Skip to main content
. 2020 May 9;13(3):247–263. doi: 10.1007/s12178-020-09630-2

Table 3.

Predictive models informing risk assessment tools and statistical summaries

Authors (year) No. of cases Variables considered for model Variables in final model C-index AUC/ROC R2 Positive predictive value (PPV) Validation cohort Model (application) available
Asher et al. (2017) [27] 4694 NR Age; gender; race; smoking status; employment; occupation type; past surgery; number of spinal levels; ASA grade; BMI; diabetes; osteoporosis; CAD; education level; anxiety; depression; dominant symptom; motor deficit; symptom duration; principal spinal diagnosis; workers’ compensation; liability insurance for disability; insurance type; ambulatory status; lumbar fusion, anterior, posterior or combined approach; preoperative ODI, EQ-5D, and NRS-BP/LP scores 0.71 (good) NR NR NR BS http:// statcomp2.vanderbilt.edu:37212/app_0/
Devin et al. (2018) [29] 4689 Age; gender; ethnicity; insurance status; education level; occupation level; occupation intensity; employment type; workers’ compensation status; smoking status; diabetes mellitus; CAD; PVD; anxiety; depression; osteoporosis; median BMI; ASA grade; symptom duration; predominant symptom; ambulation; presence of motor deficit; presence of cervical instability; pseudarthrosis; adjacent-segment disease; ACDF; disc herniation; foraminal stenosis; central stenosis; minimally invasive surgery; number of levels fused; NDI; VAS; NASS satisfaction score at 3 months Age; full time vs part time; education level; employment; occupation intensity; workers’ compensation; baseline NDI; symptom duration; predominant symptom; levels fused 0.81 (excellent) NR NR NR BS Nomogram created
Genov et al. (2018) [28] 291,272 Age; gender; BMI; education Level; smoking; pain at rest and in motion; pain intensity, duration and frequency; analgesic use; VAS; situational and Personal Anxiety Index; Beck Depression Index; Tampa Scale of Kinesiophobia; Pain Catastrophizing Scale; pain threshold and pain tolerance Mild dynamic pain on 1st postoperative day: motion pain before surgery (VAS); expected pain (VAS); pain threshold; gender; **Constant chronic pain syndrome 5–7 months after surgery: age; motion pain on the 1st postoperative day (VAS); **Constant NR NR NR

1st-postop day pain: PPV: 70%, 95% CI (63–76)

CPS 5–7 months after surgery PPV: 65%; 95% CI (59–71)

BS Formula given; MS Excel
Hegarty and Shorten (2012) [24] 53 Gender; age; VAS; PPI; McGill Word; RMF; Pain Catastrophizing Scale; Anxiety score; Depression score Age, RMF, PPI 0.658 NR NR NR BS NR
Khor et al. (2018) [5] 1965 NR Age; sex; race; insurance; ASA score; smoking status; diagnoses; prior surgery status; prescription opioid use; asthma; and baseline PRO scores 0.66–0.79

ODI: 0.66

NRS-BP: 0.79

NRS-LP: 0.69

NR NR 15% of sample https://becertain.shinyapps.io/lumbar_fusion_calculator
Lee et al. (2016) [31] 376 ED-5Q domains: mobility (M); self-care (S); usual activities (A); pain/discomfort (P); anxiety/depression (D) Regression equation for ED-5Q subcategories to determine risk: S + A + 2xP + D > = 9 NR NR NR NR Positive predictions 86% with cutoff value = 9 S + A + 2xP + D > = 9 domains from ED-5Q
Lubelski et al. (2018) [3] 952 Age; gender; race; marital status; BMI; smoking status; income; comorbidities; presenting symptoms; symptom duration; type of surgery; indications for surgery; total number of operated spinal levels 30-day ED visit: race; BMI; zip code; median income; surgical approach; spondylosis; total number of operated levels 30-day readmission: race; BMI; diagnosis of disc herniation or myelopathy; surgical approach; number of levels; terminal level; history of cancer 90-day reoperation: BMI; presenting symptoms; presence of disc herniation; surgical approach; presence of diabetes; change ED-5Q: herniation; myelopathy; surgical approach; diabetes mellitus; preoperative EQ-5D; preoperative PDQ change PHQ-9: age; herniation; surgical approach; diabetes mellitus; preoperative PHQ-9 change in PDQ: BMI; smoking status; spondylosis; surgical approach; preoperative EQ-5D

*3-day ED visit: 0.63

*30-day readmission: 0.78

*90-day Reoperation: 0.91

NR

EQ-5D: 0.43

PHQ-9: 0.35

PDQ: 0.47

NR BS http://riskcalc.org/PatientsEligibleforCervicalSpineSurgery/
McGirt (2017) [6] 7618 Age; gender; BMI; race; education level; history of surgery; smoking status; comorbid conditions; ASA grade; symptom duration; predominant presenting symptom; workers’ compensation; liability insurance; insurance status; ambulatory ability; baseline PROs; surgery-specific variables: number of levels, need for arthrodesis and surgical approach Age; gender; BMI; race; education level; history of prior surgery; smoking status; comorbid conditions; ASA grade; symptom duration; indication for surgery; number of levels surgically treated; history of fusion surgery; surgical approach; receipt of workers’ compensation; liability insurance; insurance status; ambulatory ability; baseline PROMs

Disability: 0.69

QOL: 0.69

Back pain: 0.67

Leg pain: 0.64

NR NR NR NR Not functional http://statcomp2.vanderbilt.edu:37212/app_0/
McGirt (2015) [23] 1803 Age; gender; race; BMI; employment; symptom duration; prior surgeries; narcotic use; smoking status; insurance; acute or chronic; neurologic symptoms; past medical history; baseline PRO scores; fracture; deformity; spinal diagnosis Age; gender; baseline ODI; prior surgery; narcotic use (days); private insurance; number of levels involved; ASA grade ≥ 2; hypertension; arthritis; diabetes; BMI; MSPQ; fusion NR

0.72–0.84 for 5 predictive models

0.82 for unplanned postoperative outcome

Complications: 0.82*

Readmission: 0.79*

Inpatient rehabilitation: 0.84*

Return to work: 0.83 Unplanned outcome: 0.78*

12-month ODI: 0.51

Validated—12-month ODI: 0.47

NR 20% of cohort NR
Merali et al. (2019) [26]

6-month: 605

12-month: 583

24-month: 539

Patient demographics; clinical presentation; surgical treatment; detailed medical co-morbidities

6 months: 41 features

12 months: 108 features

24 months: 101 features

NR

SF-6D

6 months: 0.71

12 months: 0.70

24 months: 0.73

mJOA

6 months: 0.73

12 months: 0.73

24 months: 0.67

NR

SF-6D

6 months: 0.90

12 months: 0.98

24 months: 0.92 mJOA

6 months: 0.82

12 months: 0.91

24 months: 0.96

70% training/validation,30% testing dataset NR
Sharma et al. (2019) [30] 191 Age; sex; race; BMI; diagnosis; EQ-5D score; history of spinal trauma; history of spinal surgery; operative approach (anterior vs posterior); number of spinal levels fused; comorbid dyslipidemia, obesity, hypertension, diabetes mellitus, tobacco use disorder Pre-op EDQ score, sex, pre-op diagnosis (degenerative, idiopathic, iatrogenic), previous spinal surgery history, obesity, sex-by-obesity interaction term

0.739

BS sample = .764

0.89 NR NR BS Nomogram provided in paper
Siccoli et al. (2019) [25] 635 Baseline PROs; sex; age; height; weight; BMI; smoking status; regular alcohol intake; ASA grade; prior surgery at index level; single or multi-level decompression; index levels; present of concomitant spondylolisthesis; predominant presenting symptoms (claudication or radiculopathy) NR NR

6 weeks:

ODI: 0.71

NRS-BP: 0.75

NRS-LP: 0.87

*Reop overall: 0.63

*Reop at level: 0.70

*Prolonged surgery: 0.61

*Extended stay: 0.77 12 months:

ODI:0.73

NRS-BP: 0.72

NRS-LP: 0.75

*Reop overall: 0.66

*Reop at level: 0.61

*Prolonged surgery: 0.54

*Extended stay: 0.58

NR

6 weeks:

ODI: 0.73

NRS-BP: 0.84

NRS-LP: 0.79

12 months:

ODI: 0.80

NRS-BP: 0.76

NRS-LP: 0.72

BS; 70% training, 30% testing NR
Staartjes et al. (2018) [32] 422 Age; gender; height (cm); weight (kg); BMI; smoking status; alcohol consumption; recreational drug use; ASA grade; prior discectomy at the index level; location of herniation; classification as a far-lateral or sequestered herniation; classification as broad bulging disc without annular rupture compressing nerve roots; presence of stenosis or spondylolisthesis at index level; baseline PROMs—ODI, NRS-BP, NRS-LP Age; BMI; gender: ASA score; smoking status; prior discectomy at index level; sequestered disc herniation; NRS leg pain; NRS back pain; ODI; predicted likelihood of achieving MCID in leg pain, back pain, and functional disability NR

NRS-LP = .87 (deep learning), 0.78 (regression)

NRS-BP = .90 (deep learning), 0.55 (regression) ODI = 0.84 (deep learning), 0.72 (regression)

NR NR Training/validation/testing cohorts of 60%/20%/20% of data NR

NR not reported, ACDF anterior cervical discectomy and fusion, ASA American Society of Anesthesiologists, BS bootstrapping, BMI body mass index, CAD coronary artery disease, ODI Oswestry Disability Index, NDI Neck Disability Index, ED-5D EuroQOL-5 Dimension Questionnaire, PRO patient-reported outcome, PVD peripheral vascular disease, MCID minimal clinically important difference, NASS North American Spine Society, NRS Numeric Rating Scale, BP back pain, LP leg pain, VAS Visual Analog Scale, QOL quality of life, PDQ Pain/Disability Questionnaire, PHQ-9 Patient Health Questionnaire-9, PPI persistent pain intensity, Reop reoperation, RMF Roland-Morris Functional score

*Statistic reported as a component of a given risk assessment model but does not fit inclusion criteria and therefore is not discussed in this review

**Constant value included in model