Outcome measures
Perceived ability to perform daily hand activities was rated with the ABILHAND Questionnaire (stroke version) [
4]. The ABILHAND consists of 23 common bimanual activities (see Table
2) that are rated as impossible (0 point), difficult (1 point) or easy (2 points). Items not attempted within the last three months are set as not applicable. The items are ordered hierarchically, from the most difficult items to the easiest, and they are also rated according to the level of bimanual involvement: A = breakable into unimanual sequences; B = requires stabilization with the more affected upper extremity; and C = requires fine bimanual dexterity [
4]. The ABILHAND is Rasch analyzed [
4], which means that ordinal data can be converted into an unidimensional interval scale, and presented in logits (i.e., log odds units) that ranges from plus to minus around zero as the center of the scale [
22]. The higher the logit value, the better the self-perceived ability to use the upper extremities in daily hand activities. In this study, the Swedish version of the ABILHAND was used [
23], which has been shown to have acceptable test-retest reliability for persons with mild to moderate impairments of the upper extremities after stroke [
8]. Due to cultural differences the item
‘Peeling potatoes with a knife’ was changed to
‘Peeling potatoes’ (as a potato-peeler is commonly used in Sweden) and in the item
‘Tearing open a pack of chips’ the following words were added ‘
or a candy-bag’ (because older persons in Sweden more rarely eat chips) [
8]. After the participants had responded to the ABILHAND the logits were obtained by entering the raw scores into an online data analysis module (
http://www.rehab-scales.org/) established for chronic stroke patients [
4].
Pain (present or not) was recorded by asking the participants if they perceived daily or almost daily pain in their more affected upper extremity.
Spasticity was assessed by the response to resistance of passive movement according to the Modified Ashworth Scale (MAS) [
24]. The participants’ spasticity was assessed in an upright sitting position and was classified as present if the elbow, wrist or fingers had a score on the MAS larger or equal to 1 point. The MAS has been shown to have high intra-rater reliability of the upper extremity for persons with stroke [
25].
Grip strength was measured with the digital dynamometer Grippit (Catell AB, Hägersten, Sweden,
http://www.catell.se/). The Grippit is a portable device that is wirelessly connected to a computer. Grip strength was measured three times with the participants seated with the forearm supported on a table on a foam cushion (the shoulder in 30° flexion, the elbow in 90° flexion and the wrist in 0° to 15° dorsiflexion) according to a standardized test protocol [
26]. Each contraction lasted 3 s with a 60 s rest interval between each repetition. The highest value in Newton (N) of the three contractions was recorded as the maximal grip strength. Measurements of grip strength with the Grippit dynamometer have been shown to have an acceptable test-retest reliability for persons with mild to moderate impairments of the upper extremity after stroke [
26], and grip strength has also been found to be a representative measure of the entire upper extremity muscle strength after stroke [
27].
Active touch (somatosensation) of the hand was assessed with the The Shape/Texture Identification test (STI-test) [
28] (Össur Nordic AB, Uppsala, Sweden,
http://www.ossur.se/). Active touch means that identification of different shapes and textures is done by active hand movements. Compared to passive touch, active touch has the advantage of reflecting how somatosensation is integrated during hand movements. The STI-test includes three shapes (cube, cylinder or hexagon) and three textures (one, two or three raised metal dots placed in a row) in three difficulty levels (decreasing sizes). According to the standardized test instructions [
28], the participants were seated behind a screen and identified the shapes (presented randomly size for size) by the index finger. Thereafter the textures were presented and identified in the same way. The score of the STI-test ranges from 0 to 6 points per hand and a higher score indicates better somatosensation [
28,
29]. The STI-test has been shown to have high test-retest reliability for persons with mild to moderate impairments of the upper extremity after stroke [
29].
Dexterity was assessed by the modified Sollerman Hand Function test (mSHFT) [
30] (Catell AB, Hägersten, Sweden,
http://www.catell.se/). The mSHFT assesses manual dexterity by common pinch and volar grips. It is a short version of the 20-item original Sollerman Hand Function Test [
31,
32] and consists of the three items most strongly correlated with the total score [
30]. The items in the mSHFT are: number 4) picking up 4 coins of different sizes from a purse; number 8) putting 4 nuts in decreasing size on bolts; and number 10) buttoning 4 buttons in decreasing sizes. These items are performed as unimanual tasks and the ability to grasp the object correctly, the time to complete the item and the quality of the movement are both assessed on a 5-point scale (0 to 4 points). The total sum score ranges between 0 and 12 points for each hand (where 12 approximates normal dexterity) [
30]. The mSHFT has been found to be valid and reliable for persons with mild to moderate impairments of the upper extremity after stroke [
33].
Perceived participation, i.e., a person’s engagement in meaningful life situations, was rated by the participation domain of the Stroke Impact Scale 3.0 (SIS; Swedish version), that can be used as a separate scale [
34,
35]. SIS Participation is interview-based and includes eight items: impact of stroke on work, social activities, quiet recreations, active recreations, role as a family member, religious activities, life control and ability to help others. The items are scored on a 5-point scale from 1 (limited all of the time) to 5 (limited none of the time). The mean for the items is calculated as a composite score and converted into a percentage value (from 0 to 100) [
35], and a higher percentage value indicates higher perceived participation. SIS has been shown to be reliable and valid in persons with stroke [
34,
36].
Life Satisfaction was rated by the Life Satisfaction Questionnaire (LiSat-11) [
37]. LiSat-11 is interview-based and includes one item that assesses the level of global satisfaction with life as a whole and 10 items that assess the level of domain-specific satisfaction. In the present study, only the item of global satisfaction with life as a whole was used. The responses were rated on a six-graded scale: 6 = very satisfied; 5 = satisfied; 4 = rather satisfied; 3 = rather dissatisfied; 2 = dissatisfied; and 1 = very dissatisfied. In this study, the score was dichotomized into two categories; dissatisfied (score 1 to 4) and satisfied (5 and 6) according to Fugl-Meyer et al. [
37].
Procedures
Prior to the assessments the participants were asked about their age, handedness, social situation (if they lived alone or together with another person) and their vocational situation (not working or in work at least 20 h per week).
The assessments were performed in the following order: 1) perceived pain 2) ability to perform daily hand activities (ABILHAND) [
4]; 3) spasticity (MAS) [
24]; 4) perceived participation (SIS-Participation) [
34]; 5) dexterity (mSHFT) [
30]; 6) active touch (STI-test) [
28]; 7) life satisfaction (LiSat-global satisfaction) [
37]; and 8) grip strength (Grippit) [
26]. Each test took about 10 min to complete and a short rest (approximately 5 min) was allowed between the tests. All assessments were performed on one occasion according to the standardized test procedures of each test in a quiet and separate room of the hospital by an experienced physiotherapist (first author). Data on time of stroke onset, type of stroke (ischemic or hemorrhagic) and side of paresis were verified from the medical records.
Statistics
Descriptive statistics, such as frequencies, means and standard deviations (SD) and medians and minimum and maximum (min-max) were calculated for demographic and clinical characteristics of the participants, perceived ability to perform daily hand activities (logits) and potentially associated factors. The distribution of the participants’ ratings (easy, difficult, impossible or not applicable) of the 23 items in the ABILHAND was presented in percent.
Before the linear regression analyses were conducted correlations between the ordinal and continuous potentially associated factors were calculated using the Spearman correlation (rho) to investigate the strength of their associations.
Perceived ability to perform daily hand activities (continuous: logits) and the potentially associated factors were analyzed by linear regression models. The potentially associated factors were: age (continuous); gender (categorical: female vs male); social situation (categorical: living together with another vs living alone); vocational situation (categorical: working vs not working); affected hand (categorical; dominant vs non-dominant); perceived pain in the more affected upper extremity (categorical: present vs not present); spasticity in the more affected upper extremity (categorical: present vs not present); grip strength in the more affected hand (continuous: newton); active touch in the more affected hand (ordinal scale; 0 to 6 points); dexterity in the more affected hand (ordinal scale; 0 to 12 points), perceived participation (ordinal scale; 0 to 100 %) and life satisfaction (categorical: satisfied vs dissatisfied).
The multivariate regression building was made with a generous inclusion criterion (p ≤0.20) so that no potential variable was excluded in the early stages. First, the associations with perceived ability to perform daily hand activities were evaluated for one variable at the time. Secondly, the variable with the lowest p-value (if ≤0.20) was kept and thereafter the other variables were tentatively added, one at a time. Thirdly, the two variables with the lowest p-values (if both ≤0.20) were included and the remaining factors were tentatively added, one at a time. This procedure was continued as long as the p-values for all included factors were ≤0.20. This selection strategy, to evaluate the model at each step, was chosen as it increases the understanding of the different variables significance.
The explained variances (adjusted R2) after successive addition of the factors are given in the final multivariate model. To ensure the linearity, scatterplots were visually inspected for the bivariate associations. In addition, model assumptions were checked by means of residual analysis.
Data were analyzed with the IBM SPSS Statistics version 22 (IBM Corporation, Armonk, New York, United States).