Estimating Local Costs Associated With Clostridium difficile Infection Using Machine Learning and Electronic Medical Records

Modest Clostridiodes difficile infection prediction using machine learning models in a tertiary care hospital

Previous studies have shown promising results of machine learning (ML) models for predicting health outcomes. We develop and test ML models for predicting Clostridioides difficile infection (CDI) in hospitalized patients. This is a retrospective cohort study conducted during 2015-2017. All inpatients tested for C. difficile were included. CDI was defined as having a positive glutamate dehydrogenase and toxin results. We restricted analyses to the first record of C. difficile testing per patient. Of 3514 patients tested, 136 (4%) had CDI. Age and antibiotic use within 90 days before C. difficile testing were associated with CDI (P < 0.01). We tested 10 ML methods with and without resampling. Logistic regression, random forest and naïve Bayes models yielded the highest AUC ROC performance: 0.6. Predicting CDI was difficult in our cohort of patients tested for CDI. Multiple ML models yielded only modest results in a real-world population of hospitalized patients tested for CDI.

Machine learning in infection management using routine electronic health records: tools, techniques, and reporting of future technologies

BACKGROUND

Machine learning (ML) is increasingly being used in many areas of health care. Its use in infection management is catching up as identified in a recent review in this journal. We present here a complementary review to this work.

OBJECTIVES

To support clinicians and researchers in navigating through the methodological aspects of ML approaches in the field of infection management.

SOURCES

A Medline search was performed with the keywords artificial intelligence, machine learning, infection∗, and infectious disease∗ for the years 2014-2019. Studies using routinely available electronic hospital record data from an inpatient setting with a focus on bacterial and fungal infections were included.

CONTENT

Fifty-two studies were included and divided into six groups based on their focus. These studies covered detection/prediction of sepsis (n = 19), hospital-acquired infections (n = 11), surgical site infections and other postoperative infections (n = 11), microbiological test results (n = 4), infections in general (n = 2), musculoskeletal infections (n = 2), and other topics (urinary tract infections, deep fungal infections, antimicrobial prescriptions; n = 1 each). In total, 35 different ML techniques were used. Logistic regression was applied in 18 studies followed by random forest, support vector machines, and artificial neural networks in 18, 12, and seven studies, respectively. Overall, the studies were very heterogeneous in their approach and their reporting. Detailed information on data handling and software code was often missing. Validation on new datasets and/or in other institutions was rarely done. Clinical studies on the impact of ML in infection management were lacking.

IMPLICATIONS

Promising approaches for ML use in infectious diseases were identified. But building trust in these new technologies will require improved reporting. Explainability and interpretability of the models used were rarely addressed and should be further explored. Independent model validation and clinical studies evaluating the added value of ML approaches are needed.

Incidence and Outcomes Associated With Clostridium difficile Infections: A Systematic Review and Meta-analysis

IMPORTANCE

An understanding of the incidence and outcomes of Clostridium difficile infection (CDI) in the United States can inform investments in prevention and treatment interventions.

OBJECTIVE

To quantify the incidence of CDI and its associated hospital length of stay (LOS) in the United States using a systematic literature review and meta-analysis.

DATA SOURCES

MEDLINE via Ovid, Cochrane Library Databases via Wiley, Cumulative Index of Nursing and Allied Health Complete via EBSCO Information Services, Scopus, and Web of Science were searched for studies published in the United States between 2000 and 2019 that evaluated CDI and its associated LOS.

STUDY SELECTION

Incidence data were collected only from multicenter studies that had at least 5 sites. The LOS studies were included only if they assessed postinfection LOS or used methods accounting for time to infection using a multistate model or compared propensity score-matched patients with CDI with control patients without CDI. Long-term-care facility studies were excluded. Of the 119 full-text articles, 86 studies (72.3%) met the selection criteria.

DATA EXTRACTION AND SYNTHESIS

Two independent reviewers performed the data abstraction and quality assessment. Incidence data were pooled only when the denominators used the same units (eg, patient-days). These data were pooled by summing the number of hospital-onset CDI incident cases and the denominators across studies. Random-effects models were used to obtain pooled mean differences. Heterogeneity was assessed using the I2 value. Data analysis was performed in February 2019.

MAIN OUTCOMES AND MEASURES

Incidence of CDI and CDI-associated hospital LOS in the United States.

RESULTS

When the 13 studies that evaluated incidence data in patient-days due to hospital-onset CDI were pooled, the CDI incidence rate was 8.3 cases per 10 000 patient-days. Among propensity score-matched studies (16 of 20 studies), the CDI-associated mean difference in LOS (in days) between patients with and without CDI varied from 3.0 days (95% CI, 1.44-4.63 days) to 21.6 days (95% CI, 19.29-23.90 days).

CONCLUSIONS AND RELEVANCE

Pooled estimates from currently available literature suggest that CDI is associated with a large burden on the health care system. However, these estimates should be interpreted with caution because higher-quality studies should be completed to guide future evaluations of CDI prevention and treatment interventions.

Direct cost of health care for individuals with community associated Clostridium difficile infections: A population-based cohort study

Background

Even though the incidence of community-acquired Clostridium difficile infection (CDI) is reported to be increasing, few studies have reported on the healthcare costs of community-acquired CDI. We estimated cost of care for individuals with community-associated CDI and compared with that for matched controls without CDI in the time period of six months before to one year after CDI.

Methods

All individuals in the province of Manitoba, diagnosed with CDI between July 2005 and March 2015 were matched up to 4 individuals without CDI. Health care utilization and direct costs resulting from hospitalizations, physician reimbursement claims and prescriptions were determined from the population based provincial databases. Quantile regressions were performed to determine predictors of cost of individuals with community associated CDI.

Results

Of all CDIs, 30–40% in each period of the study had community-associated CDI; of which 12% were recurrent CDIs. The incremental median and 90^th percentile cost of care for individuals with community-associated CDI was $800 and $16,000 respectively in the six months after CDI diagnosis. After adjustment for age, co-morbidities, sex, socioeconomic status and magnitude of health care utilization prior to CDI, the median incremental cost for recurrent CDI was $1,812 and that for a subsequent episode of CDI was $3,139 compared to those with a single community-associated CDI episode. The median cost for a prescription of Vancomycin was $316 (IQR 209–489).

Conclusions

Health care costs of an episode of community-associated CDI have been much more than the cost of antibiotic treatment. Our study provides population-based data for formal cost effectiveness analysis for use of newer treatments for community-associated CDI.

Introduction to Machine Learning in Digital Healthcare Epidemiology

To exploit the full potential of big routine data in healthcare and to efficiently communicate and collaborate with information technology specialists and data analysts, healthcare epidemiologists should have some knowledge of large-scale analysis techniques, particularly about machine learning. This review focuses on the broad area of machine learning and its first applications in the emerging field of digital healthcare epidemiology.