The recent resurgence of tuberculosis (TB) in the United States has led to a number of nosocomial outbreaks of this disease. 1 Emergency department

Transcription

1 1056 TUBERCULOSIS Sokolove et al. TUBERCULOSIS PRESENTATION BRIEF REPORTS The Emergency Department Presentation of Patients with Active Pulmonary Tuberculosis PETER E. SOKOLOVE, MD, LORCA ROSSMAN, MD, STUART H. COHEN, MD Abstract. Objective: To determine the clinical presentation of emergency department (ED) patients with active pulmonary tuberculosis (TB). Methods: This was a retrospective medical record review of adult patients, identified through infection control records, diagnosed as having active pulmonary TB by sputum culture over a 30-month period at an urban teaching hospital. The ED visits by these patients from one year before to one year after the initial positive sputum culture were categorized as contagious or noncontagious, using defined clinical and radiographic criteria. The medical records of patients with contagious visits to the ED were reviewed to determine chief complaint, presence of TB risk factors and symptoms, and physical examination and chest radiograph findings. Results: During the study period, 44 patients with active pulmonary TB made 66 contagious ED visits. Multiple contagious ED visits were made by 12 patients (27%; 95% CI = 15% to 43%). Chief complaints were pulmonary 33% (95% CI = 22% to 46%), medical but nonpulmonary 41% (95% CI = 29% to 54%), infectious but nonpulmonary 14% (95% CI = 6% to 24%), and traumatic/orthopedic 12% (95% CI = 5% to 22%). At least one TB risk factor was identified in 57 (86%; 95% CI% = 76 to 94%) patient visits and at least one TB symptom in 51 (77%; 95% CI = 65% to 87%) patient visits. Cough was present during only 64% (95% CI = 51% to 75%) of the patient visits and hemoptysis during 8% (95% CI = 3% to 17%). Risk factors and symptoms that, if present, were likely to be detected at triage were foreign birth, homelessness, HIV positivity, hemoptysis, and chest pain. Conclusions: Patients with active pulmonary TB may have multiple ED visits, and often have nonpulmonary complaints. Tuberculosis risk factors and symptoms are usually present in these patients but often missed at ED triage. The diversity of clinical presentations among ED patients with pulmonary TB will likely make it difficult to develop and implement high-yield triage screening criteria. Key words: tuberculosis; emergency department; triage; symptoms. ACADEMIC EMERGENCY MEDICINE 2000; 7: From the Division of Emergency Medicine (PES) and Division of Infectious Diseases (SHC), UC Davis School of Medicine, Sacramento, CA; and Department of Emergency Medicine, Alameda County Medical Center, Highland Campus, Oakland, CA (LR). Received December 20, 1999; revision received April 4, 2000; accepted April 20, Presented at the SAEM annual meeting, Washington, DC, May Address for correspondence and reprints: Peter E. Sokolove, MD, Division of Emergency Medicine, UC Davis Medical Center, 2315 Stockton Boulevard, PSSB 2100, Sacramento, CA Fax: ; pesokolove@ucdavis.edu The recent resurgence of tuberculosis (TB) in the United States has led to a number of nosocomial outbreaks of this disease. 1 Emergency department (ED) personnel are at increased risk of contracting TB because patients often present to the ED with undiagnosed disease. Purified protein derivative (PPD) skin test conversion rates for ED personnel have been reported to be as high as 30% overall, with risk of conversion increasing as a function of time of employment. 2 It has previously been demonstrated that there is frequently a delay in the identification and respiratory isolation of patients with pulmonary TB. 3,4 One possible explanation for this is that the clinical presentation of patients in the ED may be atypical. This can be particularly true for HIV-infected patients, in whom clinical symptoms may be confused with Pneumocystis carinii pneumonia (PCP), sputum stains and cultures may be less sensitive, and radiographic findings are frequently atypical Even immunocompetent patients may present to the ED with nonspecific symptoms or for illness or injury that is coincidental to having active pulmonary TB. In order to more readily identify patients who may be at risk for pulmonary TB, emergency physicians must be familiar with the clinical presentation of this disease. Understanding patient presentation is also an essential step in the development of ED triage screening protocols as recommended by the Centers for Disease Control and Prevention. 12 One such protocol was only moderately sensitive (63%) for isolating patients with pulmonary TB at ED triage. 13 This study was performed to determine the ED presentation of patients with active pulmonary TB. METHODS Study Design. We reviewed the TB surveillance records of the Division of Epidemiology and Infection Control in order to identify adult patients diagnosed as having active pulmonary TB by sputum culture at our institution between January 1994 and June We then conducted a retrospective medical record review to select patients who presented to the ED for care from one year before to one year after their initial positive sputum culture. This study was deemed exempt from informed consent by the Human Subjects Review Committee at the UC Davis Medical Center. Study Setting and Population. The UC Davis Medical Center is an urban teaching hospital in Sacramento County with an ED that has an annual census of approximately 65,000 patients. Our institution is the de facto public hospital in the Sacramento geographic area, which serves a population of 400,000

2 ACADEMIC EMERGENCY MEDICINE September 2000, Volume 7, Number within city limits and about 1.5 million in the surrounding area. Approximately 60% of ED patients are medically indigent or insured through government programs. In 1995, the Sacramento metropolitan area had a tuberculosis case rate of 10.0 cases per 100,000 population. 14 In order to select patients with active, potentially contagious pulmonary TB, each patient visit was reviewed and categorized as either a contagious (capable of disease transmission) or noncontagious visit. A patient was considered to be contagious if any of the following criteria were met: 1) positive sputum culture for Mycobacterium tuberculosis during hospitalization, 2) any visit subsequent to the positive culture but prior to starting antimycobacterial therapy, 3) any visit prior to a positive sputum culture when the chest radiograph (CXR) was suggestive of pulmonary TB, or 4) any visit within one month prior to the initial positive sputum culture if a CXR was not taken. Using these criteria, patient visits were categorized by the consensus of the three authors, which included an infectious diseases specialist (SHC). Study Protocol. We abstracted the medical records of patients having contagious visits to the ED to determine demographic information, chief complaints, presence of TB risk factors and symptoms, physical examination and CXR findings, antibiotic resistance patterns of M. tuberculosis isolates, and whether the patient was placed in respiratory isolation. Chief complaints were taken from the triage nursing note and were categorized as either pulmonary (e.g., dyspnea, cough, hemoptysis), infectious but nonpulmonary (e.g., fever), medical but nonpulmonary (e.g., chest pain, abdominal pain, vomiting), or traumatic/orthopedic (e.g., extremity pain, back pain, motor vehicle collision). Chief complaints were categorized as pulmonary if any pulmonary symptoms were noted at triage, even if other complaints were also noted. We reviewed the triage nursing note, ED report, and initial admitting team note in order to determine the presence of TB risk factors and symptoms, as well as physical exam findings. For the risk factors of homelessness and foreign birth, we also reviewed the initial ED registration sheet to determine current address and place of birth. Risk factors, symptoms, and exam findings were considered to be present only if documented in the medical record. Cough was considered to be present if patients reported a cough of any duration. When TB risk factors or symptoms were present, we recorded the location where these positive findings were first noted (e.g., at triage, in the ED, or at the time of hospital admission). Chest x-ray readings were taken from the dictated attending radiologist note. For data analysis, CXR readings were recorded as upper lobe infiltrate, diffuse interstitial infiltrate, other infiltrate, mediastinal lymphadenopathy, cavitary lesion, mass or coin lesion (not cavitary), pleural effusion, other finding (e.g., rib fracture, hyperinflation), or normal. These categories of CXR readings were taken from a prior investigation of a TB triage screening procedure. 14 All positive findings were recorded for each CXR. Medical record review was performed by a single author (LR) using a standardized data collection form. Data Analysis. Descriptive statistical analysis was performed using Microsoft Excel version 4.0 (Microsoft Corporation, Redmond, WA). The frequency of patient chief complaints, TB risk factors, TB symptoms, and physical exam findings were determined. Corresponding 95% confidence intervals (CIs) were calculated using Stata 5 (Stata Corporation, College Station, TX). RESULTS There were 44 patients with active pulmonary TB treated in the ED during the study period. These 44 patients made a total of 134 visits within one year of their initial positive sputum culture (median 1, IQR 1:3, range 1 30). Sixty-six (49%) of these visits were categorized as contagious. The number of contagious visits per patient ranged from 1 to 7 (median 1). Of the 44 patients, 12 (27%, 95% CI = 15% to 43%) made two or more and 3 patients (7%, 95% CI=1to19%) made three or more contagious ED visits. Patient visits to the ED while having active pulmonary TB had the following characteristics. Patients had a mean SD age of years, and 86% were male. The distribution of patient ethnicity was white 42%, Asian 24%, Hispanic 20%, African American 8%, Native American 3%, and other ethnicity 3%. Chief complaints recorded at triage were categorized as pulmonary 33% (95% CI = 22% to 46%), medical but nonpulmonary 41% (95% CI = 29% to 54%), infectious but nonpulmonary 14% (95% CI = 6% to 24%), and traumatic/orthopedic 12% (95% CI = 5% to 22%). Patients were found to have a variety of chief complaints (Table 1). The frequency of various risk factors, symptoms, and physical exam findings detected in patients with active pulmonary TB are shown in Table 2. These values reflect documentation in either the triage nursing note, the ED record, or the initial admission note. Treating health care providers identified at least one TB risk factor in 57 (86%; 95% CI = 76% to 94%) patient visits, and in 51 (77%; 95% CI = 65% to 87%) visits, the patient had at least one symptom of TB. For three patient visits (5%; 95% CI = 3% to 17%) no risk factors or symptoms were identified. Of the 89 total risk factors that could be identified from the medical records, 66% were initially documented at triage, 24% in the ED, and 10% on admission. Foreign birth and homelessness were always documented at triage, while recent PPD positive results, immunosuppressive medications, and TB exposure were never documented at triage. The proportions of other risk factors first noted at triage were HIV positivity (82%), history of TB (47%), and intravenous drug use (17%). When not documented at triage, TB exposure and HIV positivity were always documented in the ED. Recent incarceration was only first documented on the admission note. The proportions of other risk factors first documented on the admission note were immunosuppressive medications (67%), recent PPD positivity (40%), intravenous drug use (17%), and history of TB (13%). Of the 191 total symptoms that could be identified from the medical

3 1058 TUBERCULOSIS Sokolove et al. TUBERCULOSIS PRESENTATION TABLE 1. Patient Chief Complaints during 66 Contagious Tuberculosis (TB) Visits to the Emergency Department Chief Complaint Number of Visits Frequency (95% CI) Pulmonary 22 33% (22%, 46%) Dyspnea 9 14% (6%, 24%) Cough (any duration) 7 11% (4%, 21%) Hemoptysis 3 5% (1%, 13%) Abnormal chest radiograph 1 1.5% (0%, 8%) Referred for pulmonary TB 1 1.5% (0%, 8%) Spiders in lung 1 1.5% (0%, 8%) Medical nonpulmonary 27 41% (29%, 54%) Abdominal pain 6 9% (3%, 19%) Chest pain 3 5% (1%, 13%) Seizures 3 5% (1%, 13%) Weak 3 5% (1%, 13%) Hematemesis 2 3% (0.4%, 11%) Syncope 2 3% (0.4%, 11%) Found down 2 3% (0.4%, 11%) Dizzy 1 1.5% (0%, 8%) Groin pain 1 1.5% (0%, 8%) High home fingerstick glucose result 1 1.5% (0%, 8%) Losing voice 1 1.5% (0%, 8%) Neck mass 1 1.5% (0%, 8%) Tremor 1 1.5% (0%, 8%) Infectious nonpulmonary 9 14% (6%, 24%) Fever 4 6% (2%, 15%) Sore throat 3 5% (1%, 13%) Infected hand 1 1.5% (0%, 8%) Thigh abscess 1 1.5% (0%, 8%) Traumatic/orthopedic 8 12% (5%, 22%) Back pain 3 5% (1%, 13%) Gunshot wound to thigh 1 1.5% (0%, 8%) Hip pain 1 1.5% (0%, 8%) Multiple burns 1 1.5% (0%, 8%) Motor vehicle collision 1 1.5% (0%, 8%) Shoulder pain 1 1.5% (0%, 8%) records, 24% were initially documented at triage, 47% in the ED, and 29% on admission. The proportions of symptoms first noted at triage were hemoptysis (80%), chest pain (56%), cough (38%), dyspnea (33%), fever (19%), and weight loss (5%). Chills, night sweats, and malaise were never documented at triage. When not documented at triage, hemoptysis was always documented in the ED. The proportions of other symptoms first documented on the admission note were night sweats (80%), malaise (56%), chills (52%), weight loss (40%), fever (26%), dyspnea (20%), and chest pain (6%). Fifty-two patients (79%) had a CXR taken during their ED visits, all of which had at least one positive finding. The distribution of CXR findings was upper lobe infiltrate (45%), diffuse interstitial infiltrate (22%), other infiltrate (29%), cavitary lesion (16%), mass or coin lesion (15%), pleural effusion (13%), other findings (29%), and mediastinal lymphadenopathy (2%). Thirteen CXRs (25%; 95% CI = 14% to 39%) were atypical for pulmonary TB (defined as absence of either an upper lobe infiltrate, a diffuse interstitial infiltrate, or a cavitary lesion). These CXRs had the following findings: pleural effusion (4), upper lobe nodules (3), middle or lower lobe infiltrates (3), mediastinal lymph adenopathy only (1), pulmonary nodule (1), and degenerative skeletal changes (1). Of the 44 patients, 36 (82%) were admitted on their first ED visits and 42 (95%) were ultimately admitted during a contagious visit. During a contagious visit, 35% of the patients were placed in respiratory isolation in the ED. An additional 11% were first isolated on the ward, while 55% were never isolated. Mycobacterium tuberculosis antimicrobial resistance patterns were available for 42 of the 44 patients (95%) with active pulmonary TB, of which six isolates (14%; 95% CI = 5% to 29%) showed resistance to at least one of the five primary antimycobacterial drugs, and two isolates (5%; 95% CI = 0.6% to 16%) showed resistance to both isoniazid and streptomycin. DISCUSSION In our study population, the patients with active, potentially contagious, pulmonary TB often had multiple ED visits, with 27% having two or more and 7% having three or more visits during the study period. It is possible that certain patients at risk for contagious TB had delayed identification due to the presence of confounding comorbid disease or familiarity to the ED staff. For example, a 45-year-old male patient with frequent alcohol-related visits presented to the ED 30 times during the study period. Tuberculosis was not considered in this patient until he had made six potentially contagious visits. Another patient who had a history of alcohol abuse made seven visits during a one-year period, and all were potentially contagious and unrecognized as such. We were surprised to find a wide variety of chief complaints in these patients. Only 36% of the patients reported any pulmonary complaint at triage (cough, shortness of breath, hemoptysis), whereas most patients presented with a variety of general medical complaints (e.g., chest pain, abdominal pain, fever). The high degree of variability of chief complaints appears to be due to two factors. First, some patients with active pulmonary TB have chief complaints that are nonspecific, but probably caused by TB infection (e.g., chief complaints of chest pain, fever, weakness, or abdominal pain). Second, many ED visits by patients with active pulmonary TB are for illness or injuries that are coincidental to having TB infection (e.g., gunshot wound to thigh, seizures, multiple burns, syncope). This highlights one of the difficulties in identifying patients with active TB at triage. At

4 ACADEMIC EMERGENCY MEDICINE September 2000, Volume 7, Number some institutions, ED screening protocols for TB are applied only to patients complaining of cough. Although TB patients with an active cough are generally more contagious than those without a cough, 12 even a perfectly sensitive protocol for evaluating coughing patients would have missed nearly two-thirds of the contagious patients in this study. Risk factors for TB were present in 86% of contagious visits and symptoms in 77%. Despite this high prevalence, few TB risk factors and symptoms were reliably identified at triage, more commonly being first documented in the ED medical record or inpatient admission notes. Risk factors that, if present, were likely to be detected at triage were foreign birth, homelessness, and HIV positivity. The high rate of identification of foreign birth and homelessness was due to documentation by ED registration rather than by nursing at triage. This patient information may or may not have been available to the triage nurse. Thus, HIV positivity was the only risk factor reliably detected at triage itself. Hemoptysis and chest pain were often first detected at triage, but the remaining TB symptoms were detected at triage in fewer than 50% of patients. It appears that the identification of constitutional symptoms of TB was particularly rare at triage. Efforts to better recognize these risk factors and symptoms might improve the early identification of TB patients. Among our patients with active pulmonary TB, 80% had an abnormal physical exam. Most commonly, this involved the presence of either fever or an abnormal lung exam on auscultation. The finding of a normal physical exam (absence of fever, tachypnea, cachexia, adenopathy, or abnormal findings on lung exam) in 20% of patients illustrates that physical exam alone cannot be used to exclude active pulmonary TB. Chest radiographs were obtained in 79% of patients, and of these, a high proportion (75%) had findings typical of TB, with upper infiltrates, cavitary lesions, or diffuse interstitial patterns. Indeed, a CXR consistent with pulmonary TB may have been the initial clue that suggested the diagnosis in a number of cases. In the 21% of visits when a CXR was not obtained, the diagnosis may TABLE 2. Patient Tuberculosis (TB) Risk Factors, Symptoms, and Physical Exam Findings during 66 Contagious Visits to the Emergency Department Number of Visits Frequency (95% CI) Risk factors Foreign born 28 42% (30%, 55%) Homelessness 18 27% (17%, 40%) History of TB 15 23% (13%, 35%) HIV positivity 11 17% (9%, 28%) PPD * within 2 years 5 8% (3%, 17%) Chemotherapy or steroids 3 5% (1%, 31%) Jail within 2 years 1 1.5% (0%, 8%) TB exposure 1 1.5% (0%, 8%) None of the above 9 14% (6%, 24%) Symptoms Cough 42 64% (51%, 75%) Dyspnea 30 45% (33%, 58%) Fever 27 41% (29%, 54%) Chills 25 38% (26%, 51%) Weight loss 20 30% (20%, 43%) Chest pain 18 27% (17%, 40%) Night sweats 15 23% (13%, 35%) Malaise 9 14% (6%, 24%) Hemoptysis 5 8% (3%, 17%) None of the above 15 23% (13%, 35%) Physical exam findings Abnormal lung exam 38 58% (45%, 70%) Temperature > 100 F 16 24% (15%, 36%) Wasting 12 18% (10%, 30%) Respiratory rate > 24 breaths/min 9 14% (6%, 24%) Generalized adenopathy 4 6% (2%, 15%) None of the above 13 20% (11%, 31%) *PPD = purified protein derivative positive result. have been further delayed. Overall, during a contagious visit only 35% of the study patients were placed in respiratory isolation in the ED, which is similar to the 51% ED isolation rate reported by Moran et al. 3 While formal PPD conversion rates were not calculated, three ED personnel converted their PPD to positive during the study period. This included one physician in 1994, one clerk in 1995, and one registration worker in It is interesting to note that, despite many visits by patients with apparently unrecognized active pulmonary TB, few health care workers suffered adverse consequences. This may have been due to a number of factors, including proper ventilation in the ED, limited direct exposures to these patients, and a low degree of infectivity among some patients. LIMITATIONS AND FUTURE QUESTIONS This study has a number of limitations, many inherent in its retrospective design. We relied on medical record documentation to determine the presence of TB risk factors and symptoms, and patients may have had risk factors or symptoms that were either never assessed or never documented. This would lead us to underestimate the prevalence of these historical items. We sought to minimize this possibility by including the hospital admission history and physical note, on the presumption that it might contain a more thorough history. This usually increased the number of historical items identified, and for certain items (night sweats, malaise, recent incarceration, immunosuppressive medications), the risk factor or symptom was usually documented only on the admission note. The increase in recognition of more subtle factors on the admission note may be the result of increased interviewer sensitivity after learning the results of a CXR obtained in the ED. There is also a potential pa-

5 1060 TUBERCULOSIS Sokolove et al. TUBERCULOSIS PRESENTATION tient recall bias, as patients would be more likely to report positive responses once they are being admitted for pulmonary TB. Thus, a patient with chest pain, admitted for a suspicious cavitary lesion on CXR, might acknowledge having night sweats and recent incarceration in jail when questioned by an admitting physician concerned about TB. Similarly, it is impossible to know whether factors assessed in the ED were asked before or after the results of the CXR were known. Our study did not have a control group, so we are unable to compare the prevalences of risk factors, complaints, and physical findings in TBinfected patients with those among the entire ED population. Our objective, however, was to determine the clinical presentation of ED patients with active pulmonary TB. In addition, we deliberately chose liberal criteria for categorizing patients as contagious, and are unable to definitively say whether all patients having contagious visits could in fact transmit TB. Because one goal of this study was to identify factors that might be useful in developing screening protocols, we believed that it was important to classify cases that were questionable as contagious visits. It should be noted that the nonindependence of patient visits might have influenced our findings. Because some patients had multiple visits, their presentation had a greater influence on the described presentation of the entire population. However, we chose to analyze our data by patient visits, rather than patients, because this more accurately reflects the way patients present to the ED. We used hospital infection control records to identify patients for enrollment into the study. Because patients with less classic presentations might be less likely to be admitted and eventually diagnosed as having TB, there is a potential selection bias toward identifying patients with more typical presentations of the disease. The hospital s infection control records are based on positive TB cultures obtained primarily in the inpatient setting, and we may therefore have selected for a group of patients ill enough to require admission and inpatient treatment. Similarly, we may have missed patents who were seen in our ED while contagious, but eventually diagnosed as having TB elsewhere. All of these factors would likely result in the true presentation of ED patients with active TB being even less typical than described in our study patients. CONCLUSIONS Patients with active pulmonary TB may have multiple ED visits, and often have nonpulmonary complaints. Tuberculosis risk factors and symptoms are usually present in these patients, but often missed at ED triage. One-third of the patients in this study denied having a cough, and hemoptysis was rare. Chest radiographs usually appeared typical of TB. The diversity of clinical presentations among ED patients with pulmonary TB will likely make it difficult to develop and implement high-yield triage screening criteria. The authors thank Margaret Morita, CIC, from the Division of Epidemiology and Infection Control for her assistance with identifying study patients. References 1. Menzies D, Fanning A, Yuan L, Fitzgerald M. Tuberculosis among health care workers. N Engl J Med. 1995; 332: Sokolove PE, Mackey D, Wiles J, Lewis RJ. Exposure of emergency department personnel to tuberculosis: PPD testing during an epidemic in the community. Ann Emerg Med. 1994; 24: Moran GJ, McCabe F, Morgan MT, Talan DA. Delayed recognition and infection control for tuberculosis patients in the emergency department. Ann Emerg Med. 1995; 26: Rao VK, Iademarco EP, Fraser VJ, Kollef MH. Delays in the suspicion and treatment of tuberculosis among hospitalized patients. Ann Intern Med. 1999; 130: Pierce JR, Sims SL, Holman GH. Transmission of tuberculosis to hospital workers by a patient with AIDS. Chest. 1992; 101: Klein NC, Duncanson FP, Lenox TH 3rd, Pitta A, Cohen SC, Wormser GP. Use of mycobacterial smears in the diagnosis of pulmonary tuberculosis in AIDS/ARC patients. Chest. 1989; 95: Samb B, Sow PS, Kony S, et al. Risk factors for negative sputum acid-fast bacilli smears in pulmonary tuberculosis: results from Dakar, Senegal, a city with low HIV seroprevalence. Int J Tuberc Lung Dis. 1999; 3: Pitchenik AE, Rubinson HA. The radiographic appearance of tuberculosis in patients with the acquired immune deficiency syndrome (AIDS) and pre-aids. Am Rev Respir Dis. 1985; 131: Perlman DC, el-sadr WM, Nelson ET, et al. Variation of chest radiographic patterns in pulmonary tuberculosis by degree of human immunodeficiency virusrelated immunosuppression. Clin Infect Dis. 1997; 25: Haramati LB, Jenny-Avitall ER, Alterman DD. Effect of HIV status on chest radiographic and CT findings in patients with tuberculosis. Clin Radiol. 1997; 52(1): Asimos AW, Ehrhardt J. Radiographic presentation of pulmonary tuberculosis in severly immunosuppressed HIV-seropositive patients. Am J Emerg Med. 1996; 14: Centers for Disease Control and Prevention. Guidelines for preventing the transmission of Mycobacterium tuberculosis in health-care facilities, MMWR. 1994; 43(RR-13): Sokolove PE, Lee BS, Krawczyk JA, et al. Implementation of an emergency department triage procedure for the detection and isolation of patients with active pulmonary tuberculosis. Ann Emerg Med. 2000; 35: Centers for Disease Control and Prevention. Reported Tuberculosis in the United States, Atlanta: CDC, 1997, p 35.