Prevalence of classification methods for patients with lumbar impairments using the McKenzie syndromes, pain pattern, manipulation, and stabilization clinical prediction rules

Abstract

Objectives

Aims were (1) to determine the proportion of patients with lumbar impairments who could be classified at intake by McKenzie syndromes (McK) and pain pattern classification (PPCs) using Mechanical Diagnosis and Therapy (MDT) assessment methods, manipulation, and stabilization clinical prediction rules (CPRs) and (2) for each Man CPR or Stab CPR category, determine classification prevalence rates using McK and PPC.

Methods

Eight physical therapists practicing in eight diverse clinical settings classified patients typically referred to rehabilitation by McKenzie syndromes (i.e. derangement, dysfunction, posture, or other), pain pattern classification [i.e. centralization (CEN), not centralization (Non CEN), and not classified (NC)], Manipulation CPR (positive, negative), and stabilization CPR (positive, negative). Prevalence rates with 95% confidence intervals (CI) were calculated for each classification category by McK, PPC, and manipulation and stabilization CPRs. Prevalence rates (95% CIs) for McK and PPC were calculated for each CPR category separately.

Results

Data from 628 adults [mean age: 52±17 years, 56% female] were analyzed. Prevalence rates were: McK — derangement 67%, dysfunction 5%, posture 0%, other 28%; PPC — CEN 43%, Non CEN 39%, NC 18%; manipulation CPR — positive 13%; Stab CPR — positive 7%. For patients positive for manipulation CPR (n = 79), prevalence rates for derangement were 89% and CEN 68%. For patients positive for stabilization CPR (n = 41), prevalence rates for derangement were 83% and CEN 80%.

Discussion

The majority of patients classified based on initial clinical presentation by manipulation and stabilization CPRs were also classified as derangements whose symptoms centralized. Manipulation and stabilization CPRs may not represent a mutually exclusive treatment subgroup but may include patients who can be initially treated using a different classification method.

Introduction

Low back pain is the second most common reason for patients seeking primary care services and is responsible for substantial economic burden exceeding 10 billion (US dollars) annually.^1,2 Because of the high prevalence and economic impact associated with low back pain, classifying patients with non-specific low back pain into homogeneous treatment subgroups to help direct treatment decisions and improve prognosis, quality of care, and patient outcomes has been recognized as an important research and clinical priority.^3,4

Two common classification approaches utilized by physical therapists and researchers are the McKenzie, or Mechanical Diagnosis and Therapy (MDT)⁵ and the Delitto or Treatment-Based Classification (TBC)⁶ systems. The MDT method is a standardized assessment including the patient’s subjective and medical history and a physical examination. One primary purpose for examining patients following McKenzie assessment is to classify patients into one of three mechanical syndromes, i.e. derangement, dysfunction, or posture based on a series of repeated end-range lumbar movement tests or positioning techniques.⁵ Repeated movement testing is guided by the patient’s symptomatic and mechanical responses observed during the examination process. Classification directs an individualized treatment program. The MDT classification system was developed to be inclusive of the majority of patients with mechanical low back pain syndromes who are referred to physical therapy regardless of age and neurological or symptom acuity status. Previous research indicates that a high proportion of patients can be classified into one of the three MDT syndromes.^7–9 A smaller percentage of patients (2.8–32.9%) who do not fit into the three mechanical syndromes are classified as ‘Other’.⁸

An important clinical symptom observed during the MDT examination process is centralization. Centralization is characterized by spinal pain and referred spinal symptoms that are progressively abolished in a distal-to-proximal direction in response to therapeutic movement and positioning strategies.^5,10 Although the validity for classifying patients into one of the three MDT mechanical subgroups has not been determined, numerous research studies support CEN as an important prognostic factor and classification category for identifying patients who respond favorably from specific CEN-based interventions.^10–20 Classification by patient response criteria including CEN and directional preference for directing treatment have been frequently recommended by clinicians and researchers.^{11,13–15,21–29} Similar to the three McKenzie syndromes, a substantial proportion of patients (estimates range between 31 and 87%) seen in physical therapy clinics for treatment of low back pain can be classified into a CEN category at intake.¹⁰

The TBC system is another common approach used by physical therapists for classifying patients based on a standardized physical evaluation process.⁶ Patients are classified into three stages based on condition severity, i.e., stage 1 or the acute stage where therapy goal is symptomatic relief; stage 2 or the subacute stage where return to normal function is a priority; and stage 3 or the advanced rehabilitation stage for patients requiring high physical demands and conditioning to perform their usual work or activities of daily living.⁶ Stage 1 of the TBC system, which identifies four basic treatment subgroups, i.e. manipulation, exercise, stabilization, and traction, using specific clinical signs and symptoms, has been extensively researched and supported in the literature.^14,30,31 The original TBC classification criteria have been recently updated and evidence supports the application of clinical prediction rules (CPRs) to classify patients into the TBC manipulation and stabilization subgroups.^14,32–35

CPRs are sophisticated probabilistic and prognostic models where a group of identified patient characteristics and clinical signs and symptoms are statistically associated with meaningful prediction of patient outcomes. Two separate CPRs were developed by researchers for identifying patients who would respond favorably to manipulation.^33,34 Flynn et al. developed the original manipulation CPR using five criteria, i.e. no symptoms below the knee, recent onset of symptoms (<16 days), low fear avoidance belief questionnaire³⁶ score for work (<19), hypomobility of the lumbar spine, and hip internal rotation ROM (>35° for at least one hip).³³ Flynn’s CPR was subsequently modified by Fritz et al. to two criteria, that included no symptoms below the knee and recent onset of symptoms (<16 days), as a pragmatic alternative to reduce clinician burden for identifying patients in primary care most likely to positively respond to thrust manipulation.³⁴ The percentage of patients classified according to the pragmatic manipulation CPR varied between 29 and 48% in the samples studied.^21,34,37 The stabilization CPR was developed using four classification criteria, which included younger age <40 years, positive prone instability test, positive aberrant trunk movements, and average straight leg ROM>91°, and initially derived by Hicks et al.³⁵ The authors reported that the stabilization CPR could be used to determine whether patients with low back pain are likely to favorably benefit from stabilization exercises.³⁵ The percentage of patients classified into the stabilization category was reported by Brennan et al. as 24%.²¹

Preliminary efficacy evidence exists for both MDT and TBC classification systems^30,38 as well as classifying patients based on CPR and CEN for treating patients with low back pain.^{11,16,21,32,35} However, there are no published studies examining outcomes between patients classified according to McKenzie syndromes, PPC, and CPR categories, which may partially explain why there is a lack of agreement between clinicians and researchers on which classification treatment method works best for which patients. Of interest, there are two case studies comparing MDT versus manipulation CPR classification for a patient with low back pain.^39,40 Both case studies identified patients who met four out of five criteria on the CPR for spinal manipulation as proposed by Flynn et al,³³ and were successfully treated by specific lumbar flexion⁴⁰ or extension exercises,³⁹ matched to a directional preference or CEN following a standard MDT assessment. Both case studies suggest that the CPR may not identify a unique or mutually exclusive treatment category for thrust manipulation. It appears plausible that some patients classified into one treatment category may also meet criteria for another treatment subgroup and benefit from either treatment or a combination of both treatments. At this time, it is unclear what proportions of patients belong to more than one Treatment-Based Classification subgroup.

Because of the lack of data comparing patient classification methods, the overall aim was to begin comparisons of common methods currently used to classify patients during the initial examination with non-specific low back pain typically referred to outpatient physical therapy settings. Specific aims were to: (1) determine the proportion of patients who could be classified by McKenzie syndromes and pain pattern classification using MDT assessment methods and clinical prediction rules for manipulation and stabilization; and (2) within each manipulation and stabilization CPR category, determine classification prevalence rates for McKenzie syndromes and pain pattern categories (PPCs). Study results may allow physical therapists to better assess the generalizability of common classification methods currently used for managing a wide range of patients seeking treatment in diverse physical therapy outpatient settings for low back pain complaints.

Methods

Design

We conducted a prospective, longitudinal, observational, cohort study. We analyzed data collected from patients with non-specific lumbar syndromes who were classified and treated by eight physical therapists (PTs) working at eight different clinical facilities. All clinicians that participated also used the Focus On Therapeutic Outcomes, Inc. (FOTO) (Knoxville, TN, USA) international medical rehabilitation data management company outcomes system.^41,42 The FOTO Institutional Review Board for the Protection of Human Subjects approved the project. The study did not include any change in clinical practice and retrospective data were analyzed; thus patient informed consent was not required for the analyses of data collected during normal clinical practice.

Procedures

Clinicians

Eight physical therapists (mean age: 42 years, range: 31–59 years; 100% males) participated. All therapists received postgraduate MDT training and achieved either diploma or credentialed educational levels. Practice settings were diverse: three PTs worked in hospital-based orthopedic outpatient clinics, four PTs were in private practice, and one PT worked in two military orthopedic outpatient clinic settings. Two therapists worked in the same practice, and one therapist worked in the military and moved from one to another military clinic during data collection. Four clinicians earned master’s degree in physical therapy, one PT earned a doctorate degree in science, and one PT earned a doctorate of physical therapy. The average number of years of clinical experience was 15 years (range: 8–40 years). Not all therapists collected data during the entire study period (July 2007–December 2009); three therapists started data collection in the summer of 2009 and four therapists were either transferred between clinics or had other non-patient educational responsibilities which interrupted their data collection. Therapists were experienced with classifying patients into McKenzie syndromes and pain pattern and CPR classification categories.

Subjects

Of the 725 consecutive patients with lumbar syndromes who were treated, 33 did not start data collection, producing a sample of 692 patients and a participation rate of 95%.⁴³ Reasons for not starting data collection included computer system was down (n = 9), cognitive deficits (n = 5), language barriers (n = 8), visual deficits (n = 2), one visit, patient referred for home program only (n = 3), and no reason given (n = 6). Of the 692 patients, 64 patients did not have classification data (Table 1).

Table 1. Patient characteristics (n = 692).

Characteristic	No classification data (n = 64)	Complete classification data (n = 628)
Age (years)	50 (18), 18, 86	52 (17), 18, 91
Gender (%)*
Male	30	44
Female	70	56
Missing	0	0
Symptom acuity (%)
Acute	20	20
Subacute	25	25
Chronic	55	54
Missing	0	0
Surgical history (%)
None	80	83
One or more	20	17
Missing	0	0
Fear of physical activities (%)
Not elevated	55	71
Elevated	19	27
Missing	26	2
Number comorbid conditions (%)
None	13	12
One	23	18
Two or three	28	32
Four or more	36	38
Missing	0	0
Payer (%)*
Litigation	0	1
Medicaid	5	1
Medicare Part B	23	22
Patient private pay	0	4
HMO	19	25
PPO	38	19
Workers’ compensation	8	2
Other	3	16
Missing	6	6
Intake functional status	51 (15), 5, 94	52 (13), 5, 96
Intake pain	6 (2), 2, 10	6 (2), 0, 10

Note: Continuous data: mean (standard deviation), minimum, maximum.

*χ² significant (P<0.05).

Patient classification methods at intake

McKenzie syndromes

Therapists classified patients at intake into one of three McKenzie syndromes (i.e. derangement, dysfunction, or posture) based on symptomatic or mechanical responses observed during repeated end-range lumbar movement testing as demonstrated by McKenzie and May.⁵ If the patient’s examination was mechanically inconclusive or did not fit the criteria for any of the three syndromes, the patient was classified into ‘Other’ category. Stenosis, surgery, and chronic pain syndrome are examples of patients classified into ‘Other’ category. Inter-tester reliability for identifying the three main McKenzie syndromes by qualified examiners attaining credentialed level of MDT training is substantial (K = 0.6–0.7).^44,45

Pain pattern subgroups

Patients were classified into three PPCs at intake: centralization (CEN), non-centralization (Non CEN) and not able to be classified (NC), which have been recommended^10,13 and operationally described.^20,29 Briefly, for PPC, patients were classified into CEN, Non CEN, or NC categories at intake by measuring changes in pain location observed during a standard MDT physical examination without consideration of symptom intensity. A body diagram and overlay template were used to quantify changes in pain location.^20,29 The patient was instructed by the examiner to shade in all areas on the body diagram where she or he currently was experiencing spinal pains and referred symptoms. Body diagrams were completed in standing before and after end-range repeated trunk movements and or positioning techniques. The overlay template was placed over the body diagram, which allowed quantification of the anatomical location of pain. Inter-rater reliability for documenting CEN and Non CEN identified during the physical examination using body diagrams/measurement template classification procedure has been reported to be almost perfect (K = 0.96–1.0).²⁹

Manipulation subgroup

Patients were classified at intake into a manipulation category using two criteria: duration of symptoms <16 days and no symptoms below the knee.³⁴ The two criteria manipulation CPR was recommended by Fritz et al. as an alternative or pragmatic application of the original Manipulation CPR developed by Flynn et al.^33,34 The original CPR was determined by four or five out of five positive findings as recommended by Flynn et al. (i.e. duration <16 days, no symptoms below the knee, Fear Avoidance Belief Questionnaire Work subscale³⁶ <19/42 points, at least one hypomobile lumbar segment, and at least one hip>35°).³³ For this study, we applied only the pragmatic manipulation CPR and modified the duration to <21 days because of our study’s standardized procedure for documenting acuity levels.²⁰ The original validation of the Man CPR included patients in the manipulation group with a median duration of symptoms at 22 days.³² The pragmatic manipulation CPR maintains sufficient accuracy compared to the five criteria CPR for identifying patients who respond favorably to manipulation and has been recommended by others for reliably classifying patients with low back pain.^21,34,37,46

Stabilization subgroup

Patients were classified at intake into the stabilization subgroup by three or four out of four criteria developed by Hicks et al.: (1) age <40 years old; (2) average left and right SLR >91°; (3) positive aberrant trunk movement; and (4) positive prone instability test.³⁵ Although inter-rater reliability for each stabilization CPR criterion was reported to range between poor and substantial (K = 0.16–0.87),^35,47 inter-rater reliability for classifying patients using a proposed classification algorithm including stabilization CPR category was moderate (K = 0.60).⁴⁶

Data analyses

We calculated differences between patients who had missing classification data and those with complete classification data (Table 1) by using chi-square tests of independence (dichotomous and categorical data) or two-sample t-tests (continuous data) using available independent variables.

We calculated prevalence rates two ways: first, prevalence rates were calculated for each classification category (i.e. derangement, dysfunction, posture, other, CEN, Non-CEN, NC, manipulation CPR, and stabilization CPR); second, prevalence rates for McKenzie syndromes and PPCs were calculated for each CPR category separately. Ninety-five percent confidence intervals (CI) were calculated for each prevalence rate.

Results

Missing classification data at intake

Characteristics of the 628 patients with complete classification data are displayed in Table 1. Compared to patients with complete classification data (n = 628), patients with no classification data (n = 64) tended to have more females (χ² = 5.1, df = 1, P = 0.02), tended to receive benefits from Preferred Provider Organizations or workers’ compensation, and tended to receive fewer benefits from ‘other’ methods of payment (χ² = 30.1, df = 9, P<0.01). The groups were not different by level of symptom acuity (P = 0.99), number of functional comorbid conditions (P = 0.79), level of fear (P = 0.81), number of surgeries (P = 0.51), age (P = 1.00), intake functional status (P = 1.00), and intake pain (P = 1.00).

Prevalence

Prevalence rates (95% CI) were: McKenzie syndromes — derangement 0.67 (0.63, 0.70), dysfunction 0.05 (0.03, 0.06), posture 0.002 (−0.002, 0.005), other 0.28 (0.25, 0.32); PPCs — CEN 0.43 (0.39, 0.47), Non CEN 0.39 (0.35, 0.43), Not classified 0.18 (0.15, 0.21); manipulation CPR — positive 0.13 (0.10, 0.15); stabilization CPR — positive 0.07 (0.05, 0.08). For patients positive for manipulation CPR (n = 79), the prevalence rates for derangement and CEN were 0.89 (0.82, 0.96) and 0.68 (0.58, 0.79), respectively. For patients positive for stabilization CPR (n = 41), the prevalence rates for derangement and CEN were 0.83 (0.71, 0.94) and 0.80 (0.68, 0.93).

Discussion

The primary findings of the study suggest that (1) a greater proportion of patients with non-specific low back pain syndromes seeking outpatient physical therapy services can be classified based on initial clinical presentation by McKenzie syndromes and by PPCs compared to classifying patients according to TBC manipulation and stabilization categories using clinical prediction rules; and (2) the majority of patients classified by manipulation and stabilization clinical prediction rules were also classified as McKenzie derangements whose symptoms centralized.

Classification of these patients is an important research priority,⁴ and validation of classification systems is a multi-step process including testing for reliability and validity.⁴⁸ Data support the reliability for classifying patients according to McKenzie syndromes,^7,44,45 pain patterns,^29,44 and clinical prediction rule criteria.^33,46,47 Preliminary efficacy data provide evidence for classifying and treating patients with non-specific low back pain according to the MDT approach,^16,38,49,50 CPR,^21,32,35 and centralization criteria.¹¹ In addition to reliability and validity, the generalizability of a classification system to guide treatment decision making must also be supported.⁴⁸

For a classification system to be clinically useful, the system must enhance patient outcomes as well as account for or include a substantial proportion of potential patients referred to rehabilitation.^9,51 For example, two recent independent studies found that McKenzie syndromes were frequently diagnosed by physical therapists for diverse and large patient populations referred to multiple clinical settings with non-specific spinal pain complaints.^8,9 Hefford reported that of 321 patients, assessed by 34 physical therapists trained in MDT methods, 92% were classified into one of the three McKenzie syndromes.⁸ Similarly, May found that of 601 patients with spine pain, assessed by 57 therapists trained in MDT in 18 countries, 83% were classified into the three McKenzie syndromes and 17% were considered non-mechanical and classified as ‘Other’ category.⁹ Seventy-two percent of the patients in our study were classified into McKenzie syndromes, which is consistent with previous published data. In addition, a large proportion of our patients (43%) were classified into a centralization pain pattern, which is also similar to previous data examining centralization prevalence rates using a standard measurement procedure to document centralization.⁵²

The proportions of patients classified by manipulation and stabilization CPRs in our patient sample were low at 13 and 7%, respectively. The low overall CPR subgroup prevalence rates in our study may be partially explained by differences in our sample’s patient case-mix and practice settings compared to the CPR derivation studies.^33,35 For instance, our sample was older (mean age: 52 years old versus mean age: 37 years old reported by Flynn et al.³³ and mean 42 years old in the study by Hicks et al.³⁵) and more chronic, i.e. 54% >3 months (versus primarily acute and subacute patient populations.^33–35 In addition, our patient sample was recruited from diverse outpatient practice settings compared to military facilities in manipulation CPR studies.^32,33 Another reason that may explain the low CPR prevalence rates found in our study was that we calculated the proportion of patients categorized into manipulation and stabilization subgroups based on the total number of patients referred for physical therapy assessment and treatment compared to previously published CPR percentage rates, which were calculated based on eligible and consenting patients.^21,32 For example, Brennan et al. reported that a total of 1052 potential patients were referred for treatment to all participating clinics, of which 268 were eligible and of those, 123 subjects provided consent.²¹ The authors reported that 48 and 24% of eligible and consenting patients were classified into the manipulation and stabilization categories respectively. When these manipulation and stabilization classification percentages reported by Brennan et al.²¹ were recalculated based on all potential patients referred for physical therapy services, the actual proportions of patients grouped into the Man and Stab categories were 6 and 3%, respectively, which is more consistent with our prevalence data.

The patient characteristics in our study appear similar to FOTO’s national normative database for lumbar patients⁵³ for age, acuity, number of medical comorbidities, payer type, fear avoidance beliefs of physical activity, and gender (data not shown but available on request) and therefore, our sample appears representative of typical patients seeking physical therapy services in large metropolitan areas. Our results indicate that although CPRs for thrust manipulation and stabilization exercises may be appropriate and effective for young and acute patients with lumbar syndromes identified in randomized trials, these CPRs appear of limited utility for a majority (approximately 90%) of patients with non-specific low back pain syndromes referred to diverse outpatient therapy clinics. In comparison, diagnosing patients according to McKenzie syndromes and pain patterns appear more generalizable, as assessed using prevalence, in the sample tested.

The majority of patients in our study classified by manipulation and stabilization CPR categories were also classified as McKenzie derangements whose symptoms centralized. This contrasts with claims by others that TBC manipulation and stabilization groups will not include patients experiencing centralization of symptoms.^14,15,21 For example, Fritz et al. developed a classification treatment algorithm including specific exercise, manipulation, and stabilization to guide initial treatment delivery based on intake clinical signs and symptoms within each subgroup.¹⁴ The algorithm uses a hierarchical decision making process where specific exercise based on centralization is considered first followed by manipulation criteria and finally stabilization.^14,21 If centralization was not observed, then by definition following the algorithm’s classification decision tree, it would not be possible to detect centralization in the other categories. The reason why centralization was observed in the majority of patients classified by manipulation and stabilization CPR in our study may be partially explained by differences in the physical examination approach to judge centralization. The therapists in our study applied MDT methods and identified centralization by repeated end-range lumbar movements using a variety of positional strategies based on the patient’s symptomatic and mechanical responses.⁵ The exact number of repeated movement tests and testing positions were not predetermined, as is the case with other centralization classification methods. For instance in the TBC system, centralization may be judged using a minimum of one or two to a maximum of 10 lumbar movements in predefined positions (i.e. lumbar active range of motion and repeated extension in standing, sustained prone extension, and repeated flexion with the patient seated),²¹ which may not accurately assess CEN.⁵⁴ In addition, for the subject to be classified into CEN in our study, centralizing symptoms had to remain better at the end of the initial evaluation.^20,29

Similar to the classification algorithm proposed by Fritz et al. and Brennan et al., setting the priority of centralization also has been recommended by other clinicians and researchers including McKenzie and May,⁵ Petersen et al.,⁵⁵ and Laslett.⁵⁶ Centralization is a key or cornerstone concept to many published and current classification algorithms; therefore, research examining and standardizing testing procedures to judge CEN may be beneficial to improve the decision making treatment process and clinical utility of current classification systems. For instance, the clinical interpretation of manipulation and stabilization CPR subgroups in our study may be more relevant for patients whose symptoms do not centralize during the initial examination.

Finally, the manipulation and stabilization CPR categories as proposed by Fritz et al., Flynn et al., Hicks et al., and others^{14,21,32–35} may not represent discrete treatment subgroups but may include patients who can be initially treated in alternative ways. It is clear from our data that a substantial proportion of patients belong to more than one of these treatment categories, i.e. 80% of patients who met the criteria for stabilization exercises also met the criteria for prescribing specific exercises based on centralization. It is possible that these patients may benefit from just specific exercise or stabilization or from a combination of both treatments. Although our study design cannot answer these questions or recommend one treatment as superior to another, our findings indicate that current classification categories do not appear to be mutually exclusive when classification methods are applied to diverse patient populations with lumbar impairments seeking treatment in a variety of outpatient settings. Clinicians may be best served by future research that emphasizes collaboration between system developers and clinicians for examining the strengths, weaknesses, limitations, and advantages of current classification algorithms with the ultimate goal of identifying the right treatment for the right patient.

Limitations

The value of a classification system should be determined by its ability to direct treatment and enhance patient outcomes. The aim of our study was simply to assess the proportion of patients with lumbar impairment who could be classified using common classification methods currently used by clinicians during the initial examination process. Comparative effectiveness between classification systems requires future research. In addition, our high prevalence rates reported for McKenzie syndromes and centralization may not be found among clinicians not specifically trained in MDT or clinicians who do not objectively judge patient response classification criteria using precise and standardized operational definitions and physical examination testing methods applied in this study. Finally, the prevalence rates for the manipulation subgroup may have been different if we applied Flynn’s five-rule criteria³³ versus the pragmatic two-rule CPR recommended by Fritz et al.³⁴ However, the two-factor CPR has accurately predicted the status of the original five-facctor manipulation CPR developed by Flynn et al.³⁷ In addition, data suggest that the prevalence rate for manipulation CPR decreases as the number of criteria used to judge manipulation CPR increases from 2 to 5.³³ Therefore, we believe that the manipulation prevalence rate reported in our study is accurate for the sample investigated. Further research would be beneficial.

Conclusions

Our data suggest that the generalizability of different classification approaches for a wide variety of patients with non-specific low back pain syndromes referred to outpatient physical therapy services favors McKenzie syndrome and pain pattern classification based on MDT methods, as assessed using prevalence, compared to categorizing patients based on manipulation and stabilization clinical prediction rule criteria. The classification methods using CPR identify a narrow range of subjects resulting in a large proportion of patients remaining unclassified, thereby limiting the generalizability of that classification approach. In addition, the majority of patients classified by manipulation and stabilization CPR criteria were also classified as McKenzie derangements whose symptoms centralized. Manipulation and stabilization CPR may not represent unique treatment categories but include patients who can be initially treated in other ways. Future research efforts should be directed towards finding commonality between classification methods versus head-to-head competitions to judge classification superiority in order to identify best treatment practices designed to enhance patient outcomes and quality of life.