A Randomized Controlled Trial of Total Knee Replacement

Published - Written by Andrea Merrill

Insight post 10-21-15Ms. Smith, as we’ll call her, was a 60-something year old woman I met on my medicine rotation during my third year of medical school. She had presented to the emergency room with fevers, chills, and pain and redness around her knee- a knee that contained hardware from her total knee replacement (TKR) three years prior. Unfortunately, workup revealed septic arthritis of her knee plus bacteremia necessitating surgery and removal of that hardware. And thus began her long road of multiple operations, prolonged antibiotic therapy, and grueling rehabilitation. As Ms. Smith reflected on her situation prior to surgery, she wondered if she had made the right decision three years ago. Maybe she should have tried a course of physical therapy first instead of choosing surgery? If she had, she certainly wouldn’t be facing this situation today, but would she still be in constant daily pain from her osteoarthritis (OA)?

These questions posed by Ms. Smith are common and valid, but until recently, there was no randomized clinical trial addressing them. This week’s issue of NEJM fills this void with a new randomized controlled trial by Skou et al. comparing TKR to non-surgical treatment for the management of moderate to severe knee OA. They randomized 100 total eligible patients from Aalborg University Hospital in Denmark — 50 to non-surgical treatment (12 weeks of individualized exercise, education, dietary advice, insoles and pain medication) and 50 to TKR followed by non-surgical treatment. The primary outcome was the between-group difference in change from baseline to 12 month KOOS4 score — a mean of scores for pain, symptoms, activities of daily living (ADL) and quality of life (QOL); there was a battery of secondary outcomes that measured additional aspects of knee function. Adverse outcomes were also recorded for each group.

Nearly 100% of the non-surgical group and about 90% of the TKR group completed the 12 month follow-up. 13 patients (26%) from the non-surgical group underwent TKR during the 12 month follow-up period, while 1 patient in the TKR group opted for non-surgical treatment only. All patients were analyzed via an intention-to-treat analysis while about 50% of each group were included in the per protocol analysis. Baseline characteristics were similar between the two groups.

For the TKR group, KOOS4 scores (on a scale from 0-100) improved by 32 units as compared to 16 in the non-surgical group, a statistically significant difference. The TKR group also had greater improvements in their scores in the subscales of the KOOS score (pain, symptoms, ADL and QOL), the Timed Get Up and Go Test, and the EQ-5D index, compared to the non-surgical group. However, serious adverse events were more common for the TKR group including 3 DVTs, 1 deep infection, and 1 supracondylar femur fracture in the TKR group.

The takeaway message from this trial is that while both TKR and a regimented non-surgical program (which includes physical therapy) resulted in clinically relevant improvements in pain and function, TKR resulted in significantly greater improvements in these outcomes but at the price of more serious advents.

So with these results in mind, what do we tell our Ms. Smiths of the world when they ask about treatment options for knee OA? The accompanying editorial by Dr. Jeffrey Katz sheds some light on this issue which he says ultimately comes down to informed patient choice. Dr. Katz comments, “For most patients, the dramatic pain relief associated with TKR provides a compelling rationale to elect surgery. Others, particularly those more risk-averse, may prefer nonoperative care. As patients vary considerably in their preferences, physicians should present the relevant data to their patients and then listen carefully.”

The authors of the first study and the editorialist are available through October 30th to answer your questions on the NEJM Group Open Forum. There is also an NEJM Quick Take video summary available.