Psychosocial and personality factors and physical measures in lateral epicondylalgia reveal two groups of “tennis elbow” patients, requiring different management

2.1 Patients and controls. Ethics and patient information

The study was undertaken by the principal author in a Norwegian primary care setting. Fifty-four subjects (16 males, 38 females, mean age 48.7 ± 7.5 years); with LE were recruited via newspapers and advertisements in the area around Brumunddal, Norway. Physiotherapists recruited 43 pain-free control subjects (18 males, 25 females, mean age 48.8 ± 8.4 years) from the region. Sociodemographic and clinical characteristics are presented in Table 1. The procedures followed were in accordance with the ethical standards of the responsible local ethic committee on human experimentation and with the Helsinki Declaration of 1975, as revised in 1983.

Subjects with LE were examined by the principal author and required to test positive in at least two of the following: (1) pain on palpation of the epicondyle; (2) pain on resisted extension of the wrist; (3) pain on passive stretching of the wrist extensor muscles; and (4) pain on resisted finger extension. Subjects were classified according to Nirschl’s criteria [2]. Exclusion criteria were a previous dislocated elbow, referred cervical pain to the forearm, neurological signs, an acute traumatic onset, cortisone injection into the CEO within the last two months, rheumatoid arthritis, and carpal tunnel syndrome. Seven pain subjects were excluded, three had only one positive provocative test, one had rheumatoid arthritis, one consumed analgesia before testing, one did not attend, and one had neurological signs.

In addition to the exclusion criteria, the control subjects were required to have negative provocative tests. Two control subjects were excluded due to positive provocative testing. All subjects spoke and understood the Norwegian language. All subjects gave written informed consent.

Subjects with LE and the control subjects had both arms tested. Participants were subjected to clinical examination, pressure pain threshold (PPT) testing, and testing of maximal load pain tolerance testing (MLT) and low load pain tolerance testing (LLT) of forearm extensors. They completed the ÖMSPQ(only pain subjects) and the Swedish Universities Scale of Personality (SSP; [21], which is a valid and reliable measure of personality) [20].

2.2 Psychosocial and personality factors

2.3 Swedish university Scales of Personality (SSP)

Personality traits were measured by the Swedish universities Scales of Personality (SSP). The SSP is a self-report personality inventory which was developed to study different personality traits and the assumed biological basis of some psychiatric disorders [21]. The SSP contains 91 items divided into 13 subscales. The subscales measure personality traits that in many ways correspond to clinical symptoms reported in dementia: anxiety proneness factors (Somatic Trait Anxiety, Psychic Trait Anxiety, Stress Susceptibility, and Low Assertiveness), Extraversion factors (Impulsivity, Adventure Seeking, Detachment, Embitterment, and Social Desirability) and Aggression-Hostility factors (Verbal Trait Aggression, Physical Trait Aggression, Trait Irritability, and Mistrust).

Each item is rated on a four-point response scale from “does not apply at all” to “apply completely”. Analysis of the correlations between the SSP scales have yielded a three-factor solution whereas factor 1 encompass scales related to anxiety proneness, factor 2 scales measuring Aggression-Hostility and factor 3 reflects the factor Extraversion.

The SSP has standardised on a representative, randomised sample from the general Swedish population [21]. Scale scores are summed and transformed to T-scores (mean = 50, SD = 10) and standardised for men and women, separately. In this study an authorised translation into Norwegian was used.

Fig. 1

Pressure pain threshold testing sites. (1) Radial styloid (used as landmark only); (2) ECRB (PPT_ECRB; 48% of the length from the CEO to the radial styloid); (3) EDC (PPT_EDC; 17% of the length from the CEO to the radial styloid); and (4) CEO (PPT_CEO) [23]. ECRB: extensor digitorum communis; CEO: common extensor origin; EDC: extensor digitorum communis; PPT: pressure pain threshold.

2.4 Clinical examination

Data collection included: dominant hand, painful or most painful arm, symptom duration, intermittent or constant pain, average pain in the last week measured on a visual analogue scale, type of work, sporting activity engaging the forearm, and presence of night pain.

2.5 Pressure pain thresholds (PPT)

Pressure pain threshold was defined as the minimum pressure needed to evoke a painful sensation [22]. PPTs were recorded at the CEO (PPT_CEO); ECRB (PPT_ECRB; 48% of the length from the CEO to the radial styloid), and EDC (PPT_EDC; 17% of the length from the CEO to the radial styloid) [23] (see Fig. 1) using an algometer (Somedic AB, Sweden) with a stimulation area of 1 cm². Subjects were familiarised with PPT measurements over upper trapezius prior to measurements being taken on the non/less painful arm or non-dominant arm first, followed by the most painful or dominant arm. Pressure increased at 30 kPa/s [5] until the patient detected a painful sensation. They then pressed a button on the algometer to record a pressure reading. This was repeated three times with 30 s intervals.

2.6 Isometric maximal load pain tolerance test (MLT)

The subject was sitting, elbow flexed between 100° and 110°, forearm pronated on the bench and fingers flexed over the edge. Isometric maximal load tolerance testing of the forearm extensors was performed with a handheld dynamometer (Nicholas Manual Muscle Tester, Lafayette Instrument Co., Lafayette, USA) over the distal phalanges. Subjects started with three repetitions of extending the hand without resistance for familiarisation, then three isometric maximal voluntary contractions with 30 s intervals. Both arms were tested. The dynamometer has been shown to have good intra-and inter-tester reliability and is valid compared to isokinetic dynamometry [24]. This test was provocative of elbow area pain in some cases. It is therefore impossible to determine whether in these cases it was a test of maximal isometric strength or a test of pain tolerance to high loading.

Fig. 2

Isometric low load paintolerancetest position.

2.7 Isometric low load pain tolerance test (LLT)

Subjects with LE commonly describe sustained holding of an object as an aggravating factor in LE. Therefore we developed a low load pain tolerance test, for the wrist extensor muscles similar to Biering-Sorensen and McGill’s methods [25, 26]. A 0.5 kg weight, necessitating a low percentage of maximal voluntary contraction (MVC), was held in the aforementioned position (see Fig. 2) with the forearm supported but the hand and weight unsupported. Subjects held the weight until they could not stand the pain anymore or until the wrist flexed. The duration of the patient’s ability to sustain this position was recorded. If subjects were able to hold the weight for 45 min the trial was stopped. The test started with the nonpainful/less painful or non-dominant hand. Both arms were tested.

2.8 Statistical methods

PPT and MLT were measured three times, and a mean used for statistical analysis. In the comparisons the non-dominant arm of the control group was compared with the injured arm of patients and vice versa to minimise a comparison with an inflated increased risk for Type-I errors, which might be the case if the injured arm was be compared with the dominant arm of the control group.

Variables were summarised using standard descriptive statistics (mean, standard deviation, and frequency). Distributions were checked for skewness and outliers. All distributions were approximately normal with no outliers.

Group differences in discrete variables, for example gender and handedness, were analysed with the chi-squared (χ²) method. If the expected cell frequency was less than five Fisher’s exact test was applied.

Differences between patients and controls in pain thresholds and MLT of the injured/non-dominant arm were analysed with analysis of variance (ANOVA) for repeated measurements (time). However, the interaction effect (group x time) was not taken into consideration. Since there was a difference in gender distribution, although not significant (cf. below), gender was entered as a covariate.

All control subjects were able to hold the weight for 45 min, thus differences between groups in LLT – less than 45 min vs. 45 min – was analysed with the χ² method. Differences between the patient groups in scores of the ÖMPSQ scale and the personality variables were analysed with one-way ANOVA.

Correlations between pain thresholds, MLT, and ÖMPSQ scores, and between the ÖMPSQ scores and the personality variables, were expressed as Pearson’s parametric correlation coefficients. To investigate the unique correlation between the ÖMPSQ score and the personality variables a regression analysis (stepwise forward) was also performed.

The level of significance was set at α ≤ 0.05 (two-tailed tests). Post hoc testings were made with the Tukey HSD test.

3.1 Socio demographic data

There were 58% (n = 25) females in the control group and 70% (n = 38) in the patient group. The difference in gender distribution was not statistically significant (χ² = 1.57, p = 0.210). Of the 54 patients 46 (85%) were working, two were at home, four retired, and two on sick-leave. Thirty-eight control subjects (88%) were working, three were at home, one on sick-leave and one was a student. The patients were classified into Nirschl’s sub-groups: group I consisted of 12 patients; group II of 15 patients; group III of 15 patients; and group IV of 12 patients. Mean symptom duration was 34 months (group I – 14.4 months; group II – 15.7; group III – 51 months; and group IV – 56.4 months). There were no significant differences between the patient and the control group in any of the sociodemographic variables.

3.2 Pressure pain thresholds

Patients had significantly lower pain thresholds at all loci than the controls, especially the threshold of the common extensor origin (Table 2). The pain thresholds were negatively related to the Nirschl grouping, including the control group, implying that the higher the classification was the lower the pain threshold. According to the analyses corrected for multiple testing, the control group and Nirschl’s pain group I did not differ significantly except for the PPTCEO with higher thresholds in the control group. Furthermore, Nirschl’s pain group II did not significantly differ from the control group, Nirschl’s pain group I, or Nirschl’s pain groups III and IV.

The pain thresholds were significantly higher in the non-injured or the dominant arm than in the injured or non-dominant arm (for PPT_CEOF(1, 90) = 27.75, p <0.001; for PPT_ECRBF(1, 90) = 12.15, p<0.001; and for PPT_EDCF(1, 90) = 9.13, p = 0.003). The differences between the arms and the groups, i.e. the interaction effect (group × arm), was, however, not significant for PPT_ECRB and PPT_EDC (cf. Table 2) but for PPT_CEO with a tendency for smaller differences for the control group than for the patient groups. This difference was, however, not significant when corrected for multiple testing.

3.3 Isometric maximal load pain tolerance test (MLT)

The MLT of the groups is presented in Table 3. The control group and Nirschl’s pain group I was significantly stronger than Nirschl’s pain group III (p <0.001). The Nirschl’s pain groups I, II and IV, although weaker, were not statistically different from either the control group or Nirschl’s pain group I.

3.4 Isometric low load pain tolerance test (LLT)

There was no significant difference between the Nirschl’s pain groups in LLT (χ² =3.49; p = 0.322). However, all the subjects of the control group were able to hold the weight for 45 min while only 14 of the patient group (26%) held the weight this long. This difference was statistically significant (Fisher’s exact test< 0.001 (see Fig. 3)).

3.5 Self-reported long-term musculoskeletal pain (ÖMSPQ)

Means and standard deviations for the four Nirschl’s pain groups are shown in Table 4. There was a highly significant difference between groups (p <0.001). The scores progressively increased the higher the Nirschl’s pain group. Nirschl’s pain group IV had the highest mean score and group I the lowest. The outcome from the Tukey test is presented in Table 4.

There were eight ÖMPSQ questions that were significant when comparisons were made between groups: “How often would you say that you have experienced pain episodes, on average, during the past 3 months?”, “I can do light work for an hour.”, “I can do ordinary household chores:”, “I can sleep at night.” (p<0.001); “How would you rate the pain you have had during the past week” (p = 0.006); “How many days of work have you missed (sick leave) because of pain during the past 12 months?” (p = 0.03); “In your view, how large is the risk that your current pain may become persistent (may not go away)?” and “I should not do my normal work with my present pain.” (p = 0.04). ÖMPSQscore significantly correlated with PPT and MLT in the painful and non-/less painful arms (Table 5).

Fig. 3

Forearm extensor muscle isometric low load pain tolerance test (minutes) (Con – control subjects; Pain gr 1 – Nirschl’s pain group one, etc.).

3.6 Relationships between pain thresholds, MLT, and ÖMPSQ scores

In Table 5 the correlations between the tests and mean ÖMPSQ score are shown. All the pain thresholds were highly correlated approaching the level that is expected when test-retest stability is studied.

The pain thresholds were also significantly correlated to MLT, i.e. the higher threshold the greater the subject’s MLT, although to a lower extent.

Both pain thresholds and MLT were also significantly but negatively correlated to mean ÖMPSQ score, i.e. the higher mean ÖMPSQ score the lower threshold and poorer MLT the patient demonstrates.

3.7 Personality (SSP)

The control group and the pain groups were also compared regarding personality. The only significant difference was found for somatic anxiety [F(4, 90) = 3.61, p = 0.009] with the highest score in Nirschl’s pain group IV and the lowest in the control group. In the Tukey analysis, the only significant difference was found for the comparison between the control group (M = 48.5) and Nirschl’s pain group IV (M =59.3). The other groups had mean scores between these scores with Nirschl’s pain group III closer to pain group IV, and Nirschl’s pain group I and II closer to the control group.^[1]

In a correlation analysis the ÖMPSQ score was significantly correlated with three anxiety proneness variables: somatic anxiety (r_xy = 0.43, p = 0.002); psychic anxiety (r_xy= 0.30, p = 0.038); and lack of assertiveness (r_xy = 0.29, p = 0.044), but none of the others.^[2] However, in a regression analysis only somatic anxiety was significantly correlated with the ÖMPSQ score.

4.1 Limitations of the study

The power of the study was too low to find small differences between groups. A higher number of participants may have led to a greater differentiation between the Nirschl’s groups using the physical measures. However, the differences that were investigated with satisfactory power seem to be the differences that are clinically meaningful.