28th March 2014, Volume 127 Number 1391

Genevieve Walls, Stephen McBride, Nigel Raymond, Kerry Read, Christin Coomarasamy, Arthur J Morris, Stephen Chambers, David Holland, David R Murdoch

Despite medical advances, infective endocarditis (IE) continues to cause substantial morbidity and mortality. The International Collaboration on Endocarditis Prospective Cohort Study (ICE-PCS), a worldwide study of IE in the 21st century, included patients recruited from New Zealand. Regional variations in presentation and management of IE exist, and knowledge of these may assist local physicians.

In this paper we present the New Zealand data from the ICE-PCS study. This is the largest series of patients with IE in New Zealand.

Methods

Design–New Zealand study sites involved in the ICE-PCS were Auckland, Counties Manukau, Waitemata, Capital and Coast, and Canterbury District Health Boards (DHBs). The methods of the ICE-PCS are described in detail elsewhere,1 and are briefly reiterated below. Study sites enrolled patients aged 18 years or over with a diagnosis of definite or probable IE, according to the modified Duke criteria.2

Study sites were required to fulfil the following criteria: minimum enrolment of 12 cases of IE per year, access to cardiac surgery, patient identification procedures ensuring consecutive enrolment and minimising ascertainment bias, high quality data, including query resolution, and Ethics Committee approval or waiver based on local standards.1 Patients were enrolled between 1 June 2000 and 1 September 2005. One-year follow-up data were collected.

A standardised case report form of 275 variables was used. These variables and definitions are described elsewhere.2,3 Data were collected by investigators at the participating site during the index hospitalisation and then collated by the coordinating centre.

Definitions—Community-acquired IE was diagnosed within 48 hours of admission in a patient who did not fulfil the criteria for healthcare-associated infection. Healthcare-associated IE was classified as nosocomial or non-nosocomial.

Nosocomial IE was diagnosed in a patient who was hospitalised for more than 48 hours before onset of symptoms or signs of IE.

Non-nosocomial healthcare-associated IE was diagnosed within 48 hours of admission in an outpatient with extensive healthcare contact, either: receipt of intravenous (IV) therapy, wound care, or specialised nursing care at home within the 30 days before IE onset; attendance at a hospital or haemodialysis clinic or receipt of IV chemotherapy within the 30 days prior to onset of IE; hospitalisation in an acute care hospital for two or more days in the 90 days prior to onset of IE; or residence in a nursing home or long-term care facility.1

Statistical analysis—Statistical analysis was performed using SAS version 9.3 software. Fisher’s exact test and Chi-squared tests were used to compare proportions. Logistic regression models were used to produce odds ratios.

Results

Internationally, 3284 patients were enrolled into ICE-PCS, with definite IE by the modified Duke criteria present in 2781. New Zealand recruited 336 patients (266 with definite IE and 70 with probable IE). Table 1 shows the baseline characteristics and medical conditions of patients in the New Zealand cohort.

The median age at enrolment of New Zealand patients was 59.5 years (range 15 to 98 years, interquartile range (IQR) 41-73), and the majority of patients were male (229 males, 68%). Prosthetic valve endocarditis occurred in 104/336 (31%); the remainder had native valve endocarditis or another type of endocarditis, e.g. related to other prosthetic intracardiac material (11/336, 3%). Around a quarter of patients (93/336, 28%) had underlying medical conditions.

Conditions predisposing to IE were common, notably underlying structural heart disease. Data on the prevalence of underlying rheumatic heart disease, prior invasive procedures and the presence of intravascular devices at the time of IE diagnosis, were available for 274 patients (274/336, 82%).

Thirty-four patients (34/336, 10%) had already had an episode of IE before the enrolment episode; eleven were previously included in the ICE-PCS. Of patients with a previous episode, 6/34 (18%) had a history of injecting drug use (IDU).

Forty patients (40/336, 12%) had healthcare-associated endocarditis: nosocomial 78% (31/40) and non-nosocomial 22% (9/40).

Table 1. Baseline characteristics and predisposing factors

1. Baseline characteristics                      

New Zealand cohort, n=336 (%)

Median age, years (range)

59.5 (15 – 98)

Male gender

229 (68)

Haemodialysis

10 (3)

Diabetes mellitus

40 (12)

HIV

1 (0.3)

Cancer

34 (10)

Native valve IE

221 (66)

Prosthetic valve IE

104 (31)

Infected pacemaker/ICD

0

2. Predisposing conditions

Current IDU

18 (5)

Previous IE

34 (10)

Invasive procedure in preceding 60 days

40/274 (15)

Long-term IV access at onset of IE

24/274 (9)

Pacemaker in situ

9 (3)

ICD

0

Congenital heart disease

45 (13)

Native valve predisposition

124 (37)

Rheumatic heart disease

12/274 (4)

HIV: human immunodeficiency virus; ICD: implantable cardiac defibrillator; IDU: injecting drug user

The clinical and echocardiographic findings in this cohort are shown in Table 2. Most patients (257/336, 76%) had fever and 120/336 (36%) had a new or changed heart murmur. Classical signs of IE were rare, with few patients exhibiting Osler’s nodes (6/336, 2%), Roth spots (5/336, 1%) or Janeway lesions (7/336, 2%).

Echocardiography was performed in 334/336 patients (99%): 191/334 (57%) had both a trans-thoracic echocardiogram (TTE) and a transoesophageal echocardiogram (TOE). Forty-one (41/334, 12%) had TOE alone and 102/334 (31%) had TTE alone.

The time of onset of infective endocarditis was known in 187/336 patients (56%) and estimated in the remainder. Eleven patients with a known date of IE onset developed IE after hospitalisation; the remaining 176 patients developed symptoms in the community. The median time between development of symptoms and hospital admission was four days (range 1-126 days, IQR 1-12 days). Most patients with a known duration of symptoms were admitted within a week of symptom onset (111/176, 63%).

The time from onset of symptoms to hospitalisation varied with the causative organism (Figure 1). Patients with S. aureus IE were more likely to present within a week of symptom onset than those with viridans streptococcus IE (OR 4.18, 95% CI 2.36–7.42). Only 5/36 patients (14%) with Enterococcus faecalis IE presented within a week of symptom onset.

Table 2. Clinical, laboratory and echocardiography findings in NZ-ICE patients

Clinical sign/laboratory finding

New Zealand patients with sign, n=336 (%)

Fever (temperature >38°)

257 (76)

New murmur

90 (27)

Worsening murmur

30 (9)

Splinter haemorrhages

65 (19)

Janeway lesions

7 (2)

Osler’s nodes

6 (2)

Roth spots

5 (1)

Conjunctival haemorrhage

6 (2)

Splenomegaly

12 (4)

Other emboli

13 (4)

Elevated CRP

266 (79)

Elevated ESR

234 (70)

Microscopic haematuria

15 (4)

Positive rheumatoid factor

5 (1)

Echocardiographic findings, n=334 patients

Aortic regurgitation

73 (22)

Mitral regurgitation

79 (24)

Tricuspid regurgitation

20 (6)

Any vegetation

229 (69)

Aortic valve vegetation

109 (33)

Mitral valve vegetation

115 (34)

Tricuspid valve vegetation

17 (5)

Chordae tendinae vegetation

6 (2)

Paravalvular abscess

43 (13)

Paravalvular perforation

10 (3)

Paravalvular fistula

4 (1)

Prosthetic valve dehiscence

8/104 (8)

Prosthetic valve regurgitation

10/104 (10)


Figure 1. Relationship between causative organism and duration of symptoms

Holland-1

 

Blood cultures were obtained from 333 patients (333/336, 99%) and were positive in 94% (314/333 patients). Data on number of bottles of blood cultures were available for 273 patients (273/333, 82%). A median of six blood culture bottles was collected per patient (range 0–22 bottles, IQR 4-8). Amongst patients with positive blood cultures, a median of four bottles per patient was positive (range 1–16, IQR 2-6) and 46% of patients (153/333) had growth from blood samples drawn at least 12 hours apart.

The causative organism was cultured from three or more sets (where one set consists of two bottles) of blood cultures in 61% of patients (204/333). All blood cultures were negative in 22/333 patients (7%); 12/22 (55%) had been on antibiotics within seven days prior to obtaining blood cultures.

Culture of excised valvular tissue or vegetations from 13 blood-culture-negative patients was positive in four: Haemophilus parainfluenzae, Haemophilus/Aggregatibacter aphrophilus and Propionibacterium acnes were the organisms cultured; one valve had polymicrobial growth. Seven patients had additional serological testing, including two blood-culture-negative patients, but these tests were non-contributory. Polymerase chain reaction (PCR) was performed on excised valvular tissue from two blood-culture-negative patients, which led to identification of Staphylococcus lugdunensis, the presumed causative organism. In total, 80 valves and three vegetations were removed at surgery and sent for culture; organisms were cultured from 42/83 specimens (51%). The 41 patients whose valves and vegetations were culture-negative had had pre-operative antibiotics for longer than the 42 whose valve cultures were positive (median 11 days versus 4 days).

Table 3 outlines the organisms causing endocarditis in this cohort. Viridans streptococci were the most common (109/336 patients, 32%), followed by S. aureus (80/336 patients, 24%). Viridans streptococci were more likely to cause native than prosthetic valve endocarditis (36% versus 24%; difference in proportions 0.12, 95%CI 0.014–0.22), as was S. aureus (28% versus 15%; difference in proportions 0.13, 95%CI 0.042–0.22). Enterococcus faecalis (OR 2.44, 95% CI 0.19–0.89) and coagulase-negative staphylococci (OR 4.17, 95% CI 0.075–0.75) were both more likely to cause prosthetic valve endocarditis than viridans streptococci.

Most S. aureus were methicillin-susceptible (MSSA, 75/80 patients, 94%). There were five cases of methicillin-resistant S. aureus (MRSA) endocarditis. Nine (9/40, 23%) E. faecalis isolates demonstrated resistance to high-level gentamicin. There were no vancomycin-resistant enterococci (VRE).

Seventy-two percent of patients with viridans streptococcus IE (79/109) had isolates susceptible to penicillin (minimum inhibitory concentration (MIC≤0.25 mg/L);4 and 2% (2/109) had IE caused by viridans streptococci resistant to penicillin (MIC >2 mg/L). Antibiotic treatment was administered to 335 patients; in most cases, antibiotic treatment involved a beta-lactam. Complete specific data on antibiotic treatment are not available for this cohort.

Surgery was performed in 110/336 patients (33%). Table 4 outlines the surgical management and findings for patients in the New Zealand ICE-PCS cohort. Five patients in whom surgery was indicated were too unwell for surgery.

Table 4. Surgical management of patients in the NZ-ICE cohort

Variables

        Number of patients n=110 (%)A

Aortic valve replacement (AVR) or repair

77/110 (70)

AVR – mechanical valve

32/77 (42)

AVR – bioprosthetic valve

43/77 (56)

Repair – no prosthesis

1/77 (1)

Unspecified aortic valve replacement/repair

1/77 (1)

Mitral valve replacement (MVR) or repair

42/110 (38)

MVR – mechanical valve

28/42 (67)

MVR – bioprosthetic valve

6/42 (14)

Repair – no prosthesis

3/42 (7)

Unspecified mitral valve replacement/repair

5/42 (12)

AVR + MVR

12/110 (11)

Tricuspid valve replacement (TVR) or repair

7/110 (6)

TVR – mechanical valve

1/7 (14)

TVR – bioprosthetic valve

2/7 (29)

Unspecified tricuspid valve replacement/repair

4/7 (57)

Pulmonary valve replacement

1 (1)

Remove infected pacemaker

0

Other procedure

16 (15)

A Several patients had more than one procedure. Other procedures: replace VSD patch (1 patient); remove vegetation (2); aortic root graft repair (1); aortic root replacement (2); coronary artery bypass graft (3); closure ventricular septal defect (VSD) (4); closure patent ductus arteriosus (1); drainage infected pericardial effusion (1); replace incompetent conduit (1); pulmonary artery homograft (1); remove infected VSD patch (1).

 

Enterococcal IE and Streptococcus bovis/gallolyticus IE were more likely to be managed medically rather than surgically; there was no significant difference between surgical versus medical management for other organisms.

Those who were managed medically were older than those managed surgically (median ages 67 and 46 years respectively), with 39/83 (47%) of those aged 15–40 years receiving surgery compared with 5/88 patients (6%) aged 73-98 years (p<0.001).

The median time from initial contact with healthcare services to surgery was four days (range 0 to 259 days, IQR 1-12). Most patients (109/110, 98%) received antibiotics before surgery. There was evidence of endocarditis at surgery or on surgical specimens in 107/110 patients (97%).

Measured outcomes in the ICE-PCS were in-hospital mortality, length of hospital stay and complications of disease or treatment. These outcomes are shown in Table 5.

In total, 20/336 (6%) patients died in hospital. The median time to death after admission was 15.5 days (range 1 to 80 days, IQR 9.5-22.5), and the median age at death was 76 years (range 39 to 90 years, IQR 61-80.5). Those with IE caused by coagulase-negative staphylococci had an increased risk of death compared with those with IE caused by viridans streptococci (RR 4.7, 95% CI 1.2, 17).

The most common complications of IE were congestive heart failure (71/336, 21%) and systemic embolisation (59/336, 17%). IE caused by coagulase negative staphylococci conferred a higher relative risk of several complications (Table 5).

There were 104 patients with prosthetic valve endocarditis. The microbiology of prosthetic valve IE is shown in Table 3. TOE was performed in 85 patients with prosthetic valve IE (85/104, 82%). Paravavular complications were identified in 17% (18/104 patients): prosthetic valve dehiscence in 8/104 patients (8%) and paravalvular regurgitation in 10/104 patients (10%)). Vegetations were identified in 60 patients (60/104, 58%). Thirty-six patients required surgery (36/104, 35%). Complications and outcomes for patients with prosthetic valve IE are shown in Table 6.

Adverse drug effects, including rash, hepatotoxicity, neutropenia and Clostridium difficile colitis, complicated the treatment of 7/336 patients (2%).

One-year follow-up data were available for 316 patients. Out-of-hospital mortality data were collected. A further 18 patients (18/316, 6%) had died at one-year follow up, two from strokes, four from heart failure, and the remaining patients for other or unknown reasons. Fourteen patients (14/316, 4%) required further surgery on their aortic valve (eight patients) or mitral valve (two patients).

Discussion

The ICE-PCS provided valuable global information on endocarditis in the 21st century, although the bulk of data was obtained from Europe (1213 patients, 44%) and North America (597 patients, 22%),1 which may have influenced descriptions of the characteristics of IE. Others have published retrospective single-centre or referral centre data on IE in New Zealand, 5-10 however we believe this is the only prospective multicentre study and the largest modern series of New Zealand patients with IE.

As noted in the international cohort, most New Zealand patients with IE were men (68%) with a median age of 60 years. A significant proportion of New Zealand patients with IE had underlying structural cardiac abnormalities, including congenital heart disease (13%), prosthetic heart valves (31%) and a native valve predisposition (37%). The prevalence of underlying rheumatic heart disease, however, was low (4%), a significant change from historical descriptions of IE, where up to 50% of patients had rheumatic heart disease,11 and less than may have been anticipated in New Zealand given the high incidence of acute rheumatic fever among Maori and Pacific Island people.12 It is possible that some IE patients with rheumatic heart disease were not included in the study because they did not live in a participating DHB’s catchment area. We were not able to match these data with specific DHBs and so are not able to comment on this. However, Counties Manukau DHB, which has one of the highest incidences of acute rheumatic fever in New Zealand,13 was included in the ICE-PCS, suggesting that the rheumatic heart disease data presented here should be representative.

Other studies of IE in New Zealand describe a variable prevalence of underlying rheumatic heart disease in the 21st century: 1 of 62 patients (2%) in a study from Dunedin Hospital5 (a tertiary referral centre with cardiothoracic surgical services) and 11 of 57 patients (19%) with IE in Tauranga6 (a peripheral hospital with no cardiothoracic surgical services). A study of IE in South Auckland in the 1980s reported an underlying prevalence of rheumatic heart disease of 45%.9

The clinical presentation of IE in the modern age is also different from historical descriptions. The majority of patients had had symptoms for less than one month before presentation, and those with S. aureus IE tended to have a shorter illness than those with IE caused by other organisms. Clinical signs of IE were less frequent in this cohort than traditional descriptions of the disease, with very few exhibiting the classical Roth spots, Osler’s nodes or Janeway lesions.

This change in the clinical presentation of IE has been noted elsewhere:1,11 IE is less frequently a sub-acute illness presenting with classical signs engendered by prolonged inflammation, but rather an illness of relatively short duration, which may have few objective clinical signs. This change may be due to a number of factors including changing patient characteristics, for example, more patients with prior healthcare contact, prosthetic valves or degenerative heart disease rather than underlying rheumatic heart disease; the increasing role of S. aureus as a causative agent; and earlier diagnosis through improved microbiological methods and readier access to echocardiography.

Viridans streptococci remained the most common causative organisms in the New Zealand cohort (32%), followed by S. aureus (24%). Internationally, S. aureus has emerged as the commonest cause of IE, accounting for 31% of IE cases worldwide and 43% of cases in the USA, compared with 17% of cases worldwide caused by viridans streptococci (9% in the USA).1

The preponderance of S. aureus IE internationally differs from historical studies, in which viridans streptococci were more prevalent.11 This is thought to reflect the changing epidemiology of S. aureus infection in the Western world, with increased contact with healthcare services,14 high rates of S. aureus bacteraemia reported among hospitalized patients,14,15 increased numbers of patients receiving prosthetic endovascular and intracardiac devices with subsequent infection,16 and increased incidence of invasive procedures, all of which are associated with S. aureus bacteraemia or endocarditis.3,14 Other risk factors for S. aureus IE (e.g. IDU), also exist globally, and contribute to making S. aureus the leading cause of IE in many parts of the world.

While there appeared to be less S. aureus IE in the New Zealand cohort than internationally, it is difficult to make direct comparisons between the groups The international study analysed data only from patients with definite IE, whereas New Zealand patients with both definite and probable IE were included in this paper. There appeared to be a lower prevalence of healthcare-associated IE, previous healthcare contact, and IDU in the New Zealand cohort, which may potentially contribute to less S. aureus IE than internationally.

Only 12% of the New Zealand cohort had healthcare-associated IE; in the North American sub-group 32% had healthcare-associated IE.1 In the New Zealand cohort there was a low prevalence of prior invasive procedures (15%), endocavitary cardiac device infection (0), and patients receiving haemodialysis (3%); in the North American cohort the prevalence was 27%, 8% and 21% respectively. The prevalence of IDU in the New Zealand cohort is also low (5%), reflecting a low prevalence of IDU in the general New Zealand population.17

The in-hospital mortality rate in the New Zealand cohort was relatively low (6%) compared with historical New Zealand data.8-10 Internationally, in-hospital mortality remains high, between 12 – 22%.1,3 The low in-hospital mortality in this cohort may possibly be explained by a lower proportion of S. aureus IE in New Zealand than internationally, but also possibly by the contribution of bias in the type of New Zealand patient who was included in the study, leading to an under-representation of severe IE in this cohort. For example, a number of patients from non-study-centre hospitals may have been too unwell to undergo surgery or transfer to a study centre. The inclusion of ‘probable’ endocarditis in this analysis may have also contributed to an apparently lower mortality. It is difficult to comment on whether this reflects an actual decline in IE mortality in New Zealand.

Only 2 of 22 (10%) blood-culture negative patients had additional serological or molecular testing to identify a causative organism (e.g. Bartonella species or Tropheryma whipplei; Coxiella burnettii, a well recognised cause of culture-negative endocarditis, is not endemic in New Zealand). Overseas studies show that these fastidious organisms cause culture-negative endocarditis relatively frequently, and they should be borne in mind as part of the differential diagnosis.18 Molecular methods are increasingly used for elucidating the microbiological aetiology of IE and would almost certainly be found to be more widely used if this study were repeated.

This study was not specifically designed to look at IE in New Zealand; however, it is a large, prospective study involving several local centres. All sites were major centres with ready access to cardiothoracic surgical services. This may have introduced observational bias towards patients with more severe disease or a requirement for surgery, or, conversely, towards those who were well enough to undergo surgery. A truly representative picture of IE in New Zealand would involve both tertiary centres and peripheral hospitals.

Another limitation is the lack of data on medical management of IE in New Zealand. Other interesting data, which were initially collected by local investigators in this cohort but are no longer available for analysis, include information about dentition and dental procedures, antimicrobial therapy used, previous valve surgery for IE, duration between IE episodes and indications for and findings at surgery.

Conclusion

The ICE-PCS provides valuable information on IE in the modern era, and most of the findings in the international cohort apply to IE in New Zealand. IE in the modern age is less likely to be a prolonged illness with classical clinical signs and is more likely to present acutely.

In the New Zealand cohort, there was a ‘traditional’ predominance of viridans streptococci over S. aureus as causative organisms, though the increasing role of S. aureus in IE observed internationally may soon be relevant to New Zealand as well. This large contemporary series of IE in New Zealand provides a reference for local practitioners assessing and managing infective endocarditis.

Abstract

Aim

The International Collaboration on Endocarditis Prospective Cohort Study (ICE-PCS) collected worldwide data on the presentation, management and outcome of infective endocarditis (IE). We present data from patients with endocarditis enrolled from New Zealand

Method

Patients who fulfilled the Duke criteria for definite or probable endocarditis were enrolled from five district health boards: Auckland, Counties Manukau, Waitemata, Capital and Coast, and Canterbury, between June 2000 and September 2005.

Results

There were 336 New Zealand patients enrolled in the ICE-PCS. Prosthetic valve endocarditis occurred in 31%. Underlying medical conditions were present in 28% of patients, but only 4% of patients had rheumatic heart disease. Forty patients (12%) had healthcare-associated endocarditis. Viridans streptococci were the most common cause of IE (32%), followed by Staphylococcus aureus (24%). Patients with S. aureus IE were more likely to present within a week of symptom onset than those with viridans streptococcus IE (OR 4.18, 95% CI 2.36–7.42). Surgery was performed in 33% of patients. In total, 20 patients (6%) died in hospital. Those with endocarditis caused by coagulase-negative staphylococci had an increased risk of death compared with those viridans streptococcus endocarditis (RR 4.7, 95% CI 1.2–17). The risk of stroke was higher in those with endocarditis caused by S. aureus and coagulase-negative staphylococci (RR 2.7, 95% CI 1.2–6.05, and 4.9, 95% CI 1.9–13, respectively).

Conclusion

While viridans streptococci remain the predominant causative organisms of IE in New Zealand, many ‘traditional’ clinical and management aspects of this disease no longer apply. This paper provides a reference for local practitioners assessing and managing IE.

Author Information

Genevieve Walls, Infectious Diseases Physician, Department of Infectious Diseases, Middlemore Hospital, Auckland; Stephen McBride, Infectious Diseases Physician, Department of Infectious Diseases, Middlemore Hospital, Auckland; Nigel Raymond, Infectious Diseases Physician, Department of Infectious Diseases, Wellington Hospital, Wellington; Kerry Read, Infectious Diseases Physician, Department of Infectious Diseases, North Shore Hospital, Auckland; Christin Coomarasamy, Statistician, Biostatistics Research Office, Counties Manukau District Health Board, Auckland; Arthur J Morris, Microbiologist, Auckland City Hospital, Auckland; Stephen Chambers, University of Otago, Christchurch; David Holland, Infectious Diseases Physician, Department of Infectious Diseases, Middlemore Hospital, Auckland; David Murdoch, University of Otago, Christchurch 

Acknowledgements

The authors acknowledge Dr Dalice Sim (Statistical Consultant, School of Mathematics, Statistics and Operations Research, Victoria University of Wellington, New Zealand) as well as the investigators of the ICE-PCS (listed in Reference 1).

Correspondence

Genevieve Walls, Department of Infectious Diseases, Middlemore Hospital, Private Bag 93311, Otahuhu, Auckland, New Zealand. Fax: +64 (0)9 2709746;

Correspondence Email

Genevieve.Walls@middlemore.co.nz

Competing Interests

Nil

References

1.       Murdoch DR, Corey R, Hoen B, et al. Clinical presentation, etiology and outcome of infective endocarditis in the 21st century: the International Collaboration on Endocarditis Prospective Cohort Study. Arch Int Med 2009;169:463-73.

2.       Li JS, Sexton DJ, Mick N, et al. Proposed modifications to the Duke Criteria for the diagnosis of infective endocarditis. Clin Infect Dis 2000;30:633-8.

3.       Fowler VF, Miro JM, Hoen B, et al. Staphylococcus aureus endocarditis: a consequence of medical progress. J Am Med Assoc 2005;293:3012-21.

4.       European Committee on Antimicrobial Susceptibility Testing. Data from the EUCAST clinical breakpoint website, last accessed 10 January 2013. http://www.eucast.org

5.       Chu J, Wilkins G, Williams M. Review of 65 cases of infective endocarditis in Dunedin Public Hospital. NZ Med J 2004;117:1-12.

6.       Wong CW, Porter G, Tisch J, Young C. Outcome and prognostic factors on 57 cases of infective endocarditis in a single centre. NZ Med J 2009;122:54-62.

7.       Morris AJ, Drinkovic D, Pottumarthy S, et al. Gram stain, culture, and histopathological examination findings for heart valves removed because of infective endocarditis. Clin Infect Dis 2003;36:697-704.

8.       Wells AU, Coe Fowler C, Ellis-Pegler RB, et al. Endocarditis in the 80s in a general hospital in Auckland, New Zealand. Quarterly J Medicine 1990;76:753-62.

9.       Peat EB, Lang SDR. Infective endocarditis in a racially mixed community: a 10 year review of 78 cases. NZMJ 1989;102:33-6.

10.    Ormiston JA, Neutze JM, Agnew TM, et al. Infective endocarditis: a lethal disease. Aust NZ J Med 1981;11:620-9.  

11.    Weinstein L, Rubin RH. Infective endocarditis – 1973. Prog Cardiovasc Dis 1973;16:239-74.

12.    Wilson N. Rheumatic heart disease in indigenous populations – New Zealand experience. Heart Lung Circ 2010;19:282-8.

13.    Milne RJ, Lennon DR, Stewart JM, et al. Incidence of acute rheumatic fever in New Zealand children and youth. J Paed Child Health 2012;48:685-91.   

14.    Naber CK. Staphylococcus aureus bacteremia: epidemiology, pathophysiology and management strategies. Clin Infect Dis 2009;48 Suppl:231-7.

15.    Wisplinghoff H, Bischoff T, Tallent SM, et al. Nosocomial bloodstream infections in US hospitals: analysis of 24,179 cases from a prospective nationwide surveillance study. Clin Infect Dis 2004;39:309-17.

16.    Cabell CH, Heidenreich PA, Chu VH, et al. Increasing rates of cardiac device infections among Medicare beneficiaries: 1990–1999. Am Heart J 2004;147:582-6.

17.    Mason K, Hewitt A, Stefonogiannis N, et al. Drug Use in New Zealand: Key results of the 2007/2008 New Zealand alcohol and drug use survey. Wellington: Ministry of Health 2010.

18.    Fournier PE, Thuny F, Richet H, et al. Comprehensive diagnostic strategy for blood culture-negative endocarditis: a prospective study of 819 new cases. Clin Infect Dis. 2010;51:131-40.