Year: 2016
Type: editorial
Abstract: This article refers to ‘Natriuretic peptide levels taken following unplanned admission to a cardiology department predict the duration of hospitalization’ by M. Asphaug et al., published in this issue on pages 1499–1505. Patients are not fractions. A 5% risk of death is not a 5% experience: what happens to an individual patient is different from the likelihood of outcomes across a population of people. Thus explaining risk to a single patient is a key clinical skill. Before risk can be explained, however, we first have to determine it. Risk-stratification tools are invaluable to the practicing physician. In the acute setting, risk stratification typically facilitates a single decision, such as admission or discharge from the emergency department, deciding whether or not to pursue further testing, or initiating more aggressive management. Ideally, risk stratification augments decision making, providing evidence-based criteria to help guide important clinical decisions. However, when used out of context, risk predictors can be misleading despite their value—akin to studying fish out of water. Trying to determine risk from an isolated marker outside of clinical context should only be performed with caution. In this issue of the Journal, Asphaug et al.1 demonstrate that elevated natriuretic peptide (NP) levels are associated with longer length of hospital stay (LOS) and all-cause mortality among patients admitted to a cardiology service. Their findings are consistent with a large body of evidence showing the tremendous prognostic power of NPs,2, 3 but the novelty of this study is their focus on LOS. Length of stay is a challenging endpoint in acute heart failure (HF) as no universal definition for when a patient is ‘ready for discharge’ exists.4 Issues such as convenience, rehabilitation or skilled nursing facility placement, or socioeconomic issues may keep a patient in the hospital longer than is medically necessary. While these issues may be less applicable to the population in this study than in the United States, they highlight the potential limitations of LOS as a clinical endpiont. Despite these limitations, LOS has importance to patients and clinicians, as those who stay in hospital longer are likely to be sicker. The authors hypothesized that NP levels would predict longer LOS. If correct, the value of this finding would depend on whether NP levels have value above and beyond other readily available prognostic information. Inevitably, clinicians will ask a corollary question: What can I do about it? If a patient is identified as high-risk, what can be done to mitigate this higher risk? Regardless, the first step is identification; concerns of next actions are irrelevant if we do not know who to target. As this is a retrospective analysis, the setting and rationale behind the study is critically important. In this single-centre administrative database study of 2978 patients with an unplanned admission to a cardiology ward, the aim was to identify a marker of longer LOS, irrespective of diagnosis: in essence, a test to separate the truly sick from the less sick. The authors hypothesized that NP values could be such a test. Thus, all patients with an unplanned cardiology admission who also had an NP value recorded were included. It is unclear whether the test was ordered for diagnostic purposes in patients with signs and symptoms of HF, although this could be assumed. However, the final primary discharge diagnosis was HF or cardiomyopathy in only 11%. Other diagnoses included arrhythmia (15%), hypertension (6%), airway infection (6%), angina pectoris or ischaemic heart disease (4%), and other (57%). Despite this broad context of multiple diagnoses, the authors found that NP levels retained their prognostic significance, supporting their contention that a single test may indeed provide a guidepost regarding hospital bed utilization. Enthusiasm for this potentially important finding must, however, be tempered with the limitations of the study, which are well-highlighted by the authors. Other variables relevant to risk assessment and management (except for age) were unknown in the database. Although the authors acknowledge the importance of other variables, such as diagnosis, aetiology, co-morbidities, gender, and laboratory values such as troponin or estimated glomerular filtration rate (eGFR), these were unaccounted for in their analysis. This makes it challenging to determine the independent value, if any, of NP levels. It is possible, perhaps even likely, that patients with elevated NP levels had worse vital signs, more co-morbidities, and lower eGFR and higher troponin levels. In the context of a single-centre study of patients with unplanned cardiology admission, varied diagnoses, and assumed appropriate testing and treatment, it is challenging to draw any firm conclusions about the value of isolated NP values. Even if potential confounders are disregarded, the value of NP levels was modest [area under the curve–receiver operating characteristic of 0.68] to predict patients with prolonged (>4 days) LOS.1 Despite the limitations of this study, the findings are supported by a robust body of evidence demonstrating the independent prognostic value of NPs.2 As important as the methodological limitations are the limitations of the applicability of the results. The authors emphasize NP levels should not be used clinically in isolation. Notably, the analysis was performed in relative isolation (there were no corrections for other predictors of outcome). In addition, only patients admitted to cardiology were included; thus these results should not be applied to other patient groups. As such, no conclusions can be drawn regarding NP values and need for admission, as no discharged patients were included. Ultimately, the authors highlight a key clinical question: given limited resources, who has the greatest need? In acute HF, we would argue a contrary perspective: the greater need is not identification of high risk, but lower risk.5 Nearly all acute HF patients are admitted. Multiple markers and risk instruments exist for hospitalized patients. Unnecessary hospitalization carries risks to patient safety, may lead to deconditioning, and even increase risk for post-discharge adverse events.6 Identifying patients safe for discharge directly from the emergency department or the hospital would conserve one of the most expensive resources in all of health care: staying in the hospital. Overall, the authors identify NP levels as a key candidate variable for future risk stratification tools or decision instruments in a broad spectrum of hospitalized cardiology patients. In general, an abnormal laboratory result, be it NPs, troponin, lactate, or creatinine, usually portends a worse prognosis than having normal results. What any single laboratory value adds to our ability to prognosticate, or what to do about it, is unclear unless placed in context. Conflict of interest: Neither author has any conflicts of interest to declare regarding the present paper. However, P.S.P. is receiving or has, in the past year, received consulting fees, honoraria, or research support from ACC, BMS, Medtronic, the Novartis, Trevena, scPharmaceuticals, Cardioxyl, Roche Diagnostics, and Relypsa. Research support has been received from Roche, Novartis, PCORI, and IUSM.
DOI: 10.1002/ejhf.658
Authors: Benton R. Hunter, Peter S. Pang
Language: English
Subfield: Cardiology and Cardiovascular Medicine
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W2545845163