Differences in Clinical Presentation among Persons with Pulmonary Tuberculosis: A Comparison of Documented and Undocumented Foreign-born to US-born Persons

Abstract

Background

Most tuberculosis (TB) cases in the US are diagnosed in foreign-born persons, and undocumented foreign-born may face particular barriers to timely access to health services. This study investigates whether differences in clinical presentations among persons with pulmonary TB (PTB) are associated with foreign-birth or documentation status.

Methods

In this cross-sectional study, we reviewed the medical records of patients diagnosed with microbiologically proven PTB at a New York City public hospital between April 1999 and March 2005. Three groups with PTB: US-born, foreign-born with documents, and undocumented, were defined at presentation and compared. Odds ratios (ORs) for the presence of symptom duration ≥8 weeks prior to hospital admission for each group were estimated using logistic regression.

Results

Among 194 subjects with newly diagnosed PTB, 61 (31%) were US-born, 62 (32%) were documented, and 71 (37%) were undocumented foreign-born. Undocumented persons presented with significantly higher frequencies of cough (p=0.020) and hemoptysis (p=0.012), and had a significantly longer median duration of symptoms compared to US-born (8 versus 4 weeks; p=0.023). No significant differences between documented foreign-born and US-born were observed. In multivariate analysis, being undocumented relative to US-born (OR_adj 4.1, 95% CI 1.7–10.2; p=0.002) and being unemployed (OR_adj 2.2, 95% CI 1.1–4.5; p=0.023) were independently associated with prolonged symptom duration ≥8 weeks.

Conclusion

An undocumented status is associated with increased frequency of cough and hemoptysis, and longer symptom duration prior to hospital evaluation for PTB. Whether reducing barriers to health services for undocumented persons could enhance TB control deserves further study.

Introduction

Despite its declined incidence, tuberculosis (TB) continues to be a social, public health, and economic burden in urban areas of the US. In fact, experts fear a resurgence of the disease due to decreased funding for TB elimination at the Centers for Disease Control and Prevention (CDC) [1]. Further, the increased incidence of multi-drug resistant TB (MDR TB) worldwide, and the occurrence of extensively drug-resistant (XDR) TB highlight the necessity for better TB control [2]. In 2006, 13,770 active TB cases were reported to the CDC, representing a TB case rate of 4.6 cases per 100,000 population in the US, a rate much higher than the Healthy People 2010 objective of 1.0 case per 100,000 population [3, 4].

The majority (57%) of TB cases nationwide are diagnosed in foreign-born persons. The TB case rate for foreign-born is almost 10 times higher than the case rate for US-born persons, 22.0/100,000 versus 2.3/100,000, respectively [4]. New York City (NYC) has a high proportion of foreign-born residents, and its TB case rate is 2.6 times higher than the national average [5]. Concomitant with a continuing decline in annual TB case rates in NYC, the proportion of foreign-born persons among reported TB cases rose from 18% in 1992 to over 70% in 2006. Whether differences in clinical presentations among persons with pulmonary TB (PTB) are associated with foreign-birth or documentation status in comparison to US-birth is unknown.

Undocumented foreign-born persons may face particular barriers to timely access to health services resulting in delayed hospital evaluation for suspected PTB compared to US-born persons. We hypothesized that among foreign-born persons with PTB the undocumented would present with signs of more advanced disease than would US-born. The objectives of this study were i) to measure the extent of disease, and the presence and duration of symptoms at the time of initial hospital evaluation for PTB, and to compare the results for documented and undocumented foreign-born to US-born persons, and ii) to assess whether documentation status among foreign-born relative to US-born has an independent association with prolonged symptom duration prior to initial hospital evaluation in persons with PTB after adjustment for other risk factors.

Methods

Design, Setting and Study Population

In this cross-sectional study, we reviewed the medical records of all identifiable patients 18–99 years old who were diagnosed with PTB between April 1999 and March 2005 at Bellevue Hospital Center (BHC). This 800-bed public hospital is part of the New York City Health and Hospitals Corporation (HHC) which is a public benefit corporation that serves 1.3 million New Yorkers and nearly 400,000 who are uninsured [6]. As the largest HHC hospital, BHC is located in midtown Manhattan, and thus evaluates most TB suspects from that community. Over the past years, including the time period studied, BHC together with Elmhurst Hospital Center, another HHC facility located in the borough Queens, NY, has been the hospital reporting the highest number of new TB cases among all facilities in NYC [7].

Records of patients who were reported to the NYC Department of Health as new probable active TB were screened, and the records of patients with microbiologically proven active TB were reviewed. Patients whose diagnosis of PTB was confirmed by growth of Mycobacterium tuberculosis in a respiratory specimen were included in the analysis. Patients were excluded from analysis if they were diagnosed with extrapulmonary TB without microbiologically proven pulmonary disease, if they were diagnosed with active PTB prior to hospital admission, or if information was missing on outcome variables or documentation status. Patients with a TB diagnosis prior to admission were excluded because BHC is a referral hospital for TB patients detained by the New York City Department of Health for non-compliance with their TB medications. Including such patients could potentially skew results because these patients are often partially treated for several months, come from outside the community and information on outcome variables such as symptom duration at the time of diagnosis is frequently vague. Furthermore, these patients are often not reported by BHC as new cases of active TB, and thus would not be detected by our screening method.

Approval for human subjects’ research was obtained from the Institutional Review Boards of the New York University School of Medicine and BHC.

Measurements

Information on reported variables was extracted from the admitting physician’s note, social worker’s note, and diagnostic test reports in the patients’ medical records. Our main variables of interest were location of birth and documentation status. The patients’ self-reported information on location of birth was extracted from the physician’s note, while self-reported information on documentation status was extracted from the social worker’s note. Statements in the social worker’s note such as “undocumented”, “no legal papers” or “no visa” were considered indicative of an undocumented status. Subjects were categorized into three groups, US-born, documented foreign-born, and undocumented foreign-born. Persons born in Puerto Rico or US Virgin Islands were considered US-born. Additional demographic factors recorded included sex, age, race as per physicians note, self-reported years in the US for foreign-born persons, health insurance and self-reported employment status, and homelessness.

Clinical characteristics included HIV status, other diagnostic test results towards establishment of PTB diagnosis, and self-reported symptoms. Chest X-ray results were recorded as either unilobar versus multilobar or miliary infiltrates with separate scoring for the presence or absence of cavitary lesions. Sputum smears for acid fast bacilli (AFB) were recorded as positive if at least one of the initial three smears was positive regardless of quantity of AFB seen per microscopy slide. Furthermore, the degree of smear-positivity was categorized into rare (8–10), few (15–20) and numerous AFB per slide. The presence of multilobar or miliary infiltrates, cavitary lesions, or smear positivity were considered potential signs for more advanced disease. Because HIV-mediated immunosuppression can impair granuloma formation, resulting in both diminished formation of pulmonary cavities and atypical infiltrates [8], we performed univariate analysis including and excluding HIV-infected subjects.

The patients’ self-reported symptoms that were recorded as potentially suggestive of PTB included the presence of cough, hemoptysis, fever, night sweats, and weight loss over 2 lbs. For each of these symptoms the patients’ self-reported duration was recorded in weeks prior to hospital evaluation. The longest duration of any one of the symptoms suggestive of PTB, as listed above, was considered the symptom duration. For multivariate analysis, symptom duration was treated as a dichotomous outcome with a cut-off of ≥8 weeks based on the median duration of 7 weeks for all subjects included in the analysis.

Statistical Analysis

Statistical analysis was performed using STATA software, version 9.2 (StataCorp, College Station, TX). A two-tailed α < 0.05 was considered to be statistically significant.

On univariate analysis, depending on distribution, we used the t test or Mann-Whitney U test when comparing two groups, and the one-way ANOVA or Kruskall-Wallis test when comparing three groups. For categorical variables we used the chi-square test without correction for continuity. In each case a summary test was used to assess differences between the three groups, a significant or near significant summary test was followed by pairwise contrasts between documented foreign-born compared to US-born, and undocumented compared to US-born persons. For the pairwise contrasts a Bonferroni corrected two-tailed α < 0.025 was used as the criterion for statistical significance.

Multivariate logistic regression models were constructed to test whether there was an independent association between the dichotomous outcome variable symptom duration ≥ 8 weeks and documented or undocumented foreign-born status with US-born as reference. Variables whose association with outcome had a p value < 0.2 in univariate analysis or which are known to have an impact on delayed PTB diagnosis were initially included in the model. Adequacy of the final model was assessed by the goodness-of-fit test.

Results

We identified 244 patients with a diagnosis of active TB between April 1999 and March 2005. After exclusions (Figure 1), 194 patients with newly diagnosed microbiologically proven PTB were included in the analysis. Among the 194 subjects evaluated, 61 (31%) were US-born, 62 (32%) were documented foreign-born and 71 (37%) were undocumented. Table 1 shows the characteristics of these three groups and apart from years in the US, lists the summary p values for the three group comparison.

Figure 1.

Reasons for exclusion of subjects from analysis

Table 1.

Characteristics of subjects with PTB (n=194) according to location of birth and documentation status

	US-born*	Foreign-born
	N=61	Documented N=62	Undocumented N=71	P value
Characteristic
Male sex, n (%)	46 (75)	48 (77)	43 (61)	0.063a
Age, mean±SD	45±12	42±12	33±10	< 0.001b
Race, n (%)
Black	34 (56)	10 (16)	5 (7)
Hispanic	22 (36)	11 (18)	12 (17)	< 0.001a
Asian	2 (3)	37 (60)	50 (70)
White	3 (5)	4 (6)	4 (6)
Years in USd, median (range)	NA	9 (1–37)	3 (1–13)	< 0.001c
Health insurede, n (%)	41 (68)	37 (61)	8 (11)	< 0.001a
Unemployed, n (%)	53 (87)	38 (61)	44 (62)	0.002a
Homeless, n (%)	15 (25)	8 (13)	2 (3)	0.001a
HIV+f, n (%)	22/44 (50)	4/37 (11)	8/37 (22)	< 0.001a

^*Includes Puerto Rico and US Virgin Islands

^aChi-square test for 3 group comparison

^bOne-way ANOVA

^cMann-Whitney test

^dInformation on years in the US available in 122/133 (92%) foreign-born persons

^eHealth insurance information available for 192 (99%) subjects

^fHIV-1/2 serology results available for 118 (61%) subjects

Comparison of Characteristics

In pairwise contrast, documented foreign-born were similar to US-born subjects with regard to sex, age, and health insurance but differed significantly in racial distribution (p<0.001). Among both documented and undocumented foreign-born persons, Asians represented the largest ethnic group, while the US-born group consisted mostly of Black and Hispanic persons. According to frequency, the countries of origin among foreign-born Asians were as follows: China 44/87 (51%), Nepal 11/87 (13%), Tibet 10/87 (12%), Philippines 7/87 (7%), and <5% for each of several other Asian countries 15/87 (17%). Documented as well as undocumented foreign-born were significantly less likely to be unemployed (p<0.001) or HIV-infected (p<0.001 and 0.008, respectively) than US-born persons. Undocumented foreign-born were significantly younger and less likely to have health insurance than US-born persons (p<0.001).

Comparison of Diagnostic Test Results

No significant differences among the 3 groups were observed regarding the presence of multilobar or miliary infiltrates, cavitary lesions or sputum smear-positivity (Table 2). Furthermore, no significant differences among the 3 groups were observed in the degree of smear-positivity. This did not change when HIV co-infected subjects were excluded from univariate analysis.

Table 2.

Diagnostic test results and symptoms in subjects with PTB (n=194) according to location of birth and documentation status

	US-born*	Foreign-born
	N=61	Documented N=62	Undocumented N=71	P value
Diagnostic test results
Multilobar or miliary infiltrates, n (%)	41 (67)	44 (71)	44 (62)	0.541a
Presence of cavitary lesions, n (%)	20 (33)	23 (37)	26 (37)	0.886a
Smear+c, n (%)	38 (62)	35 (56)	47 (66)	0.512a
Presence of symptoms suggestive of active PTBd
Cough, n (%)	40 (66)	46 (74)	59 (83)	0.069a
Fever, n (%)	32 (52)	32 (52)	46 (65)	0.224a
Night sweats, n (%)	19 (31)	16 (26)	31 (44)	0.081a
Weight losse, n (%)	28 (46)	29 (47)	35 (49)	0.920a
Hemoptysis, n (%)	9 (15)	11 (18)	24 (34)	0.018a
Symptom duration suggestive of active PTB
Symptom durationf, median weeks (range)	4 (0–36)	4 (0–104)	8 (1–104)	0.014b

^aChi-square test for 3 group comparison

^bKruskall-Wallis test

^cAt least one of initial three sputum smears positive for AFB regardless of quantity

^dSymptoms self-reported by patients

^eself-reported weight loss of more than 2 lbs

^fLongest duration of any one or several symptoms suggestive of PTB

Presence of symptoms suggestive of PTB

Table 2 shows the presence of symptoms suggestive of PTB, and the summary p values for the comparisons of the 3 groups. Due to the near-significant difference for the presence of cough between the 3 groups (p=0.069) the pairwise contrast was performed. Presentation with cough was significantly more frequent in undocumented foreign-born compared to US-born (p=0.02), while there was no significant difference between documented foreign-born and US-born (p=0.297). Similarly, for the presence of hemoptysis which was significantly different in the 3 group comparison (p=0.018), the pairwise contrast showed that hemoptysis was significantly more present in undocumented foreign-born compared to US-born persons (p=0.012), while there was no significant difference between documented and US-born (p=0.653). Other symptoms suggestive of PTB like fever and night sweats showed a trend of higher frequency in undocumented foreign-born compared to US-born without reaching statistical significance.

Duration of symptoms suggestive of PTB

In univariate analysis, the duration of symptoms prior to hospital evaluation for PTB was significantly different among the 3 groups (p=0.014; Table 2). Undocumented had significantly longer median symptom duration (8 weeks) than US-born (4 weeks; p=0.023) while there was no significant difference between documented foreign-born and US-born persons.

Table 3 shows the adjusted odds ratios (OR_adj) and 95% confidence intervals (CI) for foreign-born/documentation status relative to US-born with prolonged symptom duration of ≥8 weeks. In multivariate analysis, being undocumented compared to US-born was independently associated (OR_adj 4.1, 95% CI 1.7–10.2; p=0.002) with symptom duration ≥8 weeks. In addition, being unemployed (OR_adj 2.2, 95% CI 1.1–4.5; p=0.023) was independently associated with prolonged symptom duration. Neither lack of health insurance nor HIV-infection was significantly associated with prolonged symptom duration; neither were homelessness nor any of the other demographic variables. Race was taken out of the model because of multicollinearity since the foreign-born were predominantly Asian and the US-born were almost all Black and Hispanic. To control for potential confounding by race we constructed a logistic regression model for Asians alone, the largest ethnic group among the foreign-born with PTB. Table 4 shows the adjusted association of an undocumented status relative to a documented status with prolonged symptom duration for Asians born outside the US (OR 3.3, 95% CI 1.0–10.6; p=0.045).

Table 3.

Adjusted association^a of birth location and documentation status with duration of symptoms ≥8 weeks in patients with PTB

Variable	Adjusted OR	95% CI	P value
Birth location/documentation status
US-born	Reference
Foreign-born/documented	1.01	0.46–2.24	0.976
Foreign-born/undocumented	4.14	1.68–10.22	0.002
Health insuredb	1.51	0.73–3.10	0.240
Unemployed	2.24	1.12–4.46	0.023
HIV statusc
HIV negative	Reference
HIV positive	0.74	0.31–1.81	0.511
HIV unknown	0.81	0.41–1.57	0.525

^aLogistic regression model with 192 observations

^bInformation on health insurance available on 192/194 (99%) subjects

^cHIV-1/2 serology results available for 118 (61%) subjects

Table 4.

Adjusted association^a of documentation status with duration of symptoms ≥8 weeks in foreign-born Asians with PTB

Variable	Adjusted OR	95% CI	P value
Documentation status
Documented	Reference
Undocumented	3.30	1.03–10.61	0.045
Health insuredb	1.13	0.34–3.77	0.841
Unemployed	1.77	0.69–4.53	0.233

^aLogistic regression model with 86 observations

^bInformation on health insurance available on 86/87 (99%) subjects

Discussion

To our knowledge, this is the first study that compared the clinical presentation of documented and undocumented foreign-born to US-born persons with PTB at the time of initial hospital evaluation. Contrary to hypothesized expectation, we found no significant differences between undocumented and US-born persons regarding the presence of multilobar or miliary infiltrates, cavitary lesions or smear-positivity. However, being an undocumented foreign-born person was significantly associated with an increased frequency of cough, hemoptysis, and symptom duration ≥8 weeks, compared to those who were US-born. In contrast, documented foreign-born persons did not differ from US-born in these respects.

The epidemiologic profile of our study population was largely similar to that reported for TB cases in NYC during the study period, although the proportion of homeless and HIV-infected subjects as well as Asians was larger in our study population [9]. This is likely due to the fact that our study site is a public hospital which is located in midtown Manhattan, NY, close to homeless shelter facilities and Chinatown. Because HIV infection and homelessness are known risk factors for TB, and might be associated with the clinical presentation of TB, we performed univariate analyses with inclusion as well as exclusion of HIV-infected subjects, and adjusted for HIV-infection and homelessness in multivariate analysis. We further controlled for potential confounding by Asian race. In addition, our reported frequencies of smear-positivity, cavitary lesions, and symptoms associated with PTB were similar to the frequencies found in other US studies [10–12], and the median reported duration of symptoms prior to hospital evaluation in our study population was similar to the average patient delay of 5–10 weeks prior to TB diagnosis reported in other US-based studies [11–14]. We therefore believe that our results might allow inferences towards other US urban areas with large immigrant populations.

It is well known that persons with smear-positive cavitary disease transmit TB more frequently than those with most other forms of PTB [15]. Furthermore, studies have shown that a delay in TB diagnosis is associated with greater transmission of infection to contacts [10, 15]. Ash et al. found that during the delay between symptom onset and TB diagnosis, a person exposes on average 8 contacts [12]. Our findings suggest that the difference in time to presentation may not impact the course of disease in undocumented persons but could, in addition to the higher frequency of cough, lead to an increased exposure of close contacts. Although population-based molecular epidemiologic studies performed in the US and Europe have not demonstrated significant associations between foreign-born strains and recent TB transmission, these studies have not specifically examined transmission among close contacts of undocumented persons [16–20]. A detailed investigation would be needed to further investigate whether undocumented who, as our data suggests, may face different barriers to health care access, transmit TB more than documented foreign-born persons.

The underlying reason for the significantly higher frequency of hemoptysis in undocumented foreign-born compared to US-born persons is not clear to us. The presence of hemoptysis is often associated with cavitary lesions, and therefore could be considered a sign of advanced PTB [21]. However, the presence of cavitary lesions on chest X-rays in our study population was similar among the three groups, and detailed information on size of cavitary lesions was not available for most subjects. For the majority of persons regardless of ethical and cultural background hemoptysis is an extremely worrisome symptom. It is conceivable that the occurrence of hemoptysis would lead undocumented foreign-born persons to overcome their potential barriers for seeking healthcare. Larger studies would be needed to further investigate whether the higher frequency of hemoptysis in undocumented persons is due to more advanced disease at the time of hospital evaluation, due to major health concerns initiating immediate health care seeking behavior, or both.

The significantly positive association between an undocumented status with prolonged symptom duration could be due to patient factors which have been shown to be associated with a delay in TB diagnosis. Among these are: Being of non-Caucasian race [10]; having a primary language other than English [11]; being afraid of immigration authorities [12]; being concerned about costs [12]; being unemployed [12]; and having a lower level of education [13]. Although undocumented immigrants in our study population were significantly less likely to have health insurance or be of Caucasian race, we found neither of these to be significantly associated with prolonged symptom duration. We did find an independent positive association between unemployment and prolonged symptom duration, which is consistent with results from previous studies [12]. A detailed analysis of factors potentially contributing to the prolonged symptom duration of undocumented foreign-born persons was beyond the scope of this study and would require a larger and prospective investigation.

Our study was limited by the retrospective nature of the design that did not permit us to evaluate factors such as English language speaking ability which was not described in detail in the records of most patients. Furthermore, we were not able to assess treatment outcomes for our study population because part of our study population received TB treatment outside of Bellevue Hospital, and thus this information was not available to us. Bellevue Hospital Center serves a predominantly poor and disadvantaged population, and therefore our results might not be generalizable to non-public institutions. The fact that lack of health care insurance was not significantly associated with outcome may reflect this limitation. Our study was further limited by the self-reported information on documentation status. It is conceivable that undocumented persons might report themselves as documented due to fear of immigration authorities. However, this would create a bias towards smaller observed differences between documented and undocumented.

Undocumented foreign-born persons have been the ongoing focus of intense policy debate over the past several years. A recent study, analyzing data from the 2003 California Health Interview Survey showed that undocumented Mexicans and other undocumented Latinos reported less use of health care services compared to their US-born counterparts [22]. Our data showing significantly longer symptom duration prior to hospital evaluation for PTB in undocumented but not documented foreign-born compared to US-born persons is consistent with these findings. This raises the concern that current political developments with aggressive measures targeting undocumented foreign-born persons could further jeopardize timely access to health services in this disadvantaged population.

In conclusion, an undocumented status is associated with increased presentation of cough, hemoptysis, and longer symptom duration prior to hospital evaluation for PTB. Whether identifying and reducing barriers to health services for undocumented persons can enhance case finding and TB control deserves further study.