ARTICLE IN PRESS

AVERAGE VERSUS MAXIMUM GRIP STRENGTH: WHICH IS

MORE CONSISTENT?
S. G. HAIDAR, D. KUMAR, R. S. BASSI and S. C. DESHMUKH
From the Orthopaedic Department, City Hospital, Dudley Road, Birmingham, UK

The average of three consecutive measurements is the most frequently used method for grip
strength assessment. The purpose of this study was to compare the consistency of the maximum
value with that of the average value of three consecutive measurements of grip strength. One
hundred healthy volunteers participated in this study. Three measurements of grip strength were
taken on two occasions separated by 2 weeks. For each hand, two average values and two maximum
values were obtained. Ninety-five per cent limits of agreement for the average method were – 8.3
( 23%) to +7.2 (+20%) kg and for the maximum method were – 8.8 ( 23%) to + 8
(+21%) kg. Both methods of grip strength assessment were found to be highly consistent with no
statistically significant difference.
Journal of Hand Surgery (British and European Volume, 2004) 29B: 1: 82–84
Keywords: grip strength, average, maximum, consistency

INTRODUCTION Instrument

Hand grip strength assessment is simple and reliable
(Fong and Ng, 2001) and used commonly by hand
surgeons and physiotherapists. It provides a quantita-
tive evaluation and is therefore incorporated in many of
the outcome assessment tools that are used to evaluate
hand and upper arm function.
The American Society of Hand Therapists recom-
mends that the average of three consecutive measure-
ments of grip strength is used (Bohannon, 1991). This
contrasts with The American Society for Surgery of the
Hand recommendation to record all three measurements
(Smith and Benge, 1985). In a review of 122 articles,
Bohannon (1991) found that the most frequently used
method to determine grip strength was the average of
multiple measurements and the second most common
was the maximum of multiple measurements. In Smith
and Benges’ survey (1985), 44% of respondents used the
average and 36% used the maximum. An extensive
review of the literature failed to reveal any strong
evidence to support the use of the average of three
measurements over their maximum.
A Jamar dynamometer was used in this study. This is
widely used in orthopaedic and physiotherapy practice
(Bohannon, 1991; Smith and Benge, 1985) and is
considered to be the most reliable instrument for
measuring grip strength (Mathiowetz et al., 1984).
Participants were allowed to choose the most comfor-
table handle position for their hand size (Smith and
Benge, 1985) and the same handle position and the same
dynamometer were used on both testing occasions
(Flood-Joy and Mathiowetz, 1987).
Testing position
The testing position (Fig 1) was as recommended by The
American Society of Hand Therapists (Bohannon,
1991). The participant was seated with the shoulder in
adduction, the elbow in 901 of flexion, the forearm in
midprone position and the wrist in 301 of extension and
neutral radioulnar deviation. In this position the elbow
is most stable and the wrist extension allows greater
mechanical advantage for flexor tendons resulting in
maximum grip strength (Fong and Ng, 2001).

PATIENTS AND METHODS
Subjects
One hundred hospital worker volunteers (50 women and
50 men) participated in this study. The mean age was 34
(range, 21–58) years for women and 37 (range, 23–63)
years for men. Three women and five men were left-
handed. All participants were healthy with no upper
extremity disability. The age, gender and hand dom-
inance of each were recorded.
Procedure
Testing always started with the dominant hand.
Participants were given instructions to hold the handle
and squeeze as hard as they could (Mathiowetz et al.,
1984) for up to 6 seconds and then relax (Kamimura and
Ikuta, 2001). Three consecutive measurements of grip
strength were obtained for each dominant and non-
dominant hand with a 1 minute interval between
measurements. This interval produces less fatigue (7%)
than 15 and 30 seconds intervals (10% and 12%,
respectively) (Trossman and Li, 1989). The average

82
 ARTICLE IN PRESS

GRIP STRENGTH ASSESSMENT 83

Table 1—The means of the averages and the maximums of both men and
women in kilograms
Dominant hand Non-dominant hand
Men Women Men Women
Mean SD Mean SD Mean SD Mean SD
Average
Right hand 46 9.9 29 7.8 42 9.8 26 7.5
Left hand 42 10.7 26 6.5 41 12.3 27 6.7
Maximum
Right hand 49 9.9 31 8.3 44 11.1 28 7.7
Left hand 45 12.4 28 6.9 46 11.4 30 7.4

15
Fig 1 The testing position.
10

Difference of averages
value of the three measurements and their maximum 5
value were recorded. The test was repeated after 2 weeks
by the same examiner at the same time of day to avoid 0
diurnal variation (Wright, 1959). 0 10 20 30 40 50 60 70 80
-5

Statistical analysis -10

For each hand, the average and the maximum -15

values were obtained at each testing occasion. Ninety- Mean of averages
five per cent limits of agreement (Bland and Altman, Fig 2 Scatter of the means of averages against their differences.
1986) were calculated for the average values and also for
the maximum values. The 95% limits of agreement for
%
two measurements are defined as: d72SD; in which d% is 20
the mean of the difference between the two measure- 15
Differrence of maximums

ments and SD is the standard deviation of this

difference. 10

It is important to plot the differences of the two 5

measurements against their means to determine if there 0
is a systematic variation between the measurements (i.e. 0 10 20 30 40 50 60 70 80
the random variation differs according to the mean -5

value). If the random variation increases with the mean -10

value, analysis of the differences after a logarithmic -15
transformation is more appropriate (Altman, 1991). Mean of maximums

Fig 3 Scatter of the means of maximums against their differences.

RESULTS
The mean results for women and men are summarized in
Table 1. Thirty-two participants used the third handle
position (22 men and 10 women), 67 participants the
second position (28 men and 39 women) and one woman
the first position.
In all tests, the maximum grip strength was achieved
at the first measurement in 76% of participants, the
second measurement in 16% and the third measurement
in 8%. The dominant hand had a higher maximum than
the non-dominant hand in 64% of the participants,
whereas both had equal strength in 12%.
The 95% limits of agreement for the average method
were –8.3 ( 23%) to +7.2 (+20%) kg. For the
maximum method they were –8.8 ( 23%) to +8
(+21%) kg. Figs 2 and 3 demonstrate that the random
variation of the differences (of averages and maximums,
respectively) does not differ according to the mean
value. In addition, both figures show similar scatter.
DISCUSSION
Although grip strength assessment is an easy and
important clinical test, there is no consistency in practice
with regard to the method used for its evaluation. This
study investigated the consistency of the maximum and
average of three consecutive measurements of grip
strength. Only two studies have previously compared
 ARTICLE IN PRESS
84
these two methods. Mathiowetz et al. (1984) studied
27 women and reported higher correlation coefficients
(Pearson product-moment) for the average (0.929
for left hand and 0.883 for right hand) than maximum
(0.915 for left hand and 0.822 for right hand). On the
contrary, in Hamilton et al.’s (1994) study which
included 33 subjects (17 women and 16 men), the
difference between the correlation coefficients (inter-
class) of the average (0.954 for left hand and 0.964
for right hand) and that of the maximum (0.933 for
left hand and 0.954 for right hand) was insignificant.
Both studies included small number of participants
and the first one only women. In addition, the use
of the correlation coefficient in both papers to assess
agreement was inappropriate, as it assesses the degree
of association between two variables, not the agreement
between them (Bland and Altman, 1986). Evaluation
of the agreement between two variables is best provided
by calculating the 95% limits of agreement (Altman,
1991). As differences are usually normally distributed
(Bland and Altman, 1986), it is expected that 95%
of differences of the assessed two variables lie be-
tween the limits of agreements. This method is
easy to perform and to interpret and it is the
method of choice for assessing the repeatability of a
single method, comparing the repeatability of two
methods and comparing measurements made by
two observers.
The 95% limits of agreement of both the average
method and the maximum method of recording grip
strength in this study are similar. As the Jamar
dynamometer is accurate to within 73% (1–1.5 kg)
(Bohannon, 1991), it can be concluded that both
methods have high repeatability and the small difference
between them is negligible. We favour the maximum as
it represents an achieved value and does not require
calculations in a busy clinical sitting. However, one
measurement is not enough as approximately 25% of
THE JOURNAL OF HAND SURGERY VOL. 29B No. 1 FEBRUARY 2004
participants recorded their maximum grip strength at
the second or third measurement.
Acknowledgements
The authors thank Dr Tim Marshall for his advice regarding statistical tests.
References
Altman DG (1991). Practical statistics for medical research. Chapman & Hall,
London.
Bland JM, Altman DG (1986). Statistical methods for assessing agreement
between two methods of clinical measurement. The Lancet, 307–311.
Bohannon RW (1991). Hand grip dynamometers: issues relevant to application.
Journal of Human Muscle Performance, 1: 16–36.
Flood-Joy M, Mathiowetz V (1987). Grip-strength measurement: a comparison
of three Jamar dynamometers. The Occupational Therapy Journal of
Research, 7: 235–243.
Fong PW, Ng GY (2001). Effect of wrist position on the repeatability and
strength of power grip. The American Journal of Occupational Therapy,
55: 212–216.
Hamilton A, Balnave R, Adams R (1994). Grip strength testing reliability.
Journal of Hand Therapy, 7(3): 163–170.
Kamimura T, Ikuta Y (2001). Evaluation of grip strength with a sustained
maximal isometric contraction for 6 and 10 seconds. Journal of
Rehabilitation Medicine, 33: 225–229.
Mathiowetz V, Weber K, Volland G, Kashman N (1984). Reliability and validity
of hand strength evaluations. The Journal of Hand Surgery, 9A: 222–
226.
Smith RO, Benge MW (1985). Pinch and grasp strength: standardization of
terminology and protocol. American Journal of Occupation Therapy,
19: 531–535.
Trossman PB, Li P-W (1989). The effect of the duration of intertrial rest periods
on isometric grip strength performance in young adults. The Occupa-
tional Therapy Journal of Research, 9: 362–378.
Wright V (1959). Some observations on diurnal variation of grip. Clinical
Science, 18: 17–23.
Received: 7 March 2003
Accepted after revision: 19 September 2003
Dr S.G. Haidar, Orthopaedic Department, City Hospital, Dudley Road, Birmingham, B18
7QH, UK. Tel.: +44-121-5075390; fax: +44-121-5075483; E-mail: thehaidars@aol.com
r 2003 The British Society for Surgery of the Hand. Published by Elsevier Ltd. All rights
reserved.
doi:10.1016/j.jhsb.2003.09.012 available online at http://www.sciencedirect.com