Setting and participants
This study was conducted at the Medical University of Graz Department of Orthopaedics and Trauma. A total number of 189 people were asked between October 2016 and June 2017 to participate in the study. Participants were defined either as patients with traumatic hip fracture, followed by a hip fracture surgery (HFS), or total hip arthroplasty (THA) admitted to surgery, and people without a current surgery, which comprised the control group. The study was performed according to the convention of the Declaration of Helsinki, 1964 and approved by the local ethics committee (EK-28-515 ex 15/16). Written informed consent prior to the investigation was obtained from all participants, after they had received detailed information about the investigation and the study protocol.
Testing conditions
Anthropometric measurements were conducted using a calibrated electronic scale (Seca Modell 799, Germany) to measure weight (participants were asked to remove shoes), and body height measured to the nearest 0.1 cm with an anthropometer (GPM 100, Rudolf Martin Antropometer, Switzerland). Calf circumference (CC) was taken using a tape measure. Measurements were performed according to the International Society for the Advancement of Kinanthropometry (ISAK) protocol [
20]. The testing time (between 3 and 7 PM) was kept constant, following circadian rhythms; room temperature was also kept constant and the laboratory was kept quiet to exclude any interrupting noise. Participants were required to abstain from caffeine, alcohol and heavy meals for two hours prior to testing. After familiarizing participants with the experimental protocol, disposable ECG electrodes were attached at three points on the participant´s chest. During the entire experimental procedure, patients had to sit quietly in a comfortable chair that was adjusted for each person, without speaking or moving abruptly. Our protocol consisted of an adaptation period followed by a 3 min measurement at rest, to record baseline results; this was measurement time point (MTP) 1. Thereafter, participants were instructed to press a grip strength dynamometer as strongly as possible for three seconds, twelve times with twelve seconds break after each turn, to produce physical stress. Finally, a further 3 min measurement at rest followed, to record recovery results (MTP 2); this concluded the investigation. Prior to and also following ECG recordings, blood pressure was measured using a standardized hospital device (Boso Clinicus I, Germany).
Data acquisition and preprocessing
Continuous ECG was recorded using the exercise physiology and software system Powerlab 8/35, with LabChart Pro Software all from ADInstruments (Sydney, Australia), with a sampling rate of 1000 Hz. Grip strength, derived from a grip strength transducer, ECG and respiration frequency, derived from a chest-strap, were indicated by three channels on the device. Disposable electrodes were fixed at the thorax (2-lead, 1 channel position), and bipolar limb derivation using an Eindhoven Lead II set-up was chosen. The recorded binary data were saved as European Data Format, EDF [
21]. The R-wave detection was carried out using a revised MATLAB-function (MATLAB®, Mathworks Natick, Massachusetts, USA) and the immediate beat to beat heart rate was also calculated using this function [
21]. Artifact handling was performed semi-automatically by a visible check of every signal, in combination with a Matlab-function which identified these signals according to the following criteria: (1) ectopic beats, (2) physiological limits and (3) maximal percentage of change in relationship to standard deviation of the signal. Therefore, we used time series with equidistant time steps, after resampling beat to beat values at 4 Hz, using piecewise cubic spline interpolation. Single artifacts were replaced by linear interpolation and only time series with 85% validity were accepted.
The signal measurement and the subsequent processing were performed according to international recommendations of
linear (time domain, TD: the changing of signals over time, and frequency domain, FD: the frequency of signals in a special range) and
nonlinear HRV parameters [
22].
In regards to linear parameters, Time domain variables of HRV were generated using the standard deviation of normal to normal R-R intervals, SDNN, in ms (R is the peak of a QRS complex—heartbeat, HR), which is responsible for the variations in heart beat, to reflect sympathetic and extreme vagal tone. The root mean square of successive heartbeat interval differences, RMSSD, in ms, correlates with the frequency domain variable ´high frequency´, HF, listed below, and estimates variations in the HR. Frequency domain variables, using the power spectral density, were differentiated into low frequency, LF, in ms2, defined by the power of the “low”-frequency band 0.04–0.15 Hz, and HF, in ms2, the power of the “high”-frequency band 0.15–0.4 Hz. Due to skewed distributions of frequency domain indexes, a natural logarithm, ln, transformation was applied; LnLF is related to mainly sympathetic factors, lnHF is related to the vagal influence, and lnLF/HF is considered to reflect the sympathovagal balance.
Nonlinear parameters, describing the SD of an average variability around a mean, are expressed in the standard deviation of the short-term NN, normal to normal R-R intervals, interval variability, SD1, in ms, the standard deviation of the long-term NN interval variability, SD2, in ms, and the ratio between SD2 and SD1, SD2/SD1.
Statistical analyses
Data are presented as means, M, ± standard deviation, SD, as well as minimum and maximum values. Normal distribution was checked by normal probability plots and homogeneity of variance was checked by Levene’s test. LF and HF were log transformed to achieve normal distribution. Group differences at baseline and recovery were investigated using ANOVAs (Analysis of Variance). The differences in change of groups over time was evaluated by a 3 (group: HFS, THA, control; between-subjects factor) × 2 (time: baseline, recovery; within-subjects factor) repeated measures ANOVA. The chi square test was chosen to validate group affiliations and gender distribution. A probability of p < 0.05 was considered to be significant. All statistical analyses were performed using the Statistical Packages for the Social Sciences, SPSS, IBM SPSS Statistics Version 24 for Windows (SPSS Inc., Chicago, USA).