LUMBAR SURGERY IN WORK-RELATED CHRONIC LOW BACK PAIN: CAN A CONTINUUM OF CARE ENHANCE OUTCOMES?

Abstract

Background Context

Systematic reviews of lumbar fusion outcomes in purely workers’ compensation (WC) patient populations have indicated mixed results for efficacy. Recent studies on lumbar fusions in the WC setting have reported return-to-work rates of 26–36%, re-operation rates of 22–27%, and high rates of persistent opioid use two years post-surgery. Other types of lumbar surgery in WC populations are also acknowledged to have poorer outcomes than in non-WC. The possibility of improving outcomes by employing a biopsychosocial model with a continuum of care, including post-operative functional restoration in this “at risk” population, has been suggested as a possible solution.

Purpose

To compare objective socioeconomic and patient-reported outcomes for WC patients with different lumbar surgeries followed by functional restoration, relative to matched comparison patients without surgery.

Study Design/Setting

A prospective cohort study of chronic disabling occupational lumbar spinal disorder (CDOLD) patients with WC claims treated in an interdisciplinary functional restoration program.

Patient Sample

A consecutive cohort of 564 patients with pre-rehab surgery completed an functional restoration and was divided into groups based on surgery type: lumbar fusion (F group, n = 331) and non-fusion lumbar spine surgery (NF group, n = 233). An unoperated comparison group was matched for length of disability (U group, n = 349).

Outcome Measures

Validated patient-reported measures of pain, disability and depression were administered PRE- and POST-rehab. Socioeconomic outcomes were collected via a structured one-year POST interview.

Methods

All patients completed an intensive, medically-supervised FRP, combining quantitatively-directed exercise progression with a multimodal disability management approach. The writing of this manuscript was supported in part by NIH Grant 1K05-MH-71892 and no conflicts of interest are noted among the authors.

Results

The F group had a longer length of disability compared to the NF and U groups (M = 31.6, 21.7, and 25.9 months, respectively, p < .001). There were relatively few statistically significant differences for any socioeconomically-relevant outcome among groups, with virtually identical POST-rehab return-to-work (F=81%, NF=84%, U=85%, p =.409). The groups differed significantly after surgery on diagnosis of major depressive disorder and opioid dependence disorder, as well as patient-reported depressive symptoms and pain intensity PRE-rehab. However, no significant differences in patient-reported outcomes were found POST-rehab. PRE-rehab opioid dependence disorder significantly predicted lower rates of work return and work retention, as well as higher rates of treatment-seeking behavior. Higher levels of PRE-rehab perceived disability and depressive symptoms were significant risk factors for poorer work return and retention outcomes.

Conclusions

Lumbar surgery in the WC system (particularly lumbar fusion) have the potential achieve positive outcomes that are comparable to CDOLD patients treated non-operatively. This study suggests that surgeons have the opportunity to improve lumbar surgery outcomes in the WC system, even for complex fusion CDOLD patients with multiple prior operations, if they control post-operative opioid dependence and prevent an excessive length of disability. Through early referral of patients (who fail to respond to usual post-operative care) to interdisciplinary rehabilitation, the surgeon determining this continuum of care may accelerate recovery and achieve socioeconomic outcomes of relevance to the patient and WC jurisdiction through the combination of surgery and post-operative rehabilitation.

INTRODUCTION

Lumbar spine surgery rates in the workers’ compensation (WC) setting for chronic disabling occupational lumbar disorder (CDOLD) patients have been rising steadily, especially lumbar spine fusion procedures.^{1, 2} There has also been a debate in the scientific literature about the long-term effectiveness of spinal fusion surgery for chronic (especially discogenic) low back pain.^{3, 4} Findings, such as low work return, high re-operation rates, and highly prevalent opioid dependence have been demonstrated in different WC jurisdictions,^5–7 and have led to limitations on recommendations for spinal fusion in WC claims by the largest evidence-based national guidelines, resulting in a lower surgical approval rates in a growing number of state jurisdictions mandating use of these guidelines.^{8, 9} It has been suggested that similar guidelines will be incorporated into group health and Medicare settings with the expansion of national health insurance. A large population-based study of fusions within the Washington State WC setting revealed a 22% re-operation rate and a low return-to-work rate of 36%, at 2-years post-surgery.⁷ A more recent study in the Ohio WC jurisdiction showed similar results, with a 26% return to work rate and a 27% re-operation rate, as well as a opioid dependence disorder (ODD) still noted in 85% of fusion patients 2 years postoperative.⁶ Outcomes in the Utah WC jurisdiction were comparable. ^{10, 11}

In contrast, a previous prospective cohort study demonstrated high rates of work status improvement (>80%) and low rates of re-operation and recurrent injury claims for both fusion and discectomy patients in a chronic lumbar WC cohort.¹² This particular study utilized a “continuum of care approach,” with administration of functional restoration after surgery for these patients. An important question is whether lumbar spinal surgery combined with subsequent interdisciplinary rehab can improve the results of surgery alone. Such a continuum of care protocol is now quite standard in the treatment of other musculoskeletal disorders of the extremities, especially the knee.^13–16 National treatment guidelines, such as the Official Disability Guidelines (ODG), utilize a biopsychosocial “continuum of care” model, which recommends appropriate steps and limits for sufficient preoperative care and testing, surgical decision-making, postoperative rehab and, in some chronic pain cases, interdisciplinary functional restoration.⁹ Therefore, the major goal of the present study was to take an important step in further evaluating the potential efficacy of such a combined approach with a “worst case” cohort of CLBP patients with WC claims.

Recent studies have documented the importance of psychosocial factors and their impact on outcomes following lumbar surgery,^{11, 17–20} and the impact of psychosocial factors in the development and perpetuation of chronic pain and disability has been widely documented in the literature.^{21, 22} Because lumbar fusion surgery for nonspecific chronic low back pain (CLBP) is usually performed on patients already demonstrating extensive chronic pain/disability behaviors in the WC setting, the need to pay heed to a biopsychosocial model when contemplating surgery, or managing them in post-operative rehabilitation, is becoming more apparent.^{8, 23, 24} The ODG recognizes the length of disability as being one of the most critical risk factors for poor outcomes and higher costs linked to delayed recovery.^{8, 9} Even under optimum conditions, objective outcome measures declined with increasing length of disability, making this factor a primary selection criterion for determining a “worst case” WC claim cohort.

The present study evaluated several objective socioeconomic outcomes in a prospective consecutive cohort of lumbar surgical patients with WC claims, operated on after already developing CLBP disability despite extensive non-operative care (average 16–19 months from injury to index surgery). They were all referred to an interdisciplinary functional restoration program (FRP) after demonstration of persistent post-operative disability following a trial of usual postop rehab (averaging 11–18 months postop). Patients able to successfully resume occupational activities after usual post-operative physical therapy were excluded from this study, leaving only patients demonstrating persistent pain/disability after fusion, discectomy, decompression and/or artificial disc replacement (ADR), and after failure to respond to usual postop reconditioning and interventions.

METHODS

Patients

A consecutive cohort of 3,888 patients, admitted to a functional restoration program between the years 1992 and 2003, received rehabilitation for chronic musculoskeletal disorders. All patients in this cohort had WC claims (state or federal U.S. jurisdictions), with or without attached third party claims, or non-WC disability payments (long-term disability, short-term disability, Social Security, etc.). To be eligible for this study, all patients had a claim of a lumbar injury that was recognized as compensable (i.e., accepted for treatment within the jurisdiction rules), with a specific injury date. There were 822 patients without compensable spinal injuries who were excluded from the study, resulting in a sample of 3,066 patients with compensable spinal injuries. Initial injury claims ranged from minor sprains to major traumatic events (falls from a height, crush injuries or explosions). Program participation criteria included: 1) four or more months elapsed since a work-related injury; 2) non-surgical care failed to improve symptoms sufficient to allow full return-to-work; 3) surgery had not produced resolution, or simply was not an option; 4) severe pain and functional limitations remained; and 5) patients had the ability to communicate in English or Spanish.

Of this sample, those who failed to complete the functional restoration program were excluded (n = 472) as these patients had a low rate of outcome contact and had not completed a full dose of a rehab treatment. The final cohort consisted of 2,594 program completers, divided into 3 groups as follows:

• F group: At least one lumbar fusion or artificial disc replacement (ADR) surgical procedure after injury claim for degenerative disease with indications including discography, instability or failed prior lumbar surgery, but excluding lumbar fractures, dislocations, tumors, infectious etiologies or spondylolisthesis (N = 331);

• NF group: At least one lumbar surgery (discectomy or decompression) after injury claim, but no lumbar fusion or ADR (N = 233);

• U Group: A mixed group of patients with no PRE-rehab spinal surgery (N=2,021), some of whom had PRE-rehab surgery to a non-spinal compensable body part (N=191). All patients in this group who had any type of non-spinal surgery were excluded. From the remaining patients (N=1,830), an unoperated comparison group (N=349) was established, matched to the F and NF groups on length of disability. The remaining non-surgical patients (N=1,481) were excluded from the analysis (per Figure 1 flow chart).

Figure 1.

Inclusion and Exclusion Criteria

Both surgical groups may have had other surgeries to the same or other levels prior to the WC claim under study, or may have had a second or subsequent procedure (see Table 1). The study was granted an exemption from IRB review because all data were collected from the standard medical records. Figure 1 demonstrates the mechanisms by which the groups were derived, in the CONSORT style.

Table 1.

Types of surgeries for the 2 subgroups (N = 564).

Variable	NF group N = 233	F group N = 331
Type of PRE-rehab surgery†
discectomy	135 (56.3%)	41 (8.4%)
micro-discectomy	57 (23.8%)	11 (2.3%)
percutaneous discectomy	5 (2.1%)	2 (0.4%)
spinal decompression	21 (8.8%)	7 (1.6%)
ADR/TDR	0	7 (1.4%)
IDET	20 (8.3%)	1 (0.2%)
fusion, unspecified	0	35 (7.2%)
fusion, anterior	0	125 (25.7%)
fusion, posterior interbody	0	59 (12.1%)
fusion, posterior lateral	0	43 (8.8%)
fusion, 360	0	95 (19.5%)
pseudarthrosis repair	0	12 (2.5%)
hardware removal	1 (0.4%)	27 (5.6%)
bone or spinal stimulator removal	1 (1.4%)	12 (2.5%)
other decompression+fusion	0	9 (1.6%)
Total surgeries	240 (100%)	486 (100%)

^†Percentage of total number of surgeries is reported; each patient may have had more than one surgery

Note: All patients had workers’ compensation (WC) lumbar injury claims. All NF patients had an index discectomy, decompression, while all F patients had an index lumbar fusion procedure following the WC claim. Other surgical procedures may have predated the WC injury (including other spinal levels) or followed the index procedure under the same claim, accounting for the surgical distribution in this table.

Table 2 summarizes the demographic characteristics of the U, NF, and F groups. The U group had a lower proportion of males than the F and NF groups (p < .001). The NF group was more likely to have a single spinal injury (as opposed to multiple spinal injuries or a combination of spinal and extremity injuries), compared to both the F and the U groups (p < .001). Finally, the F group was more likely to have had multiple PRE-rehabilitation surgeries (p < .001), and had a longer time from surgery to rehab (p < .001), relative to the NF group. In claiming this is a “worst case” cohort, the authors wish to emphasize the extremely long period of time (involving partial and/or total disability) from injury to index surgery (averaging 16–19 months), in addition to 11–18 months between surgery and functional restoration admission, with the average total disability time ranging from 22–32 months for all three groups. Table 1 presents the types of spine surgery and their frequency under the WC claim being treated for the NF and F groups.

Table 2.

Demographics for the 3 subgroups (N = 913)

Variable	Unoperated matched comparison group (U) N = 349	Non-fusion spinal surgery (NF) N = 239	Spinal fusion surgery (F) N = 325	p-value	effect size
Age
mean (SD)	43.5 (9.9)	44.5 (9.6)	44.3 (8.7)	.404
Gender
n (% male)	198 (56.7%)	176 (75.5%)d	225 (68%) d	<.001	w = 0.16a
Race
n (%)				.801
Caucasian	236 (69%)	155 (68%)	219 (67%)
African-American	38 (11.1%)	34 (14.9%)	43 (13.1%)
Hispanic/Latino	65 (19%)	38 (16.7%)	61 (18.7%)
Other	3 (0.9%)	0 (0.4%)	4 (1.2%)
Job demand
n (%)				.072
Sedentary-light	41 (13%)	25 (10.7%)	27 (8.2%)
Light-medium	88 (27.9%)	48 (20.6%)	76 (23.2%)
Medium-heavy	120 (38.1%)	91 (39.1%)	136 (41.5%)
Heavy-very heavy	66 (21%)	69 (29.6 %)	89 (27.1%)
Length of disability (months)
mean (SD)	25.9 (21.2)	21.7 (23.1)	31.6 (23.4) d,e	< .001	ηp2 = 0.03a
Time from injury to surgery (months)
mean (SD)	N/A	19.1 (61.8)	16.3 (19.8)	.474
Time from surgery to rehab (months)
mean (SD)	N/A	10.8 (19.7)	17.6 (18.2)e	< .001	ηp2 = 0.03a
Injured spinal region
n (%)				< .001	w = 0.18a
Single spinal	217 (62.2%)e	192 (82.4%)d	233 (70.4%)d,e
Multiple spinal	59 (16.9%)e	25 (10.7%)d	42 (12.7%)
Spinal + extremity	73 (20.9%)e	16 (6.9%)d	56 (16.9%)e

Variable	Unoperated matched comparison group (U) N = 349	Non-fusion spinal surgery (NF) N = 233	Spinal fusion surgery (F) N = 331	p-value	effect size
Number of PRE-rehab surgeries
n (%)				< .001	w = 0.25a
none	349 (100%)	0 (0%)	0 (0%)
single	N/A	194 (83.3%)	199 (60.1%)e
multiple	N/A	39 (16.7%)	132 (39.9%)e
Order of fusion surgery
n (%)	N/A	N/A
primary surgery			222 (67.1%)
second surgery			26 (7.9%)
third or later surgery			11 (3.3%)
multiple fusion surgeries			35 (10.6%)

^asmall effect size;

^bmedium effect size;

^clarge effect size

^ddifferent from no surgery matched comparison group;

^edifferent from non-fusion surgery group

Procedure

The functional restoration program was a medically-supervised interdisciplinary program, combining a quantitatively-directed exercise progression protocol with a multimodal disability management approach. Patients admitted to the program first underwent a comprehensive physical and psychosocial evaluation, including functional and psychosocial assessment. Patients then participated in a structured exercise program, administered by physical and occupational therapists, which was guided by repeated objective physical measurements. The disability management program included: cognitive-behavioral therapy designed to promote pain coping skills; medical case management to facilitate vocational reintegration; biofeedback and relaxation training to assist stress management, and educational sessions to improve knowledge of musculoskeletal disorders and to encourage health and fitness maintenance after program discharge. The program was medically-supervised to allow for simultaneous medication management (including psychotropic medications and opioid taper), as well as limited interventional procedures (if needed). The specific program descriptions of its procedures has been published in multiple outcome studies documenting its effectiveness in providing functional improvement, often cited in national treatment guideline references.^{8, 9, 25–32} The effectiveness of functional restoration methods have also been well documented in the literature.^33–41

Measures

Patients completed a battery of psychosocial patient-reported outcome measures at admission to, and discharge from, the functional restoration program. In addition, the Structured Clinical Interview based on the Diagnostic and Statistical Manual of Mental Disorders (SCID) was administered by a trained clinician upon program admission, and patients were evaluated for Axis I psychiatric disorders.^{42, 43} For the purposes of this study, the major Axis I psychiatric disorders occurring with a high prevalence in a chronic pain population were reported: major depressive disorder; anxiety disorders; and substance use disorders.⁴⁴ Psychosocial patient-report measures were administered at program admission (PRE) and discharge (POST). Measures included the Beck Depression Inventory (BDI), a measure of depressive symptoms.⁴⁵ BDI scores of 20 or greater have been found to indicate moderate to severe depressive symptoms.⁴⁶ The Million Visual Analog Scale (MVAS) was used to measure patient-perceived disability,⁴⁷ and scores of 70 or greater indicate moderate to severe disability.⁴⁸ Pain was measured with the pain intensity visual analog scale (VAS), a 10 cm line with endpoints of 0=no pain and 10=the worst possible pain. Scores on the pain VAS of 6 or greater indicate severe pain and are associated with poorer rehabilitation outcomes.^{48, 49} All patient-reported measures have been shown to be reliable and valid for use with chronic pain patients.⁵⁰

One-year POST-rehabilitation socioeconomic outcomes were collected using a structured telephone interview.^{51, 52} These one-year socioeconomic outcomes were obtained with a 94% successful contact rate for this cohort, and included:

• return-to-work ( i.e., whether the patient returned to work at any point in the post-discharge year);

• work retention (whether the patient was still employed at the one-year follow-up);

• percentage seeking treatment from a new healthcare provider;

• number of visits to a new healthcare provider;

• new surgeries to original area of injury since program discharge;

• new compensable injury claims (with or without work loss);

• and workers’ compensation case settlement.

These socioeconomically-relevant outcomes have been consistently reported from this program in the past, and have been shown to be reliable discriminant indicators of patients who complete the program, relative to those who refuse treatment or do not complete it.^{25, 26, 53, 54}

Source of Funding

The writing of this manuscript was supported in part by Grant 1K05-MH-71892 from the National Institutes of Health, which focuses on evidence-based assessment and treatment of musculoskeletal pain and the monitoring of valid outcomes.

Data Analyses

Statistical analyses of differences among the three groups were performed using the Chi-Square statistic for categorical variables, and one-way ANOVAs for continuous variables. All significant p-values reported were complemented by the reporting of the Cohen’s w effect size for categorical variables,⁵⁵ and the partial eta-squared effect size statistic for continuous variables. The POST psychosocial patient-reported measures were evaluated using a ANCOVA procedure, with PRE-rehab scores as a covariate . For the one-year outcome variables, multivariate logistic regression analyses were used to assess possible risk factors for adverse outcomes across the entire cohort. In addition to lumbar fusion surgery, predictor variables in these regression analyses included any substantial psychosocial comorbidity and relevant demographic differences between the groups.

RESULTS

PRE- and POST-Rehabilitation Psychosocial Characteristics

Patients in the NF group had lower PRE-rehabilitation mean levels of depressive symptoms as measured by the BDI, compared to the F group (15.9 vs. 19.1, p = .002). In addition, patients in the NF groups were less likely to meet the diagnostic criteria for major depressive disorder compared to the F and the U groups (43.4% vs. 59% and 56.5%, respectively, p =.002). Patients in the F group had a significantly higher prevalence of post-injury opioid dependence disorder relative to the NF group and the U group (31.3% vs. 19.4% and 17.3%, respectively, p < .001). Upon completion of the program, both groups were comparable on all psychosocial measures. PRE and POST psychosocial characteristics of the cohort are summarized in Table 3.

Table 3.

Psychosocial testing results (N = 913).

Measure	U group N = 349	NF group N = 233	F group N = 331	p-value	effect size
Opioid dependence disorder n(%)	48 (17.3%)	38 (19.4%)	94 (31.1%)d,e	<.001	w = 0.15a
Major depressive disorder n(%)	157 (56.5%)	85 (43.4%)d	178 (58.9%)e	.002	w = 0.13a
Anxiety disorder n(%)	26 (9.4%)	9 (4.6%)	31 (10.3%)	.070
Beck Depression Inventory (BDI)
PRE-rehab	17.3 (10.8)	15.9 (9.9)	19.1 (10.7)e	.002	ηp2 = .014a
POST-rehabb	10.1 (7.8)	10.0 (7.8)	9.9 (7.8)	.955
Mean (SD)
Million Visual Analog Scale (MVAS)
PRE-rehab	95.2 (23.7)	91.9 (25.5)	93.4 (25.4)	.306
POST-rehabb	64.4 (27.7)	68.2 (27.7)	65.8 (27.7)	.292
Mean (SD)
Pain intensity visual analog scale
PRE-rehab	6.8 (1.9)	6.4 (2.2)	6.5 (2.0)	.023	ηp2 = .008a
POST-rehabb	4.6 (2.1)	4.9 (2.1)	4.8 (2.1)	.201
Mean (SD)

^asmall effect size;

^bNote: POST-rehab scores are adjusted for PRE-rehab scores;

^cdifferent from no surgery matched comparison group;

^ddifferent from non-fusion surgery group

One-Year Socioeconomic Outcomes

Table 4 summarizes the one-year POST-rehabilitation objective socioeconomic outcomes of the three groups. There were no significant differences among the groups in work-related outcomes, including both return-to-work and work retention. This finding provided contrast to much lower work status outcomes in prior WC spine surgery publications.^6–8 In addition, there were no differences in rates of additional surgery (1.2–3.7%) or new compensable injuries. These non-significant differences among groups may hide the fact that the additional surgery rates for the surgical groups ranged from 3.3–3.7%, well below fusion re-operation rates of 22–27% in recent lumbar fusion WC claim studies. Because a minority of patients may have had a non-lumbar compensable injury associated with their lumbar claim, some of these surgeries may have been to the cervical spine or extremities, further lowering the actual re-operation rate from what is historically found in this CDOLD cohort. Recurrent WC injury claims in the year following treatment completion at 1.4–2.4% are also well below historical rates of more than 30–60% after recovery from WC injuries.^56,57 The F group was more likely to seek treatment from a new provider after discharge from rehabilitation, relative to the U and NF groups (34.4% vs. 21.7% and 25.6% respectively, p = .001), with more medical visits (p = .032). Analysis of groups seeking healthcare from new providers following the FRP over 3 decades has identified opioid dependence disorder and opioid-seeking as the primary reason given by such patients, consistent with the higher rates of opioid dependence for the F group.^6,58

Table 4.

One year socioeconomic outcomes (N = 856)

Outcome	Unoperated matched comparison group (U) N = 349	Non-fusion spinal surgery (NF) N = 233	Spinal fusion surgery (F) N = 331	p	Effect size
Return to work n (%)	271 (85.2%)	177 (83.5%)	242 (81.2%)	.409
Work retention n (%)	256 (81%)	162 (76.8%)	223 (74.8%)	.172
New surgery n (%)	4 (1.2%)	8 (3.7%)	10 (3.3%)	.135
New injuries n (%)	6 (1.9%)	3 (1.4%)	7 (2.4%)	.751
Case settlement n (%)	319 (96.4%)	206 (93.2%)	292 (94.2%)	.222
Treatment seeking from new providers n (%)	72 (21.7%)	56 (25.6%)	105 (34.4%)d,e	.001	w = 012a
Number of visits to new providers				.032	w = 0.16a
none	260 (79%)	163 (74.4%)	200 (66.2%)d,e
1–5 visits	38 (11.6%)	30 (13.7%)	46 (15.2%)
6–10 visits	13 (4%)	13 (5.9%)	23 (7.6%)
more than 10 visits	18 (5.5%)	13 (5.9%)	33 (10.9%)d,e
n (%)

Note: Contact rate for program completers was 94%.

^asmall effect size;

^bmedium effect size;

^clarge effect size

^ddifferent from no surgery matched comparison group;

^edifferent from non-fusion surgery group

Risk Factors for Adverse Outcomes

Multivariate logistic regression analyses were conducted across the entire cohort of patients. The results of this analysis are shown in Table 5, with all non-significant predictors removed from the model. Patients in the F group were 1.94 times more likely to seek treatment from a new provider after rehabilitation than the U group, after controlling for gender and opioid dependence. The most robust risk factors for poorer outcomes was the presence of post-injury opioid dependence disorder, which predicted lower rates of return-to-work and work retention, as well as higher rates of treatment-seeking behavior at the one-year follow-up interview. In addition, PRE depressive symptoms on the BDI were predictive of both return-to-work and work retention, while PRE perceived disability on the MVAS predicted work retention, and female gender predicted treatment-seeking behavior.

Table 5.

Logistic regression analyses for predicting one-year outcomes (N = 913)

Socioeconomic outcome	Predictor variables	B (SE)	OR	95% CI
Failure to return to work
	Opioid dependence	0.76 (0.22)	2.13**	[1.37, 3.31]
	PRE-rehab BDI ≥ 20	0.54 (0.21)	1.72**	[1.14, 2.60]
	PRE-rehab surgery group
	U group†
	NF group	0.11 (0.28)	1.12	[0.64, 1.95]
	F group	0.22 (0.25)	1.24	[0.77, 2.01]
Failure to retain work
	Opioid dependence	0.66 (0.21)	1.94**	[1.30, 2.90]
	PRE-rehab BDI ≥20	0.45 (0.19)	1.57*	[1.07, 2.30]
	PRE-rehab MVAS ≥ 85	0.65 (0.24)	1.92**	[1.21, 3.07]
	PRE-rehab surgery group
	U group†
	NF group	0.33 (0.25)	1.38	[0.84, 2.28]
	F group	0.34 (0.22)	1.41	[0.91, 2.18]
Treatment seeking
	Opioid dependence	0.45 (0.19)	1.56*	[1.07, 2.28]
	Gender (female)	0.60 (0.18)	1.82**	[1.29, 2.59]
	PRE-rehab surgery group
	U group†
	NF group	0.14 (0.23)	1.15	[0.73, 1.82]
	F group	0.66 (0.20)	1.94**	[1.31, 2.87]

Note: The original regression models included the following covariates: gender, job demand, length of disability, type of injury, opioid dependence disorder, major depressive disorder, PRE-rehab BDI, and PRE-rehab pain intensity. The models reported in the Table include only significant predictors of the one-year outcomes.

^*p < .05,

^**p < .01,

^†reference category

DISCUSSION

Most previous studies of lumbar surgery in CLBP (particularly lumbar fusion) have concentrated on comparisons between the surgery and a non-operative comparison group.^59–62 However, few studies focus on (or even include) WC claim patients. With the exception of two studies,^{41, 59} no other previous studies have looked at a population of extreme lengths of disability after WC claims receiving one or more fusions or other spinal procedure in a continuum of care (i.e., surgery followed by active rehabilitation). Under current WC national treatment guidelines, a continuum of care is contemplated, beginning with a surgical decision made only after early non-operative care has had an adequate trial.^{8, 9} When usual postop reconditioning therapy and/or injections fails to result in recovery to pre-injury function, a more intensive interdisciplinary program, such as a functional restoration program (or chronic pain management), is recommended.^8,9 No prior study has evaluated such a continuum of care in comparison to patients treated non-operatively with similar extreme lengths of disability prior to interdisciplinary rehabilitation.

In demonstrating no statistically significant differences in most socioeconomic outcomes of relevance to the WC system, very important differences were shown between these surgical cohorts and comparisons in the spine literature of lumbar surgery in the WC system whose surgeons or other providers offered only “usual,” more limited post-operative care. Work return and retention rates in these fusion and non-fusion groups were substantially higher than the 26–36% reported 2 years post-surgery in recent studies of large WC cohorts from Washington and Ohio State jurisdictions.^{6, 7, 10} While there may have been differences in state jurisdiction rules and economies during the study period of 1992–2003, those differences cannot account for a 2–3 fold increase in the percentage returning to work and retaining work at the end of a year, particularly for this cohort demonstrating extreme periods of disability. On average, the 1-year outcomes were taken more than 2 years after the index spine surgical procedure, making this cohort study results comparable to that from other state jurisdictions.

Similar unusual findings were noted in the re-operation rates. The studies from Washington, Utah and Ohio demonstrate 22–27% re-operation rates for lumbar fusion cases.^{6, 7,10} In this present study, prior to the opportunity to participate in interdisciplinary rehabilitation 16.7% of the non-fusion patients had a second or subsequent surgical procedure, while 39.9% of the fusion patients had a second or subsequent surgery, with 17.2% of fusion patients having 3 or more surgeries PRE-rehab. In the year following the functional restoration program, however, only 3.3% of fusion, and 3.7% of non-fusion, patients had an additional surgical procedure. In addition, for those returning to work after the rehabilitation program, an unusually low rate of new WC injury claims was found for the post-surgical patient population. In contrast to many prior studies, only 1.4% of non-fusion patients and 2.4% of fusion patients reported a new WC injury claim from their job.⁵⁷ These results are comparable to the 1.9% new injury claims among patients in the non-surgical comparison group, who were matched for duration of disability.

The increasing trend in the prescription of opioids has generated research into long-term negative effects associated with opioid use.⁶⁰ Two recent publications have documented that the use of opioids within the WC setting, for both acute and chronic back pain, is significantly associated with increased risk for work disability at one-year.^{44, 61–63} While surgical patients from this patient population have been identified as having increased rates of post-rehabilitation healthcare utilization,⁶⁴ the increased treatment seeking by patients may be in the service of maintaining their opioid dependence.⁶³ Further support for this hypothesis comes from the fact that data on more recent 2010–11 outcomes has seen the treatment seeking from new healthcare providers decrease 60–80% into the single digits (as measured on a quarterly basis), possibly due to more effective pain medication management and patient-negotiated opioid tapering.^58,63,65

Selection bias is a frequent criticism of prospective cohort studies, such as the present one. For example, it is quite possible that the fusion group patients might have had the same outcomes if they entered the functional restoration program before or without surgery. In a sense, the U groups speaks to that, because the objective work and reoperation outcomes were identical for that group and the surgical groups with an equivalent length of disability. We have no way of knowing how many of the U group had similar surgical indications as the NF or F groups, which makes conclusions about cause and effect relationships difficult. What is striking is that all the groups did about the same despite an extremely long period of disability. It remains for future studies to establish a cause and effect relationship, but it is clear from this study that neither fusion or other types of lumbar surgery necessarily lead to suboptimal outcomes. It should also be noted that the elapsed time between injury and admission to rehabilitation was an indication that this is a “worst case” cohort of fusion and non-fusion CDOLD patients, who were referred only if they exhibited persistent disability over a significant period of time. This is demonstrated by the extreme length of disability for all groups, as well as the majority of surgical patients having an elapsed time of eleven to eighteen months between surgery and intensive rehabilitation admission. This time, was not controlled between the groups, this contributing to another potential source of bias. It is noteworthy that the development and promulgation of national treatment guidelines are predicated on a biopsychosocial continuum of care for their recommendations, with patients “dropping out” of the guideline as they recover with the least intensive treatments (that usually begin with acute non-operative care, move to surgical and postop care, and ending in pain management for only a small group who fail to recover). Although no selection bias was evident for demographic and occupational factors in terms of program completion, it should be noted that the efficacy of this rehabilitation program has been previously demonstrated by a comparison of program completers versus non-completers.^53,54 For example, program non-completers were 10 times less likely to return to work, 7 times less likely to retain work, and 7 times more likely to have re-operations, relative to program completers. In terms of the concept of a “continuum of care,” these findings generally reflect that program dropouts have chosen to pursue disability benefits offered under state WC jurisdictions or federal benefits like Social Security Disability. By having patients identify themselves as disability-prone on a biopsychosocial basis through persistent pain behaviors, limited effort and non-compliance, non-completion also provides late evidence to surgeons and state administrators of the need to consider case closure rather than persistent interventions.

CONCLUSIONS

Overall, lumbar surgery, including fusion, was not a risk factor for adverse outcomes, including critical outcomes such as return-to-work and work retention one-year post-functional restoration program, when a continuum of care that included interdisciplinary functional restoration was offered to patients and completed. Poorer outcome rates across the cohort were more accurately predicted by several known risk factors for persistent chronic pain and disability, especially the presence of post-injury opioid dependence disorder. Controversy has existed for decades over the poorer outcomes in the WC claim setting for lumbar spine surgery, particularly fusions. The problem with most prior studies is that they rely on the surgery as a stand-alone entity, without considering other aspects of a continuum of care. This study, following a biopsychosocial model, and evidence-based recommendations of national treatment guidelines (currently specific to the WC claim setting), demonstrated that a continuum of care produced surgery outcomes that were as positive as those involving unoperated-matched patients. Once a surgical decision has been made and acted upon, it is possible for the surgeon to influence the postoperative rehabilitation process to achieve similar outcomes, rather than leaving those decisions to others. As national health insurance and Medicare/Medicaid cost-overruns grab more headlines, poor objective outcomes for ongoing disability and healthcare utilization are likely to adversely impact the spine surgeon’s options. Fortunately, this manuscript suggests that a surgeon’s attention to a continuum of care, that includes attention to the patient’s compliance, opioid use, and cooperation with intensive rehabilitation designed to assist vocational reintegration, can dramatically impact outcomes of socioeconomic relevance, not simply patient-reported outcomes. Although the availability of quality functional restoration program remains a limiting factor on testing the hypothesis raised in this manuscript and in national guidelines, sufficient quality programs are available to further test the findings raised in this preliminary study, to provide greater evidence-based support for the continuum of care concept. In the future, such attention by surgeons may also impact national evidence-based guidelines currently restricted to WC claims, but soon likely to impact the vastly larger group health and Medicare populations falling under the supervision of federally-administered and influenced healthcare programs.