The Spine Blog

Friday, March 16, 2018

Large administrative database studies depend on ICD codes to define procedures, comorbidities, and adverse events. The reliability of ICD coding has been questioned, but the research community has generally assumed the coding errors were random and would not confound results when huge numbers of patients were involved. In fact, random coding errors would tend to bias results towards the null and risk missing an association that could be real. On the other hand, non-random coding errors could potentially introduce confounders and bias results towards finding a statistically significant association when none is actually present. In order to better understand this phenomenon, McLynn and colleagues evaluated nearly 800 lumbar fusion patients treated at Yale New Haven Hospital from 2013-2016. They were classified as obese or non-obese based on two definitions: BMI > 30 as defined by the hospital record or having an ICD billing code indicating obesity. Approximately 50% of the patients were obese based on the hospital record. They performed a similar analysis for morbid obesity (BMI > 40 or codes indicating morbid obesity). After identifying the two cohorts of obese patients (defined either by hospital record of BMI or ICD billing code), they found that the sensitivity of the ICD code for obesity was 43% and the specificity was over 99%. The sensitivity of the ICD code for morbid obesity was 61%. Using NSQIP data for comorbidities and adverse events, they found that the ICD coding for obesity was significantly more sensitive for patients with diabetes, ASA class III or IV, those with a post-operative venous thromboembolism, or any adverse event. In a multivariate analysis controlling for age and ASA class, they found a stronger relationship between obesity and adverse events when obesity was defined by ICD code rather than by hospital record. How morbid obesity was defined (hospital record vs. ICD coding) had much less of an effect on the association between morbid obesity and adverse events.

The authors have provided relatively strong evidence that ICD coding errors for obesity are not random at their institution, and this non-random error could introduce confounding into analyses looking at the association between obesity and outcomes when obesity is defined based on ICD coding. If coders systematically code for obesity more frequently in patients with comorbidities and adverse events, this will bias analyses to indicate a stronger relationship between obesity and adverse events than actually exists. The main concern with this study is that it looks at coding practices at a single institution, and how comorbidities like obesity are coded likely varies substantially across institutions. It would certainly be interesting to repeat the study somewhere else. The study also raises the question of the potential for non-random coding errors for other comorbidities, which could have a similar effect. Coders at many institutions are under increasing pressure to code greater numbers of comorbidities in order to maximize case mix index and payments. How this will change the accuracy of comorbidity coding going forward remains to be seen. Readers should be left with a healthy skepticism about large administrative database studies based on ICD coding. These studies should be viewed as hypothesis generating, but prospective studies using clinically derived data are required to confirm the hypotheses generated with administrative databases.

Please read the study by McLynn et al. in the April 1 issue. Does this change how you interpret results of large administrative database studies?

Adam Pearson, MD, MS

Associate Web Editor

Friday, March 9, 2018

Multiple studies have suggested that depressed patients do worse following spine surgery.1 Prior investigations have defined depression in different ways, including patient self-report and scores on various health surveys. While the spine community generally agrees that depression is a predictor of worse outcomes, no spine surgery study has used the PROMIS depression questionnaire to diagnose depression. In order to assess the PROMIS depression questionnaire as a predictor of spine surgery outcomes, Merrilll and colleagues evaluated baseline and 6-month post-operative outcomes including the PROMIS physical function, pain, and depression inventories and the Oswestry Disability Index (ODI) in 111 patients undergoing lumbar laminectomy. The patients were classified as depressed (PROMIS depression score > 50) or non-depressed (depression score < 50) based on their baseline PROMIS depression score. Note that this measure has a mean value of 50 in the general population. Not surprisingly, the depressed patients had significantly worse baseline and 6 month scores on all outcome measures. In an analysis that did not control for baseline differences, the depressed patients actually had a greater degree of improvement on the PROMIS physical function score and depression score compared to the non-depressed group. There were no differences in change scores on the other outcome measures. In a subgroup analysis comparing outcomes between patients with depression scores of 50-60 (mild depression) and over 60 (more severe depression), the severity of depression had no effect on outcomes.

This is an interesting study as it uses the PROMIS depression score as a way to diagnose depression. One concern with this paper is that it classified depression as having a score above the mean, indicating that 50% of the population would be diagnosed with depression using this standard. While that seems like a low bar to classify depression, their analysis demonstrated that severity of depression did not affect outcomes. An important finding is that depressed patients improved to a similar or greater degree than non-depressed patients, though their absolute scores were worse. This makes it clear that depressed patients do benefit from surgery, though patients and providers should have reasonable expectations about their likely outcomes. These findings have some similarities to the Spine Patient Outcomes Research Trial (SPORT), which demonstrated that the treatment effect of surgery (the difference in improvement between surgical and non-operative patients) is similar for depressed and non-depressed patients.2-4 SPORT defined depression as an SF-36 mental component score of < 35, a more stringent definition than that used in the current study. The SPORT analysis also controlled for baseline ODI score, and that analysis showed a smaller change score for the depressed patients. The current study showed similar or greater change scores for the depressed patients, but this is likely due to their worse baseline scores, which were not controlled for in the current analysis. This paper demonstrates that the PROMIS depression score may be a reasonable tool to screen for depression in the spinal stenosis population. The appropriate cut-off for defining depression in this population remains unknown. The real question is whether or not treating depression pre-operatively improves post-operative outcomes. Future studies will have to address that.

Adam Pearson, MD, MS
Associate Web Editor



1.            Slover J, Abdu WA, Hanscom B, Weinstein JN. The impact of comorbidities on the change in short-form 36 and oswestry scores following lumbar spine surgery. Spine (Phila Pa 1976) 2006;31:1974-80.

2.            Pearson A, Lurie J, Tosteson T, et al. Who should have surgery for an intervertebral disc herniation? Comparative effectiveness evidence from the spine patient outcomes research trial. Spine 2012;37:140-9.

3.            Pearson A, Lurie J, Tosteson T, Zhao W, Abdu W, Weinstein JN. Who should have surgery for spinal stenosis? Treatment effect predictors in SPORT. Spine 2012;37:1791-802.

4.            Pearson AM, Lurie JD, Tosteson TD, Zhao W, Abdu WA, Weinstein JN. Who should undergo surgery for degenerative spondylolisthesis? Treatment effect predictors in SPORT. Spine (Phila Pa 1976) 2013;38:1799-811.


Friday, March 2, 2018

Most patients undergoing lumbar discectomy do very well, though approximately 10% experience a reherniation in the decade following their index procedure. Multiple studies have looked at risk factors for reherniation, and the most consistent predictors have been a large annular defect and a large remaining disk volume. In order to better synthesize the existing literature, Dr. Miller and colleagues performed a meta-analysis evaluating the relationship between annular defect width (medial-lateral) and reherniation or reoperation. Their literature search yielded 7 relatively high quality studies that have studied this topic, and they classified annular defects as small (<6 mm) or large (> 6mm). Over 1,600 patients were included, with a median follow-up of about 3 years. The authors did not report the actual reherniation and reoperation rates and instead reported the odds ratios (OR). The pooled OR for reherniation for large relative to small annular defects was 2.5, and the OR for reoperation was 2.3. One study by Carragee et al. had ORs of 9.2 and 8.3 for reherniation and reoperation, with most of the other studies reporting ORs between 1.5 and 3.5. Only one study (Wera et al.) reported ORs of less than 1 (0.76 for both outcomes), and the surgeons in this study performed more aggressive subtotal discectomy (as compared to sequestrectomy), which may have reduced the risk of reherniation. Multivariate analysis evaluating other risk factors was limited by what was reported in the studies and the relatively small number of reherniation patients. This limited analysis suggested that subtotal discectomy and age over 45 may reduce the risk of reherniation associated with a large annular defect to some degree. The authors concluded that annular defect width was a strong predictor of re-operation.

The authors have done a nice job synthesizing the current literature on the topic, and their results show rather conclusively that annular defect width is a strong risk factor for lumbar disk reherniation. It would have been nice if a more robust multivariate analysis had been performed to determine the relative strength of different predictors, but this was not possible due to limitations in what was reported by the different studies and the relatively low number of patients who experienced a reherniation. While this paper yielded information that might be helpful for counseling patients about their risk of reherniation following surgery, it is unlikely to change practice as annular defect size is not an easily modifiable risk factor. Surgeons have experimented with annular closure systems, though these have not been widely adopted. These data may encourage surgeons to minimize the size of the annulotomy they create in cases where the annulus remains intact, though most surgeons likely do this already. This study helps to confirm the belief that a large annular defect is a risk factor for recurrent disk herniation. Unfortunately, there is probably not much we can do about it.

Please read Dr. Miller’s article on this topic in the March 1 issue. Does this change your view of risk factors for recurrent disk herniation? Let us know by leaving a comment on The Spine Blog.

Adam Pearson, MD, MS

Associate Web Editor

Sunday, February 25, 2018

Adult deformity is challenging to treat as non-operative treatment tends to be ineffective, while surgery involves long fusions in an older population and is associated with a high rate of complications. When evaluating the value of any treatment, the benefits of the procedure need to be compared to the costs. The most formal cost-effectiveness analyses determine the incremental cost-effectiveness ratio (ICER) for the treatment, which is calculated by dividing the incremental cost of the procedure by the incremental benefit. The benefit is quantified as quality adjusted life years (QALYs), which are calculated by multiplying the utility of a health state (ranging from 0 to 1, with 0 representing death and 1 perfect health) by the duration of time in that health state. Not surprisingly, there is very little data looking at quality of life gains and cost-effectiveness in adult deformity surgery. In order to help fill this void, Scheer and colleagues from the University of Virginia retrospectively calculated utilities and QALYs following surgical and non-operative treatment of adult spinal deformity. They did this by converting SF-6D scores to utilities through a commonly used statistical conversion. While the baseline characteristics of the surgical and non-operative groups were markedly different, the authors performed a propensity score matching procedure that yielded two cohorts of 44 patients that had completed 3 years of follow-up. Over three years, the surgery patients experienced 1.903 QALYs (average utility of 0.634) compared to 1.749 QALYs (average utility of 0.583) for the non-operative group. That suggests an incremental gain of 0.154 QALYs associated with surgery in the 3 years following the operation.

The authors have done a nice job publishing some of the only QALY data available from adult spinal deformity patients undergoing surgical and non-operative treatment. The authors seemed somewhat surprised that the QALY gains associated with surgery were relatively low. The results can be compared to the SPORT cost-effectiveness analyses for disk herniation, spinal stenosis, and degenerative spondylolisthesis.1 Over 4 years, surgery resulted in QALY gains of 0.34 for disk herniation, 0.22 for stenosis, and 0.34 for degenerative spondylolisthesis. These values are greater than those reported in the current study for adult deformity, but the SPORT data included an additional year of follow-up. If the adult deformity data is extrapolated out for another year, this suggests deformity surgery would result in about 0.2 QALYs gained (assuming no changes in health state from year 3 to year 4). Given the high rate of complications associated with adult deformity surgery—especially compared to the low rates observed following surgery for disk herniation or spinal stenosis—it is not surprising that the QALY gains were more modest. The authors indicated that they are going to perform a cost-effectiveness analysis using these data. A quick back of the envelope calculation using an incremental cost of $100,000 for surgery divided by 0.2 QALYs gained yields an ICER of $500,000 per QALY for adult deformity surgery. Most cost-effective interventions have ICERs below $100,000 (i.e. about $60,000/QALY for spinal stenosis and degenerative spondylolisthesis and $20,000/QALY for disk herniation according to SPORT), so these data suggest that adult deformity surgery is less cost-effective than other well-accepted spine interventions. Given what we know about the costs and outcomes following major deformity correction, that conclusion comes as no surprise.

Please read the article by Scheer et al. in the March 1 issue. Does this change how you consider quality of life changes following major adult deformity surgery? Let us know by leaving a comment on The Spine Blog.

Adam Pearson, MD, MS

Associate Web Editor


1.            Tosteson AN, Tosteson TD, Lurie JD, et al. Comparative effectiveness evidence from the spine patient outcomes research trial: surgical versus nonoperative care for spinal stenosis, degenerative spondylolisthesis, and intervertebral disc herniation. Spine 2011;36:2061-8.


Thursday, February 15, 2018

Lumbar fusion is associated with substantial blood loss, especially in multilevel procedures with deformity correction. Efforts to reduce blood loss have included controlled hypotension, intra-operative blood salvage, and the use of bipolar sealers. More recently, pharmacologic agents like tranexamic acid (TXA) have been shown to significantly reduce blood loss, especially in scoliosis surgery. While not available for use in the United States, batroxobin is derived from snake venom and cleaves fibrinogen to fibrin, which stimulates clot formation. In order to determine the effect of TXA and batroxobin on blood loss during single level TLIF, Dr. Nagabhushan and colleagues from India performed a double blinded, placebo controlled RCT in which 100 patients were randomized to TXA, batroxobin, TXA + batroxobin, or placebo. They found that the three pharmacologic treatment groups had less blood loss than the placebo group, and the combination group had the least intra-operative and total blood loss (intra-operative + drain collection). While the pharmacologic treatments reduced blood loss, there were no significant differences in post-operative hemoglobin levels or the need for blood transfusion. The authors concluded that both pharmacologic agents helped reduce blood loss and that they likely had a synergistic effect by working through different branches of the clotting cascade.

Tranexamic acid has been widely adopted for deformity surgery on the basis of multiple studies showing its efficacy. While it has a clear benefit in multilevel deformity surgery, its potential benefit for smaller operations such as single level lumbar fusion remains unknown. Now that surgeons have adopted lower thresholds for transfusion (i.e. hemoblogin levels of less than 7 g/dL), the need for blood transfusion following single level fusion is rare. As such, it is unclear if there is a clinically meaningful benefit to pharmacologic blood conservation for single level fusion. This paper does a nice job showing that pharmacologic agents can reduce blood loss. However, the authors did not report the transfusion rates for the different groups and only noted that there were no significant differences in the rates. My guess is that the transfusion rate was negligible. The literature demonstrating the benefits of pharmacologic blood conservation is relatively strong, and TXA appears to be a very safe drug without major side effects. Less is known about batroxobin, and this drug requires further study prior to widespread adoption. The spine surgery community needs to work on determining the indications for TXA and other blood conservation agents. If the costs are low and side effects minimal, there seems little downside to using these medications. However, there may also be little benefit to using them for surgeries with anticipated blood loss less than 500 mL.

Please read Dr. Nagabhushan’s article on this topic in the March 1 issue. Does this change how you consider the role of pharmacologic blood conservation agents in lumbar surgery? Let us know by leaving a comment on The Spine Blog.

Adam Pearson, MD, MS

Associate Web Editor