Contact Investigations and the Continued Commitment to Control Tuberculosis

The incidence of tuberculosis (TB) in the United States has steadily declined over the past 9 years.¹ However, disease outbreaks and pediatric cases continue to occur, indicating the ongoing transmission of Mycobacterium tuberculosis. To continue to control TB transmission and prevent outbreaks, a major priority is to improve contact investigations. While the basic steps for contact investigations are well established, implementation of these steps can be challenging.

In this issue of THE JOURNAL, Reichler and colleagues² describe contact investigations conducted in 1996 in 5 areas of the United States that, in the authors' opinion, had the "best-established contact investigation programs and best-organized records." Their study revealed many deficiencies in the conduct of these investigations. First, although contact investigations should be conducted for all infectious TB patients, the authors found no record of a contact investigation for 3% of patients. Second, although all persons can be presumed to have at least 1 close contact, 11% of patients had no close contacts identified. A major gap was the identification of contacts for patients residing in homeless shelters—only 50% of whom had contacts noted in their case record. Third, contact investigations will disrupt transmission only if identified contacts are evaluated and provided treatment for active disease or infection. Yet, Reichler and colleagues have also found that 39% of contacts were either not evaluated or incompletely screened.³ In addition, key information was lacking in most contact records. For example, only 17% of records documented the infectious period of the TB case and only 10% recorded the human immunodeficiency virus (HIV) status of the contact. Another study by Marks et al⁴ found similar gaps in contact investigations of patients with TB from US urban areas.

Given the apparent deficiencies in contact tracing and the reality of limited public health resources, improving the focus of contact investigations should be a high public health priority. Identification of characteristics of contacts who are most likely to develop infection or disease could potentially enable public health workers to better prioritize those contacts in need of evaluation.

In a related article in this issue, Bailey and colleagues⁵ describe a model incorporating case, contact, and environmental factors that predict contacts with a positive tuberculin skin test (TST) result. The authors used data from 292 cases and their 2941 contacts reported to the Alabama Department of Public Health in 1998 to derive their model, which aims to focus and streamline contact investigations. The model suggests that contacts of TB cases with positive sputum smear results or cavitary disease are more likely to have a positive TST result than contacts of patients without these clinical findings. Nonwhite race and hours of exposure to the case patient also predicted a positive TST result in contacts. Characteristics of contacts that were negatively correlated with a positive TST result were female sex and age 15 years or younger.

While Bailey et al should be applauded for their approach, certain limitations of their study and the data interpretation deserve consideration. One important limitation in their assessment of transmission is that contacts with positive TST results included both those who tested positive at their initial TST screening and those who tested positive at the 3-month follow-up screening. As the authors note, individuals who test positive at the initial TST include those recently infected as well as those who were infected in the distant past but who were not previously detected. Since the background rate of TST positivity varies by demographics, using the initial TST results could have introduced bias in assessing recent transmission.

The higher proportion of positive TST results among adults compared with children could reflect a higher background TST positive prevalence in an adult population. The finding that child contacts were less likely to have an initial positive TST result could be a consequence of the lower likelihood of remote infection in children, and not a lower likelihood of recent infection. Similarly, the proportion of contacts with positive TST results could be higher among nonwhite individuals because of prior exposure to TB, and not necessarily because they have a greater likelihood of becoming infected after recent exposure to a person with TB. For this reason, applying their model to make decisions about limiting the evaluation of contacts according to these demographic factors could be problematic. For example, based on their model, a 1-year-old white girl would be accorded a lower priority for contact investigation than a 25-year-old Asian or black man. The use of this model to determine which contacts to screen deserves careful consideration.

When prioritizing contacts for evaluation, it is important to consider not only the risk that transmission occurred, but also the risk that infection will progress to disease. Tuberculosis case contacts who are younger than 5 years should have high priority for investigation because (1) contact investigations are a primary method of preventing childhood tuberculosis, (2) the progression from infection to disease is high among young children and the frequency of disseminated disease and meningitis is greater in young children than in adults, and (3) infected children are a reservoir of future infectious cases.⁶ Preventing disease in child contacts of TB cases is such high priority that it is recommended that children who initially have a negative TST result but have contact with an infectious patient start treatment for assumed latent TB infection.⁷ Bailey et al determined that their model predicted a positive TST result with a sensitivity of 89%, suggesting that 11% of contacts with a positive TST result would not be identified. A key issue is determining what proportion of infected children younger than 5 years would be represented among these 11%, and therefore would fail to be identified for evaluation by the model.

Contact investigations are a major cornerstone of public health practice because they can detect new TB cases and prevent future cases. Because newly infected contacts are at substantial risk for progression to disease, contact investigations should be prioritized over other TB screening efforts.⁸ Coincident with this prioritization, efforts are needed to enhance the effectiveness of contact investigations. Of paramount importance is improving the elicitation of contact information from case patients. Techniques such as the social network approach, routinely used in sexually transmitted disease control, deserve evaluation.⁹ Once all contacts have been identified, selecting those who are at highest risk of becoming infected and progressing to disease requires systematic collection of salient information.

Bailey et al identified and created clear definitions for some transmission factors that are often difficult to ascertain, including the frequency and duration of exposure and features of the exposure environment(s). The need for complete information collection to assess contact risk was underscored by Reichler et al, who found that there was documentation of the sputum smear results in only 38% of contact records and of the length of exposure to patients in only 1% of contact records. Modern information systems can facilitate systematic collection of data and tracking of patients and contacts. Molecular methods to track transmission of M tuberculosis also have complemented traditional public health approaches to better define case, environment, contact, and bacterial characteristics that pose the highest risk for transmission.^{10 - 13} Recently developed diagnostic tests that are logistically simpler and have greater specificity than the TST could also refine and focus contact investigations.¹⁴

The final and necessary steps in contact investigations are completely evaluating contacts for disease and beginning therapy when infection or disease is identified. Case contacts have a high risk of developing active TB in the first 2 years following infection.¹⁵ Approaches to overcome barriers posed by certain patient and contact populations, including directly observed therapy for both disease and latent infection, have been effective in improving treatment completion and may warrant wider implementation.^{4 ,16}

Clinicians have an important role in interrupting TB transmission. To ensure timely initiation of contact investigations, clinicians should collaborate with health departments and promptly report persons with suspected or confirmed TB. Because the report of a suspected or confirmed TB case triggers the contact investigation, public health departments are dependent on community physicians to initiate prompt contact investigations. Delays in the diagnosis and reporting of TB impede the prompt initiation of contact investigations. Clinicians also can play an important role in ensuring that contacts of TB patients undergo initial and follow-up skin testing, as well as radiographic evaluation if warranted, and that they initiate and complete therapy.

Maintaining a successful and effective contact investigation program is no small task. For example, in the year 2000, local public health jurisdictions in California conducted more than 2900 contact investigations involving more than 19 000 contacts of smear-positive pulmonary TB cases (unpublished data). The cost of contact investigations could be reduced and the benefits enhanced by assessing these current efforts, evaluating new approaches, and applying those that are shown to be effective. The 2 articles in this issue of THE JOURNAL are good examples of studies that assess current efforts and evaluate new approaches, and convey the important message that strengthening and focusing contact investigations is essential. With strong collaboration between front-line clinicians and public health departments, the tide of TB can be stemmed.