2.1 Study Objectives

The aim of this study was to describe the incidence and characteristics of HF patients hospitalized in France and to estimate the hospital healthcare consumption and the associated costs of HF patients.

2.2 Study Design

This was a retrospective cohort study that included patients with HF aged 18 years or older hospitalized in France (metropolitan and French overseas territories) between January 1, 2019 and December 31, 2019. The index date was the first hospital admission date with an HF diagnosis with or without decompensation during the study period. The look-back period was defined as the 5 years before the index date, during which comorbidities and risk factors were collected. The follow-up period was set to 1 year following the index date.

2.3 Data Source

Hospitalization data were retrieved from the French national hospital discharge database, referred to as the Programme national de Médicalisation des Systèmes d'Information (PMSI). The PMSI collects pseudo-anonymized patient data on hospitalizations (overnight, inpatient stays, and day stays—excluding outpatient consultations) and ambulatory care requiring hospitalization or treatment session (e.g., dialysis). More precisely, the PMSI database includes hospital discharge summaries for each inpatient stay including administrative information (age, gender, and unique identifier of each patient), medical procedures coded in the common classification of medical acts (CCAM) and the following diagnoses coded according to the 10th revision of the World Health Organization international classification of diseases (ICD-10):

Primary diagnosis (PD) is a health condition that justifies the patient’s admission to the medical unit determined on discharge from the medical unit

Associated diagnosis (AD) is a condition, symptom or any other reason for hospital care coexisting with primary diagnosis, or constituting:

o

a distinct additional health problem (another condition)

o

complication of the primary diagnosis

o

complication of the treatment of the main morbidity

Related diagnosis (RD) has a role, in association with the primary diagnosis, and when the latter is not sufficient, to report on the care of the patient in medico-economic terms. Its determination is based on three principles:

o

It is only necessary to mention when the primary diagnosis is coded within the chapter XXI of ICD-10

o

The related diagnosis is a chronic or long-term disease or permanent state, present at the time of the stay

o

The related diagnosis answers the question: for which disease or condition was the care recorded as primary diagnosis carried out

Data from the following four hospital wards are available in the PMSI: Medicine, Surgery, and Obstetrics (MSO); Home Hospitalizations (Hospitalisation à Domicile, HAD); Post-Acute Care and Rehabilitation (Soins de Suite et de Réadaptation, SSR); and Psychiatry (Recueil d'Informations Médicalisé pour la Psychiatrie, RIM-P). In this study, the data were retrieved for the MSO ward alone.

Drugs and medical devices given during a hospitalization are directly included in the Diagnosis-Related Group (DRG) tariffs. It is therefore not possible to know precisely which drugs/medical devices were prescribed and their costs. However, innovative and high-cost drugs and medical devices, also called ‘liste-en-sus’ drugs and medical devices, are collected using 7-digit codes in this database.

2.4 Patient Identification

HF patients without cardiac decompensation were identified with ICD-10 codes I50, I420, I110, I130, or I132, and without ICD-10 codes J9600 (code available since March 2019), R570, K720, or R601, as the primary or associated diagnosis. As an ICD-10 code for cardiac decompensation in HF patients does not exist, it was identified using an algorithm for acute HF defined by the French Health Insurance Fund [13]. According to the algorithm, patients with cardiac decompensation were defined as those who were hospitalized during the study period with ICD-10 codes J9600, R570, K720, R601, I50, I110, I130, I132, or J81 as the PD or AD. EF status was identified with the corresponding ICD-10 codes (ICD-10 codes for p-EF [i.e., EF ≥ 50] were I5000, I5010; ICD-10 codes for r-EF [i.e., EF < 40] were I5002, I5012; ICD-10 codes for mildly reduced EF [i.e., 40 ≤ EF < 50] were I5001 and I5011; ICD-10 codes for unspecified EF were I5009 and I5019). In this study, the mildly reduced EF group was merged with the preserved EF group in statistical analyses. Type 2 diabetes mellitus (T2D) and chronic kidney disease (CKD), known for their important interrelation with heart failure [14], were identified using ICD-10 codes (T2D: E11, CKD: N18), and the algorithm defined by the French Health Insurance Fund for end-stage renal disease (presented in the Electronic Supplementary Material [ESM]).

2.5 Outcomes and Variables2.5.1 Patients’ Socio-Demographic and Clinical Characteristics

Age, gender, EF status, and comorbidities including T2D and CKD were described for all patients included in this study. T2D and kidney diseases were considered in this study for their important interrelation with heart failure [14].

2.5.2 Incidence and Prevalence of Heart Failure

Incident patients of HF were those diagnosed with HF during the study inclusion period (January 1, 2019–December 31, 2019) with no prior diagnosis of HF during the previous 5 years (January 1, 2014–December 31, 2018). Therefore, the incidence was calculated with the following equation and expressed as number of cases per 100,000 inhabitants:

$$} = \frac}\;2019}}} \ge 18}\;2019}} \times 00,000.$$

Prevalent patients of HF were those diagnosed with HF during the study inclusion period (January 1, 2019 to December 31, 2019) and with prior diagnosis of HF during the look-back period (January 1, 2014–December 31, 2018). Therefore, the prevalence was calculated with the following equation and expressed as number of cases per 100,000 inhabitants:

$$} = \frac}2019}01}2014}31}2018}}} \ge 18}2019}} \times 00,000$$

The general population aged ≥ 18 years in France in 2019 was estimated at 52,408,636 inhabitants by the French National Institute of Statistics and Economic Studies (INSEE).

2.5.3 Mortality and Cardiovascular Death Rates

In-hospital death is identified using a specific code (‘9’) when describing reason for hospital discharge. Cardiovascular death was defined as in-hospital death with a cardiovascular event during the last hospitalization.

All-cause mortality and cardiovascular death rates were calculated using the following equations:

$$\mathrm-\mathrm=\frac}} \times 100$$

$$} = \frac}}}} \times 100.$$

2.5.4 Healthcare Resource Utilization (HCRU)

All outcomes were considered in the year after the index date (up to December 31, 2020, at the latest). The outcomes of the study included the frequency and length of HF hospitalizations per patient in the MSO ward; use of high-cost innovative drugs and medical devices (‘liste-en-sus’ drugs and medical devices identified using 7-digit codes) per patient; type and frequency of the most commonly used medical procedures during HF hospitalizations; and the cost estimations of events listed above from the hospital perspective.

HF hospitalization outcomes were further stratified based on the presence or absence of cardiac decompensation, EF status (r-EF and p-EF), and incident/prevalent patients. P-EF and r-EF subgroups were further stratified based on the presence or absence of comorbidities (patients with T2D and CKD [T2D+/CKD+], with T2D and without CKD [T2D+/CKD−], without T2D and with CKD [T2D−/CKD+], and without T2D and without CKD [T2D−/CKD−]).

2.5.5 Hospital Costs in 2019

The cost analysis was conducted using the hospital perspective for year 2019. Under a hospital perspective, the healthcare resources including stays, medical procedures, and innovative or expensive health products were considered.

To approximate the production cost of a hospitalization, the preferred source of data is the National costs study (ENCC) based on DRG [15]. Economic data are based on accounting agreements from the sample of French hospitals that participated in ENCC. The ENCC database provides the direct cost related to the hospitalization (structure costs are not included). The calculation of the cost was conducted as follows:

$$}\left( }} \right) \, = \, \sum_ \left( }\_}_} - }}} \left( } \right) \, *}\_}_ } \right),$$

where i is the DRG associated with the stay; 1 DRG = 1 cost depending on the type of hospital (public or private) and the year in which the health resources were used.

2.6 Data Collection

Data on patient-related variables such as sociodemographic characteristics (age, sex, and residence), type of hospital stay, dates of hospital admission and discharge, diagnoses at hospital admission and discharge, medical procedures, ‘liste-en-sus’ drugs and medical devices, length of hospitalization, comorbidities, and all-cause and cardiovascular death were retrieved from the PMSI database.

2.7 Statistical Analysis

Data extraction, management, and statistical analysis were carried out by IQVIA using SAS® software, version 9.4 (SAS Institute, North Carolina, USA).

All the analyses were descriptive. Continuous variables were described by their number (of valid cases, of missing values if any), mean, standard deviation (SD), and median, interquartile range (IQR), minimum, and maximum.

Categorical variables were reported using the total number (of valid cases, of missing values if any) and relative percentage per category. Missing values on type of EF were also reported in percentage.

The prevalence and incidence of HF were estimated per 100,000 inhabitants. All cost analyses were carried out per patient and per year.