Study cohort
Of the 232 patients initially enrolled 40 patients did not conclude the study. 30 patients did not conclude the proposed follow up protocol and 10 patients died during the follow-up. At the time of the analysis, 192 patients had been included in the study (Figure 1). Table 3 shows the baseline characteristics of the study cohort. Overall, 79.7% of all patients were male, and the mean age (± SD) was 64 ± 12 years. The duration of the HF was 3 years [2-5]. Common comorbidities included hypertension (69.3%), diabetes (37.5%), chronic renal failure (25.8%), chronic obstructive lung disease (20.3%), and atrial fibrillation (34.9%). Of those with heart disease, 50% had coronary artery disease, although a majority of patients (69.8%) had no HF hospitalizations in the year before enrollment. Most patients, 83.7% from Table 3 (NYHA I is 36.1% and NYHA II is 47.6%) were assessed as NYHA Class I or II, reflecting prevalently a mildly symptomatic HF cohort. We found levels of NT-proBNP of 984 [393-2334] pg/mL, and hs-TnT levels of 15 [8-27] ng/mL. We found HF hospitalization readmission rates of 30.2% and 21.9% visited the ED during the 12 months prior to the intervention. The calculate risk for the study groups are showed in the supplementary table 1
The subgroup analysis showed that the patients in the highest risk group were more likely to be older, had more comorbidities, and their heart disease was at a more advanced stage. We found levels of NT-proBNP of 599 [244-1211] pg/mL, 2045 [860-3664] pg/mL and 3494 [1503-8541] pg/mL (p<0.001) and hs-TnT levels of 11 [6-16] pg/mL, 24 [17-40] pg/mL and 53 [39-68] pg/mL (p<0.001) for the low, medium and high risk patients respectively. Improvement in NT-proBNP levels were found after 12 months of follow-up in the post-intervention period, 392 [192-949] pg/mL, 1923 [800-3685] pg/mL and 2283 [1263-4409] pg/mL for the low, medium and high risk patients respectively.
For the low risk group 91.4% of the patients were in the lowest functional class (NYHA class I or II), 76.5% for the medium risk group and 50% for the high-risk groups. No differences between groups were found in the LVEF or in the use of therapies included Angiotensin-receptor-neprilysin-inhibitor (ARNI), Angiotensin Converting Enzyme Inhibitor (ACEI)/Angiotensin Receptor Blocker (ARB) or beta-blockers (BB).
Clinical Outcomes
Primary outcomes: Table 4 compares the main outcome of rate of admission 30 days, 6 months and 12 months between pre- and post-intervention periods, analyzing patients by risk groups. Of the total, 7.8% had been admitted at least once in the 30 days prior to the baseline visit, which reduced to 1% in the 30 days following the intervention (p=0.002). The respective proportions being admitted in the pre-intervention and post-intervention periods, in the low-, medium- and high-risk groups were 5.1% and 0% (p=NA), 9.5% and 1.6%, (p=0.125), and 25% and 8.3% (p=0.625). Overall, a significant reduction was observed when comparing 6 months in the pre-intervention period (20.3%) and 6 months post-intervention (6.3%) (p<0.001). Of the 192 patients, 30.2% of the sample had been admitted at least once in the pre-intervention period; and in the post-intervention period, this number decreased to 10.4% in a year (p<0.001). The respective proportions in the low-, medium- and high-risk groups before and after the intervention were: 22.2% and 5.1% (p<0.001); 38.1% and 14.3% (p=0.125); and 66.7% and 41.7% (p=0.453).
When we consider the absolute data, in the cohort of 192 patients, we found a significant reduction (65%) in the number of patients admitted during the post-intervention period (20 patients; 10.4%) compared with the pre-intervention period (58 patients; 30.2%) (p<0.001). The difference was significant for the low, medium and high-risk groups. We also found a significant reduction (41%) in the number of hospital admissions during the post-intervention period 46 vs 78 for the pre-intervention period (p<0.001). The hospital admission length of stay during the pre-intervention period was 15 [6-29] days and 9 [2-21] days for the post-intervention period (supplementary material table 2).
Secondary outcomes: The rate of visits in the ED also decreased in the post-intervention period (Table 4a). In the 12 months before the study, the number of visits was 64, which decreased to 20 after the intervention, a reduction of 68.8% (p<0.001). A marked functional improvement was observed in the post-intervention period (Table 4b). In total, 31.1% of the patients improved at least one class in NYHA score, 61.6% remained the same, and 7.3% got worse. The number of asymptomatic patients also increased by 10%.
Costs analysis
Table 2 compares the total care associated cost and the specific components during the follow-up between the groups. The overall cost of applying the new follow-up intervention for the cohort of 192 patients was € 139,717 lower compared with standard care pre-intervention (Figure 2). We found a significant cost reduction in most of the categories considered. The most important cost reduction was related to costs associated with hospitalization, demonstrating a significant reduction of 78% (p<0.05).
There was a reduction of 69% in the number of ED visits (reduction in total cost for the post-intervention group of €17,249 compared with the pre-intervention group), and there was a significant reduction in the costs associated with primary care visits and with medication. There was a corresponding incremental cost related to the use of biomarkers (€1,728) and HFU visits (€28,567).
Utilizing the personalized biomarker approach produced a total of 113.6 QALYs (95% CI 108.5 to 118.2) compared with 109.1 QALYs (95% CI 104.2 to 113.4) for regular care, an increment of 4.5 QALYs (95% CI 2.9 to 6.1). The new approach was dominant (both less costly and more effective). The sensitivity analysis indicates that the new approach is the most cost-effective decision (Figure 3).
The budget impact analysis showed a potential saving between €-704,028 per 1000 patient-years (p-y) (95% CI 1,141,654 to -273,829) when the savings per patient was translated to the overall patient population in the best-case scenario (100% of the HF patients conducted using our new approach), to €‑352,014 p-y (95% CI -570,827 to -136,914) in the worst-case scenario (100% of the HF patients conducted using our new approach), with a medium-case scenario (50% of the HF patients conducted using our new approach) with a potential savings of €-528,021 (95% CI -856,240 to -205,371). Based on the 80,000 hospital admissions for HF that occur every year in Spain [18], the budget impact, considering only direct costs, could be between €-56,322,308 (95% CI -91,332,342 to ‑21,906,326) in the best-case scenario and €-28,161,154 (95% CI -45,666,171 to ‑10,953,163) in the worst-case scenario.