Moberg picking-up test in patients with inflammatory joint diseases: A survey of suitability in comparison with button test and measures of disease activity




To assess and compare the suitability of Moberg pickup test (MPUT) and button test (BT) as indicators for functional impairment in patients with inflammatory joint diseases.


Measurements for 369 patients attending a rheumatology outpatient clinic were collected. In addition to MPUT and BT, measurements collected were grip strength, tender and swollen joint counts, visual analog scales for pain and disease activity, Health Assessment Questionnaire, C-reactive protein levels, and erythrocyte sedimentation rates.


We found a significant relationship between MPUT and BT. Both tests show the same pattern of correlations with the other parameters, although all correlations are higher for MPUT. There is a significant sex and learning effect for the BT, which implies a confounding of hand function and motor abilities. A significantly higher proportion of patients was unable to complete BT.


MPUT and BT measure comparable aspects of hand function. In several theoretical and practical aspects, MPUT seems superior to BT in arthritis. It is necessary to evaluate its value in long-term followup.


Quantitative assessments of functional status, such as grip strength (GST) and walking time, have been widely used for several decades in studies on rheumatoid arthritis (RA) (1). These measures, as well as the button test (BT) (2) and questionnaires regarding activities of daily living (3, 4), are effective in documenting significant morbidity. Declines in functional status have been reported in most patients over the course of 9 years (5–7).

To evaluate patients with RA, quantitative assessment of functional status can provide valuable data on disability (7) and can be used as a long-term functional outcome measure (1). Quantitative assessment of functional status in patients with RA has been approached through simple, rapidly completed physical measures of performance, such as GST, walking time, and the BT (2). Those measures have advantages compared with self-report questionnaires: Some patients find it difficult to complete self-report questionnaires, especially those with low formal education levels who have an increased risk of developing progressive RA (8). Furthermore, clinicians might find it problematic to rely only on self-reported data of functional status as adjunct to clinical decisions (2).

Moreover, GST, walking time, and BT have been found to be highly reproducible measures concerning their interobserver and intraobserver reliability when used according to a standard protocol (2).

To assess hand function in patients with RA, GST is widely used and BT is less often used (7, 9–11). Compared with specific tests with a broad focus on different aspects of function, GST and BT values can be obtained quickly, which is especially important in busy clinic settings and long-term evaluations of RA patients.

Nevertheless, these tests have serious drawbacks. Patients with RA report pain when performing GST, which has to be done 3 times with each hand according to the standard protocol (2, 7). In joint protection instructions in occupational therapy, patients with RA are told to avoid maximum grip force and to respect their pain as an indicator to stop (12). To use GST as a measurement in rheumatology might be questionable under this perspective. Some patients use these arguments or complain about pain and refuse to have GST assessed at all. In addition, GST measures a single dimension of function whose importance for performing everyday tasks may be limited and thus severely impairs the applicability of the test for long-term observations.

BT is less often used in clinical practice than GST. The standard protocol requires a standardized button board (2) that is difficult to obtain. Wear and tear, replacement, or repair of the board change the performance conditions for the patients.

The Moberg pickup test (MPUT) was considered a possible alternative measure for assessment of functional status of the hand in patients with RA. MPUT consists of 12 small objects that have to be picked up while time is taken with a stop watch (Figure 1). A standard protocol has been established and interrater reliability has been found to be good (13). When administered blindfolded, MPUT has been described to assess sensory function grip of the hand (13). When administered with open eyes, MPUT could be used to assess precision grip in patients with RA. MPUT has not been used in RA patients so far. The aim of this study was to assess and compare the suitability of MPUT as an indicator of functional impairment in patients with inflammatory joint diseases. In this analysis, MPUT was compared with BT as an already established test for quantitative assessment of functional status in patients with inflammatory joint disease.

Figure 1.

The 12 small objects of Moberg pickup test: These objects have to be picked up and put into the small container while time is taken with a stopwatch.



We assessed 369 patients with inflammatory joint disease attending a rheumatology outpatient clinic. Inflammatory joint disease was classified as RA (n = 252) and non-RA (n = 17). RA was classified according to the American College of Rheumatology (ACR, formerly American Rheumatism Association) criteria (14). Non-RA included patients with other inflammatory joint diseases, such as psoriatic arthritis, connective tissue disease, or undifferentiated oligoarthritis. Furthermore, patients with early arthritis in each group, RA (n = 38) and non-RA (n = 62), were analyzed separately. The term early is defined as the presence of symptoms of inflammatory joint disease for no longer than 3 months. Furthermore, for the analysis, the patients were divided into groups according to demographic criteria, such as age and sex. Due to missing data, only 366 patients were analyzed in the age groups. The patients represent a heterogeneous population of patients, as can be expected in a rheumatology outpatient clinic.

Inclusion criteria were as follows: 1) inflammatory joint disease; 2) involvement of the upper extremity defined as pain and/or soft tissue swelling in at least 1 joint (finger joints, wrist, elbow, or shoulder) (15); and 3) no history of any neuromotor disease that could possibly affect motor function.

Data collection.

Every patient was assessed once. Data were gathered using a standardized assessment that consisted of MPUT (13), BT (2), GST (16), and a core set of parameters for assessing RA (17): tender and swollen joint count (32 joints) (15), visual analog scales (VAS) for general pain and disease activity, Health Assessment Questionnaire (HAQ) (18) in a validated German version (19), C-reactive protein levels (CRP), and erythrocyte sedimentation rate (ESR).

Not all variables could be obtained for every patient at the time of MPUT due to the survey character of the data collection. MPUT was obtained for 333 patients, BT for 124 patients; 100 patients took both tests.

BT was administered to fewer patients than MPUT because of difficulties to obtain the standardized button board and because of the material weakness of the button board.

Clinical tests.

MPUT was performed according to a standard protocol once with open eyes only, instead of being performed 2 times, both blindfolded and with open eyes (13). For more detailed information on the standard protocol of MPUT, see Appendix A.

Due to time constraints in a busy clinic setting, only the dominant hand was tested, but 2 attempts were performed according to the protocol (13). Failures were recorded as 300 seconds. For more detailed information, see Appendix A.

The BT was administered according to the standard protocol as described in the literature (2). Both hands were tested, the patient always starting with the dominant hand. Times were recorded (in seconds) separately for both hands; for some analyses, times were averaged (according to the standard protocol). Failures were recorded as 300 seconds.

GST was measured with a vigorimeter (16): Patients were sitting, shoulder in neutral position, elbow 90° flexed, thumb upwards and outside of the fist. Patients were tested 3 times on both hands and average values for both hands were recorded. The middle sized rubber bulb (diameter 43 mm) was used for all patients.

Tender and swollen joint counts were performed for 32 joints (15) by the same trained joint assessor (TAS). Furthermore, patients were asked to assess their general level of pain and their disease activity on a 100-mm VAS. In addition, the therapist had to assess the patients' disease activity also on a 100-mm VAS (20). For HAQ, patients were asked to fill in a German HAQ score form (19). CRP levels were measured in mg/dl.

Statistical analyses.

The data were analyzed using version 1.2.3 of the statistical package R (21). Mean, standard deviation, median, interquartile range, and skewness were calculated for both MPUT and BT. Histograms and quantile-quantile plots clearly showed that the values of both tests were strongly non-normally distributed. Therefore, 95% confidence intervals for mean, standard deviation, and skewness were calculated by bootstrapping the data, using the library boot of R (22). The difference in the percentages between the MPUT and the BT were tested using Fisher's exact test (23).

MPUT and BT values were compared by diagnosis, age, and sex of patients (Wilcoxon rank-sum test and the Kruskal-Wallis test). The possibility of interactions between the 3 grouping factors was explored graphically, using conditioning plots; additionally, analysis of variance (ANOVA) models were fitted to the transformed test values.

Scatterplots showed that relationships between MPUT and BT and all other tests were monotonous, but not linear. Therefore, Spearman rank correlation coefficients, together with approximate confidence intervals based on the Fisher z-score transform (23), were calculated. The linear relationships between MPUT and BT and between repetitions of both tests were expressed using simple linear regression.



Descriptive statistics and confidence intervals for MPUT and BT are given in Table 1. The upper left panels of Figure 2 and Figure 3 show the corresponding histograms of the values.

Table 1. Statistical measures for the values of MPUT and BT*
StatisticMPUT value (95% CI)BT value (95% CI)
  • *

    MPUT = Moberg pickup test; BT = button test; 95% CI = 95% confidence interval.

Mean14.5 (14.0–15.1)28.3 (26.4–30.6)
Standard deviation7.1 (5.81–9.54)10.9 (9.02–15.3)
Skewness3.0 (2.3–5.0)1.4 (0.5–5.4)
Median (interquartile range)12.0 (5.0)26.3 (12.4)
Unable to perform, n610
Unable to perform, %1.8 (0.66–3.88)8.1 (3.93–14.33)
Figure 2.

Histogram of population Moberg pickup test (MPUT) scores and boxplots of group MPUT scores by diagnosis, gender, and age (clockwise from upper left to lower left panel). RA = rheumatoid arthritis.

Figure 3.

Histogram of population button test (BT) scores and boxplots of group BT scores by diagnosis, gender, and age (clockwise from upper left to lower left panel). RA = rheumatoid arthritis.

Group comparisons.

The boxplots of MPUT values by diagnosis, age, and sex are shown in Figure 2. As can be seen, the values in RA patients were higher than in non-RA patients. In addition, there was an increase in values with age. The differences between diagnostic and age groups were significant (P < 10−4 and P = 0.02, respectively). Differences between men and women were unremarkable and nonsignificant. Figure 3 shows boxplots of the BT values for the same groups. These exhibit the same significant influence of diagnosis and age (P = 0.01 and P = 0.04, respectively) as MPUT, but additionally, there was a highly significant (P = 0.005) sex effect. No significant interactions between the demographic factors in regard to the MPUT and BT values were found either by visual inspection or in the ANOVA models.

Learning effect.

The scatterplots (data not shown) show the strength of agreement between first and second attempt, which is clearly stronger for the MPUT (R2 = 0.87 for the linear regression line) than for the BT (R2 = 0.38). However, for the MPUT both attempts were made with the dominant hand, while for the BT, the first attempt was made with the dominant hand and the second attempt with the nondominant hand.


Figure 4 shows the Spearman rank correlations between MPUT and BT and the other tests, together with 95% confidence intervals. Correlations with CRP and ESR measurements and the swollen joint count are not significant at a 95% level for both MPUT and BT, whereas the VAS score by the therapist is significant for MPUT, but not for BT. All other tests correlate significantly with both MPUT and BT at a 95% error level. The highest correlation for both MPUT and BT was with the HAQ values. Overall, the correlations with the other tests follow the same pattern for MPUT and BT, with the MPUT correlations always slightly higher than the BT correlations. The exceptions are the GST subtests, which correlate clearly stronger with MPUT than with BT.

Figure 4.

Spearman rank correlations between Moberg pickup test (MPUT; dark bars) and button test (BT; light bars) and all other tests. The thin error bars show approximate 95% confidence intervals for the correlation coefficients. Note that for ease of visual comparison, the absolute values of the correlations are shown. The direction of the relationship is indicated by the signs at the right border of the plot. HAQ = Health Assessment Questionnaire; GST.dom = grip strength dominant hand; GST.sub = nondominant hand; VAS.pat = visual analog scale; VAS.pain = disease activity as assessed by patient; TE = tender joint count; VAS.ther = as assessed by therapist; SW = swollen joint count; CRP = C-reactive protein; ESR = erythrocyte sedimentation rate.

Test comparison.

A significantly higher percentage of patients were unable to finish the BT as compared with the MPUT (P = 0.004). For those patients who completed both tests (n = 100), the Spearman rank correlation coefficient between MPUT and BT values is 0.52, with a 95% confidence interval of 0.35–0.64.


In general, MPUT and BT show the same pattern of correlation with the other parameters, though the correlations are slightly stronger for MPUT than for BT. The most notable exception is GST on both hands, which correlates clearly stronger with the MPUT values. This allows a more consistent interpretation of the MPUT in relation to the other parameters. MPUT provides additional information compared with the BT. This notion is based on the following grounds.

There is a significant linear relationship between MPUT and BT; the measure of determination (R2 = 0.37) implies that this relationship explains approximately 37% of the variability of the values for both tests. We interpret this as the amount of variability due to the common aspects of functional ability that are measured by both tests, whereas the rest is due to different aspects of functional ability and random variation between patients. Under this assumption, we have 1) regressed MPUT on BT and 2) regressed BT on MPUT; the residuals from these regressions can be seen as corrected test values for 1) MPUT and 2) BT, after removal of the shared aspects of functional ability. Figure 5 shows the correlations of these corrected test values with the other tests: the corrected BT values show no significant correlation with any of the other parameters, whereas the corrected MPUT values are still significantly correlated with grip strength for the dominant hand and to a lesser degree also with ESR and CRP. From this we conclude that 1) MPUT measures specific aspects of functional ability that are not described by BT, but that correlate negatively with grip strength, CRP, and ESR; and 2) if BT measures any comparable specific aspects, these are not related to the other parameters.

Figure 5.

Spearman rank correlations between the residuals of 1) regressing Moberg pickup test (MPUT) on button test (BT; dark bars) and 2) regressing BT on MPUT (light bars). The thin error bars show approximate 95% confidence intervals for the correlation coefficients, and the direction of the correlations is indicated by the signs at the right border of the plot. See Figure 4 for definitions.

Both MPUT and BT values exhibit the population effects that one would expect from clinical practice: a tendency for slightly higher values for RA patients, and a slight increase of values with the age of the patients. Additionally, there is a strong sex effect for BT, which is completely absent from MPUT. A possible explanation is that the BT depends not only on functional ability, but also on skill, because there is no connection between sex and degree of impairment, and the skills required for the BT are more traditionally associated with women. This sex effect complicates test standardization for the BT.

Both MPUT and BT show a significant learning effect in that the time for the second attempt of the patient is on average shorter than for the first attempt. The MPUT protocol accounts for this learning effect by requiring the patient to use the dominant hand on both attempts. By recording only the better time, it allows patients to have a trial run first.

In case of the BT, the learning effect is strong enough to overcome the disadvantage of having to use the nondominant hand for the second attempt. This raises the question to what degree the value for the second attempt is influenced not only by learning the necessary motor skills, but also by the ability of the patient to use the nondominant hand for fine motor tasks. This problem is confounded by taking the average time of both attempts, so that both the learning ability and the degree of handedness of the patient influence the final value.

Interestingly, even though the MPUT does not measure performance of the nondominant hand, its values correlate better with grip strength of the nondominant hand than the BT values. From our clinical experience, MPUT refers to a domain of hand function that seems to be relevant for daily life, such as picking up money from a table or picking up paper clips. Thus, MPUT might have more relevance for daily life than BT (buttons have to be opened and closed by using one hand only, which is not done in daily life) and GST (using maximum grip force should be avoided in daily life by patients with RA). Nevertheless, further research is needed to determine to what level MPUT might be representative of hand function in general.

The limitation for all 3 (MPUT, BT, and GST) is that they are administered in a different environment according to a standard protocol that might not allow for the patient to use his or her normal coping strategies. The relevance for daily life can be questioned under this perspective.

Because fine motor tasks may be particularly severely affected by RA's stiffness and finger joint immobility, a test that measures these domains may have great practical value for RA patient assessment. So far, no gold standard exists for measuring hand function in rheumatology and the available tests have limitations in standardization (BT), practicability (Jebsen-Taylor-Hand-Function Test &lsqbr;24&rsqbr;) and practical value (grip strength). To define a gold standard for functional testing, an international consensus on a set of functional outcome measures in rheumatology would be highly useful.

In conclusion, we found in our comparative survey that both MPUT and BT show population effects as expected (higher values for RA and older age). MPUT, compared with BT, exhibits no sex and a smaller learning effect and correlates better with several core set parameters. A significant linear relationship between MPUT and other function-related tests, BT and grip strength, could be demonstrated. Additionally, MPUT seems to have more relevance to everyday life than grip strength and BT. Thus, we consider MPUT a possible alternative to GST and BT for measuring functional ability of patients with inflammatory joint disease. Further research is needed to determine the usefulness of MPUT in monitoring RA patients over long-term periods.


Moberg pickup test (MPUT) was performed according to a standard protocol with open eyes only (13). Patients were sitting in front of a table, 12 small objects were randomly placed 15 cm from the edge of the table, and a small container was put in front of the patient. No specific surface was used, but the objects were placed on the table. The 12 small objects consist of 2 paper clips (length 3.3 and 2.8 cm), 1 nail (length 5 cm), 1 safety pin (length 3.8 cm), 2 nuts, 1 wing nut, 2 coins (diameter 28 cm and 2.2 cm), 1 screw (length 2.3 cm), 1 O-ring (diameter 1.4 cm), and 1 key (length 5.5 cm). The 12 items can be seen in Figure 1.

The tested patient was instructed to put the small objects into the container as quickly as possible beginning with any of these objects and was not allowed to do any trick movements, such as sliding an object to the edge of the table. The test was timed with a stop watch and time recorded in seconds. Failures were recorded with 300 seconds.

Patients were instructed as follows: “When I tell you so, using 1 hand only, starting with the dominant hand, please put these objects into the small container in front of you. You have to take 1 object at a time and you are not allowed to slide these objects to the edge of the table. I will time you while you are doing this.”