Handgrip force steadiness in young and older adults: a reproducibility study

A method comparison study with the gold standard is usually warranted when introducing a new method [39]. We used a test-retest design with seven days between each session to test the reproducibility of the WBB method [39]. The term “reproducibility” refers to the variation in measurements made on a subject under changing conditions [39]. Reproducibility was the ability of our method to give consistent results on two occasions separated in time. This includes the ability to distinguish between subjects in a sample on repeated tests, referred to as reliability [39, 40], and the degree to which two measurements made on the same subject are comparable, referred to as agreement [40]. Following the Guidelines for Reporting Reliability and Agreement Studies, we use the terms reliability and agreement [40].

Our study-population consisted of a group of old adults for reproducibility testing and a smaller group of young adults for age-related comparison. The old participants were recruited by telephone lists in Ålesund municipality in Møre and Romsdal County, Norway. Participants were included if they were 55 years or older, judged themselves to be healthy and willing and able to be tested twice, one week apart. Participants were excluded if they had neuromuscular diseases (e.g. Parkinson’s disease), sequelae from stroke, suffered from dementia, did not understand Norwegian, suffered from acute illness or had recent (within 6 months) surgery.

The younger participants were recruited at university campus at the University of Oslo. The inclusion and exclusion criteria were the same for both groups, but age was between 20 and 30 years for the younger participants. All participants gave oral consent and the protocol was reviewed and declared not mandatory for submission by the Regional Committee for Medical and Health Research Ethics in Norway (2016/1505/REK Nord). The North Regional Ethics Committee, Norway, did not provide a full ethical evaluation and formal approval due to the ethically benign nature of our work (method development study). Still, all participants were required to provide oral consent to participate in the study and, thus, participant registration was synonymous with documented consent. None of the authors had access to any personal or potentially identifying participant information.

The WBB is a rectangular-shaped platform with one force transducer in each corner. Data was wirelessly streamed to a computer (Samsung Chronos series 7, Windows 8) and onto FysioMeter software (version 1.0.8, Bronderslev, Denmark) via a Bluetooth Human Interface Device. Each of the transducers channels delivered 16-bit digital data at approximately 100 Hz. These were filtered using 4th order Butterworth filter (cut-off 20 Hz). The software records and visualizes the data in real-time on a force-time curve.

Prior to testing we measured weight and collected information on height, handedness, smoking history and number of prescribed drugs taken daily. All test procedures were performed independent of each other during home visits in the old participants and on the Oslo University campus for the young participants. The raters (Andreas Wahl Blomkvist (AWB) and Fredrik Eika (FE), medical doctors) were involved in all tests performed. Initially, maximum isometric handgrip strength was measured for each hand in order to determine MVC using a method previously described and validated [36]. The participants were seated on a standard chair and held the WBB on their lap (Fig. 1). The order of handgrip steadiness measurements at 5%, 10% or 25% of MVC as target force was randomized to avoid order effects. Steadiness at low percentages of MVC (e.g. 5% and 10%) was chosen because it has been shown to vary the most with age and gender [15, 41]. 25% of MVC was chosen because Lodha et al. [5] recommended this measure as most objective assessment of steadiness when making comparisons between chronic stroke patients and age-matched controls. The software visualized a horizontal yellow target line (20 s in duration) corresponding to the target force in a force-time curve window. An ascending line of five seconds preceded this yellow target line, which ended with a five seconds descending line as shown in Fig. 2. Participants were told to squeeze the corner of the WBB with sufficient force such that their visual force curve, produced in real-time (black graph in Fig. 2), was superimposed on the yellow target line. Furthermore, the participants were instructed to focus on the task and to avoid speaking during the test. Starting with the left hand and alternating between hands, six measurements per target force were recorded in total (three for each side), giving 18 measurements per person per session. The five seconds long intro and outro, as well as the three initial seconds of the flat yellow line, were omitted from analysis to give the participant time to adjust to the flat yellow line. In total, three times 17 s of data were used for each hand and each target percentage (i.e. 5%, 10% and 25% of MVC).

For the old participants, a retest session was performed one week later with a same time-of-day approach using MVC values from the first session. Following randomization of starting target force, handgrip steadiness was tested as described above.

Statistical analyses were performed using the Statistical Package for the Social Sciences version 22 (SPSS Inc., Chicago, Illinois). Each side was analysed separately using the three times 17 s of data for each session. From these data, CV was calculated in two ways [1]: SD divided by the mean force (CVM) and [2] SD divided by the target force (CVT), i.e. the actual value of the 5%, 10% or 25% of MVC. CVM and CVT were multiplied by 100 to convert into percentages. Furthermore, FysioMeter estimated the area between the target force line and the force produced. This was done by calculating the approximated area between target force line (yellow) and the force-time line (black) for ~10 ms increments of time. Thus, for each side we derived three variables (i.e. CVM, CVT and Area) for each percentage of MCV (5%, 10% and 25%) for each session. Every variable and their between-session difference were tested for normality visually (histogram) and statistically (Shapiro-Wilk test). Two-sample t test was used to test for differences between the old and young participants. Next, paired t test was used to explore systematic bias between test and retest sessions. If the normality assumption was violated, Wilcoxon signed-rank test was used.

The difference between individual results and the mean result from both sessions was plotted in a simple scatter plot to check for heteroscedasticity prior to further analysis. Reliability was assessed by calculating intra-class correlation coefficient (ICC) with 95% confidence interval using absolute agreement in a two-way mixed model and the results of a single measurement. The results were interpreted on the recommended scale of poor (< 0.69), fair (0.70–0.79), good (0.80–0.89) and high (0.90–1.00) [42] reliability. Agreement was assessed by calculating the standard error of measurement (SEM), limits of agreement (LOA) and smallest real difference (SRD). SEM and LOA were calculated by multiplying the SD of the difference between each test session with \( \sqrt{1-\mathrm{ICC}} \) and 1.96, respectively. SRD was calculated by multiplying the SEM value with 1.96 • √2. Lastly, the agreement measures are given in percentages by dividing them with the mean value of both test sessions multiplied by 100.