RESULTS
Characteristics of the patients. We studied 9658 adults with incident HF, most of them women (5157 patients, 53.4%), and most of them with HFrEF (5167 patients, 53.5%) (Table 1 suppl ).
Before matching, there were significant differences between the baseline characteristics of patients in the high income level and those in the low/middle income level (Table 1 suppl ). These differences among the baseline characteristics of HF patients by income were independent of the type of HF, as they were also present in patients with HFrEF (Table 2 suppl ) and with HFpEF (Table 3 suppl ). As no difference of mortality between patients with low and middle income was observed, we determined the optimal cut-off point for the net annual household income as 30,000 euros. Therefore, we have performed our analyses by comparing PS matched patients belonging to the highest NAHIL (>30,000 euros) to those belonging to the low/middle NAHIL (up to 30,000 euros).
After matching, all the measured baseline covariates were balanced between HF patients with a high NAHIL vs. those with a low/middle NAHIL (Table 4 suppl ), as well as for patients with HFrEF (Table 1 ), and with HFpEF (Table 2 ).
Relationship between the income level and mortality. The number of deaths by income after matching are presented in Table 5 suppl . After matching, during a median follow-up of 8.11 years (interquartile range, 3.37–10.62), 5616 patients died (78.9%) and 4195 patients died of a cardiovascular cause (59.0%).
The patients with HF and the highest NAHIL presented a survival longer than that of the patients with a low/middle NAHIL (RR of death for highest NAHIL [95% CI]: 0.86 [0.80-0.92], P <0.001;Figure 1A ). Similarly, the patients with HFrEF in the highest NAHIL presented a survival longer than that of the HFrEF patients with a low/middle NAHIL (RR of death for highest NAHIL: 0.88 [0.82-0.95], P <0.001; Figure 1B ), and the patients with HFpEF in the highest NAHIL presented a survival longer than that of the HFpEF patients with a low/middle NAHIL (RR of death for highest NAHIL: 0.82 [0.75-0.90], P <0.001; Figure 1C ). The HF patients in the highest income groups showed a cardiovascular mortality significantly lower than that of the patients in the low/middle income groups (RR of death for highest income in HF patients [95% CI]: 0.84 [0.76-0.98], P <0.001; Figure 1A suppl. RR of death for highest income among HFrEF patients [CI 95%]: 0.87 [0.81-0.93], P <0.001; Figure 1B suppl. RR of death for highest income among HFpEF patients [95% CI]: 0.82 [0.77-0.88], P <0.001; Figure 1C suppl ).
Relationship between the income and the hospitalizations. The hospitalizations and 30-day readmissions are presented in Table 6 suppl and Table 7 suppl , respectively. Figure 2 suppl presents the cumulative number of admissions for HF worsening, per 100 patient years (Figure 2A suppl ), and that of 30-day readmission for HF worsening (Figure 2B suppl ) in the matched groups. The estimated effects of SES for hospitalization and 30-day readmission for HF by each method considered are presented in Table 8 suppl . Rate ratios for recurrent hospitalizations and 30-day readmissions were lower than those obtained by the proportional-hazards models.
Multivariate relationship between the income and the prognosis.In our primary analysis using an intention-to-treat approach, highest NAHIL exposure was associated with a 16% lower relative risk (RR) of all-cause death, a 15% lower RR of cardiovascular death, a 17% lower RR of hospitalization for HF, a 35% lower RR of hospitalization for a CV cause, a 30% lower risk of hospitalization for HF, and a 34% lower RR of 30-day readmission for HF, compared with patients in the low/middle NAHIL, even after adjustment for sociodemographic characteristics, comorbidities, longitudinal use of medications, and propensity to be in the highest income group (Table 3 ). In the second type of analysis, time-dependent exposure to high NAHIL was associated with an even lower adjusted risk of all-cause and cardiovascular death, of hospitalization for HF, and of 30-day readmission for HF, compared with periods with a low/middle NAHIL (Table 9 suppl ). Similarly, high NAHIL was associated with a reduced mortality (all-cause and cardiovascular), hospitalizations and 30-day readmissions in patients with HFrEF, HFpEF, as well as in men and women with HF (Table 3 and Table 9 suppl ).
This favorable relationship of high NAHIL to mortality was maintained independently that the patients, before their inclusion or during the follow-up, presented cardiovascular events or other comorbidities in Table 1 and Table 2 (highest adjusted hazard ratio [HR] 0.84; 95% CI: 0.78-0.90, vs. highest HR 0.74; 95% CI: 0.68-0.80; P < 0.01 in all cases).
Subgroup analyses. These analyses have been performed for patients with HF (Figure 3A suppl ), HFrEF (Figure 3B suppl ) and with HFpEF (Figure 3C suppl ). Although, age, sex, atrial fibrillation, BMI and comorbidity negatively influenced the effects of income on the prognosis of HF (29), this benefit was also observed in the subgroups of patients over 70 years, women, diabetics, patients with chronic renal disease or with atrial fibrillation, overweight patients and those with an elevated burden of comorbid conditions (Figure 3A-C suppl ).