Because of the growing frequency of exposure to patients with active tuberculosis in hospitals, the Centers for Disease Control and Prevention now requires, among other measures, that hospitals become serious about periodic tuberculin testing of employees [3]. Most new tuberculous infections of health care workers should be detected by skin test conversion in the required annual retesting of nonreactors. In otherwise healthy persons, the infection can then be cured with 93% to 98% certainty by administration of isoniazid for 6 to 12 months before tuberculosis has developed [4-6].

After a period during which the importance of annual screening of hospital employees was debated [7, 8], the concern over occupational exposure of health care workers is now intensifying as the incidence of tuberculosis increases [9-13]. This is occurring at a time when, because of good public health work in recent decades, the prevalence of tuberculin reactivity, and thus some acquired immunity from an earlier infection, is at its lowest point ever in the United States.

Inevitably, physicians will be called on more frequently to advise individuals and infection control nurses on the indications for preventive therapy after occupational exposure to tuberculosis. It has been well shown that the greater the exposure, the greater the chance for developing clinical tuberculosis [14, 15]. Persons at risk for developing tuberculosis should be given appropriate preventive therapy [5], but those at little risk should not be subjected to the potential of drug toxicity. The problem is deciding who among the exposed persons is at significant risk and who is not.

Because isoniazid is not completely without risk, it should be given only if the risk for tuberculosis warrants it. Therefore, the first decision to make is whether the exposure is significant. My objective is to review data from six intense exposures of hospital personnel and to provide a guide for the most effective use of isoniazid to prevent tuberculosis in such situations.

Background

It is widely accepted that a healthy person with a positive tuberculin skin test result is at reduced risk for reinfection from subsequent exposure [16]. However, no recent studies support this premise, and there are several reports of well-documented reinfection [17-19]. The growing incidence of drug-resistant tuberculosis in urban hospitals raises this previously academic question to one of practical importance.

Animal studies have consistently shown that an inactive primary tuberculous infection affords great immunity against developing tuberculosis from reinfection [20, 21], and epidemiologic studies suggest a similar protection in humans [22, 23]. However, several prominent European epidemiologists contend that exogenous reinfection is the principal cause of adult tuberculosis in developing countries where reexposure is extremely common [24, 25].

A controlled trial to determine the protection against reinfection from a healed tuberculous infection is precluded on ethical grounds. Thus, I have approached the question by studying "experiments of nature," that is, inadvertent heavy tuberculosis exposures of health care workers for whom accurate records of skin test results before and after exposure are available. Efforts at a meta-analysis to supplement our experience were frustrated by wide variations in conditions, criteria for conversion of the tuberculin skin test result, and a paucity of information on the tuberculin skin test result status of all exposed persons before exposure. Such information is essential to differentiate tuberculin skin test reactors of long standing from skin test conversions related to the exposure.

Data Sources and Definitions

For this paper I draw on 5 outbreaks that ran their course before the chemotherapy era, 6 from the recent literature, and 22 among nursing home residents (Table 1). The recent outbreaks were all aborted by prompt preventive therapy, thus preventing many cases that otherwise would have developed.

It is important to define terms as used in this paper. The definition of a positive intradermal tuberculin skin test result has changed over time from an induration 5 mm from 1 unit of purified protein derivative 50 years ago to 10 mm from 5 units for the past 30 years. Now, an induration 5 mm is considered a positive result when human immunodeficiency virus (HIV) infection is suspected. Nonreactor and tuberculin skin test-negative indicate a reaction smaller than a given author's cut-point (Table 2). Reactor and skin test-positive signify a documented positive reaction as defined by the author before the exposure. Converter and conversion mean an increase of specified amount (shown for the six recent studies in Table 2) in the size of the skin test result after exposure from a reaction of less than 5 mm before exposure. In the studies in which I was involved, we used an increase 15 mm as recommended for elderly persons in nursing homes [26] because it more reliably distinguishes conversions from simple "boosted" or "recalled" reactions. Nonconverter signifies persons who remain skin test negative at least 8 weeks after exposure and may be considered uninfected [27]. Preventive chemotherapy or chemoprophylaxis means a course of isoniazid 300 mg daily for 6 to 12 months. Tuberculosis case refers to someone with clinical tuberculosis or isolation of Mycobacterium tuberculosis from sputum at least 8 weeks after exposure.

Reinfection tuberculosis refers to disease that develops as a result of reexposure to tuberculosis of a known reactor. The terms normal and healthy simply mean not infected by HIV or otherwise seriously immunocompromised, regardless of tuberculin status.

In analyses of results of exposure of nursing home residents in Arkansas, the term no exposure is used to designate nursing homes in which no infectious case (sputum smear-positive) and no tuberculin conversions had been found in 7 years. Heavy exposure designates nursing homes in which an infectious case (sputum smear- and culture-positive) was found in association with skin test conversion in 25% of nonreactors.

The strength of the data reviewed here lies in the availability of reliable skin test results before exposure, which permits accurate classification of each person by skin test result status. The weakness lies in the paucity of data on the outcome of exposed tuberculin reactors in older reports and in the fact that the full impact of the exposure of nonreactors is obscured by prompt treatment with isoniazid in most converters.

Outbreaks in the Era before Chemotherapy

In her review of 109 epidemics of tuberculosis, Lincoln [28] emphasizes that for a tuberculosis outbreak to occur, "a large proportion of the individuals exposed must be tuberculin negative. This state is usually associated with a lack of acquired immunity seen in. geographic areas where the rate of tuberculosis is very low" [28].

Few physicians have experienced the devastation once seen after a heavy exposure to tuberculosis because outbreaks today are generally aborted by preventive therapy with isoniazid. Two classic papers remind us of the havoc such outbreaks can wreak when they run their natural course (Table 1). Heimbeck [29] reported that 210 (95%) of 220 tuberculin skin test-negative nursing students converted to positive during 3 years of training and that 48 (22%) of these developed tuberculosis. This is strikingly different from 3 of 200 initially skin test-positive students (1.5%) who developed tuberculosis, possibly as reactivation of their earlier infection. Hyge [30] reported that 70 (67%) of 105 skin test-negative students converted to positive after exposure to an infectious teacher and that 41 (59%) of these developed tuberculosis. Of 130 skin test-positive classmates, only 2 (1.5%) developed tuberculosis, again possibly from reactivation of an earlier infection. No nonconverter developed the disease.

Finally, among the 109 outbreaks summarized by Lincoln [28], 3 involved adults and provided the information necessary for epidemiologic analysis (Table 1). Among these, 176 (30%) of 591 who were skin test negative before exposure converted to positive; 78 (44%) of these developed tuberculosis. Although the author did not explicitly state that no tuberculosis developed among the 134 persons who had positive skin test results before exposure, the point of the article is that few, if any, did.

Experience in More Recent Outbreaks

Catanzaro [31] reported on 51 health care workers who had been exposed to tuberculosis during bronchoscopy of a patient with pulmonary tuberculosis. Tuberculin skin test conversion occurred in 14 (31%) of the 45 who were skin test negative, 2 (14%) of whom developed tuberculosis before chemotherapy could be started. The 12 treated converters, 31 nonconverters, and 6 known reactors, none of whom was treated, remained well. (The number of reactors exposed and follow-up of all persons were obtained by personal communication with Dr. A. Catanzaro).

Kantor and colleagues [32] reported that 9 (16%) of 56 tuberculin skin test-negative health care workers showed skin test conversion after exposure to a patient with tuberculosis. Three (33%) had developed tuberculosis before the problem was detected. One person was not treated because he was skin test positive as a result of a BCG (bacille Calmette–Guérin) vaccination at age 10 years. He developed primary tuberculosis (pleural effusion) 5 months after exposure. In the 8 years since exposure, no tuberculosis has developed among the 6 treated converters, the 47 nonconverters, and the 24 known reactors, none of whom was treated (Personal communication. Kantor HS).

Haley and colleagues [33] described an outbreak of tuberculosis in an intensive care unit after health care workers were exposed to a patient with pulmonary tuberculosis. Tuberculin skin test conversion developed in 23 (21%) of 112 skin test-negative health care workers; 6 (26%) of these developed tuberculosis before preventive therapy could be given. The 16 treated converters, 89 nonconverters, 21 known reactors, and 4 workers with a history of BCG vaccination remain well, despite receiving no treatment (Personal communication. Luby JP).

Frampton [34] reported that after inadvertent exposure to a patient with tuberculosis of the skin, 12 (20%) of 59 skin test-negative health care workers converted to positive. Three declined isoniazid as preventive therapy and two developed pulmonary tuberculosis. The 9 treated converters, 47 nonconverters, and 10 known reactors were not treated and developed no disease (Personal communication. Frampton M).

In 1992, we described an outbreak among 70 health care workers on a ward with a patient who had 2 weeks of daily irrigation and dressing changes of a large abscess of one thigh [35]. The abscess was later found to be tuberculous, but there was no clinical or radiographic evidence of pulmonary involvement. Of 60 health care workers who were skin test negative before exposure, 36 (63%) showed skin test conversion (>15 mm induration); 5 (14%) of these developed disease before preventive treatment could be started (9 to 12 weeks after exposure). All cases and converters were treated with isoniazid and rifampin, and no more cases of tuberculosis developed. No tuberculosis has developed in 10 years among the 24 nonconverters and 10 known reactors who were not treated.

Since our report of this outbreak [35], cultures from the finger ulcer of the index case and from sputum of two other cases among the converters were found to be viable in a laboratory freezer at the Centers for Disease Control and Prevention. These were shown to have identical restriction fragment-length polymorphism fingerprints (Data provided by Crawford JT and Cave MD).

In this issue of Annals, Templeton and colleagues [36] describe a hospital outbreak in which all five nonreactors involved in an autopsy of a patient with unrecognized tuberculosis had skin test conversions with indurations greater than 15 mm (mean induration, 24 mm). Within 8 weeks, M. tuberculosis was cultured from the sputum of two (40%) of these and were shown by restriction fragment-length polymorphism to be identical to those of the index case. These two health care workers were treated for tuberculosis with isoniazid and rifampin, and the other three were treated preventively with isoniazid alone. None developed clinical disease. The five who were known tuberculin reactors before exposure remain well, despite receiving no therapy.

Experience among Nursing Home Residents

Based on our database of more than 60 000 residents of 230 nursing homes in Arkansas [27, 37], a cohort analysis was done on the effect of heavy tuberculosis exposure on the incidence of active disease among 1160 known tuberculin reactors. It is unlikely that nursing home residents would be more resistant to reinfection than health care workers. This should allow some generalization of the conclusions from the findings to assess the likelihood of reinfection in yet another setting.

In 22 "heavy exposure" nursing homes, the tuberculin skin test result of 363 nonreactors converted to a reaction of 15 mm. All but 52 of these were treated for 1 year with isoniazid, and none developed tuberculosis. However, of the 52 converters whose physicians elected not to treat them with isoniazid, 10 (19% [95% CI, 12% to 29%]) developed tuberculosis. Yet, only 5 (2.4% [CI, 0.8% to 5.5%]) of the 209 skin test-positive residents in this heavy-exposure nursing home developed tuberculosis. In 76 nursing homes with no infectious case, 10 (1.1% [CI, 0.5% to 2%]) of the 921 skin test-positive residents developed the disease within a year of admission (P = 0.17). These values are consistent with the previously reported 2% to 3% risk of skin test-positive nursing home residents for developing active tuberculosis from recrudescence of old infection [37].

Discussion

Top Discussion Author & Article Info References

It is widely stated that, if not treated, about 5% of tuberculin skin test converters will develop tuberculosis within 1 or 2 years and that another 5% will develop it later in life. This generalization is based on conversions 5 mm or 10 mm induration, cut-points that produce a considerable overestimation of new infections. In the six recent hospital outbreaks Table 1, 19 (19%) of 98 skin test converters developed tuberculosis within the few weeks required to recognize the problem and institute preventive chemotherapy. The intensity of exposure is attested by the fact that 98 (29%) of 336 nonreactors showed skin test conversion. It seems safe to say that more would have developed tuberculosis had not 78 (80%) of the converters been treated promptly with isoniazid.

It is noteworthy that despite such heavy exposures, no tuberculosis developed in nonconverters or those who were skin test-positive before exposure. In the former, this could be because of too little exposure or because of innate resistance as suggested from our work in nursing homes [38, 39]. Variation in innate resistance to tuberculosis was also obvious in the 1927 experience in Lubeck, Germany, where 71% of 251 newborn infants survived enormous oral inocula of virulent M. tuberculosis [40]. However, the remarkable protection of the skin test-positive health care workers described here is presumably mediated by prompt activation of macrophages by T cells sensitized by an earlier infection [41, 42].

None of the six recent studies [31-36] is large enough to settle the issue of protection of normal tuberculin reactors from reinfection. However, taken together Table 1, the message seems clear: Healthy tuberculin reactors are well protected from the development of active tuberculosis after heavy exposure. The degree of protection is most clearly seen when conversion is defined as induration 15 mm. This is because of the greater specificity of this cut-point for new infection, albeit with possible sacrifice of sensitivity [43].

Although our experience includes no one with a history of BCG vaccination, two other recent outbreaks include five such persons [31, 32]. One of the five (20%) developed tuberculous pleural effusion 5 months after exposure, which is about what one would expect among untreated converters.

Specific antibody-mediated immunity provides reliable protection from reinfection with smallpox, measles, and mumps [44]. Protection against tuberculosis is less clear because the disease may develop either from a new infection or recrudescence of an old one. The two mechanisms can be distinguished only if a record of previous skin test results is available. However, the data presented here suggest a similar immunity after infection by M. tuberculosis, mediated by prompt activation of macrophages by previously sensitized T cells [41, 42].

On the basis of the data reviewed here, a case of tuberculosis could be prevented for every 2 to 5 converters treated with isoniazid, particularly if an induration cut-point 15 mm is used. This is far less than the 1% to 3% incidence of generally nonfatal toxicity from isoniazid. However, to prevent a case of tuberculosis among exposed skin test-positive persons, 50 to 100 persons must be treated, which would mean the risk for drug toxicity would often be greater than the risk for tuberculosis [5]. However, it should be remembered that the American Thoracic Society and Centers for Disease Control and Prevention guidelines recommend preventive therapy for reactors younger than age 35 years, regardless of exposure [13].

Clinical Application of These Observations

On discovery of a heavy tuberculosis exposure of a group of persons, the first task is to assemble a list of all those exposed along with their skin test results before exposure. It is best to wait about 8 weeks after exposure before retesting nonreactors to allow for skin test conversion to occur. However, if a heavy exposure is discovered immediately, it is advisable to start preventive therapy in nonreactors who were most heavily exposed even before retesting. Persons who are healthy and have remained skin test negative after 8 weeks require no attention beyond reassurance. However, HIV-infected persons who were exposed should be given preventive therapy regardless of skin test reaction because of their reduced ability to react and their exquisite susceptibility to tuberculosis.

Persons who were probably exposed and have skin test reactions 10 mm after exposure can be divided into groups on the basis of skin test results before and after exposure Figure 1:

View larger version (19K): [in this window] [in a new window]   Figure 1. General guidelines for managing the situation after an inadvertent exposure to tuberculosis. Previously known tuberculin skin test reactors and those who remain tuberculin skin test negative 8 weeks after exposure require no therapy unless otherwise indicated by, for example, age or other risk factors. Those with tuberculin skin test conversions of 15 mm or greater should be treated preventively, regardless of age, provided they are clinically healthy and the chest radiograph shows no evidence of tuberculosis. Those with less definite evidence of new infection require individual consideration. Skin test reactors with no skin test record before exposure should be treated as if they had been nonreactors before exposure. Asterisk indicates that for persons younger than age 35 years, preventive therapy should be considered regardless of exposure. Dagger indicates that if the patient has been very heavily exposed, isoniazid therapy should be started. This therapy can be discontinued if the skin test result is still negative.

2. Reactors who have been exposed for whom results before exposure are not available: In the absence of data by which to determine skin test conversion, prudence dictates treating this group as converters.

3. Persons who were skin test positive before exposure: The evidence cited here shows clearly that, even in the face of heavy exposure, a previous infection with M. tuberculosis, as shown by a skin test result of greater than 10 mm induration, affords excellent protection of a healthy person against reinfection and that preventive therapy is not indicated. A recent report by Small and colleagues [45] shows that HIV infection renders even previously infected persons highly susceptible to reinfection if exposed. I do not doubt that exogenous reinfection with M. tuberculosis occasionally occurs in persons without HIV infection. However, the data presented here suggest that exposure to tuberculosis rarely causes disease in healthy tuberculin reactors. Certainly, reinfection is not common enough to justify routine use of preventive chemotherapy for known reactors after exposure. Judgment should be exercised in persons whose immunity may be impaired by other serious health problems.

Special Considerations

There is no evidence that multidrug-resistant tubercle bacilli are more invasive than drug-susceptible strains. Thus, the data presented here suggest that known reactors would be at very little risk for reinfection even by multidrug-resistant organisms. In this situation, there is no consensus on whether simply to observe converters for disease and treat them vigorously if it develops or to give them preventive therapy with more toxic and unproven new drugs [46].

One must consider the exceptions to any useful rule. Occasional instances of tuberculosis resulting from reinfection have developed in persons without HIV infection [17-19]. At least some of these patients were chronically ill, alcoholic, drug abusers, malnourished, and heavy smokers [19] (Personal communication. Nardell EA). Such factors may contribute to reduced local, if not systemic, resistance to infection and should be considered in the management of individual persons after exposure to tuberculosis.

Several of the world's most eminent tuberculosis epidemiologists have contended that reinfection is the predominant mechanism for tuberculosis among adults in developing countries [24, 25]. However, their data are drawn from population studies in which prevalence of tuberculosis is very high and distinguishing new infection from recrudescent disease is impossible. Malnutrition and various coinfections could account for many cases, whether caused by recrudescence of old infection or by exogenous reinfection. The data reviewed here are derived from detailed study of specific groups of healthy persons whose courses were followed individually rather than in aggregate. Our data agree with Styblo's statement that exogenous reinfection of normal persons is negligible in developed countries [47].

Medical students and house officers often ask whether it is safer to be skin test negative or skin test positive. There is no one answer to this question because it depends on the situation of each individual. Persons living in the suburban or rural United States with almost no chance of exposure to tuberculosis are safer if they harbor no potentially harmful tubercle bacilli, that is, if they are skin test negative. Those who will be working in the medical or nursing profession but in situations in which only an occasional exposure is likely are probably as safe one way as the other. On the other hand, those who will be working in urban hospitals, particularly those caring for patients with multidrug-resistant disease, clearly would be safer if they had cellular immunity derived from a previous natural infection, that is, to be skin test positive. Those who are skin test negative in such situations certainly must be monitored at least annually for conversion and treated preventively if conversion occurs.

I hope this study of "experiments of nature" in tuberculosis will prompt hospitals and long-term care facilities to keep better tuberculin skin test records on their employees [3] (including medical staff [48] and students [49]) on which to base advice in the event of an unexpected exposure. I also hope it will be of help to physicians when they must advise health care workers after a heavy exposure to tuberculosis.