Association between Charlson Comorbidity Index and positive blood cultures at a tertiary-care hospital in Indonesia

TL;DR

This study found no link between comorbidities and positive blood cultures in hospitalized patients in Indonesia, suggesting other factors should guide testing.

Contribution

The study investigates the utility of the Charlson Comorbidity Index for predicting positive blood cultures in a low-resource setting.

Findings

No significant association was found between CCI score and positive blood cultures after adjusting for confounders.

Common pathogens included Staphylococcus aureus, Acinetobacter spp., and Klebsiella pneumoniae.

Antibiotic use prior to testing may modify the relationship between CCI and blood culture positivity.

Abstract

Blood culture (BC) tests are a scarce resource in low- and middle-income countries (LMICs); therefore, prioritization based on likelihood of positive results might be beneficial. We aimed to determine whether comorbidities in the Charlson Comorbidity Index (CCI) were associated with positive BC tests among patients with suspected hospital-acquired bacteremia. We analysed a retrospective cohort from health records at Dr. Wahidin Sudirohusodo Hospital, Makassar, Indonesia from 2015-2018. We applied multivariable logistic regression to identify associations between CCI score and the outcome of the first BC taken two calendar days after admission, adjusting for confounders. The primary analysis considered BCs positive for all pathogens. Of 3,875 adult patients who had their first BCs taken two calendar days after hospital admissions, 786 (20.3%) had their first BCs positive for any pathogen. Those included 371 patients who had their first BCs positive for Staphylococcus aureus (n = 133; 35.9%), Acinetobacter spp. (n = 84; 22.6%), Klebsiella. pneumoniae (n = 58; 15.6%), Escherichia coli (n = 63; 17.0%) and Pseudomonas aeruginosa (n = 33; 8.9%). There was no association between increasing CCI score and positive BC (OR 1.01, 95%CI: 0.96-1.06, p = 0.69) after adjustment for age, sex and other potential confounders. There was some indication that antibiotic use prior to BC test acted as an effect modifier between CCI score and positivity of BC (p = 0.17). In this single-hospital study, no significant association was observed between CCI score and positive BC taken two calendar days after hospital admission. We suggest that other factors need to be investigated to guide BC testing, and that improving diagnostic and antibiotic stewardship, including increasing resources for BC testing prior to antibiotics among hospitalized patients are needed in LMICs.

Figures