INVESTIGATING AND PREDICTING EARLY LUMBAR SPINE SURGERY OUTCOMES

Abstract

Objective

To examine short-term changes in patients’ clinical status following Lumbar Spine Surgery (LSS), and to explore presurgical variables that predict surgical outcomes.

Methods

Prospective cohort study. A total of 46 patients underwent LSS. Patients completed following questionnaires one week before LSS and two weeks following discharge from hospital: back and leg visual pain analogue scale, Ronald Morris questionnaire (RMQ), Modified Somatic Perception questionnaire (MSPQ), SF-36, Fear-Avoidance Beliefs Questionnaire, Beck’s Depression Inventory, EuroQol questionnaire, and patient-perception of improvement. Regression models were constructed to examine predictors of pain, function, quality of life, and patient-perception of improvement at 2-weeks postsurgery.

Results

Patients demonstrated significant improvement in back and leg pain and function. MSPQ and symptom duration were significant predictors of back pain, while type of diagnosis and use of opioids were significant predictors of leg pain. Baseline MSPQ and RMQ were significant predictors of postoperative RMQ. MSPQ, gender, and back pain were significant predictors of quality of life. Back pain, leg pain, depression, smoking, and worker’s compensation were significantly associated with patient-perception of improvement.

Conclusions

This preliminary study could be viewed as a directory to identify potential risk factors for unfavorable outcomes at early stages following LSS.

INTRODUCTION

Spinal disorders are one of the most common health problems globally, causing significant impact on individuals, community, and health care system.^{1, 2} Lumbar Spine Surgery (LSS) is usually considered a treatment of certain spinal disorders when conservative management fails. Lumbar spinal stenosis is the most common reason for spinal surgery.³ Spinal stenosis is a narrowing of the central vertebral canal or lateral foramina affecting one or multiple levels of lumbar vertebra/e, either one side or both sides of spinal foramina.⁴ Lumbar disc herniation, prolapse, protrusion, or extrusion accounts for less than 5% of all low back problems, but are one of the most common causes of nerve root pain resulting in surgical interventions.⁵ Spondylolisthesis, defined as the anterior shifting of one vertebra in comparison to adjacent vertebrae, is another condition that may require surgery if conservative approaches fail to resolve symptoms of displacement, and if neural compression is manifested.

Several studies have suggested that the gain from surgery is limited^6–18 in spite of LSS being reported as more efficient treatment option than nonsurgical treatment,.^19–21 possibly due to the complexity and variability of spinal pathology and variation in surgery selection criteria.. Therefore, more unified selection criteria should be considered for spinal surgeries to optimize surgical outcomes that may be identified by studying multiple possible predators influencing clinical outcomes.

Several factors have been used to measure the success of surgery such as postsurgical pain, function/disability, return to work, and quality of life (QOL).^{6–11, 22} In addition, patient-perception is important to define success from the patients perspective regarding improvement of symptoms and achievement of their goals.^{10, 23} Because clinical outcomes are multidimensional, a comprehensive assessment of clinical outcomes after LSS could lead to a more accurate assessment.

Studies have shown positive benefits in optimizing the outcomes of spinal surgeries when outpatient rehabilitation following LSS is implemented.^24,25 Patients who underwent active rehabilitation following LSS had better improvement in pain and function in comparison to usual care.^{26, 27} However, prescription of outpatient rehabilitation services following LSS is not a standard care and is based on surgeons’ preference and practice pattern.^{28, 29} This discrepancy could be related to the variability in patient recovery. Detecting early LSS outcomes and identifying presurgical factors that predict early LSS outcomes would be an asset to surgeons to guide appropriate rehabilitation services following LSS to optimize surgery outcomes.

Past studies have primarily investigated presurgical variables to predict surgical outcomes with follow up periods from six weeks to 10 years. Studies assessing long-term surgical outcomes are useful in defining the natural course of symptoms and long-term effects of surgeries. However, predicting outcomes soon after surgery may be beneficial in determining direct effects of surgery without possible confounding factors, patients’ early recovery that may affect hospital readmission or long-term outcomes and need for early intervention and possible outpatient rehabilitation to gain further improvement. Understanding presurgical variables and surgery effects during acute recovery phase may also assist clinicians and caregivers in providing reasonable expectations and identify patients who may experience early difficulty after surgery.. No study to date has investigated the postsurgical outcomes as early as two weeks following hospital discharge.

The aims of this pilot study were to (i) examine the short-term change in patients’ clinical status after LSS and (ii) explore possible presurgical predictors of surgery outcomes as early as two weeks following lumbar diskectomy, laminectomy, or fusion. We hypothesized that patients will demonstrate significant reduction in back pain, leg pain, medication usage, and increase in function during this acute recovery period. Secondly, we hypothesized that sociodemographic and presurgical clinical and psychological variables will be significant predictors of postsurgical pain, physical function, QOL, and patients’ perception of improvement.

METHODS

This is an exploratory, prospective, and observational cohort study conducted at the University of Kansas Hospital. Patients undergoing LSS were prospectively assessed with self-reported questionnaires at one week before the surgery and two weeks following hospital discharge.

Study Cohort

Consecutive patients with spinal disorders, scheduled for spinal surgery, were identified by a neurosurgeon or an orthopedic surgeon for study participation and were approached in person during their outpatient visit approximately one week before the date of surgery. Subjects were eligible for inclusion in this study if they were 18 years of age or older, were candidates for one of the following three commonly performed posterior lumbar surgeries: diskectomy, laminectomy or fusion, and could understand written and spoken English. They were excluded if they had spinal tumor or infection, spine trauma that caused movement limitation, significant head trauma, psychiatric disorders or severe memory problems reported by the patients, or any neurological disease that affected their cognition and/or movement. After meeting the inclusion and exclusion criteria, subjects were consented for the study. The study was approved by the University of Kansas Medical Center Human Subjects Committee.

Procedure

Medical records were reviewed using a data extraction form to standardize the procedure. Medical records reviewers were not involved in the clinical assessment or the treatment of the patients. The data collected from the medical records were: age, gender, race, marital status, body mass index (BMI), worker’s compensation, physical work load (unemployed, light physical workload, or heavy physical workload), smoking history (smoker, nonsmoker), alcohol use (yes, no), comorbidities, previous LSS, preoperative surgeon diagnosis (disc disorders, stenosis, or spondylolisthesis), symptom’s duration, preoperative analgesics used, and proposed operative procedure.

At baseline, patients were to complete the self-reported questionnaires:

Visual Analogue Scale (VAS)³⁰ is used to measure pain intensity on a 10-cm horizontal line, anchored by the verbal descriptors “no pain” (score = 0) and "pain as bad as it could be (score = 10). Patients were asked to rate their back and leg pain averaged separately over the last week.

Modified Somatic Perception Questionnaire (MSPQ)³¹ is a13-item scale designed to measure somatic complaints and autonomic perception in patients with back pain.³² Patients rate how much they have been bothered by each symptom/ item on a 4-point Likert scale (0=not at all, 3= extremely bothered). Higher scores demonstrate greater somatic complaints.

Roland-Morris Questionnaire (RMQ)³³: is a disability scale with 24-items. Patients select a yes or no response for each item. Higher scores reflect higher disability.

Beck’s Depression Inventory (BDI)³⁴ is a 21-item scale to measure attitudes and symptoms of depression. Patients rate how applicable each item relates to their case, starting with 0 as unrelated to and 3 as totally related. Higher scores indicate greater depressive symptoms or attitudes.

Short Form (36) Health Survey (SF-36)³⁵ has two major components: physical component score (PCS) with questions related to physical health and Mental Component Score (MCS) with questions related to mental health. Higher scores indicate better general health and QOL.

Fear-Avoidance Beliefs Questionnaire (FABQ)³⁶ measures fear of movement (5-items) and work (11-items) related to low back pain. Patients indicate their response on 0 to 6 Likert scale, 0= completely disagrees, 3=unsure, and 6=completely agree. Higher scores reflect higher level of fear avoidance.

Two weeks after hospital discharge, patients rated their back and leg pain on VAS scale, and completed RMQ. In addition they completed the following questionnaires:

European Quality of Life Scale (EuroQol-5D) (© 1990 EuroQol Group. EQ-5D™)^{37, 38} has five generic health status items: Mobility, Self-Care, Pain/Discomfort, Usual activities, and Anxiety/ Depression; each item scored on a 3-point response scale: no problem, moderate problem, or severe problem.^{37, 38} The scores from five items are then converted to 0.0–1.0 index. A higher index value indicates better QOL.

Patient-perception of improvement: In addition, subjects rated their perceived improvement of back pain, leg pain, function, and general health using a 7-point Likert scale 1=very much improved and 7=very much worse. The scores were then dichotomized into better or worse (unchanged or feeling worse).

Statistical analysis

IBM SPSS Statistics 20.0 (SPSS Inc. Chicago, IL) was used for all statistical analyses. To examine the changes from presurgery to postsurgery, we used a paired t-test for continuous variables and the McNemar test for categorical variables. We used analysis of variance and chi-squared test to compare baseline difference in the 3 types of surgeries.

We initially used univariate regression analysis to identify potential presurgical predictors for each primary outcome of pain, function, QOL and general health. Variables meeting significant level, 0.15, were considered as possible predictors in final regression models. We then used a series of multivariate linear regression to construct predictor models for back pain, leg pain, RMQ score, and EuroQol-5D index. Controlled stepwise regression was used to control for presurgical pain in the back and leg pain models and presurgical RMQ in the functional model. We used multivariable logistic regression (enter method) to predict patient-perception of improvement (improvement or no improvement) in back pain, leg pain, function, and general health.

RESULTS

A total of 106 patients were qualified for the study. Eight patients did not proceed with surgery for various reasons. 38 patients did not complete the baseline questionnaire and other 14 did not complete the follow up questionnaire. A total of 46 patients completed the study (43%). Testing the baseline difference between the patients who completed the study and patients who did not complete the study showed no significant differences in age (p=.41), BMI (p=.10), symptoms duration (p=.91), lower back pain intensity (p=.79), leg pain intensity (p=.80), gender (p=.55), type of diagnosis (p=.41), and number of patients using opioids (p=.32). The demographic and baseline characteristics for patients who completed the study are summarized in table 1.

Table 1.

Patient baseline characteristics

Variable	Mean ±SD or proportion %
Age	59.20 (12.46)
Sex (Women)	56.50%
Race
White or Caucasian	96%
Black or African American	4%
Body mass index (kg/m2)	33.38 (7.14)
Marital status
Married	63%
Divorced/Widowed/separated	22%
Single	15%
Worker’s compensation (yes)	9%
Work load
Unemployed/light load	65.2%
Heavy load	34.8%
Current low back pain duration (months)	21.20 (19.89)
Type of surgery
Discectomy	15.2%
Laminectomy	39.1%
Fusion	45.7%
Previous back surgery (Yes)	34.80%
Diagnosis
Disc disorders	31.1%
Stenosis	53.3%
Spondylolisthesis	15.6%
Use of opioid
No	47.8 %
Yes	52.2%
Smoking (yes)	21.7%
Visual Analogue Scale
Back pain	6.26 (2.83)
Leg pain	6.02 (2.53)
Length of stay	3.06 (2.07)
Back pain description
Dull	47.8%
Sharp	23.9%
Both	28.3%
Ronald Morris Questionnaire	14.68 (5.39)
Fear Avoidance total score	37.74 (18.55)
Modified Somatic Perception Scale	5.89 (4.75)
Beck’s Depression Inventory	10.03 (8.34)
SF-36 physical component	26.87 (8.12)
SF-36 mental component	48.92 (12.23)

Subjects baseline differences according to the type of surgery are described in table 2. There was no significant baseline difference in gender, BMI, pain level, and number of spinal levels operated between the three types of surgeries. Conversely, patients who received diskectomy were significantly younger, and patients who had discectomy and laminectomy had significantly less length of stay than fusion (p<.001)

Table 2.

The baseline differences in subjects who received different types of surgery

		Discectomy	Laminectomy	Fusion	p
Age (mean (SD))		45.86 (12.17)	64.61 (11.21)	59.20 (12.46)	.002*
BMI (mean (SD))		32.86 (6.79)	34.19 (6.33)	32.86 (8.12)	.833*
Pain (mean (SD))		4.71 (3.73)	5.78 (2.84)	7.19 (2.18)	.081*
Levels operated (n)	• 1	5	8	8	.086†
	• 2	2	7	6
	• 3	0	3	7
	• Total	7	18	21
Gender (female %)		43%	56%	62%	.393†
Length of stay		1.86 (1.35)	2.59 (1.23)	4.90 (2.00)	P<.001*

^*ANOVA,

^†Chi square test

In comparison to presurgery levels, there was statistically significant improvement (Table 3) in back pain (p<0.001), leg pain (p<0.001), and function (p<0.001) at two weeks following LSS. The change in type of analgesic used by patients and the amount of opioid consumed was not significant following LSS. Of the 46 patients, 35 reported improvement in back pain, 39 in leg pain, 35 in function, and 34 in general health, while the remaining subjects reported either no change or worsening of symptoms.

Table 3.

Changes in back and leg pain intensity, and function, somatic and type of medications used from presurgery to postsurgery

	Presurgery mean (SEM) or proportion%	Postsurgery mean (SEM) or proportion%	Mean difference (95% CI)	p
Back pain	6.26 (.42)	2.74 (.32)	3.52 (2.69 to 4.35)	<.001*
Leg pain	6.02 (.37)	2.15 (.36)	3.87 (2.89 to 4.85)	<.001*
Ronald Morris	14.68 (.81)	11.98 (.98)	2.71 (1.09 to 4.33)	.002*
Opioid use	No 47.8%	No 43.5%		.774†
	Yes 52.2%	Yes 56.5%
Opioid Dose	1.40 (2.05)	2.28 (2.84)	.87 (−.11 to 1.85)	.08*

SEM: Standard Error of Mean,

^*Paired-t-test,

^†McNemar test, p<0

Back Pain and Leg Pain Models

Significant predictors of postoperative back pain were MSPQ score (p=.042) and symptom duration (p=.020) with preoperative back pain intensity controlled in regression model (Table 4). These results showed that for a one-point increase in MSPQ score, postoperative back pain is expected to increase by 0.14 points on VAS scale (95%CI=.01 to .27). When the duration of symptoms is increased by one month, back pain is likely to increase by .04 points (95%CI=.01 to .06). The model predicted 41.5% of the variation in postoperative back pain.

Table 4.

Linear regression models for back pain, leg pain, function and quality of life at 2 weeks postsurgery

Model	Predictor	Β (95% CI)	S.E	Change in R square	p
Back Pain model	Constant	1.01(−.90 to 2.93)	.94		.289
	Back VAS	.14 (−.09 to .37)	.11	.089	.216
	MSPQ	.14 (.01 to .27)	.07	.172	.042
	Symptoms’ duration	.04 (.01 to .06)	.01	.223	.020
	Total			.484
Leg pain model	Constant	.70 (−1.37 to 2.78)	1.02		.494
	Leg VAS	.13 (−.17 to .42)	.15	.001	.359
	Diagnosis*	.62 (−.36 to 1.59)	.55	.141	.007
	Opioid use†	.78 (−.51 to 2.07)	.73	.12	.042
	Total			.256
Functional model (RMQ)	Constant	2.26 (−2.7 to 7.19)	2.42		.357
	RMQ	.52 (.13 to .90)	.19	.396	.010
	MSPQ	.48 (.02 to .93)	.23	.075	.040
	Total			.471
Quality of Life model (EuroQol- 5D)	Constant	.85 (.80 to .88)	.02		<.001
	MSPQ	−.01 (−.02 to −.01)	.01	.341	<.001
	Gender*	−.07 (−.13 to −.02)	.03	.085	.007
	Back pain	−.01 (−.02 to .00)	.01	.072	.044
	Total			.498

VAS: Visual Analogue Scale, MSPQ: Modified Somatic Perception Questionnaire, RMQ: Ronald Morris Questionnaire, EuroQol-5D: European Quality of Life Scale,

^*Reference= Spondylolisthesis,

^†Reference= Yes,

^‡Reference = female, p<0.05

Diagnosis and preoperative use of opioids were the only significant predictors for postoperative leg pain (p=.007 and 042, respectively) in the model with preoperative leg pain intensity controlled (Table 4). Patients with a diagnosis of spondylolisthesis are likely to have .62 points (95%CI=−.360 to 1.59) higher leg pain on VAS scale. Patients with preoperative use of opioids are likely to have more leg pain by .78 points (95%CI= − .51 to 2.07). The model explained 25.6% of the variation in postoperative leg pain.

Functional Model

Baseline RMQ score was a significant predictor (p=.010) of postoperative RMQ and explained 39.6% of the variation in the model (Table 4). For every one point increase in preoperative RMQ scores, the expected increase in postoperative RMQ was .52 (95%CI=.13 to .90). Preoperative MSPQ was also a significant predictor (p=.040). An increase in one point on MSPQ scale is likely to increase RMQ score following surgery by.48 points (95%CI=.02 to .93).

Quality of Life Model

MSPQ score, female gender, and preoperative back pain intensity were significant predictors of the EuroQol-5D index (p<.001, p=.007, p=.044 respectively) (Table 4). When the MSPQ score is increased by one score, the EuroQol-5D index is likely to decrease by 0.011 units (95%CI= −.02 to −.01). Female patients have 0.07-unit lower (95%CI= −.13 to −.02) EuroQol-5D index than males. Finally, when preoperative back pain is increased by one point on VAS scale, the EuroQol-5D index is likely to decrease by 0.01 units (95%CI= −.02 to −.00). The model explained 49.8% of the variation in EuroQol-5D index.

Patients’ Perception of Improvement

Patients with more preoperative back pain (odd ratio (95%CI) =5.36 (1.45 to 19.83), p= .012), less preoperative leg pain (odd ratio (95%CI) = .183 (.04 to .76), p= .019), and high preoperative depression scores (BDI) (odd ratio (95%CI) = .76 (.59 to .97), p= .026) were more likely to have a negative impression of back pain improvement (Table 5). Smokers (odd ratio (95%CI) =.19 (.04 to .97), p=.046) were more likely to have a negative perception of improvement in function and in general health status (odd ratio (95%CI) =.34 (.06 to 1.82), p=.035) as compared to non-smokers (Table 5). Patients with a higher BMI (odd ratio (95%CI) =1.134 (.99 to 1.30), p=.041) and who received worker’s compensation (odd ratio (95%CI) =.16 (.01 to 2.00), p=.035) were more likely to have a negative perception of their general health improvement (Table 5). There was no significant predictor for perception of improvement in leg pain.

Table 5.

Logistic regression model for patients’ perception of improvement in back pain, leg pain, function and general health at 2 weeks postsurgery

Model	Predictor	Β	Odd ratio (95% CI)	S.E	p
Back pain improvement	Back VAS	1.68	5.36 (1.45 to 19.83)	.67	.012
	Leg VAS	−1.70	.18 (.05 to .76)	.73	.019
	BDI	.28	.76 (.59 to .97)	.13	.026
	Constant	3.62	37.34	2.44	.139
Leg pain improvement	Smoker*	−1.61	.20 (.05 to 2.77)	1.34	.230
	Symptom’s duration	−.07	.93 (.86 to 1.01)	.04	.091
	Back pain	−1.15	.32 (.08 to 1.30)	.72	.111
	Leg pain	.13	1.14 (.75 to 1.74)	.22	.547
	Constant	12.20	3.85	1.97	.080
Function improvement	Back pain	−.25	.79 (.55 to 1.13)	.18	.789
	Leg pain	−.09	.93 (.66 to 1.29)	.17	.925
	Smoker*	−1.68	.19 (.04 to .97)	.56	.046
	Constant	4.04	56.35	1.58	.347
General health improvement	BMI	.20	1.22 (1.00 to 1.49)	.10	.041
	Worker’s compensation†	−5.81	.01 (.00 to .87)	2.89	.045
	Smoking*	−5.71	.01 (.00 to 1.14)	2.98	.035
	Somatic perception	.62	1.85 (.89 to 3.85)	.37	.099
	Constant	−5.63	.01	3.35	.093

^*Reference = smoker,

^†Reference = yes, p<0.05

Discussion

To our knowledge, this is the first study to examine clinical symptoms and predict clinical outcomes at two weeks post-hospital discharge following LSS. Patients reported significant improvement in back and leg pain, and function without change in opioids use. Duration of symptoms and somatic complaints predicted back pain whereas type of diagnosis and use of analgesics predicted leg pain. Somatic complaints also predicted function and QOL. Presurgical functional status highly influenced postsurgical function. Moreover, female gender and less back pain intensity predicted better QOL. Finally, patients’ perception of improvement following surgery was influenced by preoperative back and leg pain, depression, smoking, general health status, high BMI and workers’ compensation.

A previous study of LSS reported that changes in pain and function are more apparent at six weeks in comparison to six months.³⁹ Our results are in agreement with this study and showed that patients experienced significant reduction in back and leg pain, and improvement in function as early as two weeks post-hospital discharge. Back and leg pain improvements were clinically significant.⁴⁰ The type of analgesics used by patients was not significantly different from pre-surgery to post hospital discharge, suggesting that improvement in back and leg pain was independent of analgesics use.

Preoperative back pain has been shown to be a predictor of postoperative pain in numerous studies.⁴¹ The average symptoms’ duration in our sample was approximately 21 months and was a significant predictor of postoperative back pain even though back pain might be influenced by surgical incision during this period. Previous studies showed inconclusive findings: preoperative back pain has been associated with symptom duration for more than one⁴² and two years.⁴³ In contrast, Sigmundsson et al.,⁴⁴ reported no significant effect of symptom duration on postoperative back pain.

Preoperative diagnosis of spondylolisthesis and use of opioids were significant predictors of leg pain intensity. These results are in line with Sigmundsson et al.,⁴⁴ showing that patients with spondylolisthesis were more likely to have a higher reduction in leg pain following LSS, and preoperative use of analgesic predicted higher postsurgical leg pain after one year follow up.

MSPQ predicted multiple outcomes including back pain, RMQ score, and EuroQol-5D. Previous studies have also reported MSPQ to be a significant predictor of short- and long-term outcomes following LSS.^{6, 41, 45} MSPQ was a significant predictor of pain and function at one year follow-up and had the highest prediction accuracy in comparison to all other psychological tests.⁴⁶ Also, depressive symptoms predicted negative improvement in back pain. MSPQ and depression scores are often used together to identify patients with distress. This combination has shown to be a significant predictor of functional disability and pain at 6 and 12 months following LSS.⁴⁷ This combination also significantly predicts changes in back and leg pain, and functional ability after one year post LSS.⁴⁵ Distressed patients may perceive a higher level of pain and develop anxiety or negative behavior.³⁹ Therefore, distressed patients will tend to avoid functional and social activities and rate their function and QOL lower than non-distressed patients.

Preoperative back pain and female gender were significant predictors of postsurgical QOL. Brunstrom and colleagues⁴⁸ found that in general population, women rate their QOL significantly lower than men, and specifically the depression/anxiety and pain subscale indices of the EuroQol-5 D tend to be lower in women in comparison to men.

Smokers have a negative perception of improvement on health and functional status. Nicotine is known to effect early vascularization and consequently delay healing and bone graft union.^{49, 50} Trief and colleagues¹¹ reported that smoking predicted lower general health and functional status at 12 and 24 months following lumbar fusion. Other studies reported that smokers had greater use of analgesics, less walking ability, and inferior QOL after surgery.⁵¹ Our findings also showed a decrease in reported functional improvement in smokers in the early postoperative period.

Having worker’s compensation predicted negative improvement in QOL. There are many studies showing that work related variables like worker’s compensation,^{16, 45} high physical workload,^{14, 52} preoperative work status,⁵³ number of days off from the work⁵⁴, and weekly wages were predictors of LSS outcomes.⁵⁴ Our study found this to be a predictor of diminished perception of early outcome.

This is an observational pilot study and results should be validated with larger studies. Other clinical outcomes such as assessment of personality and patients’ expectation from the surgery, and objective physical examination should also be tested before surgery and included in a regression model, as these variables have shown some predictive values in other studies.^{10, 40, 44, 52, 55} Finally, different surgery selection criteria between the surgeons may influence clinical outcomes, but the selection of patient cohort from neurosurgery and orthopedic surgery clinics used herein may increase the external validity of our study.

CONCLUSION

The study suggests clinical improvement as early as two weeks post hospital discharge following LSS. Symptom duration, type of diagnosis, and preoperative use of analgesics could predict back and leg pain intensity after surgery. Females and patients who had high preoperative back pain could have inferior quality of life. Most importantly, of all variables examined, somatic perception predicted majority of outcomes (i.e. back pain, function and QOL) and could be considered as a routine assessment prior to surgery to predict post-surgery outcomes. Preoperative back and leg pain intensity, high depression symptoms, smoking, and having worker’s compensation suggest negative patient-perception of improvement after LSS.