Introduction

Total knee arthroplasty (TKA) is an effective pain relief, cost-effective, and quality of life intervention for advanced-stage knee osteoarthritis (KOA)1,2. Postoperatively, it has also been reported to improve peripheral and central sensitization3. However, approximately 20% of patients experience persistent postoperative pain (PPP) lasting more than 3 months after TKA4,5. Pain intensity, in particular preoperative pain intensity, is relevant in these cases6,7,8. In addition, acute postoperative pain has also been reported as an associated factor9,10,11. Therefore, perioperative pain management is important to prevent PPP.

Pain involves multidimensional mechanisms and requires an individualized approach according to the factors involved12. In clinical situations, patients undergoing TKA express various descriptions of pain according to individual factors. For example, “throbbing” and “sharpness” are common complaints during the joint inflammation and postoperative inflammatory phases. Patients with nerve damage often express descriptions derived from sensory abnormalities such as “tingling,” “pricking,” and “numbness.” Thus, the description of pain contains information that reflects its etiology and is important for deciding the intervention strategy. Several studies have shown that there is a pathological classification by the description of pain13,14,15, and that the intervention effect differs according to its characteristics14,16,17,18. Although there have been reports analyzing the description of pain in the acute phase of TKA19, to the best of our knowledge, no studies have reported its relationship with PPP. Clarification of the changes before and after TKA and the description of pain associated with prolonged pain may assist in determining interventions that are appropriate for each patient’s condition during the perioperative period.

Therefore, in this study, we aimed to clarify changes in the description of pain before and after TKA and their association with PPP.

Results

Patient characteristics and preoperative/postoperative assessments are shown in Tables 1 and 2 and Figure 1. The Mann–Whitney U test and Chi–squared test / Fisher's Exact Test showed the pain duration (p=0.036), blood loss (p=0.002) and TKA type (P=0.045) significant differences in the presence or absence of PPP at 6 months postoperatively. The Friedman test and multiple comparisons (Bonferroni correction) test showed that the Numerical Rating Scale (NRS) score improved significantly from preoperative to 6 months postoperatively (p<0.001). The Wilcoxon signed-rank test showed significant postoperative improvement in the short-form McGill pain questionnaire-2 (SFMPQ-2) total score (p < 0.001) and the three subclasses (continuous pain, p=0.013; intermittent pain, p = 0.004; and affective pain, p = 0.013). Regarding the description of pain, four items (throbbing pain, p <0.001; sharp pain, p<0.001; aching pain, p=0.001; and tiring/exhausting, p=0.002) significantly improved, and two items (tender, p=0.025 and itching, p=0.013) significantly worsened.

Table 1 Characteristics of the participants.
Full size table
Table 2 Comparison of the preoperative/postoperative description of pain.
Full size table
Figure 1
figure 1

Description of pain preoperatively (red dot plot and left side of the boxplot) and 2 weeks postoperatively (green dot plot and right side of the boxplot).

Full size image

A Generalized Linear Mixed Model (GLMM) of Poisson distribution with NRS at 3 months postoperatively as the objective variable, individual differences (individual ID) as the random effect, and Bayesian estimation using the Markov chain Monte Carlo (MCMC) method (two models: SFMPQ-2 subclass and description of pain) were analyzed. The results showed that, in the description of the pain model, four preoperative items, shooting pain (estimate: 0.42, 95% confidence interval [CI] 0.06–0.80), aching pain (estimate: 0.50, 95% CI 0.02–1.04), caused by touch (estimate: −0.68, 95% CI −1.41–−0.01), and numbness (estimate: 0.42, 95% CI 0.05–0.78) [individual ID for random effects (estimate: 0.37, 95% CI 0.02–0.92)], and one postoperative item, cramping pain (estimate: 0.50, 95% CI 0.03–1.02) [individual ID for random effects (estimate: 0.44, 95% CI 0.03–0.96)] were significantly associated with the NRS at 3 months postoperatively (Table 3). GLMM with the NRS score at 6 months postoperatively as the objective variable did not yield convergence of the MCMC.

Table 3 GLMM for the association between NRS at 3 months and descriptions of pain.
Full size table

The association between the preoperative/postoperative 2-week description of pain and the presence or absence of persistent pain at 3 and 6 months postoperatively was analyzed with a Bernoulli distribution. As the GLMM did not converge with the MCMC, the Generalized Linear Model (GLM) was used to exclude random effects (Table 4). As a result, only cramping pain at 2 weeks postoperatively was associated with the presence of persistent pain at 3 months (estimate: 1.42, 95% CI 0.60–2.37) and 6 months (estimate: 0.95, 95% CI 0.21–1.78) postoperatively. None of the preoperative descriptions showed any significant associations.

Table 4 GLM for the association between persistent postoperative pain and descriptions of pain.
Full size table

Discussion

In this study, we focused on the description of pain, investigated changes before and after TKA, and examined their relationship with persistent pain at 3 and 6 months postoperatively. The main findings showed the following: (1) descriptions of pain, such as “throbbing,” “sharp,” “aching,” and “tiring/exhausting,” improved with TKA, while “tender” and “itching” worsened slightly; (2) some preoperative descriptions (“shooting,” “aching,” “caused by touch,” and “numbness”) and one postoperative description (“cramping pain”) were associated with NRS at 3 months; and (3) only postoperative “cramping pain” was associated with the presence of persistent pain at 3 and 6 months.

The median preoperative descriptions “throbbing” and “aching” were moderate to severe20, but improved significantly postoperatively. “Throbbing” and “aching” pains are caused by joint arthritis, suggesting that the joint problem has been improved by TKA21,22,23. Since pain is also associated with fatigue24, it is considered that those with “tiring/exhausting” pain descriptions showed improvements in pain intensity. On the other hand, those with pain descriptions of “tender” and “itching” showed slight worsening. Two weeks after surgery, when the wound healed and the staples was removed, the residual peripheral/central sensitization caused by the surgical invasion may have caused these descriptors. However, Graven-Nielsen et al. showed that peripheral/central sensitization improves approximately 5–28 weeks postoperatively3, and these temporarily worsening pain symptoms improve over time. In the SFMPQ-2 subscale, only neuropathic pain alone showed no improvement. As arthroplasty does not improve nerve problems, patients with preoperative neuropathic pain should be considered for perioperative treatment, which includes pharmacotherapy18 and electrical stimulation25,26.

The GLMM analysis of Poisson distribution (description of pain model) with individual ID as a random effect and 22 descriptions as fixed effects showed that several preoperative (“shooting pain,” “aching pain,” “caused by touch,” and “numbness”) and postoperative (“cramping pain”) descriptions were associated with the NRS at 3 months. However, the GLM analysis of the Bernoulli distribution showed that only postoperative “cramping pain” was associated with the presence of PPP at 3 and 6 months postoperatively.

The pain expression “cramping pain” is described when the muscles are over-contracted. This type of pain is typically caused by fatigue associated with overuse27. Muscle weakness after TKA has been attributed not only to peripheral skeletal muscles, but also to central nervous system control, including the spinal cord and cerebral cortex28,29. This may also contribute to the occurrence of cramping pain. Pain expressions reflecting motor control problems have also been observed in patients with phantom limb pain17, intractable pain after nerve injury16, and central post-stroke pain30. Some studies have reported that these patients show improvements with interventions for the sensory-motor system using mirror therapy and virtual reality environments16,17,30. In addition, interventions that consider the central nervous system for perioperative TKA/KOA, such as repetitive transcranial magnetic stimulation31, transcranial direct current stimulation32, neuromuscular electrical stimulations33, motor imagery34, have been reported to be effective. Therefore, it is necessary to verify whether such specific interventions that take into account central nervous system control can improve cramping pain and prevent persistent pain in the management after TKA.

This study had several limitations. First, psychological factors6,7,8,35, central sensitization36, and neuropathic pain7 associated with PPP were not evaluated in detail. In addition, characteristics of the participants, pain duration, blood loss, TKA type associate with absent or present PPP at 6 months postoperatively in this study. These factors are probably reflected in the individual ID that was found to be associated with pain intensity at 3 months postoperatively in the GLMM. Second, the description of pain at 3 and 6 months postoperatively was not measured. It remains to be clarified whether the same description of pain as at 2 weeks postoperatively was continued to be used. Future studies could analyze these associations by assessing catastrophic thinking37,38,39, anxiety and depression40, central sensitization-related symptoms36,39,41,42, and neuropathic pain7, as well as by examining pain descriptions at 3 and 6 months. Third, central/peripheral neuromuscular activity measurements, such as functional magnetic resonance imaging, motor-evoked potentials, and electromyography, were not performed. Future studies should include these techniques while considering the underlying neuromuscular activity.

This study focused on the description of pain and clarifying its association with PPP after TKA. We conclude that, in perioperative pain management, it is important to carefully listen to the patient's complaints and determine the appropriate intervention strategy for the clinical condition.

Methods

Participants

The participants were hospitalized patients who underwent primary unilateral or bilateral TKA at Kyowakai Hospital between April 2018 and December 2023. Participants received standardized anesthesia, surgery, postoperative pain management, and guideline-based physical therapy43,44. The exclusion criteria were dementia, higher brain dysfunction, inability to respond to the questionnaire adequately, and revision TKA. The study was approved by the Kyowakai Hospital Ethical Review Committee (approval number: Kyorin18-1), and all participants provided written informed consent in accordance with the Declaration of Helsinki.

Measures

Participant characteristics (age, sex, body mass index [BMI], preoperative pain duration, unilateral or bilateral surgery, Kellgren-Lawrence [KL] grade, operative time, blood loss, wound size, and TKA type) were collected from patient charts. A few days before and two weeks after surgery when the wound is healed and the staples are removed, the NRS was used to assess pain intensity, and the SFMPQ-2 was used to description of pain. In addition, the NRS was obtained 3 and 6 months postoperatively via mail or telephone. The preoperative and 2-weeks postoperative NRS and SFMPQ-2 were assessed in all participant, but lack of contact resulted in one missing NRS score at postoperative 3 or 6 months.

The NRS is the most commonly used measure of pain intensity with an 11-point scale ranging from 0 (no pain) to 10 (worst imaginable pain). The minimal clinically important difference is 22% for acute pain45 and 33% for chronic pain46. In this study, persistent pain was defined as moderate or severe pain (NRS≥3) persisting at 3 or 6 months postoperatively20,47,48.

The SFMPQ-2 is used to assess the intensity and description of pain. It consists of 22 sensory expressions of pain, with each item rated on an NRS scale of 0–10. The higher the intensity of pain, the higher the total score. The SFMPQ-2 has four subclasses (continuous pain, intermittent pain, affective pain, and neuropathic pain)49.

Statistical analysis

Characteristics of participants were compared using the Mann–Whitney U test for continuous variables and Chi-square test or Fisher’s exact test was used for categorical variables.

The NRS assessed the preoperative and 2 weeks/3 months/6 months postoperative scores and compared them using the Friedman test (post hoc Bonferroni correction). In addition, each item was compared using the Wilcoxon signed-rank test preoperatively and 2 weeks postoperatively.

The association between SFMPQ-2 results preoperatively and 2 weeks postoperatively and pain intensity at 3 and 6 months postoperatively was analyzed by GLMM with individual ID as a random effect. A Bayesian estimation of the posterior distribution was performed using the MCMC algorithm50. MCMC is a method for obtaining an approximate solution to the integral through Monte Carlo integration using random numbers generated by a Markov chain, which enables a reasonable estimation for small sample sizes and missing values. The Z-score for each item was used in the analysis. An uninformative prior distribution was used, the number of iterations was set to 2000, burn-in period to 1000, number of chains to four, and 95% CI were given. Rhat values were used to evaluate MCMC convergence, wherein a Rhat value of <1.1 suggested good convergence.

We created two models in the analysis: (1) SFMPQ-subclass: continuous pain, intermittent pain, neuropathic pain, and affective pain; and (2) description of pain: “throbbing pain,” “shooting pain,” “stabbing pain,” “sharp pain,” “cramping pain,” “gnawing pain,” “hot-burning pain,” “aching pain,” “heavy pain,” “tender,” “splitting pain,” “tiring/exhausting,” “sickening,” “fearful,” “punishing/cruel,” “electric-shock pain,” “cold-freezing pain,” “piercing,” “pain caused by touch,” “itching,” “tingling or pins and needles,” and “numbness.” The association between each model and the NRS at 3 and 6 months postoperatively was analyzed using a Poisson distribution. Furthermore, the associated preoperative and postoperative items were analyzed by a Bernoulli distribution with the presence or absence of persistent pain (NRS≥3) with 3 and 6 months postoperatively as the objective variable. Statistical analyses were performed using the R ver. 4.1.3.