pe riod of six years of un su per vised ther apy was fol lowed by a pe riod of six years of DOT with isonia zid, ri fampin and pyrazina mide (plus eth am bu tol or strep to my cin or ca preo my cin un til iso lates were con firmed to be fully sen si tive to the first three drugs).There was a dra matic re duc tion in pri mary re sis tance from 13.0% to 0%, in ac quired re sis tance from 10.3% to 0%, in mul tidrug re sis tance from 6.1% to 0% and in re lapse from 20.9% to 0%.Dur ing DOT, medi ca tions were ad min is tered under ob ser va tion Mon day through Fri day for the first two to four weeks of ther apy, and then twice weekly for the next six months.About 30% of the pa tients in both treat ment pe ri ods had a his tory of al co hol abuse.
In Den ver in an other popu la tion with a high (57%) in cidence of al co hol abuse, DOT was given with isonia zid, rifampin, pyrazi na mide and strep to my cin given daily for two weeks, then in high doses twice weekly for an other six weeks fol lowed by isonia zid and ri fampin twice weekly for an other 18 weeks (71).Thir teen of the 160 pa tients were lost to follow-up (mostly be cause they moved out of Den ver) and four died.In the re main der there were no treat ment fail ures and only two re lapses (both of them al co hol ics who be came culturenegative af ter four weeks of ther apy but then missed some ap point ments).In pa tients with pul mo nary tu ber cu lo sis, sputum cul tures be came nega tive af ter a me dian of 4.6 weeks, and 100% by 20 weeks.
Stud ies of in ter mit tent DOT (thrice-weekly ther apy for the en tire treat ment pe riod) in Hong Kong (72), (thrice-weekly ther apy for three months fol lowed by twice-weekly ther apy for three months) in Ma dras (73), and in Ro ma nia (74) are as effec tive as pro grams that be gin with daily ob served ther apy for the first two to four weeks.
In con trast, in a self-administered regi men planned to last six months, 39% of the pa tients dropped out of the study, and in a nine-month regi men 49% dropped out (75).The cost of the six-month regi men was US$327 per pa tient, whereas the Den ver DOT regi men cost US$397 per pa tient but could be reduced to US$283 if two chest x-rays and two cul tures were deleted, by re ly ing on the greater ef fi cacy of DOT (76).In New York City only 11% of the tu ber cu lo sis pa tients dis charged from one hos pi tal fol lowed up in clinic (77), but ther apy comple tion rates of up to 98% are ex pe ri enced when DOT is used (78).
Non com pli ance can be ex pected to oc cur in pa tients with psy chi at ric ill ness, al co hol ism, drug ad dic tion or who are home less (66), but can not be pre dicted on the ba sis of age, sex, so cio eco nomic status or es ti mates by health per son nel (79).In ter views with in tra ve nous drug us ers in Brook lyn, New York showed that DOT is ap pro pri ate be cause many of them had lit tle knowl edge about tu ber cu lo sis or their own tu ber culo sis status, and were afraid of de ten tion for tu ber cu lo sis and there fore stopped treat ment (80).Com ple tion rates for ther apy in one study of home less pa tients in the United States improved from 44% to 89% when the pro vi sion of shel ter and food vouch ers was added to DOT (81).(There has been no research on re ac ti va tion rates if free hous ing is dis con tin ued once the pa tient has com pleted treat ment.) The four es sen tial as pects of DOT are that ther apy should be highly patient-centred (the pa tient should ex pe ri ence no bar ri ers to re ceiv ing ther apy), re sis tance should be care fully moni tored and treated, a moni tor ing and in for ma tion sys tem should be me ticu lously main tained, and ther apy can not be dele gated by the DOT worker to a fam ily mem ber be cause that per son may con nive with the pa tient or be un der pres sure from the pa tient not to com plete the ther apy (eg, a child repeat edly spits out the tu ber cu lo sis medi ca tions) (82).
Op tions for con trol other than DOT are treat ment of all known ac tive tu ber cu lo sis cases in the home coun try for six months be fore im mi gra tion (the cur rent Aus tra lian pol icy); a pay ment on com ple tion of ther apy to new ar ri vals who need pro phy laxis (on the model of the $1,000 pay ment in France for com ple tion of an te na tal care); or re quir ing new ar ri vals to post a bond, which would be re paid plus in ter est on com pletion of ther apy, but these are less likely to en sure com ple tion of ther apy.

CON CLU SIONS
Ex po sure to tu ber cu lo sis is wide spread among im migrants to Can ada.Ac tive tu ber cu lo sis is most likely to develop in im mi grants from coun tries with a high preva lence and dur ing the first five years in Can ada, but risks re main higher over the fol low ing 40 years than they do for the popu lation born in Can ada.
Can ada is not op ti mally or gan ized to de tect and fol low patients who are at risk for de vel op ing tu ber cu lo sis be cause Can ada has no na tional ref er ence cen tre to track move ment be tween prov inces, and there is sepa ra tion of the re spon si bility for sur veil lance (Labo ra tory Cen tre for Dis ease Con trol) from that for treat ment (phy si cians and pub lic health de partments).
DOT should be in sti tuted for each ac tive case of tu ber cu losis, and the name of each pa tient con tinu ing or com plet ing ther apy should be cir cu lated at the be gin ning of each month to treat ing phy si cians and pub lic health de part ments, per mitting prompt re in sti tu tion of DOT by pub lic health de part ments in those pa tients who dis con tinue ther apy (83).

TA BLE 5 Dos age rec om men da tions for the ini tial treat ment of tu ber cu lo sis in chil dren and adults
THO MAS AND GUSHU LAK 250 CAN J INFECT DIS VOL 6 NO 5 SEP TEM BER/OCTO BER 1995

la tions ap proved by the Ca na dian Tho racic So ci ety, the Ca na dian Lung As so cia tion, the Tu ber cu lo sis Di rec tors of Can ada and Health Can ada †
Newly ar rived in di vidu als with a pre vi ous his tory of tu ber cu lo sis or an ab nor mal chest x-ray sus pi cious of in ac tive tu ber cu lo sis ‡ should give:• a com plete his tory and physi cal ex ami na tion• a chest x-ray• three spu tum (or gas tric lavage) speci mens for smear and cul ture for My co bac te rium tu ber cu lo sis• a tu ber cu lin test or • two sta ble chest x-rays more than six months apart The Regu *Ref er ence 84; † Ref er ence 85; ‡ In 1992 12,000 im mi grants had such an x-ray.VDRL Ve ne real Dis ease Re search Labo ra tory test

TA BLE 4 Ini tial treat ment regi men op tions for chil dren and adults
Isonia zid, ri fampin, pyrazi na mide, eth am bu tol or strep to my cin daily for eight weeks.Stop eth am bu tol or strep to my cin when sensi tiv ity to isonia zid and ri fampin is dem on strated Then isonia zid plus ri fampin daily (or two to three per week by DOT) for 16 weeks in ar eas where isonia zid re sis tance not documented ≤ 4%To tal treat ment du ra tion should be six months AND three months be yond cul ture con ver sion to nega tive* Op tions 1, 2 or 3 can be used, but should be con tin ued for nine months AND at least six months be yond cul ture con ver son to nega tive Preg nancy: Nei ther strep to my cin (may cause con geni tal deaf ness) nor pyrazi na mide (ab sence of tera to gen ic ity has not been dem on strated) should be used.The regi men is for nine months, and un til drug sus cep ti bili ties are known, the pre ferred ini tial regi men is isonia zid, ri fampin and eth am bu tol Lac ta tion: Isonia zid, ri fampin and eth am bu tol are safe be cause the con cen tra tions passed into breast milk are very low.Con versely, breast feed ing while tak ing these drugs con fers no pro tec tion on the baby

If the pa tient is clini cally, smear-or culture-positive af ter three months
65a tu ber cu lo sis medi cal ex pert should be con sulted DOT Di rectly ob served ther apy; HIV Hu man im mu no de fi ciency vi rus.*TheAmeri can Acad emy of Pe di at rics (ref er ence 64) rec om mends op tion 1, and in the case of bone, joint, mili ary or men in gitic tu ber cu lo sis, 10 months of isonia zid and ri fampin; † The strong est evi dence from clini cal tri als is the ef fec tive ness of all four drugs admin is tered for the full six months.There is weaker evi dence that strep to my cin can be dis con tin ued af ter four months if the iso late is sus cep ti ble to all drugs.The evidence for stop ping pyrazi na mide be fore the end of six months is equivo cal for the thrice/week regi men, and there is no evi dence of the ef fec tive ness of this regi men with eth am bu tol for less than the full six months.Data from ref er ence65 65he Ca na dian Tho racic So ci ety 1994, Ta ble 2 (ref er ence 6) has the fol low ing dif fer ences: (i) the up per daily limit for strep to my cin for chil dren is 40 mg/kg; (ii) a 'usual daily dose' of isonia zid 300 mg, ri fampin 600 mg, pyrazi na mide 1500 mg, eth am bu tol 800 to 1200 mg and strep to my cin 1000 mg is speci fied, rather than the 'max imum dose' of the Ameri can author ity; (iii) the doses for 'twice -weekly ther apy' for isonia zid are 900 to 1200 mg, pyrazi na mide 2500 mg, eth am bu tol 2400 mg and strep to mycin 1000 mg.The Ca na dian Pe di at ric So ci ety, Ta ble 1 (ref er ence 87) has the fol low ing dif fer ences: (i) the up per daily dose for isonia zid is 15 mg/kg and for strep to mycin is 40 mg/kg, and the range for pyrazi na mide is 20 to40mg/kg; (ii) the dos age for strep to my cin for twice-weekly ther apy is 20 to 40 mg/kg.The Ameri can Acad emy of Pe di at rics (ref er ence 64) has the fol low ing dif fer ences: (i) the up per daily dose for isonia zid is 15 mg/kg; (ii) the daily dos age for pyrazi na mide is 20 to 40 mg/kg; (iii) theup per daily dos age for strep to my cin is 40 mg/kg.†Chil dren 12 years old or younger; ‡ Eth am bu tol is not rec om mended for chil dren whose vis ual acu ity can not be measured (usu ally six years old or younger), but should be con sid ered for chil dren with or gan isms re sis tant to all other drugs, when sus cep ti bil ity to eth am bu tol has been dem on strated or is likely.Data from ref er ence65