Design and study setting
The study consists in a two-years retrospective observational analysis of all the ID-cons provided in a large tertiary hospital in Milan, Italy. San Carlo Borromeo Hospital (HSC) is a non-teaching public hospital with 494 beds and 20,000 admissions/years. HSC does not contain a transplant or haematology-oncology unit but does have a neurosurgery unit. Thus, possible source and aetiology of infections might account for these characteristics. Additionally, HSC does not have an ID-unit but rather a consult service that is staffed by ID-specialists from the Unit of Infectious Diseases of San Paolo Hospital since 2016.
Infectious Diseases consultations were provided once a week from September 2016 to September 2017 (weekly ID-cons period) while from October 2017 the service of ID-consultation was provided on a daily basis (daily ID-cons period). All the ID-cons were performed by the same team of ID-consultants. ID-cons is defined as any request by the treating non-ID physician for ID advice with bed-side evaluation of the patient resulting in a written statement by the ID consultant.
This observational analysis received approval by our local Ethics committee (Comitato Etico Milano Area 1) in 01/17/2019 with number of protocol: 57183/2018. Written informed consent was obtained for enrolled patients.
Study procedures and definitions
We collected in a dedicated database demographics, clinical conditions and microbiological findings of all the hospitalized patients for whom an ID consultation was asked, from September 1st, 2016 to October 31st, 2018. Patients’ medical histories and mortality were reviewed by clinical charts to identify risk factors of infections (alcoholism, radio-chemotherapy, use of steroids, injection drug use), and comorbidities (cardio-vascular disease, dementia, liver cirrhosis, cancer, chronic obstructive pulmonary disease, chronic renal failure, diabetes, HCV). Comorbidities were evaluated according to the Age-adjusted Charlson Comorbidity Index (ACCI), a validate prognostic tool that predict the risk of death of patients with several comorbidities [16].
Hospital Units were grouped into three departments:
- Medical Department, which includes: Cardiology, Gastroenterology, Internal Medicine Units, Oncology, Nephrology, Neurology, Pulmonary, Psychiatry and Rehabilitation Unit;
- Surgical Department: General Surgery, Neurosurgery, Obstetrics/gynecology, Orthopedic Unit, Urology and Vascular Surgery;
- Emergency Department: Coronary Unit, Intensive and Sub-intensive Units, Emergency Rooms, Emergency Medicine Unit and Stroke Unit.
Infections were classified as: 1) healthcare-associated infections (HAI), in case of infections associated with hospitalization or other medical treatment that appear 48 h or more after hospital admission; 2) community-acquired infections (CAI), in case of patients admitted for an infection acquired before hospitals’ admission [17]. Classification of an infection into one of the two groups, HAI or CAI, was made by the Id consultant by combining clinical presentation with radiological and microbiological findings.
At each consultation all the antimicrobial therapies were reviewed and discussed with the treating physician. Interventions on antibiotic therapies were collected into the database according to the following classification: start of ABT, no need of ABT, confirmation of ABT and modification of ABT (including dosage optimization, change of ABT, de-escalation, intensification and discontinuation of ABT). De-escalation therapy was defined as: i) switching from combination to monotherapy; ii) narrowing spectrum of activity. An opposite definition was applied to intensification therapy.
A therapy was considered appropriate in case of “confirmation of ABT” by the ID consultant in terms of dose, duration, penetrability and choice of regimen. In case of microbiological findings, appropriateness was assessed based on of in vitro susceptibility data. Assessment of appropriateness was performed on the basis of internal guidelines that refers in turn to national and international guidelines. Conversely, “modification of ABT” was considered a marker of inappropriate therapy. In order to evaluate the appropriateness of antibiotics prescribed by the non-ID specialist physician, only first consultations per patient with an already ongoing antibiotic therapy were analyzed.
Inclusion/exclusion criteria and study period
All the hospitalized patients for whom the treating physician asked for ID consultation, from September 1st, 2016 to October 31st, 2018 were included in the. Age < 18 years was the only criteria for exclusion.
We evaluated the impact of one year of weekly service of ID-cons (09/2016-09/2017) versus one year of daily service of ID-cons (10/2017-10/2018).
Outcomes
Process outcomes estimate the performance of the study. Process outcomes of the study are: i) number-of-ID-cons per 100 bed days (bd), ii) days-from-admission-to-first-ID-cons, iii) type of ABT-intervention and iv) appropriateness of ABT prescription (evaluated only on first ID evaluation).
Primary outcomes of the study are: (i) the reduction of overall ABT consumption and (ii) the reduction of ABT consumption by department and by ABT classes expressed as defined-daily-dose (DDD)/100bd.
The secondary outcome is a non-significant increment of overall and sepsis-related in-hospital mortality (as death/patient’s admissions) from 2017 to 2018.
Statistical analysis
Categorical variables were presented as absolute numbers (percentages) while continuous variables as median (interquartile range; IQR). Clinical characteristics of patients were represented as median and interquartile range. Chi-square and Wilcoxon test were used as appropriate. ATB consumption was expressed as DDD/100bd. In order to evaluate the impact of our intervention on the outcome a sensitivity analysis including units with high number of ID cons/100bd (≥ 25th percentile of the ID-cons distribution) was performed. Differences in patients’ characteristics and process outcomes were evaluated by Mann-Whitney and Chi-square tests while differences in antimicrobial consumption between the two time periods were evaluated by Wilcoxon test for paired data. A p-value < 0.05 was considered statistically significant. Statistical analyses were performed with SAS software (version 9.2).