Prevention and management of osteoporosis and osteoporotic fractures in persons with a spinal cord injury or disorder: A systematic scoping review

Abstract

Objectives: The primary objective was to review the literature regarding methodologies to assess fracture risk, to prevent and treat osteoporosis and to manage osteoporotic fractures in SCI/D.

Study Design: Scoping review.

Settings/Participants: Human adult subjects with a SCI/D.

Outcome measures: Strategies to identify persons with SCI/D at risk for osteoporotic fractures, nonpharmacological and pharmacological therapies for osteoporosis and management of appendicular fractures.

Results: 226 articles were included in the scoping review. Risk of osteoporotic fractures in SCI is predicted by a combination of DXA-defined low BMD plus clinical and demographic characteristics. Screening for secondary causes of osteoporosis, in particular hyperparathyroidism, hyperthyroidism, vitamin D insufficiency and hypogonadism, should be considered. Current antiresorptive therapies for treatment of osteoporosis have limited efficacy. Use of surgery to treat fractures has increased and outcomes are good and comparable to conservative treatment in most cases. A common adverse event following fracture was delayed healing.

Conclusions: Most of the research in this area is limited by small sample sizes, weak study designs, and significant variation in populations studied. Future research needs to address cohort definition and study design issues.

Introduction

In 2016, 282,000 persons in the United States (U.S.) were living with a spinal cord injury (SCI). Approximately 17,000 new cases of SCI occur each year.¹ Sublesional loss of bone mineral density (BMD) is present in the majority of patients with a traumatic motor complete SCI, and incidence rates for osteoporotic fractures in this population range between 2 to 3 fractures per 100 patient-years.^2–5 These fractures contribute to both excess morbidity⁶ and mortality.⁷ Many patients experience more than one long bone fracture following injury.^8,9

However, approaches to management of osteoporosis and osteoporotic fractures in patients with a SCI vary,¹⁰ and unlike the general population for whom there are established guidelines for management of osteoporosis,¹¹ there is only one previously published screening paradigm to identify persons with a SCI who are at greatest risk for fracture.¹² Scoping reviews are used by researchers to examine the extent and nature of reported research and identify gaps in existing literature. Scoping reviews can be most helpful when there are few randomized clinical trials available, as is the case for management of osteoporosis in SCI. Therefore, a scoping review¹³ was undertaken to identify optimal strategies in three domains: (1) assessment of fracture risk, (2) treatment of osteoporosis, and (3) management of osteoporotic fractures. Secondary objectives were to identify research gaps and to make recommendations for future research.^14,15

Materials and methods

We followed an approach recommended by scoping review methodologists^14–19 and utilized a six-stage methodological framework for scoping studies illustrated in Figure 1.

Figure 1.

Six stage methodological framework for scoping studies.

1. Identify the research question

The scoping review team included experts in SCI medicine and health services research experts. The objective of the review was to identify best practices described in the literature. A set of research questions that would address the objective of the review were identified:

1. Can Dual Energy X-ray Absorptiometry (DXA) measurements and/or clinical factors identify persons with a SCI at highest risk for fracture?

2. What secondary causes of osteoporosis should be considered in persons with a SCI?

3. What is the role of non-pharmacological modalities in the treatment of SCI-related bone loss?

4. What is the evidence for efficacy and safety of pharmacological treatments used for postmenopausal osteoporosis in treating osteoporosis in persons with a SCI?

5. Is surgical or nonsurgical treatment of extremity fractures in SCI preferred and what patient or fracture level characteristics should be considered in determining treatment?

6. What key areas should be identified for further research in this area?

Criteria for inclusions and exclusions were identified including focusing on studies that included adults (18 years and older), patients with SCI/D, and covering the concepts of interest within this review.

2. Utilizing carefully selected search terms to identify potentially relevant studies

The search strategy recommended by the Joanna Briggs Institute (JBI) was used to identify relevant qualitative and quantitative articles.²⁰ A medical research librarian conducted a search within PubMed and Cochrane Clinical Trials databases using the following Medical Subject Headings (MeSH) terms:

• “Osteoporosis” or “bone density” or “fractures” AND “spinal cord injuries” or “traumatic myelopathy” or “spinal cord trauma.”

• “Bone and lower extremity” or “bones of lower extremity” AND “spinal cord injuries” or “traumatic myelopathy” or “spinal cord trauma.”

• “Bone therapy” or “drug therapy” AND “spinal cord injuries” or “traumatic myelopathy” or “spinal cord trauma.”

All search terms used are listed in Figure 2.

Figure 2.

MeSH terms.

Relevant articles were restricted by language (English), date (from 1966–2016), and human studies. Article titles and abstracts were independently reviewed by two authors. Articles that included fractures at the same time as the SCI, non-United States Food and Drug Administration (FDA) approved medications, ultrasound technology or QCT alone, children, and those that did not address one of the specific research questions were excluded.

3. Reviewing studies to ensure they met inclusion

A total of 438 manuscripts were identified based on the initial searches, of which 29 were excluded due to duplication (21), inclusion of non-FDA approved drugs (6), axial fractures at time of SCI (1), and incorrect population (e.g. children, non-SCI) (1). Title and abstract review excluded 107 articles for the following reasons: irrelevant to the research question, axial fractures only and incorrect population (e.g. children, non-SCI). Independent review for inclusion and exclusion yielded 92% initial agreement between reviewers and 100% agreement after consensus discussion.

The 302 articles were included for full-text review. Similar to the abstract review, two authors reviewed each article for inclusion. There was 100% initial agreement between the two reviewers for inclusion based on the full-text review. Following complete review of the articles, an additional 129 were excluded (Figure 3, Preferred Reporting Items for Systematic Reviews and Meta-Analysis [PRISMA] Flow Diagram). A manual literature search of PubMed and Cochrane Clinical Trials databases and of the grey literature (including World Cat, Google and Google Scholar [first five pages], National Technical Information Service/National Technical Reports Library [NTIS/NTRL], Agency for Healthcare Research and Quality [AHRQ] Grey Source Index, and U.S. Registry of Clinical Trials and European Union [EU] Registry of clinical trials) was performed to retrieve additional relevant articles including methodology for scoping reviews; osteoporosis in the general healthy population without a SCI; other key articles from the SCI literature; vertebral fractures occurring after the SCI; key references embedded in manuscripts identified in the original literature search by the medical research librarian; and updated articles through August 2017. This yielded an additional 53 articles that were included in the scoping review. In cases of planned but unpublished literature reviews (Cochrane databases) identified through grey literature searches, authors were contacted for further information. A total of 226 articles were included in the scoping review.

Figure 3.

PRISMA flow diagram.

4. Final assessment and data extraction

The data extraction sheet was developed in Microsoft Excel and contained 17 distinct tabs with related variables in each tab with a total of 253 variables. A glossary and instruction set were developed to standardize the abstraction process among abstractors.

5. Collating, summarizing, and reporting results

Collating

The abstraction forms allowed collecting information from each manuscript, while combining the information to facilitate early insights emerging from the manuscripts across all the reviews (e.g. number of manuscripts detailing medications as risk factors for osteoporosis).

Summarizing

An extraction tool was utilized to summarize the extracted information (see Supplementary Appendix).

Reporting

Thoughtful development of the variable definitions, extraction tabs, and planned analyses facilitated the reporting of the final results of the scoping review. A thematic analysis was used to report the qualitative findings.

6. Consulting with stakeholders

SCI providers were consulted prior to initiating the review to discern the scope of the review and identify key search terms and data elements to be abstracted from each article. SCI providers were also involved in the data review and abstraction.

Results

For a number of studies, information about the study design (i.e. pre-post (1 group vs. 2 or more groups), repeated measures (1 group vs. 2 or more groups), RCT, or observational (prospective, cross-sectional, or retrospective)) and study participants (i.e. demographic and SCI characteristics) are presented in Table 1. The results of these studies are presented in more detail in Tables 2 through 8. Table 1 also provides the number of the table where each study’s results can be found.

Table 1. Overview of study designs and subject characteristics.

Authors	Table number (where results are discussed)	Study type1	Sample Size2	Sex M/F	Mean Age (or range)3	Duration of Injury4	Extent of Injury (or AIS)	Anatomic Level5/ Etiology (if reported)
McMaster et al. (1975)	8	Retrospective observation	44	n.s.	n.s.	acute + chronica	C + I	C-L
El Ghatit et al. (1981)	7	Case report	2	2/0	39	chronic	C	T4, T12
Freehafer et al. (1981)	7	observation	18	n.s.	n.s.	n.s.	n.s.	n.s./ traumatic + Nontraumatic
Nottage (1981)	7	Retrospective observation	30	23/7	33	acute, n.s. if chronica	C + I	C3-L4
Baird et al. (1984)	7	Case report	1	1/0	50	chronic	C	T12/ traumatic
Baird et al. (1986)	7	Retrospective observation	9	9/0	47	acute + chronic	C + I	C5-T12
Garland et al. (1986)	7	Retrospective observation	28	18/10	30	acute	C + I	C-L
Biering–Sorensen et al. (1988)	2	Cross-sectional observation	26	24/2	20–65	chronic	C + I	C5-L4/ traumatic + nontraumatic
Garland et al. (1988)	7	Retrospective observation	46	13/33	27.9	n.s.	n.s.	para + tetra
Ingram et al. (1989)	7	Retrospective observation	25	n.s.	(13–70)	≥ 1 year	n.s.	C-L/ traumatic + nontraumatic
Leeds et al. (1990)	5	Pre-post (1 group)	6	6/0	n.s.	chronic	n.s.	tetra
Sobel et al. (1991)	8	Case report	1	1/0	43	14 years	n.s.	T8
Keating et al. (1992)	7, 8	Case report	1	1/0	42	2 years	n.s.	para
Garland et al. (1993)	2	Cross-sectional observation	28	28/0	35.7, 32.2	chronic	C	C4 –T12/ traumatic
Freehafer et al. (1995)	7	Retrospective observation	n.s.	n.s.	n.s.	n.s.	n.s.	n.s.
Wilmet et al. (1995)	3	Prospective observation	31	24/7	32.5	< 8 weeks	Cb	T2-L3/ traumatic
BeDell et al. (1996)	5	Pre-post (1 group)	12	12/0	34	> 2 years	(A)	C5-T12/ traumatic
Needham-Shropshire et al. (1997)	4	Pre-post (1 group)	16	13/3	28.8	n.s.	C	T4-T11
Szollar et al. (1998)	3	Prospective observation	176	176/0	30.4	0–39 years	C + I	para + tetra
Vestergaard et al. (1998)	3	Cross-sectional observation	438	309/129	43, 40	acute + chronic	C + I	C-L/ traumatic + nontraumatic
de Bruin et al. (1999)	4	Repeated measures (2 ≥ groups)	13	13/0	32.7	< 5 weeks	(A-D)	C4-L1
Belanger et al. (2000)	5	Pre-post (1 group)c	14	11/3	32.4	> 1 year	(A-C)	C5-T6/ traumatic
Liu et al. (2000)	3	Cross-sectional observation	64	64/0	52	acute + chronic	C + I	n.s./ traumatic
Lazo et al. (2001)	2	Cross-sectional observation	41	n.s.	56	chronic	(A-D)	C2-L1
Wood et al. (2001)	3	Cross-sectional observation	22	22/0	36.4	n.s.	C	T3-T12/ traumatic
Sniger & Garshick (2002)	6	Case report	1	1/0	47	n.s.	(D)	C4/ traumatic
Eser et al. 2003	5	Pre-post (≥ 2 groups)	38	34/4	32.9, 33.8	acute	(A, B)	C5-T12/ traumatic
Bubbear et al. (2004)	6	Case report	4	n.s.	n.s.	n.s.	C + I	n.s.
Clasey et al. (2004)	3	Cross-sectional observation	29	21/8	38.5	n.s.	C	< C6
Garland et al. (2004)	3	Cross-sectional observation	152	n.s.	n.s.	n.s.	C + I	para + tetra
Goktepe et al. (2004)	4	Cross-sectional observation	34	34/0	28.4, 28.5	chronic	(A-C)	T1-T12
Zehnder et al. (2004)	6	RCT	55	55/0	38.8, 37.9	4 < 6 months, 51 ≥ 6 months	(A-B)	T1-L3
Ben et al. (2005)	4	Pre- post (1 group)c	20	16/4	34	< 12 months	n.s.	para + tetra
De Brito et al. (2005)	6	RCT	19	15/4	30.9, 30.8	> 6 months	(A-C)	para + tetra/ traumatic
Mulsow et al. (2005)	3	Retrospective observation	158	n.s.	n.s.	n.s.	C	n.s.
Shields & Dudley-Javoroski (2006)	5	Repeated measures (1 group)c	7	7/0	27.7	< 6 weeks	(A)	C5-T10
Shojaei et al. (2006)	2	Cross-sectional observation	132	132/0	37.4	chronic	C	C-L
Clark et al. (2007)	5	Repeated measures (≥ 2 groups)	33	n.s.d	31.0, 8.6	acute	(A-D)	C4-T12/ traumatic + nontraumatic
Gilchrist et al. (2007)	6	RCT	31	22/9e	(17–55)	acute	(A-Db)	C4-L2
Shapiro et al. (2007)	6	RCT	18	14/4	30.1, 28.4	10-12 weeks	(A, B)	C2-T12/ traumatic
Alekna et al. (2008)	4	Prospective observation	54	44/10	34.6	acute	(A, B)	C2-L1/ traumatic
Dudley–Javoroski & Shields (2008)	5	Repeated measures (1 group)c	12	12/0	36.2	acute + chronic	(A, B)	C5-T11
Goktepe et al. 2008	4	Cross-sectional observation	71	60/11	29.80, 32.09, 31.02	> 1 year	(A, B)	para + tetra/ traumatic + nontraumatic
Frotzler et al. (2009)	5	Prospective observation	5	4/1	38.6	chronic	(A)	T4-T7
Griffin et al. (2009)	5	pre/post-test (1 group)	18	13/5	40	≥ 1 year	C + I	C4-T7
Morse et al. (2009)	8	Prospective observation	n.s.	n.s.	n.s.	n.s.	(A-D)	para + tetra
Forrest et al. (2010)	5	RCT	11	n.s.	n.s.	x̅ = 3.5 years	(A, B)	n.s.
Groah et al. (2010)	5	RCT	26	22/4	26.2, 31.1	<12 weeks	(A, B)	C4- T12
Bubbear et al. (2011)	6	RCT	14	9/5e	29.3	< 3 months	C + I	C4-L3
Chain et al. (2012)	4	Cross-sectional observation	25	25/0	36, 30	> 1 year	C + I	C5-C7/ traumatic
Lichy et al. (2012)	4	Case report	1	1/0	32	1.5 years	(A-Ca)	T4/ traumatic
Sugi et al. (2012)	7	Retrospective observation	9	n.s.	n.s.	chronic	n.s.	N.S./ traumatic + nontraumatic
Astorino et al. (2013)	4	Pre-post	13	11/2	29.4	11 < 3 years 2 ≥ 3 years	C + I	C4-L1
Bishop et al. (2013)	7	Retrospective observation	396	99%/1%	60	n.s.	n.s.	para + tetra
Carbone, Chin, Lee, et al. (2013a)	3	Retrospective observation	7,447	7,447/0	57.18, 58.73	acute + chronic	C + I	para + tetra/traumatic
Carbone Chin, Lee, et al. (2013b)	3	Retrospective observation	7,447	7,447/0	57.14, 58.91	≥ 2 years	C + I	para + tetra/ traumatic
Karimi et al (2013)	4	Case report	3	n.s.	32-35	n.s.	C	T12-L1
Uehara et al. (2013)	7	Case series	15	9/6	52.7	n.s.	(A)	C5-L1
Carbone, Chin, & Burns et al. (2014)	3	Retrospective observation	12,389	12,389/0	54.2, 54.6	≥ 2 years	C + I	para + tetra
Carbone, Chin, Lee et al. (2014)	3	Retrospective observation	6969	6969/0	26.10	n.s.	n.s.	para + tetra
Dolbow et al. (2014)	5	Case report	1	n.s.	60	2 years	(A)	T6
Gibbons et al. (2014)	5	Case reportg	24	n.s.h	58, 32, 37	chronic	(A)	C6-T10
Gordon et al. (2014)	6	Pre-post	12	10/2	34	> 1 year	(A-C)	C3-5 – L2
Lala et al. (2014)	2	Cross-sectional observation	70	50/20	48.8	chronic	(A-D)	C1-L1/ traumatic
Wang et al. (2014)	8	Prospective observation	22	14/8	41.95	n.s.	n.s.	C3-L4
Bauman et al. 2015	6	Repeated measures (≥ 2 groups)	13	11/2	25.5, 33	< 16 weeks	(A, B)	para + tetra
Bethel et al. (2015)	7	Retrospective observation	1281	1281/0	n.s.	>2 years	C + I	para + tetra/ traumatic
Frotzler et al. (2015)	7	Retrospective observation	107	73/34	50.6	x̅ = 20.1 ± 12.2	(A-D)	para + tetra/ traumatic
Kostovski et al. (2015)	4	Prospective observation	31	31/0	34	acute	(A-E)	C4-L5/ traumatic
Wang et al. (2015)	8	Prospective observation	15	n.s.	46.6	n.s.	(A-D)	C-L
Bethel, Weaver, et al. (2016)	3	Retrospective observation	1281	1281/0	56	≥ 2 years	(A-D)	para + tetra/ traumatic + nontraumatic
Gifre et al. (2016)	6	Prospective observation	14	14/0	39	8-21 months	(A-C)	para + tetra/ traumatic
Morse et al. (2016)	3	Prospective observation	152	133/19	55.1	n.s.	(A-D)	n.s.
Abderhalden et al. (2017)	2	Retrospective observation	552	n.s.d	54, 52.2	≥ 2 years	(A-D)	para + tetra/ traumatic + nontraumatic
Deley et al. (2017)	5	Case report	1	0/1	36	2 years	(A)	T4-T5

M, male; F, female; C (except where presented as AIS grade in parentheses), complete; I, incomplete; para, paraplegic; tetra, tetraplegic; (for extent of injury:) n.s., not specified.

¹Repeated measures studies included 3 or more repeated measurements. 2 or more groups means the study design involved a comparison of 2 or more groups of participants, single group means that the study design involved only single group of participants, case report means only individual cases were examined.

²Sample sizes are reported for SCI participants only unless the comparison between SCI and non-SCI participants was relevant to the scoping review.

³For some studies, age is presented separately for each participant group as reported in the original articles.

⁴If a specific duration or range of durations of injury was indicated in the original article, it was included in the table. Otherwise, duration of injury was categorized as acute, chronic, or both as indicated in the original article.

⁵Anatomic levels: C (cervical), T (thoracic), L (lumbar)

^aSome of the fractures occurred at the time of spinal cord injury.

^bSome subjects became less impaired over the course of the study.

^cWithin-subjects: trained leg vs. untrained leg

^dsample contains males and females, but the number of each was not reported.

^ePremenopausal.

^fPostmenopausal.

^g1 case compared to untrained SCI group & non-SCI group.

^hFES rower was a male with T4 ASIA A injury of 13.5 years.

Six studies in Table 2 specifically examined BMD as a risk factor for fractures in SCI. All studies found a relationship between lower BMD and increased risk of fracture in persons with SCI. Twenty studies examined risk factors related to osteoporosis and/or fractures in SCI (see Table 3). Four addressed different clinical factors, suggesting that white race, history of fractures, being female and over 50 years of age, and having a family history of fractures were risks for fracture after SCI. A fifth study found a positive relationship between low BMI and osteoporosis. Twelve studies examined SCI- related factors. Those with complete injuries, with injuries of longer duration, and who were not ambulatory were at greater risk for fracture. Three studies examined medication as a risk factor, finding that opioids and anticonvulsants increased risk while thiazides appeared to be protective.

Table 2. DXA-derived BMD and fracture risk.

Study	Risk factor	Results (unless otherwise specified, the results state how the risk factor was associated with fracture risk)
Biering–Sorensen et al. (1988)	Osteoporosis of femoral neck	Patients with osteoporosis of femoral neck have greater number of fractures events compared to those with normal BMD or osteopenia of femoral neck
Garland et al. (1993)	Net average BMD of the distal femur and proximal tibia.	Was significantly lower in SCI patients who had a lower limb fracture than in both healthy age-matched controls and SCI patients with no prior history of a lower extremity fracture
Lazo et al. (2001)	BMD at femoral neck	Each 0.1 g/cm2 decline in BMD at the femoral neck increased the risk of fracture over two-fold in patients within each age group
Shojaei et al. (2006)	BMD at the lumbar spine and femoral neck	Each reduction of 0.1 g/cm2 of BMD increased the risk of fracture up to 27.7% in the lumbar spine and 53.9% in the femoral neck
Lala et al. (2014)	Low areal BMD at the knee and geometrical changes at the tibia	Were significantly associated with fragility fractures after adjusting for motor complete injury
Abderhalden et al. (2017)	Low BMD at the hip	Associated with more incident fractures in patients with incomplete SCI

Table 3. Clinical, SCI-related and medication use as risk factors for osteoporosis and/or fractures.

Study	Risk Factor	Results (unless otherwise specified, the results state how the risk factor was associated with fracture risk)
Clinical risk factors
Vestergaard et al. (1998)	Family history of fracture	Fractures were more frequent in male patients with a family history of fractures
Garland et al. (2004)	Low BMI	Was associated with increased likelihood of osteoporosis. The odds ratio for BMI indicated that every unit increase in BMI lowered the odds of being in the osteoporotic group by 11.29%
Carbone, Chin, Burns et al. (2014)	White race	Patients with an incident lower extremity fracture were significantly more likely to be white compared to patients without an incident lower extremity fracture
Bethel, Weaver, et al. (2016)	Prevalent fracture	Increased risk for both incident upper and lower extremity fractures
Bethel, Weaver et al. (2016)	Females over 50 years old	Increased risk for lower extremities fracture
SCI-related risk factors
Wilmet et al. (1995)	Ambulatory status	Patients who became partially ambulatory at 1 year post SCI showed less bone loss at the pelvis relative to patients who were still immobilized
Szollar et al. (1998)	Duration of injury	Longer duration of injury associated with lower BMD at the femoral regions
Liu et al. (2000)	Duration of injury	Z-scores at the femoral neck were significantly lower with longer duration of injury
Wood et al. (2001)	Motor complete paraplegia	81.8% of these patients had Z scores less than or equal to -1.0 and 22.7% less than or equal to -2.0 at the hip
Clasey et al. (2004)	Duration of injury	Inversely related to BMD of arm, leg and trunk
Mulsow et al. (2005)	Ambulatory status	Patients with complete SCI had increased incidence of lower extremity fractures compared to patients with SCI who subsequently attained mobility
Carbone, Chin, Lee et al. (2014)	Complete injury	Is associated with increased risk for fractures
Bethel, Weaver et al. (2016)	Duration of injury	Longer duration of injury was associated with increased fracture risk
Bethel, Weaver et al. (2016)	paraplegia	Associated with an increased risk of lower extremity fractures but a decreased risk of upper extremity fractures
Bethel, Weaver et al. (2016)	Complete injury	Associated with an increased risk of lower extremity fractures but a decreased risk of upper extremity fractures
Morse et al. (2016)	Ambulatory status	Patients using a wheelchair more than 50% of the time lost BMD at the knee at a rate 1.45% of that of ambulatory patients
Abderhalden et al. (2017)	Chronic injury > 2 years	80% of 500 pts with chronic injury had either osteopenia or osteoporosis
Medication risk factors
Carbone, Chin, Lee et al. (2013a)	Opioid use	Was significantly associated with risk of lower extremity fractures. Shorter duration of use (<6 months) and higher doses were positively related to fracture risk
Carbone, Chin, Lee et al. (2013b)	Anticonvulsant use	There are significant positive relationships between fractures and overall use of anticonvulsants, use of the benzodiazepine subclass, polytherapy compared with monotherapy, Temazepam, alprazolam, and diazepam
Carbone, Chin, Lee et al. (2014)	Thiazide use	Was associated with a one quarter reduced risk of lower extremity fracture at any given point in time

Table 4 provides studies that examined the role of exercise in management of osteoporosis in SCI. Three of the 10 studies found no change following exercise, two involved standing and one used a standing frame with stimulation. The other 7 studies reported either increased BMD or less BMD loss than controls. These studies used walking, standing or other physical activity interventions. The data suggest that exercise may be beneficial, but details regarding type, frequency and duration are less clear.

Table 4. The role of exercise in treatment of osteoporosis.

Authors	Intervention or treatment	Frequency of intervention or treatment	Duration of intervention, treatment, or study follow-up	Results (unless otherwise stated, results state the effect of the intervention or treatment)	Anatomic Location
Needham-Shropshire et al. (1997)	Walking frame with stimulator	n.s.	Series of 32 treatments with additional 8 weeks	Did not change BMD relative to baseline	Hip (femoral neck, Ward's triangle, and greater trochanter)
de Bruin et al. (1999)	Standing with or without walking	1 hour/day, ≥ 5 days/week	n.s.	Decreased loss in BMD compared to immobilized controls	tibia
Goktepe et al. (2004)	Wheelchair basketball	n/a	n/a	BMD was higher in wheelchair basketball players than age & duration of injury-matched controls	Radius
Ben et al. (2005)	Standing	30 minutes/ session, 3 times per week	12 weeks	Did not reduce bone loss in trained leg compared to untrained leg	Proximal femur
Alekna et al. (2008)	Standing	1 or more hours/day, ≥ 5 days/week	2 years	Greater increase in BMD from baseline relative to no-standing matched controls	Lower extremities
Goktepe et al. (2008)	Standing	More than one hour per day in one group, less than one hour per day in another group.	n/a	No differences in BMD among patients standing for more than one hour per day, less than one hour per day, and non-standing patients	Lumbar spine and proximal femur
Chain et al. (2012)	Self-reported physical activity	n/a	n/a	Onset of physical activity and the number of hours of exercise were positively associated with BMD	Lumbar spine and radius
Lichy et al. (2012)	Robotic body weight-supported treadmill training	3 times/ week	3 months	Lower BMD in the more neurologically impaired leg	Distal femur and proximal tibia
Astorino et al. (2013)	Intense activity-based therapy	2-3 hours/ day, ≥ 2 ds/ week	6 months	Decreased BMD loss compared to what would normally be expected in persons with recent SCI	Lower extremities
Karimi et al. (2013)	Walking with an orthosis	n.s.	n.s.	A higher percentage of loads were transmitted by the body than the orthosis, suggesting that walking with orthosis could improve BMD	Lower extremities
Kostovski et al. (2015)	Self-reported physical activity	n/a	n/a	Self-reported levels of physical activity in the three months preceding and one year following the SCI in one study were correlated with a reduction in bone loss	Proximal femur and lumbar spine

n/a, not applicable; n.s., not specified.

We identified 14 papers in which authors examined the use of FES on osteoporosis (see Table 5). Twelve used FES cycling and two examined FES rowing. Results were somewhat mixed; 4 studies did not find BMD improvement as a result of FES, while the others found increased BMD due to FES or reduced BMD loss compared to control. One study found that the effects of FES cycling were sustained for one year after stopping or reducing FES cycling.

Table 5. The role of FES in treatment of osteoporosis.

Authors	Intervention or Treatment	Frequency of Intervention or Treatment	Duration of intervention, treatment, or study follow-up	Results	Anatomic location
Leeds et al. (1990)	FES cycling	Max. 30  min, 3 days/week	6 months–3 years	Intervention did not increase BMD from baseline	Proximal femur
BeDell et al. (1996)	FES cycling	3–4 times/week (with quadriceps strengthening exercise)	8 weeks	Did not increase BMD from baseline	Lumbar spine and hip
Belanger et al. (2000)	FES cycling (one limb with resistance and the other without resistance)	One hour/session, 5 days/week	24 weeks	Training increased BMD from baseline compared to controls with no difference by training type (with or without resistance)	Proximal tibia & distal femur
Eser et al. (2003)	FES cycling	Maximum of 30 minutes, 3 times/ week	6 months	The intervention group did not show decreased cortical BMD loss relative to non-cycling controls	Tibia
Shields & Dudley-Javoroski (2006)	FES cycling	5 sessions/week	≥ 2 years	BMD higher in trained leg than untrained leg	Leg
Clark et al. (2007)	FES cycling	Discontinuous	5 months	Did not show decreased BMD loss at 6 months relative to non-cycling controls	Total body, spine, or hip
Dudley–Javoroski & Shields (2008)	FES cycling (unilateral soleus training)	Over 8,000 soleus contractions per month	4.5–6 years	BMD higher in trained leg than untrained leg	Tibia
Frotzler et al. (2009)	FES cycling	n.s.	1 year	Benefits of FES cycling were partly preserved after stopping or reducing FES for 1 year	Distal femur
Griffin et al. (2009)	FES cycling	Max. 30  min, ≤30 minutes, 2–3 times per week	10 weeks	Intervention did not increase BMD from baseline	n.s.
Forrest et al. (2010)	FES cycling	3-4 times per week (60 sessions total)	3-5 months	Decreased BMD loss compared to standing alone	Total hip & proximal femur
Groah et al. (2010)	FES cycling	1 hour, 5 days/ week, plus usual care	6 weeks	Intervention decreased BMD loss relative to patients randomized to usual care alone	Distal femur
Dolbow et al. (2014)	FES cycling	Average of 2.9 sessions/week	12 months	Increased BMD from baseline; decreased total body fat mass, increased lean mass	Total body
Gibbons et al. (2014)	FES rowing	n.s.	> 8 years (prior to study)	Trabecular BMD higher in FES-rowing trained participant than in untrained SCI patients but lower than in healthy controls	Proximal tibia
Deley et al. (2017)	3 months of FES-strengthening, then 9 months of FES-rowing	3 times per week	1 year	Increased BMD from baseline	n.s.

n.s., not specified.

Table 6 provides information on pharmacological therapies for bone loss over 6 to 24 months. Five studies examined alendronate; three found an increase in BMD over baseline. In one case report and one case series, alendronate use increased BMD at the hip. In a randomized control study, alendronate increased total body BMD compared to the control group. One study found stable BMD at the hip and the tibia compared to controls who lost BMD, and another found less bone loss at the hip in the treatment group. Zoledronic acid was tested in three studies, two found decreased bone loss the hip in the treatment group. However, the third study, which included measurements at the knee, a critically important site of fractures in SCI, actually found loss of BMD in the treatment group versus the control. A study of denosumab found increased BMD in the hip while a study of teriparatide found no change in BMD at the hip.

Table 6. Pharmacological therapies.

Authors	Intervention or treatment	Duration of intervention, treatment, or study follow-up	Results	anatomic location
Alendronate
Sniger & Garshick et al. (2002)	10 mg/day	2 years	Increased bone mass from baseline	Lumbar spine (greatest increase) and hips
Bubbear et al. (2004)	10 mg/day	12–24 months	Increased BMD from baseline	Lumbar spine (greatest increase), total hip, & femoral neck (absolute hip BMD low)
Zehnder et al. (2004)	10 mg /day with 500 mg Calcium/d	24 months	BMD at the distal tibial epiphysis, tibial diaphysis and total hip remained stable in the experimental group whereas it decreased significantly in the control group with calcium alone compared to baseline	Hip and tibia
de Brito et al. (2005)	10 mg /day with 1000 mg Calcium/d	6 months	Intervention lead to greater increase in BMD from baseline relative to calcium alone	Total body
Gilchrist et al. (2007)	70 mg/week with Vitamin D	3 months	Reduced bone loss compared to placebo controls. These effects were sustained 6 months after cessation of alendronate	Hip (all areas)
		12 months		total hip, femoral neck, trochanter, and femoral shaft
Zoledronic Acid
Shapiro et al. (2007)	Single dose	6 months	Decreased bone loss & maintained parameters of bone strength compared to placebo controls	Proximal femur
		12 months	Stabilized BMD compared to placebo controls	Intertrochanteric region and femur shaft but not femoral neck
Bubbear et al. (2011)	4 mg once	1 year	Decreased BMD loss compared to usual care alone	Spine and total hip, but not femoral neck
Bauman et al. (2015)	5 mg once within 16 weeks of injury	6 & 12 months	Increased BMD loss compared to no-therapy controls matched for duration & completeness of injury, age, & gender	Knee
			Reduced bone loss compared to controls	Hip
Denosumab
Gifre et al. (2016)	60 mg every 6 months with Calcium & Vitamin D	1 year	Increased BMD and decreased bone turnover markers (with no adverse events noted)	Total hip, femoral neck, & lumbar spine
Teriparatide
Gordon et al. (2013)	20 µg/day with a robotic-assisted stepping device 3 times/week, followed by 6 months of teriparatide alone	12 months	No significant change in BMD	Lumbar spine or hip

Surgical and nonsurgical management approaches to fracture management is reported for 15 studies in Table 7. In general outcomes were good and similar between approaches. One exception may be for hip fractures for which surgery appeared to result in better outcomes. There were cases where surgical intervention was required due to delayed healing following conservative management. Studies are needed to determine which fractures and patient characteristics predict delayed healing. These may be the cases that would benefit from immediate surgical intervention.

Table 7. Surgical vs. nonsurgical approach to management of osteoporotic fractures.

Study	Number of Fractures	Fracture characteristics	Approach	Outcomes	Comments
El Ghatit et al. (1981)	2	1-Displaced femoral fracture 2-Comminuted proximal and distal tibia fracture	Nonsurgical: Posterior splinting with a Mud bed	No complications with good healing reported	none
Freehafer et al. (1981)	27	lower extremity	Nonsurgical: Splints made of soft materials	Rapid union with no important complications	This approach allowed good healing with early rehabilitation.
Nottage et at. (1981)	18	Site: Hip: 9, femoral shaft: 6, fibula: 1, tibia: 2	Majority were treated surgically. Open reduction with internal fixation was most commonly done for hip fractures.	Complication rates were high in the follow up periods of 2 years, (20%–40%) with open or closed treatment of extremity fracture	Fractures treated by nonoperative splinting and skin care healed well even in patients with long-standing paraplegia.
Baird et al. (1984)	1	Distal femoral shaft	Leg lengthening device with external fixation (Wagner device)	11 weeks after application of the device, the fracture showed radiographic signs of healing	Early return to prefracture functional capacities and reduced hospital stay may be achieved.
Baird et al. (1986)	9	Femoral shaft fractures: 8 were closed, 1 was open	Nonsurgical: External fixation applied 1–6 weeks post fracture. Mean treatment period was nine weeks	Fracture healed in 7 out of 8 pts who completed treatment. Two complications occurred. 1-Superficial pin-track drainage resolved with antibiotics. 2- communition of the distal femoral fracture fragment when the distal pins were inserted	7 patients were able to resume using their wheelchair during treatment of their fracture
Garland et al. (1986)	34	Tibia	Non-operative Early open reduction and internal fixation Type III open injuries treated with either surgical or non-surgical methods.	Non noperative fractures healed at a prolonged rate, while open reduction and internal fixation enhanced the rate and time to union. Open fractures had delayed unions and non-unions regardless of whether operative or conservative treatment was used	Open reduction and internal fixation is a justifiable alternative to nonoperative treatment in the uncomplicated tibia fracture regardless of neurologic lesion for improved medical and fracture care.
Garland et al. (1988)	53	Upper extremity (long bone) fractures	24 fractures treated surgically and 29 conservatively.	Similar outcomes irrespective of treatment modality except for, conservative management of displaced radial fractures was associated with higher frequency of non-union	Outcomes of management assessed were union, avoidance of angulation deformity and preservation of pain free range of motion. Surgical approach maybe associated with less medical complications like thrombosis and decubitus ulcers.
Ingram et al. (1989)	33	Lower extremity, 6 were femoral neck fractures	29 fractures managed conservatively and 4 surgically.	Optimal results achieved with conservative management of fractures with soft splints except for femoral neck fractures	5 femoral neck fractures were initially treated conservatively, two required surgeries later and one had a non-union. The remaining two healed well.
Sugi et al. (2012)	11	Lower extremity fractures	Surgical approach: 7 fractures of distal femur and proximal tibia required ORIF with locked plates and screws. 4 mid-shaft tibia fractures required intramedullary nailing.	All fractures achieved complete union, and no complications were reported. The average follow up was 28 months	Surgical approach for LE fractures can improve quality of life post injury and facilitate a rapid return to activities.
Bishop et al. (2013)	n.s.	Femur fractures. Majority are shaft or distal femur fractures	Surgically and non-surgically. The distal femur fractures in the SCI population were less likely to be operated on.	There were minimal differences in 3-month mortality between patients treated nonoperatively and operatively. Factors associated with higher mortality were older age, higher Charlson comorbidity index, and proximal femur fracture	The nonoperative group had a higher rate of adverse events than the surgical group.
Uehara et al. (2013)	19	Femoral and tibial fractures	Conservative management with hinged soft plastic knee brace for 80 days.	Successful treatment with average time for union of 80 days. No difference reported in ADLs pre- and post-fracture	Misalignment occurred in one patient only.
Bethel et al. (2015)	1979	345 upper extremity fractures 1634 lower-extremity fractures (incident)	Minority of patients (9.4%) underwent surgical treatment	No differences in mortality outcome at one year between surgical and nonsurgical approach	Amputations and disarticulations accounted for 19.7% of all surgeries (1.3% of all fractures), and the majority of these were done more than 6 weeks following the incident fracture.
Frotzler et al. (2015)	156	Lower extremity fracture. Site: Femur (60.9%) lower leg (39.1%) 75% of fractures were extra-articular	Conservative fracture management was applied in 16.7% of the cases and consisted of braces or a well-padded soft cast.	Fracture-associated complications were low and did not differ between surgical and conservative approach	Femur fractures were treated with locking compression plates in 48.2% of cases. Lower leg fractures were mainly managed with external fixation (48.8%)
Frotzler et al. (2015)	n.s.	Lower extremity fracture. Site: Femur (60.9%) lower leg (39.1%)	In almost 50% of femoral fractures, treatment with locking compression plates is used.	Complication rate with surgical treatment was low and similar to conservative treatment’s complication rate	None

n.s., not specified.

Complications of osteoporotic fractures in SCI are presented in Table 8. Complications were common, often related to delayed healing and non-union of fracture. Two of the seven studies were case reports of adverse outcomes following femur fractures including amputation. The few studies reporting surgical and conservative management of fractures suggest adverse events may be less after surgical management, but these were uncontrolled studies for which other factors may have contributed to outcomes.

Table 8. Complications from osteoporotic fractures.

Study	Approach: surgical vs. non-surgical	Fracture characteristics	Complications
McMaster et al. (1975)	Surgical approach applied to 19 fractures	4 hip fractures, 26 femur fractures, and 13 tibia fractures. 3 humerus and 2 forearm fractures also co-occurred with spinal cord injury	Delayed union (1), wound infection (1), anesthesia-related cardiac arrest (1) and urinary tract infection (2)
	Non-surgical approach applied to 30 fractures		Delayed union (5), non-union (2), malposition (2), cast pressure area (4)
Sobel et al. (1991)	Closed fixation in a long leg circular plaster cast	Supracondylar femur fracture	Vascular occlusion at Hunter’s canal requiring an above the knee amputation
Keating et al. (1992)	Conservative	Bilateral sub capital femoral fracture	Abundant callus formation following fracture
Morse et al. (2009)	87% of the fractures were treated with bed rest, with or without extension bracing. Surgery was performed in 3 cases (9%), all of which were proximal femur/hip fractures.	30 fractures in 30 patients: Site: most common was tibia/fibula (47.5%) then distal femoral metaphysis (20%) and then the proximal femur (15%)	Fracture non-union/delayed healing (25%) and pressure ulcers (25%) were the most common. Other included increased muscle spasms, pain, autonomic dysreflexia, and heterotopic ossification at the fracture site. Complications occurred frequently with 53% of the fracture events resulting in 21 documented complications
Angoules et al. (2012)	n.s.	n.s.	Lengthening hospitalizations, skin breakdown and increased spasticity from malunions
Wang et al. (2014)	Surgical management with intramedullary nailing	22 femoral fracture patients with SCI compared to 22 subjects with femoral fracture without SCI	Accelerated fracture healing (with an average time of 22.86 weeks) and enhanced callus formation were seen in patients with SCI
Wang et al. (2015)	Surgical treatment with closed, antegrade, reamed intramedullary nailing technique	8 Tibial and 7 femoral fractures in 15 patients with SCI compared to 15 tibial or femoral fracture patients without SCI	Callus production at 4 and 8 weeks post op was slower in patients with SCI compared to patients without SCI

n.s., not specified.

Discussion

Bone mineral density (BMD) by central dual energy x-ray absorptiometry (DXA) is a standard measure of bone health which can be used to identify osteoporosis, determine fracture risk and measure potential response to osteoporosis treatment. In persons with a SCI, one study showed that low lumbar and hip BMD by DXA at the time of injury is associated with a decline in BMD at these sites one year after injury.²¹

Identification of those persons with a SCI who are at highest risk for fracture is important to properly guide treatment strategies. In persons without a SCI, DXA is a powerful predictor of fracture, with fracture risk doubling for each standard deviation below peak bone mass.^22,23 Although DXA is able to globally predict fracture risk, DXA measurements done at a particular skeletal site are most predictive of fractures at that site (site-specific fracture prediction).²⁴

In clinical practice within the Veterans Administration, the most common sites of DXA measured for persons with SCI do not differ from the general population and include the lumbar spine and hip,²⁵ despite suggestions that BMD at the posterior anterior (PA) lumbar spine underestimates bone loss^26–28 and that hip contractures²⁹ may substantially limit technical quality of DXA measurements at the hip. Despite being rarely performed in current clinical practice, it is possible to measure BMD at the femur and tibia.³⁰ The distal femur is a more sensitive bone site for assessing bone loss by DXA than the proximal femur.³¹

Low BMD at the femoral neck and hip^32–34 and at the distal femur and proximal tibia, the most common sites for fracture in the SCI population,² was associated with higher risk for fractures.^35,36

Other skeletal changes that occur following injury also may impact fracture risk. Elevated bone resorption occurs following SCI,^37–39 and diminished osteoblastic activity occurs later following injury.⁴⁰ Torsional stiffness and strength at the proximal tibia in patients with SCI ASIA A-B decreases exponentially until reaching a new steady state within 2 years post-injury.^41,42 Endosteal resorption appears to be more significantly affected by SCI than periosteal surfaces.⁴³ Changes in bone geometry also occur following SCI.⁴⁴

A number of clinical factors including white race,⁷ female sex, prevalent fractures,⁴⁵ low BMI,⁴⁶ and family history of fractures.⁴⁷ SCI-specific risk factors including motor complete lesions⁴⁸ reduced ambulation,^48–50 paraplegia, and longer duration of injury are associated with lower BMD and increased risk for fracture.^{25,28,45,51–53}

Similarly, several medications that contribute to osteoporosis in the general healthy population without a SCI⁵⁴ also contribute to SCI-related bone loss. Use of opioids⁵⁵ and anticonvulsants⁵⁶ (in particular, enzyme-inducing ones) is associated with increased risk for fracture. Use of thiazides is inversely related to lower extremity fracture risk.⁵³ It has been hypothesized that chlorthalidone might also reduce risk of fracture in patients with SCI⁵⁷ although no studies have directly examined this. Chlorothiazide has been reported to decrease calciuria.⁵⁸ In one cohort, use of lipophilic statins was associated with gains in BMD at the knee.⁵⁹ Large doses of methylprednisolone administered acutely following spinal cord injury are theoretically a concern for fracture risk.⁶⁰

The relationship of medical comorbidities to fracture risk in persons with a SCI is controversial. One study reported that comorbid conditions were not significant predictors of fracture-related hospitalization.³ However, others have reported that the Charlson comorbidity score is positively related to incident fractures.^45,53

Secondary medical causes of osteoporosis are common in healthy postmenopausal women and elderly men, and laboratory testing is routinely done to identify them.^61,62 Disorders of calcium metabolism are among the most prevalent.^61,63

Women with a SCI may have similar risk factors for osteoporosis as in the general healthy population.⁶⁴ A retrospective cohort study of patients with chronic SCI and a T-score of less than –2 identified a secondary cause of osteoporosis in 28% of patients; the most common causes were parathyroid, thyroid disease, vitamin D deficiency, hypogonadism, or chronic liver disease.¹² Craven et al. recommended screening for these abnormalities in patients with SCI.⁶⁵ Other studies suggested that patients with a SCI who have low vertebral BMD should undergo a workup for secondary causes of osteoporosis as this is not routinely expected.^66,67

Low parathyroid (PTH) levels and hypercalcemia occur acutely following complete SCI,⁶⁸ although PTH tends to increase with duration of the complete SCI.² In 45 men and women with a chronic SCI, there was a non-significant trend for higher levels of PTH to be inversely associated with cortical thickness at the tibia.⁶⁹ In patients with a chronic SCI, PTH levels at 36 months were inversely associated with BMD at the femoral neck and femur in patients with paraplegia and tetraplegia with different extents of injury.⁷⁰ In one report of patients with an acute complete SCI, baseline levels of PTH and vitamin D were not significantly associated with BMD at the lumbar spine or hip at 12 month follow up.²¹

Vitamin D deficiency has been noted in the majority of patients with chronic and acute SCI,^71,72 which does not vary by sex⁷³ or ASIA scores.⁷⁴ Prevalence estimates range from approximately one-third to over 90% of patients.^75–77 Lower vitamin D levels were noted in African-Americans and in those with incomplete spinal cord injury.⁷⁶ However, the relationship of low levels of Vitamin D to fracture risk in patients with a SCI is uncertain. In a study of persons with chronic SCI, including both motor complete and incomplete injuries, serum 25-hydroxyvitamin D (25(OH) D) levels were not significantly associated with femur or tibia BMD.⁶⁹

Hypogonadism may be present in 25% of men with a SCI.⁴⁹ One study showed that free testosterone levels were not significantly associated with BMD of the lumbar spine, femoral neck, or distal femur; and surprisingly, total testosterone levels were inversely correlated with BMD at these sites.⁷⁸ No reports were identified that determined the relationship of disorders of calcium metabolism, hypogonadism, or other laboratory testing¹¹ to incident fractures in this population.

The mechanical unloading that occurs following SCI inhibits Wnt/beta-catenin signalling with upregulation of sclerostin.^79,80 Although sclerostin levels may initially increase after SCI, in the long- term circulating sclerostin may serve as a biomarker of osteoporosis severity and not a mediator of ongoing bone loss.⁸¹ Surprisingly, in chronic SCI, circulating sclerostin levels in one study were positively associated with BMD.⁵⁰

The muscle-bone-fat interface deserves consideration as muscle limb size and function have been correlated with bone strength in chronic SCI.⁸² There may be an even more substantial loss of muscle than cortical bone after SCI although muscle volume does correlate with cortical bone volume and bone mineral content.⁸³ In one study, adipose-derived adiponectin was inversely related to BMD of the knee in male wheelchair users with chronic SCI.⁸⁴

Several studies reported that physical activity and exercise attenuate bone loss and are associated with higher BMD.^85–92 Lower extremity fractures were more common in patients with a complete SCI.⁹³ Furthermore, Vitamin D levels and calcium intake were higher in active patients with SCI when compared to sedentary controls.⁸⁹ In contrast, others have not found substantial clinical benefits on bone from modified weight-bearing.^94–97

It is possible that while weight-bearing and increased mobility are clearly important for skeletal health in SCI,⁹⁸ they are not sufficient to maintain BMD of the lower extremity.^99,100 Further, there may be individual variations in response to activity based therapies.^101,102 Considerations for weight bearing interventions should include the timing of the intervention, length and intensity of the treatment (should be sufficiently long and high), and the need to be maintained over the long term.¹⁰³ Moreover, for individuals with SCI who participate in activities involving mechanical loading of the lower limbs, an assessment of risk, including history of fracture, BMD, and degree of loading, should be performed and recommendations made accordingly.¹⁰⁴

Exoskeleton use has substantially facilitated ambulation of patients with SCI^105–107 and is associated with increases in BMD of the tibia,¹⁰⁸ suggesting benefit for osteoporosis. However, fractures can occur while using these devices. In one study, 3.4% of persons with a SCI sustained a fracture while undergoing training to use a powered exoskeleton.¹⁰⁵ Additional case reports of fractures while using exoskeleton ambulatory devices^109,110 have been published, suggesting that low BMD may be a relative contraindication for the use of these devices.¹⁰⁵ Ongoing protocols using ReWalk devices (ReWalk Robotics, Inc. Marlborough, MA) are obtaining DXA scans to measure BMD around the knee; thus, further information regarding the utility of DXA measurements at the knee to predict fractures should be forthcoming. It is important to note these anticipated data may not include those with very low BMD as a T-score below –3.0 and knee BMD <0.70 gm/cm² are exclusion criteria for the ongoing ReWalk study.¹¹¹

Electrical stimulation and functional electrical stimulation (FES) may also increase loads on the skeleton without ambulation, although studies of their efficacy in addressing osteoporosis in persons with a SCI are mixed. In a meta-analysis of 19 studies of patients with acute and chronic traumatic SCI, FES cycling did not significantly decrease BMD loss in acute SCI; whereas, for persons with a chronic SCI, BMD demonstrated a significant increase.¹¹² Timing and persistence of use are important. In one study, when subjects started training within 6 weeks of SCI, torque and fatigue properties were preserved for more than 3 years.¹¹³ In another study, BMD returned to baseline after 6 months without FES.¹¹²

Some studies report no benefits with regular FES.^114–118 In contrast, others have reported that regular FES training in patients may improve BMD^{113,119–126} (see Table 5 for further descriptions of study findings). Studies assessing the efficacy of FES for osteoporosis treatment were often single-group, repeated measures designs with interventions. Samples ranged in size from 1 to 38, and most often included patients with SCI of at least one-year duration.

Considerations for FES efficacy in osteoporosis in SCI should include attention to duration, frequency, and power of the output.^{112,127–131} Moreover, it appears that if training is not maintained, then improvements in BMD are lost.¹³² Guidelines include intensities with loads of 1–1.5 times body weight, sessions of 3 or more per week for a duration of several months to 1 or more years, and safety considerations to prevent fractures¹³³ as fractures have been reported while using FES.¹³⁴ Another consideration that might partially explain the differences in efficacy of electrical stimulation reported among studies, is that DXA might not be sufficiently sensitive to capture changes.²⁸ Others have suggested that there are regional changes in BMD with FES which are not fully captured but may be important.¹³⁵

Mechanical vibration has also been utilized to address BMD loss, again with mixed results. Standing with combined whole body vibration may improve BMD in the trunk and spine of subjects with incomplete SCI.¹³⁶ However, in a 12 week study, mechanical vibration applied to the forearm did not significantly increase BMD at this site.¹³⁷ Likewise, no differences in either BMD or bone architecture occurred after 1 year of controlled vibration in patients with a chronic SCI.¹³⁸ A case study demonstrated that maximal isometric quadriceps testing with the subjects’ knee locked at 90 degrees can be hazardous and cause patellar fracture.¹³⁹

Similar to FES, the specifics of the whole-body vibration platform chosen including intensity and frequency of vibration, signal transmission, and joint position and angle may play a role in efficacy.^140,141

Other non-pharmacological therapies including pulsed ultrasound, electromyostimulation and acupuncture have not shown efficacy in significantly increasing BMD in SCI.^142–145

The majority of fractures in persons with a SCI occur during simple activities of daily living including transfer activities.^146,147 One study noted that 51% of fractures were due to a fall from wheelchair.³ In another study, the most common fracture cause was a fall; other causes included physiologic ranging/loading of the joint (23%), rolling in bed (4%) and lower limb twisting during transfers (2%).⁵⁹

Fall prevention is essential in this population. Education to promote safe wheelchair use has been suggested.¹⁴⁸ However, there are no studies on whether training in safe use of wheelchairs or other modalities for fall prevention can reduce incident fracture rates in persons with a SCI. There are studies underway examining fall prevention in those with incomplete SCI who use walking aids.¹⁴⁹

The estimated prevalence of cigarette smoking is between 19–40% in patients with a SCI.¹⁵⁰ 22% of Veterans with a SCI are current smokers.¹⁵¹ To date, no data directly tie cigarette smoking to incident fracture risk in persons with a SCI. Alcohol consumption patterns are similar to those of the general population relative to sex and age^152,153 although one report suggested that more than 50% of patients with a SCI had alcohol consumption patterns that ranged from moderate to heavy drinking.¹⁵⁴ A retrospective cohort study identified a high prevalence of significant alcohol consumption (17 servings per week) among patients with chronic SCI.¹² In one study, increased alcohol use was a significant predictor of hospitalization for a fracture in SCI patients.¹⁴⁸

Given the considerable smoking and alcohol use reported in the SCI population, counselling on smoking cessation and monitoring alcohol use should be a focus of care. Cigarette smoking may be harmful to bone density and fracture healing,¹⁵⁵ has direct toxic effects on bone, may alter calcium, cortisol and estradiol levels and increase levels of free radicals.¹⁵⁵ The effects of alcohol consumption on bone are dose dependent. High consumption increases osteocyte apoptosis, increases bone marrow fat, and changes body composition,¹⁵⁶ and may increase fall risk.¹⁵⁷ Limiting alcohol use in SCI may reduce the risk for fracture, and should be part of patient education regarding risk.¹⁴⁸

There are differences in calcium homeostasis between persons with a SCI, especially in regards to duration of SCI, and the general healthy population.⁷⁵ However, calcium supplements in doses of 500 mg¹⁵⁸ or 1 000 mg per day¹⁵⁹ in persons with a SCI have been reported to be safe and well tolerated.

In patients with both acute and chronic SCI, 2000 IU of vitamin D3 daily and 3.25 g calcium carbonate were sufficient to increase vitamin D levels to normal in 6 of 7 patients after 3 months of treatment.¹⁶⁰ In a randomized, placebo-controlled trial of 40 patients with chronic, complete SCI, treatment with 1-alpha-hydroxyvitamin D2 for 24 months increased leg BMD by 6% in the treated compared with the control group.¹⁶¹

Restriction of calcium intake after acute SCI should not be considered as this will further worsen the calcium and phosphate metabolism leading to hypercalciuria or hypercalcemia.^162,163 One clinical trial chose not to administer calcium or vitamin D supplements due to concerns about increased risk of renal calculi following SCI as hypercalciuria maybe be present in the acute phase.¹⁶⁴ However, increasing calcium intake to 1000 mg per day with liberalization of sodium intake did not affect urinary or serum calcium levels.⁴⁰ Importantly, with use of antiresorptive agents, hypocalcemia may occur. Thus, it has been suggested that before initiating these therapies, vitamin D levels should be in the normal range determined before drug initiation.¹⁶⁵

There is very little data on the relationship of other dietary intakes to osteoporosis for persons with a SCI. Higher protein intake is inversely associated with osteoporosis in the general population; however, this relationship may vary by source, with animal protein sources negatively associated with hip fractures in healthy men.¹⁶⁶ In contrast, in another study including men and women with a traumatic SCI, higher protein intake was associated with lower BMD of the lumbar spine.¹⁶⁷ The efficacy and safety of sodium restriction,¹⁶⁸ magnesium,¹⁶⁹ and vitamin K intake¹⁷⁰ for BMD and fracture prevention in SCI has not been reported. Since patients with a SCI are at increased risk for thrombosis relative to healthy controls,¹⁷¹ this could have implications as to whether vitamin K supplementation is safe in patients with a SCI.

There have been several reviews that have addressed the efficacy of pharmacological therapies for osteoporosis in SCI with BMD as the outcome measure.^{112,143,172–175} Bisphosphonates are the most common class of drugs that have been studied in SCI-related osteoporosis.^{158,159,176–178,164,179,180} Very few studies examined denosumab use and the changes of its target protein RANKL after SCI ^181–183 and one study addressed the use of teriparatide.¹⁸⁴

A systematic review and meta-analysis of 19 studies concluded that bisphosphonates are effective at treating SCI-related bone loss.^112,175 Another meta-analysis suggested that bisphosphonates were efficacious only if treatment was initiated immediately after SCI.¹¹² In contrast, others concluded that there was insufficient data to recommend bisphosphonates for fracture prevention in either acute or chronic SCI patients.¹⁷⁴

DXA has limitations and cannot accurately detect changes in volumetric measures of bone mineral or changes specifically in trabecular and cortical bone.¹⁸⁵ Modalities other than DXA including QCT might be considered as primary outcome measures to assess the efficacy of pharmacological measures in this population.¹⁸⁶ There is a need to study combination therapy (for example, an anabolic agent followed by an antiresportive agent) or in combination with mechanical stimulation.¹⁸⁷ It has also been suggested that the mechanisms that induce SCI-related osteoporosis are substantially different from postmenopausal osteoporosis for which all currently available agents were developed; perhaps drugs must be developed specifically for use in SCI-related bone loss.¹⁸⁸

Potential side effects from long-term use of bisphosphonates and denosumab include osteonecrosis of the jaw (ONJ)¹⁸⁹ and atypical femoral fractures (AFF)¹⁹⁰ which have prompted clinicians to consider “drug holidays” after 5 or 10 years of use.^191–193 There were no reports of ONJ or AFF following these therapies in patients with a SCI identified in these searches. However, most studies of bisphosphonate/denosumab use were of short duration, and both ONJ and AFF are more likely to occur with prolonged use.

Traditionally, management of osteoporotic extremity fractures in persons with a SCI was done non-surgically with satisfactory results reported.^194–198 More recent reports, however, suggest that surgical management of appendicular fractures in persons with a SCI may be increasing without any detrimental impact on mortality¹⁹⁹ or fracture related complications⁸ and may even be associated with fewer medical complications.²⁰⁰

Distal femur and proximal tibia fractures are the most common fracture sites in persons with a SCI.¹⁹⁹ Treatment of lower extremity fractures has been addressed in a few case series describing both the surgical approach²⁰¹ and the conservative approach.¹³⁴ Surgical management of femur fractures decreased the rate of complications but the mortality rate did not differ when compared to patients who received a non-surgical approach.²⁰² Several case reports noted good outcomes with both surgical and conservative management of femoral fracture.^8,42,203

Relative to tibia fractures, studies reported less orthopedic and medical complications²⁰⁴ and improvement in quality of life post-injury²⁰⁵ when fractures were treated surgically early compared to nonsurgical treatment. In a case series of 19 fractures, maintenance of activity of daily living was described with non-surgical treatment of tibial fractures.⁴²

Complications from fractures can arise from either surgical or nonsurgical management.^3,206,207 Reducing immobilization following surgical treatment should be considered to reduce bone loss.²⁰⁸ A fracture complication reported in one study was abundant callus formation following fracture²⁰³ in SCI.²⁰⁹ However, a study of 15 patients with SCI who had surgery for a tibial or femoral fracture, found that callus production in these patients was actually slower compared to patients without a SCI.²¹⁰

Early surgical management for lower extremity osteoporotic fractures in SCI is becoming more prevalent and may be associated with better outcomes than traditional conservative approaches. However, to date, there are no large scale randomized trials addressing the efficacy and safety of management of appendicular osteoporotic fractures in persons with a SCI comparing surgical versus nonsurgical managements. Reports of fracture management are generally limited to case reports or case series. There are several factors to consider when deciding how to treat an extremity fracture in patients with a SCI. These include the potential impact on future function of the limb and the overall function of the patient.²¹¹ Fractures may affect patients’ levels of function differently.²¹² In non-ambulatory patients, exact anatomical reduction of fractures may not be necessary, but maintaining limb length and correcting rotation and pelvic obliquity are important. In ambulatory patients, maintaining limb length and minimizing deformity are important.¹³⁴ One study suggested that while shortening and angulation of healed fractures were acceptable in non-ambulatory patients, rotational deformities might be unacceptable due to prolonged sitting.¹⁹⁶ One editorial review suggested that perhaps surgical treatment of lower extremity fractures may be useful in preventing major fracture related complications.²¹³

Physical therapy is often provided following fracture treatment to regain or improve function of the affected limb. In patients with a SCI, PT is an essential aspect of the rehabilitation process²¹⁴ and is guided by domains identified by the International Classification of Functioning, Disability and Health but there is very little information on effectiveness in SCI.²¹⁵ In one case report, 6 months after sustaining a distal femur fracture due to falling from a wheelchair, a patient was “cleared by the orthopedic surgeon” to safely resume cycling with FES.²¹⁶ There is a planned Cochrane review to assess the effects of physical therapy interventions for the prevention of fractures after SCI which lists a number of possible interventions that might be helpful;²¹⁷ however, this review has yet to be completed (personal communication, Rada, 2017).

Limitations

There are several limitations to this work. To start, scoping studies themselves have limitations as the quality of evidence is not considered.^13,218 A priori, the authors had decided to perform a scoping study because many of the studies in osteoporosis in SCI are simply descriptive, have small sample sizes, and lack control groups. In part, because of this, to broaden the scope of available knowledge, the grey literature was incorporated including abstracts, theses, and other material that may not be widely disseminated. A second limitation is that there are a number of registered clinical trials examining the effects of pharmacological therapies for osteoporosis including risedronate,²¹⁹ alendronate,²²⁰ alendronate in those previously treated with teriparatide,²²¹ teriparatide,²²² denosumab,²²³ and zoledronic acid,^224,225 some of which are actively recruiting, but do not have published results available for this review. Similarly, there is a report of a planned trial using a 12-week program of FES-assisted cycling versus passive cycling to prevent muscle atrophy and maintaining skeletal integrity after spinal cord injury.²²⁶ Further, we are aware that some articles of DXA, particularly related to fracture threshold, were not included in our scoping review. In the general population, fracture threshold is no longer considered to be a key facture, with the focus instead on the continuum of risk. This continuum also should be explored in the SCI population. A third limitation is that available studies in the literature do not consider prevention and treatment of osteoporosis separately, and this may be considered for future research.

Conclusions

DXA-defined low bone mineral density and a combination of clinical and demographic factors including Caucasian race, female sex, prior fracture history, medication use (corticosteroids, opioids, anticonvulsants) and SCI-specific characteristics including longer injury duration, paraplegia and completeness of injury can identify those at highest risk for incident osteoporotic fractures. Screening for Vitamin D deficiency, hyperparathyroidism, thyroid disease and hypogonadism should be considered in the workup for secondary causes of osteoporosis in SCI.

The effects of non-pharmacological measures including weight-bearing, electrical stimulation, functional electrical stimulation and vibration on SCI-induced osteoporosis are mixed, but these measures are insufficient as a sole modality for osteoporosis prevention. Ultrasound, electromyostimulation and acupuncture do not seem to play a role in preventing or treating osteoporosis. Further research is needed to explore whether restricted alcohol use or cigarette smoking reduce fracture risk in SCI. Bisphosphonates are the most common class of drug used to treat SCI-related osteoporosis, but data is mixed on their efficacy for BMD outcomes, and no study has shown a reduction in incident fractures with these medications. Surgical management for lower extremity osteoporotic fractures in SCI may be associated with fewer complications when compared to conservative management.

There are important gaps relative to the clinical management of osteoporosis in persons with a SCI. These include recommendations for treatment of osteoporosis, including non-pharmacological and pharmacological interventions. There is a clear need for trials designed to directly assess fracture prevention as the outcome measure for efficacy of pharmacological treatment of osteoporosis in patients with a SCI. A second major gap area is identification of which patient and fracture characteristics predict whether surgical or conservative management of lower extremity fractures should be done. Given the gaps in the existing literature and the difficulty in performing large scale randomized trials to address these issues, many of these gap areas may be best addressed using a multifaceted approach that includes a combination of literature synthesis, large longitudinal observational studies, and expert opinion.

Acknowledgements

The authors would also like to thank Billy Houke, Medical Librarian, Charlie Norwood VA, Augusta, Georgia for his assistance with this project.

Disclaimer statements

Contributors None.

Funding This work was funded by the Department of Defense (DOD) Grant #SC150092 and Health Services Research & Development, Department of Veterans Affairs Grant#: 1 I01 HX 002090-01A2.

Declaration of interest None.

Conflict of interest All authors who contributed to this paper have no conflict of interest.

Context Osteoporotic fractures in persons with a spinal cord injury or disorder (SCI/D) contribute to excess morbidity and mortality. However, to date, it is unclear how to best manage osteoporosis and osteoporotic fractures in this population.

Ethics Approval None.

ORCID

B. Catharine Craven http://orcid.org/0000-0001-8234-6803

Cara Ray http://orcid.org/0000-0002-1383-8801

Laura Carbone http://orcid.org/0000-0003-3370-4680