Reliability and validity of the K-force grip dynamometer in healthy subjects: do we need to assess it three times?

Abstract

Introduction

Digital dynamometers to assess grip strength are becoming more common in research and clinical settings. The aim of the study was to assess validity and reliability of the K-force dynamometer compared to the Jamar dynamometer. We also aimed to assess differences over the course of three measurements.

Methods

Twenty-seven healthy participants were included. Three trials with the K-force and Jamar dynamometers were completed. Testing order was randomised. Intraclass correlation coefficients (ICCs) with absolute agreement assessed reliability and validity. Standard error of the measurement (SEM) and minimal detectable change (MDC₉₅) were calculated. Concurrent validity was assessed using Pearson’s correlations and ICCs. Differences between the three repetitions were assessed using one-way repeated measures ANOVAs.

Results

Both the K-force and the Jamar presented excellent intra-rater reliability with ICCs ranging from 0.96 to 0.97. The SEM ranged from 1.7 to 2 kg and the MDC from 4.7 to 5.7 kg for both dynamometers. The concurrent validity of the K-force was high (r ≥ 0.89). However, the K-force underestimated the grip strength by 4.5–8.5 kg. There was no change in grip strength with either dynamometer over the course of three trials.

Conclusions:

The K-force is reliable, but it underestimates grip strength by 4.5–8.5 kg compared to the Jamar dynamometer. K-force can be used to monitor progress over time but cannot be used to compare results against normative data. The use of a single measurement when assessing grip strength is sufficient when assessing healthy subjects.

Introduction

Grip strength (GS) has been used as a measure of physical function ¹ associated with upper limb and total muscle strength, frailty, sarcopenia and cardiovascular outcomes.^2–4 The utility of GS data obtained in healthy populations as a predictor for mortality risk, type 2 diabetes, length of hospital stay and post-surgical complications has also been demonstrated.^5–8 Grip strength measurement tools can also be used to assess GS, establish baselines and monitor progress throughout rehabilitation. Several instruments including sphygmomanometers and hand grip dynamometers are utilised to assess grip strength, however, the Jamar hand dynamometer (Jamar Hydraulic Hand Dynamometer, Sammons Preston, USA) is the most commonly used.

The Jamar hand dynamometer presents excellent reliability and is viewed as the gold standard for GS assessment. ⁹ This dynamometer has also been widely utilised to establish normative data for different age groups, which inform clinical decisions.^10–12 Specific methodology is vital to ensure reliability and accurate results when testing GS. The protocol recommended by the American Society of Hand Therapists (ASHT) has been widely used. ¹³ Grip strength testing should be performed in sitting, with the feet supported, arm unsupported, shoulder adducted, elbow flexed to 90°, wrist and forearm in neutral and wrist positioned between 0–30° of dorsiflexion and 0–10° of ulnar deviation. Three repetitions are recommended with 15 seconds rest in between each repetition. ¹⁴ A protocol employing three repetitions may result in testing fatigue (when consecutive strength outcomes decrease), or a learning effect (when consecutive strength outcomes increase). Investigation of differences between three consecutive repetitions will help determine if fatigue or a learning effect is relevant in a repeated measures protocol. If no differences exist across grip strength measures, then one repetition of grip strength assessment may improve clinical efficiency.

Although the Jamar is considered the gold standard in GS testing, technological advances have meant that computerized grip force assessment can be implemented with benefits such as efficiency in reporting and downloading of results. The K-force grip (Kinvent, Montpellier, France) dynamometer is lighter in weight (0.2 kg) compared to the Jamar (1.4 kg) and the grip shape is elliptical. The K-force can be connected to a smartphone showing real time results quickly and allowing users to connect with motivational games. The smaller and lighter K-force tester may prove useful in assessing and training grip strength in functional and elevated shoulder positions.

The K-force has shown high reliability and a high correlation with the Jamar hand dynamometer, however, previous investigations have reported lower GS measurements compared to the Jamar.^15,16 Differences between the testing devices must be recognised as much of the normative and predictive data available have been generated using the Jamar grip strength test.^11,12 When clinical decisions and allocation of resources are informed by GS measurements obtained using different dynamometers, it is important to determine whether results are valid. Any systematic errors must be identified and accounted for if GS is to be used as a determinant of current and future health status. ⁴ Thus, assessment of the validity, reliability and analysis of GS results achieved in healthy populations with the K-force in comparison with the Jamar device will help inform clinical decisions related to GS.

This study aims to establish the concurrent validity of the K-force instrument against the Jamar in a healthy population and to confirm the reliability of both instruments. It was also of interest to determine whether there was a significant difference between the three repeated GS tests within each instrument. We hypothesised that (i) both the Jamar and the K-force instruments would be reliable, (ii) K-force would not be valid compared to the Jamar, and (iii) there would be no difference between the three grip strength tests.

Methods

Study design

This study examined the reliability and concurrent validity of the K-force grip dynamometer (Kinvent, Montpellier, France) using the Jamar Handheld dynamometer (Jamar Hydraulic Hand Dynamometer, Sammons Preston, USA) as the gold standard. The Guidelines for Reporting Reliability and Agreement Studies (GRRAS) were followed for the completion of this study. ¹⁷ Ethical approval was sought and obtained from the Auckland University of Technology Ethics Committee (AUTEC number: 21/319).

Instruments

The K-force dynamometer (Figure 1) is a digital grip dynamometer that measures grip strength and transmits the results to a mobile phone (K-force App). It is reported to have a precision of 100 grams and a sampling frequency of 75 Hz. ¹⁸ The Jamar hand dynamometer is an analogue dynamometer that can be easily calibrated with a known weight. The Jamar hand dynamometer has been widely used in the literature and normative data are available.^10–12

Figure 1.

K-force dynamometer.

Participants

Healthy volunteers responded to social media posts, advertisements and verbal invitations from the research team. The inclusion criteria for this study were adults aged between 18 and 55 years of age. Participants were excluded if they had any pain with gripping, or any other neurological or musculoskeletal condition of the upper limb. Individuals who were unable to provide written consent for any other reason (including not being fluent in English) were also excluded.

Procedure

Participants’ demographics were recorded at baseline. The order of testing was randomised using an excel spreadsheet. Grip testing was performed using the dominant/non-dominant hand for the K-force and the Jamar dynamometer. The second handle position was utilised for the Jamar dynamometer. ¹⁴ A Physiotherapist Specialist (MO) with 27 years of upper limb assessment and treatment experience, completed data collection. Participants were asked to “squeeze as hard as possible” for the dynamometer tests. Participants performed the tasks sitting in a chair with feet placed on the ground, shoulder in neutral, elbow bent to 90° and elbow unsupported, forearm in a neutral position, and wrist between 0° and 30° dorsiflexion and between 0° and 15° of ulnar deviation. Participants performed a maximum contraction for 5 seconds three times, with a 30-second rest interval. They then repeated with the other hand. If they were randomised to perform the first test with the digital dynamometer, they then repeated the contractions using the Jamar dynamometer and vice versa. Testing of both dynamometers was conducted on the same day with a 10-minute interval between devices.

Data analysis

Data analyses were performed in R version 4.0.4 and RStudio v 1.2.1335. Validity and intra-rater reliability were examined using intra-class correlation coefficients (ICC) and 95% confidence intervals (ICC2,3). Measurement error and responsiveness of the tests were determined by calculating the standard error of the measurement (SEM) and the minimal detectable change (MDC₉₅: 1.96 × √2 × SEM) with a 95% confidence interval. ¹⁹ Concurrent validity between the K-force and the Jamar dynamometer was assessed using Pearson’s r correlations and ICC for absolute agreement. The MDC₉₅ for the Jamar dynamometer was utilised to determine whether the measurement difference between the K-force and the Jamar hand dynamometer was within an acceptable range. To provide a graphical representation of the agreement between the average K-force and Jamar measurements, Bland-Altman plots were created. ²⁰ The difference between the three trials was examined using repeated measures ANOVA. The Shapiro-Wilk test was utilised to assess data normality. Data were also screened for sphericity. In case of non-normally distributed data, log₁₀ transformations were completed prior to statistical analyses.

An a-priori power calculation was performed to determine the sample size required to identify a large association (Cohen’s effect size = 0.5) between the measurements recorded using the Jamar and K-Force dynamometers using Pearson’s r. This estimate was based on the results of a prior study assessing the convergent validity of Jamar and K-Force dynamometers. ¹⁶ Using G*power 3.1.9.7 software, ²¹ the alpha level was set to 0.05, power to 0.80. The hypothesis was set to one-tail. With these parameters, the required sample size was 23 participants.

Results

Participants and grip strength measurement’s reliability

A total of 27 healthy participants were included in the present study. Demographic data are reported in Table 1. Log₁₀ transformations were utilised across the whole dataset to normalise data. The intra-rater reliability analysis provided similar results on both the non-transformed and transformed data. To facilitate the interpretation and implementation of MDC₉₅ thresholds, the intra-rater reliability tables present the results from non-transformed data (See Table 2).

Table 1.

Demographic data for the participants included.

Category	All participants
Age (y) mean (SD)	37 (12)
Height (m) mean (SD	1.7 (0.1)
Weight (kg) mean (SD)	72.7 (13.8)
BMI (kg/m2) mean (SD)	25 (3.7)
Sex, n (%)
Male	10 (37)
Female	17 (63)
Dominant hand, n (%)
Right	25 (93)
Left	2 (7)
Work status, n (%)
Full time	24 (89)
Part time	3 (11)
Self-reported work Type, 22 n (%)
Sedentary	6 (22)
Light	4 (15)
Medium	14 (52)
Heavy	2 (7)
Very heavy	1 (4)

Legend: y = Years; m = Metre; kg = Kilogram; n = Number of participants; % = Percentage.

Table 2.

Grip strength (kg) for the dominant and non-dominant hand are reported as mean (standard deviation).

	Trial 1 mean (SD)	Trial 2 mean (SD)	Trial 3 mean (SD)	Average (SD)	ICC (95% CI)	SEM	MDC95
Dominant
K-force	31.8 (9)	31.7 (9.3)	31.5 (9.8)	31.6 (9.2)	0.96 (0.93 to 0.98)	1.9	5.2
Jamar	39.5 (12)	40.1 (11.8)	38.6 (11.5)	39.4 (11.7)	0.97 (0.95 to 0.99)	2.0	5.7
Non-dominant
K-force	31.3 (10.1)	30.6 (10)	30.2 (8.9)	30.7 (9.6)	0.97 (0.95 to 0.99)	1.7	4.7
Jamar	37.9 (11.6)	37.8 (10.3)	36.8 (10.3)	37.5 (10.6)	0.97 (0.94 to 0.98)	1.9	5.2

Legend: MDC = minimal detectable change, SEM = standard error of the mean.

Grip strength was assessed using both dynamometers for the dominant hand and non-dominant hand. Reliability was assessed across the three trials for each condition. The results are reported in Table 2.

Concurrent validity

The concurrent validity of the K-force against the Jamar dynamometer was assessed using Pearson’s correlation and ICCs. The two devices and trials were highly correlated (lowest r = 0.89). The absolute agreement between the devices was poor with 95% CI ranging from −0.07 to 0.95 (See Table 3) as the K-force grip consistently underestimated the Jamar results. In particular, the K-force underestimated grip strength to a larger extent at higher levels of grip strength (See Figure 2). Visual representation of differences in grip strength obtained through the K-force and Jamar dynamometers are provided in the Bland-Altman plots below (See Figure 2).

Table 3.

Intra-class correlation Coefficients with 95% CI between K-force and Jamar dynamometers for dominant and non-dominant hand.

	Trial 1	Trial 2	Trial 3
Dominant	0.7 (−0.07–0.92)	0.7 (−0.07–0.92)	0.76 (−0.06–0.93)
Non-dominant	0.79 (−0.05–0.95)	0.76 (−0.06–0.94)	0.75 (−0.06–0.93)

Figure 2.

Bland-Altman plots for the dominant and non-dominant hand to visualise differences in average measurements between the K-force and the Jamar hand dynamometers. The black horizontal line represents the average difference between measurements (kg) and the red dashed lines are the lower and upper confidence intervals of the average difference. The blue dotted lines represent the MDC₉₅ of the Jamar dynamometer. As evident from the graph, the average difference was beyond the MDC₉₅ levels.

Repeated measures

Grip strength was not significantly different across the three trials for the digital dynamometer on the dominant side (F_(1.3,33) = 0.29, p = 0.7, generalized eta squared = 0.0003). Equally, there was no difference across the three trials for the K-force (F_(2,52) = 2.4, p = 0.1, generalized eta squared = 0.002) and Jamar dynamometer (F_(2,52) = 2.6, p = 0.08, generalized eta squared = 0.002) on the non-dominant hand. The only condition which showed a statistically significant difference across the three trials was measured using the Jamar device on the dominant side (F_(2,52) = 4.8, p = 0.01, generalized eta squared = 0.003), with the second trial presenting a higher grip strength compared to the third trial (t₍₂₇₎ = 2.8, p = 0.03).

Discussion

The results of this study supported our hypothesis that both the Jamar and K-force dynamometers were reliable. The K-force was not valid compared to the Jamar dynamometer, because it underestimated grip strength by a level that surpassed the MDC of 5.75 kg. Clinicians should be aware of this consistent underestimation compared to the Jamar and interpret the grip measurements accordingly. Finally, there were no differences over the course of three trials, suggesting that a learning effect or fatigue was not a factor for either the K-force or Jamar dynamometer.

Our findings are consistent with previous evidence showing high levels of reliability (ICC) for both the Jamar ¹³ and the K-force^15,16 hand dynamometers. The ICC and 95% confidence interval values for both devices were beyond 0.9 suggesting excellent intra-test reliability. The SEM of the measurement provides an estimate of the error associated with the measurement and it ranged from 1.7 to 2.0 kg for both devices. In addition, the MDC₉₅ for both devices ranged between 4.7 and 5.7 kg, suggesting that improvements of grip strength beyond these thresholds are unlikely to be due to measurement error (i.e. 5% chance of any change in grip strength due to measurement error). These results are similar to what has been previously reported in the literature, with MDC₉₅ for grip strength around 6 kg. ²³

In terms of validity, the K-force measurements presented a high correlation (r ≥ 0.89) with Jamar readings. However, the K-force underestimated grip strength, especially at higher levels of grip force. Absolute agreement between the two devices was poor with ICCs 95% CI crossing zero (see Table 3). The K-force underestimated grip strength on average between 7 and 8.5 kg compared to the Jamar. Previous research has reported similar findings with the K-force underestimating grip force between 4.5 and 8 kg.^15,16 This suggested that absolute levels of grip strength measured with the K-force cannot be directly compared to normative values or utilised to indirectly calculate the lifting capacity of patients. ²⁴ These findings may be due to the smaller handle size of the K-force compared to the Jamar hand dynamometer (second handle position). ¹⁴ Integrating the K-force dynamometer into clinical practice should therefore be done with caution considering that its measurement estimation is well beyond 5.7 kg.

Over the course of three trials, grip strength remained stable. The only exception to this occurred when the dominant hand was tested with the Jamar, during which the third trial was lower compared to the second one (1.4 kg lower). Overall, these findings suggest that one maximum measurement should suffice when the aim is to assess maximum grip strength in healthy populations. ¹³ More than one measurement may be necessary in symptomatic populations. ²⁵

Clinically, the use of a digital grip meter such as the K-force, as an exercise tool is of great interest. The ability to train participants at a certain level of grip strength may increase adherence to resistance training guidelines and improve strength outcomes. ²⁶ The connection of devices such as K-force to mobile phones would also allow opportunities to gamify rehabilitation, improving patients’ compliance. In addition, cloud synchronisation of exercise sessions may allow clinicians to track progress remotely. Despite the lack of validity compared to the Jamar dynamometer, the K-force could be used to assess within-patient improvements over time.

Limitations

The present study is not without its limitations. It is not possible to comment on the inter-rater reliability of these tools as only one assessor collected data from participants. Healthy participants were included in this study as grip strength is commonly assessed in healthy people for workplace screening, collection of baseline measures and as part of general strength screening; however, this may not be translated to a pathological population.

Conclusion

The K-force and the Jamar hand dynamometer demonstrate excellent reliability. The K-force underestimated grip strength compared to the Jamar. As there were no differences associated with repeated trials of grip strength, one measurement is acceptable in healthy people to improve testing efficiency. Exercise prescription based on measurements taken through the K-force (e.g. exercise putty prescription) is likely biased towards under prescription of resistance level. Similarly, grip strength measurements obtained through the K-force cannot be compared to normative data or utilised to predict mortality/frailty risk due to a consistent and clinically relevant underestimation of strength. Modifications to the size and shape of the device or software modifications are required for the device to provide readings which are similar to the Jamar dynamometer.

ORCID iD

Nico Magni https://orcid.org/0000-0001-8140-3001