Tuberculosis Control: the Relevance of Classic Principles ...sjryan/PPP/readings/Kaye and Frieden... · SPD, and TB Prevention, Centers for Disease Control and Preven-tion, Atlanta,

Epidemiologic ReviewsCopyright © 1996 by The Johns Hopkins University School of Hygiene and Public HealthAll rights reserved

Vol. 18, No. 1Printed in U.S.A.

Tuberculosis Control: the Relevance of Classic Principles in an Era ofAcquired Immunodeficiency Syndrome and Multidrug Resistance

Katherine Kaye1 and Thomas R. Frieden12

INTRODUCTION

The estimated 64,000 excess cases of tuberculosiswhich occurred in the United States between 1985 and1993 (1) are part of a global pandemic (2). Yearsbefore industrialized countries started to experience aresurgence in tuberculosis, the disease had persisted inneglected groups within their own borders and indeveloping nations (3). With outbreaks of multidrug-resistant tuberculosis in the United States, the threat of"casual transmission of a highly fatal disease" (4, p.648) led to a renaissance of interest in, and expansionof funding for, tuberculosis epidemiology, studies onthe emergence of multidrug-resistant strains, and con-trol programs in the United States. Global tuberculosiscontrol programs, however, remain woefully under-funded (5) and are often poorly structured.

Even in the age of multidrug resistance and theacquired immunodeficiency syndrome (AIDS), theclassic principles of tuberculosis control are as rele-vant as ever. These principles include diagnosis andappropriate and complete treatment of persons withactive tuberculosis, and, after treatment is effective,contact investigation and preventive therapy as re-sources permit. Indeed, both the tuberculosis epidemicand the emergence of drug-resistant strains of Myco-bacterium tuberculosis have been propelled, in part,by neglect of these basic principles by medical andpublic health establishments.

In this review, we summarize the findings of recentepidemiologic studies on the resurgence of tuberculo-sis and the emergence of multidrug-resistant strains,

Received for publication November 7, 1995, and accepted forpublication April 22, 1996.

Abbreviations: AIDS, acquired immunodeficiency syndrome;BCG, Bacille Calmette Guerin; CDC, Centers for Disease Controland Prevention; HIV, human immunodeficiency virus; RFLP, restric-tion fragment length polymorphism.

'Bureau of Tuberculosis Control, New York City Department ofHealth, New York, NY.

2Division of Tuberculosis Elimination, National Center for HIV,SPD, and TB Prevention, Centers for Disease Control and Preven-tion, Atlanta, GA.

Reprint requests to Dr.Katherine Kaye, Bureau of TuberculosisControl, New York City Department of Health, 125 Worth Street,New York, NY 10013.

and the implications of these findings for tuberculosisprevention and control. In the last section of thisreview, we describe how New York City's Bureau ofTuberculosis Control analyzed and responded to thecity's epidemic, and review critical gaps in our knowl-edge and ability to control tuberculosis.

DEFINITIONS

A verified case of tuberculosis is, for the purposes ofpublic health surveillance and prevention, defined by1) the isolation of M. tuberculosis in a culture grownfrom a clinical specimen, 2) the finding of acid-fastbacilli in clinical specimens for which a culture has notbeen or cannot be obtained, and/or 3) a clinical pre-sentation consistent with tuberculosis (e.g., a positivetuberculin skin test, symptoms compatible with tuber-culosis, and an abnormal or unstable chest radiograph)in a patient who is started on two or more antituber-culosis drugs (6, 7).

Multidrug-resistant tuberculosis is usually definedas disease caused by M. tuberculosis organisms whichare resistant to at least isoniazid and rifampin, the twomost effective antituberculosis medications. Drug re-sistance may be primary, in a patient who has neverbefore been treated for tuberculosis and was infectedby drug-resistant bacilli, or secondary (acquired), in apatient who initially had drug-susceptible M. tubercu-losis but received inappropriate or inadequate treat-ment, was nonadherent to a prescribed regimen, orabsorbed prescribed medication poorly, and thus de-veloped drug-resistant tuberculosis organisms (8).

NATURAL HISTORY

Infection

Generally, tuberculosis infection depends on expo-sure to an active case while progression from infectionto disease depends on the immune status of the in-fected person (9). Most tuberculosis infections resultfrom inhalation of airborne droplets containing thebacillus M. tuberculosis (10); these droplets are emit-ted most efficiently by patients with laryngeal tuber-culosis, cavitary disease on chest radiographs, and

52

Tuberculosis Control 53

sputum samples which show acid-fast bacilli onsmear. M. tuberculosis is an acid-fast mycobacterium;of the four species in the M. tuberculosis complex (M.tuberculosis, Mycobacterium bovis, Mycobacteriumafricanum, and Mycobacterium microti) (11), it is theone primarily responsible for tuberculosis in humans.

An undiagnosed and untreated smear-positive pa-tient has been estimated to infect 10-14 persons peryear (12). The likelihood of infection following expo-sure to infectious droplets increases with increasingconcentration of M. tuberculosis droplet nuclei in theair and increasing length of exposure (9, 13).

The efficacy of the Bacille Calmette Guerin (BCG)vaccine in protecting against tuberculosis is variableand partial, although the vaccine does appear to offerprotection against disseminated and meningeal tuber-culosis in children (14).

Risk of disease progression

Once in the lungs, M. tuberculosis evokes a nonspe-cific inflammatory response (primary complex) in thelung tissue and in adjacent lymph nodes (15). Approx-imately 10 percent of infected persons develop diseaseat some point in their lives. Of those who developactive disease, most (50-80 percent) do so within 2years of infection (16, 17).

Young children are especially likely to progress toactive disease after infection (18). Active cases amongyoung children are sentinel health events and indicaterecent transmission within a community (9).

Characteristics of the host which affect the likeli-hood of progression from infection to active diseaseinclude infection with human immunodeficiency virus(HIV), fibrotic lung lesions, silicosis, some malignan-cies, hemophilia, immunosuppressive medical treat-ment, low body weight, diabetes, heavy smoking, andcertain genotypes (human leukocyte antigens Al l -B15 and DR2) (9).

Treatment to prevent disease progression, usuallywith isoniazid, is recommended in the United Statesfor all persons younger than 35 years of age who areinfected with tuberculosis. Persons at high risk forprogressing to active disease should receive preventivetherapy, regardless of age; those at high risk includerecent skin test converters, close contacts to activecases, and persons with any of the risk factors men-tioned above, especially HIV infection (19).

Reactivation of latent infection had been consideredthe most important mechanism in the pathogenesis ofactive tuberculosis, but some recent studies suggestthat rapid progression of recent infection may be animportant cause of active disease in San Francisco,California, and New York, New York (20, 21). Inaddition, exogenous reinfection has been the mecha-

nism responsible for two outbreaks of multidrug-resistant tuberculosis among homeless and HIV-infected patients (22, 23).

The effect of treatment on the natural history oftuberculosis

Without treatment, the case fatality rate for all formsof tuberculosis combined is 50-60 percent (12). Casefatality for tuberculous meningitis and miliary tuber-culosis, however, may approach 100 percent.

Treatment for tuberculosis requires a minimum of 6months, using a combination of isoniazid, rifampin,pyrazinamide, and either ethambutol or streptomycinfor 2-3 months, and then, if the patient's strain of M.tuberculosis is susceptible, continuing with isoniazidand rifampin for at least 4 more months (14). Thisshort-course regimen replaces earlier regimens whichrequired 12-24 months of treatment. If patients withdrug-sensitive M. tuberculosis are treated appropri-ately and are fully compliant with their medical regi-men for its entire duration, cure rates exceed 95 per-cent (12).

Development of drug resistance

Shortly after the discovery of antituberculosis drugs,it was recognized that therapy with a single agent canlead to drug resistance and treatment failure. In M.tuberculosis, resistance arises by spontaneous single-step mutation, which occurs at different frequenciesfor different medications. Previous treatment withantituberculosis drugs has been found to be the mostimportant predictor of drug resistance (24-26). Mul-tidrug resistance has drastic implications for the treat-ment of tuberculosis and for disease control programs,as patients with multidrug-resistant tuberculosis gen-erally require at least 18 months of treatment withantituberculosis drugs, which are more costly andtoxic than those used routinely, and may be morelikely to relapse even after longer periods of treatment.

SURVEILLANCE CHALLENGES

It remains impossible to measure the full impact oftuberculosis on human well-being in much of thedeveloping world because systems for surveillance ofmorbidity and mortality are inadequate (2). In somedeveloping countries, detection of most tuberculosiscases may be entirely on the basis of physical exam-ination and history; even where sputum microscopyand/or chest radiography are available, these proce-dures may be performed without adequate quality con-trol or be inaccessible to populations at highest risk.Overreliance on radiography and underutilization ofsputum smears for diagnosis and monitoring of treat-

Epidemiol Rev Vol. 18, No. 1, 1996

54 Kaye and Frieden

ment are common. In most industrialized countries, onthe other hand, cases are confirmed through bacterio-logic culture and species identification of mycobacte-ria. Susceptibility testing of microorganisms is notavailable in many developing countries, while it is aroutine procedure in several industrialized ones.

The New York City experience suggests that even indeveloped countries, the enumeration of smear- andculture-negative tuberculosis cases may be extremelylabor-intensive. In New York City, health-care provid-ers are legally mandated to report all suspected tuber-culosis cases. Most verified cases (in 1994, approxi-mately 83 percent) are confirmed by positive culturereports from laboratories (27). If suspected cases arenot confirmed by positive cultures, health departmentstaff must determine whether they may be confirmedclinically (e.g., through the demonstration of patientimprovement in symptoms and radiographs while onantituberculosis medication).

International comparisons of tuberculosis morbidityand mortality are tenuous. To better estimate the bur-den of morbidity imposed worldwide by tuberculosis,a method has been devised to estimate the annual riskof tuberculosis infection through skin test surveys ofpersons who have not been vaccinated with BCG (2,28). It has been estimated that 50 smear-positive tu-berculosis cases per 100,000 population occur for ev-ery 1 percent increase in the annual risk of tuberculosisinfection (2, 29). Limitations to this method includevariability between regions in the specificity of tuber-culin skin tests, in access to medical care, and in otherfactors that affect progression from infection to activedisease (30), as well as the labor-intensiveness andexpense of such surveys and increasing coverage withthe BCG vaccine, which complicates interpretation ofthe skin test.

ADVANCES IN MICROBIOLOGY

Tuberculosis diagnosis in industrialized countrieshas already been strengthened by rapid detection,identification, and susceptibility testing of specimensfrom highly contagious smear-positive patients (31).This approach incorporates use of broth-based mediawith radiometric or fluorometric indicators to detectM. tuberculosis, a nucleic acid probe to identify spe-cies, and radiometric susceptibility testing.

Recently developed tools for DNA amplification(e.g., polymerase chain reaction and transcription-mediated amplification) hasten identification of M.tuberculosis directly from respiratory specimens (32),rather than after growth of mycobacteria in culture.Smears and cultures will still be needed for grading ofinfectivity and susceptibility testing, respectively. The

clinical utility and cost effectiveness of DNA ampli-fication techniques have not yet been demonstrated.

Restriction fragment length polymorphism (RFLP)analysis (33) has been used to determine whethermicroorganisms cultured from patients with active tu-berculosis are genetically related. This technique hasbeen helpful in outbreak investigations (23) and hasbeen applied in population-based investigations to es-timate the proportion of cases which arise from recenttransmission of tuberculosis (20, 21). RFLP analysis,while helpful, complements, but does not replace, con-ventional epidemiologic analysis.

RECENT EPIDEMIOLOGY

National trends

In the United States, tuberculosis incidence reachedits nadir in 1985 when there were 22,201 cases (9.3 per100,000 population). Since 1953, when national datawere first collected, annual numbers of cases had, witha few minor exceptions, declined steadily. Between1974, when there was a change in the surveillancedefinition for tuberculosis, and 1985, the average an-nual decrease in cases was approximately 5 percent(9). Between 1985 and 1992, the national number ofcases increased by almost 3 percent per year to reach26,673 in 1992 (10.5 cases per 100,000 population);this upturn resulted in an estimated excess of 64,000cases, compared with the number expected based onearlier trends (1, 6). Starting in 1993, the nationalnumber of cases has declined annually, to 25,313 (9.8cases per 100,000 population) in 1993 (6), to 24,361cases (9.4 per 100,000 population) in 1994 (34), and to22,813 cases in 1995 (35).

The national increase between 1985 and 1992 wasdriven by an upsurge of tuberculosis in large cities(36). Trends in tuberculosis incidence in New YorkCity heralded the national epidemic by 6 years: Afterreaching a low of 1,307 cases in 1978, the number ofcases started to rise, to reach a peak of 3,811 cases(52.0 per 100,000 population) in 1992 (27, 36).

Along with the increase in cases between 1985 and1992, there was a shift in their age distribution. By1992, reactivation of tuberculosis among the elderlyno longer accounted for the highest percentage of totalcases; instead, most cases occurred among personsaged 25-44 years (36, 37). The increasing incidenceof primary and reactivated tuberculosis among HTV-infected adults contributed to this shift in age distri-bution, as did increasing numbers of young adult im-migrants and refugees with tuberculosis.

In the 1960s (38) and 1970s (39), levels of singleand multidrug-resistant tuberculosis were relativelylow, although variation between localities and ethnic/



racial groups was considerable. In 1976, 8.6 percent ofa national sample of M. tuberculosis cultures showedprimary drug resistance to one or more antituberculo-sis drugs, with the highest levels observed amongAsian and Hispanic patients (20.7 and 15.0 percent,respectively) (39).

Although community-based transmission ofmultidrug-resistant tuberculosis had been documentedearlier (40, 41), institutional outbreaks between 1988and 1991 focused attention on the problem of emerg-ing drug resistance. In 1989, after exposure in a resi-dential drug-treatment facility to an HIV-infected manwith smear-positive multidrug-resistant tuberculosis,22 percent of contacts had documented skin-test con-versions (42). Between 1988 and 1990, breaks in iso-lation precautions and delayed diagnosis of both tu-berculosis and multidrug resistance contributed to thenosocomial spread of multidrug-resistant tuberculosis,primarily among HIV-infected persons, in several hos-pitals in New York City and Miami, Florida (43-45).By 1992, the Centers for Disease Control and Preven-tion (CDC) had investigated outbreaks of multidrug-resistant tuberculosis in eight hospitals and a statecorrectional system. The mortality rate for the 297patients who had multidrug-resistant tuberculosis inthese outbreaks was approximately 70 percent, and themedian interval from tuberculosis diagnosis to deathranged from 4-16 weeks (43-46). More recently,however, lower mortality has been documented amongmultidrug-resistant tuberculosis patients who werepromptly diagnosed and treated with appropriatedrugs, especially if they were not HIV-infected (47,48).

In response to outbreaks of multidrug-resistant dis-ease, the CDC included as part of the National ActionPlan to Combat Multidrug-Resistant Tuberculosis (49)the recommendation that local surveillance systemscollect information on drug susceptibility patterns ofinitial and final M. tuberculosis isolates from all per-sons with culture-positive tuberculosis. The CDC re-ported that initial drug susceptibility results wereavailable for 81.7 percent of M. tuberculosis isolatesfor 1994 cases—8.0 percent of isolates were resistantto at least isoniazid, and 2.2 percent were resistant toat least isoniazid and rifampin (34).

International trends

Trends in tuberculosis incidence in industrializedcountries have varied. Case notification rates continueto decline in some western European countries (Bel-gium, Finland, Germany, and Portugal) and in Japan(where, however, the rate of decline is slowing); rateshave plateaued in other European countries (GreatBritain and Sweden) and in Australia; and rates have

increased in Denmark, Austria, Ireland, The Nether-lands, Norway, Italy, France, and Switzerland (2, 50,51).

Tuberculosis rates are higher in countries in easternEurope and the former Soviet Union than in westernindustrialized nations, and the quality of surveillancein eastern European and former Soviet bloc countriesis variable (2). While tuberculosis incidence continuesto decline in some eastern European and former Sovietbloc countries, the rate of decline has slowed in sev-eral; in Romania and in some of the central Asian andBaltic countries, tuberculosis incidence rates have in-creased (2, 52).

Tuberculosis remains a leading cause of mortalityand morbidity in developing countries. Seven percentof all deaths in developing countries are attributed totuberculosis; 95 percent of the 90 million new tuber-culosis cases which have occurred or will occur duringthe 1990s will be in developing countries (2, 32).

The impact of the HIV epidemic on tuberculosis insub-Saharan Africa became apparent during the 1980swhen, after years of steady decline, reported cases oftuberculosis increased dramatically (2, 53). The greatmajority of the estimated 305,000 excess tuberculosiscases which have occurred worldwide and which areattributable to HIV infection have thus far occurred inAfrica (31, 53). Because the largest proportion of theworld's tuberculosis-infected people live in Asia, andbecause HIV infection is increasing in several largeAsian cities, the number of people in Asia who coulddevelop tuberculosis because of dual infection maymultiply nearly sevenfold (54).

In some developing countries, low adherence ratesand availability of antituberculosis drugs without pre-scription have led to rates of acquired and primarydrug resistance which are higher than those seen indeveloped countries (2, 55). The spectre of risingmultidrug-resistant tuberculosis, and especially of rif-ampin resistance, threatens to turn back the clock ontuberculosis treatment and control by 30 years, dim-ming the prospect for international control. Fortu-nately, as will be documented later in a description ofNew York City's recent experience, effective controlprograms have demonstrated that it is possible to bothprevent and reverse increases in multidrug-resistanttuberculosis.

ECONOMIC IMPACT

The resurgence of tuberculosis is associated witheconomic costs which are staggering for both devel-oped and developing countries. In the United States, a1992 analysis estimated that the average hospital costper tuberculosis patient was approximately $25,000(56, 57); in 1990, it was estimated that inpatient care


56 Kaye and Frieden

for patients with multidrug-resistant tuberculosis ap-proached $100,000 per patient (40, 56). In New YorkCity, Arno et al. (57) found that the public sectorshouldered a large share of the $179 million spent in1990 for tuberculosis-related hospitalizations (57).

Because the burden of tuberculosis in developingcountries falls most heavily on adults during the mostproductive years of their lives, its cost in terms of theWorld Bank's latest development indicator (theDALY, or disability-adjusted life year) is enormous(31, 58). Treatment of tuberculosis in developingcountries is an extremely cost-effective health inter-vention and an appropriate investment in develop-ment. The great expense involved in curing patientswith multidrug-resistant tuberculosis means thatspread of this form of the disease in developing coun-tries puts tuberculosis programs back into the pre-antibiotic era, as ineffectively treated patients remainsources of infection in their communities.

REASONS FOR THE REEMERGENCE OFTUBERCULOSIS

Host susceptibility

HIV infection. By the year 2000, it is projected thatworldwide as many as 17,454,000 people will be in-fected with HIV, of which 1,495,000 will be in NorthAmerica (59).

Some of the earliest epidemiologic evidence sug-gesting that HIV infection had contributed to the re-surgence of tuberculosis was ecologic in nature—itwas noted that the highest incidence of tuberculosisoccurred in geographic areas and among racial/ethnicgroups with higher numbers of cases of AIDS (60, 61).For example, between 1980 and 1989, New York hadmore AIDS patients than any other city, and the inci-dence of tuberculosis increased by 68 percent (60-63). Because HIV-infected persons have a high fre-quency of extrapulmonary tuberculosis, the increase inproportion of extrapulmonary cases which occurredbetween 1984 and 1989 may have been due to increas-ing incidence of disease in dually-infected persons(61, 63). That HIV infection greatly increased the riskfor developing active tuberculosis was demonstratedmost convincingly by the finding that progression toactive tuberculosis is far higher among HIV-infectedpersons with positive skin tests than among personsinfected only with tuberculosis (60, 64, 65).

HIV infection increases the incidence of active tu-berculosis by increasing the risk of reactivation oflatent tuberculosis, of rapid progression to active dis-ease following recent infection with tuberculosis, andof exogenous reinfection (22, 32). Findings of clus-tered M. tuberculosis strains among HIV-infected per-

sons in two separate studies (20, 21) suggest that muchactive tuberculosis in HIV-infected persons may bedue to rapid progression of recent infection.

Delayed diagnosis of tuberculosis in dually-infectedpeople may be caused by atypical presentation, some-times resembling that of other opportunistic infections,on the chest radiographs of HIV-infected patients withpulmonary tuberculosis (44). Some reports haveshown, however, that delayed diagnosis was moreoften due to errors in routine management than toatypical presentation (60, 66, 67).

A HIV seroprevalence survey of all patients whoattended 20 tuberculosis clinics in cities throughoutthe United States between 1988 and 1989 revealed amedian clinic seroprevalence rate of 3.4 percent, withhighest rates (over 20 percent) in the Northeast andAtlantic coastal areas (68).

Substance abuse. Injecting drug use is a risk fac-tor for the progression of tuberculosis infection toactive disease (17), and the high prevalence of HIVinfection among injecting drug users further increasesthat risk. Buskin et al. (69) found that, after adjustmentfor age and smoking status, heavy drinkers were twiceas likely as nonhabitual drinkers to have tuberculosis.

Population migration

Migration from areas of the world with relativelyhigh tuberculosis prevalence to areas with relativelylow prevalence may be an important contributor to therising incidence of tuberculosis in industrialized coun-tries. In some of the western European countries,where tuberculosis rates have recently increased, largeproportions of cases (38 percent in Denmark, 41 per-cent in The Netherlands, and 51 percent in Switzer-land) occurred among immigrants (2, 50).

In the United States between 1977 and 1979,foreign-born persons accounted for 15 percent of tu-berculosis cases reported from 11 sites (70). McKennaet al. (3) reported that, from 1986 to 1993, the UnitedStates experienced increases in the number and pro-portion of foreign-born tuberculosis cases, from 4,925cases in 1986 (21.6 percent of the total) to 7,346 casesin 1993 (29.6 percent of the total).

California's recently approved Proposition 187 re-quires publicly funded health-care providers to denynonemergency care to undocumented immigrants andto report them to government officials. Such legisla-tion could lead to increased tuberculosis morbidity andmortality by deterring persons with active tuberculosisfrom seeking treatment (71).

Events which place large populations of dislocated,malnourished people in crowded living conditions arelikely to increase tuberculosis case rates (2). Combatin El Salvador (72) and the former Yugoslavia, and



wars in parts of Africa, have lead to forced migrationsand may facilitate the spread of tuberculosis.

Living conditions

Homelessness. Tuberculosis disproportionatelyaffects the poor and disenfranchised, and has histori-cally been associated with homelessness. In the 1980sand 1990s, the problem of homelessness in manyAmerican cities was exacerbated by the deinstitution-alization of the mentally ill (without development ofcommunity-based housing), the increasing scarcity ofjobs for unskilled persons, and drug abuse (36). Trans-mission of tuberculosis may be increased by the hous-ing of homeless persons in large congregate shelters(73). A survey of homeless men attending a shelter-based clinic in New York City showed that 42.8 per-cent of those screened were infected with M. tubercu-losis and 6 percent had active disease (74). Clinicalpresentation of patients in one outbreak of multidrug-resistant tuberculosis in a homeless shelter in Boston,Massachusetts, as well as phage-typing of their M.tuberculosis cultures, suggested that exogenous rein-fection rather than reactivation of earlier infection wasmainly responsible for the outbreak, and thus providedevidence for transmission of tuberculosis within theshelter (22). Because taking antituberculosis drugsmay be a low priority for homeless persons, and be-cause many such persons may be quite mobile, theachievement of high treatment completion rates in thispopulation requires a concerted effort on the part ofpublic and private providers (73).

Correctional settings. Inmates are at increased riskfor infection by tuberculosis and other respiratory dis-eases in poorly ventilated and crowded correctionalfacilities (36, 75), and because they are often afflictedby HIV-infection, substance abuse, and homelessness,they are also at increased risk for disease progression.Ensuring continuity of treatment for inmates with ac-tive tuberculosis, and appropriate management of theircontacts within and outside the correctional system, ishampered by frequent movement between and withinfacilities and by unexpected discharges (36).

An outbreak of multidrug-resistant disease occurredamong prisoners and a guard in the New York Statecorrectional system (36, 76). In 1989, to address theincreased risk of tuberculosis among prisoners, theCDC published recommendations for the control oftuberculosis in correctional facilities (36, 77).

Demographic trends

Increasing urbanization in developing countries.By the year 2000, 71 percent of the world's urbanpopulation will live in developing countries; if present

conditions prevail, one quarter of these city dwellerswill live in extreme poverty (78). In some countries,certain health indices (e.g., nutritional status) areworse among the urban poor than among their ruralcounterparts (79). As more people live in the crowdedand poorly ventilated dwellings common in slums ofthe developing world, rates of tuberculosis may in-crease still further. Mobility of residents and lack ofaccessible medical and public health services withinsome of these poor urban communities may make itdifficult to implement tuberculosis control programs.

Changing population age structure. Regions withrecent high rates of population growth now have in-creasing proportions of their population in age groupswith the highest likelihood of tuberculosis morbidityand mortality. Even if age-specific tuberculosis inci-dence rates remain constant, therefore, numbers oftuberculosis deaths and new cases will increase in thedeveloping world (2).

Public health programs and medical practices

The medical and public health establishments to-gether share some responsibility for the resurgence oftuberculosis, as they countenanced the dismantling oftuberculosis control programs and neglected soundinfection control practices (80).

Brudney and Dobkin (73) chronicled the series ofcuts in public health expenditures in New York Citywhich contributed to the abysmal 11 percent rate oftherapy completion by tuberculosis patients at a publichospital in 1988. Hospital tuberculosis beds wereeliminated, yet resources intended for outpatient carenever materialized; in the midst of New York City'sfiscal crisis in the 1970s, state and federal fundingfor tuberculosis control activities also decreased.Reichman (81) popularized the phrase "the U-shapedcurve of concern" to describe the interaction betweenpublic health funding and health indices—as incidenceof a targeted condition decreases, public health fund-ing diminishes to the point where the condition onceagain begins to increase. A major decline in fundingoccurred when federal project grants, specifically fortuberculosis control, were replaced by block grants tothe states, resulting in reallocation of federal moniespreviously used for tuberculosis control to other pro-grams (81).

Medical mismanagement of patients withmultidrug-resistant tuberculosis (for example, throughlate diagnosis, treatment with drugs against whichtuberculosis organisms are resistant, or failure to ad-minister medication in a program of directly observedtherapy) may cause these patients to remain infectiouslonger (24, 44).


58 Kaye and Frieden

Lapses in adherence to guidelines for preventingtuberculosis in health-care settings have caused sev-eral nosocomial outbreaks of drug-susceptible anddrug-resistant tuberculosis among patients and health-care workers (43-45). As summarized by Menzies etal. (82), factors which facilitated transmission of tu-berculosis in hospital included delayed diagnosis(which leads to late institution of isolation), poor ven-tilation with positive pressure isolation rooms, andaerosolization of bacilli through procedures such asmechanical ventilation, bronchoscopy, and autopsy.

The resurgence of tuberculosis has also been facil-itated by failure on the part of the medical and publichealth professions to make adequate and appropriateuse of a simple tool, tuberculin skin testing (83), andto follow up with preventive therapy for infected pa-tients at high risk for progression to active disease.

IMPLICATIONS FOR CONTROL

Effective treatment of active disease

The cornerstone of an effective tuberculosis controlprogram is the detection and treatment of persons withactive tuberculosis, particularly smear-positive dis-ease. The World Health Organization (84) and theCDC (14) recommend short-course (6-8 month) ther-apy, consisting of a 2-3 month multidrug initial phasefollowed by a 4 - 6 month continuation phase. Short-course regimens have proven more cost effective thanconventional regimens of 12-18 months, and patientsare far more likely to complete shorter courses oftreatment.

To help ensure that recommended regimens arecompleted, directly observed therapy has been advo-cated internationally as the standard of care (84). Thissimple intervention (i.e., having a specially trainedworker observe patients as they swallow each dose oftheir antituberculosis medicines) has been found toreduce the frequency of primary and acquired drugresistance and relapse (85). By reducing the emer-gence of resistance to rifampin, directly observed ther-apy ensures that one of the most effective weaponsin our antituberculosis armory remains useful for thenext generation of tuberculosis patients. Savingsaccrued through preventing the emergence ofmultidrug-resistant tuberculosis more than compensatefor costs of implementing directly observed therapyprograms (86). With almost 1,400 fewer tuberculosiscases, including 300 fewer multidrug-resistant cases,in 1995 than in 1992, New York City saved hospital -ization costs of more than $40 million in the past yearalone, and more than $110 million since 1992 (87).

In 1993, the World Health Organization declaredtuberculosis a global health emergency (5). They em-

phasized both the large death toll from tuberculosis,which kills more people than any other single infec-tious agent, and that tuberculosis is easily curable. Inthe years since this declaration, funding for tubercu-losis control has increased and some national pro-grams have improved their structure and management.The World Health Organization promotes a strategy ofdirectly observed therapy, short course (DOTS) (84).The "S" for short course could also represent othercritical elements of a control strategy—Smear-baseddiagnosis, with quality assurance of microscopy find-ings, Sputum examinations for follow-up to documentcure, and Systematic monitoring with Simplified reg-isters to ensure program accountability. The WorldHealth Organization estimates that in the decade of the1990s, more than 30 million people will die fromtuberculosis. Many of these deaths could be preventedusing directly observed therapy.

Screening and prevention

Adherence to, and completion of, a course of pre-ventive therapy by persons infected with tuberculosishas been documented to be highly effective in prevent-ing progression to tuberculosis disease (88). Preven-tive therapy becomes more complicated in cases ofexposure to drug-resistant cases and infection with M.tuberculosis strains that are possibly drug resistant.When the infecting strain is thought to be resistant toisoniazid but susceptible to rifampin, rifampin may beused for preventive therapy (89). If the infecting strainis thought to be resistant both to isoniazid and rif-ampin, infected contacts should be categorized accord-ing to their risk of developing active tuberculosis—those at high risk should receive preventive therapywith two drugs to which the probable infecting strainis susceptible, though the efficacy of such preventiveregimens remains to be determined; "watchful wait-ing" may be more appropriate for those who are not atsuch high risk (90).

In developing countries, resources must be priori-tized for case detection and improvement of cure ratesamong smear-positive cases. Preventive therapy, how-ever, may be appropriate for identified HIV-infectedpersons (84). In areas with high prevalence of HIVinfection, institution of effective preventive therapyprograms may prolong the productive lives of youngadults, helping to maintain family and social structure.

Infection control in health-care settings

The infection control guidelines issued in 1994 bythe CDC (90) recommend a "hierarchy of controlmeasures" to decrease transmission of tuberculosis.The most important level of the hierarchy includes



administrative measures that are designed to ensure 1)rapid identification, isolation, diagnosis, and treatmentof persons likely to have tuberculosis, 2) effectivework practices among health-care workers, 3) educa-tion of health-care workers about tuberculosis, and 4)screening of health-care workers. The second level inthe hierarchy includes engineering controls, such asventilation, air flow, air filtration, and ultraviolet ger-micidal irradiation. The final level includes the use ofpersonal respiratory protective equipment, whichshould meet standard criteria for filtration efficacy andachievement of good facial fit.

Interventions for HIV-infected persons

Measures must be taken to reduce the exposure ofHIV-infected persons to those with infectious tuber-culosis, and to improve tuberculosis prevention andtreatment for HIV-infected patients (36). Persons withinfectious tuberculosis should not be housed in con-gregate settings, and HIV-infected individuals shouldnot be housed near persons with tuberculosis (36).

Improved screening and treatment services forimmigrants and refugees

The increasing proportion of foreign-born tubercu-losis patients among cases counted in the United Stateshighlights the need for more effective internationalcontrol programs and for increased support of thoseprograms by developed nations. Improved manage-ment of persons entering the United States with activetuberculosis requires better screening of immigrationapplicants overseas, better communication of screen-ing results to public health authorities in the UnitedStates, and better evaluation by local health-care pro-viders of immigrants with suspicious chest radio-graphs but negative sputum smears (3). Location andevaluation of recent immigrants will require greatercooperation between local health departments and theUS Immigration and Naturalization Service. Reduc-tion of rates of tuberculosis infection, however, mustbe accomplished overseas.

Surveillance

Effective tuberculosis control requires that re-sources be allocated for surveillance as well as fordirect service delivery. Population-based surveillancecan improve case management and allow accuratemonitoring of trends in tuberculosis and multidrug-resistant tuberculosis. Surveillance should include in-formation on susceptibilities of M. tuberculosis, wherefeasible.

THE TUBERCULOSIS EPIDEMIC IN NEW YORKCITY: ANALYSIS AND RESPONSE

Because the upsurge in New York City of tubercu-losis cases preceded the national epidemic by 6 years,the city offered a laboratory for analysis of contribut-ing factors and for piloting interventions. We outlinebelow how some early epidemiologic observationswere translated by the health department, with guid-ance from the CDC, into new policies and program-matic interventions, and describe changes in thecourse of the tuberculosis epidemic in New York City.Then, from the perspective of health department-basedtuberculosis controllers, we suggest avenues for fur-ther epidemiologic study.

A survey of all M. tuberculosis isolates in New YorkCity in April 1991 showed that the proportion whichwere drug resistant had increased dramatically sincethe last survey 8 years earlier (24). By 1991, 19percent of tested isolates showed resistance to two ormore drugs, and 26 percent to at least isoniazid.

In response to these findings, the New York CityDepartment of Health strengthened its surveillanceoperations by mandating, in 1991, that results of anysusceptibility tests performed be reported to the healthdepartment, and incorporating into its routine func-tions active surveillance for positive cultures and sus-ceptibility results at mycobacteriology laboratories.The health department thus became a source of essen-tial information for physicians throughout the city whocould call with questions about prior treatment andsusceptibility results of their patients.

Simultaneously, political leaders were informedabout the impact that an epidemic of drug-resistanttuberculosis could have on the city, and on methodsfor control. It was emphasized that, unlike many otherhealth and social problems, tuberculosis was curableand the epidemic reversible.

Recognizing that the single most important steptoward controlling the tuberculosis epidemic and theemergence of multidrug resistance was treatment ofactive cases until completion of therapy, the Bureau ofTuberculosis Control emphasized directly observedtherapy and instituted cohort reviews to monitor thetherapy of every active tuberculosis case in New YorkCity. Directly observed therapy was established as thestandard of care at New York City chest clinics, andhealth-care providers outside the public system wereinformed about how to refer their patients into directlyobserved therapy programs, which were established byboth the New York City and the New York Statehealth departments. For those few patients who areunable or unwilling to adhere to prescribed treatment,a credible threat of detention was established, withpassage of regulations which balanced individual lib-


60 Kaye and Frieden

erties against protection of the public safety—a lockedtuberculosis treatment ward was created at a municipalhospital.

During the early part of the tuberculosis epidemic inNew York City, large numbers of HIV-infected per-sons were housed in commercial hotels while awaitingpermanent placement. The health department estab-lished directly observed therapy and directly observedpreventive therapy in the hotels for both HIV-infectedand HIV-negative clients; high rates of patient adher-ence were achieved, in part through the use of finan-cial incentives (36).

The impact of these interventions is reflected infigure 1. Since the peak of the recent epidemic in1992, tuberculosis cases have declined by 35.8 per-cent, from 3,811 in 1992 (52.0 cases per 100,000persons) to 2,445 in 1995 (33.4 cases per 100,000persons) (87). Multidrug-resistant tuberculosis caseshave declined by approximately 75 percent since1992, from 441 (11.6 percent of the total) in 1992 toapproximately 108 (4.4 percent of the total) in 1995(27, 87, 91-93).

As the recent tuberculosis epidemic has beenbrought under better control, there has been a sharpdrop in US-born cases, while foreign-born cases havecontinued to increase in number and in proportion oftotal cases. The number of foreign-born cases recordedin 1995, 1,010 (41 percent of the total), is 37 percentgreater than the number reported in 1993, 739 whenforeign-born cases comprised 23 percent of the total.The New York City Department of Health has estab-lished an Immigrants and Refugees Unit to improvefollow up and treatment of persons who enter theUnited States with radiologic evidence of tuberculosis,assuring complete evaluation and treatment of more

than 90 percent of such persons within weeks of theirarrival in New York City.

Although there is a rich history of tuberculosis ep-idemiology and control, large gaps remain in both ourknowledge about tuberculosis and in our ability toapply this knowledge to controlling the epidemic.

On the therapeutic side, an effective vaccine againsttuberculosis would be one of the greatest medicaldiscoveries of the century. Better antituberculosismedications would also be helpful—only a tuberculo-sis specialist would use the term "short course" for atreatment regimen lasting 6 months or more.

On the epidemiologic side, we know surprisinglylittle about where and why most tuberculosis infec-tions occur. We cannot identify those smear-positivepersons who are most likely infectious, although weknow that there is marked variability between smear-positive persons in terms of the number of contactsthey infect. We have only a poor understanding of hostcharacteristics which help determine infection, and wedo not know whether they differ from those whichdetermine progression to disease. We do not knowwhether different strains of M. tuberculosis vary intheir ability to be transmitted or in their virulence.

Although RFLP analysis has expanded our knowl-edge of the molecular epidemiology of tuberculosis, ithas also raised many questions: Why do some strainsappear to be common "background" strains? Why dosome strains have few insertion sequences? Are somestrains less genetically "plastic" than others?

Perhaps the most significant and intriguing area ofour current ignorance centers on the proportion ofpatients with tuberculosis disease arising from recenttransmission. Classically, only 10-20 percent of pa-tients were thought to have been infected or reinfected

c 4

1

o0 i i i i i i i i i i i i i i i i i i i i i i i T i i n n

FIGURE 1. Tuberculosis in New York, New York, 1960-1995. •cases.

-, cases; D patients on directly observed therapy; • multidrug-resistant



within the 2 years before disease onset (80). Recentstudies indicate that this proportion may be 30-40percent (20, 21), and in New York City at the peak ofthe epidemic, it may have exceeded 50 percent (NewYork City Department of Health, unpublished data). Itwas because so much tuberculosis disease resultedfrom recent transmission that New York City was ableto control the disease so rapidly—improving treatmentoutcomes quickly interrupted transmission of infectionand progression to active disease.

In developing countries, on the other hand, mostadults have already been infected with tuberculosis,and as a result may have some level of immunity.Thus, in high-prevalence countries, unless reinfectionis common, the proportion of tuberculosis resultingfrom recent transmission may be lower.

HIV infection adds to the complexity of this ques-tion. We believe that New York City experienced abiphasic HTV-associated epidemic. The first phase,beginning in the mid-1980s through the early 1990s,was associated with reactivation of disease in HIV-infected persons who had been infected with tubercu-losis years or decades previously. The second, muchmore explosive, stage, from 1990 to 1992, was char-acterized by microepidemics in hospitals, prisons, andcommunities, with a very rapid increase in cases andas many as three generations of cases in a year.Whether this second phase would have been lessmarked if more HIV-infected persons had prior tuber-culosis infection, and, thus, some degree of protectiveimmunity, is unknown.

The proportion of patients with disease resultingfrom recent transmission is of great practical impor-tance. If, in developing countries, only 10-20 percentof cases are from transmission in the past 2 years, then,no matter how effective the treatment programs, newcases are unlikely to decrease by more than 5-10percent per year, although prevalence may decreaserapidly by cure of cases. If, on the other hand, 30-50percent of cases are from recent transmission, as wethink was the case in New York City, tuberculosis indeveloping countries may be brought under controlmuch more quickly through intensive emphasis onidentification of cases and completion of therapy, fol-lowing the pattern seen in New York City, whereUS-born cases decreased by nearly two-thirds in 3years.

In calling for the development of new methods fortuberculosis control, Comstock (94) pointed out thatmuch of our knowledge on the natural history oftuberculosis comes from studies conducted in the earlyand middle parts of this century. New epidemiologicinvestigations are needed to answer basic questions

about risks for infection and disease progression in the1990s.

Maintenance of adequate tuberculosis control pro-grams requires that policy makers and politicians con-tinue the fight against tuberculosis. It remains to beseen whether the recent brush with a major publichealth disaster leads to any change in the "U-shapedcurve of concern" (81). As our recent epidemic beginsto recede, policy makers must remain aware that, aslong as tuberculosis persists at unacceptably high ratesin underserved groups here in the United States and indeveloping countries, the potential for a new epidemicremains.

ACKNOWLEDGMENTS

The authors thank Muriel Chariot and Fong Fong Miaofor their assistance in the preparation of this manuscript. Weare also grateful to Dr. Patricia Simone (Deputy Director,Program Services Branch, Division of Tuberculosis Elimi-nation, CDC) for her review of the manuscript.

REFERENCES

1. Expanded tuberculosis surveillance and tuberculosismorbidity—United States, 1993. MMWR Morb Mortal WklyRep 1994;43:361-6.

2. Raviglione MC, Snider DE Jr, Kochi A. Global epidemiologyof tuberculosis: morbidity and mortality of a worldwide epi-demic. JAMA 1995;273:220-6.

3. McKenna MT, McCray E, Onorato I. The epidemiology oftuberculosis among foreign-born persons in the United States,1986 to 1993. N Engl J Med 1995;332:1071-6.

4. Joseph S. Tuberculosis, again. (Editorial). Am J Public Health1993;83:647-8.

5. TB Programme. TB, a global emergency: WHO report on theTB epidemic. Geneva, Switzerland: World Health Organiza-tion, 1994. (WHO/TB/94-177.)

6. Reported tuberculosis in the United States, 1993. Atlanta, GA:Centers for Disease Control and Prevention, 1994.

7. Wharton M, Chorba TL, Vogt RL, et al. Case definitions forpublic health surveillance. MMWR Morb Mortal Wkly Rep1990;39(RR-13):l-43.

8. Gangadharam PRJ. Drug resistance in tuberculosis. In:Reichman LB, Hershfield ES, eds. Tuberculosis: a compre-hensive international approach. New York, NY: MarcelDekker, 1993:293-328.

9. Rieder HL, Cauthen GM, Comstock GW, et al. Epidemiologyof tuberculosis in the United States. Epidemiol Rev 1989;11:79-98.

10. Nardell EA. Pathogenesis of tuberculosis. In: Reichman LB,Hershfield ES, eds. Tuberculosis: a comprehensive interna-tional approach. New York, NY: Marcel Dekker, 1993:103-22.

11. Grosset JH. Bacteriology of tuberculosis. In: Reichman LB,Hershfield ES, eds. Tuberculosis: a comprehensive interna-tional approach. New York, NY: Marcel Dekker, 1993:49-74.

12. Murray CJL, Styblo K, Rouillon A. Tuberculosis in develop-ing countries: burden, intervention and cost. Bull Int UnionTuberc Lung Dis 1990;65:6-24.

13. Houk VN, Kent DC, Baker JH, et al. The Byrd Study: in-depthanalysis of a micro-outbreak of tuberculosis in a closed envi-


62 Kaye and Frieden

ronment. Arch Environ Health 1968;16:4-6.14. Treatment of tuberculosis and tuberculosis infection in adults

and children. American Thoracic Society and Centers forDisease Control and Prevention. Am J Respir Crit Care Med1994;149:1359-74.

15. Daniel TM, Ellner JJ. Immunology of tuberculosis. In:Reichman LB, Hershfield ES, eds. Tuberculosis: a compre-hensive international approach. New York, NY: MarcelDekker, 1993:75-101.

16. Murray C, Styblo K, Rouillon A. Tuberculosis. In: JamisonDT, Mosely WH, Measham AR, et el., eds. Disease controlpriorities in developing countries. New York, NY: OxfordUniversity Press (for the World Bank), 1993:233-59.

17. Comstock GW, Cauthen GM. Epidemiology of tuberculosis.In: Reichman LB, Hershfield ES, eds. Tuberculosis: a com-prehensive international approach. New York, NY: MarcelDekker, 1993:23-48.

18. Starke JR, Jacobs RF, Jereb J. Resurgence of tuberculosis inchildren. J Pediatr 1992;120:839-55.

19. The use of preventive therapy for tuberculous infection in theUnited States. Recommendations of the Advisory Committeefor Elimination of Tuberculosis. MMWR Morb Mortal WklyRep 1990;39(RR-8):9-12.

20. Small PM, Hopewell PC, Singh SP, et al. The epidemiology oftuberculosis in San Francisco: a population-based study usingconventional and molecular methods. N EngI J Med 1994;330:1703-9.

21. Alland D, Kalkut GE, Moss AR, et al. Transmission of tuber-culosis in New York City: an analysis by DNA fingerprintingand conventional epidemiologic methods. N Engl J Med 1994;330:1710-16.

22. Nardell E, Mclnnis B, Thomas B, et al. Exogenous reinfectionwith tuberculosis in a shelter for the homeless. N Engl J Med1986;315:1570-5.

23. Small PM, Shafer RW, Hopewell PO, et al. Exogenous rein-fection with multidrug-resistant Mycobacterium tuberculosisin patients with advanced HIV infection. N Engl J Med1993;328:1137-44.

24. Frieden TR, Sterling T, Pablos-Mendez A, et al. The emer-gence of drug-resistant tuberculosis in New York City. N EnglJ Med 1993;328:521-6.

25. Shennan DH. Resistance of tubercle bacilli to isoniazid, PASand streptomycin, related to history of previous treatment.Tubercle 1964;45:l-6.

26. Costello HD, Caras GJ, Snider DE Jr. Drug resistance amongpreviously treated tuberculosis patients: a brief report. AmRev Respir Dis 1980;121:313-16.

27. Tuberculosis in New York City, 1994: information summary.New York, NY: New York City Department of Health, Bureauof Tuberculosis Control, 1995.

28. Styblo K, Meijer J, Sutherland I. Tuberculosis SurveillanceResearch Unit report no. 1: the transmission of tuberclebacilli; its trend in a human population. Bull Int Union Tuberc1969;42:1-104.

29. Styblo K. The relationship between the risk of tuberculosisinfection and the risk of developing infectious tuberculosis.Bull Int Union Tuberc 1985 ;60:117-19.

30. Rieder HL. Methodological issues in the estimation of thetuberculosis problem from tuberculin surveys. Tuber LungDisease 1995;76:114-21.

31. Inderlied CB. Antimycobacterial susceptibility testing: presentpractices and future trends. Eur J Microbiol Infect Dis 1994;13:980-93.

32. Porter JDH, McAdam KPWJ. The re-emergence of tubercu-losis. Annu Rev Public Health 1994;15:303-23.

33. van Embden JDA, Cave MD, Crawford JT, et al. Strainidentification of Mycobacterium tuberculosis by DNAfingerprinting: recommendations for a standardized method-ology. J Clin Microbiol 1993 ;31:406-9.

34. Tuberculosis morbidity—United States, 1994. MMWR MorbMortal Wkly Rep 1995;44:387-9, 395.

35. Tuberculosis morbidity—United States, 1995. MMWR Morb

Mortal Wkly Rep 1996;45:365-70.36. Fujiwara PI, Frieden TR. Tuberculosis epidemiology and con-

trol in the inner city. In: Rom WN, Garay SM, eds. Tubercu-losis. Boston, MA: Little Brown, 1996:99-111.

37. Tuberculosis statistics in the United States—1992. Atlanta,GA: Centers for Disease Control and Prevention, 1994.

38. Doster B, Caras GJ, Snider DE. A continuing survey ofprimary drug resistance in tuberculosis, 1961 to 1968. AUnited States Public Health Service cooperative study. AmRev Respir Dis 1976; 113:419-25.

39. Kopanoff DE, Kilburn JO, Glassroth JL, et al. A continuingsurvey of tuberculosis primary drug resistance in the UnitedStates: March 1975 to November 1977. A United States PublicHealth Service cooperative study. Am Rev Respir Dis 1978;118:835-42.

40. Outbreak of multidrug-resistant tuberculosis—Texas, Califor-nia, and Pennsylvania. MMWR Morb Mortal Wkly Rep 1990;39:369-72.

41. Interstate outbreak of multidrug-resistant tuberculosis involv-ing children—California, Montana, Nevada, Utah. MMWRMorb Mortal Wkly Rep 1983;32:516-18.

42. Transmission of multidrug-resistant tuberculosis from anHIV-positive client in a residential substance-abuse treatmentfacility—Michigan. MMWR Morb Mortal Wkly Rep 1991;40:129-31.

43. Nosocomial transmission of multidrug-resistant tuberculosisamong HIV-infected persons—Florida and New York,1988-1991. MMWR Morb Mortal Wkly Rep 1991;40:585-91.

44. Pearson ML, Jereb JA, Frieden TR, et al. Nosocomial trans-mission of multidrug-resistant Mycobacterium tuberculosis: arisk to patients and health care workers. Ann Intern Med1992;117:191-6.

45. Edlin BR, Tokars JI, Grieco MH, et al. An outbreak ofmultidrug-resistant tuberculosis among hospitalized patientswith the acquired immunodeficiency syndrome. N Engl J Med1992;326:1514-21.

46. Snider DE Jr, Dooley SW. Nosocomial tuberculosis in theAIDS era with an emphasis on multidrug-resistant disease.Heart Lung 1993;22:365-9.

47. Telzac EE, Sepkowitz K, Alpert P, et al. Multidrug-resistanttuberculosis in patients without HIV infection. N Engl J Med1995;333:907-ll.

48. Park MM, Davis AL, Schluger NW, et al. Outcome ofMDR-TB patients, 1983-1993: prolonged survival with ap-propriate therapy. Am J Respir Crit Care Med 1996;153:317-24.

49. National action plan to combat multidrug-resistant tuberculo-sis. MMWR Morb Mortal Wkly Rep 1992;41(RR-ll):5-48.

50. Raviglione MC, Sudre P, Rieder HL, et al. Secular trends oftuberculosis in western Europe. Bull World Health Organ1993;71:297-306.

51. Hargreaves J. Tuberculosis notifications in Australia, 1992.Commun Dis Intell 1994;18:330-7.

52. Raviglione MC, Rieder HL, Styblo K, et al. Tuberculosistrends in eastern Europe and the former USSR. Tuber LungDis 1994;75:400-16.

53. Sudre P, ten Dam HG, Kochi A. Tuberculosis: a global over-view of the situation today. Bull World Health Organ 1992;70;149-59.

54. Tuberculosis, HIV on collision course in Asia. World HealthOrganization, Office of Information, press release, August 10,1994. (WHO/63).

55. Kleeborg HH, Olivier MS, eds. A world atlas of initial drugresistance. 2nd rev ed. Atlanta, GA: Centers for DiseaseControl, 1984.

56. Bloom BR, Murray CJL. Tuberculosis: commentary on areemergent killer. Science 1992;257:1055-64.

57. Arno PS, Murray CJL, Bonuck K, et al. The economic impactof tuberculosis in hospitals in New York City: a preliminaryanalysis. J Law Med Ethics 1993;21:317-23.

58. The global burden of disease, 1990. In: World development



report, 1993: investing in health. New York, NY: OxfordUniversity Press, 1993:appendix B.

59. The HIV pandemic: status and trends. In: Mann J, TarantolaDJM, Netter TW. AIDS in the world, 1992. Cambridge, MA:Harvard University Press, 1992:11-108.

60. Barnes PF, Bloch AB, Davidson PT, et al. Tuberculosis inpatients with human immunodeficiency virus infection.N Engl J Med 1991,324:1644-50.

61. Cantwell MF, Snider DE, Cauthen GM, et al. Epidemiology oftuberculosis in the United States, 1985 through 1992. JAMA1994;272:535-9.

62. Tuberculosis and acquired immunodeficiency syndrome—New York City. MMWR Morb Mortal Wkly Rep 1987;36:785-90, 795.

63. Chaisson RE, Schecter GF, Theuer CP, et al. Tuberculosis inpatients with the acquired immunodeficiency syndrome: clin-ical features, response to therapy, and survival. Am RevRespirDis 1987;136:570-4.

64. Selwyn PA, Sckell BM, Alcabes P, et al. High risk of activetuberculosis in HIV-infected drug users with cutaneous an-ergy. JAMA 1992;268:504-9.

65. Selwyn PA, Hartel D, Lewis VA, et al. A prospective study ofthe risk of tuberculosis among intravenous drug users withhuman immunodeficiency virus infection. N Engl J Med 1989;320:545-50.

66. Kramer F, Modilevsky T, Waliany AR, et al. Delayed diag-nosis of tuberculosis in patients with human immunodefi-ciency virus infection. Am J Med 1990;89:451-6.

67. Flora GS, Modilevsky T, Antoniskis D, et al. Undiagnosedtuberculosis in patients with human immunodeficiency virusinfection. Chest 1990;98:1056-9.

68. Onorato IM, McCray E. Prevalence of human immunodefi-ciency virus infection among patients attending tuberculosisclinics in the United States. J Infect Dis 1992;165:87-92.

69. Buskin SE, Gale J, Weiss N, et al. Tuberculosis risk factors inadults in King County, Washington, 1988 through 1990. AmJ Public Health 1994;84:1750-6.

70. Powell KE, Meador MP, Farer LS. Foreign-born persons withtuberculosis in the United States. Am J Public Health 1981;71:1223-7.

71. Ziv TA, Lo B. Denial of care to illegal immigrants: Proposi-tion 187 in California. N Engl J Med 1995;332:1095-8.

72. Barr RG, Menzies R. The effect of war on tuberculosis: resultsof a tuberculin survey among displaced persons in El Salvadorand a review of the literature. Tuber Lung Dis 1994;75:251-9.

73. Brudney K, Dobkin J. Resurgent tuberculosis in New YorkCity: human immunodeficiency virus, homelessness, and thedecline of tuberculosis control programs. Am Rev Respir Dis1991;144:745-9.

74. McAdam JM, Brickner PW, Scharer LL, et al. The spectrumof tuberculosis in a New York City men's shelter clinic(1982-1988). Chest 1990;97:798-805.

75. Hoge CW, Reichler MR, Dominiquez EA, et al. An epidemicof pneumococcal disease in an overcrowded, inadequately

ventilated jail. N Engl J Med 1994;331:643-8.76. Transmission of multidrug-resistant tuberculosis among im-

munocompromised persons in a correctional system—NewYork, 1991. MMWR Morb Mortal Wkly Rep 1992;41:507-9.

77. Prevention and control of tuberculosis in correctionalinstitutions: recommendations of the Advisory Committee forthe Elimination of Tuberculosis. MMWR Morb Mortal WklyRep 1989;38:313-20, 325.

78. Harpham T, Stephens C. Urbanization and health in develop-ing countries. World Health Stat Q 1991;44:62-9.

79. Bradley DJ, Stephens C, Harpham T, et al. A review ofenvironmental health impacts in developing country cities.Washington, DC: The World Bank, 1992.

80. Hamburg MA, Frieden TR. Tuberculosis transmission in the1990s. (Editorial). N Engl J Med 1994;330:1750-l.

81. Reichman LB. The U-shaped curve of concern. (Editorial).Am Rev Respir Dis 1991; 144:741-2.

82. Menzies D, Fanning A, Yuan L, et al. Tuberculosis amonghealth care workers. N Engl J Med 1995,332:92-8.

83. McAnulty JM, Fleming DW, Hawley MA, et al. Missedopportunities for tuberculosis prevention. Arch Intern Med1995;155:713-16.

84. TB Programme. DOTS stops TB at the source: WHO reporton the tuberculosis epidemic, 1995. Geneva, Switzerland:World Health Organization, 1995. (WHO/TB/95.183).

85. Weis SE, Slocum PC, Blais FX, et al. The effect of directlyobserved therapy on the rates of drug resistance and relapse intuberculosis. N Engl J Med 1994;330:1179-84.

86. Iseman MD, Cohn DL, Sbarbaro JA. Directly observed treat-ment of tuberculosis: we can't afford not to try it. N EnglJ Med 1993;328:376-8.

87. New York City's tuberculosis cases decline but still highest innation. New York, NY: New York City Department of Health,press release, March 22, 1996.

88. Comstock GW, Woolpert SF. Preventive treatment of un-treated, nonactive tuberculosis in an Eskimo population. ArchEnviron Health 1972;25:333-7.

89. Management of persons exposed to multidrug-resistant tuber-culosis. MMWR Morb Mortal Wkly Rep 1992;41(RR-11):61-71.

90. Guidelines for preventing the transmission of Mycobacteriumtuberculosis in health-care facilities, 1994. MMWR MorbMortal Wkly Rep 1994;43(RR-13): 1-132.



93. Tuberculosis in New York City, 1995: information summary.New York, NY: New York City Department of Health, Bureauof Tuberculous Control, 1996.

94. Comstock GW. Tuberculosis: is the past once again prologue?Am J Public Health 1994;84:1729-31.


Documents

Tuberculosis Control: the Relevance of Classic Principles ...sjryan/PPP/readings/Kaye and Frieden... · SPD, and TB Prevention, Centers for Disease Control and Preven-tion, Atlanta,