Worst pain intensity and opioid intake during the early postoperative period were not associated with moderate-severe pain 12 months after total knee arthroplasty – a longitudinal study

Introduction

Moderate-severe pain persisting beyond 3 months after total knee arthroplasty (TKA) remains a major issue affecting 10–34% of patients [1]. A number of preoperative risk factors have been identified as predictors of significant chronic pain following TKA, including psychological factors [2], pain intensity and widespread pain [3], female gender and younger age [4], higher body mass index (BMI) [5], and long-term use of opioids [6, 7]. In addition, patients’ acute postoperative pain experiences have been identified as a potential contributor to chronic pain across diverse surgery types [8], including TKA [9], [10], [11], [12]. However, prior studies of the association between acute and chronic pain following TKA have not accounted for patients’ preoperative pain level, a key potential confounder that is known to predict pain following TKA [3, 13]. Moreover, prior studies have been limited by short follow-up periods (i.e., six months) [9], [10], [11], despite a systematic review [14] showing that clinically important improvement occurs in the 6–12 month recovery period. Thus, a 12 month follow-up period would allow for a more reliable determination of chronic pain. Research is therefore needed to determine whether acute pain intensity during the early postoperative period is independently associated with moderate severe pain 12 months after TKA, even when accounting for the potentially confounding effect of preoperative pain.

Opioids are an important component of multimodal perioperative pain management [15], and thus, their association with the development of moderate-severe postoperative pain following TKA also needs to be evaluated. Unfortunately, long-term, and even short-term, opioid administration, as well as high doses may result in patients becoming more sensitive to pain, potentially leading to more severe pain over time [16]. According to one study [17], use of opioids for more than one month prior to surgery was associated with more severe pain at rest and walking during the first six postoperative days after TKA. Long-term use of oral opioid medications for knee pain before surgery increased the risk of chronic pain for up to two years after TKA [6, 7]. These findings suggest that patients’ preoperative opioid intake may be a potential risk factor for long-term pain and may be a potential confounder when assessing the relationship between postoperative opioid intake and moderate-severe pain following TKA. To our knowledge, only one study [18] found an association between a higher intraoperative remifentanil dose and severe pain one year after cardiac surgery.

Based on our review of the current evidence on the relationships between early postoperative pain intensity and postoperative opioid intake and chronic postoperative pain, we aimed to test the following hypotheses: (1) high levels of worst pain intensity in the early postoperative period are independently associated with moderate-severe pain 12 months after TKA; and (2) higher opioid intake in the early postoperative period is independently associated with moderate-severe pain 12 months after TKA.

Methods

This preplanned sub-study is part of a larger, prospective longitudinal study at Lovisenberg Diaconal Hospital, Oslo, Norway between October 2012 and September 2014 [19]. Inclusion criteria: primary TKA for osteoarthritis, age ≥18 years, able to read/write in Norwegian. Exclusion criteria: dementia or revision surgery.

Patients received information about the study by mail. Upon admission for TKA, a study nurse screened and formally invited eligible patients to participate. All patients signed a consent form.

All patients received similar posterior cruciate-retaining fixed modular-bearing implants. Unless contraindicated, patients received a neuraxial block with bupivacaine for anesthesia, followed by epidural analgesia with continuous infusions of bupivacaine 1 mg/mL, adrenaline 2 μg/mL, and fentanyl 2 μg/mL (5–12 mL/h), oral acetaminophen 1 g every 6 h, and celecoxib 200 mg plus controlled-release oxycodone 5–20 mg every 12 h. Alternatively, patients received total intravenous anesthesia and a continuous femoral nerve block with bupivacaine 2.5 mg/mL 4–10 mL/h. Ketamine infusion 1.5 μg/kg/min was available during the first 48 postoperative hours. Regional blocks were usually removed on postoperative day (POD)2 [19].

Results

The initial study sample included 202 patients. Fifteen patients were excluded due to incomplete follow-up data, leaving 187 patients (93%) for analysis. At 12 months, 40% of patients rated their worst pain intensity in the moderate-severe range (NRS ≥ 4), and their mean worst pain score was 3.2 (SD=2.4). Demographic and clinical data are summarized in Table 1.

As illustrated in Figure 1, patients’ mean worst pain intensity on the day of surgery was 4.3, with a peak of 5.9 on postoperative day two (POD2), followed by a decrease to 5.0 on both POD3 and POD4.

Figure 1:

Mean value trajectory for postoperative ratings of worst pain intensity after TKA (n=187).

NRS, numeric rating scale; POD, postoperative day.

As shown in Figure 2, patients’ mean daily morphine-equivalent dose of opioids on the day of surgery was 12.4 mg, with a peak of 15.6 mg on POD2 and a decrease to 10.7 mg on POD3.

Figure 2:

Mean value trajectory for postoperative opioid intake in morphine milligram equivalent (MME) dose after TKA (n=187).

POD, postoperative day.

We examined the univariate associations of the potential covariates with moderate-severe pain 12 months after TKA (see Table 2). Two covariates were significantly associated with moderate-severe pain 12 months after TKA: higher BMI (OR=1.1, 95% CI [1.0–1.2], p=0.03) and higher preoperative pain (OR=1.5, 95% CI [1.2–1.8], p=0.001). The remaining covariates did not reach the level of statistical significance in the univariate analyses and were not entered into the multivariate models.

Table 3 shows the estimates from the univariate and the multivariable logistic regression models for worst pain intensity on POD 0 to 4. The univariate analyses showed that worst pain intensity on POD2 (OR=1.2, [95% CI 1.0–1.3], p=0.027) and POD3 (OR=1.2, [95% CI 1.0–1.3], p=0.05) were significantly associated with moderate-severe pain 12 months after TKA. The multivariable logistic regression model included worst pain intensity on POD 0 to 4 as possible predictive factors and statistically significant covariates from the univariate analyses (i.e., BMI and preoperative pain) as possible confounders. Contrary to our first hypothesis, in the final model, the intensity of worst postoperative pain on days 0–4 was no longer significantly associated with moderate-severe pain 12 months after TKA. However, preoperative pain, but not BMI, remained statistically significant.

Table 4 shows the results from the univariate and multivariable logistic regression models for opioid intake on POD 0 to 3. The univariate analyses showed that postoperative opioid intake on days 0–3 was not significantly associated with moderate-severe pain at 12 months. The multivariable logistic regression model included postoperative opioid intake on days 0–3 as possible predictive factors and statistically significant covariates from the univariate analyses (i.e., BMI and preoperative pain) as possible confounders. Contrary to our second hypothesis, postoperative opioid intake on days 0–3 was not independently associated with moderate-severe pain 12 months after TKA in the final model, but both preoperative pain and BMI remained statistically significant.

Discussion

This study builds on prior studies of the relationships of early postoperative worst pain intensity and opioid intake to moderate-severe pain 12 months after TKA. However, a unique contribution of our study is that we controlled for multiple potential confounders, most importantly, preoperative pain, which is a known predictor of postoperative pain. This analysis adds to our understanding of the relationship between acute worst pain intensity and opioid intake trajectories, and moderate-severe pain 12 months after TKA. Contrary to our hypotheses, our study did not reveal any statistically significant associations between either early worst postoperative pain intensity or opioid intake and moderate-severe pain 12 months after TKA, when controlling for the potential confounders including preoperative pain and BMI.

Over the first 3–4 postoperative days, trajectories of patients’ worst pain intensity and opioid intake followed a similar pattern, with relatively low intercepts, rising to peaks on POD2, followed by slight declines on POD3. The peak in worst pain intensity on POD2 coincided with the end of regional nerve blocks. As highlighted in a prior review [24], rebound pain after discontinuation of nerve block effects is a well-known issue that may be avoided by using a multimodal pain management regimen tailored to individual patients through systematic pain assessment [25]. In our study, patients’ worst pain intensity levels on POD1 and POD2 were 5.3 and 5.9, respectively. These pain levels are comparable to previous studies following TKA [26, 27] or mixed surgical procedures [28, 29], which reported pain intensity levels ranging from 4.0 to 9.0, indicating suboptimal pain management. To optimize and individualize postoperative pain management, it is critical to determine whether pain is a result of poor pain management or modifiable preoperative risk factors for pain.

We hypothesized that greater worst pain intensity during the early postoperative period would be independently associated with moderate-severe pain 12 months after TKA. The univariate associations between some of the postoperative ratings of worst pain intensity (i.e., POD2 and POD3) and pain at 12 months were consistent with findings from four similar studies [9], [10], [11], [12] that found acute pain intensity in the days and weeks following surgery increased the risk of moderate-severe postoperative pain in TKA patients. However, these associations did not remain statistically significant once we controlled for body mass index and preoperative pain. Interestingly, none of the previously mentioned studies controlled for preoperative pain, which may explain at least in part why our findings were not in line with previous studies. Furthermore, variation in the design of previous studies, such as using pain assessment scores at 2 and 8 weeks after TKA as predictor variables [11], using a retrospective design [12] or evaluating pain only six months after TKA [9], [10], [11], may also have contributed to the differences in findings. Our study’s prospective design, earlier postoperative assessment points, and adjustment for key confounders may at least partially explain the differing results across studies. Our results from the multivariable models are consistent with previous research in patients undergoing total hip arthroplasty surgery (THA) [30, 31], which found no link between acute and chronic postoperative pain intensity. However, hip and knee arthroplasty are not directly comparable, as THA patients are generally characterized by lower acute postoperative pain scores than TKA patients [32].

Postoperative opioid intake on days 0–3 was not independently associated with moderate-severe pain 12 months after TKA in either univariate or multivariable analyses. To our knowledge, no studies have evaluated the relationship between opioid intake in the immediate postoperative period and chronic postoperative pain following TKA. However, a study in cardiac surgery patients [18] found a link between higher intraoperative remifentanil doses and more severe thoracic pain after 12 months, suggesting that there is a dose-dependent relationship between higher intraoperative opioid doses and increased incidence of pain one year after surgery. While this finding contradicts our results, it could be explained by differences in study populations, study designs, and surgical procedures. Furthermore, in our study, patients’ mean daily opioid intake was relatively low (<20 morphine milligram equivalent), which could reflect the potential opioid-sparing effect of the relatively complex multimodal analgesia [33]. Despite this, some patients experienced moderate-severe pain and were given a low-dose N methyl-D-aspartate receptor antagonist ketamine infusion as supplemental treatment within the first 48 h after surgery. According to a recent meta-analysis [34], ketamine is effective in reducing postoperative opioid intake in the first 24–48 h, along with decreased pain intensity. Ketamine is also hypothesized to decrease the probability of acute pain transitioning to chronic pain [35]. Neither pain intensity nor opioid intake during the early postoperative period were linked to the development of moderate-severe pain 12 months after TKA. A possible explanation could be the use of a multimodal analgesia approach in this study, which resulted in adequate acute pain management and lower opioid doses [36].

The results of this study are consistent with a systematic review [3] that identified preoperative pain as a risk factor for persistent pain. Moreover, the significant association between higher BMI and moderate-severe pain found in our study is in line with a study by Fisher et al. [5]. However, in another systematic review [37] that evaluated the effect of higher BMI on chronic postoperative pain, only four of the included studies found a significant association, while seven studies found no association, suggesting that this factor warrants more research.

This study’s major strengths are its prospective cohort design and low attrition rate. Patients were included from all regions of Norway, which may have increased the generalizability of our findings. Despite our efforts to include all relevant confounders, some factors identified in previous studies, including psychological factors [2] and time spent in severe pain [38], were not measured in this study.

In conclusion, neither worst pain intensity nor opioid intake in the early postoperative period was associated with moderate-severe pain 12 months after TKA after controlling for relevant confounders. More research that addresses the limitations of this and other studies is needed to further advance our understanding of modifiable risk factors for chronic pain after TKA.