Surveillance for Clostridium difficile infection: ICD-9 coding has poor sensitivity compared to laboratory diagnosis in hospital patients, Singapore

Evidence of within-facility patient-patient Clostridiodes difficile infection spread across diverse settings

Previous studies have suggested that a hospital patient's risk of developing healthcare facility-onset (HCFO) Clostridioides difficile infections (CDIs) increases with the number of concurrent spatially proximate patients with CDI, termed CDI pressure. However, these studies were performed either in a single institution or in a single state with a very coarse measure of concurrence. We conducted a retrospective case-control study involving over 17.5 million inpatient visits across 700 hospitals in eight US states. We built a weighted, directed network connecting overlapping inpatient visits to measure facility-level CDI pressure. We then matched HCFO-CDIs with non-CDI controls on facility, comorbidities and demographics and performed a conditional logistic regression to determine the odds of developing HCFO-CDI given the number of coincident patient visits with CDI. On average, cases' visits coincided with 9.2 CDI cases, which for an individual with an average length of stay corresponded to an estimated 17.7% (95% CI 12.9-22.7%) increase in the odds of acquiring HCFO-CDI compared to an inpatient visit without concurrent CDI cases or fully isolated from both direct and indirect risks from concurrent CDI cases. These results suggest that, either directly or indirectly, hospital patients with CDI lead to CDIs in non-infected patients with temporally overlapping visits.

Epidemiology of Clostridium (Clostridioides) difficile Infection in Southeast Asia

This review describes the current understanding of Clostridium (Clostridioides) difficile infection (CDI) in southeast Asia regarding the prevalence of CDI, C. difficile detection methods, antimicrobial susceptibility profiles, and the potential significance of a One Health approach to prevention and control. Our initial focus had been the Indochina region, however, due to limited studies/surveillance of CDI in Indochina, other studies in southeast Asian countries and neighboring Chinese provinces are presented here for comparison. Clostridium (Clostridioides) difficile infection is one of the most common causes of hospital-acquired gastroenteritis worldwide. Since its discovery as a cause of pseudomembranous colitis in 1978, C. difficile-related disease has been more prevalent in high-income rather than low-income countries. This may be because of a lack of knowledge and awareness about the significance of C. difficile and CDI, resulting in underreporting of true rates. Moreover, the abuse of antimicrobials and paucity of education regarding appropriate usage remain important driving factors in the evolution of CDI worldwide. The combination of underreporting of true CDI rates, along with continued misuse of antimicrobial agents, poses an alarming threat for regions like Indochina. C. difficile ribotype (RT) 027 has caused outbreaks in North America and European countries, however, C. difficile RT 017 commonly occurs in Asia. Toxin A-negative/toxin B-positive (A-B+) strains of RT 017 have circulated widely and caused outbreaks throughout the world and, in southeast Asia, this strain is endemic.

Clostridioides difficile epidemiology in the Middle and the Far East

OBJECTIVES

Clostridioides difficile is an important pathogen of healthcare-associated gastrointestinal infections. Recently, an increased number of C. difficile infection (CDI) surveillance data has been reported from Asia. The aim of this review is to summarize the data on the prevalence, distribution and molecular epidemiology of CDI in the Middle and the Far East.

METHODS

Literature was drawn from a search of PubMed up to September 30, 2021.

RESULTS

The meta-analysis of data from 111 studies revealed the pooled CDI prevalence rate in the Middle and the Far East of 12.4% (95% CI 11.4-13.3); 48 studies used PCR for CDI laboratory diagnoses. The predominant types (RT)/sequence type (ST) differ between individual countries (24 studies, 14 countries). Frequently found RTs were 001, 002, 012, 017, 018 and 126; RT017 was predominant in the Far East. The epidemic RT027 was detected in 8 countries (22 studies), but its predominance was reported only in three studies (Israel and Iran). The contamination of vegetable and meat or meat products and/or intestinal carriage of C. difficile in food and companion animals have been reported; the C. difficile RTs/STs identified overlapped with those identified in humans.

CONCLUSIONS

A large number of studies on CDI prevalence in humans from the Middle and the Far East have been published; countries with no available data were identified. The number of studies on C. difficile from non-human sources is limited. Comparative genomic studies of isolates from different sources are needed.

Effect of clinical versus administrative data definitions on the epidemiology of C. difficile among hospitalized individuals with IBD: a population-based cohort study

Background

Hospitalization admissions and discharge databases (DAD) using the International Classification of Diseases (ICD) codes are often used to describe the epidemiology of Clostridioides difficile infections (CDI) among those with Inflammatory bowel disease (IBD), even though DAD CDI definition can miss many cases of CDI. There are no data comparing the assessment of the epidemiology of CDI among those with IBD by DAD versus laboratory diagnosis. We used a population-based dataset to determine the effect of using DAD versus laboratory CDI diagnosis on CDI assessment among those with IBD.

Methods

We linked the University of Manitoba IBD Epidemiology Database to the provincial CDI laboratory dataset for the years 2005–2014. Time trends of CDI were assessed using joinpoint analyses. We used stratified logistic regression analysis to assess factors associated with CDI among individuals with IBD.

Results

Time trends of CDI among hospitalized individuals with IBD were similar when using DAD or the laboratory CDI diagnosis. Prior hospital admission and antibiotic exposure were associated with CDI using either of the CDI definitions, 5-ASA use was associated with CDI using DAD but not laboratory diagnosis, whereas corticosteroid exposure was associated with laboratory-based CDI diagnosis. Using laboratory results as gold standard, DAD had a sensitivity and specificity of 75.4% and 99.6% for CDI among those with IBD.

Conclusions

Using ICD codes in the DAD for CDI provides similar epidemiological time trend patterns as identifying CDI in the laboratory dataset. Hence, ICD codes are reliable to determine CDI epidemiology among hospitalized individuals with IBD.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12876-022-02223-y.

Prevalence and molecular subtyping of Blastocystis in patients with Clostridium difficile infection, Singapore

Background

Blastocystis is a common anaerobic colonic protist in humans with controversial pathogenicity. Clostridium difficile (C. difficile) is the commonest cause of infectious diarrhea in healthcare settings. The prevalence and subtype (ST) characteristics of Blastocystis in patients with C. difficile infection (CDI) are rarely documented. Therefore, the present study was conducted to investigate the prevalence and subtype characteristics of Blastocystis in patients with suspicion of CDI in Singapore.

Methods

Fecal samples were collected from 248 patients presenting with suspected CDI from a single tertiary hospital in Singapore. C. difficile was diagnosed through positive glutamate dehydrogenase (GDH) with or without toxin A/B using enzyme immunoassay methods. The prevalence and subtype genetic characteristics of Blastocystis were determined by polymerase chain reaction (PCR) amplification and analysis of the barcode region of the SSU rRNA gene.

Results

The proportion of C. difficile in patients with healthcare-associated diarrhea in this study was 44% (109/248). Among the 109 C. difficile-positive patients, 59 (54.1%, 59/109) tested positive for toxigenic C. difficile, which was considered CDI. Based on the sequence analyses of the barcode region of the SSU rRNA gene, 10.1% (25/248) of the patients were found to be Blastocystis-positive, and three subtypes were identified: ST7 (64%, 16/25), ST1 (20%, 5/25), and ST3 (16%, 4/25). Remarkably, we found five patients with Blastocystis and C. difficile coinfection, and further subtype analysis showed two with ST7, two with ST1, and one with ST3.

Conclusions

To the best of our knowledge, this is the first study to investigate the subtype distributions of Blastocystis in patients with CDI in Singapore. We found ST7 to be the predominant subtype in diarrheal patients. The pathogenicity of ST7 has been strongly suggested in previous in vitro and mouse model experiments, further confirming its potential pathogenicity to humans.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13071-021-04749-8.

Evaluation of the accuracy of diagnostic coding for influenza compared to laboratory results: the availability of test results before hospital discharge facilitates improved coding accuracy

Background

Assessing the accuracy of diagnostic coding is essential to ensure the validity and reliability of administrative coded data. The aim of the study was to evaluate the accuracy of assigned International Classification of Diseases version 10-Australian Modification (ICD-10-AM) codes for influenza by comparing with patients’ results of their polymerase chain reaction (PCR)-based laboratory tests.

Method

A retrospective study was conducted across seven public hospitals in New South Wales, Australia. A total of 16,439 patients who were admitted and tested by either cartridge-based rapid PCR or batched multiplex PCR between January 2016 and December 2017 met the inclusion criteria. We calculated the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of ICD-10-AM coding using laboratory results as a gold standard. Separate analyses were conducted to determine whether the availability of test results at the time of hospital discharge influenced diagnostic coding accuracy.

Results

Laboratory results revealed 2759 positive influenza cases, while ICD-10-AM coding identified 2527 patients. Overall, 13.7% (n = 378) of test positive patients were not assigned an ICD-10-AM code for influenza. A further 5.8% (n = 146) patients with negative test results were incorrectly assigned an ICD-10-AM code for influenza. The sensitivity, specificity, PPV and NPV of ICD-10-AM coding were 93.1%; 98.9%; 94.5% and 98.6% respectively when test results were received before discharge and 32.7%; 99.2%; 87.8% and 89.8% respectively when test results were not available at discharge. The sensitivity of ICD-10-AM coding varied significantly across hospitals. The use of rapid PCR or hospitalisation during the influenza season were associated with greater coding accuracy.

Conclusion

Although ICD-10-AM coding for influenza demonstrated high accuracy when laboratory results were received before discharge, its sensitivity was substantially lower for patients whose test results were not available at discharge. The timely availability of laboratory test results during the episode of care could contribute to improved coding accuracy.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12911-021-01531-9.

Value of the International Classification of Diseases code for identifying children with biliary atresia

BACKGROUND

Although identifying cases in large administrative databases may aid future research studies, previous reports demonstrated that the use of the International Classification of Diseases, Tenth Revision (ICD-10) code alone for diagnosis leads to disease misclassification.

PURPOSE

We aimed to assess the value of the ICD-10 diagnostic code for identifying potential children with biliary atresia.

METHODS

Patients aged <18 years assigned the ICD-10 code of biliary atresia (Q44.2) between January 1996 and December 2016 at a quaternary care teaching hospital were identified. We also reviewed patients with other diagnoses of code-defined cirrhosis to identify more potential cases of biliary atresia. A proposed diagnostic algorithm was used to define ICD-10 code accuracy, sensitivity, and specificity.

RESULTS

We reviewed the medical records of 155 patients with ICD-10 code Q44.2 and 69 patients with other codes for biliary cirrhosis (K74.4, K74.5, K74.6). The accuracy for identifying definite/probable/possible biliary atresia cases was 80%, while the sensitivity was 88% (95% confidence interval [CI], 82%-93%). Three independent predictors were associated with algorithm-defined definite/probable/possible cases of biliary atresia: ICD-10 code Q44.2 (odds ratio [OR], 2.90; 95% CI, 1.09-7.71), history of pale stool (OR, 2.78; 95% CI, 1.18-6.60), and a presumed diagnosis of biliary atresia prior to referral to our hospital (OR, 17.49; 95% CI, 7.01-43.64). A significant interaction was noted between ICD-10 code Q44.2 and a history of pale stool (P<0.05). The area under the curve was 0.87 (95% CI, 0.84-0.89).

CONCLUSION

ICD-10 code Q44.2 has an acceptable value for diagnosing biliary atresia. Incorporating clinical data improves the case identification. The use of this proposed diagnostic algorithm to examine data from administrative databases may facilitate appropriate health care allocation and aid future research investigations.

Propensity-Matched Cost of Clostridioides difficile Infection Overdiagnosis

Background

Clostridioides difficile is the leading health care–associated pathogen, but clinicians lack a test that can reliably differentiate colonization from infection. Health care costs attributed to C. difficile are substantial, but the economic burden associated with C. difficile false positives is poorly understood.

Methods

A propensity score matching model for cost per hospitalization was developed to estimate the costs of both true infection and false positives. Predictors of C. difficile positivity used to estimate the propensity score were age, Charlson comorbidity index, white cell count, and creatinine. We used polymerase chain reaction (PCR) cycle threshold to identify and compare 3 groups: (1) true infection, (2) C. difficile colonization, and (3) C. difficile negative.

Results

A positive test was associated with $3018 higher unadjusted hospital cost. Among the 3 comparisons made with propensity-matched negative controls (all positives [+$179; P = .934], true positives [–$1892; P = .100], and colonized positives), only colonization was associated with significantly increased (+$3418; P = .012) cost. Differences in lengths of stay (all positives 0 days, P = .126; true 0 days, P = .919; colonized 1 day, P = .019) appeared to underly cost differences.

Conclusions

In the first C. difficile cost analysis to utilize PCR cycle threshold to differentiate colonization, we found high propensity-matched hospital costs associated with colonized but not true positives. This unexpected finding may be due to misdiagnosis of non–C. difficile diarrhea or unadjusted factors associated with colonization.

Clostridioides difficile infections in Alberta: The validity of administrative data using ICD-10 diagnostic codes for CDI surveillance versus clinical infection surveillance

BACKGROUND

Clostridioides difficile infection (CDI) is one of the most common health care-associated infections. This study assessed the validity of the Discharge Abstract Database (DAD) compared to a traditional clinical surveillance method for identifying CDI.

METHODS

Retrospective analysis of all DAD records with International Statistical Classification of Diseases and Related Health Problems 10th Revision (ie, ICD-10) diagnostic code A04.7 (enterocolitis due to CDI) between April 2015 and March 2019 were compared to a clinical dataset of positive inpatient CDI for all acute care facilities in Alberta, Canada. Sensitivity and positive predictive values were calculated using R version 3.6.0.

RESULTS

The DAD had a sensitivity of 85.0% (95% confidence interval: 84.1%-85.8%) and a positive predictive value of 80.0% (95% confidence interval: 79.2%-80.0%). The CDI rate per 1,000 admissions over the study period was 28% higher in the DAD compared to Infection Prevention and Control surveillance.

DISCUSSION

The DAD does not distinguish symptomatic cases from asymptomatic cases and so indicators to identify symptomatic disease would need to be applied, potentially through a linkage to antibiotic treatment orders available in patient management systems.

CONCLUSIONS

The DAD is moderately sensitive for identifying symptomatic CDI cases in Alberta, Canada and caution should be applied when interpreting rates based on administrative data.

Outcomes of C difficile infection in solid-organ transplant recipients: The National Inpatient Sample (NIS) 2015-2016

BACKGROUND

Clostridioides (formerly Clostridium) difficile infection (CDI) is one of the leading causes of morbidity and mortality worldwide. Solid organ transplant (SOT) recipients are at an increased risk for CDI. A recent study showed an overall improvement in mortality amongst hospitalized individuals with CDI, but it is unclear if this benefit extends to SOT recipients.

METHODS

We scrutinized the 2015 and 2016 National Inpatient Sample (NIS), the largest all-payer inpatient database in the United States for CDI data in patients with SOT. SOT was defined as any recipient who had received a heart, lung, liver, intestinal, kidney, pancreas, or combined thoracic and/or abdominal organ transplantation. Baseline characteristics, comorbidities, and concomitant diagnosis of pneumonia or urinary tract infection were adjusted for in our analysis. Primary outcomes included inpatient mortality, hospital length of stay and total hospital charges.

RESULTS

A total of 105 780 hospital discharges of SOT recipients were included. The incidence of CDI was 3554 (3.36%) among SOTs. CDI was associated with a higher inpatient mortality (OR 1.85, 95% CI 1.56-2.20, P < .01), longer length of hospital stay (mean difference 5.07 days, 95% CI 4.43-5.71, P < .01) and higher total hospital charges (mean difference 43 958 US dollars, P < .01).

CONCLUSION

Our study found that CDI is associated with poorer overall outcomes among hospitalized SOT recipients. However, there was a possible improving trend of the outcomes when compare to previous studies.

Global burden of Clostridium difficile infections: a systematic review and meta-analysis

Background

Clostridium difficile is a leading cause of morbidity and mortality in several countries. However, there are limited evidence characterizing its role as a global public health problem. We conducted a systematic review to provide a comprehensive overview of C. difficile infections (CDI) rates.

Methods

Seven databases were searched (January 2016) to identify studies and surveillance reports published between 2005 and 2015 reporting CDI incidence rates. CDI incidence rates for health care facility-associated (HCF), hospital onset-health care facility-associated, medical or general intensive care unit (ICU), internal medicine (IM), long-term care facility (LTCF), and community-associated (CA) were extracted and standardized. Meta-analysis was conducted using a random effects model.

Results

229 publications, with data from 41 countries, were included. The overall rate of HCF-CDI was 2.24 (95% confidence interval CI = 1.66-3.03) per 1000 admissions/y and 3.54 (95%CI = 3.19-3.92) per 10 000 patient-days/y. Estimated rates for CDI with onset in ICU or IM wards were 11.08 (95%CI = 7.19-17.08) and 10.80 (95%CI = 3.15-37.06) per 1000 admission/y, respectively. Rates for CA-CDI were lower: 0.55 (95%CI = 0.13-2.37) per 1000 admissions/y. CDI rates were generally higher in North America and among the elderly but similar rates were identified in other regions and age groups.

Conclusions

Our review highlights the widespread burden of disease of C. difficile, evidence gaps, and the need for sustainable surveillance of CDI in the health care setting and the community.

Clostridium difficile-associated Diarrhea in Developing Countries: A Systematic Review and Meta-Analysis

Introduction

The prevalence of Clostridium difficile infection is rapidly increasing worldwide, but prevalence is difficult to estimate in developing countries where awareness, diagnostic resources, and surveillance protocols are limited. As diarrhea is the hallmark symptom, we conducted a systematic review and meta-analysis to determine the prevalence and incidence of C. difficile infection in patients in these regions who presented with diarrhea.

Methods

We conducted a systematic literature search of MEDLINE/PubMed, Scopus, and Latin-American and Caribbean Health Sciences Literature databases to identify and analyze data from recent studies providing prevalence or incidence rates of C. difficile-associated diarrhea in developing countries within four regions: Africa–Middle East, developing Asia, Latin America, and China. Our objectives were to determine the current prevalence and incidence density rates of first episodes of C. difficile-associated diarrhea in developing countries.

Results

Within the regions included in our analysis, prevalence of C. difficile infection in patients with diarrhea was 15% (95% CI 13–17%) (including community and hospitalized patients), with no significant difference across regions. The incidence of C. difficile infection in 17 studies including this information was 8.5 per 10,000 patient-days (95% CI 5.83–12.46). Prevalence was significantly higher in hospitalized patients versus community patients (p  = 0.0227).

Conclusion

Our prevalence estimate of 15% is concerning; however, low awareness and inconsistent diagnostic and surveillance protocols suggest this is markedly underestimated. Enhanced awareness and management of C. difficile infection in patients with diarrhea, along with improvements in infection control and surveillance practices, should be implemented to reduce prevalence of C. difficile-associated diarrhea in developing countries.

Funding

Pfizer Inc.

Electronic supplementary material

The online version of this article (10.1007/s40121-019-0231-8) contains supplementary material, which is available to authorized users.

Retrospective analysis of long-term gastrointestinal symptoms after Clostridium difficile infection in a nonelderly cohort

Elderly patients and those with comorbid conditions are at high risk for poor outcomes after Clostridium difficile infection (CDI) but outcomes in a healthier, nonelderly population are not well described. We sought to investigate gastrointestinal diagnoses and CDI during hospitalizations in the 24 to 36 months after an initial episode of CDI in nonelderly patients in a cohort with an overall low prevalence of comorbid conditions. We performed a retrospective analysis of hospital admissions from 2010–2013 using the Truven MarketScan database of employment-based private insurance claims. Subjects <65 years of age and their adult dependents (> = 18 years old); a CDI diagnosis in 2011 (index date); at least 12 months of pre-index continuous enrollment; and 24–36 months of continuous post-index enrollment were included. The 12 months of each subject’s enrollment prior to the index date for a CDI served as the reference period for the analyses of that subject’s post-CDI time periods. Hospital claims during the follow-up period were evaluated for gastrointestinal diagnoses and/or CDI ICD-9 codes. The risk of gastrointestinal diagnoses was assessed using Cox proportional hazards models adjusted for a pre-specified set of baseline demographic and clinical factors. During 2011, 5,632 subjects with CDI met the inclusion criteria for our study. The risk of gastrointestinal diagnoses in patients with a CDI diagnostic code for the same admission was almost 8-fold higher 3 months post-CDI (hazard ratio (HR) = 7.56; 95% confidence interval (CI): 2.97–19.19) than for subjects without CDI and remained statistically significant until month 24 (HR = 1.47; 95% CI = 1.04–2.08). After CDI, patients remained at risk for gastrointestinal symptoms with CDI for up to two years. There is an important, long-term healthcare burden after CDI in this population.

Validity and Reliability of Administrative Coded Data for the Identification of Hospital-Acquired Infections: An Updated Systematic Review with Meta-Analysis and Meta-Regression Analysis

OBJECTIVE

To conduct an updated assessment of the validity and reliability of administrative coded data (ACD) in identifying hospital-acquired infections (HAIs).

METHODS

We systematically searched three libraries for studies on ACD detecting HAIs compared to manual chart review. Meta-analyses were conducted for prosthetic and nonprosthetic surgical site infections (SSIs), Clostridium difficile infections (CDIs), ventilator-associated pneumonias/events (VAPs/VAEs) and non-VAPs/VAEs, catheter-associated urinary tract infections (CAUTIs), and central venous catheter-related bloodstream infections (CLABSIs). A random-effects meta-regression model was constructed.

RESULTS

Of 1,906 references found, we retrieved 38 documents, of which 33 provided meta-analyzable data (N = 567,826 patients). ACD identified HAI incidence with high specificity (≥93 percent), prosthetic SSIs with high sensitivity (95 percent), and both CDIs and nonprosthetic SSIs with moderate sensitivity (65 percent). ACD exhibited substantial agreement with traditional surveillance methods for CDI (κ = 0.70) and provided strong diagnostic odds ratios (DORs) for the identification of CDIs (DOR = 772.07) and SSIs (DOR = 78.20). ACD performance in identifying nosocomial pneumonia depended on the ICD coding system (DORICD-10/ICD-9-CM  = 0.05; p = .036). Algorithmic coding improved ACD's sensitivity for SSIs up to 22 percent. Overall, high heterogeneity was observed, without significant publication bias.

CONCLUSIONS

Administrative coded data may not be sufficiently accurate or reliable for the majority of HAIs. Still, subgrouping and algorithmic coding as tools for improving ACD validity deserve further investigation, specifically for prosthetic SSIs. Analyzing a potential lower discriminative ability of ICD-10 coding system is also a pending issue.

Risk factors for Clostridium difficile infections - an overview of the evidence base and challenges in data synthesis

Background

Recognition of a broad spectrum of disease and development of Clostridium difficile infection (CDI) and recurrent CDI (rCDI) in populations previously considered to be at low risk has renewed attention on differences in the risk profile of patients. In the absence of primary prevention for CDI and limited treatment options, it is important to achieve a deep understanding of the multiple factors that influence the risk of developing CDI and rCDI.

Methods

We conducted a review of systematic reviews and meta–analyses on risk factors for CDI and rCDI published between 1990 and October 2016.

Results

22 systematic reviews assessing risk factors for CDI (n = 19) and rCDI (n = 6) were included. Meta–analyses were conducted in 17 of the systematic reviews. Over 40 risk factors have been associated with CDI and rCDI and can be classified into three categories: pharmacological risk factors, host–related risk factors, and clinical characteristics or interventions. Most systematic reviews and meta–analyses have focused on antibiotic use (n = 8 for CDI, 3 for rCDI), proton pump inhibitors (n = 8 for CDI, 4 for rCDI), and histamine 2 receptor antagonists (n = 4 for CDI) and chronic kidney disease (n = 4 for rCDI). However, other risk factors have been assessed. We discuss the state of the evidence, methods, and challenges for data synthesis.

Conclusion

Several studies, synthesized in different systematic review, provide valuable insights into the role of different risk factors for CDI. Meta–analytic evidence of association has been reported for factors such as antibiotics, gastric acid suppressants, non–selective NSAID, and some co–morbidities. However, despite statistical significance, issues of high heterogeneity, bias and confounding remain to be addressed effectively to improve overall risk estimates. Large, prospective primary studies on risk factors for CDI with standardised case definitions and stratified analyses are required to develop more accurate and robust estimates of risk effects that can inform targeted–CDI clinical management procedures, prevention, and research.

Learning bundled care opportunities from electronic medical records

OBJECTIVE

The traditional fee-for-service approach to healthcare can lead to the management of a patient's conditions in a siloed manner, inducing various negative consequences. It has been recognized that a bundled approach to healthcare - one that manages a collection of health conditions together - may enable greater efficacy and cost savings. However, it is not always evident which sets of conditions should be managed in a bundled manner. In this study, we investigate if a data-driven approach can automatically learn potential bundles.

METHODS

We designed a framework to infer health condition collections (HCCs) based on the similarity of their clinical workflows, according to electronic medical record (EMR) utilization. We evaluated the framework with data from over 16,500 inpatient stays from Northwestern Memorial Hospital in Chicago, Illinois. The plausibility of the inferred HCCs for bundled care was assessed through an online survey of a panel of five experts, whose responses were analyzed via an analysis of variance (ANOVA) at a 95% confidence level. We further assessed the face validity of the HCCs using evidence in the published literature.

RESULTS

The framework inferred four HCCs, indicative of (1) fetal abnormalities, (2) late pregnancies, (3) prostate problems, and (4) chronic diseases, with congestive heart failure featuring prominently. Each HCC was substantiated with evidence in the literature and was deemed plausible for bundled care by the experts at a statistically significant level.

CONCLUSIONS

The findings suggest that an automated EMR data-driven framework conducted can provide a basis for discovering bundled care opportunities. Still, translating such findings into actual care management will require further refinement, implementation, and evaluation.

Hospital discharge abstracts have limited accuracy in identifying occurrence of Clostridium difficile infections among hospitalized individuals with inflammatory bowel disease: A population-based study

BACKGROUND

Hospital discharge databases are used to study the epidemiology of Clostridium difficile infections (CDI) among hospitalized patients with inflammatory bowel disease (IBD). CDI in IBD is increasingly important and accurately estimating its occurrence is critical in understanding its comorbidity. There are limited data on the reliability of the International Classification of Diseases 10th revision (ICD-10) (now widely used in North America) CDI code in determining occurrence of CDI among hospitalized patients. We compared the performance of ICD-10 CDI coding to laboratory confirmed CDI diagnoses.

METHODS

The University of Manitoba IBD Epidemiology Database was used to identify individuals with and without IBD discharged with CDI diagnoses between 07/01/2005 and 3/31/2014. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of ICD-10 CDI code was compared to laboratory CDI diagnoses recorded in a province wide CDI dataset. Multivariable logistic regression models were performed to test the predictors of diagnostic inaccuracy of ICD-10 CDI code.

RESULTS

There were 273 episodes of laboratory confirmed CDI (hospitalized and non-hospitalized) among 7396 individuals with IBD and 536 among 66,297 matched controls. The sensitivity, specificity, PPV and NPV of ICD-10 CDI code in discharge abstracts was 72.8%, 99.6%, 64.1% and 99.7% among those with IBD and 70.8%, 99.9%, 79.0% and 99.9% among those without IBD. Predictors of diagnostic inaccuracy included IBD, older age, increased co-morbidity and earlier years of hospitalization.

CONCLUSIONS

Identification of CDI using ICD-10 CDI code in hospital discharge abstracts may not identify up to 30% of CDI cases, with worse performance among those with IBD.

Validation of the V66.7 Code for Palliative Care Consultation in a Single Academic Medical Center

BACKGROUND

Use of administrative data to study the effectiveness of specialized palliative care is limited by the lack of a reliable method to identify patients receiving palliative care consultation. The International Classification of Diseases, Ninth Revision (ICD-9) code V66.7 has been used, but its validity for this purpose is unknown.

OBJECTIVE

To examine the validity of the ICD-9 code V66.7 for identifying whether hospitalized patients received palliative care consultation.

DESIGN

Retrospective cohort study.

SETTING/SUBJECTS

All patients of age ≥18 years admitted to a single academic medical center between August 2013 and August 2015.

MEASUREMENTS

Sensitivity and specificity of the V66.7 code for palliative care consultation for all patients and several a priori identified subgroups. The reference standard was the presence of a palliative care consultation note in the electronic medical record.

RESULTS

Of 100,910 admissions, 1999 received a palliative care consultation (2.0%) and 1846 (1.8%) had usage of the V66.7 code. Sensitivity and specificity for the V66.7 code were 49.9% and 99.1%, respectively. Sensitivity was considerably higher for certain subgroups, such as patients with dementia (76.3%) and metastatic cancer (66.3%); addition of age restrictions further improved sensitivity while maintaining high specificity. Specificity was substantially lower for patients who died during hospitalization (sensitivity 53.9%, specificity 75.1%).

CONCLUSIONS

In a single center, the ICD-9 code V66.7 had poor sensitivity and high specificity for identifying hospitalized patients who received a palliative care consultation. Appropriate use of this code for this purpose should take these characteristics into consideration.

Effects of antiviral treatment on influenza-related complications over four influenza seasons: 2006-2010

PURPOSE

The objective of the study is to evaluate the effect of antiviral treatment, pre-existing diseases, and sociodemographic factors on the risk of influenza-related complications and healthcare utilization.

METHODS

Case data was obtained from U.S. MarketScan Research Databases. Cases had a clinical diagnosis of influenza between 2006 and 2010 and continuous healthcare insurance from 90 days before to 30 days after diagnosis. Logistic regression models were applied to explore the impact of antiviral treatment on complications and healthcare utilization. Modified generalized estimating equation regression models in propensity score matched samples were used to address the robustness of the study.

RESULTS

Analyses included 1,557,437 cases from four influenza seasons. In each season, 34.82%-43.42% of patients received antiviral treatment, mostly oseltamivir. On average, 1.86% of patients were hospitalized, 9.56% visited the emergency room and 41.14% made ≥2 outpatient visits. The incidence of complications ranged from 17.62 to 19.67 per 100 patient-months. The relative risk of complications was increased in patients aged 0-4 years and those with pre-existing diseases, including asthma, Parkinson's disease, and cystic fibrosis. Overall, patients receiving antiviral treatment had an 11% reduction in the risk of complications. Among oseltamivir-treated patients, the risk of complications was significantly reduced by 81% in those treated ≤2 days after diagnosis compared with later. Antiviral treatment significantly reduced the risk of hospitalization, emergency room visits and need for ≥2 outpatient visits by 29%, 24% and 11%, respectively. The propensity score matching method improved the strength of the study.

CONCLUSIONS

Early treatment with antivirals, and specifically oseltamivir, significantly reduced the risk of influenza-related complications and healthcare utilization. However, lacking information about disease severity and the time from onset of symptoms to fulfillment of a prescription may bias the outcomes.

Electronic health records (EHRs): supporting ASCO's vision of cancer care

ASCO's vision for cancer care in 2030 is built on the expanding importance of panomics and big data, and envisions enabling better health for patients with cancer by the rapid transformation of systems biology knowledge into cancer care advances. This vision will be heavily dependent on the use of health information technology for computational biology and clinical decision support systems (CDSS). Computational biology will allow us to construct models of cancer biology that encompass the complexity of cancer panomics data and provide us with better understanding of the mechanisms governing cancer behavior. The Agency for Healthcare Research and Quality promotes CDSS based on clinical practice guidelines, which are knowledge bases that grow too slowly to match the rate of panomic-derived knowledge. CDSS that are based on systems biology models will be more easily adaptable to rapid advancements and translational medicine. We describe the characteristics of health data representation, a model for representing molecular data that supports data extraction and use for panomic-based clinical research, and argue for CDSS that are based on systems biology and are algorithm-based.

Identification of Recurrent Clostridium difficile Infection Using Administrative Codes: Accuracy and Implications for Surveillance

OBJECTIVE

To develop an algorithm using administrative codes, laboratory data, and medication data to identify recurrent Clostridium difficile infection (CDI) and to examine the sensitivity, specificity, positive and negative predictive values, and performance of this algorithm.

METHODS

We identified all patients with 2 or more International Classification of Diseases, Ninth Edition, Clinical Modification (ICD-9-CM) codes for CDI (008.45) from January 1 through December 31, 2013. Information on number of diagnosis codes, stool toxin assays (enzyme immunoassay or polymerase chain reaction), and unique prescriptions for metronidazole and vancomycin was identified. Logistic regression was used to identify independent predictors of recurrent CDI and a predictive model was developed.

RESULTS

A total of 591 patients with at least 2 ICD-9 codes for CDI were included (median age, 66 years). The derivation cohort consisted of 157 patients among whom 43 (27%) had recurrent CDI. Presence of 3 or more ICD-9 codes for CDI (odds ratio, 2.49), 2 or more stool tests (odds ratio, 2.88), and 2 or more prescriptions for vancomycin (odds ratio, 5.87) were independently associated with confirmed recurrent CDI. A classifier incorporating 2 or more prescriptions for vancomycin and either 2 or more stool tests or 3 or more ICD-9-CM codes had a positive predictive value of 41% and negative predictive value of 90%. The area under the receiver operating characteristic curve for this combined classifier was modest (0.69).

CONCLUSION

Identification of recurrent episodes of CDI in administrative data poses challenges. Accurate assessment of burden requires individual case review to confirm diagnosis.

Infect Control Hosp Epidemiol 2015;36(8):893–898

Building bridges across electronic health record systems through inferred phenotypic topics

OBJECTIVE

Data in electronic health records (EHRs) is being increasingly leveraged for secondary uses, ranging from biomedical association studies to comparative effectiveness. To perform studies at scale and transfer knowledge from one institution to another in a meaningful way, we need to harmonize the phenotypes in such systems. Traditionally, this has been accomplished through expert specification of phenotypes via standardized terminologies, such as billing codes. However, this approach may be biased by the experience and expectations of the experts, as well as the vocabulary used to describe such patients. The goal of this work is to develop a data-driven strategy to (1) infer phenotypic topics within patient populations and (2) assess the degree to which such topics facilitate a mapping across populations in disparate healthcare systems.

METHODS

We adapt a generative topic modeling strategy, based on latent Dirichlet allocation, to infer phenotypic topics. We utilize a variance analysis to assess the projection of a patient population from one healthcare system onto the topics learned from another system. The consistency of learned phenotypic topics was evaluated using (1) the similarity of topics, (2) the stability of a patient population across topics, and (3) the transferability of a topic across sites. We evaluated our approaches using four months of inpatient data from two geographically distinct healthcare systems: (1) Northwestern Memorial Hospital (NMH) and (2) Vanderbilt University Medical Center (VUMC).

RESULTS

The method learned 25 phenotypic topics from each healthcare system. The average cosine similarity between matched topics across the two sites was 0.39, a remarkably high value given the very high dimensionality of the feature space. The average stability of VUMC and NMH patients across the topics of two sites was 0.988 and 0.812, respectively, as measured by the Pearson correlation coefficient. Also the VUMC and NMH topics have smaller variance of characterizing patient population of two sites than standard clinical terminologies (e.g., ICD9), suggesting they may be more reliably transferred across hospital systems.

CONCLUSIONS

Phenotypic topics learned from EHR data can be more stable and transferable than billing codes for characterizing the general status of a patient population. This suggests that EHR-based research may be able to leverage such phenotypic topics as variables when pooling patient populations in predictive models.

Ambulatory-treated Clostridium difficile infection: a comparison of community-acquired vs. nosocomial infection

The purpose of this study was to identify the clinical outcomes of ambulatory-treated Clostridium difficile infection (CDI) and risk factors associated with community-associated CDI (CA-CDI). Adult patients diagnosed with CDI in the institutional or ambulatory-care setting between 1 April 2005 and 30 April 2011, with no other CDI diagnosis in the previous 180 days, and who purchased an ambulatory, anti-CDI agent within 7 days of CDI diagnosis were included. A total of 1201 patients were included with 914 (76%) and 287 (24%) identified with CA-CDI and nosocomial CDI (N-CDI), respectively. Patients with N-CDI were more likely to have had a recurrent CDI (P = 0·043) and died from any cause (P < 0·001). Patients with CA-CDI were younger, healthier, and had fewer traditional risk factors compared to patients with N-CDI. To prevent CA-CDI, clinicians should be aware that patients at risk for CA-CDI are unique from those at risk for N-CDI.

Use of the i2b2 research query tool to conduct a matched case-control clinical research study: advantages, disadvantages and methodological considerations

Background

A major aim of the i2b2 (informatics for integrating biology and the bedside) clinical data informatics framework aims to create an efficient structure within which patients can be identified for clinical and translational research projects.

Our objective was to describe the respective roles of the i2b2 research query tool and the electronic medical record (EMR) in conducting a case-controlled clinical study at our institution.

Methods

We analyzed the process of using i2b2 and the EMR together to generate a complete research database for a case–control study that sought to examine risk factors for kidney stones among gastrostomy tube (G-tube) fed children.

Results

Our final case cohort consisted of 41/177 (23%) of potential cases initially identified by i2b2, who were matched with 80/486 (17%) of potential controls. Cases were 10 times more likely to be excluded for inaccurate coding regarding stones vs. inaccurate coding regarding G-tubes. A majority (67%) of cases were excluded due to not meeting clinical inclusion criteria, whereas a majority of control exclusions (72%) occurred due to inadequate clinical data necessary for study completion. Full dataset assembly required complementary information from i2b2 and the EMR.

Conclusions

i2b2 was critical as a query analysis tool for patient identification in our case–control study. Patient identification via procedural coding appeared more accurate compared with diagnosis coding. Completion of our investigation required iterative interplay of i2b2 and the EMR to assemble the study cohort.

Temporal phenome analysis of a large electronic health record cohort enables identification of hospital-acquired complications

OBJECTIVE

To develop methods for visual analysis of temporal phenotype data available through electronic health records (EHR).

MATERIALS AND METHODS

24 580 adults from the multiparameter intelligent monitoring in intensive care V.6 (MIMIC II) EHR database of critically ill patients were analyzed, with significant temporal associations visualized as a map of associations between hospital length of stay (LOS) and ICD-9-CM codes. An expanded phenotype, using ICD-9-CM, microbiology, and computerized physician order entry data, was defined for hospital-acquired Clostridium difficile (HA-CDI). LOS, estimated costs, 30-day post-discharge mortality, and antecedent medication provider order entry were evaluated for HA-CDI cases compared to randomly selected controls.

RESULTS

Temporal phenome analysis revealed 191 significant codes (p value, adjusted for false discovery rate, ≤0.05). HA-CDI was identified in 414 cases, and was associated with longer median LOS, 20 versus 9 days, and adjusted HR 0.33 (95% CI 0.28 to 0.39). This prolongation carries an estimated annual incremental cost increase of US$1.2-2.0 billion in the USA alone.

DISCUSSION

Comprehensive EHR data have made large-scale phenome-based analysis feasible. Time-dependent pathological disease states have dynamic phenomic evolution, which may be captured through visual analytical approaches. Although MIMIC II is a single institutional retrospective database, our approach should be portable to other EHR data sources, including prospective 'learning healthcare systems'. For example, interventions to prevent HA-CDI could be dynamically evaluated using the same techniques.

CONCLUSIONS

The new visual analytical method described in this paper led directly to the identification of numerous hospital-acquired conditions, which could be further explored through an expanded phenotype definition.

Accuracy of ICD-10 codes for surveillance of Clostridium difficile infections, France

The sensitivity and specificity of surveillance for Clostridium difficile infections according to International Classification of Diseases, 10th revision, codes were compared with laboratory results as standard. Sensitivity was 35.6%; specificity was 99.9%. Concordance between the 2 methods was moderate. Surveillance based on ICD-10 codes underestimated the rate based on laboratory results.