Predictive ability of positive clinical culture results and International Classification of Diseases, Ninth Revision, to identify and classify noninvasive Staphylococcus aureus infections: a validation study

Novel strategies to diagnose prosthetic or native bone and joint infections

INTRODUCTION

Bone and Joint Infections (BJI) are medically important, costly and occur in native and prosthetic joints. Arthroplasties will increase significantly in absolute numbers over time as well as the incidence of Prosthetic Joint Infections (PJI). Diagnosis of BJI and PJI is sub-optimal. The available diagnostic tests have variable effectiveness, are often below standard in sensitivity and/or specificity, and carry significant contamination risks during the collection of clinical samples. Improvement of diagnostics is urgently needed.

AREAS COVERED

We provide a narrative review on current and future diagnostic microbiology technologies. Pathogen identification, antibiotic resistance detection, and assessment of the epidemiology of infections via bacterial typing are considered useful for improved patient management. We confirm the continuing importance of culture methods and successful introduction of molecular, mass spectrometry-mediated and next-generation genome sequencing technologies. The diagnostic algorithms for BJI must be better defined, especially in the context of diversity of both disease phenotypes and clinical specimens rendered available.

EXPERT OPINION

Whether interventions in BJI or PJI are surgical or chemo-therapeutic (antibiotics and bacteriophages included), prior sensitive and specific pathogen detection remains a therapy-substantiating necessity. Innovative tests for earlier and more sensitive and specific detection of bacterial pathogens in BJI are urgently needed.

Comparative Effectiveness of Switching to Daptomycin Versus Remaining on Vancomycin Among Patients With Methicillin-resistant Staphylococcus aureus (MRSA) Bloodstream Infections

BACKGROUND

Patients with methicillin-resistant Staphylococcus aureus bloodstream infections (MRSA BSI) usually receive initial treatment with vancomycin but may be switched to daptomycin for definitive therapy, especially if treatment failure is suspected. Our objective was to evaluate the effectiveness of switching from vancomycin to daptomycin compared with remaining on vancomycin among patients with MRSA BSI.

METHODS

Patients admitted to 124 Veterans Affairs Hospitals who experienced MRSA BSI and were treated with vancomycin during 2007-2014 were included. The association between switching to daptomycin and 30-day mortality was assessed using Cox regression models. Separate models were created for switching to daptomycin any time during the first hospitalization and for switching within 3 days of receiving vancomycin.

RESULTS

In total, 7411 patients received vancomycin for MRSA BSI. Also, 606 (8.2%) patients switched from vancomycin to daptomycin during the first hospitalization, and 108 (1.5%) switched from vancomycin to daptomycin within 3 days of starting vancomycin. In the multivariable analysis, switching to daptomycin within 3 days was significantly associated with lower 30-day mortality (hazards ratio [HR] = 0.48; 95% confidence interval [CI]: .25, .92). However, switching to daptomycin at any time during the first hospitalization was not significantly associated with 30-day mortality (HR: 0.87; 95% CI: .69, 1.09).

CONCLUSIONS

Switching to daptomycin within 3 days of initial receipt of vancomycin is associated with lower 30-day mortality among patients with MRSA BSI. This benefit was not seen when the switch occurred later. Future studies should prospectively assess the benefit of early switching from vancomycin to other anti-MRSA antibiotics.

Variable validity of computer extracted problem lists for complications of diabetes mellitus within the VA Greater Los Angeles Health System

PURPOSE

Electronic health data in the form of International Classification of Disease, Ninth Revision (ICD-9) codes is routinely used for clinical research, yet the accuracy of specific diagnoses is largely unknown. The purpose of this study is to assess the validity of computer extracted problem lists for diabetic retinopathy (DR) and other complications of diabetes mellitus (DM) within the VA Greater Los Angeles Health Administration (VHAGLA).

METHODS

The study population consisted of patients at the VHAGLA with an ICD-9 diagnosis of DM between Jan 1st 1999 and March 22nd 2016 with visits to the eye clinic. Fifty patients either with or without an ICD-9 diagnosis of DR were randomly selected. The accuracy of ICD-9 codes for DR, as well as related co-morbidities such as hypertension, hyperlipidemia, coronary artery disease (CAD), and cerebrovascular accident (CVA), were assessed through chart review.

RESULTS

A total of 3193 patients met our inclusion criteria. Of the 50 patients with an ICD-9 diagnosis of DR, the positive predictive value (PPV) was 0.7. For 50 patients without a ICD-9 diagnosis of DR, the negative predictive value (NPV) was 0.9. Of the other co-morbid medical conditions, NPV ranged from a low of 63% for obesity to a high of 98% for CVA and CAD.

CONCLUSION

Validity of ICD-9 diagnoses of diabetic complications in this VA population varied considerably, with DR demonstrating moderate agreement, obesity being more under-documented, and CVA and CAD being more consistently documented. These discrepancies should be considered when using billing codes for research purposes.

Extracting Concepts Related to Homelessness from the Free Text of VA Electronic Medical Records

Mining the free text of electronic medical records (EMR) using natural language processing (NLP) is an effective method of extracting information not always captured in administrative data. We sought to determine if concepts related to homelessness, a non-medical condition, were amenable to extraction from the EMR of Veterans Affairs (VA) medical records. As there were no off-the-shelf products, a lexicon of terms related to homelessness was created. A corpus of free text documents from outpatient encounters was reviewed to create the reference standard for NLP training and testing. V3NLP Framework was used to detect instances of lexical terms and was compared to the reference standard. With a positive predictive value of 77% for extracting relevant concepts, this study demonstrates the feasibility of extracting positively asserted concepts related to homelessness from the free text of medical records.

Clinical and molecular epidemiology of community-onset invasive Staphylococcus aureus infection in New Zealand children

Our aim was to describe the epidemiology and incidence of community-onset invasive S. aureus disease in children presenting to our hospital, and to compare the clonal complexes and virulence genes of S. aureus strains causing invasive and non-invasive disease. The virulence gene repertoire of invasive disease isolates was characterized using DNA microarray and compared with the virulence gene repertoire of non-invasive S. aureus isolates. Over the study period, 163 children had an invasive S. aureus infection. There was no difference in the distribution of clonal complexes or in the prevalence of genes encoding virulence factors between invasive and non-invasive isolates. Future research should include a strong focus on identifying the host and environmental factors that, along with organism virulence factors, are contributing to the patterns of invasive S. aureus disease observed in New Zealand.

Incidence, trends and demographics of Staphylococcus aureus infections in Auckland, New Zealand, 2001-2011

Background

New Zealand has a higher incidence of Staphylococcus aureus disease than other developed countries, with significant sociodemographic variation in incidence rates. In contrast to North America, the majority of disease is due to methicillin-susceptible S. aureus (MSSA), although relatively little is known about the comparative demographics of MSSA and methicillin-resistant S. aureus (MRSA) infections in New Zealand.

Methods

Our objectives were to describe the trends, incidence and patient demographics of all S. aureus infections in patients presenting to our institution between 2001 and 2011, and compare the epidemiology of MSSA and MRSA infections. We identified all patients with S. aureus infections over the study period. A unique S. aureus infection was defined as the first positive S. aureus culture taken from the same patient within a thirty-day period. Standard definitions were used to classify episodes into community- or healthcare-associated S. aureus infection.

Results

There were 16,249 S. aureus infections over the study period. The incidence increased significantly over the study period from 360 to 412 per 100,000 population (P < 0.001), largely driven by an increase in community-associated non-invasive MSSA infections. When compared with MSSA infections, patients with non-multiresistant MRSA infections were more likely to be older, have hospital-onset infections and be Māori or Pacific Peoples.

Conclusions

Our work provides valuable baseline data on the epidemiology and trends of S. aureus infections in New Zealand. The significant increase in community-associated S. aureus infections is of public health importance. Future studies should investigate the reasons underlying this concerning trend.

Using natural language processing on the free text of clinical documents to screen for evidence of homelessness among US veterans

Information retrieval algorithms based on natural language processing (NLP) of the free text of medical records have been used to find documents of interest from databases. Homelessness is a high priority non-medical diagnosis that is noted in electronic medical records of Veterans in Veterans Affairs (VA) facilities. Using a human-reviewed reference standard corpus of clinical documents of Veterans with evidence of homelessness and those without, an open-source NLP tool (Automated Retrieval Console v2.0, ARC) was trained to classify documents. The best performing model based on document level work-flow performed well on a test set (Precision 94%, Recall 97%, F-Measure 96). Processing of a naïve set of 10,000 randomly selected documents from the VA using this best performing model yielded 463 documents flagged as positive, indicating a 4.7% prevalence of homelessness. Human review noted a precision of 70% for these flags resulting in an adjusted prevalence of homelessness of 3.3% which matches current VA estimates. Further refinements are underway to improve the performance. We demonstrate an effective and rapid lifecycle of using an off-the-shelf NLP tool for screening targets of interest from medical records.

Use of electronic health record data to identify skin and soft tissue infections in primary care settings: a validation study

Background

Epidemiologic studies of skin and soft tissue infections (SSTIs) depend upon accurate case identification. Our objective was to evaluate the positive predictive value (PPV) of electronic medical record data for identification of SSTIs in a primary care setting.

Methods

A validation study was conducted among primary care outpatients in an academic healthcare system. Encounters during four non-consecutive months in 2010 were included if any of the following were present in the electronic health record: International Classification of Diseases, Ninth Revision (ICD-9) code for an SSTI, Current Procedural Terminology (CPT) code for incision and drainage, or a positive wound culture. Detailed chart review was performed to establish presence and type of SSTI. PPVs and 95% confidence intervals (CI) were calculated among all encounters, initial encounters, and cellulitis/abscess cases.

Results

Of the 731 encounters included, 514 (70.3%) were initial encounters and 448 (61.3%) were cellulitis/abscess cases. When the presence of an ICD-9 code, CPT code, or positive culture was used to identify SSTIs, 617 encounters were true positives, yielding a PPV of 84.4% [95% CI: 81.8–87.0%]. The PPV for using ICD-9 codes alone to identify SSTIs was 90.7% [95 % CI: 88.5–92.9%]. For encounters with cellulitis/abscess codes, the PPV was 91.5% [95% CI: 88.9–94.1%].

Conclusions

ICD-9 codes may be used to retrospectively identify SSTIs with a high PPV. Broadening SSTI case identification with microbiology data and CPT codes attenuates the PPV. Further work is needed to estimate the sensitivity of this method.

Reduction of methicillin-resistant Staphylococcus aureus infection among veterans in Atlanta

OBJECTIVE

Describe local changes in the incidence of community-onset and hospital-onset methicillin-resistant Staphylococcus aureus (MRSA) infection and evaluate the impact of MRSA active surveillance on hospital-onset infection.

DESIGN

Observational study using prospectively collected data.

SETTING

Atlanta Veterans Affairs Medical Center (AVAMC).

PATIENTS

All patients seen at the AVAMC over an 8-year period with clinically and microbiologically proven MRSA infection.

METHODS

All clinical cultures positive for MRSA were prospectively identified, and corresponding clinical data were reviewed. MRSA infections were classified into standard clinical and epidemiologic categories. The Veterans Health Administration implemented the MRSA directive in October 2007, which required active surveillance cultures in acute care settings.

RESULTS

The incidence of community-onset MRSA infection peaked in 2007 at 5.45 MRSA infections per 1,000 veterans and decreased to 3.14 infections per 1,000 veterans in 2011 ([Formula: see text] for trend). Clinical and epidemiologic categories of MRSA infections did not change throughout the study period. The prevalence of nasal MRSA colonization among veterans admitted to AVAMC decreased from 15.8% in 2007 to 11.2% in 2011 ([Formula: see text] for trend). The rate of intensive care unit (ICU)-related hospital-onset MRSA infection decreased from October 2005 through March 2007, before the MRSA directive. Rates of ICU-related hospital-onset MRSA infection remained stable after the implementation of active surveillance cultures. No change was observed in rates of non-ICU-related hospital-onset MRSA infection.

CONCLUSIONS

Our study of the AVAMC population over an 8-year period shows a consistent trend of reduction in the incidence of MRSA infection in both the community and healthcare settings. The etiology of this reduction is most likely multifactorial.

Predictors of Staphylococcus aureus Rectovaginal Colonization in Pregnant Women and Risk for Maternal and Neonatal Infections

BACKGROUND

Staphylococcus aureus infections are increasing among pregnant and postpartum women and neonates, but risk factors for S. aureus colonization in pregnancy and the association between maternal colonization and infant infections are not well defined. We sought to identify risk factors for maternal S. aureus rectovaginal colonization and assess colonization as a risk factor for infections among mothers and infants.

METHODS

We conducted a retrospective cohort study of pregnant women and their infants. Demographic and clinical data, including S. aureus infections that occurred in mothers from 3 months before to 3 months after delivery and in infants during the first 3 months of life, were extracted from electronic medical records. Predictors for maternal S. aureus rectovaginal colonization were assessed through multivariable logistic regression analysis.

RESULTS

The cohort included 2702 women and 2789 infants. The prevalence of maternal rectovaginal colonization with methicillin-susceptible S. aureus and methicillin-resistant S. aureus (MRSA) was 13% and 0.7%. Independent predictors of colonization included multigravidity, human immunodeficiency virus seropositivity, and group B Streptococcus colonization. S. aureus colonization was associated with an increased risk of infection in mothers (odds ratio [OR], 3.5; 95% confidence interval [CI], 1.4-8.8) but not in their infants (OR, 1.9; 95% CI, .6-5.6). The frequency of S. aureus infections was 0.8% in mothers and 0.7% in infants.

CONCLUSIONS

S. aureus rectovaginal colonization was associated with an increased risk of infections in women but not in their infants. The frequency of MRSA infections was low. These data suggest that routine MRSA screening of pregnant women may not be indicated.

Staphylococcus aureus infections in US veterans, Maryland, USA, 1999-2008

Trends in Staphylococcus aureus infections are not well described. To calculate incidence in overall S. aureus infection and invasive and noninvasive infections according to methicillin susceptibility and location, we conducted a 10-year population-based retrospective cohort study (1999-2008) using patient-level data in the Veterans Affairs Maryland Health Care System. We found 3,674 S. aureus infections: 2,816 (77%) were noninvasive; 2,256 (61%) were methicillin-resistant S. aureus (MRSA); 2,517 (69%) were community onset, and 1,157 (31%) were hospital onset. Sixty-one percent of noninvasive infections were skin and soft tissue infections; 1,112 (65%) of these were MRSA. Ten-year averaged incidence per 100,000 veterans was 749 (± 132 SD, range 549-954) overall, 178 (± 41 SD, range 114-259) invasive, and 571 (± 152 SD, range 364-801) noninvasive S. aureus infections. Incidence of all S. aureus infections significantly increased (p<0.001), driven by noninvasive, MRSA, and community-onset infections (p<0.001); incidence of invasive S. aureus infection significantly decreased (p<0.001).