The psychological features of patellofemoral pain: a cross-sectional study

2.1 Participants

The PFP participants were a subset of those included in the Foot Orthoses versus Hip eXercises (FOHX) trial [13]. They were recruited across the greater Brisbane region through advertising in physiotherapy clinics, posters in public areas, and social media outlets. In accordance with the FOHX trial’s recruitment process, volunteers were assessed through a two-part screening process conducted by a registered physiotherapist with post-graduate qualifications. Volunteers were first screened by a phone assessment, and then underwent a physical examination. Inclusion criteria for the PFP group were: (1) aged 18–40 years; (2) 6 weeks or more of PFP symptoms that were provoked by at least two of squatting, jogging, running, hopping, jumping or stair ambulation; (3) worst pain over the last week ≥3/10 on a numerical pain rating scale (0=no pain, 10=worst pain imaginable); (4) pain on palpation around the patella; and (5) none of the following: concomitant knee injuries, pathologies or surgery; patellar instability or evidence of knee swelling; hip, pelvis or lower back pain; current use of anti-inflammatory medication; or treatment in the last 12 months [13].

Pain-free control volunteers were recruited through the same method as the PFP group minus the clinical advertisements. They were aged 18–40 years and had been pain-free for the past 3 months. Controls were excluded if they had been unable to complete normal daily duties due to body pain in the past 3 months, or had any chronic conditions affecting their physical function.

2.2 Psychological measures

Patellofemoral pain and pain-free groups completed self-rated questionnaires pertaining to anxiety, depression and pain catastrophizing, with only the PFP group completing a fear of movement or re-injury questionnaire.

The Hospital Anxiety and Depression Scale (HADS) is a 14-item questionnaire (two 7-item subscales) developed to detect states of anxiety and depressive symptoms in clinical settings [14]. Participants are instructed to answer each item in response to how they have felt over the last 7 days. Each item is scored on a 4-point scale, and summed to give a total score from 0 to 21. Although the developers of the scale recommended categorising a score of 8–10 as a potential case and 11–21 as a definite case, a literature review of over 700 studies concluded that a subscale score of ≥8 identifies cases of anxiety and depression [15].

The Pain Catastrophizing Scale (PCS) was used to measure pain-related catastrophizing [16]. It has 13 items that yield an overall score and three subscale scores (rumination, magnification and helplessness). Scores greater than 24 have been associated with higher pain ratings following multi-disciplinary intervention [17]. Reliability and validity of the PCS have been established and confirmed [16], [18], [19].

The Tampa Scale for Kinesiophobia (TSK) is a 17-item questionnaire designed to assess fear of movement or (re)injury [20]. Items are scored on a 4-point Likert scale, and responses summed to yield a net score between 17 and 68, with higher values indicating a greater fear of movement. As an operational definition, 37 has been proposed as the cut-off between individuals with high- and low-fear [21]. The TSK was intended for use in patients with chronic pain, and thus was not completed by the control group. The TSK demonstrates excellent test-retest reliability and correlates significantly with other fear-avoidance measures [22], although its construct validity has previously been questioned [23].

2.3 Measures of symptoms and function

Patient-reported measures of symptoms and function were completed by the PFP group only. Pain severity was measured using a numerical rating scale (NRS). Participants were asked to rate their worst knee pain experienced during the previous week across 11 points, where 0=“no pain” and 10=“worst pain imaginable”.

The Kujala Patellofemoral Score (KPS) was used to measure symptoms and function [24]. The KPS comprises 13 questions related to PFP. Participants are asked to answer each question by selecting a response that best corresponds to their level of activity and symptoms. A final score is calculated as the sum of all items (0–100), where lower scores indicate worse symptoms and disability.

The KOOS was used to characterise the cohort and subgroup PFP participants into high and low severity groups [25]. The original KOOS comprises 42 questions across five subscales that relate to symptoms, pain, function in activities of daily living, sport and recreation function, and knee-related quality of life. We also included the 11-item patellofemoral pain and osteoarthritis subscale (KOOS-PF), which was recently developed for use in PFP conditions [26]. Items are scored from 0 to 4, each of the six subscale scores are calculated independently, and transformed to give a score from 0 (extreme problems) to 100 (no problems).

Statistical analysis

Group data were reported in tables as mean and standard deviations or median and inter-quartile ranges. Differences between groups were reported as the relevant point estimate of effect and 95% confidence interval. All analyses were performed using Stata 14.0 (Statacorp, College station, TX, USA).

For aim 1, the PFP and pain-free groups were compared using two methods. First, t-tests were employed for between-group mean comparisons. Second, recommended measure cut-offs were applied to the psychological measures, dichotomizing both groups. The prevalence of impairment was then presented as a percentage of cases for each group. Once obtained, prevalence rates were used to calculate relative risks with 95% confidence intervals. Relative risks were considered to be either small=2, medium=3 or large=4 [27].

For aim 2, PFP data were analysed in subgroups, determined by k-means cluster analysis, conducted using all six KOOS subscales as multiple variables. Numbers of subgroups, ranging from 1 to 5 were considered, with the Calinski-Harabasz pseudo-F index used to determine the optimal clustering solution. The two-cluster solution produced the highest index score (94.4). Thus, the PFP cohort was partitioned into two groups; less-severe or more-severe PFP. Pain severity, KPS, KOOS, TKS, PCS and HADS were compared between less-severe and more-severe PFP subgroups and the control group using one-way analysis of variance (ANOVA). Bonferroni corrected post-hoc analyses were conducted where main effects were statistically significant (p<0.05).

For aim 3, relationships between psychological factors were assessed using Pearson correlations and Sidak adjustment methods, and were reported as r-values. Effect sizes of correlations were considered as follows: 0.0=trivial; ≥0.1=small; ≥0.3=medium; ≥0.5=large; ≥0.7=very large; ≥0.9=nearly perfect [28]. Multivariate backward stepwise linear regression was conducted (p-out >0.05) to explore how psychological factors, as independent variables, were contributing to disability as defined by the KOOS subgroups (dependent variable). The F statistic and R² value for the model were reported along with standardised coefficients (β) and p-values for each psychological variable.

PFP versus controls

No between-group mean differences were found for the HADS-A, HADS-D or PCS (Table 1). When explored as percentages, the prevalence of anxiety and depression were, respectively, 8% and 6%, higher in the PFP group. The prevalence of pain catastrophizing did not differ between groups (Table 2).

Subgroups of PFP severity

Individuals with PFP were partitioned into a less-severe group (n=57) and a more-severe group (n=43) (Table 1). Significantly higher pain severity (p≤0.001) and consistently worse symptoms and disability (KPS and all KOOS scales; p≤0.001) seen in the more-severe group helped validate the group split. Both groups were comparable in terms of demographic features (Table 3).

Although there were no significant differences in TKS, PCS or HADS between those with less-severe PFP and controls, those with more-severe PFP had significantly higher HADS-D (p=0.01) and PCS (p=0.01) scores than the controls (Table 3). The more-severe group also had significantly worse scores for the TSK (p≤0.001), HADS-D (p≤0.001) and PCS (p≤0.001) than the less-severe group. The largest difference in prevalence was observed for kinesiophobia, which was 27% more prevalent in the more-severe PFP group.

Relationships between psychological factors and their contribution to disability

A range of effect sizes for psychological factor correlations were seen. Anxiety and depression were the only constructs to hold a large correlation (r=0.52). Catastrophizing held medium correlations to all other psychological constructs (ranging from r=0.37 to r=0.47). The only small correlation was between anxiety and kinesiophobia (r=0.29). All results in Table 4.

The multivariate regression revealed that kinesiophobia [β (95% CI), −0.27 (−0.265 to −0.274)] and depression [−0.22 (−0.211 to −0.228)] were the only psychological constructs significantly contributing to the overall KOOS results. The model [F(2,97) 10.07, p≤0.001] explained 17% of the total variance (Table 5).

Clinical implications

Subgrouping our PFP cohort according to severity of symptoms on the KOOS scale, identified significant differences in psychological factors. Those with more-severe PFP had higher levels of kinesiophobia, depression and catastrophizing. Thus, clinicians need to be aware of the possibility of this relationship between high levels of disability and psychological impairment. Psychological factors have been associated with higher pain severity 4 weeks after lower limb trauma [49], and with a lack of improvement and worsening disability following physiotherapy intervention [50]. If identified, psychological impairments could be targeted with specific treatments that reduce the negative influence these factors are likely to have on the outcomes of active management strategies [51].

Strengths and limitations

The strengths of this study include that we: (a) recruited a consecutively sampled PFP cohort from the general public; (b) employed a control group of comparable age, sex and activity levels; and (c) used accessible and easy to administer psychological measures, which have been used extensively in clinical practice and musculoskeletal pain research. While the creation of subgroups is also a strength of this study, the method by which we clustered stands as one of many. We used all subscales of the KOOS as this allowed us incorporate a multitude of factors (pain, symptoms, physical function and quality of life) when defining our clusters. Had the clusters been derived from other condition-related or demographic measures or physical features, results may have differed.

Limitations of this study relate to the nature of the recruited PFP group, the use of recommended measure cut-offs, and the size of the sample. For this study, we focussed on psychological factors in a cohort of PFP who presented without co-existing comorbidities and pain in other body regions. Inclusion of those with PFP plus additional co-morbidities may capture a different demographic in terms of self-perceived disability and psychological factors. As cut-off points of the psychological measures have not been established in PFP groups, we used values from other musculoskeletal cohorts. Whether these cut-offs are appropriate for PFP remains to be determined. Where we dichotomised our PFP cohort on the basis of these cut-offs, our data needs to be interpreted with a degree of caution. Finally, any null effects that we report might also be a function of sample.

Future research

There are several potential avenues for further research into the psychological features of PFP. One research direction involves the study of social factors pertaining to education, employment and relationship status alongside psychological factors. Social factors have been found to influence the experience of musculoskeletal pain [52], [53] and will add further depth to our understanding of PFP subgroups. A second research approach should investigate the causal aspects of psychosocial factors in PFP. This would require prospective cohort studies that are more likely to validate temporal relationships. If high levels of severity of symptoms and disability are a precursor to psychological problems, then early intervention aimed at reducing the severity of PFP may prevent the development of psychological problems and their impact on overall well-being. Previous PFP research has shown that high levels of patient-reported disability can be prognostic of poor long-term outcome [54]. Thus, a third research direction could explore the prognostic capabilities of psychological features in PFP and how attending to these might impact on treatment outcomes.