A pharmacoepidemiological approach to the study of prognosis in acute and chronic heart failure
Instituto de Saúde Pública da Universidade do Porto (ISPUP) Unidade de Investigação e Desenvolvimento Cardiovascular (UIDC/FM/UP) Faculdade de Medicina da Universidade do Porto (FM/UP)
Unanswered questions regarding the management of heart failure (HF), including those related with the use of diuretics and the unintended effects of drugs such as non-steroidal anti-inflammatory drugs (NSAIDS) and beta2-adrenergic agonists, must be studied by observational studies.
Recent studies suggest an association between larger diuretic doses and higher morbidity and mortality, beyond the fact that higher diuretic dose reflects to a large extent more severe HF. Non-adherence to prescribed diuretics may lead to decompensation of HF. NSAIDs are commonly used drugs and increase the risk of hospitalization for acute HF.
Among subjects with LVSD, inhaled beta-agonists were associated with an increased risk of heart failure hospitalization, and all-cause mortality.
This project aims to study the intended and unintended effects of pharmacological exposures on heart failure prognosis, accounting for their potential role as precipitants of decompensation in chronic HF as well as the effects of continued exposure on hospitalization and mortality at long term, comprising 2 specific objectives:
1) To quantify the effect of non-adherence to treatment with loop diuretics and use of NSAIDs and beta-2-agonists as triggering factors for decompensation of HF.
2) To quantify the independent association between loop diuretic maintenance dose and the occurrence of all-cause death or hospitalization due to HF.
These objectives will be accomplished using two different methodological approaches: case-crossover (retrospective case-timecontrol study; case-time-control analysis nested in a prospective cohort) and retrospective cohort designs.
In case-crossover designs each subject experiencing the outcome is his own control, overcoming most of the between-person confounding. Within-person confounding by transient factors is still possible. The case-time-control design is a variant of the casecrossover, comprising the selection of an independent control group that is subjected to a similar within-person analysis, providing a less biased OR estimate for the precipitating effect of the exposures under study.
Cases for case-crossover analysis (n=375) are subjects hospitalized at a teaching tertiary care hospital, due to decompensated chronic HF, defined by an increase of at least one New York Heart Association (NYHA) class in patients with chronic HF. Controls (n=200) are stable chronic HF patients (no hospitalization due to decompensated HF in the previous 3 months) followed at the HF outpatient clinic of the same hospital.
Casetime exposure is the week preceding the outcome, for cases, or the time of enrollment in the study, for controls. The control-time exposure window will be the fourth week before the index time (occurrence of the outcome for cases and time of enrollment in the study for controls). In both groups, exposure will be assessed retrospectively by self-report.
A case-time-control study with prospective assessment of exposure will be conducted, nested within a cohort of patients undergoing monthly evaluations of the time-varying exposures of interest. This is a prospective case-crossover design aiming to overcome the limitations of the retrospective study regarding the potential misclassification of exposure. Patients from the HF outpatient clinic (n=350) will be followed for 2 years, with monthly assessment of the exposures. Eighty cases of hospitalization due to decompensated HF are expected over a 2-year period and will be included in a case-crossover study able to test for different exposure windows. Among the remaining cohort participants, 80 controls (matched for sex and NYHA class) will be selected for the case-time-control analysis.
A randomly selected subcohort (n=88) will undergo 3-month evaluations of adherence to HF pharmacological treatments using electronic monitoring, allowing the validation of self reported adherence to treatment with diuretics.
The association between diuretic maintenance dose and the HF prognosis will be estimated using a retrospective cohort (prospective designs can shape the prescription patterns). A propensity score will allow a finer adjustment for confounding by indication.
Chronic stable systolic HF patients under optimized therapy will be selected from the outpatient clinic (n=400), among those enrolled after 1999. A combined endpoint (all-cause death or hospitalization due to decompensated HF) will be used.
This project has the potential to provide answers to unsolved issues in the management of chronic HF. The study designs are adequate to the complexity of the questions under study and the apparent redundancy of the methodological approaches to be used contributes to the achievement of internally valid conclusions.
The evaluation of a population of chronic HF patients across the broad spectrum of severity of the syndrome contributes to the external validity of the attained results. The research team has adequate experience in both clinical and epidemiological areas covered by the project.