An Integrative Literature Review of Heart Rate Variability Measures to Determine Autonomic Nervous System Responsiveness using Pharmacological Manipulation

Abstract

Background:

Heart rate variability (HRV) is defined as the difference in timing of intervals between successive heart beats and is used as a surrogate measure to the responsiveness of the autonomic nervous system. A review and synthesis of HRV as an indicator of autonomic nervous system responsiveness to pharmacologic stimulation/blockade of sympathetic and/or parasympathetic nervous system branches has not been completed.

Purpose:

The aim of this integrative review is to synthesize research examining pharmacological modulation of the autonomic nervous system and the response of time domain, frequency domain and non-linear measures of HRV.

Conclusions:

Sympathetic nervous system blockade resulted in a consistent decrease in the standard deviation of normal-normal interval metric across studies. Stimulation of the parasympathetic nervous system was associated with an increase in several time, frequency, and non-linear HRV indices, while blockade of the parasympathetic nervous system led to a decrease in similar indices.

Clinical Implications:

Recommendations to improve the reproducibility of future HRV research are provided for standardization of recording, analysis and metric decisions, and more thorough reporting of HRV indices in published studies. Alterations in autonomic nervous system input to the cardiovascular system are associated with an increased risk for adverse patient outcomes and increased mortality; therefore, understanding the influence of pharmacologic autonomic nervous system modulation on HRV indices and important considerations for reproducible HRV research design will inform future translational research on cardiovascular risk reduction.

Assessment of heart rate as a method to evaluate cardiovascular health status is common practice in healthcare and clinical research.¹ Heart rate variability (HRV) is defined as the difference in timing of intervals between successive heart beats (difference between successive RR intervals) and is used as a surrogate measure to gauge the reactivity of the autonomic nervous system to maintain cardiovascular homeostasis under varying physiological conditions.² HRV was initially used as an early identifier of fetal distress,³ and then gained clinical relevance when changes in HRV were studied as a predictor/correlate of adverse cardiovascular events.^4,5 HRV continues to be used as a standard risk stratification measure in persons with cardiovascular disease.^6,7 As the standards of measurement and interpretation for HRV measurement have not been updated in its entirety since their release in 1996,⁵ the field lacks a recent official consensus for the degree of HRV value change (time domain, frequency domain, non-linear methods) that indicates a clinically meaningful change in autonomic nervous system activity.

The autonomic nervous system is a neurologic control system that unconsciously regulates multiple body functions such as respiration, digestion, and cardiac control.⁸ The parasympathetic nervous system (PNS) is the branch of the autonomic nervous system associated with rest and is largely regulated by the Vagus nerve and central/peripheral muscarinic receptors that are activated by the neurotransmitter acetylcholine.⁸ The sympathetic nervous system (SNS) is the branch of the autonomic nervous system associated with “fight or flight” response, and SNS cardiac accelerators originate from the thoracic spine at the level of T1-T4/5 and are activated when the adrenal medulla is stimulated to release catecholamines (i.e., epinephrine and norepinephrine) via preganglionic sympathetic fibers.⁹ Catecholamine-mediated SNS stimulation of cardiac β1-adrenergic receptors (β1-AR) and α1-adrenergic receptors (α1-AR) increase heart rate and contractility (chronotropy and inotropy, respectively), and work in conjunction with the PNS to maintain cardiac homeostasis.¹⁰

To study HRV responses to PNS/SNS modulation, researchers can use pharmacologic interventions, such as the administration of SNS and PNS modulating medications. Pharmacologic agents used to stimulate or suppress the autonomic nervous system act on α1-ARs, β1-ARs and muscarinic receptors (see Supplemental Table S1). As many of the pharmacologic agents used to modulate the autonomic nervous system are specific to the SNS or PNS, evaluating HRV response to these medications is a strategy to quantify responses of different HRV indices to simulation and blockade of the SNS or PNS in a controlled manner. This is important to assess because there is varied predictive accuracy of HRV for parsing sympathetic versus parasympathetic actions in the literature, with more discrepancy in the ability of different HRV metrics to isolate SNS responsiveness and reactivity.^2,5,11

Clinicians and scientists quantify HRV by a variety of indices including time domain, frequency domain and non-linear HRV measures (Table 1). Inconsistency in HRV measurement periods, sampling rates, analysis protocols, and standards in published research can contribute to HRV results that are difficult to compare and reproduce. Although there have been standards published on HRV analysis in humans,^2,5 there is a lack of implementation of standardized HRV protocols in research making it difficult to ascertain the level of HRV responsiveness to pharmacologic manipulation. Therefore, the aim of this integrative review is to synthesize research examining pharmacological modulation of the autonomic nervous system and the response of time domain, frequency domain, and non-linear measures of HRV. Both pre-clinical studies on rodents and human studies will be examined to increase the sample size of the review and to compare and contrast similarities and differences between the tightly controlled pre-clinical studies and translational human studies that may encompass higher variability across participants. The overall goal is to summarize the consistency of currently used HRV indices in characterizing cardiac SNS and PNS influence to identify valid HRV measures and provide recommendation to aid in future rigorous and reproducible HRV research.

Table 1.

Common Heart Rate Variability Metrics

Measure	Definition
Time-domain Metrics: calculate the interval (in milliseconds) between each heart beat originating from the sinoatrial node (often referred to as normal-normal intervals, or N-N intervals). The mean, standard deviation and average variability of these intervals are calculated for time-based HRV analyses.
RR	Mean RR interval
SDNN	Standard deviation of RR intervals (SDNN) in the recording segment. Unspecific indicator of the total amount of variability between successive RR intervals, where a reduction of SDNN indicates a reduction in HRV and PNS input to the heart.
rMSSD	Square root of the absolute value of mean squared differences of successive RR intervals (so that negative and positive changes are considered equally; rMSSD). Measure of variability between successive normal to normal interbeat intervals and mainly reflects PNS activity to heart.
Frequency-domain Metrics: evaluate the power spectra in the ECG recording by fitting a cosine wave to the frequency signal and using Fast Fourier analysis to digitally describe the signal.
HF Humans: 0.15-0.4 Hz Rodents: 1.5-5 Hz	The high frequency (HF) domain represents PNS activity, highly influenced by respiratory oscillations. Almost completely reflects PNS input to the heart/sinoatrial node.
HFnu	The normalized high frequency (HFnu) spectral metric is calculated after the initial calculation of the absolute HF and LF spectral bands. The normalized HF metric is calculated by the following equation: HFnu = HF / (LF + HF)
LF Humans: 0.04-0.15 Hz Rodents: 0.15-1.5 Hz	Low frequency (LF) spectral power reflects vagal and sympathetic activity via the baroreflex loop. Influenced by both PNS and SNS input to the heart/SA node.
LFnu	The normalized low frequency (LFnu) spectral metric is calculated after the initial calculation of the absolute HF and LF spectral bands. The normalized LF metric is calculated by the following equation: LFnu = LF / (LF + HF)
LF/HF Ratio	The ratio of the LF and HF bands. The primary contributors of the LF/HF ratio are PNS, SNS, and baroreflex activity.
VLF Humans: 0.0033-0.04 Hz	Uncertainty about the exact mechanisms that generate the very low frequency (VLF) band. Thought to be influenced by the heart’s intrinsic nervous system, SNS, and PNS.
Ultra-Low Frequency Humans: <0.003 Hz	Requires a 24-hour recording period because slow acting physiological processes are thought to generate ultra-low frequency bands.
Non-Linear Metricsᶲ: describe the temporal dynamics of variability analyses and evaluate HRV as a three-dimensional signal. Nonlinear HRV analysis methods have been proposed in an attempt to capture the complex and chaotic nature of autonomic nervous system input to the heart for heart rate control.
Approximate Entropy	Approximate entropy measures the complexity and regularity of a time series.
Poincare plot	Each successive RR interval plotted against the prior (i.e., RR(n) vs. RR(n+1)). Dispersion on points in a two-dimensional scatter plot is mentioned geometrically and denoted by two measures: SD1 and SD2.
Poincare plot: SD1	SD1 is the perpendicular dispersion to the line of identity- this short term HRV is suggested to be associated with cardiac vagal control.
Poincare plot: SD2	SD2 is the length of the plot along the line of identity; this long term HRV is suggested to be associated with cardiac sympathetic control.
Sample Entropy	Sample entropy measures the signal complexity and regularity.
Symbolic Analysis	Symbolic analysis converts the RR interval and systolic arterial pressure in a series of symbols and evaluates the dynamics of three consecutive symbols, classifying them by variation.
Symbolic Analysis 0V%	Symbolic analysis with zero variation.
Symbolic Analysis 1V%	Symbolic analysis with one variation.
Symbolic Analysis 2LV%	Symbolic analysis with two like variations.
Symbolic Analysis 2UV%	Symbolic analysis with two unlike variations.

Abbreviations. HF: high frequency; HFnu: normalized high frequency; HRV: heart rate variability; Hz: hertz; ms: milliseconds; LF: low frequency; LFnu: normalized low frequency; SDNN: standard deviation of RR intervals.

^ᶲNonlinear metrics reported are metrics reported in the updated 2015 HRV position paper¹⁵ in addition to symbolic analyses. Additional HRV metric responses are listed in Supplemental Table 3. Frequency band rages for humans are based on recommendations by Shaffer & Ginsberg² and rodents by Thireau et al.⁴⁵

Methods

Whittemore and Knafl’s¹² integrative methodology was utilized to evaluate and synthetize the literature, and includes problem identification, literature search, data evaluation, data analysis and presentation of findings. An integrative literature review allows for the inclusion of different sources, such as experimental and non-experimental research, which brings in various research perspectives and techniques related to the phenomenon of concern to create an abundant and meaningful dataset.¹²

We searched PubMed, CINAHL, Scopus and Web of Science for original research articles published between 2002 and 2022 to capture primary sources reporting HRV research over the past 20 years. Subject heading, title, and keyword searches were completed using variations of the words: “heart rate variability, alpha-1 adrenergic receptor, beta-1 adrenergic receptor, muscarinic receptor”, along with additional search terms to target drugs that bind to alpha-1 adrenergic, beta-1 adrenergic, and muscarinic receptors, including “AB adrenergic stimulation”, “AB beta-adrenergic antagonists”, and “Muscarinic Antagonist”, among others (see Supplemental Table S2 for search strategies). A total of 3,371 articles resulted from the combined searches.

The review was limited to human subject and rodent studies published in English with original articles investigating HRV response to pharmacologic activation or inactivation of alpha-adrenergic receptors, beta-adrenergic receptors and/or muscarinic receptors. The intent of pharmacologic agonism or antagonism of the receptors is to simulate or block the sympathetic and parasympathetic branches of the autonomic nervous system to evaluate the response of HRV parameters. Human study subjects with autonomic or cardiac dysfunction, such as spinal cord injury or cerebrovascular incident, were excluded so pharmacologic influence on the autonomic nervous system was not confounded by other autonomic processes. Animal studies that used anesthetized animals for recording were excluded. Literature was screened based on PRISMA’s article selection guidelines in three stages: titles and citations, abstracts, and full-text of article.¹³ Title and abstract exclusion criteria included: date out of range, not HRV focused, no pharmacologic intervention, and review papers. Full text exclusion criteria are outlined in Figure 1. Full-text articles that did not meet the above exclusion criteria were included in the final analysis.

Figure 1. Records Screened and Included.

PRISMA flowchart for all manuscripts that were screened and included in the final literature review.

The quality of the selected studies was assessed using the Crowe Critical Appraisal Tool, version 1.4.¹⁴ This tool is used to evaluate individual study quality using several categories: preliminaries (title and reproducibility), introduction, design, sampling, data collection, ethical matters, results, and discussion. The total score is expressed as a value and percentage of the denominator 40. The time and frequency domain indices that we chose to synthesize in this scoping review are consistent with the measures reported in the HRV Task force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology,⁵ and were included in the data synthesis for each study unless they were not reported in the primary literature source. The non-linear HRV indices reported are those that were covered in the 2015 update,¹⁵ and were included in the data synthesis for each study unless again not reported in the primary literature source. Additionally, symbolic non-linear HRV indices were reported as they were present in several manuscripts. See Table 1 for an overview of all non-linear metrics analyzed in the literature synthesis.

Results

The 19 studies listed in Table 2 met the review criteria, and quality appraisal of these studies is outlined in Supplemental Table S3. Of these manuscripts, 11 were human HRV studies^16–26 and eight were pre-clinical rodent HRV studies^27–34 (Table 2). In 18 of the studies, data on adult humans or animals were reported, while data on children (median age 10 years old) were reported in one study. The average quality scores for all studies, human only studies, and animal only studies were 86.11 ± 7.23%, 85.18 ± 8.16, and 87.38 ± 6.00, respectively (Supplemental Table S3). For human studies, the lowest category score was sampling (3.55 ± 0.52), and the highest category score was ethical matters (4.73 ± 0.47). For animal studies, the lowest category score was discussion (3.75 ± 0.89), and the highest category score was data collection (4.63 ± 0.52). The recording time analyzed for HRV analysis is listed in Table 2, and a summary of all HRV metric responses to pharmacological stimulation or blockade that were reported in each manuscript are listed in Supplemental Table S4.

Table 2.

Overview of Included Manuscripts

Author, Year	Aim	Drug	Branch of ANS Targeted	Sample	Recording Length Analyzed	Time Domain	Frequency Domain	Nonlinear
Akbari et al., 2014	Aim was to assess various doses of oral oxybutynin on cardiac ANS modulation by measuring short-term HRV indexes during supine rest	Oxybutynin 2.5 mg PO, 5 mg PO, 10 mg PO or placebo	PNS Blockade	8 healthy male subjects 20-23 years old	5 minutes	Mean NN, SDNN, PNN50%, RMSSD	HF n.u., LF n.u., LF/HF ratio	NA
Beckers et al., 2006	Aims were to examine the influence of several ANS blocking agents on non-linear HRV indices and explore the physiological mechanisms contributing to these non-linear fluctuations	SNS Blockade #1: Phentolamine (10 mg/kg IV); SNS Blockade #2: Propranolol (4 mg/kg IV); PNS Blockade: Atropine (1 mg/kg IV)	SNS Blockade; PNS Blockade	13 male Wistar rats, 280-300g (4-5/group)	At least 20 minutes	Mean RR, SDANN1, RMSSD, pNN50	LF power, HF power, total power, HF n.u., LF n.u., LF/HF	DFA alpha 1, DFA alpha 2, ApEn, fractal dimension
Bolea et al., 2014	Evaluate sensitivity of correlation dimension, sample and approximate entropies to change in ANS regulation of the heart	PNS Blockade: Atropine (0.03mg/kg IV) supine and stand position; SNS Blockade: Propranolol (0.2 mg/kg IV) supine and stand position; Combined ANS blockade: atropine (0.03mg/kg IV) + propranolol (0.2 mg/kg IV) supine and stand position	SNS Blockade, PNS Blockade, Combined ANS Blockade (SNS/PNS)	13 male subjects (19-39 years old, mean age 21)	5 minutes	NA	LF power, HF power, LF n.u., HF n.u.	ApEn, SampEn, correlation dimension
Castiglioni et al., 2011	Purpose was to separately assess cardiac vagal and sympathetic (cardiac vs. vascular) influences on fractal properties of HR and BP (atropine, propranolol and clonidine) as described by scale exponents of detrended fluctuation analysis	PNS Blockade: Atropine (40 mcg/kg IV); SNS Blockade: Propranolol (200 mcg/kg IV); PNS Stimulation: Clonidine (6 mcg/kg PO); Combined ANS Blockade: Atropine (40 microgram/kg IV) + Propranolol (200 mcg/kg IV) and Atropine (10 mcg/kg IV) reinforcing dose	SNS Blockade, PNS Blockade, PNS Stimulation, Combined ANS Blockade (SNS/PNS)	9 healthy male physicians (age 25-46 years)	15-20 minutes	Mean RR, SDNN	VLF power, LF power, HF power, LF/HF	DFA alpha 1, DFA alpha 2
Cepeda et al., 2018	Test the hypothesis that changes in non-linear HRV indices correlate with the physiological effect of progressive low dose atropine on RR intervals (RRi) and on traditional time domain variabilities in healthy individuals	PNS Stimulation: 6 progressive doses of atropine (Cumulative: 1.4, 2.2, 3.2, 4.4, 5.8, 7.2 ug/kg IV) until endpoint reached. Endpoint: Infusion stopped if HR > by 10 BPM	PNS Stimulation	14 healthy human subjects: male= 8, female=6. Age average 26	3 minutes	Mean RR, SDNN, RMSSD	RSA	SD1, SD2, Symbolic analyses 0V%, 1V%, 2LV%, 2UV%, deceleration capacity, acceleration capacity, PIP, IALS, PSS, PAS
Cysarz et al., 2015	Evaluate the ability of symbolic analysis to monitor cardiac ANS regulation during ANS challenges (SNS and PNS modulation) and which symbolic transformation is the most helpful	PNS Stimulation: Low dose atropine (2 mcg/kg IV); PNS Blockade: High dose atropine (15 mcg/kg IV). Reflex PNS Stimulation: Phenylephrine infusion (1.2 mcg/kg/minute IV); Reflex SNS Stimulation: Nitroprusside infusion (1 mcg/kg/minute IV)	Reflex SNS Stimulation; Reflex PNS Stimulation; PNS Stimulation; PNS Blockade	Autonomic challenge 1= 10 healthy male subjects (range 23-42 years). Autonomic Challenge 2= 7 males, age 31±7 years. Challenge 1 and challenge 2 used different subjects.	5 minutes	RR, SDNN	ln(LF), ln(HF), ln(LF/HF)	Symbolic analyses 0V%, 1V%, 2LV%, 2UV%
Elstad et al., 2011	To evaluate whether LF oscillations in HRV (0.1 Hz) and cardiac output buffer fluctuations at approximately 0.1 Hz in arterial BP (Meyer waves). Tested how LF power in MAP changes when HRV is removed by pharmacological cardiac ANS blockade	PNS blockade: atropine (0.04mg/kg IV) in head up tilt 30 degrees. Combined ANS blockade: Propranolol (1 mg + 0.2 mg/kg IV) and atropine (0.04mg/kg IV) in supine position.	PNS Blockade, Combined ANS Blockade (SNS/PNS)	10 healthy volunteers (5 men, 5 women) age 25 +/− 4	5 minutes	NA	LF power	NA
Fazeli et al., 2019	To assess the cardiac PNS activity before and after 12 weeks of oxybutynin treatment in children with overactive bladder	PNS Inhibition: Oxybutynin 0.4-0.6 mg/kg PO for 12 weeks.	PNS Inhibition	10 children (7 females, 3 males). Median age 10 years (range 5-18).	5 minutes	NA	HF power, total power	NA
Karmakar et al., 2011	To evaluate the measure CCM by analyzing changes in autonomic function on Poincare plot descriptors (SD1 and SD2) during head up tilt, atropine infusion and transdermal scopolamine patch	PNS Stimulation: Scopolamine (1.5 mg transdermal patch for 24 hours); PNS Blockade: Atropine sulfate (0.12 mg/minute IV for 5 minutes then 0.24 mg/minute IV)	PNS Stimulation; PNS Blockade	5 healthy subjects, sex not listed. Mean age 30.2 ± 7.2	20 minutes	NA	NA	SD1, SD2, complex correlation measure,
Kur’yanova et al., 2014	To reveal specific features of HRV changes under conditions of ß1-AR blockade and ɑ1-AR stimulation in outbred rats	SNS Blockade: Atenolol (2.5 mg/kg IP). Reflex PNS Stimulation: Phenylephrine (0.3 mg/kg IP); Control groups received 0.1 ml/100g saline according to the same schedule.	SNS Blockade; Reflex PNS Stimulation	5-week-old rat pups and 3.5-4-month adult rats. Only adult data reported. Control group: 15 males, 12 females; Atenolol group: 15 males, 13 females; Phenyl group: 15 males, 13 females	Continuous series of 300 RR interval	RRi variability range	VLF power, LF power, HF power,	NA
Miyabara et al., 2017	To identify sensitive HR and BP variability parameters from a given set of well-known methods for the quantification of CV ANS function after several ANS blockades	SNS blockade #1: Metoprolol (1 mg/kg IV); SNS blockade #2: Prazosin (1 mg/kg IV); PNS Blockade #1: Methylatropine (0.5 mg/kg IV); PNS Blockade #2: Hexamethonium (10 mg/kg IV); Combined ANS Blockade: Metoprolol + Methylatropine;	SNS Blockade; PNS Blockade; Combined ANS Blockade (SNS/PNS)	Male Sprague Dawley rats (12-14 weeks old). 6 rats in each drug group were studied	30 minutes for baseline and 60 minutes after drug administration	MeanNN, cvNN, RMSSD	VLF power	FWSHANNON, POLVAR3
Pereira-Junior et al., 2010	Test a noninvasive and inexpensive method for ECG recording in conscious rats, assessing its feasibility for HRV analysis	SNS blockade: L-Propranolol (4 mg/kg IP); PNS Blockade: Atropine (2 mg/kg IP)	SNS Blockade; PNS Blockade	6 male WKY rats (330-380g)	3 minutes	Mean RR	HF power, LF power	NA
Picard et al., 2009	To describe the relationship between vagally mediated bradycardia and HRV indices in young and older healthy individuals	PNS Stimulation: Low-dose atropine- 8 bolus doses given (cumulative doses from 0.4 to 7.2 micrograms/kg IV).	PNS Stimulation	N=61: 34 young healthy volunteers (aged 21-29, 47% female), 27 older volunteers (mean age 63 ± 1.1, 48% female)	3 minutes	Mean RR, (SDNN and RMSSD measured but not reported)	Total power and respiratory power measured but not reported	NA
Sayin et al., 2016	To evaluate the best way to calculate cardiac autonomic tone in rats and evaluate HRV response to ANS blockers in rats	SNS Blockade: Atenolol (2 mg/kg SQ); PNS Blockade: Methylatropine (2mg/kg SQ); Combined ANS Blockade: Atenolol (2 mg/kg SQ) + Methylatropine (2mg/kg SQ)	SNS Blockade, PNS Blockade, Combined ANS Blockade (SNS/PNS)	8 WKY male rats (46 weeks old)	5 minutes	NA	LF, HF	NA
Schiffers et al, 2010	Aim was to investigate the effect of the widely used anti-muscarinic drug tolterodine extended release on human HRV	PNS Blockade: Tolterodine ER 4 mg PO, 8 mg PO or placebo	PNS Blockade	30 healthy female volunteers (mean age 23.7 ± 2.3 years)	5 minutes	RMSSD, SDNN	VLF, LF, HF, total power, LF/HF	NA
Silva et al., 2009	Investigated the agreement between parametric autoregressive method (AR) and nonparametric FFT methods of HRV calculation after pharmacological blockade of the ANS in rats	PNS blockade: Methylatropine (1mg/kg IV); SNS Blockade: propranolol (1mg/kg IV). Combined ANS blockade: Propranolol (1mg/kg IV) followed by Methylatropine (1mg/kg IV)	SNS Blockade, PNS Blockade, Combined ANS Blockade (SNS/PNS)	10 male WKY rats (18-weeks)	10 minutes	Mean PI	VLF, HF power, HF n.u., LF power, LF n.u., total power, LF/HF	NA
Silva et al., 2017a	Compare spectral and symbolic analysis as tools to assess the ANS cardiac modulation in conscious rats	SNS blockade: Methylatropine (2mg/kg IV bolus followed by 0.4 mg/kg/hour infusion); PNS Blockade: Atenolol (4mg/kg IV bolus followed by 0.8 mg/kg/hour infusion); Reflex PNS Stimulation: Phenylephrine (1 ug/50 uL/minute); Reflex SNS Stimulation: Sodium Nitroprusside (0.6 ug/10uL/minute)	SNS Blockade, PNS Blockade; Reflex SNS Stimulation; Reflex PNS Stimulation	36 male WKY rats (280-300g)	10 minutes	Mean RR, SDNN, RMSSD	VLF, HF power, HF n.u., LF power, LF n.u., total power, LF/HF	Symbolic analyses 0V%, 1V%, 2LV%, 2UV%
Silva et al., 2017b	To study the effect of cardiac autonomic control on the nonlinear HRV methods multiscale entropy (MSE) and detrended fluctuation analysis (DFA) during periods properly long enough (60 minutes) to perform nonlinear HRV analysis.	SNS Blockade: Atenolol (4mg/kg bolus then 0.8 mg/kg/hour); PNS Blockade: Methylatropine (2 mg/kg bolus then 0.4 mg/kg/hour) Combined ANS Blockade: Atenolol (4mg/kg bolus then 0.8 mg/kg/hour) + Methylatropine (2 mg/kg bolus then 0.4 mg/kg/hour)	SNS Blockade, PNS Blockade, Combined ANS Blockade (SNS/PNS)	18 adult male WKY rats (280-300g)	60 minutes	RMSSD, SDNN, mean RR	NA	DFA alpha short, alpha mid, alpha long, SampEn
Tan et al., 2009	Examined 1. If fractal properties of the heart period measured over 20 min are reproducible within individuals across different experimental sessions and 2. If the physiologic effects of combined SNS and PNS blockade are reliably reflected by fractal indices of heart period	Combined Autonomic Nervous System Blockade (both SNS and PNS): Atenolol (0.2 mg/kg IV) + Atropine (0.04 mg/kg IV)	Combined ANS Blockade (SNS/PNS)	10 healthy volunteers (7 men, 3 women); 23-28 years of age	20 minutes	SDNN	NA	DFA

Abbreviations. ɑ1-AR: alpha-1 adrenergic receptor; ANS: autonomic nervous system; ß1-AR: beta-1 adrenergic receptor; BP: blood pressure; BPM: beats per minute; CV: cardiovascular; ECG: electrocardiogram; HF: high frequency; HR: heart rate; HRV: heart rate variability; IP: intraperitoneal; kg: kilogram; LF: low frequency; mg: milligram; n.u.: normalized units; PNS: parasympathetic nervous system; PO: by mouth; SNS: sympathetic nervous system; SQ: subcutaneous injection; ug: microgram; uL: microliter.

Sympathetic Nervous System Blockade

A total of fourteen studies evaluated HRV response to sympathetic blockade using propranolol and clonidine in human studies,^17,18 and propranolol, DL-propranolol hydrochloride, atenolol, metoprolol, phentolamine, and prazosin in pre-clinical rodent studies.^{28–30,32,33} Propranolol and atenolol were the primary medications used to block effects of the SNS by binding to β-ARs, while phentolamine, clonidine, and prazosin were used to block ɑ-ARs (Table 2; see Supplemental Table S1 for medication routes and dosages).

Pre-Clinical Rodent Studies

Time Domain HRV Indices

In pre-clinical rodent studies that used β-AR blocking medication (i.e., propranolol, atenolol, metoprolol), pharmacologic SNS blockade was associated with an increase in the mean RR (of the QRS complex) interval (msec) in most studies,^{27,29–31,33,34} with the exception of Silva et al.,³² who reported no RR interval change and Kur’yanova et al.²⁸ who only reported a change in female rats. Additionally, the SDNN decreased in the three animal studies using β-AR blockers who reported this metric.^27,33,34 rMSSD decreased with pharmacologic SNS blockade in one pre-clinical study using atenolol ³³ and another when prazosin was used,²⁹ but pharmacologic SNS blockade was not associated with a change in rMSSD in the other rodent studies.^27,29,34

Frequency Domain HRV Indices

In most animal studies SNS blockade was associated with a decrease in total power in msec²,²⁷ LF power in msec², ^27,30,31,33 HF power in msec²,^28,31,33 normalized LF in n.u.,^27,33 and the LF/HF ratio.^27,32,33 Normalized HF power in n.u. increased in two studies,^27,33 but did not change in another ³² when β-AR blocking medication was administered to rodents.

Non-linear HRV Indices

After β-AR blocker administration to rodents, the non-linear HRV indices Symbolic analysis 0V% and Symbolic analysis 1V% decreased,³³ Symbolic analysis 2UV% increased,³³ while there was no change in the Symbolic analysis 2LV%.³³ Additionally, approximate entropy was reported in one pre-clinical study and increased with propranolol and phentolamine administration.²⁷ There was no response to β-AR blocker in sample entropy (i.e. a measure of the heart rate signal complexity and regularity) when used as a standalone metric,¹⁷ but when sample entropy was split into multiple scales, sample entropy (scales 1-5) had no response, but sample entropy (scales 6-30) decreased with pharmacologic SNS blockade (Supplemental Table S4).³⁴

Human Studies

Time Domain HRV Indices

In humans, pharmacologic SNS blockade was associated with an increase in mean RR interval (msec),¹⁸ but no change in SDNN.¹⁸ No other time domain HRV indices were reported with SNS blockade.

Frequency Domain HRV Indices

Human studies using frequency domain indices reported no change in VLF power in msec²,¹⁸ LF power in msec²,^17,18 HF power in msec²,^17,18 or normalized LF in n.u. ¹⁷ with pharmacologic SNS blockade. SNS blockade with propranolol was not associated with a change in the LF/HF ratio,¹⁸ but when investigators administered clonidine to the same participants, there was a decrease in the LF/HF ratio.¹⁸

Non-linear HRV Indices

After β-AR blocking medication, the non-linear HRV metric approximate entropy increased when subjects were standing, but there was no change when they were supine in one study.¹⁷ In the same study, there was no response to β-AR blockade in sample entropy non-linear indices in both the sitting and standing positions.¹⁷

Parasympathetic Nervous System Stimulation

Researchers evaluated HRV response to PNS stimulation in eight studies using a variety of medications including low-dose atropine, clonidine, scopolamine, and phenylephrine in humans and pre-clinical studies. Low-dose atropine and scopolamine work by centrally activating muscarinic receptors, while clonidine works though central modulation of ɑ-ARs (Table 2; see Supplemental Table S1 for medication routes and dosages).

Human Studies

Time Domain HRV Indices

All human studies that used medications to stimulate the PNS reported an increased RR interval,^{18–20,23,24} and rMSSD also increased with PNS stimulation in the human study that evaluated this metric.¹⁹ Responses of the SDNN metric to PNS stimulation in humans were mixed with one study reporting a SDNN increase,¹⁹ and two studies reporting no change associated with medication administration.^18,20

Frequency Domain HRV Indices

In humans, clonidine administration did not result in a change in any frequency domain HRV measure including absolute VLF, LF, and HF power and normalized LF.¹⁸ Two studies measured frequency domain HRV indices with low-dose atropine, and administration resulted in a decreased normalized LF power,²⁰ but no change in normalized HF or normalized LF/HF.^19,20

Non-linear HRV Indices

In humans, scopolamine and low-dose atropine at doses of 2.2, 3.2, and 4.4 micrograms/kilogram increased the non-linear Poincare plot metric SD1 in both studies reporting this metric.^19,23 The SD2 response to pharmacologic PNS stimulation had conflicting results where there was no change in SD2 with scopolamine administration or at 3.2-4.4 micrograms/kilogram,^19,23 but when atropine was infused at 2.2 micrograms/kilogram there was an increase in the SD2 Poincare plot metric.

Symbolic analyses 0V%, 1V% and 2LV%, along with DFA alpha1 and alpha2 did not change with pharmacologic PNS stimulation.^18–20 The Symbolic analysis 2UV% metric increased in one human study using phenylephrine and low-dose atropine for PNS stimulation,²⁰ but there was no change in Symbolic analysis 2UV% in the other low-dose atropine human study.¹⁹

Parasympathetic Nervous System Blockade

Sixteen pre-clinical and human studies evaluated HRV response to PNS blockade using atropine, methylatropine, oxybutynin, tolterodine, and hexamethonium. All medications used were acetylcholine antagonists by blocking muscarinic receptors, except for hexamethonium, which acted on nicotinic acetylcholine receptors (Table 2; see Supplemental Table S1 for medication routes and dosages).

Pre-Clinical Rodent Studies

Time Domain HRV Indices

Atropine and methylatropine administration in rodents resulted in a decrease in the mean RR interval,^29–34 except for one study.²⁷ SDNN and rMSSD also decreased in most studies when atropine or methylatropine was administered to rodents,^29,33,34 except in the one study that did not report a change in the RR interval.²⁷

Frequency Domain HRV Indices

Pharmacologic PNS blockade with atropine, methylatropine and hexamethonium were associated with a decrease in absolute total spectral power,²⁷ absolute VLF power,^29,33 absolute LF power,^27,31,33 and absolute HF power.^27,30,31,33 Normalized frequency unit responses to pre-clinical pharmacologic PNS blockade studies had conflicting results, with a decrease in normalized LF power in two studies,^27,33 and no change in the other,³² along with an increase in normalized HF power in two studies^27,33 and a decrease in normalized HF power in the same study previously reporting no change in normalized LF power.³² The LF/HF ratio also had various associations with PNS blockade including an increase in one study,³² a decrease in one study,²⁷ and no change in another study.³³

Non-linear HRV Indices

Only one study reported Symbolic analysis non-linear indices with pharmacologic PNS blockade, but in this study, methylatropine was associated with a decrease in Symbolic analysis 2UV% and no change in 0V%, 1V% and 2LV% ³³. Sample entropy scales 1-5 and scales 6-30 decreased,³⁴ while approximate entropy increased,²⁹ after atropine was administered for pharmacologic PNS blockade (Supplemental Table S4).

Human Studies

Time Domain HRV Indices

In humans, atropine and methylatropine administration resulted in a decrease in the mean RR interval.^18,20,21 The SDNN also decreased with atropine/methylatropine administration.^18,20 rMSSD responses varied across medications used to block the PNS where oxybutynin administration was associated with an increase,¹⁶ whereas tolterodine was associated with a decrease.²⁵

Frequency Domain HRV Indices

There were consistent responses to pharmacologic PNS blockade in humans using absolute power frequency domain HRV indices. VLF power,^18,25,29 LF power,^17,18,21 and HF power^17,18,22,25 all decreased when researchers administered PNS blocking medication. Normalized frequency units had less coherent results, where atropine was associated with an increase¹⁷ and decrease²⁰ in normalized LF power, a decrease in normalized HF power in the one study that reported this metric with atropine administration,²⁰ and oxybutynin administration was associated with an increase in normalized HF power.¹⁶ In response to pharmacologic PNS blockade, the LF/HF ratio increased in most human studies,^16,20,25 with the exception of Castiglioni et al.¹⁸

Non-linear HRV Indices

After atropine/methylatropine administration, approximate entropy, sample entropy, and Poincare plot indices SD1 and SD2 all decreased in humans.^17,23 Symbolic analyses were performed in associated with PNS blockade in one human study, and atropine administration was associated with an increase in 0V% and a decrease in 1V%, 2LV%, and 2UV%.²⁰

Combined Blockade

Six studies evaluated HRV response to a combined PNS/SNS blockade using atropine or methylatropine and a β-AR blocking medication (propranolol, metoprolol or atenolol; Table 2; see Supplemental Table S1 for medication routes and dosages).

Pre-Clinical Rodent Studies

Time Domain HRV Indices

With combined PNS/SNS blockade, there was a decrease in the RR interval,^29,34 decrease in SDNN,³⁴ and rMSSD decreased in one study,³⁴ but did not change in the other study reporting this metric.²⁹

Frequency Domain HRV Indices

One study reported frequency domain HRV indices after combined PNS/SNS blockade in rodents, but only one metric was reported. When methylatropine and metoprolol were administered to block the SNS and PNS, there was an associated decrease in absolute VLF power.²⁹

Non-linear HRV Indices

Combined PNS/SNS blockade was associated with a decrease in sample entropy (scales 6-30) but not scales 1-5 in rodents.³⁴

Human Studies

Time Domain HRV Indices

With combined PNS/SNS blockade, there was a decrease in the RR interval²¹ and SDNN²⁶ associated with β-AR blocking medication plus atropine administration in humans. No other time domain HRV indices were reported in human studies of combined pharmacologic autonomic nervous system blockade.

Frequency Domain HRV Indices

Combined PNS/SNS blockade led to a decrease in both absolute frequency domain HRV indices including absolute LF power^17,21 and absolute HF power.¹⁷ When frequency band indices were normalized, there was no change reported in normalized LF power when atropine and propranolol were administered to humans for combined pharmacological autonomic nervous system blockade.¹⁷

Non-linear HRV Indices

The non-linear indices approximate entropy and sample entropy both decreased with combined PNS/SNS blockade in humans (Supplemental Table S4).¹⁷

Sex as a Biological Variable

Eleven manuscripts reported data on male subjects only, six included females and males, one included female subjects only, and one did not report the sex of the participants (Table 2). In one study evaluating HRV response to pharmacologic PNS stimulation in groups of young (24.1 ± 0.4 years) and older (62.8 ± 1.1 years) subjects,²⁴ a sex-specific subgroup analysis revealed that women in the young group and men in the older group demonstrated higher average correlations between the change in HRV indices and the change in the RR interval (change from baseline to peak vagotonic dose) than average correlations across all subjects. A greater number of males in the younger group showed a negative correlation between SDNN, respiratory sinus arrhythmia (physiologic variation in heart rate with inhalation and expiration) and total spectral power, while more male participants in the older group showed negative correlations between rMSSD and respiratory sinus arrhythmia, although statistical differences between males and females were not calculated by the authors.²⁴

The one pre-clinical study that included male and female rats quantified time and frequency domain HRV metric responses to SNS blockade with atenolol stratified by sex.²⁸ Female rats had a significant decrease in mean HR and absolute HF spectral power, after SNS blockade compared to baseline, while male rats had no significant differences in these indices compared to baseline.²⁸ After pharmacologic SNS blockade, female rats also had significantly lower mean heart rate, RR interval variability range, and absolute HF spectral power compared to control rats while the same differences were not reported in males.²⁸

Discussion

The aim of this integrative review was to review and summarize time domain, frequency domain, and non-linear HRV responses to pharmacologic autonomic nervous system modulation in pre-clinical and translational research studies. The main findings were: 1) SNS blocking medications generally resulted in a decrease in SDNN. There were not consistent frequency domain results for SNS blocking medications across humans and animals; 2) stimulation of the PNS was associated with an increase in rMSSD, absolute LF and HF power, and the Poincare plot metric SD1; 3) pharmacologic PNS blockade decreased SDNN, rMSSD, absolute VLF power, absolute LF power, absolute HF power, sample entropy, and Poincare plot metrics SD1/SD2 consistently across studies; and 4) combined autonomic nervous system blockade (pharmacologic blockade of PNS and SNS) was associated with a decrease in absolute VLF power, LF power, and HF power (see Text Box 1 for a summary of HRV responses in reviewed studies).

Summary of HRV Metric Responses.

• PNS stimulation:

  • Time domain- mean RR and rMSSD increase in both pre-clinical and human studies
  • Frequency domain- normalized HF and the LF/HF ratio did not change with pharmacologic PNS stimulation, although absolute LF and HF power increased in animals, but not humans
  • Non-linear- increase in the Poincare plot metric SD1 in the two human studies that reported it. Symbolic analysis metrics 0V% and 2LV% did not change in animals or humans, 1V% did not change in humans but decreased in rodents, and the response of the 2UV% metric increased in rodents and varied across humans with PNS stimulation research.
• PNS blockade:

  • Time domain- mean RR, SDNN and rMSSD consistently decreased with pharmacologic modulation in pre-clinical and human studies
  • Frequency domain- pharmacologic PNS blockade was also associated with a decrease in multiple frequency domain HRV indices in both rodents and humans, and the most consistent results occurred in the absolute spectral power indices including a decrease in absolute VLF power, absolute LF power, and absolute HF power.
  • Non-linear- increase in Symbolic analysis 0V% with PNS blockade in animals, but no change in these indices in humans. Similarly, there was a decrease in Symbolic analysis indices 1V%, 2LV%, 2LV% and 2UV% with atropine/methylatropine administration in rodents, and none of these indices were significantly altered in humans, with the exception of 2UV% that also decreased when atropine was administered for PNS blockade.
• SNS stimulation:

  • There were no studies using medications to stimulate the SNS through α-AR, β-AR or muscarinic receptors in the reviewed primary sources
• SNS blockade:

  • Time domain- the most consistent time domain HRV metric in both pre-clinical and human studies was SDNN with a consistent decrease across rodent studies and no change in humans.
  • Frequency domain- agreement across associations of pharmacologic SNS blockade with frequency domain indices were much more consistent in humans versus rodents, but a majority of these studies reported no frequency domain metric changes associated with SNS blockade. Conversely, absolute LF and HF power decreased in rodents when SNS blocking medication was administered, but the studies reporting a decrease versus no change in metric response were mixed.
  • Non-linear- there was a large variety of non-linear indices reported in both pre-clinical and human studies associated with SNS blockade so assessing agreement across studies was challenging. There was no change in SampEn and an increase in ApEn in the studies reporting these metrics.

The major themes noted while reviewing HRV responses to pharmacologic modulation of the autonomic nervous system were lack of measurement and reporting standardization including medication doses/route, recording length, and indices selected for time/frequency domain indices, and a variety of newly created non-linear measures for HRV quantification. Furthermore, a significant issue that decreased the overall quality scores of human HRV studies was small and homogenous (male only) sample sizes, in addition to the omission of data analysis procedures. Please see Table 3 for recommendations associated with each discussion theme.

Table 3.

Recommendations for Conducting Heart Rate Variability Research Based on Literature Review Synthesis Results

Theme	Issue	Recommendation
Interpretation of HRV Research Results	The results of different analysis metrics of HRV (time, frequency, nonlinear) are not consistent in the literature	HF power has consistently been a robust marker of PNS influence on the heart, and the absolute and normalized HF frequency band results were consistent across multiple studies, especially with PNS blockade. Conversely, the interpretation of the LF band frequency domain measure, and its association with SNS modulation has been controversial in the literature. Both LF and HF are influenced by PNS modulation, and therefore responses in LF should not be directly paralleled to changes in SNS modulation.2 Similarly, the LF/HF ratio should not be described as sympatho-vagal (or SNS/PNS balance), as PNS innervation to the heart influences both the LF and HF frequency bands. Sayin et al.31 also identified differences after PNS stimulation and SNS blockade from baseline heart rate at rest, respectively, were not captured by differences in the LF/HF ratio. When interpreting the responses of the LF band and LF/HF ratio, conclusions and physiologic significance should be interpreted cautiously, especially when making conclusions associated with SNS stimulation of the heart. A reproducible non-invasive measure to quantify SNS influence to the heart is needed and is a topic of current research.46,47 Reported non-linear indices varied greatly across studies, and some non-linear indices (i.e., approximate entropy) require a correction factor in order to be clinically meaningful.48 Evaluating benchmarking and calculation of non-linear indices in previously published research will ensure HRV results are reported in a reproducible and clinically relevant manner.
Sex as a Biological Variable	Females were not consistently included in the synthesized studies and studies were not powered to evaluate male and female differences	Researchers should consider larger sample sizes enrolling both male and female participants. As the limited research that included males and females reported large variance across sexes, larger sample sizes may be needed to evaluate sex-specific differences more comprehensively in heart rate and HRV response to pharmacologic autonomic nervous system modulation, so results are not skewed by outliers. Furthermore, HRV differences in healthy human males and females in time and frequency-based HRV metrics have been established in a meta-analysis evaluating several thousand participants.49 Examining sex as a biological variable with HRV analyses is necessary to adequately determine if males and females have the same HRV responses when pharmacologic interventions are used to modulate the SNS and/or PNS.
Medications used for ANS Modulation	Doses and administration routes differ across studies make cross study comparisons difficult	Future research protocols should include pre-defined heart rate and/or blood pressure metric endpoints to confirm adequate medication dose has been achieved (with minimal side effect profiles) when pharmacological modulation of the autonomic nervous system is used to evaluate HRV response. Half-life of the medication in context of route, dose, and clinical characteristics of the receiving human or rodent population should be considered when determining total recording length, optimal time to determine heart rate or blood pressure response, and the desired recording window to capture HRV response to the pharmacologic intervention.
Recording Length	Heart rate recording length varied significantly across studies	Different measures of HRV have different recommendations for minimum recording length. For example, the Task Force of the European Society of Cardiology recommended a minimum of one minute of recording is needed to assess HF, 2 minutes for LF, 5 minutes for SDNN and rMSSD, and 20 minutes for analyses derived from the Poincare plot.5 Studies published after the 1996 HRV guidelines have suggested shorter periods of time (60 seconds) may be acceptable for certain time domain analyses,50,51 although frequency domain and non-linear indices still require more recording time. Furthermore, it is suggested for the ultra-low frequency domain HRV analysis, that a minimum of 24 hours recorded should be analyzed for accurate results 2. The varying lengths of recording time again need to be considered when comparing HRV indices across studies, as longer recordings have been associated with increased HRV due to the increase in data available 2. For studies incorporating non-linear measures, recording periods of at least 60 minutes are recommended to adequately capture enough data to evaluate non-linear HRV indices.34 Importantly, when using HRV as measure for cardiovascular risk stratification, longer recording times may be required until future validation of short-term HRV recording occurs, as short-term HRV is not consistently predictive of cardiovascular outcomes compared to long-term measurements at this time.43,44
Methods Reporting	Research and data collection methods were not consistently reported across studies	In addition to recording for adequate periods of time, rigorous and transparent methods in HRV data collection, cleaning and analysis is important to assess to ensure reliable and valid results. Components of proper analysis include use of a sampling rate of at least 250 Hz and removal of artifact.35 Sampling rates of 500 Hz were recommended in the initial Task Force Guidelines in HRV research,5 but recent recommendations state that 250 Hz may be an adequate sampling rate in human subjects.35 Guidelines for reporting on HRV have been published by Quintana et al.,35 where the Guidelines for Reporting Articles on Psychiatry and Heart rate variability (GRAPH) checklist was introduced on minimum reporting standards. To improve transparency and study reproducibility, future research with HRV indices should minimally consider reporting on participant selection (intervention and control group selection, inclusion criteria, demographics), interbeat interval collection (hardware/software details, data collection details including time and sampling rate), interbeat interval analysis and cleaning (calculation, artifact identification, data loss, removal/imputation, and cleaning), and HRV calculation (method of HRV analysis, frequency bands used).35
Frequency Band Parameters	In frequency-based HRV research, frequency band parameters varied greatly and were not always reported	As resting heart rates differ significantly between rodents and humans,52 the recommended frequency band parameters also differ and should be clearly reported. In humans, there are established recommendations for frequency band ranges to evaluate spectral power in the HF (0.15-0.4 Hz), LF (0.04-0.15 Hz), and VLF (0.0033-0.04 Hz) bands.2 There is more variance for recommended frequency band ranges in rodent data, but a manuscript focused on this topic concluded the previously recommended HRV parameters in mice (HF [1.5-5 Hz] and LF [0.15-1.5 Hz] bands) were optimal to adequately capture parasympathetic modulation to the heart as the HF frequency band range was broader and could capture the average peak around 2.5 Hz.45 Frequency domain measures can be expressed including absolute power (ms2), relative power (%), or in normalized units (see Table 1 for definitions). Future studies should consider sharing their raw and transformed data, and we recommend including time and frequency domain HRV analyses with reporting non-linear HRV indices, especially when new non-linear analysis methods are introduced.
Environmental Variables	Variables that may influence HRV response were not always reported in the synthesized studies	In HRV research, great care should be taken to maintain stable environmental and behavioral conditions as to not confound the autonomic nervous system tone or response to stimuli and therefore recommendations on a standard postoperative waiting period may be of interest to future pre-clinical researchers. Time of day the HR data was collected for HRV analysis is also important to standardize and report. Measurements taken later in the day can be influenced by many different factors including stress, exercise, caffeine, and smoking.

Abbreviations. ɑ1-AR: ANS: autonomic nervous system; ß1-AR: beta-1 adrenergic receptor; HF: high frequency; HR: heart rate; HRV: heart rate variability; LF: low frequency; PNS: parasympathetic nervous system; SNS: sympathetic nervous system.

Measurement Differences and Considerations: Sex as a Biological Variable

Female subjects were seriously underrepresented in both human translational research and pre-clinical rodent research of the studies evaluated. In most of the translational research that recruited both males and females,^19,21,22 researchers performed HRV analyses at the whole group level and sex-specific differences in HRV indices or responses to pharmacologic autonomic nervous system modulation were not evaluated. Additionally, sample sizes were small in the studies that recruited both males and females ranging from ten^21,22 and 14 participants¹⁹ to a maximum of 61 participants.²⁴ Therefore, these studies were likely not powered to detect sex differences, even if sex-specific HRV analyses were performed. The study incorporating males and females with the largest sample size stated that the sex-specific subgroup differences in average correlations between the change in HRV indices and the change in the RR interval were driven by one outlier for SDNN and rMSSD in the young female group and a higher percentage of negative correlations for SDNN, rMSSD, respiratory sinus arrhythmia and total spectral power in the young and older male groups.²⁴ Conversely, one study recruited female subjects only,²⁵ which also prevented sex-specific comparisons. In the pre-clinical HRV research, only one study included both male and female rats,²⁸ but all HRV indices that the researchers reported stratified responses to SNS blockade and PNS stimulation by sex. Disparate responses to autonomic nervous system-modulating medications between males and females were seen in commonly reported HRV indices including mean RR interval, absolute HF power, absolute LF power and absolute VLF power,²⁸ further justifying that including sex as a variable is necessary for rigorous HRV research. Additional pre-clinical studies powered to detect sex differences, if present, will be important to replicate these findings.

Measurement Differences and Considerations: Autonomic Nervous System Modulating Medications

Different doses and administration of drugs make cross study comparisons challenging as study subjects may have had varying autonomic nervous system responses across doses/routes. For example, four human studies administered medications by mouth,^16,18,22,25 one by transdermal medication application,²³ and eight administered intravenously by one-time intravenous push^{17,18,20,21,24,26} or continuous infusion.^19,20,23 In pre-clinical animal studies, two studies administered medication through the intraperitoneal route,^28,30 one subcutaneously,³¹ and five intravenously by one-time intravenous push^27,29,32 or continuous infusion.^33,34 Different medication routes not only have different pharmacokinetic implications, but medications subcutaneously or by mouth might have vastly different peak response times across subjects, which makes the appropriate time for HRV data collection to synchronize with peak medication response somewhat subjective.

Differing doses of medications administered by the same route also occurred. For example, researchers used low-dose atropine for PNS stimulation in two studies,^19,24 but different initial doses were used in each protocol. Fortunately, both studies used HR response criteria as an endpoint to evaluate adequate stimulation of the PNS before data collection.^19,24 In other studies that employed varying doses of the same medication for autonomic nervous system stimulation or blockade, physiologic heart rate or blood pressure endpoints to determine adequate autonomic nervous system response were not used.

Measurement Differences and Considerations: Recording Length and Sampling Rate

Many of the studies that evaluated HRV response to autonomic nervous system-modulating medications used a combination of more than one HRV analysis group (i.e., time domain, frequency domain, non-linear) in their study. For studies that included time domain HRV indices, a majority of studies analyzed 5 minutes of electrocardiogram data^{16,20,25,28,31} or longer,^{18,27,29,32–34} but there was a large variance in the recording period analyzed. Three studies used 3 minutes of electrocardiogram data,^19,24,30 two studies used 10 minutes,^32,33 two studies used 20 minutes^18,27 and two studies used 30–60 minutes.^29,34 For studies that included frequency domain HRV indices, a majority of studies again analyzed 5 minutes^{16,17,20,25,28,31} or more of electrocardiogram data.^{18,27,29,32,33} Otherwise, two studies used 3 minutes of electrocardiogram data,^24,30 two studies used 10 minutes,^32,33 two studies used 20 minutes^18,27 and one study used 30-60 minutes of data for analysis.²⁹ For non-linear HRV analyses, Silva et al.³⁴ stated in their research aims that longer periods of data collection are necessary to capture enough data to accurately quantify non-linear HRV metric trends. Although there was still a large variance of recording time used for non-linear analysis methods, more studies used longer recording times for analysis. One study used 3 minutes of recording for analysis,¹⁹ three studies used 5 minutes of recording,^17,20,28 one study used 10 minutes,³³ four studies used 20 minutes,^18,23,26,27 and one study used 60 minutes.³⁴

Adequate sampling rate is an additional consideration that is important to consider in the evaluation of HRV research as low sampling rates may miss beat to beat heart rate responses to autonomic nervous system modulating medications that may influence HRV results. Not all of the studies reviewed reported sampling rate in their methods (this data was lacking in 5 manuscripts), but signal data was captured at a rate of 250 Hz or higher in the manuscripts that reported sampling rate. With non-linear HRV analyses, there was a large range of different indices introduced or reported, so it was difficult to compare the results across studies. Furthermore, non-linear HRV responses were not always compared to time or frequency domain analysis to evaluate agreement. Finally, the methods used in data cleaning or handling of artifacts in the recording may bias HRV results and are important to consider when evaluating HRV responses across studies.³⁵ However, in many of the manuscripts reviewed, the methods and workflows used to detect and correct for artifact were not explicitly described.

Measurement Differences and Considerations: Environment/Metadata

Different environmental conditions for both animals and humans can additionally alter results. Sayin et al.³¹ allowed rodents two weeks to recover after telemetry transmitters were implanted for HRV recording and analysis, while Miyabera et al.²⁹ and Silva et al.³² started experiments less than three days after surgical telemetry implantation. Time of day that results were collected is another factor that might bias HRV results. Circadian changes in intrinsic heart rate and HRV have been reported in rats,^36,37 and humans,³⁸ and results should be interpreted based on the time the measurement was obtained. The physiological processes that generate short-term measurements include vagal modulation, respiratory sinus arrhythmia, and baroreceptor reflex.¹¹

Limitations

One limitation in our synthesis of HRV research is that not all studies were well powered and some of the studies had lower quality scores based on the Crowe Critical Appraisal Tool. Although the Crowe Critical Appraisal Tool is a continuous metric used to objectively evaluate research studies, no cut points or guidelines exist to determine “good” versus lower quality studies, and therefore all studies were included in this review regardless of their quality score. The intervention, measurement, and analysis variability across studies challenged our ability to compare HRV metric responses to pharmacologic autonomic nervous system stimulation or blockade across studies. Therefore, although trends in time, frequency and non-linear HRV metric responses to autonomic nervous system modulating pharmacological challenges are reported, varying HRV responses should be examined in light of the study design and measurement/analysis considerations. Nevertheless, we were able to find consistent responses to autonomic nervous system modulating medication across the time domain, frequency domain and non-linear HRV indices which identified consistent and reliable HRV metrics that would be of clinical utility in future research. This review also provided an opportunity to identify study design and HRV analysis variations across studies to make design, analysis and reporting recommendations to improve future research. Another limitation is we only included studies that were reported in English, which may reduce the generalizability of the review to non-English speaking populations. Nevertheless, many of the manuscripts included data from populations where English is not the primary language and therefore, we believe we were able to synthesize data from different populations in our final analysis.

Conclusions and Clinical Implications

In this review, we synthesized time domain, frequency domain, and non-linear HRV metric responses to pharmacologic autonomic nervous system modulation. We found consistent time domain responses to SNS blocking medications, but the frequency domain HRV responses to SNS blockade were more variable. Stimulation and blockade of the PNS was associated with more consistent changes in time domain, frequency domain, and some non-linear HRV indices across studies, while combined autonomic nervous system blockade (blockade of both PNS and SNS) was associated with a change in frequency domain HRV indices only. Therefore, selection of HRV analysis indices should be guided by the research question and branch of the autonomic nervous system of interest to ensure the HRV indices chosen reliably capture HRV responses. In the limited research that included comparisons stratified by sex, disparate HRV metric responses were seen across males and females emphasizing the need to include sex as a biological variable in all future research incorporating HRV. Recommendations to improve the reproducibility of future HRV research and to facilitate improved comparison across studies were made including standardization of recording, analysis and metric decisions, and more thorough reporting of HRV indices in published studies. Publication of high-quality research will continue to validate the current understanding of the interaction between pharmacologic modulation of the autonomic nervous system and the different components of HRV metric responses.

Noninvasive measures to stratify and predict risk for cardiovascular complications are consistently reported in the literature,^39–42 supporting the demand for the evaluation of the validity and reproducibility of physiologic biomarkers like HRV in translational research to inform clinical care. As alterations in the autonomic nervous system are associated with an increased risk for adverse patient outcomes and mortality, using noninvasive indices like HRV as biomarkers to capture autonomic nervous system status will inform future translational research on cardiovascular risk reduction. Nevertheless, although HRV has been used in cardiovascular risk assessment for several decades,^4,5 questions surrounding practical considerations in HRV monitoring and interpretation remain.^2,11,35

This review synthesized articles that systematically manipulated the autonomic nervous system through pharmacologic stimulation or blockade of the SNS and/or PNS, and used this collective information to provide recommendations for future researchers interested in the use of HRV as a measure of cardiovascular physiology or pathophysiology. We determined that careful attention to experimental setup and recording length are incredibly important if using HRV as a biomarker for cardiovascular risk reduction, as short-term HRV measures have not yet been shown to be consistently predictive of cardiovascular outcomes compared to long-term measurements.^43,44 Furthermore, consistent practices in sampling rates, analysis strategies and HRV metrics reported are integral for the field to adequately assess HRV in future research to make informed decisions about clinical care. Our findings reflect that HRV metrics consistently reflect stimulation and blockade of the PNS, supporting their use as a biomarker when determining PNS reactivity. Ongoing validation of appropriate recording length to use HRV as a cardiovascular risk stratification tool and the use of HRV measures in the evaluation of the SNS provide opportunities for future translational research.

List of Abbreviations:

ɑ-AR

Alpha-Adrenergic Receptor

ß-AR

Beta-Adrenergic Receptor

HRV

Heart Rate Variability

HF

High Frequency

LF

Low Frequency

PNS

Parasympathetic Nervous System

rMSSD

Root Mean Square of the Standard Deviation of Normal-Normal Intervals

SDNN

Standard Deviation of Normal-Normal Intervals

SNS

Sympathetic Nervous System

VLF

Very Low Frequency