The Symptoms-based Algorithm for Early Detection of Systolic Heart

Failure

Krishna Ari Nugraha

1,2

, M. Rizki Fadlan

1,2

, Dea Arie Kurniawan

1,2

, Liemena Harold Adrian

1,2

Faris Wahyu Nugroho

1,2

, Puspa Lestari

1,2

, Seprian Widasmara

1,2

, Anita Surya Santoso

1,2

and Mohammad Saifur Rohman

1,2

Department of Cardiology and Vascular Medicine, Faculty of Medicine, Brawijaya University – Dr. Saiful Anwar General

Hospital, Malang, East Java, Indonesia

Brawijaya Cardiovascular Research Center, Brawijaya University, Malang, East Java, Indonesia

{krishnaari22, naldaf, deaariekurniawan, liemenaharold, fariswahyunugroho, pusparyath, seprian.w,

Keywords: Heart Failure, Algorithm, Self-assessment.

Abstract: Heart failure is one of the global health problem priorities and is largely caused by late recognition of the

symptoms. Early detection is paramount to diagnosing heart failure; thus, a simplified algorithm is required.

The objective is to examine the accuracy of a symptoms-based algorithm for early detection of systolic heart

failure. We developed a symptom-based algorithm, compared to typical echocardiography examination. The

algorithm model in this study consisted of four symptoms with the highest association to systolic heart failure.

To evaluate outcomes in a larger population, we performed the derivation phase to assess the sensitivity and

specificity of this algorithm. The derivation phase was tested on 477 heart failure patients. All symptoms in

the algorithm—dyspnoea on exertion (DOE), paroxysmal nocturnal dyspnoea (PND), orthopnoea and leg

oedema—occur significantly more often among patients with systolic heart failure, compared to those with

diastolic heart failure (p < 0.05). The algorithm obtained an area under the curve and gave a sensitivity of

83.9% and a specificity of 81.1%. The symptom-based algorithm provides good outcomes for early detection

of systolic heart failure and are feasible to be developed into a self-assessment application for heart failure

patients with reduced ejection fraction.

1 BACKGROUND

Heart failure is one of the global health problem

priorities because of the high morbidity and mortality

among sufferers (Ponikowski et al., 2014; Tripoliti et

al., 2017; Ponikowski et al., 2016). Currently, at least

26 million people in the world are living with heart

failure. In Indonesia, based on data from Riset

Kesehatan Dasar (Riskesdas, 2018), the prevalence of

heart failure is 0.3%. This number will continue to

grow, along with the increasing prevalence of heart

failure risk factors, such as diabetes and hypertension.

Heart failure also contributes to increasing the burden

of national healthcare costs every year.

Although it has such a large impact on the health

burden in society, awareness of heart failure is poor.

Thus, numerous premature deaths occur, even though

most types of heart failure are preventable and a

healthy lifestyle can further reduce risk. The

incidence of premature deaths could be prevented if

people have an understanding of how to recognize

symptoms and seek immediate medical attention,

even after heart failure has developed (Ponikowski et

al., 2014; Conrad et al., 2018; Devroey and Van

Casteren, 2011).

There is little development in improving the

progression of heart failure severity, which is largely

due to non-effective approaches for early detection of

heart failure in testing interventions. Lifestyle and

pharmacologic interventions may be effectively

developed by analysing the early detection of heart

failure (Devroey and Van Casteren, 2011; Roberts et

al., 2015). Unfortunately, heart failure is a clinically

complex and heterogeneous disease that is

challenging to detect in routine care due to the

diversity of alternative explanations for symptoms. A

simple algorithm is required to help the general

population be able to early detect heart failure based

Nugraha, K., Fadlan, M., Kurniawan, D., Adrian, L., Nugroho, F., Lestari, P., Widasmara, S., Santoso, A. and Rohman, M.

The Symptoms-based Algorithm for Early Detection of Systolic Heart Failure.

DOI: 10.5220/0009388300380041

In Proceedings of the 4th Annual International Conference and Exhibition on Indonesian Medical Education and Research Institute (The 4th ICE on IMERI 2019), pages 38-41

ISBN: 978-989-758-433-6

on symptoms by using smartphone applications (Van

Riet et al., 2016).

2 METHOD

The data for this study were collected from the Saiful

Anwar Hospital Heart Failure Registry between 2016

and 2019. All patients hospitalized with the diagnosis

of heart failure were eligible for this study. We

included a patient with documented heart failure who

had an echocardiography examination performed

during hospitalization. All study participants were

given informed consent.

This study used a data survey of patients,

including history, physical examination, 12-lead

ECG and chest X-ray at the time of the initial patient

examination. Patients received an echocardiography

examination. The quantitative two-dimensional

(Biplane or Simpson) method was to assess the left

ventricular ejection fraction (LVEF) of the patients in

the study. Patients with LVEF ≤40% were

categorized as systolic dysfunction, while those who

had LVEF >40% were classified as diastolic

dysfunction, and preserved EF further was grouped as

diastolic heart failure group.

The investigators were

blinded to the echocardiography results, and they

reviewed all medical records pertaining to the patient

after completion of the clinical evaluation described

above.

To determine the symptoms with the highest

association to the diagnosis of systolic heart failure,

we performed a backward stepwise logistic

regression, with the diagnosis of heart failure based

on LVEF data as a dependent variable. The

symptoms with the highest association that were

found to be significant were dyspnoea on exertion

(DOE), paroxysmal nocturnal dyspnoea (PND),

orthopnoea and leg oedema. We built and validated

an algorithm model on our prior study from the basis

of these four symptoms with the highest association

to systolic heart failure. To evaluate outcomes in a

larger population, we performed the derivation

phase to assess the sensitivity and specificity of this

algorithm.

Figure 1: Symptom-based algorithm for systolic heart

failure.

3 RESULTS

This study was the derivation phase of our prior

study. We tested the algorithm on 477 heart failure

patients derived from the Saiful Anwar Hospital

Heart Failure Registry between 2016 and 2019. In this

study, 284 patients were male and there was no

significant difference in mean age between the two

groups. The average age of the systolic heart failure

group was 56.05 years (SD = 11.8), and 56.59 years

(SD = 13.2, P = 0.02) was the average age of those

with diastolic heart failure (table 1).

Traditional risk factors associated with systolic

HF analysed in this study include diabetes mellitus

type II, a history of coronary artery disease (CAD),

impaired renal function and the presence of atrial

fibrillation. Statistically, the frequency of the four risk

factors analysed in the two study groups was not

significantly different (p > 0.05). In this study,

patients with type II DM and atrial fibrillation were

more prevalent in the systolic heart failure

population, whereas CAD and kidney function

disorders were more prevalent in patients with

diastolic heart failure.

Algorithms that have been created based on tests

in the previous validation phase include four

symptoms: dyspnoea on exertion (DOE), paroxysmal

nocturnal dyspnoea (PND), orthopnoea and leg

oedema. This algorithm was then tested in the study

population in the derivation phase, and it appears that

patients with systolic heart failure had experienced all

yes

Orthopnoea

Diastolic Systolic

Paroxysmal

nocturnal

dyspnoea

Dyspnoea on exertion

Bilateral leg edema

The Symptoms-based Algorithm for Early Detection of Systolic Heart Failure

four symptoms more often than others with systolic

heart failure, compared to those with diastolic heart

failure (p < 0.05).

Table 1: Patient characteristics according to HF

classification (n = 477).

Demographic

and clinical

features

HF classification P

value

Systolic HF Diastolic HF

Men/ women 131/61 153/132 0.02

Age (years) 56.05(±11.8) 56.59(±13.2) 0.65

DM (%) 33.5 30.5 0.36

Known CAD

(%)

27.7 33.1 0.21

Renal

dysfunction

(%)

1 3.2 0.12

AFib (%) 18.3 16.4 0.64

Symptoms

(%)

DOE 93.8 47 0,00

PND 88 8.1 0,00

Orthopnea 68.2 46 0,00

Leg edema 43.8 26 0,00

Figure 2: ROC of the derivation phase.

As the final result from the receiver operating

curve analysis obtained area under the curve (AUC)

0.895 (95% CI, 0.867-0.922) the value obtained is

better than the results of the previous validation

phase. This algorithm provides a sensitivity of 83.9%

and a specificity of 81.1%.

4 DISCUSSION

Heart failure is one of the main cardiovascular

diseases in the world and has also become a major

concern in developing countries. The outreach of the

health system and the lack of awareness in the

population of cardiovascular risk factors results in

treatment delays, which ultimately increases

mortality. As shown in Devroey and Van Casteren's

study (2011), heart failure is one of the problems in

primary health care that requires careful history-

taking and physical examination.

Unfortunately, the symptoms of heart failure are

very diverse, so additional modalities are needed for

diagnosis. The algorithm to diagnose heart failure

according to ESC guidelines in a non-acute setting is

based on the previous clinical history of the patient,

the presenting symptoms, physical examination and

resting ECG. If all elements are normal, heart failure

is impossibly found in the patients. Plasma

Natriuretic Peptides should be measured when one

element is detected as being abnormal. This

measurement provides the opportunity for doctors to

detect those patients who need an echocardiography.

Clinical symptoms, such as dyspnoea on exertion

(DOE), paroxysmal nocturnal dyspnoea (PND),

orthopnoea and leg oedema, are arranged in an

algorithm, giving AUC 0.895 (95% CI, 0.867-0.922).

This is consistent with Devroey and Van Casteren's

study that DOE and leg oedema are heart failure

predictors with good sensitivity and specificity. The

combination of peripheral oedema, breathlessness on

exercise and pulmonary rales had good specificity to

detect heart failure, but low sensitivity.

This study attempted to create a simpler approach

with the symptom-based diagnostic tool so that

patients are able to do an accurately self-assessment

and then quickly seek medical advice. This symptom-

based algorithm was developed in an effort to

facilitate the common population in recognizing heart

failure symptoms that they may have experienced

before.

With high sensitivity and specificity, this

algorithm is expected to be able to reduce the

diagnostic delay that has happened thus far. We

expect that the heart failure morbidity and mortality

rate could be reduced in the future. This algorithm is

also expected to be used as a simple tool to help

medical practitioners in the early detection of systolic

heart failure. For further research, this algorithm

model can be tested in the general population,

especially for primary healthcare patients, as a first-

line screening tool before using more advanced

diagnostic modalities.

5 CONCLUSIONS

The symptom-based algorithm provides good

outcomes for early detection of systolic heart failure.

The 4th ICE on IMERI 2019 - The annual International Conference and Exhibition on Indonesian Medical Education and Research Institute

With the massive development of smartphone-based

technology, this algorithm is feasible to be developed

into a self-assessment application for heart failure

patients with reduced ejection fraction.

REFERENCES

Conrad, N., Judge, A., Tran, J. K., Mohseni, H., Hedgecott,

D., Perez-Crespillo, A., … Rahimi, K. (2018).

Temporal trends and patterns in heart failure incidence:

A population-based study of 4 million individuals.

The Lancet, 391(10120), P572–P580. http://doi.org/

10.1016/S0140-6736(17)32520-5

Devroey, D., and Van Casteren, V. (2011). Signs for early

diagnosis of heart failure in primary health care.

Vascular Health and Risk Management, 2011(7), 591–

596. http://doi.org/10.2147/VHRM.S24476.

Ponikowski, P., Anker, S. D., Al Habib, K. F., Cowie, M.

R., Force, T. L., Hu, S., … Filippatos, G. (2014). Heart

failure: Preventing disease and death worldwide. ESC

Heart Failure, 1(1), 4–25. http://doi.org/10.1002/

ehf2.12005.

Ponikowski, P., Voors, A. A., Anker, S. D., Bueno, H.,

Cleland, J. G. F., Coats, A. J. S., … van der Meer, P.

(2016). 2016 ESC Guidelines for the diagnosis and

treatment of acute and chronic heart failure: The Task

Force for the diagnosis and treatment of acute and

chronic heart failure of the European Society of

Cardiology (ESC). European Heart Journal, 37(27),

2129–2200. http://doi.org/10.1093/eurheartj/ehw128

Roberts, E. J., Ludman, A. R., Dworzynski, K. J.,

Al-Mohammad, A. undefined, Cowie, M. undefined,

McMurray, J. undefined, and Mant, J. undefined.

(2015). The diagnostic accuracy of the natriuretic

peptides in heart failure: Systematic review and

diagnostic meta-analysis in the acute care setting.

BMJ, 350 (mar04 22), h910. http://doi.org/10.1136/

bmj.h910.

Tripoliti, E. E., Papadopoulos, T. G., Karanasiou, G. S.,

Naka, K. K., and Fotiadis, D. I. (2017).

Heart Failure: Diagnosis, Severity Estimation and

Prediction of Adverse Events Through Machine

Learning Techniques. Computational and Structural

Biotechnology Journal, 15, 26–47. http://doi.org/

10.1016/j.csbj.2016.11.001

van Riet, E. E., Hies, A. W., Limburg, A., Landman, M. A.,

Kemperman, H., and Rutten, F. H. (2016). Extended

prediction rule to optimise early detection of heart

failure in older persons with non-acute shortness of

breath: A cross-sectional study. BMJ Open, 6(2),

e008225. http://doi.org/10.1136/bmjopen-2015-008225.

The Symptoms-based Algorithm for Early Detection of Systolic Heart Failure