Desire for newest treatment may affect patient selection, outcomes

In recent years, the issue of bias in medical studies has gained national attention. Although much of that attention has focused on the way that funding can affect the reporting of results, other factors may also bias results. Physicians and patients, influenced in subtle ways by cutting-edge devices and treatments, can make selection decisions that affect the outcomes of clinical trials.

For example, a recent clinical trial compared cervical total disk replacement (TDR) and anterior cervical discectomy and fusion (ACDF). In the Food and Drug Administration (FDA) investigational device exemption (IDE) trial, the reoperation rate at 2 years for the TDR group was 2.9 percent, compared to 12.1 percent for the single-level ACDF group, although both groups had excellent outcomes (Table 1).

“The reoperation rate for the ACDF group seemed alarmingly high to us based on our clinical practice,” said Kern Singh, MD, at the North American Spine Society (NASS) Annual Meeting.

In response, Dr. Singh and his colleagues conducted a study comparing reoperation rates between patients in the ACDF group of the clinical trial and patients from their own practice who met the same inclusion criteria.

“We wanted to understand the unique factors that affect the decision making of both the surgeon and the patient within and outside of an FDA IDE study,” Dr. Singh said.

Taking a closer look at reoperation rates
The patient demographics of the two groups were similar in most categories (Table 2). Patients in the clinical practice group had been diagnosed with spondylotic radiculopathy or myelopathy and underwent ACDF with an allograft and plate. Final follow-up for these patients occurred contemporaneously with the IDE trial group.

“Our rate of reoperation within 2 years of ACDF was approximately 2.1 percent. This was in comparison to the 12.1 percent reoperation rate in the IDE trial,” said Dr. Singh. “Even at the mean of 3.5 years, the reoperation rate after ACDF was just 4.2 percent in a routine clinical practice (Table 3).”

Causes for the discrepancy
According to Dr. Singh, factors that influence this difference are outside the study itself. “We have to be cautious of unintentional biases in an IDE trial. Surgeon- and patient-driven desire to enroll in the trial may unintentionally bias the control group, which in this case received ACDF.”

For example, patients whose stated preference was disk replacement may be randomized to the control group. The fusion procedure reduces neck and arm pain, but the patients believe they received the inferior treatment.

“In this case, patients may feel better but perceive that they would be more improved with the more novel treatment. That establishes a bias against fusion,” said Dr. Singh.

“Likewise, if a patient with two-level disease is randomized to the TDR group and has a disk replacement on one level and a fusion on the second level, the surgeon may unintentionally suggest that any arm pain the patient experiences after the surgery is due not to the TDR but to the fusion, which is accelerating the degeneration. The surgeon may then indicate a revision is needed,” Dr. Singh explained.

Another reason for the discrepancy may be that the two groups had different thresholds for revision. “For example, it may be easier for the surgeon to admit failure and reoperate in the control group than in the TDR group,” he continued.

Better indicators of failure
Dr. Singh and his colleagues do not believe revision is a strong end point because the decision to operate is elective and subjective. “This is supported by the significant difference in threshold to reoperate within our own clinical practice using a very similar technique,” he stated.

Criteria that may be better indicators for failure of treatment include oral pain medication requirements, postoperative epidural/selective nerve root injections, and the need for additional physical therapy.

“Rates of additional surgery after ACDF varied 5-fold between our routine clinical practice and an IDE trial. I urge everyone to interpret the claims with caution. The difference likely may reflect enrollment biases on the part of both the surgeon and patient. We clearly need to understand this process better,” Dr. Singh concluded.

“Factors affecting re-operations after ACDF within and outside of an FDA IDE cervical TDR trial” received a NASS Best Paper award. Co-authors include Frank M. Phillips, MD; Daniel K. Park, MD; Jay Shah, BS; Howard S. An, MD and Edward J. Goldberg, MD.

The co-authors have the following disclosures: Dr. Singh—Stryker Spine, Pioneer; Frank M. Phillips, MD—DePuy, Nuvasive, Kyphon Inc., AxioMed, Flexuspine, Cervitech, Stryker, Arcus, Spinal Motion, Spinal Kinetics, TissueLink; Howard S. An, MD—U & I Inc., Stryker, DePuy Spine, Medtronic Sofamor Danek, Johnson & Johnson, Spinal Kinetics Inc., Pioneer Surgical Inc., DePuy Spine Biologics Inc.; Edward J. Goldberg, MD—Stryker, Biomet. Daniel K. Park, MD, and Jay Shah, BS, did not report any disclosures.

Annie Hayashi is the senior science writer for AAOS Now. She can be reached at hayashi@aaos.org