Femoral nerve infusion after primary total knee arthroplasty

Pilot study and power calculation

We first performed a prospective, non-randomised, non-blinded observational pilot study in 76 patients who underwent a TKA under spinal anaesthesia at our institution. We examined VAS pain scores and opiate doses in patients receiving a CFNI compared with single-shot nerve block plus PCA. We found that on average there was a difference of 10 mm on the VAS for pain (sd 1.6). A 10 mm VAS difference is widely accepted as the minimum for clinical relevance. To detect this difference in a randomised controlled trial at any time point, using an unpaired t-test, with an alpha of 0.05 and a power of 80%, we required 40 patients per group.

Surgical technique

All TKAs were performed by one of four consultant orthopaedic surgeons using a medial parapatellar approach. The patella was everted and no patellar denervation was performed. A posterior-stabilised prosthesis was used without a mobile bearing and the patella was left unresurfaced.

Institutional review, ethical and trial registration approval

We obtained approval from the Institutional Review Board (Otago District Health Board) following Local Ethics Committee approval (LRS/08/08/030). We also registered the study as a clinical trial (ACTRN12609000406202).

Recruitment

From January 2010 until October 2011, patients listed for a primary unilateral TKA for osteoarthritis were invited to participate in the study via an information sheet posted at the same time as their appointment, to attend the pre-operative assessment clinic one week prior to the surgery date. Patients were approached by a research nurse at the clinic and those willing to participate were consented for the trial. The inclusion criterion was a unilateral primary TKA for osteoarthritis. Exclusion criteria are listed in Table I.⁹

Randomisation

The randomisation was by sealed envelope technique and independent third party:

50 cards for each study arm were placed in sealed envelopes by an independent nurse, shuffled and placed within a randomisation box.

Randomisation was undertaken in recovery by a nurse who was not involved in the study in any other way (she withdrew the envelope, prepared the drug infusion and completed the randomisation paperwork).

Blinding

Patients, surgeon, research nurse, medical statistician, ward nurses and physiotherapists were blinded to the intervention as the infusions were prepared by a recovery nurse on the day of surgery who had no further contact with the patient.

Femoral nerve infusion catheter

The groin was shaved and prepared with a chlorhexidine solution followed by a local anaesthetic with 1% lignocaine. An 18 G Touhy Nerve block needle (Braun CONTIPLEX, Melsungen, Germany) was then introduced under ultrasound and nerve stimulator guidance (placement determined by adequate motor response of patellar twitch 0.6 mA) directly distal to the inguinal ligament. Once in position 15 ml 0.25% bupivacaine was injected around the femoral nerve in aliquots. The catheter was then advanced 3Â cm to 4 cm beyond the needle tip and an IV3000 dressing with surrounding tape applied, thus securing it to the lower abdominal wall. The catheter/block was placed prior to the subarachnoid anaesthetic to avoid masking the intra-neural injection.

Subarachnoid blockade and sedation/light general anaesthetic

A standard spinal technique was used with 0.5% bupivacaine plain/heavy (10 mcg to 20 mcg) plus intrathecal morphine 1.5 mcg/kg (max 150 mcg). The patient received sedation with a midazolam or propofol infusion titrated to effect. General anaesthesia was provided with fentanyl/propofol induction and sevoflurane maintenance and intra-operative analgesia with paracetamol 1 g IV after induction.

Surgery

Patients received a standard cemented condylar TKA via a medial parapatellar approach by a consultant orthopaedic surgeon.

Intervention

Randomisation was performed in the post-anaesthetic care unit (PACU). A femoral nerve infusion was prepared in PACU and commenced at 10 ml/hr with normal saline in the control group and 0.125% bupivacaine infusion at 10 ml/hr in the intervention group. The infusion continued for 48 hours. Patients were mobilised in the initial 48 hours under strict supervision either by a physiotherapist or a nurse. Complications of motor weakness, local anaesthetic toxicity, persistent pain from the infusion site, localised inflammation or signs of infection, haematoma and catheter dislodgment, were recorded. Leakages were dealt with by application of a firm dressing and IV3000, with a sterile swab placed over the insertion site.

Post-operative analgesia

Patients were prescribed paracetamol 1 g four times a day regularly, oxynorm 5Â mg to 10 mg as needed up to every three hours commencing on day 0, oxycontin SR 10 mg twice a day from day 1 and laxsol 2 tablets commencing on day 0. On the evening after surgery if a dose of more than 20 mg of oxynorm was required or the patient had severe pain not controlled by oral analgesia, a fentanyl PCA was started.

On the first post-operative day, if pain was not controlled, an increase of oxycontin to 20 mg twice a day with oxynorm 5 mg to 10 mg as needed every three hours, and fentanyl PCA available as a second line, was allowed. On the second post-operative day, a reassessment of oxycontin requirements was done based on the last 24 hours of use. If the pain remained poorly controlled the patient commenced a fentanyl PCA. To prevent potential confounding errors, anti-inflammatories, clonidine and tramadol were avoided.

Outcome measures

The primary end point was peak pain at rest over the first 72 hours after the operation. Pain was measured by a VAS 0 mm to 100 mm scale at four-hourly intervals during the first 24 hours, then at 30, 36, 42, 48, 54, 60, 66, and 72 hours thereafter. Participants were not woken to assess their pain, which explains some of the missing data. Secondary end points were range of movement (ROM), side effects (sedation, pruritus, nausea and vomiting, leakage, toxicity, bleeding, infection,inflammation, catheter blockage), cumulative narcotic dose in morphine equivalents, infusion rates and reasons for any changes, duration of in-patient stay and reason for any discharge delays.

Significance was set at the 5% level. No adjustment for multiple testing was done. The primary outcomes were tested by using a mixed model to allow for the repeated measurements of pain. For continuous data, Mann–Whitney tests were used to test for differences in non-normally distributed data (VAS and use of analgesia) and unpaired t-tests for normally distributed data (knee ROM and length of hospital stay.) Chi-squared tests were used to test for differences in proportions. All analysis was done using Stata 10 (StataCorp, 4905 Lakeway Drive, College Station, Texas), and with participants in the group to which they were randomised.

Recruitment and demographics

Of the 226 patients who were invited to participate in the study, 34 declined and 75 were excluded for the following reasons: comorbidities (33), age > 85 years (11), exceeded daily morphine dose (11), BMI > 40 (9), declined spinal anaesthetic (5), morphine allergy (5) and unable to comprehend the VAS score (1). In total, 113 were found to be eligible for inclusion in the study. Of these, 31 were excluded prior to randomisation for the following reasons: anaesthetist not trained in femoral nerve catheter technique (27), failed spinal anaesthetic (2) and failed femoral nerve catheter insertion (2). The remaining 86 patients were randomised and 42 of the 43 completed the study protocol in each group. One patient in each group was withdrawn from the study as a result of a cardiac event and excessive uncontrolled pain, i.e. requiring more adjunct analgesia than that prescribed in the protocol.

With the exception of a predominance of males in the treatment group, there were no important differences between either group in terms of baseline demographics and clinical characteristics (Table II).

Primary end point

There was no difference between the groups in pain experience over the first 72 hours after the operation (difference 1.5 mm, 95% confidence interval (CI) -3.4 to 6.4). The confidence interval did not reach the minimum clinically important difference (10 mm) so the differences are unlikely to be important. Adjusting these results for gender did not show any difference. At all stages there was no significant difference (p = 0.01) in the pain VAS score between the control and treatment groups (Fig. 1) -3.4 to 6.4). The confidence interval did not reach the minimum clinically important difference (10 mm) so the differences are unlikely to be important. Adjusting these results for gender did not show any difference. At all stages there was no significant difference (p = 0.01) in the pain VAS score between the control and treatment groups (Fig. 1). The confidence interval did not reach the minimum clinically important difference (10 mm) so the differences are unlikely to be important. Adjusting these results for gender did not show any difference. At all stages there was no significant difference (p = 0.01) in the pain VAS score between the control and treatment groups (Fig. 1) so the differences are unlikely to be important. Adjusting these results for gender did not show any difference. At all stages there was no significant difference (p = 0.01) in the pain VAS score between the control and treatment groups (Fig. 1) in the pain VAS score between the control and treatment groups (Fig. 1). At 24 and 48 hours there were no significant differences (p = 0.07) although at 24 hours, the treatment group had a pain score 10 mm less than the control group.

Fig. 1

Visual analogue pain scores for intervention and control group over time (blue dots, intervention group; red triangles, control group.

Secondary end points

There was no significant difference in total analgesia used in morphine equivalents at any time point during the study or overall. Furthermore, there were no significant differences in side effects between the two groups (p = 0.2 and p = 0.35, respectively) (Table III). There was no significant difference in the occurrence of motor block between the two groups (all p-values > 0.05) (Table IV) and no falls occurred. There were no infections and no haematomas at the catheter site. There was no significant difference in knee ROM between groups, on any day or at time of discharge (p = 0.45, 0.78, 0.52, 0.36 for days 1, 2, 3 and discharge, respectively) (Table V). There was no significant difference in length of hospital stay between groups (p = 0.87) (Table VI).

Discharge criteria included independent transfers and activities of daily living, satisfactory wound healing, crutch walking for 20 metres and ability to walk up and down stairs. Four patients in each group had delayed discharge. The reasons for delay in discharge were variable but independent of the study protocol.