Le Infezioni in Medicina, n. 3, 427-433, 2021

doi: 10.53854/liim-2903-14

ORIGINAL ARTICLES

Achromobacter spp. bacteremia outbreak related to contaminated furosemide ampoules

Rajalakshmi Arjun1, Kalpana E John2, Vettakkara Kandy Muhammed Niyas1, Sreerekha R Nair3, Viji Mohan2 , Raveendran Sarala Ratheesh2

1Department of Infectious Diseases, Kimshealth, Thiruvananthapuram, Kerala, India;

2Department of Microbiology, Kimshealth, Thiruvananthapuram, Kerala, India;

3Department of Hospital Infection Control, Kimshealth, Thiruvananthapuram, Kerala, India

Article received 23 May, 2021; accepted 27 July, 2021

Corresponding author
Rajalakshmi Arjun
E-mail: dr.a.rajalakshmi@gmail.com

SummaRY

Nosocomial outbreaks related to medication contamination are reported world-wide. A sudden increase in cases of Achromobacter spp. bacteremia led to an outbreak investigation in our setting. Line listing and environmental sampling led to identification of contaminated furosemide ampoules as the source. Molecular identification helped in species identification and in this outbreak more than one species was identified. Prompt withdrawal of the contaminated batch of ampoules curtailed the outbreak.

Key words: Achromobacter, nosocomial infections, bacteremia, medication contamination, outbreak.

INTRODUCTION

Achromobacter xylosoxidans is an aerobic, motile, oxidase-positive, non-fermenting Gram-negative rod widely distributed in the environment. In 1971 Yabuuchi and Ohyama first described this pathogen from purulent ear discharge in patients with chronic otitis media. They later proposed the name A. xylosoxidans which encompasses 2 different subspecies A. xylosoxidans subsp. xylosoxidans and A. xylosoxidans subsp. denitrificans [1]. Achromobacter is not a usual component of human endogenous flora and is commonly found in water sources, such as well water, hospital faucets, swimming pools, and moist soil. They can contaminate intravenous (IV) fluids, disinfectants and water in humidifiers [2]. Apart from primary bacteraemia, other clinical presentations include meningitis, urinary tract infections, abscesses, osteomyelitis, corneal ulcers, prosthetic valve endocarditis, peritonitis, pneumonia and wound infection. Majority of cases reported in literature have been associated with healthcare-associated infections and in patients with haematological malignancy and those who are immunosuppressed [3, 4]. Bacteremia out­break in an ICU setting is a concern warranting urgent investigation, since nosocomial bacteremias are associated with increased morbidity, mortality, length of hospital stay and cost. The aim of the present paper is to describe an outbreak of Achromobacter bacteremia observed in our hospital. Here we describe the efforts and procedures adopted to identify the source of the infection. We have also reviewed the literature regarding the epidemiology and previous outbreaks due to Achromobacter spp. causing invasive nosocomial infections.

OUTBREAK DESCRIPTION

In April 2020, increasing number of bacteremia due to Achromobacter spp was noted, particularly in our ICUs. A trend of Achromobacter bacteremia over the past 2 years was analysed. There were total three cases of Achromobacter bacteremia in 2019. Between March and mid- April 2020, there were six cases Achromobacter bacteremia in cardiac and medical ICUs, which was above the usual occurrence, suggesting an outbreak. Soon an outbreak investigation team was formed which included hospital infection control team, infectious diseases specialists, microbiologist, lead in clinical services and nurse lead, to help identify all cases of Achromobacter bacteremia and collect the clinical details both retrospectively and prospectively till the outbreak ended. Case was defined as one in whom one or more blood cultures grew Achromobacter spp. In the latter half of April and in May, six more cases were identified taking the total number of Achromobacter bacteremia to twelve. Line listing of the cases was done.

Patient characteristics

There were 12 patients with Achromobacter bacteremia between March and May 2020; the details are shown in Table 1. Eleven of them were in cardiac or medical ICUs while one was in the ward. Majority were males (10 out of 12) and age ranged from 30 to 81. All patients had primary bacteraemia. Duration from admission till blood culture positivity ranged from 1 to 16 days; 7 out of 12 had bacteraemia within 3 days of admission. Only 6 underwent some form of interventional procedure, either vascular or surgical and were not uniform; 5 had hemodialysis. In those who had hemodialysis, blood cultures which grew Achromobacter spp. were drawn before the initiation of dialysis. Six patients had central venous catheter and arterial line; the rest had peripheral vascular catheter only. All of them received at least two or three IV medications like inotropes, diuretics, proton pump inhibitors, heparin, antibiotics, fluids and saline flush. Intravenous diuretic, furosemide was used in all patients for pulmonary edema, either as bolus or infusion.

Table 1 - Details of those who had Achromobacter bacteremia during the outbreak.

Case	Procedure done	Day(s) from admission to positive blood culture sampling	IV Medications administered prior to blood culture draw	Achromobacter spp. isolated
1	Nil	2	Saline, furosemide, enoxaparin, pantoprazole	A. denitrificans
2	Nil	1	Furosemide, pantoprazole, calcium gluconate	A. denitrificans
3	CAG and DVR	9	Furosemide	A. denitrificans A. xylosoxidans
4	CAG	3	Pantoprazole, furosemide	A. denitrificans
5	PTCA	7	Pantoprazole, heparin, furosemide	A. denitrificans
6	Nil	5	Furosemide, pantoprazole, 25% dextrose	A. xylosoxidans
7	Nil	<1	Furosemide	A. denitrificans
8	Nil	<1	Furosemide	A. denitrificans
9	CAG	4	Furosemide	A. denitrificans
10	Nil	<1	Furosemide	A. xylosoxidans
11	Nil	<1	Furosemide	A. denitrificans A. xylosoxidans
12	Sphenoidal mass excision, ICD insertion	16	Furosemide	A. xylosoxidans

Abbreviations: IV- intravenous, CAG- coronary angiogram, PTCA- percutaneous transcatheter coronary angioplasty, DVR, double valve replacement, ICD- intercostal drainage.

Blood cultures were done by inoculating in the BacT/ALERT® FA aerobic and FN anaerobic bottles and incubated in an automated system (BacT/ALERT 3D, bioMerieux, France) for five days. Bacterial identification and antimicrobial susceptibility testing were done using Vitek2, according to the Clinical Laboratory Standards Institute (CLSI) guidelines. Selected samples were further analysed to confirm species by molecular techniques by VITEK MS MALDITOF (bioMérieux) and whole genome sequencing (WGS) of DNA done on Illumina HiSeqX by Next Generation Sequencing technique. Blood culture from 10 patients grew either one of the species- A. denitrificans and A. xylosoxidans; both the species were identified in 2 patients. All isolates were resistant to quinolones, aminoglycosides and cotrimoxazole but were susceptible to ceftazidime, piperacillin tazobactam and carbapenem. All were treated with appropriate antibiotics according to the susceptibility results, and majority received piperacillin tazobactam. Four patients expired; all four had underlying heart failure.

Possible sources of Achromobacter infection

Further epidemiological details were collected based on the literature search of common sources of Achromobacter outbreak. Literature search was done with regard to the epidemiology and previous outbreaks due to Achromobacter spp. using PubMed Central and key words Achromobacter, outbreak and bacteremia. Subsequent cross references of important papers were done. Achromobacter spp. has been implicated in outbreaks of nosocomial infections associated with contaminated solutions including chlorhexidine atomizer, deionized water, intravenous preparations and contamination during diversion of morphine infusion [5-8]. With this information possible sources that were considered for the current bacteremia outbreak included contaminated chlorhexidine solution, water used in hemodialysis system, common procedures, personnel involved, instruments and medical supplies used and parenteral medications administered. Locations in the hospital where the outbreak occurred was also looked at and included in the line listing. During this period, visits were made to observe procedures, specifically regarding preparation and administration of intravenous medications and interview the area staff about the level of knowledge and training in the procedures they did.

Clinical variables looked at and noted were age, sex, hospitalization dates, location of admission, underlying diagnosis, dates when blood culture grew Achromobacter spp, intravenous medications received, presence and type of central venous catheter (CVC), symptoms, types of invasive procedures, and sample from which Achromobacter spp. was isolated.

Investigation for environmental contamination

Sample for culture was collected from chlorhexidine-alcohol skin disinfectant, de-ionised water used in humidifier and from work surfaces. Environmental samples were inoculated into MacConkey purple broth, and incubated at 37oC for up to 48 hrs. Turbidity/colour change to yellow were indicative of bacterial growth. Such samples were sub-cultured to blood agar and MacConkey agar, and the growth obtained was identified using Vitek 2 identification system. Samples from commonly used IV preparations were noted from the line listing and sent for cultures. Samples of IV medication solution were inoculated in thioglycolate broth, incubated for 18-24 hours at 37oC, and observed for turbidity. Samples showing turbidity were sub-cultured to blood agar and MacConkey agar, and the growth obtained was identified using Vitek 2 identification system.

RESULTS

Chlorhexidine-alcohol skin disinfectant used for hand disinfection, as well as the one used for pre-procedure skin disinfection from the affected unit were found to be sterile. The de-ionised water used in humidifier and dialysis water source for the affected area yielded no growth. Neither did the environmental samples collected from work surfaces.

All IV medications that the patients received prior to taking blood culture were listed out. Majority of the patients had IV drugs like inotropes, pantoprazole, heparin, saline infusion, saline flush, but the one medicine that all patients uniformly received was furosemide (Table 1). Furthermore, data was collected from pharmacy regarding details of batches and the brands of furosemide which were despatched to these patients during this period. All of them had received one particular brand and batch of furosemide.

Samples of available 2 brands of furosemide (Brand 1 and 2), pantoprazole, heparin, single use saline flush, and saline bottles were given for culture. Brand 2 of furosemide, pantoprazole, heparin, single use saline flush and saline did not yield growth. Culture of furosemide Brand 1, belonging to the same batch of furosemide which were administered to these patients yielded growth of Achromobacter spp. Four more ampoules of the same batch were cultured. All 4 ampoules grew non-lactose fermenting colonies in MacConkey agar and turbidity in the broth which were identified by Vitek2 as 2 different species, A. denitrificans and A. xylosoxidans. Thus, the source of the outbreak of Achromobacter spp. bacteraemia was identified as contaminated furosemide injection preparations.

Following identification, the pharmacy was instructed to withhold dispatch of the contaminated batch of furosemide and recall was done from all clinical areas. Following this, no new cases of Achromobacter bacteraemia were noted in the subsequent months.

Three random isolates (2 patient samples and one furosemide ampoule) were also tested by MALDITOF and WGS. The characteristic of the isolates that were obtained by WGS was interpreted using Kraken2 tool, the results of which are shown in Table 2. Multiple Achromobacter spp. as well as a discrepancy in identification of genus by Vitek 2 and molecular technique were noted in this outbreak.

Table 2 - Details of the clinical and contaminated furosemide sample isolates.

Sample	Method of species identification			Antibiotic susceptibility by Vitek
	Vitek 2	MALDI-TOF	WGS	Ctz	Cpm	AG	Cipro	TMP-SMX	BL-BLI	Carba
Patient sample 1	A. denitrificans	A. insolitus	A. insolitus	S	R	R	R	R	S	S
Patient sample 2	A. xylosoxidans	A. insolitus	A. denitrificans	S	R	R	I	R	S	S
Furosemide ampoule	A. denitrificans	A. xylosoxidans	A. xylosoxidans	ND	ND	ND	ND	ND	ND	ND

Abbreviations: Ctz: ceftazidime, Cpm: cefepime, AG: aminoglycosides, Cipro: ciprofloxacin, TMP-SMX: cotrimoxazole, BL-BLI: betalactam- betalactamase inhibitor, Carba.: Carbapenem, S: sensitive, R: resistant, ND: not done.

DISCUSSION

Between March and April 2020, we had 12 cases of Achromobacter bacteremia in our institution which warranted an outbreak investigation. Seven of them had bacteremia within three days after admission and four had blood cultures taken on day of admission growing Achromobacter. Previous studies on Achromobacter infections have pointed to its occurrence commonly in hospital setting [4, 9]. Though seven cases had a very short period of hospitalization before occurrence of bacteremia, nosocomial infection was strongly considered, as their initial presentation were as cardiac failure and not with fever. More over community onset Achromobacter endogenous infection is uncommon. Any contaminated agent directly coming in contact with blood can cause bacteremia in a short interval. Pseudo-bacteremia due to contamination of blood culture during venipuncture, in the preparation of culture media, or during laboratory processing of the culture was also considered, but many developed symptoms of infection later on.

In one of the earliest review articles on A xylosoxidans infections, Chandrasekar, et al. highlighted the pathogenic potential of this organism in immune-compromised host and therefore they should not be disregarded as contaminants [3]. Achromobacter nosocomial infections in newborn carry significant mortality. In one of the earliest reports in 1960, Linde et al., in his analysis of 205 cases of congenital heart surgery, described 2 Achromobacter endocarditis due to contaminated heart- lung machine [10]. Aerosolization of the organism resulting in pneumonia, meningitis and secondary bacteremia has also been described in new born units [11, 12].

Nosocomial bacteremias due to Achromobacter due to contaminated environmental sources, devices or IV preparations are summarized in Table 3. In initial reports contaminated water sources and equipment were noted as the source. Reverdy, et al., in his study describes the epidemiological investigation on Achromobacter spp. outbreak in ICU setting leading to identification of A. xylosoxidans in deionized water from the faucets of the hemodialysis system and the outbreak was controlled by disinfecting the deionized water system with sodium hypochlorite solution [13]. Gomez-Cerezo J, et al. describe 54 cases of A xylosoxidans bacteremia of which 52 episodes were nosocomial and the source of infection in this series was identified as tap water and the contaminated hands of two healthcare workers. As several studies have implicated faucet aerators as reservoirs of nosocomial pathogens, by removing faucet aerators and instituting the use of gloves the outbreak was curtailed [5]. In another outbreak, Tena, et al. explain that catheter related bacteremia due to A xylosoxidans where the source was an atomizer with diluted chlorhexidine which was used for skin disinfection. The isolates from blood and atomizer were tested by pulse field gel electrophoresis (PFGE) to confirm atomizer as the source [6]. In another outbreak, A xylosoxidans contaminated transducers caused 15 cases of hospital infection. The problem arose from the re-use of disposable equipment after disinfection with benzalcone [14, 15].

Table 3 - Prior reports of Achromobacter bacteremia in hospital settings due to contaminated environmental sources or IV preparations

Ref. No	Authors	Article	Nature of infection	No. of cases	Source of the outbreak
10	Linde LM, et al., 1960	Bacterial endocarditis following surgery for congenital heart disease.	Endocarditis	2	Heart lung machine
12	Hanson. L A, et al., 1966	Achromobacter Septicemia in a Premature- Clinical and Bacteriological Aspects.	Septicaemia with meningitis	1	Aerosol from an infected humidifier
13	Reverdy ME, et al., 1984	Nosocomial Colonization and Infection by A. xylosoxidans.	Peritonitis pneumonia with septicemia	2	Deionized water
14	Gahrn-Hansen B, et al., 1988	Outbreak of infection with A. xylosoxidans from contaminated intravascular pressure transducers.	Bacteremia	15	Intravascular pressure transducer
8	Reina J, et al., 1998	Nosocomial outbreak of A. xylosoxidans associated with a diagnostic contrast solution.	Bacteremia	5	Contamination of IV contrast
7	Kim M, et al., 2002	Blood stream infections in out-patient oncology office.	Bacteremia	12	Probably related to contaminated multi dose vials of heparin and saline flushes
6	Tena D, et al., 2005	Outbreak of long-term intravascular catheter-related bacteremia due to A. xylosoxidans in a hemodialysis units	Bacteremia	4	Atomizer containing diluted chlorhexidine

Line-listing helped in suggesting that intravenous furosemide, which was administered for pulmonary edema was common for all patients. The only IV medication that every patient had received prior to blood culture sampling was furosemide. Several reports on Gram negative bacteremia due to contaminated IV preparations have been reported previously and majority are non-lactose fermenters; two of them have been related to Achromobacter species [16-21]. Reina, et al. in their brief communication have reported a nosocomial outbreak of A xylosoxidans related to administration of intravenous contrast agent. The authors had suspected contrast solution contamination, but could not prove it [8]. Multidose vials of heparin or saline flushes were identified as a source and confirmed by PGFE in another outbreak of blood stream infections due to Alcaligenes xylosoxidans, in an out-patient oncology setting [7].

In some outbreaks, even after extensive search, source of the outbreak could not be identified [22]. In a study from India published in 2005 by Kumar, et al., during one-year period between 1996 and 1997, they found 20 strains of A. xylosoxidans from clinical specimens, and majority of the patients were neonates. Pulse field gel electrophoresis confirmed that 90% belonged to a single genotype. In this period environmental surveillance did not point to any potential source and transmission probably occurred from person to person over a period of time [23]. In another outbreak study, 34 episodes of bacteremia occurred in 22 neonates during a 6-month period. The investigation was not able to single out the source of the outbreak. Despite intensive efforts to control the outbreak containment could be achieved only after the neonatal intensive care unit was relocated [24].

In the current outbreak two different species of Achromobacter were identified by Vitek 2 and in 2 patients, each one of their blood culture sample grew different species. This can be explained by contamination of the furosemide ampoules with multiple species of Achromobacter or due to difficulty in species identification by Vitek 2. Species level identification of non-lactose fermenting Gram negative rods by conventional method is still a challenge and non-molecular tests may be inadequate for reliable identification [25, 26]. Molecular methods to confirm the species identification showed variation as depicted in Table 2. It is noted that there may be species mismatch between conventional method and MALDITOF MS with non-fermenting Gram-negative rods and WGS is considered the reference standard for species identification.

In the current outbreak, recall of the contaminated batch of furosemide ampoules resulted in control of Achromobacter bacteremia. The manufacturer and distributor were alerted about details of contaminated batch of ampoules. In the following months no further cases of Achromobacter bacteremia were reported from this facility [26].

In conclusion, this evaluation highlights the importance of microbiology alerts in identifying early and rapid control of the nosocomial outbreaks. Contamination of IV preparations with non-lactose fermenters have been previously reported and should be considered in any outbreak of bacteremia. Species identification of Gram-negative non-lactose fermenter can be a challenge and may require molecular methods to confirm it.

Sources of support

None.

Acknowledgements

We would like to acknowledge the support rendered by our organization KIMSHEALTH for arranging molecular diagnostics and the labs in helping us with it namely MALDI-TOF, MS method, done at ML laboratories, Coimbatore, Tamil Nadu, India. and WGS at the Medgenome lab, Bangalore, Karnataka, India.

Conflicts of interests

None.

REFERENCES

[1] Yabuuchi E, Kawamura Y, Kosako Y, Ezaki T. Emendation of genus Achromobacter and Achromobacter xylosoxidans (Yabuuchi and Yano) and proposal of Achromobacter ruhlandii (Packer and Vishniac) comb. nov., Achromobacter piechaudii (Kiredjian et al.) comb. nov., and Achromobacter xylosoxidans subsp. denitrificans (Rüger and Tan) comb. nov. Microbiol Immunol. 1998; 42 (6), 429-38.

[2] Spear JB, Fuhrer J, Kirby BD. Achromobacter xylosoxidans (Alcalagines xylosoxidans subsp. xylosoxidans) bacteremia associated with a well-water source: case report and review of the literature. J Clin Microbiol. 1988; 26, 598-9.

[3] Chandrasekar PH, Arathoon E, Levine DP. Infections due to Achromobacter xylosoxidans. Case report and review of the literature. Infection. 1986; 14 (6), 279-82.

[4] Duggan JM, Goldstein SJ, Chenoweth CE, Kauffman CA, Bradley SF. Achromobacter xylosoxidans bacteremia: report of four cases and review of the literature. Clin Infect Dis. 1996; 23 (3), 569-76.

[5] Gómez-Cerezo J, Suárez I, Ríos JJ, et al. Achromobacter xylosoxidans bacteremia: a 10-year analysis of 54 cases. Eur J Clin Microbiol Infect Dis. 2003; 22 (6), 36-43.

[6] Tena D, Carranza R, Barbera JR, et al. Outbreak of long-term intravascular catheter-related bacteremia due to Achromobacter xylosoxidans subspecies xylosoxidans in a hemodialysis unit. Eur J Clin Microbiol Infect Dis. 2005; 24 (11), 727-32.

[7] Kim MJ, Bancroft E, Lehnkering E, Donlan RM, Mascola L. Alcaligenes xylosoxidans Bloodstream Infections in Outpatient Oncology Office. Emerg Infect Dis. 2008; 14 (7), 1046-52.

[8] Reina J, Antich M, Siquier B, Alomar P. Nosocomial outbreak of Achromobacter xylosoxidans associated with a diagnostic contrast solution. J Clin Pathol. 1998; 41 (8), 920-1.

[9] Pérez E, Pérez BJS, Corbella L, Orellana MA, Fernández-Ruiz M. Achromobacter xylosoxidans bacteremia: clinical and microbiological features in a 10-year case series. Rev Esp Quimioter. 2018; 31 (3), 268-73.

[10] Linde LM, Heins HL. Bacterial endocarditis following surgery for congenital heart disease. N Engl J Med. 1960; 263 (2), 65-9.

[11] Foley JF, Gravelle CR, Englehard WE, Chin TDY. Achromobacter septicemia - fatalities in prematures. I. Clinical and epidemiological study. Am J Dis Child. 1961; 101 (3), 279-88.

[12] Hanson LA, Kjellander J, Lindsjö A. Achromobacter septicemia in a premature. Clinical and bacteriological aspects. Acta Paediatr Scand. 1966; 55 (3), 337-40.

[13] Reverdy ME, Freney J, Fleurette J, et al. Nosocomial colonization and infection by Achromobacter xylosoxidans. J Clin Microbiol. 1984; 19 (2), 140-3.

[14] Gahrn-Hansen B, Alstrup P, Dessau R, et al. Outbreak of infection with Achromobacter xylosoxidans from contaminated intravascular pressure transducers. J Hosp Infect. 1988; 12 (1), 1-6.

[15] McGuckin MB, Thorpe RJ, Koch KM, Alavi A, Staum M, Abrutyn E. An outbreak of Achromobacter xylosoxidans related to diagnostic tracer procedures. Am J Epidemiol. 1982; 115 (5), 785-93.

[16] Matsaniotis NS, Syriopoulou VP, Theodoridou MC, Tzanetou KG, Mostrou GI. Enterobacter sepsis in infants and children due to contaminated intravenous fluids. Infect Control. 1984; 5 (10), 471-7.

[17] Muller AE, Huisman I, Roos PJ, et al. Outbreak of severe sepsis due to contaminated propofol: lessons to learn. J Hosp Infect. 2010; 76 (3), 225-30.

[18] Singhal T, Shah S, Naik R. Outbreak of Burkholderia cepacia complex bacteremia in a chemotherapy day care unit due to intrinsic contamination of an antiemetic drug. Indian J Med Microbiol. 2015; 33 (1), 117-9.

[19] Moehring RW, Lewis SS, Isaacs PJ, Schell WA, Thomann WR, Althaus MM. Outbreak of Bacteremia due to Burkholderia contaminans linked to intravenous fentanyl from an institutional compounding pharmacy. JAMA Intern Med. 2014; 174 (4), 606-12.

[20] Held MR, Begier EM, Beardsley DS, Browne FA, Martinello RA, Baltimore RS. Life-threatening sepsis caused by Burkholderia cepacia from contaminated intravenous flush solutions prepared by a compounding pharmacy in another State. Pediatrics. 2006; 118 (1), e212-e215.

[21] Chiang PC, Wu TL, Kuo AJ, et al. Outbreak of Serratia marcescens postsurgical bloodstream infection due to contaminated intravenous pain control fluids. Int J Infect Dis. 2013; 17 (9), e718-22 .

[22] Aisenberg G, Rolston KV, Safdar A. Bacteremia caused by Achromobacter and Alcaligenes species in 46 patients with cancer (1989-2003). Cancer. 2004; 101 (9), 2134-40.

[23] Kumar A, Ray P, Kanwar M, Sethi S, Narang A. Investigation of hospital-acquired infections due to Achromobacter xylosoxidans in a tertiary care hospital in India. J Hosp Infect. 2006; 62 (2), 248-50.

[24] Turel O, Kavuncuoglu S, Hosaf E, et al., Bacteremia due to Achromobacter xylosoxidans in neonates: clinical features and outcome. Braz J Infect Dis. 2013; 17 (4), 450-4.

[25] Kaur M, Ray P, Bhatty M, et al. Epidemiological typing of clinical isolates of Achromobacter xylosoxidans: comparison of phenotypic and genotypic methods. Eur J Clin Microbiol Infect Dis. 2009; 28 (9), 1023-32.

[26] Gomila M, Prince-Manzano C, Svensson- Stadler L, et al. Genotypic and Phenotypic Applications for the Differentiation and Species-Level Identification of Achromobacter for Clinical Diagnoses. PLoS One. 2014; 9 (12), e114356.