Modeling Patient Recovery in Thoracolumbar Surgery: An Ambispective Cohort Study with 2-year Follow Up

Presented at SMISS Annual Forum 2014
By Neil Manson MD, FRCSC
With Alana Green BAH, Edward Abraham MD, FRCSC,

Disclosures: Neil Manson MD, FRCSC A; Unrestricted Research Grant, $85,000/annum, Medtronic Canada. B; Education Fees, Medtronic Canada. Alana Green BAH None, Edward Abraham MD, FRCSC A; Unrestricted Research Grant, $85,000/annum, Medtronic Canada. B; Education Fees, Medtronic Canada.,

Introduction:
Identifying appropriate candidates for thoracolumbar surgery can be complex. A multitude of demographic and surgical factors have been shown to affect patient outcomes. However, these factors are often examined on an isolated basis with insufficient statistical methodology to be clinically relevant.

Aims/Objectives:
To apply a highly validated statistical technique that has so far not been applied to this patient population, in order to identify baseline variables that predict patterns of patient recovery from baseline to two-years after thoracolumbar surgery.

Methods:
This was an ambispective cohort study. We identified 275 consecutive adult patients with no history of malignancy or litigation. Primary outcome variables were the SF-36 Mental and Physical Component Summaries (MCS, PCS) and Oswestry Disability Index (ODI) scores pre-operatively and at 6, 12 and 24 months post-operatively. A latent trajectory growth analysis was used to explore patient recovery for each outcome variable. This method groups patients based on their recovery patterns and identifies the pre-operative variables that predict an individual’s likelihood to fall into one pattern vs. another. A number of patient demographics and surgical factors were used.

Results:
There was evidence for 3 recovery patterns for PCS and MCS scores, and 4 recovery patterns for ODI scores. PCS recovery patterns were labeled High Start Improvers (HSI, 31.1%), Medium Start Improvers (MSI, 40.4%), and Low Start Non-Improvers (LSNI 28.5%). MCS patterns were similarly labeled HSI (56.6%), MSI (33.6%) and LSNI (9.8%). ODI recovery patterns were labeled Low Start Improvers (LSI, 37.7%), MSI (35.5%), HSI (22.7%) and High Start Non-Improvers (HSNI, 4.0%). For PCS scores, previous surgery and increased age predicted poor recovery while higher education predicted better recovery. For MCS outcomes, increased age, BMI and comorbidities predicted MSI recovery over HSI recovery. Patients receiving MISS were more likely to be in the HSI class than patients receiving Open surgery. For ODI scores, younger age and pre-operative exercise made patients significantly more likely to belong to the LSI group than the HSI group.

Conclusions:
This is a novel statistical technique that has thus far been underused in patient outcome modeling. These results highlight the factors that may be most important to a patient’s probable recovery pattern. It is interesting that MISS was associated with lasting mental health benefits. This information might help in counseling patients and surgical decision-making. Future research should aim to reproduce these results with a greater patient sample and to determine if patterns differ between pathologies.