We conclude that observed declines in 2009 TB case counts accurately reflect a true decline in the number of TB cases diagnosed in the United States, and that the deviation from past trends is significant. The unexpected decline in reported TB cases began rather abruptly in late 2008 or early 2009, and may have begun to reverse in mid 2009. We were able to exclude changes in electronic reporting systems as a causal factor, since declines were seen in states that used new software and well as states that did not. Independent information systems (NTSS, BioSense prescription claims, and public health laboratories) reported similar magnitude declines. Reductions in TB program staff were not associated with greater declines, and TB controllers reported no procedural changes. We found no cases unreported to the NTSS in our examination of over 5400 individual line-listed reports in 11 areas. A detailed study in two states similarly found no evidence of surveillance underreporting [16
Declines were seen both in culture-positive TB cases and in cases lacking diagnostic laboratory findings. Analyses showed declines particularly among the foreign-born, and among Hispanics and non-Hispanic blacks. Declines were observed among foreign-born persons from countries that implemented new TB procedures for overseas screening and from countries using older procedures for pre-immigration screening. Updated pre-immigration screening to identify and treat TB prior to U.S. arrival is expected to reduce TB in the United States. A recent California study reported decreased TB incidence diagnosed within 6 months of U.S. arrival in persons screened with revised pre-immigration TB screening, compared to persons screened under the older procedures [17
]. Authorized immigrants screened prior to U.S. arrival comprise approximately 500,000 arrivals in the United States each year, compared to greater than 160 million annual nonimmigrant admissions among students, temporary workers, diplomats, family members, and visitors, who are not routinely screened prior to U.S. arrival [18
]. Because immigration status was not available for analysis of NTSS data, we were not able to adjust for whether foreign-born persons diagnosed with TB in the United States were screened with the revised or older pre-immigration procedures, only whether they originated from a country that had begun to implement updated procedures prior to 2009.
The unexpected decline in TB cases among the foreign-born could have occurred because of fewer persons entering or more persons leaving the United States. U.S. Census figures estimate a -1.6% decline in the foreign-born population from 2008 to 2009. Since there was an -11.1% decline in observed compared with expected foreign-born TB cases, the decline in population does not explain the entire decline in foreign-born TB case counts. Our finding that the decline in foreign-born cases was most prominent among those whose TB occurred within 2 years of their arrival, and among persons from Mexico and Guatemala (who together comprised 26% of foreign-born cases), might indicate that the March 2009 Census population data do not accurately reflect population changes in these groups. Department of Homeland Security data suggest greater declines in the foreign-born population in the United States than do Census data, including a -6.9% decrease from 11.6 to 10.8 million unauthorized immigrants from 2008 to 2009 [19
]. Our finding is consistent with analyses showing that the decline in unauthorized immigrants in the United States is strongly associated with decreased immigration from, and increased deportation to, Mexico [20
]. Nonimmigrant authorized admissions to the United States also declined, from 175.4 million in 2008 to 162.6 million in 2009 [18
]. Financial data indicate that remittances to Mexico and Latin America began to decline in 2008 and early 2009, consistent with the timing of TB declines, and began to rebound in late 2009 [21
Case count declines among the U.S.-born were less impressive than among the foreign-born and cannot be substantially explained by less TB transmission from fewer foreign-born cases or by less transmission among the U.S.-born, since genotype clustering among the U.S.-born did not significantly decline from 2008 to 2009. A limitation of this analysis is that, although genotype clustering is thought to be associated with recent transmission, it is not a direct measure of transmission. Nevertheless, assessment of clustering did not show an abrupt decline in genotype clustering coincident with the abrupt decline in TB cases, suggesting that the decline in cases was not the result of a sudden improvement in TB control practices.
Some of the most dramatic percentage declines occurred among minorities and the socially disadvantaged. These declines may reflect a general problem of delayed access to diagnostic services related to the recent economic recession and some combination of loss of health insurance, fear that seeking medical care might lead to legal consequences, or inability to pay for services [22
]. We were not able to assess health insurance status or health seeking behavior among TB patients; however, recent national reports of overall declines in physician visits [25
] and correlation between unemployment and TB incidence [26
] suggest that the decline in reported TB cases may be further evidence of the impact of the economic recession.
Our finding that laboratory reports of culture-confirmed diagnoses of TB as a proportion of specimens submitted declined more (-9.8%) than did the number of patients who had specimens submitted for culture (-5.9%) is not consistent with a decline in the index of suspicion for TB among healthcare providers. The abruptness of the decline in cases beginning in late 2008 or early 2009 is also not consistent with this hypothesis.
Although we ruled out surveillance artifact in terms of the total case count, there were increases in missing data in some routinely collected variables, representing small numbers but large increases compared to expectation, which could affect the interpretation of our analyses. For example, if missing data for social risk factors such as injecting drug use and homelessness were complete, our finding of steep declines among persons with these risk factors might either remain unchanged (if missing responses were "no," or were distributed proportional to observed data) or be attenuated to indicate lesser declines (if missing social risk factors were "yes"). Typically, the national surveillance database becomes more complete over time, since TB programs have two years to follow and close out cases in electronic reporting. This incompleteness represents a limitation to using surveillance data for our comparison of observed case counts to expected counts based on prior year counts, which are more complete.
In 2010, provisional NTSS data include 11,181 TB cases reported in the United States, for a rate of 3.6 cases per 100,000 population, which was a decline of -3.9% from 2009 similar to the average decline in TB rates (-3.8% per year) from 2000-2008 [27
]. Although the steady decline in TB rates in the United States is evidence of continued progress in TB control, the abrupt decline in 2009 followed by a return to average decline is not consistent with sudden improvements in TB control efforts. Multiple causes coincident with economic recession in the United States, including decreased immigration and delayed access to medical care, could be related to TB declines.