Healthcare-Associated Bloodstream Infections
Trevor C. VanSchooneveld
Mark E. Rupp
Healthcare-associated bloodstream infections (HA-BSIs) are a significant and continuing problem in our present-day healthcare system. A variety of factors, including central venous catheterization, predispose patients toward development of infections involving the bloodstream. Pathogens causing these infections vary according to the primary site of infection and a variety of patient factors. Preventive efforts are generally directed at the primary site of invasion. This chapter summarizes general issues related to healthcare-associated bacteremia. More specific information can be found in chapters covering specific primary infections and pathogens.
INCIDENCE AND IMPACT
HA-BSIs are increasing in prevalence and result in significant morbidity, mortality, and economic cost. From 1975 to 1998, the proportion of healthcare-associated infections accounted for by BSIs increased from 5% to 17% (1,2). McGowan and Shulman (3) noted from 1975 through the early 1990s that the rate of HA-BSI increased dramatically from approximately 2 to 4 episodes/1,000 discharges to 15 to 20 episodes/1,000 discharges. A recent review of data from the US Nationwide Inpatient Sample estimated the rate of HA-BSI at 21.6 episodes/1,000 admissions (4). It is estimated that each year in the United States between 250,000 and 500,000 patients experience a HA-BSI and between 30,000 and 100,000 die from these infections (4,5). A recent encouraging development has been a decrease in methicillin-resistant Staphylococcus aureus (MRSA) HA-BSIs (6). The reason for this decline is not clear but possible explanations include changes in S. aureus epidemiology, the impact of hospital policies designed to decrease MRSA transmission, and widespread efforts to decrease rates of central venous catheter (CVC) infection.
The crude mortality associated with HA-BSI varies in published reports from 5% to 58% and depends on the microbial etiology and the underlying condition of the patient (3). Over a 7-year observational period from 1995 to 2002, the Surveillance and Control of Pathogens of Epidemiological Importance (SCOPE) investigators analyzed over 24,000 cases of HA-BSI from 49 medical centers, and noted a crude mortality rate of 27%, ranging from 21% for coagulasenegative staphylococci to 40% for Candida sp. (7). However, attributable mortality is more difficult to ascertain. In some studies that controlled for confounding variables such as severity of illness, BSI was not noted to increase mortality (8,9), while other investigators noted substantial increased mortality (10,11). HA-BSIs result in dramatic increases in economic cost. The length of hospital stay is extended by 1 to 4 weeks at a cost of up to $40,000 per survivor (10, 11, 12, 13, 14). There is no doubt that HA-BSI is a very significant problem associated with the current healthcare system and that efforts to better understand and prevent this problem are well warranted.
CLASSIFICATION AND DEFINITIONS
Although the definition of hospital-acquired BSI appears clear-cut, the application of the definition is, at times, confusing. HA-BSI is typically defined as the demonstration of a recognized pathogen in the bloodstream of a patient who has been hospitalized for >48 hours. BSIs can be further categorized as primary or secondary. When a microorganism isolated from the bloodstream originated from a healthcare-associated infection at another site (urinary tract, surgical site, etc.), the infection is classified as a secondary BSI. Conversely, primary BSIs occur without a recognizable focus of infection elsewhere. It should be noted, that BSIs stemming from intravascular catheters are classified as primary infections.
The Centers for Disease Control and Prevention (CDC) National Healthcare Safety Network (NHSN) previously defined BSI as “laboratory-confirmed BSI” or “clinical sepsis” (15). However, the category of clinical sepsis, which applied to infants and neonates, is no longer considered an NHSN event for BSI (16). NHSN laboratory-confirmed primary BSI must meet at least one of the following criteria:
Criterion 1: Patient has a recognized pathogen cultured from one or more blood cultures and microorganism cultured from blood is not related to an infection at another site.
Criterion 2: Patient has at least one of the following signs or symptoms: fever (>38°C), chills, or hypotension (systolic pressure ≤90 mm Hg) and signs and symptoms and positive laboratory results are not related to an infection at another site and common skin contaminant (e.g., diphtheroids, Bacillus sp., Propionibacterium sp., coagulase-negative
staphylococci, viridans group streptococci, Aerococcus sp., or Micrococcus sp.) is cultured from two or more blood cultures drawn on separate occasions.
staphylococci, viridans group streptococci, Aerococcus sp., or Micrococcus sp.) is cultured from two or more blood cultures drawn on separate occasions.
Criterion 3: Patient ≤1 year of age has at least one of the following signs or symptoms: fever (>38°C rectal), hypothermia (<37°C rectal), apnea, or bradycardia and signs and symptoms and positive laboratory results are not related to an infection at another site and common skin contaminant (e.g., diphtheroids, Bacillus sp., Propionibacterium sp., coagulase-negative staphylococci, viridans group streptococci, Aerococcus sp., or Micrococcus sp.) is cultured from two or more blood cultures drawn on separate occasions.
Although ambiguity is generally not encountered in evaluating patients with positive blood cultures, it is important to note that there is potentially wide practice variation with regard to procurement of blood cultures, and thus bias can be introduced when comparing rates of BSI from institution to institution or unit to unit (17). In general, it is felt that clinicians in the United States are very liberal in their ordering of blood cultures, and it is doubtful that many clinically significant episodes of bacteremia escape detection. However, differentiating true, clinically significant BSI from blood culture contaminants can, at times, offer a challenge to clinicians. This is discussed in greater detail in subsequent sections.
Another issue that has complicated the definition of HA-BSI is the blurring of the distinction between healthcare-associated and community-acquired infections as many therapies traditionally used only in hospitalized patients are now performed routinely in the outpatient setting. Multiple studies have attempted to better define this new category of BSI usually defined as HA-BSI. Friedman et al. (18) observed that of 504 consecutive BSIs detected at an academic medical center and two associated community hospitals, 37% were considered healthcareassociated. Likewise, Siegman-Igra et al. (19) noted that 39% of 604 BSIs occurring in settings traditionally classified as community-acquired could be more accurately classified as healthcare-associated. An analysis of over 6,600 BSIs from a national database classified 55.3% of these infections as healthcare-associated using the criteria of first positive culture within 2 days of admission and any of the following: transfer from another healthcare facility including nursing home, receiving chronic hemodialysis, prior hospitalization within 30 days, and currently on immunosuppressive medication or with metastatic cancer (20). It has been noted that HA-BSIs have similar mortality rates to HA-BSI and are more likely to be due to drug-resistant pathogens including MRSA and extended-spectrum betalactamase (ESBL)-producing Enterobacteriaceae (19,21). Table 19-1 summarizes the characteristics of BSI associated with different patient groups. These findings have significant implications for empiric antimicrobial treatment choices as patients with HA-BSI have been noted to be more likely to receive initially inadequate therapy, likely due to higher rates of resistant pathogens (22). Further research is needed to better delineate this category of patients and their unique risk factors and characteristics.
TABLE 19-1 Classification, Pathogens, and Outcomes from 6,697 Bacteremic Patients | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
CLINICAL MICROBIOLOGY AND DIAGNOSTIC TECHNIQUES
The diagnosis of BSI is dependent on the capacity to recover microbes from the blood. Most large laboratories utilize various automated blood culture systems that are reasonably comparable and are often continuously monitored. These automated systems have been reviewed elsewhere (23), and an extensive discussion is beyond the scope of this chapter. In considering the reliability of recovery of nonfastidious microbes, issues with appropriate procurement likely outweigh the type of system used.
Several factors regarding blood culture reliability and contamination should be emphasized:
Skin Preparation and Culture Technique
Inadequate skin preparation has been reported to be the most frequent cause of culture contamination (24). A variety of products are available for skin preparation. Several studies have found that use of iodine tincture results in lower rates of contamination when compared to povidone iodine, which is thought to be due to the shorter drying time and rapidity of antimicrobial activity associated with the alcohol containing iodine tincture (25, 26, 27). Similarly, a number of trials have observed that skin disinfection with alcoholic chlorhexidine resulted in fewer contaminated blood cultures than povidone iodine (28,29). However, Calfee and Farr (30) observed no significant differences in contamination rates among four different skin antiseptics including povidone iodine, povidone iodine with 70% alcohol, isopropyl alcohol, and tincture of iodine. Studies comparing tincture of iodine and alcoholic chlorhexidine have noted no difference and both have very low rates of contamination (31,32). Based on these studies, guidelines published by both the Infectious Disease Society of America and the American College of Critical Care Medicine recommend using either alcohol alone, chlorhexidine with alcohol, or tincture of iodine for skin decontamination (33,34). Following appropriate skin preparation, if the blood vessel must be palpated, it should be done with a sterile glove. A new needle should be utilized for each attempt at venipuncture (35). Blood should be promptly inoculated into culture bottles following disinfection of culture bottle top septums as they are not sterile. While a meta-analysis suggested the practice of changing needles between procurement of blood and inoculation of blood culture bottles decreased the rate of contamination from 3.7% to 2.0%, it is generally believed the risk of needle stick injury outweighs the benefit of this practice (34,36).
Blood Volume Sampled
To maximize the diagnostic yield from blood cultures, an adequate amount of blood must be sampled. In many cases, the concentration of microorganisms in the bloodstream is ≤1 colony-forming unit (CFU)/mL, and therefore 10 to 20 mL of blood should be sampled to reliably detect bacteremia (37,38). Mermel and Maki (39) calculated that the yield from blood cultures in adults increased 3% per milliliter of blood obtained. Unfortunately, the inadequate sampling of blood volume is frequent in many clinical centers (39). Interestingly, inadequate blood volume has also been associated with increased rates of culture contamination (40).
Timing and Number of Blood Cultures
The optimum time to draw blood cultures is when the number of microbes in the bloodstream is greatest, which unfortunately is 1 to 2 hours before the onset of symptoms (41). Therefore, it is recommended to obtain blood cultures as soon as symptoms occur and preferably before antimicrobials are administered. Although it is common to wait 30 to 60 minutes between obtaining culture sets, Li et al. (42) found no advantage associated with this practice. The practice of drawing blood cultures with fever spikes does not appear to increase yield either (43). Previous literature suggested that two to three blood cultures obtained over a 24-hour period could detect >99% of all bacteremias (44,45). More recent literature suggests that three to four blood cultures over a 24-hour period may be necessary to detect >99% of bacteremias and recent guidelines have recommended this practice, particularly in the critically ill (34,43,46,47). Issues regarding repetitive blood cultures, the utility of anaerobic cultures, blood-to-broth ratios, and other clinical microbiology issues have been reviewed elsewhere (48, 49, 50).
Sites for Obtaining Blood Cultures
Although it is generally recommended to avoid obtaining blood for cultures via intravascular catheters because of concern for contamination, the ease of vascular access, minimization of patient discomfort, and consideration of the catheter as a source of infection has made this a common clinical practice. Multiple studies have evaluated the utility of blood cultures drawn from catheters for the detection of BSI and a recent systematic review summarized their findings (51,52,53). Obtaining blood cultures from catheters increases the sensitivity for detection of bacteremia but is associated with increased isolation of contaminants and decreased positive predictive values. The sensitivity of a single blood culture from either a CVC or peripheral site is not considered adequate for detection of bacteremia and paired blood cultures from both sites are indicated if a blood stream infection is suspected (33). A variety of new diagnostic techniques have been developed to evaluate the source of fever/bacteremia in patients with CVCs including semiquantitative superficial cultures, differential time to positivity and differential quantitative blood cultures (54, 55, 56). These techniques, which are described in greater detail in Chapters 15 and 16, are based on the premise that patients with a catheter-associated infection have a greater burden of bacteria in blood drawn from the intravascular catheter than in blood drawn from the periphery. Recently published guidelines on the diagnosis of CVC infection consider both differential time to positivity and quantitative blood cultures acceptable methods for the diagnosis of CVC-related BSI (33). Therefore, if clinicians are using catheter-drawn blood for culture, it should be paired with a sample drawn peripherally and the sites and times of procurement should be clearly documented.
INDICATIONS FOR BLOOD CULTURES
Indications for blood cultures are not standardized, but should be obtained as a routine study whenever there is a realistic possibility of a HA-BSI. Fever is generally the most common clinical marker for serious healthcare-associated infection, and blood cultures are usually included in the evaluation of fever in hospitalized patients. However, it should be noted that fever may be absent during episodes of bacteremia in certain patient populations such as the elderly, neonates, immunocompromised hosts, and persons with end-stage renal disease. Changes in mental status or functional status may be the most prominent findings associated with bacteremia in elderly patients or patients with renal dysfunction (35,57). Likewise, bacteremia in neonates is often manifested by lethargy, feeding intolerance, apnea, cholestasis, and temperature instability rather than fever (58,59).
If a BSI is identified by blood culture, it is generally not necessary to repeat blood cultures after appropriate treatment has been initiated. Patients who fail to improve despite appropriate antimicrobial therapy should have repeat blood cultures performed to assess for persistence of infection. Also, in the evaluation of S. aureus HA-BSI, many authorities would recommend repeating blood cultures to help assess whether a patient has endocarditis or other deep-seated staphylococcal infection. An exception to this practice are BSIs due to Candida species that require repeat blood cultures to document clearance and determine length of therapy (60).
MICROBIAL ETIOLOGY OF HEALTHCARE-ASSOCIATED BSI
The microbial profile of HA-BSI has changed markedly over the past several decades in response to changes in patient population and antibiotic use. Throughout the 1970s, Enterobacteriaceae were the most common cause of HA-BSI (61). During the 1980s, a relative decrease in bacteremia due to Escherichia coli and Klebsiella pneumoniae was observed, whereas the contribution due to coagulase-negative staphylococci, enterococci, and Candida albicans increased (62). These changes were attributed to the widespread use of antibiotics with activity against Enterobacteriaceae and the increased utilization of indwelling medical devices, particularly intravascular catheters. Banerjee et al. (63), reporting on secular trends in healthcare-associated primary BSIs during the 1980s, found that, depending on the type of hospital studied (small, ≤200 beds; large, ≥500 beds; teaching vs. nonteaching), the rate of bacteremia due to coagulase-negative staphylococci skyrocketed by 161% to 754% (63). Similarly, enterococcal bacteremia increased by 120% to 197% and Candida sp. fungemia increased by 75% to 487% (63). Another trend observed during the 1980s was a shift toward more antibiotic-resistant pathogens. Increased prevalence of antibiotic-resistance was observed in Pseudomonas aeruginosa and Enterobacter cloacae resistant to third-generation cephalosporins, S. aureus and coagulase-negative staphylococci resistant to methicillin, and enterococci resistant to high levels of aminoglycosides (62).
These trends continued in the 1990s. Figure 19-1 illustrates the distribution of over 14,000 bloodstream isolates from the CDC’s National Nosocomial Infection Surveillance (NNIS) hospitals from 1990 through 1996 (64). BSI accounted for approximately 14% of healthcare-associated infections with gram-positive cocci including coagulase- negative staphylococci, S. aureus, and enterococci responsible for 56% of all HA-BSIs (64). Unfortunately, since the mid-1990s, due to limitations in time and personnel resources, fewer and fewer hospitals participated in the hospital-wide surveillance component of the NNIS system and it was discontinued in 1999. However, the NNIS system continued to track healthcare-associated infections from targeted surveillance in intensive care units (ICUs). There was little change in the relative rank order of bloodstream isolates observed in ICU patients from 1990 to 1999. Table 19-2 summarizes this information (65). Pathogens varied by type of ICU with gram-negative pathogens such as Enterobacter sp. or P. aeruginosa causing BSI more frequently in burn ICUs than other types of ICUs (11.2% and 9.5%, respectively), whereas BSI due to S. aureus and coagulase-negative staphylococci occurred with greater frequency in coronary care and cardiothoracic ICU patients (23.2% and 42.7%, respectively) than in other ICUs (65).
 FIGURE 19-1 Microbial etiology of healthcare-associated bloodstream infection from 1990 to 1996. |
NNIS has transitioned into the NHSN in the last decade and while NHSN includes a much larger number of institutions, it no longer reports HA-BSI data. Data on HA-BSIs has been less frequent as both national surveys and literature reports have focused on the syndromes responsible for HA-BSIs such as intravascular catheter infections, pneumonia, and UTI. Some literature has been published including a nationwide surveillance study (SCOPE) that described over 24,000 HA-BSIs from 1995 to 2002 (7). The grampositive pathogens coagulase-negative staphylococci, S. aureus, and enterococci were most common in both ICU and non-ICU settings (62.5% and 59.3%, respectively). Coagulase-negative staphylococci, Enterobacter sp., Serratia sp., Acinetobacter baumannii, and Candida species were more common in the ICU while S. aureus, Klebsiella sp., and E. coli were more common in the general ward (p < .001). A notable finding in this study was the high incidence of BSIs due to Candida species, accounting for nearly 10% of HA-BSIs and increasing significantly from 8% in 1995 to 12% in 2002 (p < .001, trend analysis). C. albicans was the most common species isolated (54%) and C. glabrata (19%), C. parapsilosis (11%), and
C. tropicalis (11%) were also frequently isolated (7). Other centers have noted similar trends with candidal BSIs making up 10% or more of HA-BSIs (66,67,68). Some institutions have recently described a reemergence of gramnegative pathogens causing HA-BSIs. In a tertiary care center in the US from 1999 to 2003 the number of BSIs caused by gram-negative microorganisms significantly increased from 15.9% to 24.1% (p < .001) while infections due to coagulase-negative staphylococci and S. aureus decreased over the same time period (p < .007) (66). These findings have not been described nationally, but in some centers gram-negative pathogens have eclipsed gram-positives as the most common microorganisms causing HA-BSIs (69,70). Factors possibly contributing to the increase in gram-negative pathogens include improved practices in the placement and maintenance of CVCs leading to decreased line-related gram-positive infections, increasing resistance in gram-negative isolates, and the emergence of microorganisms such as A. baumannii as major pathogens in the ICU. These trends in the etiology of HA-BSIs are described in Table 19-3.
C. tropicalis (11%) were also frequently isolated (7). Other centers have noted similar trends with candidal BSIs making up 10% or more of HA-BSIs (66,67,68). Some institutions have recently described a reemergence of gramnegative pathogens causing HA-BSIs. In a tertiary care center in the US from 1999 to 2003 the number of BSIs caused by gram-negative microorganisms significantly increased from 15.9% to 24.1% (p < .001) while infections due to coagulase-negative staphylococci and S. aureus decreased over the same time period (p < .007) (66). These findings have not been described nationally, but in some centers gram-negative pathogens have eclipsed gram-positives as the most common microorganisms causing HA-BSIs (69,70). Factors possibly contributing to the increase in gram-negative pathogens include improved practices in the placement and maintenance of CVCs leading to decreased line-related gram-positive infections, increasing resistance in gram-negative isolates, and the emergence of microorganisms such as A. baumannii as major pathogens in the ICU. These trends in the etiology of HA-BSIs are described in Table 19-3.
TABLE 19-2 Pathogens Isolated from Intensive Care Unit (IC) Healthcare-Associated Bloodstream Infections, National Nosocomial Infections Surveillance Report (NNIS), 1992-1999 (n = 21,943) | ||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ||||||||||||||||||||||
TABLE 19-3 Pathogens Isolated from Healthcare-Associated Bloodstream Infections, 1989-2003 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
As previously mentioned, during the 1980s a trend was observed indicating that HA-BSIs were increasingly being caused by antibiotic-resistant pathogens. This trend continued in the 1990s and worsened in the first decade of the 21st century. Klevens et al. (71) compared NNIS microbiologic data for the period 1990-1994 to 2000-2004 and noted significant increases in MRSA BSIs (27.0-54.1%), ceftazidime-resistant P. aeruginosa pneumonias (16.6-22.7%), and ciprofloxacin-resistant E. coli urinary tract infections (UTIs) (0.9-9.8%). The most recently published data from the NNIS system, summarizing bacterial isolates from ICU and non-ICU inpatient areas from January 1998 to June 2004, indicate an alarming prevalence of antimicrobial resistance (72). These data are shown in Table 19-4. Using data from the SCOPE study that included HA-BSI from 49 hospitals from 1995 to 2002, Wisplinghoff et al. (7) described significant increases in the isolation of MRSA (22-57%), ceftazidime-resistant P. aeruginosa (12-29%), and vancomycin-resistant Enterococcus faecium (47-70%). The rise of resistant pathogens is a global phenomenon as a survey of over 81,000 BSI from three continents noted 2- to 3-fold higher rates of MRSA (38.5%), vancomycin-resistant enterococci (13.3%), ESBL Klebsiella sp. (24.6%), and multidrug-resistant P. aeruginosa (9.0%) in HA-BSIs compared to community BSIs (73).
TABLE 19-4 Prevalence of Antimicrobial-Resistant Phenotypes Among Healthcare-Associated Pathogens Isolated in CDC’s National Nosocomial Infection Surveillance System From January 1998 to June 2004 | |||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| |||||||||||||||||||||||||||||||||||||||||
SOURCES OF BACTEREMIA
Most episodes of primary or laboratory-confirmed HA-BSI without an obvious source are thought to be due to intravascular catheters. These infections are discussed in depth in Chapters 17 and 18. Prior to the widespread use of intravascular catheters, HA-BSIs were largely secondary to infections at other sites. During the 1960s and 1970s, approximately 75% of HA-BSIs were secondary to surgical site infections, intra-abdominal infections, infections of the urinary tract, pneumonia, or skin and soft tissue infections (74,75). Approximately two-thirds of these infections were due to aerobic, gram-negative bacilli (74). As previously mentioned, in more recent years primary BSI has become more prevalent and staphylococci and enterococci have become more prominent pathogens. Pittet and Wenzel (76) noted that from 1981 to 1992 the proportion of HA-BSIs classified as primary BSIs increased from 51% to 71%. Over the same time period, the proportion of HA-BSIs due to coagulase-negative staphylococci increased from 12% to 30% and those due to aerobic gramnegative rods fell from 52% to 29%. SCOPE study data from 1995 to 2002 classified 77% of HA-BSIs as primary with 31% of primary BSI attributed to infections of intravenous catheters (7). UTIs and lower respiratory tract infections were other common sources of BSI (6.5% and 6%, respectively). Corona et al. (77) described HA-BSIs in 26 different countries and classified 58.3% as primary with 45% of primary BSIs being catheter-related. Other frequent sources of BSI included the respiratory tract (15.4%), gastrointestinal tract (8.9%), and wound infections (5.7%). While infections of intravenous catheters are the most common source of device-related HA-BSI other sources may be more common in specific populations. In the elderly and those undergoing urologic procedures, infections due to urinary catheters are the most common source of BSIs (78).
Rates of bacteremia vary due to the pathogen and site of infection. For example, although healthcare-associated UTIs are common and account for 30% to 40% of healthcareassociated infections, they result in secondary bacteremia in only 0.4% to 4% of cases (79, 80, 81). The rate of bacteremia secondary to healthcare-associated UTI appears to be higher with pathogens such as Serratia marcescens (16%) and is lowest in low virulence microorganisms such as coagulase-negative staphylococci (1.8%) (82). Allen et al. (61) found that bacteremia was associated with 3.3% of healthcare-associated UTIs, 6.2% of surgical site infections, and 8.6% of lower respiratory tract infections. Petti et al. (83) recently noted in a community hospital setting that 9.1% of surgical site infections were associated with bacteremia. S. aureus surgical site infection was associated with an almost 3-fold increased rate of bacteremia compared to other microbes (83). Table 19-5 characterizes the relative contribution of various sites to overall rates of secondary bacteremia (7,76,77,84).
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
