a b s t r a c t

Objectives: The objective of the study is to investigate the causes of febrile illness among HIV-infected adults visiting the emergency department (ED) of a designated hospital for HIV care in Taiwan, an area of a low HIV prevalence. Methods: From January 2004 to December 2012, all febrile HIV-infected adults visiting the ED were retrospectively investigated. Recent CD4 lymphocyte counts near ED visits and HIV transmission route were designated as major predictors for the analyses. All variables and clinical information were derived from chart records.

Results: Of the 196 eligible HIV-infected adults, major causes of febrile illness were lower respiratory tract infections (68, 34.7%), skin and soft tissue infections (31, 15.8%), intra-Abdominal infections (22, 11.2%), and urinary tract in- fections (11, 5.6%). There were 150 pathogens identified. Staphylococcus aureus (51, 34.0%) and Pneumocystis jirovecii (26, 17.3%) were the major pathogens. In a multivariate analysis, Injection drug use (odds ratio, 15.18; P b

.001) and skin and soft tissue infections (odds ratio, 18.45; P = .001) were independently associated with S aureus infections, and the proportion of S aureus increased steadily with CD4 lymphocyte count (? = 0.99; P = .01). Of pneumonic patients with recognized pathogens, P jirovecii pneumonia was frequently associated with patients hav- ing a CD4 lymphocyte count of less than 100 cells/mm³ (25/25, 100% vs 16/30, 53.3%; P b .001).

Conclusions: The causes of febrile illness in HIV-infected adults visiting the ED varied according to CD4 count and transmission route. Two independent risk factors, intravenous drug use and skin and soft tissue infections, were as- sociated with S aureus infections. For HIV-infected adults with lower respiratory tract infections, a CD4 lymphocyte count of less than 100 cells/mm³ was a risk factor for P jirovecii pneumonia.

(C) 2014

1. Introduction

Fever is a common complaint of emergency department (ED) vis- itors [1]. When ED clinicians face febrile patients, they must make a tentative diagnosis and evaluate the potential risk of the presence of bacterial infectious diseases, such as bacteremia or lower respira- tory tract infections. In patients with HIV infection, the immunolog- ical deficits include functional abnormalities and the quantitative depletion of CD4 T lymphocytes as well as functional and numerical alterations of B lymphocytes, monocytes, and polymorphonuclear cells [2]. Therefore, this complex disorder of the immune system in HIV-infected patients leads not only to increased susceptibility to neoplasms but also to life-threatening opportunistic infections due to fungal, viral bacterial agents [3,4].

? Funding: This study was partially supported by grants from the Ministry of Health and Welfare Taiwan (MOHW103-TDU-B-211-113002) and, in part, supported by a research grant from the Ministry of Science and Technology Taiwan (NSC102-2314-B-006 -079).

* Corresponding author. Department of Internal Medicine, National Cheng Kung Univer- sity Hospital, No. 138, Sheng Li Road, 70403, Tainan, Taiwan. Tel.: +886 6 2353535×3596; fax: +886 6 2752038.

E-mail address: [email protected] (W.-C. Ko).

Emergency departments often serve populations that have limited access to health care [5,6]. Thus, several investigations have discussed placing the responsibility of screening for HIV and providing Preventive care on EDs [7,8]. The first case of HIV infection and AIDS in Taiwan was reported in 1984 [9]. The Ministry of Welfare and Health Taiwan has provided free antiretroviral therapy to HIV-infected persons since 1997, and currently, there are 57 hospitals assigned for antiretroviral access and health care cover- age [10]. The annual number of newly reported cases of HIV infection continued to increase, from 860 cases in 2003 to 2244 cases in 2013 [10]. Overall, there were 27 799 cases reported by the end of July 2014, with a prevalence of 0.12% among general popula- tion in Taiwan [10]. It is not surprising that this increase has led to an increasing number of ED visits among HIV-infected patients and to epidemiologic changes in the febrile population visiting the ED. However, no ED-based study emphasizing the epidemiology of febrile ED visits among HIV-infected adults has been reported in the English literature. Thus, by focusing on febrile HIV-infected adults at the ED, we aimed to determine the causes of fever as well as the relationship between specific pathogens and the route of transmission or CD4 lymphocyte count.

http://dx.doi.org/10.1016/j.ajem.2014.10.013

0735-6757/(C) 2014

1. Materials and methods
   1. Study design and population

A retrospective study was conducted during the 9-year period from January 2004 to December 2012 at the ED of a medical center in south- ern Taiwan. The study was approved by the hospital institutional review board, and the informed consent requirement was waived. During the study period, there were a total of 405 565 ED visits, with a mean annual visit of 67 594. The mean duration of ED stay before admission was 18.2 (+-12.7) hours.

Study protocol

During the study period, all HIV-infected patients visiting the ED were investigated. The inclusion criteria included all of the following items: an aural temperature of at least 38.0?C measured at the ED and an age of 18 years or older. The exclusion criteria were as follows: lack of CD4 count measurement within 4 weeks before or 2 weeks after the onset of febrile illness, no identified transmission route of HIV, health care-acquired infection, contaminated cultures or lack of blood cultures taken at the ED, or no microbiological cultures of indicated specimens for infectious focus (such as no urine culture for urinary tract infection).

For the eligible adults, medical records were collected for clinical characteristics, vital signs, severity scores (eg, Simplified Acute Physiol- ogy Score II [SAPS II]), comorbidities, and initial syndrome (eg, sepsis and severe sepsis) immediately after each patient’s visit to the ED. At that time, a prior history of Highly active antiretroviral therapy (HAART), acquisition route of HIV, antimicrobial therapy and duration, microbiological results, concurrent bacteremia, infection source, further hospitalization, and the length of hospital stay were derived from the

chart records. In addition to blood cultures, microbiological cultures of the indicated specimen or other diagnostic tests, such as serological tests, imaging studies, or invasive procedures, were performed as clini- cally indicated. Their medical records were reviewed for the above clin- ical information by 2 of the authors, an ED clinician and an internist. If any discrepancies were found, the medical records would be inspected by the 2 authors together. If a patient had more than 1 infectious site of bacteremia, each site would be assigned to the infection focus. Thus, this patient may have more than 1 infection foci. Multiple bacteremic episodes in a single patient were considered distinct events, if separated by at least 14 days or different causes were listed for the respective ED visits [11].

Microbiologic studies and diagnosis

For bacteria, Biochemical tests and automatic identification systems were used for species identification. bacterial pathogens were tested for in vitro susceptibility by the Kirby-Bauer method on Mueller-Hinton agar. Antimicrobial susceptibility results were interpreted according to the updated Clinical Laboratory Standard Institute guidelines issued in 2014 [12]. Intracellular pathogens (eg, Mycoplasma pneumoniae and Legionella pneumoniae) and spirochetes (eg, Treponema pallidum) were detected by antigen, immunoglobulin M (IgM) antibody, or poly- merase chain reaction (PCR). Influenza viral infection was detected by viral cultures or reverse transcription-PCR from nasopharyngeal sam- ples [13]. Infections caused by hepatitis A virus, hepatitis B virus (HBV), cytomegalovirus, Epstein-Barr virus, or dengue virus were diag- nosed by the presence of virus specific IgM detected by captured enzyme-linked immunosorbent assay (ELISA) or virus-specific DNA or RNA by real-time PCR in blood samples. Invasive Fungal infections were diagnosed by either a positive blood culture or histologic evidence of tissue invasion on biopsy. Pneumocystis jirovecii in respiratory tract

Fig. 1. The clinical diagnosis of infectious diseases (A) and causative microorganisms (B) among HIV-infected adults with community-onset febrile illness.

Table 1

A trend of the clinical characteristics of community-onset febrile illness among 196 HIV-infected patients at the ED with regard to increasing CD4 lymphocyte count, calculated by Pearson correlation coefficient

Characteristics Category of CD4 count (cell/mm³), case number (%) P

	b 100, n = 77	100-300, n = 42	301-500, n = 39	N 500, n = 38
Male	75 (97.4)	38 (90.5)	35 (89.7)	32 (84.2)	.04
Age, mean +- SD	36.1 +- 9.5	41.2 +- 13.7	41.9 +- 13.2	44.3 +- 11.2	.05
Concurrent bloodstream infection	17 (22.1)	12 (28.6)	12 (30.8)	16 (42.1)	.04
Initial presentation at ED
Sepsis	32 (41.6)	29 (69.0)	19 (48.7)	20 (52.6)	.86
Severe sepsis	11 (14.3)	3 (7.1)	1 (2.6)	3 (7.9)	.37
Severity of illness at ED
SAPS II, mean +- SD	36.1 +- 9.5	41.2 +- 13.7	18.2 +- 10.8	19.3 +- 7.8	.19
Admission through ED
General ward	54 (70.1)	35 (83.3)	25 (64.1)	26 (68.4)	.62
Intensive care unit	8 (10.4)	1 (2.4)	2 (5.1)	0 (0)	.18
Length of stay (d), mean +- SD Total hospitalization	13.7 +- 14.1	11.5 +- 14.5	11.5 +- 15.5	8.7 +- 10.8	.34
Intensive care unit	1.5 +- 5.4	0.02 +- 0.15	0.6 +- 3.1	0 +- 0	.10
Appropriate antibiotics at ED a	30/68 (44.1)	16/28 (55.2)	20/27 (74.1)	18/24 (75.0)	.04
HAART naive	40 (51.9)	16 (38.1)	26 (66.7)	23 (60.5)	.06
Transmission risk group
MSM	17 (22.1)	9 (21.4)	4 (10.3)	1 (2.6)	.04
IDUs	8 (10.4)	14 (33.3)	22 (56.4)	21 (55.3)	.06
Bisexuals	1 (1.3)	1 (2.4)	0 (0)	2 (5.3)	.45
Comorbidities
Chronic hepatitis	17 (22.1)	17 (40.5)	19 (48.7)	21 (55.3)	.03
Liver cirrhosis	4 (5.2)	5 (11.9)	3 (7.7)	5 (13.2)	.31
Malignancy	3 (3.9)	4 (9.5)	1 (2.6)	1 (2.6)	.57
Diabetes mellitus	2 (2.6)	2 (4.8)	2 (5.1)	6 (15.8)	.13
Hypertension	1 (1.3)	3 (7.1)	4 (10.3)	11 (28.9)	.07
Heart failure	1 (1.3)	2 (4.8)	0 (0)	0 (0)	.50
Old stroke	1 (1.3)	1 (2.4)	1 (2.6)	0 (0)	.60
Coronary artery disease	0 (0)	1 (2.4)	0 (0)	1 (2.6)	.52
Chronic renal insufficiency	0 (0)	0 (0)	2 (5.1)	2 (5.3)	.10
Causes of fever Source of infection Lower respiratory tract	45 (58.4)	10 (23.8)	9 (23.1)	4 (10.5)	.09
Primary bacteremia	8 (10.4)	0 (0)	1 (2.6)	1 (2.6)	.40
Skin and soft tissue	7 (9.1)	3 (7.1)	13 (33.3)	8 (21.1)	.34
Central nerve system	4 (5.2)	0 (0)	1 (2.6)	0 (0)	.33
Intra-abdomen cavity b	3 (3.9)	14 (33.3)	3 (7.7)	2 (5.3)	.80
Bone and joint	2 (2.6)	3 (7.1)	3 (7.7)	5 (13.2)	.04
Vessel and heart	1 (1.3)	3 (7.1)	2 (5.1)	4 (10.5)	.14
Upper respiratory tract	1 (1.3)	1 (2.4)	2 (5.1)	5 (13.2)	.79
Urinary tract	0 (0)	4 (9.5)	3 (7.7)	4 (10.5)	.19
Others	2 (2.6)	2 (4.8)	2 (5.1)	1 (2.6)	.45
Noninfectious illness	4 (5.2)	2 (4.8)	0 (0)	4 (10.5)	.66
28-d mortality	7 (9.1)	3 (7.1)	1 (2.6)	2 (5.3)	.47

^a Thirty-eight patients without identified pathogens, 1 with dengue fever, and 10 with noninfectious illness were excluded.

^b Including Liver abscess and biliary tract infection.

samples (eg, sputum or bronchoscopic lavage) was detected by PCR. Toxoplasma gondii was detected by PCR in tissue samples or serology (the presence of IgM detected by ELISA). Entamoeba histolytica was di- agnosed serologically with IgM detected by ELISA.

The sources of infections were referred to as lower respiratory tract infection, urinary tract infection, Wound infection, soft tissue infection, intra-abdominal infection, or primary bloodstream infection based on the definitions proposed by the Centers for Disease Control and Preven- tion [14]. Liver abscess, cholangitis, and complicated intra-abdominal infection were grouped as intra-abdominal infections. P jirovecii pneu- monia (PJP) was diagnosed by clinical symptoms and signs as well as imaging findings with either P jirovecii identification or improvement by anti-PJP therapy. Cytomegalovirus disease was diagnosed by symp- toms and signs as well as by the coexistence of antigenemia or evidence of tissue invasion on biopsy, as previously reported [15]. Syphilis was di- agnosed by symptoms and signs as well as a positive serologic test with rapid plasma reagin (RPR Card Test; Becton-Dickinson, MD) and posi- tive results for fluorescent treponemal antibody absorption tests (SERODIA-TPHA; Fujirebio, Taoyuan, Taiwan). Toxoplasmosis was

diagnosed by symptoms and signs as well as by either the coexistence of antibody by ELISA or evidence of tissue invasion on biopsy.

Definitions

A health care-acquired infection (or bacteremia) is referred to as an infection (or bacteremic episode) that occurs in a patient who is re- ferred from another hospital or health care facility and has been hospi- talized for at least 48 hours or a patient who is discharged from any hospital less than 48 hours before a positive culture is obtained at the ED [16]. Other remaining infections (or bacteremic episodes) are re- ferred to as community-onset infections (bacteremia). Blood culture samples with potential contaminating pathogens (ie, coagulase- negative Staphylococcus, Micrococcus, Propionibacterium acnes, Peptostreptococcus, or Bacillus species), in any blood culture bottle, were regarded as contaminated according to previously described criteria [17]. The contamination of other specimens was determined in agreement by 2 of the authors. Appropriate antibiotic(s) prescribed at the ED was defined as the empirical drug(s) to which the isolated

Fig. 2. The trends of major 5 pathogens among 148 HIV-infected adults with community-onset febrile illness, with regard to an increasing CD4 count. P values were calculated by the

Pearson correlation.

pathogen was in vitro susceptible, and inappropriate antibiotic(s) to which the isolated pathogen was nonsusceptible in vitro. If in vitro sus- ceptibility was not available for specific pathogens, such as P jirovecii or virus, targeted antimicrobial therapy was considered to be appropriate provided that the dosage and route administration were in accordance with recommended therapy of the Sanford guideline [18].

Malignancy included both Hematological malignancies and solid tu- mors, and the definitions of other comorbidities were as previously de- scribed [19]. The severity of febrile illness was graded by the SAPS II [20].

Data analysis

Statistical analyses were performed using the SPSS for Windows (SPSS, Chicago, IL), version 18.0. Continuous variables are expressed as the means +- SDs and compared using the Student t test. Categorical variables, expressed as numbers and percentages, were compared by the ?² or Fisher exact test. Based on a previous report of HIV- seronegative injection Drug users (IDUs), Staphylococcus aureus infections accounted for 18% to 29% of soft tissue infection, 61% of endo- carditis, and 57% of bacteremia [21]. In addition, according to a previous study discussed with the characteristics of bacteremia in overall ED visits, S aureus accounted for approximately 10% of bacteremic patho- gens [11]. Thus, if a half of ED visits among HIV-infected individuals were IDUs, by the G*Power software, at least 124 patients would be in- cluded to disclose the significance of S aureus infections among IDUs and non-IDUs [22]. The variables with a P value of less than .1 by univariate analysis contributed to the stepwise, backward logistic regression model. Pearson correlation coefficient (?) was used to measure the strength of the association between 2 continuous variables. P b .05 was considered statistically significant.

1. causative pathogens”>Results
   1. Demographics and clinical characteristics of febrile HIV-infected adults

During the study period, the medical records of 245 HIV-infected pa- tients visiting the ED were inspected. After exclusion of afebrile patients,

the data of 218 febrile HIV-infected adults were collected. Eighteen patients were found to have discrepancies in their records, and their records were discussed by the 2 authors together. There were 22 patients with incomplete clinical data based on the exclusion criteria. Thus, only 196 febrile HIV-infected adults were eligible for the study. The mean age was 39.9 years, and 180 (91.8%) were males. HIV infection had been diagnosed in most (169, 86.2%) pa- tients before the ED visits. CD4 lymphocyte counts of 51 patients (26.0%) were available within 2 weeks after the ED visit, and those of 145 (74.0%) patients had been determined 4 weeks before the ED visit. The route of HIV transmission was categorized into 3 groups: IDUs (65 patients, 33.2%); men who have sex with men (MSM; 31, 15.9%), and others (non-IDU, non-MSM; 100, 51.0%).

Most patients (102 patients, 52.0%) lacked comorbidities. The co- morbidities that were present included chronic hepatitis (74 pa- tients, 37.8%), hypertension (19, 9.7%), liver cirrhosis (17, 8.7%),

diabetes mellitus (12, 6.1%), malignancy (9, 4.6%), end stage of

chronic kidney disease (4, 2.0%), heart failure (3, 1.5%), history of stroke (3, 1.5%), and coronary artery disease (2, 1.0%). Of 74 patients with chronic viral hepatitis, 34 (45.9%) had Hepatitis C virus infection; 27 (36.5%), HBV infection; and 13 (17.6%), concurrent HBV and Hepatitis C virus infection.

Most patients (140, 71.4%) were admitted to general wards; 11 (5.6%), intensive care units; and 45 (23.0%) were discharged from the ED and followed up in outpatient clinics. Their 28-day mortality rate was 6.6% (13/196). Because of 9 patients with multiple infection foci, there were a total of 205 infectious diagnosis of febrile illness. The final infectious diagnoses at discharge were shown in Fig. 1A. Of note, lower respiratory tract infection (68 patients, 33.1%) was the leading cause of febrile illness.

Distribution of causative pathogens

Of the 196 febrile HIV-infected patients, causative pathogens were identified for 148 (75.5%). Because of 2 episodes of polymicrobial bac- teremia, there were a total 150 microorganisms. Most (63, 42.9%) path- ogens were identified from blood, followed by airway secretion (44,

Table 2

Characteristics and outcome of community-onset febrile illness among 196 HIV-infected patients at the ED as categorized by potential transmission routes, calculated by ?² test

Characteristics Patient number (%) P

IDU, n= 65		MSM, n = 31	Other, n = 100	mon pathogen was S aureus (25, 42.4%), followed by Salmonella enterica (8, 13.6%), E coli (8, 13.6%), K pneumoniae (6, 10.2%), Pseudomonas
Male	63 (96.9)	31 (100)	86 (86.0)	.008 aeruginosa (2, 3.4%), Streptococcus pneumoniae (2, 3.4%), Aeromonas

nervous system infections, including 2 episodes of cerebral toxoplasma abscess, 2 of S aureus meningitis, and 1 of meningitis of unknown etiology. There were 57 episodes of Bloodstream infections, accounting for 29.6% of febrile illness. Of the 59 bacteremic pathogens, the most com-

Age, mean +- SD	39.2 +- 8.0	33.4 +- 8.2	42.4 +- 14.2	.001	hydrophila (2, 3.4%), Proteus mirabilis (2, 3.4%), Serratia marcescens (1,
Concurrent bloodstream infection Initial presentation at ED	28 (43.1)	6 (19.4)	23 (23.0)	.009	1.7%), E faecalis (1, 1.7%), C indologenes (1, 1.7%), and Penicillium marneffei (1, 1.7%).
Sepsis 41 (63.1)		11 (35.5)	48 (48.0)	.03
Severe sepsis 8 (12.3)		3 (9.7)	7 (7.0)	.51 3.3. A correlation of clinical characteristics or microorganisms and CD4
Severity of illness at ED					lymphocyte count
SAPS II, mean +- SD 23.3 +- 13.0		26.7 +- 9.7	32.2 +- 14.7	b.001
Admission through ED					All the febrile HIV-infected adults were categorized by CD4 lympho-

cyte count, that is, less than 100, 100 to 300, 301 to 500, and greater than 500 cell/mm³. A correlation of CD4 lymphocyte count and the pro- portion of the following variables was analyzed (Table 1): demographic data, initial presentations and severity at the ED, length of hospital stay, appropriateness of initial antimicrobial therapy, comorbidities, prior HAART, transmission risk group, microbiological etiologies of fever, and 28-day mortality. The proportion of concurrent bacteremia (? = 0.96, P = .04), appropriate antibiotic therapy at the ED (? = 0.95, P =

General ward 49 (75.4)		23 (74.2)	86 (86.0)	.55
Intensive care unit 3 (4.6)		2 (6.5)	6 (6.0)	.91
Length of stay (d), mean +- SD Total hospitalization	14.9 +- 15.9	11.7 +- 14.2	9.9 +- 12.2	.09
Intensive care unit	0.4 +- 2.4	1.0 +- 4.0	0.8 +- 4.2	.65
Appropriate antibiotics at ED a	38/52 (71.7)	12/21 (57.1)	34/73 (46.6)	.06
HAART naive Category of CD4 count (cell/mm3)	50 (76.9)	20 (64.5)	35 (35.0)	b.001 b.001
b100	8 (12.3)	17 (54.8)	52 (52.0)
100-300	14 (21.5)	9 (29.0)	19 (19.0)
300-500	22 (33.8)	4 (12.9)	13 (13.0)
N 500	21 (32.3)	1 (3.2)	16 (16.0)
Comorbidities Chronic hepatitis	47 (72.3)	3 (9.7)	24 (24.0)	b.001
Liver cirrhosis	14 (21.5)	0 (0)	3 (3.0)	b.001
Hypertension	6 (9.2)	0 (0)	13 (13.0)	.10
Malignancy	4 (6.2)	1 (3.2)	4 (4.0)	.75
Diabetes mellitus	2 (3.1)	0 (0)	10 (10.0)	.06
Heart failure	1 (1.5)	1 (3.2)	1 (1.0)	.68
Chronic renal insufficiency	0 (0)	0 (0)	4 (4.0)	.14
Old stroke	0 (0)	0 (0)	3 (3.0)	.23
Coronary artery disease	0 (0)	0 (0)	2 (2.0)	.38
Cause of fever Source of infection Skin and soft tissue	16 (24.6)	3 (9.7)	12 (12.0)	.06
Lower respiratory tract	12 (18.5)	12 (38.7)	44 (44.0)	.003
Bone and joint	11 (16.9)	1 (3.2)	1 (1.0)	b.001
Intra-abdomen b	9 (13.8)	7 (22.6)	6 (6.0)	.03
Vessel and heart	7 (10.8)	0 (0)	3 (3.0)	.03
Primary bacteremia	3 (4.6)	2 (6.5)	5 (5.0)	.93
Upper respiratory tract	3 (4.6)	0 (0)	6 (6.0)	.38
Urinary tract	2 (3.1)	1 (3.2)	8 (8.0)	.33
Central nerve system	0 (0)	2 (6.5)	3 (3.0)	.16
Others	1 (1.5)	0 (0)	6 (6.0)	.16
Noninfectious diseases	1 (1.5)	3 (9.7)	6 (6.0)	.20
28-d mortality	4 (6.2)	0 (0)	9 (9.0)	.21

.047), MSM (? = -0.96, P = .04), comorbidity with chronic hepatitis (? = 0.97, P = .03), and fever due to bone and joint infection (? = 0.96, P = .04) increased steadily with the CD4 count, whereas that of male sex (? = -0.96, P = .04) decreased with the CD4 count. The cor- relation of the CD4 lymphocyte count and the major 5 infection-causing microorganisms was determined, as shown in Fig. 2. A positive correla- tion of the proportion of S aureus and the CD4 count was significantly demonstrated (? = 0.99, P = .01).

Clinical characteristics and pathogen distribution among various categories of HIV transmission risk

^a Thirty-eight patients with unknown pathogen, 1 with dengue, and 10 with noninfectious illness were excluded.

^b Including liver abscess and biliary tract infection.

29.3%), wound discharge (21, 14.0%), tissue aspiration (13, 8.7%), stool (3, 2.0%), urine (3, 2.0%), and throat swab (1, 0.7%). Of 150 microorgan- isms, their distribution was shown in Fig. 1B. S aureus was the most common pathogen (51, 34.0%). Of the 24 isolates of non-Salmonella Enterobacteriaceae, Escherichia coli and K pneumoniae were the most common, accounting for 45.8% (11 isolates) and 33.3% (8), respectively. Rare causative microorganisms noted in each patient included T pallidum (CD4 count, 333 cells/mm³; transmission route, others), Clostridium Difficile (276 cells/mm³; others), Candida albicans (20 cells/mm³; others), Chryseobacterium indologenes (345 cells/mm³; others), Enterococcus faecalis (290 cells/mm³; others), M pneumoniae (231 cells/mm³; others), dengue virus (245 cells/mm³; others), cytomeg- alovirus (25 cells/mm³; others), Epstein-Barr virus (376 cells/mm³; others), HBV (27 cells/mm³; others), and influenza A (440 cells/mm³; MSM). Notably, methicillin-resistant S aureus (MRSA) only accounted for 3.9% (2) of 51 S aureus isolates. There were 5 episodes of central

The enrollees were categorized into 3 groups: IDUs, MSM, and others. Demographic data, initial presentation and severity at the ED, length of hospital stay, appropriate antimicrobial therapy at the ED, co- morbidities, prior HAART, various categories of CD4 lymphocyte count, cause of fever, and 28-day mortality are compared in Table 2. In the uni- variate analysis, only male sex; age; concurrent bloodstream infection; initial presentation with sepsis; SAPS II at the ED; prior non-HAART treatment; various category of CD4 lymphocyte count; no comorbidity; comorbidities with chronic hepatitis or liver cirrhosis; and fever due to lower airway infections, bone and joint infections, intra-abdominal in- fections, or vascular and cardiac infections were different among the transmission groups.

Of the total 150 infection-causing microorganisms, the distribution of the major 5 pathogens was analyzed in the transmission groups, as shown in Fig. 3. In the univariate analysis, the proportions of P jirovecii, S aureus, and S enterica were different. In further analyses, a higher pro- portion of S aureus infection (37/53, 69.8% vs 14/97, 14.4%; P b .001) and lower proportion of P jirovecii (1/53, 1.9% vs 25/97, 25.8%; P b .001) and S enterica (1/53, 1.9% vs 10/97, 10.3%; P = .10) infection were found in the patients categorized as IDU.

Risk factors associated with inappropriate antibiotic therapy at the ED

With the exclusion of 38 patients without identified pathogens, 1 with dengue fever, and 10 with noninfectious illness, the risk factors as- sociated with inappropriate antibiotic therapy at the ED were examined in Table 3. Several variables, including concurrent bloodstream infec- tion, CD4 lymphocyte less than 100 cells/mm³, lower respiratory tract infections, and P jirovecii or Mycobacterium infections, were significantly associated with inappropriate antibiotic therapy at the ED. In contrast, prior IDU, urinary tract infection, and S aureus or non-Salmonella

Fig. 3. The distribution of transmission risk among 148 HIV-infected adults with community-onset febrile illness, as categorized by major 5 pathogens, after the exclusion of 39 patients without recognized pathogens and 10 with noninfectious illness. Asterisk indicates P b .05, calculated by the ?² test.

Enterobacteriaceae infections were significantly associated with appro- priate antibiotic therapy. However, a lower respiratory infection was the only variable associated with inappropriate therapy by a multivari- ate analysis (odds ratio, 4.0; 95% confidence interval, 2.0-8.1; P b .001).

Risk factors associated with S aureus infections among HIV- infected patients

The association of several clinical variables and S aureus infection among the HIV-infected adults is examined in Table 4. The following were significantly positively associated with S aureus infection: trans- mission risk with IDU; CD4 lymphocyte count of 300 to 500 cell/mm³ or greater than 500 cell/mm³; comorbidity with chronic hepatitis; and fever due to skin and soft tissue infections, bone and joint infections, and vascular and cardiac infections. In contrast, SAPS II at the ED, trans- mission risk with MSM, CD4 lymphocyte count less than 100 cell/mm³, and fever due to lower airway infections and intra-abdominal infections were significantly negatively associated with S aureus infection. Only the following significant variables were identified by the multivariate analysis, as shown in Table 5: transmission risk with IDU, skin and soft tissue infections, lower airway infections, and intra-abdominal infec- tions. Of note, a strong relationship between IDU and S aureus infection was discovered.

Risk factors associated with PJP among HIV-infected patients with lower respiratory tract infections

Of the 55 pneumonic patients with recognized etiologies, the most common pathogen was P jirovecii (25, 45.5%), followed by Mycobacteri- um species (8, 14.5%), S pneumoniae (6, 10.9%), S aureus (3, 5.5%), non-

Salmonella Enterobacteriaceae (3, 5.5%), S enterica (3, 5.5%), and P marneffei (2, 3.6%). The association of several variables and PJP among the HIV-infected patients with lower respiratory tract infections was examined in a univariate analysis. P jirovecii pneumonia was also more frequently associated with patients with CD4 lymphocyte less than 100 cell/mm³ (25/25, 100% vs 16/30, 53.3%; P b .001) and less

frequently associated with patients with chronic hepatitis (2/25, 8.0% vs 12/30, 40.0%; P = .007). However, no significant variable was discov- ered in the multivariate analysis.

1. Discussion

In the present study, a decreasing trend of transmission risk with MSM as well as a increasing trend of comorbidity with chronic hepatitis, incidence of S aureus infection, and inappropriate antibiotic therapy at the ED was accompanied by an increased CD4 lymphocyte count. In ad- dition, a higher proportion of S aureus infections and a lower proportion of P jirovecii infections were discovered in patients with IDU. Among the febrile HIV-infected patients, 2 independent variables (eg, transmission risk with IDU and skin and soft tissue infections) were strongly associat- ed with S aureus infections, whereas 2 independent factors (eg, lower airway infections and intra-abdominal infections) were negatively asso- ciated with S aureus infections. Moreover, among the HIV-infected adults with lower respiratory tract infections, CD4 lymphocyte less than 100 cell/mm³ was a risk factor of PJP.

Fever is a subject that has been extensively discussed with regard to

etiology in the medical literature. The study populations in previous ar- ticles have varied greatly, from the general population to specific patient groups, including the elderly [23], the critically ill [24], and immuno- compromised individuals [25,26]. Emergency department overcrowd- ing is currently a common issue in many hospitals in the United States and Taiwan, and it threatens both patient safety and public health [27,28]. In an effort to assist ED clinicians in handling the growing HIV-infected population, managing potential infection, and further avoiding “over-anxiety” while managing HIV-infected patients, the present study is the first ED-based investigation focusing on the distri- bution of causes of fever in febrile HIV-infected patients visiting the ED. Of importance with regard to assisting the ED clinician, the CD4 lym- phocyte count and transmission risk of HIV were major determinants applied in this study to predict the key pathogens (eg, S aureus and P jirovecii) of febrile HIV-infected patients visiting the ED.

Table 3

Univariate analyses of risk factors of inappropriate antibiotic therapy at the ED among community-onset febrile illness in HIV-infected adults, calculated by ?² test

Table 4

Univariate analyses of risk factors of S aureus infections among community-onset febrile illness in HIV-infected adults at the ED, calculated by ?² test

Characteristics	Antibiotic therapy at the ED, n (%)		Odds ratio (95% CI)	P		Characteristics	S aureus infection, n (%)		Odds ratio (95% confidence intervals)	P
	Inappropriate,	Appropriate,					Yes,	No,
	n= 63	n= 84					n= 51	n= 97
Age >=50 y	10 (15.9)	14 (16.7)	0.94 (0.40-2.29)	.90		Age >=50 y	9 (17.6)	15 (15.5)	1.17 (0.47-2.90)	.73
Male	58 (92.1)	76 (90.5)	1.22 (0.38-3.92)	.74		Male	87 (89.7)	48 (94.1)	1.84 (0.48-7.01)	.54
Severity of illness at ED						Severity of illness at ED
Sepsis	33 (52.4)	37 (44.0)	1.39 (0.73-2.69)	.32		Sepsis	30 (58.8)	58 (59.8)	0.96 (0.48-1.92)	.91
Severe sepsis	34 (54.0)	53 (63.1)	0.69 (0.35-1.33)	.27		Severe sepsis	7 (13.7)	10 (10.3)	1.38 (0.49-3.88)	.54
SAPS II N 30 points	5 (7.9)	12 (14.3)	0.52 (0.17-1.55)	.23		SAPS II N 30 points	11 (21.6)	59 (60.8)	0.18 (0.08-0.39)	b.001
Concurrent bloodstream 10 (15.9)		46 (54.8)	0.16 (0.07-0.34)	b.001 Transmission risk group
infection					IDUs		37 (72.5)	16 (16.5)	13.38 (5.92-30.25)	b.001
Transmission risk group					MSM		3 (5.9)	18 (18.6)	0.27 (0.08-0.98)	.04
IDUs	15 (23.8)	38 (45.2)	0.37 (0.18-0.78)	.007	HAART naive		16 (31.4)	44 (45.4)	0.55 (0.27-1.12)	.10
MSM	9 (14.3)	12 (14.3)	1.00 (0.39-2.54)	1.00	Category of CD4 count (cell/mm3)
HAART naive	40 (63.5)	48 (57.1)	1.17 (0.78-1.77)	.44	b100		12 (23.5)	56 (57.7)	0.23 (0.11-0.48)	b.001
CD4 count b100 cells/mm3a	38 (60.3)	30 (35.7)	2.73 (1.40-5.37)	.003	100-300		9 (17.6)	20 (20.6)	0.83 (0.35-1.97)	.67
Lack of prior CD4 counts	20 (31.7)	31 (36.7)	0.80 (0.40-1.59)	.51	301-500		15 (29.4)	12 (12.4)	2.95 (1.26-6.93)	.01
Major comorbidities					N 500		15 (29.4)	9 (9.3)	4.07 (1.64-10.15)	.02
Chronic hepatitis	24 (38.1)	34 (40.5)	0.91 (0.46-1.77)	.77		Major comorbidities
Hypertension	5 (7.9)	8 (9.5)	0.82 (0.25-2.64)	.74	Chronic hepatitis		33 (64.7)	25 (25.8)	5.28 (2.54-10.99)	b.001
Liver cirrhosis	4 (6.3)	10 (11.9)	0.50 (0.15-1.68)	.25	Hypertension		8 (15.7)	5 (5.2)	3.42 (1.06-11.08)	.06
Diabetes mellitus	2 (3.2)	7 (8.3)	0.36 (0.07-1.80)	.30	Liver cirrhosis		4 (7.8)	10 (10.3)	0.74 (0.22-2.49)	.77
Major sources of					Diabetes mellitus		4 (7.8)	5 (5.2)	1.57 (0.40-6.11)	.50
infection					Major source of infection
Lower respiratory tract	35 (55.6)	20 (23.8)	4.00 (1.97-8.11)	b.001	Skin and soft tissue		20 (39.2)	3 (3.1)	20.22 (5.62-72.68)	b.001
Skin and soft tissue	9 (14.3)	14 (16.7)	0.83 (0.34-2.07)	.69	Bone and joint		11 (21.6)	2 (2.1)	13.06 (2.77-61.62)	b.001
Intra-abdomenb	6 (9.5)	12 (14.3)	0.63 (0.22-1.79)	.38	Vessel and heart		10 (19.6)	0 (0)	1.24 (1.09-1.42)	b.001
Bone and joint	4 (6.3)	9 (10.7)	0.57 (0.17-1.93)	.36	Lower respiratory tract		3 (5.9)	52 (53.6)	0.05 (0.02-0.19)	b.001
Urinary tract	1 (1.6)	10 (11.9)	0.12 (0.15-0.96)	.02	Urinary tract		2 (3.9)	9 (9.3)	0.40 (0.08-1.92)	.33

Microorganisms

P jirovecii	17 (27.0)	9 (10.7)	3.08 (1.27-7.48)	.01
S aureus	17 (27.0)	34 (40.5)	0.54 (0.27-1.1)	.09
MRSA	2 (3.2)	0 (0)	-	.18
Mycobacterium species	9 (14.3)	0 (0)	1.16 (1.05-1.29)	b.001
Non-Salmonella	5 (7.9)	18 (21.4)	0.32 (0.11-0.91)	.03
Enterobacteriaceae
S enterica	2 (3.2)	9 (10.7)	0.27 (0.06-1.31)	.12
Other pathogensc	4 (6.3)	5 (6.0)	1.07 (0.28-4.16)	1.00
28-d mortality	3 (4.8)	7 (8.3)	0.55 (0.14-2.22)	.52

Thirty-eight patients without identified pathogens, 1 with dengue fever, and 10 with non- infectious illness were excluded. Abbreviation: CI, confidence intervals.

^a Measurement within 4 weeks before or 2 weeks after ED visit for febrile illness.

^b Including liver abscess and biliary tract infection.

^c Defined with the proportion of less than 1% among the total microorganisms and in- cluded T pallidum, C difficile, C albicans, C indologenes, E faecalis, M pneumoniae, cytomega- lovirus, Epstein-Barr virus, hepatitis B virus, and influenza A.

The positive correlation of the proportion of inappropriate empirical antimicrobial agents and CD4 lymphocyte count was established in the present study. Thus, it is important to demonstrate the distribution of infection-causing microorganisms in febrile HIV-infected adults with various CD4 counts. P jirovecii pneumonia was associated with CD4 lym- phocyte counts less than 100 cell/mm³, in accordance with the 2009 guideline issued by Centers for Disease Control and Prevention [29] that approximately 90% of cases occurred among patients with CD4 counts less than 200 cells/mm³. However, as shown in Fig. 2, all 25 cases of PJP had CD4 counts less than 100 cells/mm³, so we choose the cutoff of 100 cells/mm³ to reflect the real clinical setting.

Among HIV-seronegative IDUs, S aureus was well known to be a major bacterial pathogen of soft tissue infections [21]. Similarly, an asso- ciation of S aureus and IDU or soft tissue infections was demonstrated among HIV-infected individuals in the present study. However, in con- trast to the predominance of MRSA infections among community- associated soft tissue infections in the United States and Taiwan [21,30,31], MRSA was rarely discovered in our study cohort. In particu- lar, a strong correlation of S aureus infection and CD4 count was demon- strated for the first time in HIV-infected individuals. We hypothesize that IDUs having the high CD4 count may contribute to this correlation. Therefore, based on our findings, if ED clinicians encounter febrile HIV-

Intra-abdomen cavity 1 (2.0) 17 (17.5) 0.09 (0.01-0.73) .006

Thirty-eight patients without recognized pathogens and 10 with noninfectious illness were excluded.

infected IDUs with skin and soft tissue infection, especially for patients with high CD4 count, empirical therapy for methicillin-susceptible S au- reus should be considered.

There were some differences in the pathogen distribution among MSM, IDUs, and other groups in our population. First, in previous re- ports, nontyphoid Salmonella infections often occur in those with low CD4 lymphocyte counts [2]. Lymphocyte CD4 counts varied among IDUs, MSM, and other groups in the present study, and notably, less IDUs had CD4 less than 100 cells/mm³. Thus, less nontyphoid Salmonella infection was discovered in IDUs. Second, similar to previous studies emphasized the strong relationship of S aureus and IDUs in HIV- seronegative or HIV-infected individuals, more S aureus infections were present in IDUs in this study. Therefore, these 2 factors were major contributors to the differences in pathogen distribution in differ- ent groups of transmission risk.

In the current study, pulmonary infections, mainly due to P jirovecii and Mycobacterium species, were independently associated with inap- propriate antibiotic therapy at the ED. We suspect that the lack of awareness of ED clinicians regarding these 2 infections may contribute to inappropriate antibiotic therapy at the ED. Nonspecific clinical or

Table 5

Multivariate analyses of risk factors of S aureus infections among community-onset febrile illness in HIV-infected adults

Patient characteristics Odds ratio 95% confidence intervals P

IDUs Source of infection	15.18	3.84-60.03	b.001
Skin and soft tissue	18.45	3.38-100.75	.001
Bone and joint	5.84	0.74-45.95	.09
Lower respiratory tract	0.17	0.03-0.89	.04
Intra-abdomen cavity	0.07	0.006-0.76	.03

Thirty-eight patients without recognized pathogens and 10 with noninfectious illness were excluded.

radiologic presentations of pulmonary infections due to either pathogen render diagnostic challenges to ED clinicians. Without specific clues, the appropriateness of subsequent Clinical decisions of antimicrobial thera- py becomes an issue difficult to address. However, empirical trimetho- prim/sulfamethoxazole can be considered for HIV-infected individuals with pulmonary infections and a CD4 lymphocyte count less than 100 cell/mm³. As for mycobacterial infections, of which tuberculosis was prevalent in Taiwan, increasing cases of nontuberculous mycobacteriosis were reported [32]. Clinical alertness and rapid micro- biological diagnosis of mycobacterial infections in HIV-infected patients should be emphasized.

1. Limitations

We interpreted our data in light of several limitations inherent in the study design. First, the retrospective nature of the study may limit the number of eligible patients. The patients with a lack of culture of the in- dicated specimen for an infectious focus, blood culture sampling, CD lymphocyte count, or risk of HIV transmission were excluded according to our study design. In addition, to ensure accuracy of the information, 2 authors reviewed the chart records. However, only 22 cases were ex- cluded, and the distribution of age (mean, 39.9 vs 42.6 years; P = .33) and sex (female: 16/196, 8.2% vs 2/22, 9.1%; P = .70) was similar in the included and excluded cases. Second, for privacy reasons, the screening test of HIV infection was not routinely performed in the study hospital. Therefore, we only included currently HIV-seropositive patients in the analyses. It is not clear whether this would have affected the distribution of the origin of fever. However, based on the univariate and multivariate analyses, the protective and risk factors of S aureus and PJP are reliable. Third, the low incidence of MRSA infection among HIV- infected patients with soft tissue infection was discovered in our study. Thus, our data may not be applicable to populations with MRSA pre- dominance among community-associated soft tissue infections. Finally, because of the low prevalence (0.12%) among the general population previously reported in Taiwan [10], although we examined all HIV- infected patients visiting the ED over a long period, only a small popula- tion of HIV-infected adults were collected in our study. Therefore, our data may not be generalizable to other communities with high preva- lence of HIV infection or HIV-uninfected individuals.

1. Conclusions

In conclusion, this study is the first ED-based investigation focusing on the distribution of causes of fever in febrile HIV-infected patients. Lower respiratory tract infection was the only independent risk factor of inappropriate antibiotic therapy at the ED. We suggest that ED clini- cians encountering febrile HIV-infected IDUs, especially with skin and soft tissue infection, should consider S aureus. In addition, among HIV- infected adults with low CD4 counts and community-onset pneumonia, trimethoprim/sulfamethoxazole can be empirically initiated for suspected PJP. A prospective study population to validate these Clinical predictors should be examined in the future to improve the quality of care at the ED.

References

1. Niska R, Bhuiya F, Xu J. National Hospital Ambulatory Medical Care Survey: 2007 emergency department summary. Natl Health Stat Report 2010:1-31.
2. Pantaleo G, Graziosi C, Fauci AS. New concepts in the immunopathogenesis of human immunodeficiency virus infection. N Engl J Med 1993;328:327-35.
3. Feinberg J, Leport C, Reiss P. Update on opportunistic infections. AIDS 1991;5 (Suppl. 2):S195-202.
4. Farizo KM, Buehler JW, Chamberland ME, Whyte BM, Froelicher ES, Hopkins SG, et al. Spectrum of disease in persons with human immunodeficiency virus infection in the United States. JAMA 1992;267:1798-805.
5. Stern RS, Weissman JS, Epstein AM. The emergency department as a pathway to ad- mission for poor and high-cost patients. JAMA 1991;266:2238-43.
6. Pane GA, Farner MC, Salness KA. Health care access problems of medically indigent emergency department walk-in patients. Ann Emerg Med 1991;20:730-3.
7. Silva A, Glick NR, Lyss SB, Hutchinson AB, Gift TL, Pealer LN, et al. Implementing an HIV and sexually transmitted disease screening program in an emergency depart- ment. Ann Emerg Med 2007;49:564-72.
8. Kelen GD, Shahan JB, Quinn TC. Emergency department-based HIV screening and counseling: experience with rapid and standard Serologic testing. Ann Emerg Med 1999;33:147-55.
9. Lin KT, Huang SH, Kao CL, Huang KM, Yu JC, Hung TP, et al. An autopsy-proved case of AIDS in Taiwan. Asian Pac J Allergy Immunol 1987;5:25-31.
10. Taiwan Centers for Disease Control. Centers of Disease Control, Taiwan/Health topics/communicable diseases and prevention/AIDS. http://www.cdc.gov.tw/ professional/info.aspx?treeid=3f2310b85436188d&nowtreeid=2285b9745a0a3cbb&tid= 65ED215C50763D72. [accessed 26 August 2014].
11. Lee CC, Lee CH, Chuang MC, Hong MY, Hsu HC, Ko WC. Impact of inappropriate em- pirical antibiotic therapy on outcome of bacteremic adults visiting the ED. Am J Emerg Med 2012;30:1447-56.
12. Clinical and Laboratory Standards Institute. Performance standards for antimicrobial disk susceptibility tests, 24th informational supplement. document M100-S24. Wayne, PA: Clinical and Laboratory Standards Institute; 2014.
13. Atmar RL, Baxter BD, Dominguez EA, Taber LH. Comparison of reverse transcription- PCR with tissue culture and other rapid diagnostic assays for detection of type A in- fluenza virus. J Clin Microbiol 1996;34:2604-6.
14. Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDC definitions for nosocomial infections, 1988. Am J Infect Control 1988;16:128-40.
15. Spector SA, McKinley GF, Lalezari JP, Samo T, Andruczk R, Follansbee S, et al. Oral ganciclovir for the prevention of cytomegalovirus disease in persons with AIDS. Roche Cooperative Oral Ganciclovir Study Group. N Engl J Med 1996;334:1491-7.
16. Cheong HS, Kang CI, Wi YM, Kim ES, Lee JS, Ko KS, et al. Clinical significance and pre- dictors of community-onset Pseudomonas aeruginosa bacteremia. Am J Med 2008; 121:709-14.
17. Lee CC, Chen SY, Chang IJ, Chen SC, Wu SC. Comparison of clinical manifestations and outcome of community-acquired bloodstream infections among the oldest old, el- derly, and adult patients. Medicine (Baltimore) 2007;86:138-44.
18. Selected pharmacologic features of antimicrobial agents. In: Gilbert DN, Moellering Jr RC, Eliopoulos GM, Chambers HF, Saag MS, editors. The Sanford guide to antimi- crobial therapy. 39th ed. Sperryville, VA: Antimicrobial Therapy; 2009.
19. Schellevis FG, van der Velden J, van de Lisdonk E, van Eijk JT, van Weel C. Comorbid- ity of chronic diseases in General practice. J Clin Epidemiol 1993;46:469-73.
20. Le Gall JR, Lemeshow S, Saulnier F. A new Simplified Acute Physiology Score (SAPS II) based on a European/North American multicenter study. JAMA 1993;270:2957-63.
21. Bassetti S, Battegay M. Staphylococcus aureus infections in injection drug users: risk factors and Prevention strategies. Infection 2004;32:163-9.
22. Faul F, Erdfelder E, Lang AG, Buchner A. G*Power 3: a flexible statistical power anal- ysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 2007;39:175-91.
23. Leibovici L, Pitlik SD, Konisberger H, Drucker M. Bloodstream infections in patients older than eighty years. Age Ageing 1993;22:431-42.
24. Valles J, Leon C, Alvarez-Lerma F. Nosocomial bacteremia in critically ill patients: a multicenter study evaluating epidemiology and prognosis. Spanish Collaborative Group for Infections in Intensive Care Units of Sociedad Espanola de Medicina Intensiva y Unidades Coronarias (SEMIUC). Clin Infect Dis 1997;24:387-95.
25. Chang FY, Singh N, Gayowski T, Wagener MM, Marino IR. Fever in liver transplant recipients: changing spectrum of etiologic agents. Clin Infect Dis 1998;26:59-65.
26. Pizzo PA. Fever in Immunocompromised patients. N Engl J Med 1999;341:893-900.
27. Shih FY, Ma MH, Chen SC, Wang HP, Fang CC, Shyu RS, et al. ED overcrowding in Taiwan: facts and strategies. Am J Emerg Med 1999;17:198-202.
28. Trzeciak S, Rivers EP. Emergency department overcrowding in the United States: an emerging threat to patient safety and public health. Emerg Med J 2003;20:402-5.
29. Kaplan JE, Benson C, Holmes KK, Brooks JT, Pau A, Masur H, et al. Guidelines for pre- vention and treatment of opportunistic infections in HIV-infected adults and adoles- cents: recommendations from CDC, the National Institutes of Health, and the HIV Medicine Association of the Infectious Diseases Society of America. MMWR Recomm Rep 2009;58:1-207.
30. Mistry RD, Shapiro DJ, Goyal MK, Zaoutis TE, Gerber JS, Liu C, et al. Clinical manage- ment of skin and soft tissue infections in the U.S. emergency departments. West J Emerg Med 2014;15:491-8.
31. Wang JL, Chen SY, Wang JT, Wu GH, Chiang WC, Hsueh PR, et al. Comparison of both clinical features and mortality risk associated with bacteremia due to community- acquired methicillin-resistant Staphylococcus aureus and methicillin-susceptible

S. aureus. Clin Infect Dis 2008;46:799-806.

1. Hsueh PR, Liu YC, So J, Liu CY, Yang PC, Luh KT. Mycobacterium tuberculosis in Taiwan. J Infect 2006;52:77-85.