Burden of pneumococcal pneumonia requiring ICU admission in France: 1-year prognosis, resources use, and costs

Background Community-acquired pneumonia (CAP), especially pneumococcal CAP (P-CAP), is associated with a heavy burden of illness as evidenced by high rates of intensive care unit (ICU) admission, mortality, and costs. Although well-defined acutely, determinants influencing long-term burden are less known. This study assessed determinants of 28-day and 1-year mortality and costs among P-CAP patients admitted in ICUs. Methods Data regarding all hospital and ICU stays in France in 2014 were extracted from the French healthcare administrative database. All patients admitted in the ICU with a pneumonia diagnosis were included, except those hospitalized for pneumonia within the previous 3 months. The pneumococcal etiology and comorbidities were captured. All hospital stays were included in the cost analysis. Comorbidities and other factors effect on the 28-day and 1-year mortality were assessed using a Cox regression model. Factors associated with increased costs were identified using log-linear regression models. Results Among 182,858 patients hospitalized for CAP in France for 1 year, 10,587 (5.8%) had a P-CAP, among whom 1665 (15.7%) required ICU admission. The in-hospital mortality reached 22.8% at day 28 and 32.3% at 1 year. The mortality risk increased with age > 54 years, malignancies (hazard ratio (HR) 1.54, 95% CI [1.23–1.94], p = 0.0002), liver diseases (HR 2.08, 95% CI [1.61–2.69], p < 0.0001), and the illness severity at ICU admission. Compared with non-ICU-admitted patients, ICU survivors remained at higher risk of 1-year mortality. Within the following year, 38.2% (516/1350) of the 28-day survivors required at least another hospital stay, mostly for respiratory diseases. The mean cost of the initial stay was €19,008 for all patients and €11,637 for subsequent hospital stays within 1 year. One-year costs were influenced by age (lower in patients > 75 years old, p = 0.008), chronic cardiac (+ 11% [0.02–0.19], p = 0.019), and respiratory diseases (+ 11% [0.03–0.18], p = 0.006). Conclusions P-CAP in ICU-admitted patients was associated with a heavy burden of mortality and costs at one year. Older age was associated with both early and 1-year increased mortality. Malignant and chronic liver diseases were associated with increased mortality, whereas chronic cardiac failure and chronic respiratory disease with increased costs. Trial registration N/A (study on existing database)

Codes used for diagnosis of pneumococcal etiology: J13 Pneumonia caused by Streptococcus pneumoniae, and B953 Other diseases classified elsewhere caused by Streptococcus pneumoniae.

Data collected
The available collected data on the hospital stay were the length of stay in days and the month of discharge, whether the patient was admitted in ICU, and the length of the ICU stay. Comorbidities were also coded.
For each initial hospital stay, the following individual de-identified data were obtained: demographics (age, gender), hospital characteristics (teaching versus non-teaching and private versus public hospital), patient's comorbidities according to ICD-10, and the severity of the patients on ICU admission (organ failures, and a general severity score such as SAPS II). Regarding the ICU stay, the number of days with each type of organ support such as mechanical ventilation, renal replacement therapy and inotropes are recorded. The length of the initial hospital stay and in-hospital mortality were also recorded.
For each patient, the characteristics of the subsequent hospital stays within the year following the initial stay were also collected, as were the alcohol and tobacco consumption reported by the patient.

Costs evaluation: variables collected
Items of healthcare consumption taken into account are hospitalizations and medications. Standard medication and device costs are included in the GHM (French acronym for "Homogeneous Group Stays") unit cost for hospitalization. Expensive in-hospital drugs (from "Liste en sus", i.e., additional list) are reported in a specific database (FICHCOMP database) for each hospital stay. These sources are integrated into our assessment of hospitalization cost. Medications and devices prescribed or delivered during outpatient visits are not be considered.
Costing are restricted to direct costs and determined from the perspective of the French social security system. Costs are attributed from official French national tariffs from 2014 to 2015, and expressed in 2016 Euros. A standard national tariff is applied to each hospitalization based on the GHM code attributed in the PMSI database. GHM tariffs include medical and related procedures, nursing care, treatments (except specific expensive drugs and implants), drugs/devices used, food and accommodation, and investment costs for hospitalized patients. The additional cost per day of hospitalization in an intensive care unit are added to GHM tariffs, when appropriate. Expensive drugs are costed using the public retail price issued from the FICHCOMP database. For private hospitals, physician's fees are also added to the GHM tariffs as physicians are paid on a fee-for-service basis (source: ENCC, French acronym for French Reference for Costs of Common Methodology, or "Echelle Nationale des Coûts à méthodologie Commune", a tool that links the medical administrative data from the PMSI and the financial data from hospitals). For public hospitals, physicians being salaried, their wages are included inside the GHM tariff.