Moderate-to-Severe Pain After Knee Arthroscopy Is Relieved by Intraarticular Saline: A Randomized Controlled Trial

Abstract

We have previously studied intraarticular (IA) analgesics compared with saline 10 mL in 2 randomized clinical trials. The patients who were given IA saline experienced rapid pain relief. Hypothetically, saline may produce a local analgesic effect by cooling or by diluting IA algogenic substances. This randomized double-blind study compared the analgesic effect of IA saline 10 mL with saline 1 mL, which should be a pure placebo. A soft catheter was left IA in 79 patients. We included 60 patients who developed moderate-to-severe pain within 1 h after knee arthroscopy under general anesthesia. A randomized, double-blind controlled comparison of IA saline 10 mL with saline 1 mL followed. Outcome measures were pain intensity, pain relief, and analgesic consumption. Within 1 h pain intensity decreased in both groups from approximately 50 to approximately 27 on a 0–100 mm visual analog scale. Pain intensity remained low and other pain outcome measures were similar during the 36-h observation period. The patients experienced equally good pain relief after IA injection of saline 10 mL and 1 mL. Our finding of a major placebo effect may have implications for the interpretationofpreviouslypublishedplacebo-controlled IA analgesia studies.

In Brief

IMPLICATIONS: In a randomized controlled trial we showed that pain after knee arthroscopy is modest and short-lived and can successfully be treated with intraarticular saline as placebo.

Intraarticular (IA) morphine provides long-lasting analgesia in small doses (0.5–5 mg) when given at the end of arthroscopic surgery (^1–3). However, several controlled studies have failed to show any analgesic effect different from placebo (^4–6). We previously performed a randomized placebo-controlled trial (RCT) without finding any clear analgesic effect of IA morphine 1 mg and 2 mg (⁶). In that study, as in all other previously published RCTs on IA analgesia, the patients were included at the end of surgery before any baseline pain intensity could be established. We believe that study sensitivity can be optimized if all included patients have at least moderate pain before study drug is given (⁷).

We have described a new technique with an IA catheter for test drug administration allowing assessment of baseline pain after the arthroscopic intervention before final inclusion (⁸). In that study, we noted that the incidence of moderate-to-severe pain was only 60%, even though perioperative analgesics or local anesthetics were not given. We also observed a marked pain relieving effect of IA saline 10 mL. This unexpected effect was reproduced in our next IA study in which morphine 2 mg was compared with saline 10 mL (⁹). Consequently, we raised the question whether saline 10 mL is an active treatment rather than an inactive placebo. Hypothetically, saline may produce a local analgesic effect by cooling or by diluting IA algogenic substances. However, this has never been studied.

The aim of the present study was to perform a randomized controlled comparison of the analgesic effect of saline 10 mL compared with saline 1 mL given IA when moderate to severe pain appeared after arthroscopic knee interventions under general anesthesia.

Methods

The Regional Medical Research Ethics Committee for Southern Norway and The Norwegian Medicines Control Agency approved the protocol of the study. The study is reported according to the CONSORT guideline (¹⁰).

Seventy-nine (of 90 screened patients) ASA physical status I–II patients, age older than 18 yr, scheduled for day-case knee arthroscopy gave their written informed consent to participate in the study. The patients were given detailed information about the two different treatments and instructed in the use of the pain scales. The patients were treated from February through November 2002 at the Department of Orthopaedic Surgery, Lovisenberg Diakonale Hospital, Oslo, Norway. Patients were excluded if they had taken any analgesic drug <12 h before inclusion (n = 4), had known morphine hypersensitivity, were pregnant, or experienced only mild postoperative pain (n = 19).

The study was a randomized, double-blind, single-dose, and parallel comparison of IA saline 10 mL (S-10) or saline 1 mL (S-1). The reasons for choosing saline 1 mL (and not 0) as a control are that the volume is so small (<0.7 mL saline is reaching the knee joint) that cooling or dilution IA probably is minimized and in future double-blind studies, a volume of 1 mL has practical advantages compared with a control treatment with 0 mL that will increase the risk of unblinding patients or observers. Gender difference in development of postarthroscopic pain has been documented (⁸). Therefore, patients were stratified for gender and randomized into 1 of 2 treatment groups of equal size in blocks of 6 by use of a list of random numbers (Moses Oakford assignment algorithm) (¹¹) by the senior author who did not interact with the patients. Block size and randomization codes were not revealed to the investigators until all measurements and calculations had been entered into the database. To maintain blinding of both patients and examiner throughout the study, syringes for each patient were prepared in the morning of surgery by a nurse not involved in the treatment or assessment of patients. The syringes were marked according to randomization information, contained in consecutively numbered opaque and sealed envelopes, which were hidden from the patient, the investigators, and the nurse who took care of the patient. The nurse who prepared the syringes also took care of test drug injection and removal of IA catheter. The patients could not observe the procedure.

The IA injection was given through a 20 gauge catheter (Epidural minipack, Portex, Hythe, Kent, UK) after removal of the filter. The catheter had been placed into the treated knee-joint cavity (confirmed arthroscopically) at end of surgery. The technique is described in detail elsewhere (⁸).

The patients were not premedicated. General anesthesia was induced with remifentanil 0.3 μg · kg⁻¹ · min⁻¹ and propofol 1.5–2 mg/kg and maintained with the same drugs supplemented with N₂O in O₂ 30%–50%. A laryngeal mask was inserted. No other analgesic drugs were given during the arthroscopic procedure.

All patients were scheduled for day-case surgery and patients with all kinds of IA preoperative diagnoses were asked for participation in this RCT. The use of tourniquet during the arthroscopic procedure was registered and analyzed as a possible risk factor for postoperative pain.

The number of patients reaching final inclusion criteria is shown in Figure 1. Pain intensity before inclusion was scored on a 5-point verbal rating scale (VRS) where 0 = no pain, 1 = mild pain, 2 = moderate pain, 3 = severe pain, and 4 = intolerable pain. After final inclusion (VRS 2 or 3) pain intensity at rest was scored on a 0–100 mm visual analog scale (VAS) with 0 = no pain and 100 = intolerable pain. Patients with VRS pain 4 (n = 3) were excluded and given immediate pain relief with IV morphine. Pain intensity (VAS) at rest was recorded 10, 20, 30, 60, and 120 min after test drug injection. Pain intensity during exercise was registered after 120 min. Patients walked 5 meters, scoring their maximum pain intensity during this exercise (VASex). The patients were then discharged from the hospital. At home, the patients scored pain intensity at rest (VAS) after 3 and 4 h and VAS and VASex the same evening (approximately 11 h) and the next morning (approximately 22 h).

Figure 1.: Flow diagram showing enrollment, intervention allocation, follow-up, and analyses. The diagram shows the number of participants and gender in each intervention group.

Sum of pain intensity differences (SPID) was calculated for each patient as an area under the curve (time in hours) (¹²). The main efficacy variable was SPID after 60 min.

Pain relief (PAR) on a 5-point VRS with 0 = no PAR and 4 = complete PAR were recorded 10, 20, 30, 60, and 120 min after test drug injection. At the same points in time the patients were asked if the pain reduction was more than 50% compared with baseline pain. All these measurements were performed with the patients at rest.

Time to first rescue analgesic drug was registered during the first 36 postoperative hours. Standard rescue analgesic was oral ibuprofen (Ibux^®, Weifa, Oslo, Norway) 400 mg or, in case of nonsteroidal antiinflammatory drug (NSAID) intolerance, acetaminophen (Paracet^®, Weifa, Oslo, Norway) 1000 mg. If severe pain was present during the early postoperative period, patients were given repeated doses of morphine (Morfin^®, Nycomed Pharma, Oslo, Norway) 2.5 mg IV.

The patients were instructed to record consumption of ibuprofen (or acetaminophen) throughout the day after surgery. Ibuprofen 400 mg and acetaminophen 1000 mg were regarded as approximately equivalent doses (¹³) and were treated as one Defined Analgesic Dose (1 DAD) in the statistical analysis.

Acetaminophen 400 mg + codeine 30 mg is a frequently prescribed analgesic for acute pain in Norway. Its pain relieving effect is comparable with acetaminophen 500 mg in patients with moderate pain (¹⁴) or ibuprofen 200 mg (¹⁵). Two tablets of Paralgin Forte^® were treated as 1 DAD in the statistical analysis for the two patients who, violating the protocol, took this analgesic after discharge from the hospital.

Patient global assessments of response to therapy (PGART) were registered at 2 h and at the end of the study on a categorical scale. The patients answered the question: “When taking both PAR and side effects into consideration, how satisfied are you with the test drug?” The alternatives were as follows:

4 = Very satisfied.

3 = Satisfied.

2 = Neutral.

1 = Not satisfied.

0 = Very unsatisfied.

At every registration point the patients were asked whether they had experienced any side effects. Onset, severity, duration, and type of side effects were recorded.

If rescue medication was given during the postoperative observation period of 2 h, the pain intensity at the time of rescue analgesic intake was carried forward as an estimate of pain intensity for the remaining part of the 2-h observation period. PAR was registered as 0 for the remainder of this observation period. After hospital discharge, the actual values of pain intensity scores, regardless of rescue medication, were used in the statistical analysis.

The patients and the researchers answered the following question at the end of the 2-h observation period: “Do you believe the test medication contained the larger—presumably active—dose or the smaller dose, i.e., placebo?” The alternatives were active dose, placebo, or uncertain.

The data are presented as mean ± SD or median and interquartile range, as appropriate. Mann-Whitney U-test was used for statistical analysis of continuous variables and ordered categorical variables. Pearson χ² test or Fisher's exact test was used for analysis of categorical variables with two categories. Yates correction for continuity was performed for 2 × 2 tables. A P value of <0.05 was considered to indicate statistical significance. All calculations were performed using SPSS^® version 11 (Statistical Packages for the Social Sciences, Chicago, IL). Sample size was estimated based on data (mean pain intensity with SD) from a pilot trial (n = 8). It was calculated that a sample size of 13 in each group would give >80% power to detect a 33% reduction in pain (SamplePower version 2.0; Statistical Packages for the Social Sciences). To account for a possible uneven distribution of patients between treatment groups as a result of the gender-stratified randomization and for possible dropouts, it was decided to include 40 patients. When 30 patients were included the senior author, who was responsible for the randomization procedure, performed an interim analysis. Based on this analysis he decided to increase the sample size to 60.

Results

Seventy-nine patients gave written informed consent. Sixty patients (76%) developed moderate-to-severe pain within 1 h (Fig. 1). There were no statistically significant differences between the treatment-groups in sex, age, weight, height, duration or type of surgery, or baseline pain intensity at final inclusion (Table 1). The use of tourniquet during arthroscopy was evenly distributed between the patients with moderate-to-severe pain (28 of 60) and the patients with no-or-mild pain (9 of 19)(P = 1.00).

Table 1: Distribution of Sex, Age, Weight, Height, Duration and Type of Procedure, and Pain Intensity at Inclusion

There was no statistically significant difference between the two treatment groups in SPID, total PAR, or number of patients with more than 50% PAR after 1 h, or PGART scores after 2 and approximately 24 h (Table 2). The 95% confidence intervals for the mean SPID were 6.5–19.0 (S-1) and 6.7–20.0 (S-10). Median time to rescue analgesic drug was 122 min in S-1 and 125 min in S-10 (P = 0.80). There was no statistically significant difference between the two treatment groups in consumption of analgesic drugs (Fig. 2). During hospital stay, 8 patients needed IV morphine (2.5–10 mg, median 5 mg). Fifty (of 60) patients needed 134 DAD during the 36-h observation period. Median consumption of rescue analgesics was 2 DAD during the 36-h observation period in both groups. As a result of NSAID intolerance 2 of these patients were given acetaminophen, and 2 patients took 2 and 4 tablets of acetaminophen 400 mg + codeine 30 mg.

Table 2: Distribution of Summed Pain Intensity Differences at 60 Minutes, Total Pain Relief at 60 Minutes and Patient Global Assessments of Response to Therapy at 2 and Approximately 24 Hours

Figure 2.: Cumulative consumption of defined analgesic doses (DAD)(mean + SD) after saline 10 mL (S-10) or saline 1 mL (S-1). 1 DAD = ibuprofen 400 mg (n = 134 doses) or acetaminophen 1000 mg (n = 5 doses), or 2 tablets of acetaminophen 400 mg + codeine 30 mg (n = 3 doses).

Pain intensity at rest and during exercise is illustrated in Figure 3. One patient in the control group experienced headache 30 min after inclusion. The questions regarding unblinding confirmed a successful blinding of patients and observers (Table 3).

Table 3: Distribution of Answers to the Question of Unblinding

Figure 3.: Pain intensity scores (0–100 mm visual analogue scale) after saline 10 mL (S-10) or saline 1 mL (S-1) during exercise (upper panel) and rest (lower panel) 0, 10, 20, 30 min, 1, 2, 11, and 22 h after study drug administration. Data presented as median and inter-quartile range.

Discussion

The design of this study made it possible to include only patients with significant baseline pain. A prevalence of moderate-to-severe postoperative pain of 60 of 79 patients (76%) is comparable with earlier results (^8,9). Including all patients would have increased the variance and reduced mean baseline pain. Patients with no or mild pain during the first 2 hours after surgery continued to have significantly less pain during the next 36 hours, and they consumed less analgesics (⁸). Except for our studies all previous RCTs on IA analgesia were preemptive in design. The documented infrequent incidence of pain after arthroscopic procedures requires a reassessment of the conclusions made from these studies, both the positive and the negative ones. Low baseline pain and the large variance will lead to increased risk of type II error (the finding of no difference when a true difference exists). Increasing the sample size may reduce this problem (⁸). A small preemptive study can produce a false positive result by chance. If more patients with no or mild postoperative pain, by chance, are allocated into the treatment group, this may look like a positive effect and lead to a false rejection of the null hypothesis (the finding of a difference when no true difference exists). The large variability in patients' response to the interventions may be an even more important reason for false positive results in small sized RCTs (¹⁶). Studies with positive results (even if false positive) are more often published than studies with negative findings. The high interest aroused by the novel concept of peripheral morphine analgesia resulted in a large number of studies, most of which may have been negative (¹⁷). The well known publication bias may have resulted in the few (by chance) positive studies being published, leaving the larger number of negative studies unpublished (¹⁸).

We did not find any differences in pain relief outcomes between the two treatment groups. Pain intensity decreased rapidly in both groups and most patients experienced more than 50% pain relief within 60 minutes. SPID at 60 minutes was the primary outcome measure. The 95% confidence intervals for the means are almost overlapping. This strongly suggests equivalence. The sample size in this study was larger than most IA RCTs, yet it was too small to conclude that the two treatments are equal. Hence, we cannot reject the null hypothesis. Post-experiment power calculation is not recommended in this situation (¹⁹).

Because the patients who all had moderate-to-severe pain were given two different doses of saline only, the consumption of rescue analgesic drugs was surprisingly small. Ten patients (17%) managed well without any other drug for 36 hours, and median consumption in the remaining 50 patients was 2 doses of ibuprofen 400 mg or acetaminophen 1000 mg. The need for analgesic drugs after such arthroscopic procedures of the knee joint is modest. This may have implications for the extensive clinical practice of preemptive analgesia in day-case surgery.

The patients experienced, on average, very good PAR. The pain after knee arthroscopic procedures under general anesthesia may decrease rapidly after a short peak. Alternatively, the observed PAR was caused by a potent placebo effect. However, we have not tested if normal saline itself even in a volume of 1 mL may be an analgesic, though we find it unlikely. The smooth natural course and a potent placebo effect are important characteristics for this pain model. Hence, it is unlikely that any active treatment (e.g., IA morphine) can be so efficient as to reach a statistically significant difference compared with placebo. At least the sample size must be larger. When reading previously published IA studies and planning new ones, the present study illustrates the importance of rigorous scientific design in future RCTs on IA analgesia (¹⁷).

We conclude that patients who had moderate-to-severe pain after knee arthroscopic procedures experienced equally good PAR after IA injection of saline 10 mL and 1 mL. This indicates a major placebo effect or a smooth natural postoperative course with minimal pain and requirements for analgesics. This may have implications for the interpretation of previously published placebo-controlled IA studies.

The authors thank the following nurses who prepared syringes on the day of surgery and later injected the test medication: Sigrid Braaten, Cecile Braein Nilsen, Nina Stange, Lene Laukvik, and Dag R. Evensen. We also thank the Orthopaedic Surgical Department at Lovisenberg Diakonale Hospital, Oslo, for cooperation in enrollment of patients.