Evaluating ambulatory function as an outcome following treatment for spinal metastases: A systematic review

Abstract

BACKGROUND CONTEXT:

Studies regarding treatment of spinal metastases are critical to evidence-based decision-making. However, variation exists in how a key outcome, ambulatory function, is assessed.

PURPOSE:

To characterize the sources and tools investigators have used to evaluate ambulatory function as an outcome following treatment of spinal metastases. We also sought to understand the ways ambulatory function has been conceptualized in prior studies.

STUDY DESIGN:

Systematic review of the literature.

PATIENT SAMPLE:

We identified 44 published studies for inclusion. Samples within these investigations ranged from 20 to 2,096 subjects.

OUTCOME MEASURES:

We describe the methods investigators have used to evaluate ambulatory function following treatment for spinal metastases.

METHODS:

We conducted a systematic review through PubMed, Scopus and Web of Science following PRISMA guidelines. We included studies that consisted of adult patients receiving operative or non-operative treatment for spinal metastases. We also required that study investigators specified post-treatment ambulatory function as an outcome. We recorded year of publication, study design, types of spinal metastases included in the study, treatments employed, and sample size. We also described the source (medical record, study-specific observer and/or provider, patient and/or participant), tool (standardized measure, quantitative, qualitative) and concept (eg, ambulatory vs. non-ambulatory; independent ambulation vs. ambulatory with assistance vs. non-ambulatory) used to assess ambulatory function.

RESULTS:

We found the plurality of studies relied on medical record documentation as their source. Amongst prospective studies, only a minority used a quantitative measure (eg, prespecified degree of walking ability) to assess ambulatory function. Most studies conceptualized ambulatory function as a dichotomized outcome, typically ambulatory versus non-ambulatory or a similar equivalent.

CONCLUSIONS:

Wide variation exists in how ambulatory function is defined in studies involving patients with spinal metastases. We suggest several improvements that will allow a more robust assessment of the quality and quantity of ambulatory function among patients treated for spinal metastases.

Introduction

Due to advances in treatment and medical management for patients with cancer that have translated into enhanced survival, the prevalence of spinal metastases has increased over the course of the last two decades [1−3]. Current projections maintain that 14 million persons in the United States are diagnosed with cancer every year [3] and nearly 70% of those with advanced disease develop spinal metastases [1,2]. Persons with spinal metastases have a mortality rate approaching 50% within twelve months’ of diagnosis and frequently experience severe pain and neurologic impairment [1,2,4,5].

The most effective means of care delivery for patients with spinal metastases is felt to be multidisciplinary, involving teams from oncology, radiation oncology, orthopedic spine surgery, neurosurgery, physiatry, pain management, and palliative care [1,4,5]. While surgical intervention for this condition has gained increased acceptance in the community subsequent to the landmark randomized trial of Patchell et al. [6], it is widely recognized that such interventions are not curative. With this reality in mind, the goals of treatment for spinal metastases typically focus on pain reduction, maintenance of quality of life and preservation of function, including both ambulatory capacity, and independence with activities of daily living [1,2,4,5].. Studies that provide high quality evidence regarding the effective treatment of spinal metastases, particularly with regard to preserving ambulatory ability and functional independence, are critical to evidence-based decision-making [6,7,8].

However, variation exists in how ambulatory function (eg, the ability to walk) is assessed in patients with spinal metastases. For example, Fehlings et al. consider independent function only in the case of ambulation without assistive devices [8], while Patchell and colleagues characterized patients using a walker or cane as independent ambulators [6]. The work of Patchell et al. and Fehlings et al. both required direct observation of a patient’s ambulatory function for grading [6,8], while Thirion et al accepted patient report [9]. In other work, substitute measures for ambulatory function such as the Frankel scale (which evaluates neurologic function) or Eastern Cooperative Oncology Group (ECOG; which uses a holistic, non-structured, assessment of overall function) performance status have been used [7,9]. This lack of standardization can impact both conclusions of an investigation as well as the translation of findings to patient care.

In this context, we performed a systematic review of the published literature to characterize the ways that prior investigators have evaluated ambulatory function as an outcome following the treatment of spinal metastases. We also sought to determine the means through which ambulatory function was assessed when directly measured by study investigators.

Material and methods

We conducted this systematic review in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRIMSA) guidelines [10]. We conducted a primary search of the English language literature through PubMed (US National Library of Medicine) using a combination of key words and Medical Subject Headings relevant to spinal metastases and ambulatory function (Appendix 1). Supplemental searches using similar approaches were conducted in Scopus and Web of Science. The initial search was concluded on September 19, 2020.

We included studies that consisted of adult (eg, aged 18 and older) participants receiving operative or non-operative treatment for spinal metastases. We also required that study investigators specified post-treatment ambulatory function as a primary or secondary outcome. In order to ensure relevance to the current practice environment, we limited consideration to studies published since the year 2000. Further, we restricted inclusion to English language publications and those that included cohorts of 20 or more subjects. Investigations that involved pediatric subjects or the treatment of primary spinal tumors, case reports and series with less than 20 participants were excluded.

The titles and abstracts of all papers identified through our search of PubMed and Web of Science were reviewed by three independent investigators (LN, NA, AJS), with the supplemental search of Scopus performed by a single reviewer (AJS). Potentially eligible articles were then reviewed independently in full to ascertain complete eligibility. For papers that published multiple studies from the same data source, we agreed a-priori to limit consideration to the most comprehensive work in terms of the number of years of inclusion, or the number of participants evaluated. We assessed inter-rater concordance regarding study eligibility using the kappa statistic. Following full text evaluation, the reviewers met to adjudicate disagreements regarding inclusion. All disagreements were resolved via consensus.

Studies deemed eligible for inclusion were abstracted to obtain year of publication, prospective or retrospective study design, types of spinal metastases included in the study, treatments employed (eg, operative, non-operative or operative, and non-operative treatments) and sample size. We also described the source (eg, medical record, study-specific observer, provider, patient and/or participant), tool (standardized measure, quantitative, qualitative) and concept (eg, ambulatory vs. non-ambulatory; independent ambulation vs. ambulatory with assistance vs. non-ambulatory) used to assess ambulatory function.

Data was exported and stored using Microsoft Excel (Microsoft Corp., Seattle, WA) which was also used for qualitative synthesis. STATA v 15.1 (STATA Corp., College Station, TX) was employed for determinations regarding inter-rater concordance.

Results

Our initial literature search returned 981 articles, 347 from PubMed, 367 from Web of Science to 267 from Scopus. After removing duplicates, 528 articles underwent initial title and abstract review (Figure), with 102 identified for full text review. Upon completion of full text review, reviewers independently identified a similar number of articles for inclusion (AJS=42, NA=48 and LN=40; kappa=0.52). After resolving discrepancies by consensus, we identified 44 studies [6–9,11–50] that met criteria for inclusion (Table 1). Included studies were published between 2000 and 2020, with 12 (27%) conducted using a prospective design. Sample sizes ranged from 20 to 2,096 subjects. Eighteen studies were from Europe and 16 from Asia, with just 7 from North America. The majority of publications (41 of 44; 93%) studied spinal metastases in participants with various types of underlying malignancies, while two investigations limited consideration to participants with metastases from prostate cancer and one evaluated subject with lung cancer metastases. Twenty-four publications (55%) involved only surgical participants. Eight studies (18%) restricted consideration to subjects receiving non-operative treatment and 12 (27%) evaluated both operative and non-operative cohorts.

Fig.

Flow diagram of articles that were selected for inclusion in this systematic review.

Table 1.

Summary of the findings from 44 studies identified by a literature search describing outcomes by metastatic source and treatment. N/S − not specified

Article and Y	Study Design	Source of Metastases	Treatment	Sample Size	Data Source	Conceptualization of Ambulatory Function	Tool
Graham et al. (2006)[49]	Prospective	Non-specific	Nonoperative	20	Direct observation	Ambulatory; non-ambulatory	Walk score Barthel Index
Rades et al. (2018)[50]	Prospective	Non-specific	Nonoperative	203	Direct observation	Ambulatory unaided; ambulatory aided; non-ambulatory	N/S
Itshayek et al. (2018)[23]	Retrospective	Non-specific	Operative	40	Medical Chart Review	Ambulatory; non-ambulatory	Walk 2 or more steps with or without aid
Schoenfeld et al. (2019)[7]	Retrospective	Non-specific	Operative and nonoperative	402	Medical Chart Review	Independent, dependent (eg, regularly requiring a cane or walker), or nonambulatory	N/S
Hirabayashi et al. (2003)[20]	Retrospective	Non-specific	Operative	81	Medical Chart Review	Ambulatory; non-ambulatory	Frankel Grade
Tateiwa D et al. (2019)[45]	Retrospective	Non-specific	Operative	31	Medical Chart Review	Ambulatory; non-ambulatory	Frankel Grade
Rades D et al. (2008)[41]	Retrospective	Non-specific	Nonoperative	2096	Medical Chart Review	Ambulatory without aid, ambulatory with aid, not ambulatory, paraplegic	N/S
Patchell et al. (2005)[6]	Prospective	Non-specific	Operative	101	Direct observation	Ambulatory; non-ambulatory	Two steps each foot even if cane/walker used
Crnalic et al. (2013)[17]	Retrospective	Prostate	Operative	68	Medical Chart Review	Ambulatory; non-ambulatory	Frankel Grade
Hoskin et al. (2019)[21]	Prospective	Non-specific	Nonoperative	686	Direct observation	Ambulatory; non-ambulatory	Medical Research Council Muscle Power Criteria
Ju et al. (2013)[25]	Retrospective	Prostate	Operative	27	Medical Chart Review	Ambulatory; non-ambulatory	Frankel Grade
Alamanda et al. (2018)[12]	Retrospective	Non-specific	Operative	55	Medical Chart Review	Ambulatory (with or without the use of assistive devices such as a cane) and non-ambulatory (ie, being wheelchair or bed bound)	N/S
Younsi et al. (2020)[48]	Retrospective	Non-specific	Operative	101	Medical Chart Review	Ambulatory; non-ambulatory	Frankel Grade
Kondo et al. (2008)[27]	Retrospective	Non-specific	Operative	96	Medical Chart Review	Ambulatory; non-ambulatory	Frankel Grade
McLinton et al. (2006)[32]	Retrospective	Non-specific	Operative and nonoperative	191	Medical Chart Review	Walking independently, walking with aid, unable to walk	N/S
Savage et al. (2014)[44]	Retrospective	Non-specific	Operative and nonoperative	127	Direct observation	Ambulatory; non-ambulatory	Modified version of Frankel Grade
Rades et al. (2010)[40]	Retrospective	Non-specific	Operative and nonoperative	324	Medical Chart Review	Ambulatory without aid, ambulatory with aid, not ambulatory, paraplegic	N/S
Watanabe et al. (2016)[47]	Retrospective	Non-specific	Nonoperative	112	Medical Chart Review	Ambulatory; non-ambulatory	American Spinal Injury Association Impairment Scale
Oshima et al. (2016)[35]	Retrospective	Non-specific	Operative and nonoperative	56	Medical Chart Review	Ability to walk with or without cane, walker, or crutch (number of steps not indicated)	N/S
Thirion et al. (2020)[9]	Prospective	Non-specific	Nonoperative	73	Direct observation and/or patient report	Ambulatory unaided; ambulatory aided; non-ambulatory	N/S
Abel et al. (2008)[11]	Retrospective	Non-specific	Operative	34	Direct observation	Ambulatory; non-ambulatory	Walk 10 m with or without aid
Moon et al. (2011)[33]	Retrospective	Non-specific	Operative	182	Medical Chart Review	Ambulatory; non-ambulatory	Nurick Score
Jiang et al. (2016)[24]	Retrospective	Non-specific	Operative and nonoperative	67	Medical Chart Review	Ambulatory; non-ambulatory	Frankel Grade
Park et al. (2013)[36]	Retrospective	Non-specific	Operative	102	Direct observation	Ambulatory (including normal ambulation, ambulation with aid, and ambulation without aid) or non-ambulatory	N/S
Rades et al. (2020)[39]	Prospective	Non-specific	Nonoperative	40	Direct observation	Ambulatory; non-ambulatory	Study-specific 5 point scale
Ohashi et al. (2017)[34]	Retrospective	Non-specific	Operative and nonoperative	82	Direct observation	Ambulatory; non-ambulatory	Walk more than 10 m with or without cane or walker
Putz et al. (2014)[38]	Retrospective	Non-specific	Operative	43	Medical Chart Review	Ambulatory; non-ambulatory	Mobility subcategory of the Spinal Cord Injury Measure
Helweg-Larsen et al. (2000)[19]	Prospective	Non-specific	Nonoperative	153	Direct observation	Ambulatory without assistance; ambulatory with assistance; paretic; paralytic	N/S
Marquardt et al. (2004)[31]	Prospective	Non-specific	Operative	34	Direct observation	Ambulatory; non-ambulatory	N/S
Guzik. (2016)[18]	Retrospective	Non-specific	Operative	57	Medical Chart Review	Unable to walk versus walking on crutches or with frame	N/S
Rades et al. (2016)[42]	Prospective	Non-specific	Operative	203	Direct observation	Ambulatory unaided; ambulatory aided; non-ambulatory	N/S
Chang et al. (2019)[15]	Retrospective	Non-specific	Operative	100	Medical Chart Review	Ambulatory; non-ambulatory	ECOG
Laufer et al. (2010)[28]	Retrospective	Non-specific	Operative and nonoperative	39	Medical Chart Review	Ambulatory; non-ambulatory	ECOG
Kim et al. (2011)[26]	Retrospective	Non-specific	Operative	57	Medical Chart Review	Ambulatory; non-ambulatory	Nurick Score
Sailhan et al. (2018)[43]	Retrospective	Non-specific	Operative	319	Medical Chart Review	Normal walking without assistance, altered walking without assistance, walking with cane, walking with two canes or walker, cannot walk	N/S
Lee et al. (2014)[29]	Retrospective	Non-specific	Operative and nonoperative	33	Medical Chart Review	Ambulatory; non-ambulatory	Ryu and/or Rock Score
Truong et al. (2020)[46]	Retrospective	Non-specific	Operative	87	Medical Chart Review	Non-ambulatory, ambulatory with help, independent ambulation	N/S
Park et al. (2016)[37]	Prospective	Lung	Operative	50	Direct observation	Ambulatory; non-ambulatory	Nurick Score
Fehlings et al. (2016)[8]	Prospective	Non-specific	Operative	142	Direct observation	Independent; non-independent	Walk 4 steps independently
Hutton et al. (2013)[22]	Retrospective	Non-specific	Operative and nonoperative	39	Direct observation	Ambulatory without aid, ambulatory with aid, not ambulatory, paraplegic	N/S
Chaichana et al. (2009)[14]	Retrospective	Non-specific	Operative	162	Direct observation	Ambulatory with or without aid	N/S
Alvarez et al. (2003)[13]	Retrospective	Non-specific	Operative and nonoperative	21	Direct observation	Gait without support, gait with cane, non-ambulatory	N/S
Conway et al. (2007)[16]	Prospective	Non-specific	Operative and nonoperative	319	Direct observation	Ambulatory unaided; ambulatory aided; non-ambulatory	N/S
Lei et al. (2016)[30]	Retrospective	Non-specific	Operative	206	Medical Chart Review	Ambulatory; non-ambulatory	Frankel Grade

We encountered broad variation in the means through which ambulatory function was assessed, including the sources, and tools used, as well as the conceptualization employed. The sources ranged from direct observation to prespecified use of substitute measures such as the Frankel Scale or ECOG performance status and non-specific determination from medical record documentation (Table 2). The plurality of retrospective works (15 of 32; 47%) documented ambulatory function as reported in medical charts. Other retrospective studies used substitute measures, such as the Frankel Scale, ECOG performance status or Ryu/Rock Neurologic Scale. It was frequently unclear how ambulatory function was determined and whether measures were made by study-specific staff, treating clinicians, or via participant report. This made it difficult for us to conduct an objective determination of bias regarding these measurements in the available studies. Two retrospective studies [26,33] employed the Nurick Scale, which is dedicated to describing ambulatory function, and two specified that participants had to be able to walk 10 meters in order to be considered ambulatory [11,34].

Table 2.

Method of determination of ambulatory function in studies included in this systematic review

Method of Determination	Description
Direct Observation	Study investigators directly observed patients to characterize ambulatory function.
Medical Chart Review	Investigators retrospectively reviewed patient encounter data and transcribe determinations as recorded by treating clinicians.
Frankel Grade	Investigators use assessment of Frankel Grade as a substitute for ambulatory function. Grades D and E are considered ambulatory.
Nurick Score	Investigators use this 5-point scale to characterize ambulatory function. Grade 5 considered non-ambulatory.
Eastern Cooperative Oncology Group (ECOG) Performance Status	Investigators use this 6-point scale (0–5) to characterize ambulatory and/or physical function. Grades 3–4 considered non-ambulatory.
Ryu and/or Rock Neurologic Scale	Investigators use this 5-point scale to characterize ambulatory function. Grades D and E considered non-ambulatory.

The majority of prospective investigations (11 of 12; 92%) indicated that they used direct observation to evaluate ambulatory function (Table 3). One prospective work used the Nurick Scale [37] and two studies that specified direct observation did allow subjects to self-report ambulatory function remotely if they were unable to come in for formal evaluation [9,21].

Table 3.

Method of determination and designations regarding ambulatory function in prospective studies that used direct observation of patients

Study (Y of Publication)	Method of Determination	Designation of Ambulatory Function
Helweg-Larsen et al. (2000)[19]	Not specified	Ambulatory without assistance; ambulatory with assistance; paretic; paralytic
Marquardt et al. (2004)[31]	Not specified	Ambulatory versus non-ambulatory
Patchell et al. (2005)[6]	Two steps each foot even if cane and/or walker used	Ambulatory versus non-ambulatory
Graham et al. (2006)[49]	Walk score >5/7; Barthel Index walk score >2/3	Ambulatory versus non-ambulatory
Conway et al. (2007)[16]	Not specified	Ambulatory unaided; ambulatory aided; non-ambulatory
Park et al. (2016)[37]	Nurick Score: Grades 3–5=ambulatory; Grades 1- 2=non-ambulatory	Ambulatory versus non-ambulatory
Fehlings et al. (2016)[8]	Walk 4 steps independently	Independently ambulatory versus not independent
Rades et al. (2016)[42]	Not specified	Ambulatory unaided; ambulatory aided; non-ambulatory
Rades et al. (2018)[50]	Not specified	Ambulatory unaided; ambulatory aided; non-ambulatory
Hoskin et al. (2019)[21]	Medical Research Council Muscle Power Criteria: Grades 1–2=ambulatory; Grades 3–4=nonambulatory	Ambulatory versus non-ambulatory
Thirion et al. (2020)[9]	Not specified (also allowed remote patient report)	Ambulatory unaided; ambulatory aided; non-ambulatory
Rades et al. (2020)[39]	Study-specific 5-point scale	Ambulatory versus non-ambulatory

Among the studies that used direct observation, several did not specify a quantitative measure a-priori, such that it was unclear how far a patient had to walk to be considered independently ambulatory or if they could use a walking aid and still be classified as independent. The randomized trial of Patchell et al. required that subjects take two steps with each foot to be considered ambulatory, but allowed for the use of a cane or walker [6]. Fehlings et al. followed a similar approach in terms of the number of steps required to demonstrate ambulatory function, but did not allow the use of a cane or walker [8]. Hoskin et al. employed the Medical Research Council Muscle Power Criteria to determine ambulatory function with Grades 1–2 classified as ambulatory and Grades 3 to 4 maintained to be non-ambulatory [21].

Nearly half of the prospective studies (5 of 12; 42%) dichotomized ambulatory function as ambulatory or non-ambulatory. Four used a rubric that consisted of: Ambulatory unaided, ambulatory aided, and non-ambulatory. Fehlings et al. classified ambulatory function as independent or not independent [8]. In that study, the category of not-independent grouped together those subjects who were non-ambulatory with individuals who walked with assistive devices. Helweg-Larsen et al. used a scheme with four tiers: Ambulatory without assistance, ambulatory with assistance, paretic, and paralytic [19].

Discussion

The recent rise in the prevalence of patients with spinal metastases means that clinicians who treat patients with spinal disorders are increasingly confronted with the need to manage this challenging condition [1−5]. Providers, whether individually or in multi-disciplinary teams, must reconcile patients’ clinical presentation, general health, and functional status, goals, and family needs with the risks and anticipated benefits of operative and non-operative treatment modalities [1,2,4,5]. Often, this conversation is conducted in the context of minimizing pain, protecting spinal stability and neurologic function, as well as preserving ambulatory status and personal independence [5,51]. It is frequently assumed that surgical intervention will result in superior preservation of ambulatory function when compared with non-operative treatment, but what this means from both a qualitative and quantitative standpoint remains poorly defined. To some this could mean that patients will be walking longer without assistive devices, or that patients could walk a reasonable distance without pain exacerbation, or the need to rest. In reality, however, this might only actually translate to avoiding paralysis, or immobility that requires use of a wheelchair. Variability in how ambulatory function is measured can translate into ambiguity in the interpretation of existing research and its application to patient care. Overestimating the effectiveness of treatment with respect to preservation of ambulatory function might mean that patients accept more intensive surgical procedures or multimodal interventions, but do not reap any meaningful benefits in quality of life. Underestimation of the effectiveness of such interventions could lead patients and providers to opt for less risky interventions that, at the same time, are not as effective at preserving functional independence.

We are presently unaware of prior reviews that have sought to systematically assess the ways in which ambulatory function has been studied following treatment for spinal metastases. We encountered broad variation within the literature in terms of the way ambulatory function was assessed, including the source of this data and the tools used for measurement. The majority of retrospective investigations either did not specify how ambulatory function was measured, or relied on a substitute such as an assessment of neurologic status or physical performance. In these papers, it was often unclear who made the determination of ambulatory function and whether, or not, this was achieved through a quantitative or qualitative assessment. Among prospective studies, only a minority pre-specified the degree of walking ability necessary to be considered ambulatory. When ambulatory function was measured using a pre-specified number of steps or distance, the threshold was fairly low. Patchell et al. [6] and Fehlings et al. [8] both required four steps in total, while Abel et al., [11] and Ohashi et al. [34] both specified the need to walk 10 meters.

When conceptualizing the ability to walk, most studies presented a dichotomized outcome such as ambulatory versus non-ambulatory or independent versus not independent. We believe this scheme is suboptimal in that it restricts appreciation for those patients who walk with assistive devices. This is an important cohort given that the need to ambulate with assistive devices is frequently a transition state and may occur as patients are recovering from the effects of operative or non-operative treatment, or slowly declining in the latter stages of the metastatic condition. We do not believe that a patient who is only able to ambulate a few steps with a walker can be considered equivalent to a patient who can walk a mile without need of an aid. Conversely, the patient who is able to mobilize around their house and in the community with a cane cannot be considered as having the same level of functional ability as a patient who requires use of a wheelchair or is bed-bound.

Patients and providers need data on factors associated with both the quantity and quality of life following treatment for spinal metastases and how the initial treatment strategy may influence outcomes. These data are essential to the formulation of objective clinical recommendations and shared decision-making [5]. Based on this systematic review of the literature, we believe several important recommendations can improve the quality of investigations in the future. First, ambulatory function should be measured using objective tests, such as a timed assessment of walking ability over a pre-specified distance. One previously validated assessment includes the Timed Up and Go (TUG) test, where patients perform a timed assessment of rising from a seated position, walking 3 meters in a straight line, turning around and returning to sit in the chair as quickly as possible [52,53]. Scores over 11.5 seconds to complete the test are considered mildly impaired, while those of >18.4 seconds are severely impaired [53]. Another example includes the 40 meters walk test, where patients simply walk a distance of 40 meters as quickly as possible [54]. It should be noted that these tests have been validated for patients with degenerative musculoskeletal and spinal conditions [52−54], but not specifically for patients with spinal metastases. Pathologic performance for these tests may be different in patients with spinal metastases and, moreover, limitations could result from the global burden of the oncologic process as opposed to the spinal metastases alone. This would call into question the extent to which poor performance on these objective tests might be reversible. Further work is needed to determine the extent to which generic cut-offs for performance on objective measures of ambulatory function apply to patients with spinal metastases and the extent to which such limitations may be reversible through effective treatment. We envision conducting research along both these lines in the future.

Formal assessment should ideally be conducted in a clinical setting, but recognizing the limitations that beset patients with spinal metastases, especially in latter stages of the disease process, we believe that the creative use of technology such as virtual platforms could be leveraged to facilitate these types of assessments if and when patients are confined to home. Second, a more granular and specific designation of ambulatory function should be provided, including independent ambulatory function, ambulation with assistive device and non-ambulatory. There are currently studies that exist in the literature that employ such granular designations of ambulatory function, including Schoenfeld et al. [7], Alvarez et al. [13] and Rades et al. [50] These ambulatory states should be considered as discrete entities and analyzed as such whenever practicable. Further research could also look to develop health utilities aligned with these functional categories.

We then envision a combined grading scheme, whereby the dedicated ambulatory state is supplemented by a modifier indicating the quality of ambulatory function. For example: Independently ambulatory, TUG test time of 10 seconds; or ambulatory with assist (cane), TUG test time of 20 seconds. The hierarchical scheme would prioritize independent ambulatory function with superior objective functional performance over independent function with lower degrees of functional performance, followed by ambulation with assistance and non-ambulatory designations. While this grading scheme appears reasonable, it should be validated. Clinicians and researchers who seek to use these grading schemes should employ trained study staff to administer standardized assessments and validate determinations against patient-centered estimates of quality of life such as the Euroquol 5-D or the PROMIS-29 in pre-operative and post-operative patients. These determinations could potentially be incorporated into clinical registries at the institutional level, or on national platforms such as the American Spine Registry or the National Neurosurgical Quality Outcomes Database (N2QOD).

We acknowledge several limitations associated with this work. In this qualitative synthesis of the literature, we were unable to address the potential for selection, indication, classification, and surveillance bias, among others, that may confound the results of the individual studies presented in this review. Further, patients in these investigations received a number of heterogeneous interventions and this precludes our ability to draw firm conclusions regarding the efficacy of any particular treatment strategy for maintenance of ambulatory function. While this investigation followed established guidelines for the conduct of a systematic review, including inclusion criteria defined a-priori and independent evaluation for inclusion with several investigators, there remains the possibility that certain articles relevant to this topic were missed inadvertently. A further limitation includes the fact that this review was limited to English language articles only.

In conclusion, our systematic review of the literature revealed wide variation in how ambulatory function is defined in studies involving patients with spinal metastases. Variability in how ambulatory function is measured can translate into ambiguity in the interpretation of existing research and its translation to clinical practice. We suggest several improvements that will allow a more robust assessment of the quality and quantity of ambulatory function among patients treated for spinal metastases in future work.