Association of Pneumonia, Wound Infection, and Sepsis with Clinical Outcomes after Acute Traumatic Spinal Cord Injury

Abstract

Pneumonia, wound infections, and sepsis (PWS) are the leading causes of acute mortality after traumatic spinal cord injury (SCI). However, the impact of PWS on neurological and functional outcomes is largely unknown. The present study analyzed participants from the prospective North American Clinical Trials Network (NACTN) registry and the Surgical Timing in Acute SCI Study (STASCIS) for the association between PWS and functional outcome (assessed as Spinal Cord Independence Measure subscores for respiration and indoor ambulation) at 6 months post-injury. Neurological outcome was analyzed as a secondary end-point. Among 1299 participants studied, 180 (14%) developed PWS during the acute admission. Compared with those without PWS, participants with PWS were mostly male (76% vs. 86%; p = 0.007), or presented with mostly American Spinal Injury Association Impairment Scale (AIS) grade A injury (36% vs. 61%; p < 0.001). There were no statistical differences between participants with or without PWS with respect to time from injury to surgery, and administration of steroids. Dominance analysis showed injury level, baseline AIS grade, and subject pre-morbid medical status collectively accounted for 77.7% of the predicted variance of PWS. Regression analysis indicated subjects with PWS demonstrated higher odds for respiratory (odds ratio [OR] 3.91, 95% confidence interval [CI]: 1.42-10.79) and ambulatory (OR 3.94, 95% CI: 1.50-10.38) support at 6 month follow-up in adjusted analysis. This study has shown an association between PWS occurring during acute admission and poorer functional outcomes following SCI.

Introduction

Annually, an estimated 53.4 persons per million in the United States sustain a traumatic spinal cord injury (SCI).¹ Studies suggest 17% of affected individuals die during acute hospital admission primarily from complications related to severe infections, with pneumonia, wound infection, and sepsis (PWS) being the most common causes.^1,2 PWS account for the majority of acute adverse events encountered by clinicians during the management of SCI, with the reported incidence rates varying between 23 and 80%.^3–8 Individuals with SCI are vulnerable to secondary infections for reasons such as pathogens in the hospital environment, impaired airway defences due to intubation, ineffective clearance of airway secretions, reduced mobility, nutritional deficiencies, and often extensive surgical procedures. Additionally, acute SCI immunodeficiency syndrome, unrelated to the aforementioned predisposing factors, has been hypothesized to exist, mediated through a complex process involving neuronal as well as humoral dysregulation of immune function.⁹ Other factors such as altered gut biome may also prove to be contributory.¹⁰

Aside from acute mortality, SCI-associated infections carry a high morbidity and economic burden. However, there is paucity of studies to determine what impact acute infections might have on long-term neurological outcome and clinical outcomes.^11,12 This information is important because affected subjects presently are surviving past acute admission and living longer. Moreover, robust evidence is lacking to dictate preventative and treatment strategies to target the most severe infections. Often, SCI-associated infections are misdiagnosed, unrecognized early or inadequately treated, which potentially compounds the risk of adverse outcomes.² Previous studies have attempted to identify risk factors for the development of secondary infections, with the majority of studies focusing on the subacute or rehabilitation phase of care.¹³ Less data are available on important risk factors during the acute inpatient admission when PWS (the most severe secondary infections) are encountered.

In this study we have used a combination of two large prospective SCI data sets to investigate two questions: 1) what is the impact of acute infectious complications on long-term functional and neurological outcome after traumatic SCI? and; 2) what are the key clinical predictors of acute infectious complications after SCI?

Methods

Study population

For this study, we pooled the data of SCI patients from the North American Clinical Trials Network (NACTN) prospective longitudinal registry¹⁴ and the Surgical Timing in Acute SCI Study (STASCIS).¹⁵ The patients were admitted into acute spinal units at university-affiliated North American hospitals. The NACTN cohort was enrolled at 11 hospitals between 2005 and 2017. The STASCIS cohort was enrolled at 6 hospitals between 2002 and 2009. Eligible subjects included those from whom informed consent was obtained, who were 18 years of age or older, and who had baseline American Spinal Injury Association Impairment Scale (AIS) grade assessment according to the International Standards for Neurological Classification of Spinal Cord Injury ([ISNCSCI], Revised 2011) within 7 days of injury.¹⁶ Ethics clearance and informed consent were duly obtained.

Independent variables

Information was recorded in the NACTN and STASCIS data sets on acute complications including PWS. The diagnosis was made by the patient's clinical team on the basis of clinical features and confirmatory microbiological or radiological findings, as applicable. Pneumonia was diagnosed if there was evidence of lung tissue inflammation from infectious microorganisms and associated parenchymal disease from radiological findings. A diagnosis of post-operative wound infection was made at the discretion of the attending surgeon based on clinical features with confirmatory microbiology at any point during 1-year follow-up. Sepsis was defined according to the criteria presented at a consensus conference by the American College of Chest Physicians/Society of Critical Care Medicine.¹⁷ For the analysis, the pooled cohort was classified into two categories based on whether a subject had at least one of these events or did not have PWS during the acute care.

Covariates

The following demographic, clinical, and treatment-related characteristics were collected in both NACTN and STASCIS and analyzed in the present study as potential confounders: age (continuous variable), sex, pre-morbid medical status (presence or absence of a history of hypertension, cardiovascular disease, diabetes, chronic pulmonary conditions, malignancy, or drug abuse), presence of penetrating injury (Yes or No), baseline AIS grade assessed within 7 days of injury (analyzed as ordinal variable), injury neurological level (cervical, thoracic, or lumbosacral), acute 24-h administration of high-dose intravenous (IV) methylprednisolone sodium succinate (MPSS; Yes or No), time from injury to decompressive surgery (dichotomized as ≤24 h or >24 h), and length of hospital stay (continuous variable).

Outcomes

The primary outcome was functional status including respiratory function and ability to walk indoors at 6-month follow-up. The functional outcomes were assessed with the respective subscales of the Spinal Cord Independence Measure (SCIM II)¹⁸ (see Supplementary Table S1 for a description). For the purposes of the analysis, the SCIM subscales were dichotomized. As shown in Supplementary Table S1, respiratory function was dichotomized as assisted breathing versus independent breathing. Ability to walk indoors (Indoor mobility) was dichotomized as dependent ambulation versus independent ambulation. Data on SCIM were available for the NACTN cohort but not for the STASCIS cohort. The secondary end-point was neurological outcome assessed as AIS grade improvement ≥1 point at 6 months. The AIS grade is a five-category ordinal scale of the clinical severity of SCI (see Supplementary Table S2 for definition).

Statistical analysis

Group comparisons were made with respect to demographic, clinical, and treatment variables using the Fisher's exact test for categorical and Mann-Whitney test for continuous variables, with examination of distributional assumptions. Predictors of PWS were assessed in a logistic regression model and ranked using dominance analysis. Candidate predictors included the seven covariate variables described above, which were included in the model based on consensus of their potential importance relative to the outcome. Dominance analysis was used to estimate the proportionate contribution each predictor makes to the explained variance of the dependent variable after considering the direct effect of the predictor and its effect when combined with the other variables in the regression model.^19,20 For the present analysis, the McFadden R² was used as the estimator of explained variance.

Penalized logistic regression models were fitted with PWS (Yes/No) as the independent variable to assess the association with AIS improvement ≥1 grade, breathing or ambulation outcomes at 6-month follow-up. The analysis for neurological improvement was adjusted for baseline AIS grade, age, and sex, and stratified for the study. The estimates in the NACTN and STASCIS data sets were pooled using a random effects meta-analysis model to account for differences in study design and between-study heterogeneity. Because SCIM data were available for the NACTN cohort only, the functional outcomes analysis was limited to NACTN. The functional outcome analysis was adjusted for baseline AIS, injury level, patient age, sex, and time to surgery. A sensitivity analysis focused on the cohort with cervical SCI who had non-penetrating injuries (blunt or crush) to further reduce heterogeneity and to reflect the patient group more likely to be included in clinical trials of SCI. Effect size is reported as odds ratio (OR) with 95% confidence interval (CI). Significance level was set at 5%. The analysis was performed with Stata software, version 13.1 (Statacorp, TX).

Results

Descriptive analysis and risk factors for PWS

Among the 1299 subjects for inclusion in the study, 1037 had an initial ISNC SCI evaluation within 7 days of injury (Fig. 1). The distribution of patients according to their time of baseline neurological examination from initial injury is shown in Supplementary Figure S1. Six-month follow-up data were available for 758 subjects (58.4%) for analysis of neurological outcome, and for 465 (56.1%) NACTN subjects for functional outcome analysis. Supplementary Table S3 compares the baseline data of subjects who had follow-up data with those without follow-up data. Subjects with missing follow-up data had higher frequencies of complete injuries, cervical SCI, were more frequently administered MPSS, and had a longer length of hospital stay, compared with subjects with follow-up data (p = 0.001).

FIG. 1.

Flowchart of patients in the analysis. NACTN, North American Clinical Trials Network; STASCIS, Surgical Timing in Acute SCI Study.

During acute hospitalization, 180 patients (14%) developed PWS (NACTN = 167 [20%]; STASCIS = 13 [3%]). Among this group, 146 (81%) were classified as pneumonia, 24 (13%) as wound infection, and 10 (6%) as sepsis. The baseline characteristics of the patients are shown in Table 1. The highest rate of PWS was seen among patients with AIS grade A injuries (61%) and the lowest rate among those with AIS grade D (11%), indicating a higher incidence of PWS with more severe injury (p < 0.001). Compared with those without PWS, participants with PWS were mostly male (76% vs. 86%; p = 0.007), or presented with mostly AIS grade A injury (36% vs. 61%; p < 0.001). As shown in Table 1, there were no statistical differences between participants with or without PWS with respect to time from injury to surgery, and administration of steroids. Table 2 presents results of the dominance analysis on the relative importance of predictors of PWS. In order of hierarchy, the three strongest predictors of PWS among the eight that we investigated were injury level, baseline AIS grade, and pre-morbid medical status (inclusive of a history of high blood pressure, diabetes, heart attack, malignancy, pulmonary disease, cerebrovascular disease, smoking history, or drug abuse). Collectively, injury level, baseline AIS grade, and pre-morbid medical status accounted for 77.7% of the predicted variance of PWS (33.35%, 32.52%, and 11.87%, respectively).

Table 1.

Descriptive Analysis Comparing the Groups with and without Pneumonia, Wound Infection, or Sepsis

Variables	Without PWS	PWS	P-value
Age, year, median (IQR)	47 (30, 60)	46 (28, 57)	0.16
Sex: Male	811 (75.6)	150 (85.7)	0.003
Female	262 (24.4)	25 (14.3)
Injury type: Penetrating	36 (5.7)	6 (5.0)	0.73
Non-penetrating	600 (94.3)	153 (95.0)
Baseline AIS: A	302 (35.5)	77 (60.6)	<0.001
B	121 (14.2)	19 (15.0)
C	155 (18.2)	17 (13.3)
D	273 (32.1)	14 (11.0)
Level: Cervical	767 (80.5)	135 (80.4)	0.04
Thoracic	137 (14.4)	31 (18.4)
Lumbar/Sacral	49 (5.1)	2 (1.2)
Steroid: Yes	484 (44.9)	68 (38.9)	0.14
No	593 (55.1)	107 (61.1)
Timing: ≤24 h	433 (45.7)	72 (48.3)	0.55
>24 h	514 (54.3)	77 (51.7)
Surgery: ANT approach	306 (31.4)	58 (35.1)	0.25
POST approach	463 (47.4)	81 (49.1)
Both	207 (21.1)	26 (15.8
LOS, days, median (IQR)	11 (7, 19)	25 (17, 38)	<0.001
Breathing: Assisted	11 (3.3)	13 (17.8)	<0.001
Spontaneous	321 (96.7)	60 (82.2)
Ambulate: With aids	214 (56.2)	71 (85.5)	<0.001
Without aids	167 (43.8)	12 (14.5)

Missing data excluded.

AIS, American Spinal Injury Association Impairment Scale; ANT, anterior approach; IQR, interquartile range; LOS, length of stay; POST, posterior approach; PWS, pneumonia, wound infection, or sepsis.

Table 2.

Importance of Predictors of Pneumonia, Wound Infection, or Sepsis during Acute Hospitalization for SCI

Predictor	Rank	Standardized weight
Injury level	1	0.3335
Initial AIS grade	2	0.3252
Pre-morbid status	3	0.1187
Steroid use	4	0.1051
Timing of surgery	5	0.0778
Sex	6	0.0362
Age	7	0.0023
Penetrating injury	8	0.0012

Standardized weight is the general dominance weight from McFadden R² normalized or standardized to be out of 100%. The standardized weights might not add up to 1 due to rounding errors.

AIS, American Spinal Injury Association Impairment Scale; SCI, spinal cord injury.

Functional outcomes

Of the 465 patients with outcomes data in NACTN, 285 (61.4%) needed some aid to walk indoors, whereas 24 of the patients (5.9%) required some breathing support 6 months post-SCI. As shown in Table 1, the proportion of patients who needed respiratory support was higher in the group with PWS compared with those without (18% vs. 3%). Similarly, the proportion of patients who needed a walking aid indoors was higher in the group with PWS (86% vs. 56%). In unadjusted analysis, PWS was associated with higher odds of needing respiratory support (OR 6.24, 95% CI: 2.71-14.35) or ambulation support (OR 4.47, 95% CI : 2.37-8.41). Results of the adjusted analysis are shown in Table 3. Upon adjusting for baseline AIS, injury level, patient age, sex, and time to surgery, the group with PWS had a four-fold higher odds for respiratory (OR 3.91, 95% CI: 1.42-10.79) or ambulatory support (OR 3.94, 95% CI: 1.50-10.38) at 6 months. Among the patients with cervical SCI, PWS was also associated with poorer respiratory (OR 2.93, 95% CI: 0.98-8.78) and ambulatory (OR 5.19, 95% CI: 1.57-17.20) outcomes. To assess whether the association of PWS with functional outcomes differ with injury severity or neurological level, we included interaction terms in the adjustment models. With respect to the need for respiratory support, we found no interaction between PWS and AIS grade (p = 0.2) or PWS and injury level (p = 0.59). We also noted a non-significant interaction for ambulation (PWS and AIS grade, p = 0.42; PWS and injury level, p = 0.53).

Table 3.

Results of the Adjusted Analysis Examining the Association of Pneumonia, Wound Infection, and Sepsis with Functional Outcomes

Predictors	Breathe with support OR (95% CI)	Ambulate with aid OR (95% CI)
PWS: No	Ref	Ref
Yes	3.91 (1.42-10.79)	3.94 (1.50-10.38)
Sex: Female	Ref	Ref
Male	0.62 (0.19-1.98)	0.79 (0.36-1.72)
Level: Cervical	Ref	Ref
Thoracic	0.74 (0.18-3.12)	1.52 (0.64-3.60)
Lumbar/Sacral	0.91 (0.04-17.79)	3.36 (1.12-10.06)
Baseline AIS: A	Ref	Ref
B	0.66 (0.15-2.87)	0.25 (0.07-0.89)
C	1.24 (0.37-4.20)	0.06 (0.02-0.18)
D	0.24 (0.04-1.46)	0.02 (0.01-0.06)
Time to surgery: >24 h	Ref	Ref
≤24 h	1.91 (0.64-5.73)	1.36 (0.71-2.61)

AIS, American Spinal Injury Association Impairment Scale; CI, confidence interval; PWS, pneumonia, wound infection, and sepsis; OR, odds ratio.

Neurological outcomes

In the pooled NACTN/STASCIS data set, the group with PWS had a similar level of AIS conversion rate at 6 months (34.6% vs. 38.0%; p = 0.66), as shown in Table 4. When the analysis was stratified for AIS grade, we also noted a trend toward lower rates of AIS grade conversion among the group with PWS than among those without PWS (Table 4). Figure 2 shows the results of stratified and pooled analyses of the relationship between PWS and neurological outcome for all subjects and for those with cervical SCI only. There was a trend toward reduced AIS ≥1 grade point improvement among the group with PWS compared with the control group for the NACTN and STACSIS cohorts in the unadjusted analysis. In the adjusted analysis accounting for baseline AIS, age, and sex, the group with PWS had lower odds for AIS ≥1 grade improvement than the control group. Among the NACTN cohort, we noted a 27% lower odds for neurological improvement (OR 0.73, 95% CI: 0.39-1.35), and a 50% lower odds among the STASCIS cohort (OR 0.50, 95% CI: 0.11-2.20), although the difference did not reach statistical significance. The pooled estimate was OR 0.69, 95% CI: 0.39-1.22 in the analysis including all subjects; and OR 0.63, 95% CI: 0.33-1.21 for the subjects with cervical SCI only.

Table 4.

ASIA Impairment Scale Grade Conversion

	A	B	C	D	E	Total	% CR
Both
A	147	44	23	17	0	231	36.4
B	5	21	25	33	5	89	70.8
C	2	3	16	72	14	107	80.4
D	0	0	4	116	73	193	37.8
Total	154	68	68	238	92	620	37.6
Control
A	114	39	19	12	0	184	38.1
B	3	17	21	30	5	76	73.7
C	1	2	16	66	13	98	80.6
D	0	0	4	106	71	181	39.2
Total	118	58	60	214	89	539	38.0
PWS
A	33	5	4	5	0	47	29.8
B	2	4	4	3	0	13	53.9
C	1	1	0	6	1	9	77.8
D	0	0	0	10	2	12	16.6
Total	36	10	8	24	3	81	34.6

Baseline AIS in rows; 6-month AIS in columns.

AIS, American Spinal Injury Association Impairment Scale; CR; conversion rate; PWS, pneumonia, wound infection, and sepsis.

FIG. 2.

Forest plot showing a trend toward poorer neurological recovery at 6 months among the group with pneumonia, wound infection, and sepsis. AIS, American Spinal Injury Association Impairment Scale; CI, confidence interval; NACTN, North American Clinical Trials Network; OR, odds ratio; SCI, spinal cord injury; STASCIS, Surgical Timing in Acute SCI Study.

Discussion

The present study demonstrates an association between severe secondary infections occurring during acute care for SCI and poorer outcomes in breathing or ambulatory function. A trend toward a poorer neurological outcome among the cohort with PWS was also observed. The evidence was derived after analyzing the most contemporary, prospective multi-center cohort of SCI studied so far.

Previous research has raised concern of a likely association between PWS and poorer long-term survival or reduced prospect for clinical outcomes. Following analysis of subjects with cervical SCI in the National Spinal Cord Injury Database, U.S., (NSCID), Failli and colleagues¹² demonstrated a lower gain in ASIA motor score, lower gain in motor or sensory levels, and lower AIS conversion rates after 1 year among the cohort with pneumonia or post-operative wound infection, relative to the group without (the control). They found pneumonia or wound infection to be independent risk factors of lower achievement in upward AIS conversion (OR 1.89, 95% CI: 1.36-2.63; p < 0.0005) or gain in motor score (OR −8.21, 95% CI: 12.29 to −4.14; p = 0.005). In a separate publication,¹¹ the same group of researchers examined the relationship with functional outcome and noted lower gain in the Functional Independence Measure (FIM) motor subscore up to 5 years after SCI (−7.4 points, 95% CI: −11.5 to −3.3) and lower chances of 10-year survival (hazard ratio 1.65, 95% CI: 1.26-2.16) in adjusted analysis. Interpreting the clinical significance of the results of the NSCID study for functional outcome is challenging. Although the FIM has been widely applied to assess activities of daily living, its clinical utility is questioned for the SCI population.²¹ The motor subscale that was used in the previous study incorporates 13 items including eating, grooming, dressing, walking, and bowel and bladder management, among others. Thus, it provides no information on functional recovery in specific domains.²¹ Moreover, the Minimal Clinically Important Difference (MCID) for the FIM has yet to be established for the SCI population, which could have provided some baseline for evaluating the clinical value of estimates of functional outcome in the NSCID study.

The present study offers more granular information regarding the association between PWS and functional outcome following acute SCI. We noted a robust four-fold higher risk for respiratory and ambulation support at 6 months among the group with PWS as compared with the control group, even after adjusting for important clinical and treatment covariates. It is possible that the results, to some extent, are indicative of greater debility among patients with PWS during acute care, which makes them less likely to commence early or undergo aggressive rehabilitation.

Previous studies suggest that SCI patients with respiratory complications have greater burden of secondary medical comorbidities and longer length of stay in acute admission.²² In our analysis, the group with PWS spent, on average, twice the number of days in acute admission than the control group. Further, in contrast to the NSCID study results, we noted a weaker association between PWS and neurological outcome, as others also have shown.^22,23

Improvement in neurological outcomes may take longer to demonstrate considering the ceiling effect of the AIS, which might explain the stronger association in the NSCID study that assessed outcome at 1 year rather than 6 months as we did in this study. Given the lower incidence of pneumonia among our cohort (14%) compared with that of the NSCID cohort (44%), we probably underestimated the strength of the relationship; even though the incidence rate in our study is still within reported values in the literature. The exclusion of sicker patients from the NACTN prospective and STASCIS studies most probably accounts for the relatively low incidence of pneumonia in the present study. A weaker association was noted between PWS and neurological outcome among the STASCIS cohort as compared with that of the NACTN, probably due to the lower incidence of pneumonia, or more restricted criteria for inclusion in the STASCIS—which was a prospective cohort study in cervical patients only exploring the effectiveness of early surgery—compared with the NACTN—which is a prospective SCI data registry with very broad eligibility criteria allowing for inclusion of patients with all levels of SCI as well as sicker polytrauma patients.

Identifying predictors of SCI-associated infections is important when developing strategies for early recognition of at-risk groups, which could be useful in guiding prevention and therapy for acquired infections. In previous studies, an increased risk of pneumonia or wound infection was associated with older age, male sex, completeness of injury, more rostrally located injury, higher comorbidity index, paraplegia, or delayed surgical intervention.^7,11,24–27 Using dominance analysis^19,20—a technique recommended as the most appropriate for ranking predictors in a regression model—we demonstrated that the combined effect of injury level, baseline severity, and patient pre-morbid status accounts for much of the risk associated with the indigence of PWS during acute care. Although SCI is associated with higher risk of acquired infections,^3–8 individuals with cervical and higher thoracic level injuries are particularly prone to pneumonia as a result of compromised respiratory function related to disruption of sensorimotor and autonomic innervation from involvement of the upper spinal segments. Systemic immune paralysis from loss of innervation to secondary lymphoid organs further increases the susceptibility.^9,28 It is likely that the degree of vulnerability to secondary infection correlates with the severity of neural damage.

Although approximately 40% of subjects in the present study received MPSS during the acute admission, the rate of infectious complications was not statistically elevated among this cohort relative to those who did not. The neuroprotective effects and safety of MPSS use following acute SCI has been the subject of several studies given the potent anti-inflammatory and immunosuppressive effects of corticosteroids. The pivotal NASCIS II and III trials demonstrated a trend toward increased mortality and morbidity rates among the group randomized to MPSS, with respect to pneumonia, sepsis, acute respiratory distress syndrome, and gastrointestinal bleeding, while demonstrating no beneficial effect for MPSS when compared with the control group.^29,30

Meta-analysis studies aggregating data from previous studies, including those of the NASCIS, have reported somewhat contrary conclusions.^31–33 In one review, the pooled evidence suggested MPSS has no long-term benefit on neurological recovery, although the risk of gastrointestinal bleeding was increased.³¹ A recently published AOSpine review and guideline demonstrated no statistically significant change in motor score with MPSS administration. However, pooled results indicated administration of MPSS within 8 h of injury resulted in improvement in mean motor score at 6 and 12 months.³² The review found no statistical differences between treatment and control groups in the risk of complications, including pneumonia. Although the results of our current study align with this recent review, it should be noted that steroids were not randomly allocated in either NACTN or STASCIS. As a result, it is possible that selection bias and confounding may distort the relationship between steroids and PWS in our study. It is also important to note that the NASCIS III 48-h MPSS protocol, which has been shown to significantly increase infection rates, was not administered to patients in either NACTN or STASCIS.³⁰

The present study has a number of limitations. It included subjects who were treated in tertiary centers with dedicated units for SCI management; therefore, in interpreting the results, the possibility of referral bias should be considered. Selection bias might have arisen from loss to follow up, particularly as systematic differences were noted between subjects with or without missing follow-up data. Although 50% of follow-up data are missing among our cohort, it should be noted that this proportion of missing data is not uncommon among multi-center clinical studies of SCI. Similar to our studies, the NSCID study reported a 44% loss to follow-up at 1 year and 69% at 5 years.¹¹ There is also the possibility of case ascertainment bias resulting from imprecision in the definition of variables, including pneumonia, wound infection, and sepsis, among others. In particular, the literature suggests SCI-associated pneumonia is often underdiagnosed,² and that the timing of neurological examination could influence the ASIA grade classification.³⁴

Further, treatment bias might have occurred; we have no information on the number of patients who were receiving respiratory therapy or support at the time they developed pneumonia. We also had no information on the duration or efficacy of intensive care unit (ICU) management or rehabilitation for the cohort with or without PWS, which have been shown to be strongly predictive of functional recovery.³⁵ To illustrate, it is possible that subjects with greater burden of associated traumatic injuries were more aggressively treated or had longer hospital or rehabilitation stays. Despite these limitations, it is noteworthy that the present study, in our opinion, is one of the most powered to examine the effect of infectious complications. The use of subjects from multiple hospitals increases the generalizability of results.

In conclusion, the findings of the present study have significant relevance for SCI management. Secondary infections are a major concern considering their high frequency during acute hospitalization, the challenges of timely recognition, and evidence of low adherence by practitioners to guidelines on microbiological testing and treatment.² Whereas PWS is the leading cause of acute mortality after SCI, studies suggest case fatality during acute admission and inpatient rehabilitation might have fallen over time.^1,36 Despite this decline, our findings linking these severe acquired infections to poorer long-term functional outcomes warrant greater vigilance and underscore the need for strategies to improve prevention, early identification, and effective treatment of secondary infections in patients with SCI. Further, we advocate that, as much as possible, early surgery is warranted considering the growing evidence of its association with lower incidence of infectious complications.³⁷ Finally, infectious complications should be considered as potential confounders when planning and analyzing interventional studies of SCI.

Author Disclosure Statement

CS discloses relationships with Medtronic (consulting, royalties, patents), Nuvasive (consulting, royalties, patents, stock ownership), Zimmer-Biomet (consulting, royalties, patents), and EOS (consulting). JH discloses relationships with Depuy (consulting), Stryker (honorarium), and Globus (honorarium). RGG discloses relationships with Vertex Pharmaceuticals (Data Safety Monitoring Board) and Insightec (Data Safety Monitoring Board). JRW discloses relationships with Stryker Canada (consulting and teaching). BNRJ, FJ, JHB, BA, RS, ET, SK, JDG, MB, and MGF have no potential conflicts to disclose.

Supplementary Material

Supplementary Table S1

Supplementary Table S2

Supplementary Table S3

Supplementary Figure S1