Risk Factors for Pressure Injury Development in Patients With Spinal Cord Injury Beyond Index Hospitalization: An Analysis of Violent Mechanism of Injury and Socioeconomic Disparity

PIs, defined by the National Pressure Injury Advisory Panel as “localized injury to the skin and/or underlying tissue, usually over a bony prominence, resulting from sustained pressure (including pressure associated with shear),” are widely documented as one of the most common complications following SCI.^1,2 Patients with SCI are particularly vulnerable to PI development, with a 14-fold increase in patients with a traumatic SCI. This is primarily due to their immobility, lack of protective sensation, and baseline reduced perfusion.^2-4 Without protective sensation and mobility, these patients no longer have an effective feedback system to sense localized ischemia and re-distribute pressure before irreversible damage can occur. Rather, they rely on devices and care providers to offload pressure, without a clear consensus on the frequency, duration, or type of pressure relief maneuvers that are most efficacious in prevention. This makes assessment of their support structure post discharge even more important for prevention.

The development of a PI is a particularly debilitating condition associated with pain, functional impairment, decreased social and psychological well-being, rehospitalization, and mortality.² Pressure injuries are a leading cause of hospital admission among patients with SCI, with a prevalence ranging from 25% to 66% in the subacute setting and 15% to 30% in chronic stages.^2,5-8 Furthermore, these wounds may persist for years post injury, with 20% to 35% of patients with SCI continuing to report nonhealing wounds up to 20 years after initial trauma.³ In addition to the health implications, these wounds result in a significant financial burden to the patient as well as an already overwhelmed US health care system; they are associated with costs of more than $17.8 billion annually.^2,9,10

Inadequate patient education and implementation of preventative measures are regarded as the primary contributors to PI incidence.^6,9,11 Standardized patient education programs demonstrate efficacy in reducing PI development and recurrence.¹¹ Education and other preventative interventions are best employed by allocating more resources to higher risk patients. To do so, it is necessary to stratify patients according to PI risk. Currently, several scoring systems exist that aim to stratify risk of PI development. The most prominently used is the Braden scale, which takes into account a patient’s sensory perception, activity, mobility, nutrition, moisture level, and friction/shear.¹² However, its applicability to both persons with trauma and SCI has not been validated, with many studies suggesting that it may not be a useful clinical tool in these populations.^12,13 Additionally, the scale takes into account factors that are hard to quantify and are fairly subjective, such as distinguishing between “occasionally moist” and “rarely moist.” A 2008 comprehensive evaluation of 7 existing PI risk assessment tools concluded none were reliable predictors in patients with SCI.¹³

While risk factors for PI development are largely cited in the literature, few studies assess the influence of sociodemographic factors or cause of the patient’s initial SCI, particularly if the patient’s injury was related to violence, on PI development. A few studies do suggest that there may be sociodemographic disparities in PI development, worsening, and recurrence, with Black male patients at highest risk.^14-16 However, most of these studies consider only nursing home patients^14-16 and lack long-term follow-up.^14,15 Given the association between lower SES and increasing health disparities, conflicts, and instability, as well as the resource-intensive nature of PI prevention that these patients may not be able to afford, further investigation is required to determine if low SES places patients at a higher risk for developing a PI.^17,18 Analysis of potential racial, social, and economic disparities utilizing longitudinal data would add to a body of evidence informing systematic changes in PI prevention and management, geared towards the most vulnerable populations.

Thus, the primary aim of this study is to assess consistently identifiable potential risk factors for long-term PI development beyond index hospitalization in patients with traumatic SCI including violent versus nonviolent mechanism of SCI and sociodemographic factors. This will serve as a first step in creating a better system for identifying these patients at index hospitalization to institute more resources for long-term, post-discharge PI prevention.

This investigation was a retrospective cohort study utilizing data from the electronic medical record system at a single level one trauma center. Inclusion criteria consisted of patients 18 years of age or older who had a traumatic SCI and were treated between January 2, 2002, and December 31, 2018, with a minimum of 1 year follow-up. Pregnant women, incarcerated patients, patients with less than 1 year of follow-up, and those with incomplete sociodemographic data were excluded. Sociodemographic data including sex, marital status, age, MHI of patients’ ZIP code, and race were collected. Injury-related data were collected, including level of SCI, AIS score, and complete versus incomplete SCI. Mechanism of injury was collected and subsequently divided into violent versus non-violent etiology. Violent mechanism of SCI was defined as gunshot wound, assault, or stab wound; suicide attempts were excluded. Outcome variables including PI incidence, PI stage (1–4), multiple/recurrent PIs, time to development of first PI from the time of initial SCI, and flap coverage were also recorded.

Data analysis was conducted using the statistical software platform SPSS Version 28.0 (IBM). Descriptive statistics as well as univariate and multivariate logistic regression analyses were conducted. A P value of less than .05 was utilized as a cut-off to determine statistical significance.

Descriptive statistics

A total of 448 patients were identified who met inclusion criteria. Of these, 355 (79%) were male and 93 (21%) female. Regarding race, 264 (59%) patients were documented as White, 151 (34%) Black or African American, 26 (6%) Hispanic, and 4 (1%) were identified in the electronic medical record as other/unknown. The mean age of patients in this study was 44 years old with a SD of 18 years (range, 18–90 years). Within the study population, 69% were unmarried (including single, divorced, or widowed) and 31% were married. Mean MHI by ZIP code was $56 075 with a SD of $20 743 (range, $14 700–$135 146) (Table 1). A total of 188 patients had complete SCI (42%) while 260 had an incomplete injury (58%). A total of 94 patients (21%) had a violent mechanism of SCI (Table 2). In total, 163 (36%) patients developed one or more PIs. The mean time to development of initial PI from the time of SCI was 3.74 years (SD, 4.07 years; range, 0–15 years). Of those with PIs, 68.7% developed multiple PIs and 20% underwent flap coverage. There was a greater percentage of patients who underwent flap surgery in the cohort that had a violent SCI (26% vs 17%). A greater percentage of male patients (40% vs 22%), unmarried patients (41% vs 28%), patients with complete SCI (59% vs 20%), and patients with violent mechanism of SCI (56% vs 31%) developed PIs. Median stage of PI was higher in patients of violent mechanism of SCI (violent = stage 4, non-violent = stage 3; P <.05).

Logistic regression

On univariate binary logistic regression, significant predictors of PI development included violent mechanism of injury (OR = 2.89, P <.001), male sex (OR = 2.46, P <.01), unmarried marital status (OR = 1.77, P <.01), complete SCI (OR = 5.51, P <.001), and increasing age at the time of SCI (OR = 1.01, P <.05) (Table 3). While there was a trend towards lower MHI by the patients’ respective ZIP code and PI development, it did not reach statistical significance (P =.077). For race, patients identifying as Black or African American was correlated with PI development, but this observation did not reach statistical significance (P =.052). Violent mechanism of SCI was a significant predictor of developing more than 1 PI over 2 separate time points (OR = 3.06, P <.01) (Figure).

A multivariate logistic regression analysis was conducted utilizing the variables found to be significant predictors of PI development on univariate analysis. Significant variables using this regression analysis included complete SCI (OR = 5.51, P <.001), male sex (OR = 2.08, P <.05), and violent mechanism of injury (OR = 2.36, P  <.01). Increasing age at the time of SCI (OR = 1.00, P =.979) and marital status (OR = 1.43, P =.146) did not reach significance in this model (Table 4).

Patients with an SCI have extensive medical, psychological, social, and financial challenges.¹⁹ Of these, PIs unfortunately remain one of the most common and devastating complications.^2,5-8,19 In the current cohort, 36% of patients developed one or more PIs, consistent with rates reported in the literature. Incidence of PI, including in persons with SCI, has continued to rise in recent years, perhaps due to an aging population, shorter hospitalizations lacking adequate counseling, or a multitude of other factors.^9,10 Regardless, the numbers indicate a shortcoming in current strategies for prevention and treatment. This highlights the need for a new method for risk stratification, as timely identification of high-risk populations enables proactive initiation of preventative interventions early during hospitalization. This also requires PI risk to be assessed by nurses, surgeons, and physicians as early as presentation to the emergency department utilizing objective, identifiable, and consistent risk factors.^4,6

The current authors’ goal was to investigate potential risk factors that would allow rapid identification of patients who have a significantly higher risk of PI development and likely lack the resources needed to prevent PI development to target more services to these patients. Hence, the authors divided the population into 2 easily distinguishable cohorts—one that experienced a violent mechanism of traumatic SCI and another that had a non-violent etiology. In North America, violent etiology accounts for approximately 13.5% of all SCIs.²⁰ In the current study, 21% of patients had a violent SCI. This higher prevalence is likely due to this institution acting as one of the only 2 level one trauma centers in Wisconsin as well as serving the Milwaukee community, which includes a large urban population and experiences a relatively higher volume of penetrating trauma. The authors found that violent mechanism of SCI was indeed a significant predictor of PI development on both univariate and multivariate logistic regression analysis, and that these patients experienced a higher median stage of PI. Additionally, there was a higher rate of need for flap surgery in this group which is associated with additional hospitalization needs and a much higher cost. Possible underlying explanations for this difference include higher incidence of paraplegia and complete injury in those with a violent cause of SCI, resulting in injuries with even more immobility and less protective sensation.^20,21

Consistent with this notion, the current study reported that complete SCI was a significant predictor of PI development (P <.001). Inferior functional status in these patients also increases reliance on attendants and devices for prevention.²² Moreover, it is possible that injuries may go unnoticed for a longer time in such patients, resulting in higher stage injuries more likely to require flap coverage. This disparity is exacerbated by the fact that patients with SCI who had violent trauma additionally tend to exhibit poor long-term social integration impairing their access to non-medical care givers.²⁰ Studies have found violent injury to be a significant risk factor for persistent depressive symptoms and psychological distress.²³ Patients experiencing these mental health sequelae from their trauma may lack many of the known protective factors for prevention of PI, such as engagement in personally meaningful activities.²⁴ These patients may also face increased barriers to necessary psychological assistance due to the implicit biases of health care staff, who may be more likely to feel these patients are somehow at fault for their injuries. Finally, prior research demonstrates that particular groups are more vulnerable to violent injury, including low SES and unmarried individuals, and ethnic minorities.²³ While the authors controlled for sociodemographic variables and still found violent etiology to be an independent predictor of PI development, there is undoubtedly interplay between these factors that the retrospective cohort may not have been powered well enough to delineate. These factors can therefore likely still help guide where resources should be devoted to preventing PIs.

While there has been much more research investigating risk factors for PI development in this population during index hospitalization, there are limited data addressing the long-term risk following discharge. In the current cohort, the authors reported the mean time to development of a patient’s first PI was nearly 4 years after index hospitalization. Therefore, previously determined clinical risk factors for PI development in patients with SCI, such as pneumonia and mechanical ventilation, are seemingly less important when assessing long-term risk factors. At the very least, these hospitalization-related factors offer an incomplete view of long-term PI risk. Similarly, some protective factors, such as the motivation to prevent negative health outcomes or self-advocacy skills, are more pertinent when a patient is no longer actively in a health care setting.²⁴

In the current study, patients treated from 2002 to 2018 with a minimum of 1 year follow-up were included. This retrospective, longitudinal design allowed for identifying factors that impact the long-term post discharge risk of PI development which has been sparsely reported in the literature. When looking through this longitudinal lens, sociodemographic variables likely contribute more after discharge when the pressure and moisture-limiting interventions are not performed by health care professionals. It is this long-term morbidity that represents an ideal opportunity for systemic prevention strategies centered on patient and caregiver education.

The authors found male sex was an independent risk factor for PI development in persons with SCI. Male sex has been previously validated as a risk factor in 2 high-power cohort studies and continues to serve as a dependable measure for increased PI risk.⁵ In the current study, unmarried marital status was a significant predictor for PI development in univariate analysis but not in multivariate analysis. There have been conflicting reports on marital status as a risk factor, although a well-designed cohort study found married status to be a protective factor.⁸ Theoretically, a spouse offers psychological and physical support to a patient with SCI, encouraging vigilant self-care as well as physically assisting with preventative care. Marital status may be a strong covariate with other sociodemographic variables including age and sex, and thus this study may not have been powered enough to detect a difference. Increasing age at the time of SCI was similarly found to be predictive of PI development on univariate analysis alone. Age has not previously been found to be a significant risk factor in the literature.⁵ Finally, the authors did not find race to be an independent predictor of PI development, consistent with prior studies which have also demonstrated that after controlling for socioeconomic and clinical factors, race is often not a significant predictor of PI.^25,26

Interestingly, while this study hypothesized that lower SES would be associated with increased PI risk, this trend did not reach statistical significance. Prior studies have demonstrated an association between individual household income and PI presence.^26,27 Prevention of PI is resource intensive, requiring frequent access to the health care system and to a variety of medical equipment often not covered by insurance.²⁴ The lack of this association in the current study may be attributed to the use of MHI by a patient’s ZIP code as a proxy for socioeconomic status. While this is a common practice in medical literature, a 2020 population-based study by Moss et al²⁸ found that area-level SES indicators, like ZIP code, were more likely to underestimate individual-level SES and are in fact poor proxies.²⁸ Therefore, future research should consider utilizing a different
indicator of SES.

While this study highlights the predictive value of violent mechanism of SCI and male sex in PI risk, these findings should be considered in the context of the current study limitations. First, this is a retrospective chart review; therefore, data were restricted to what could be acquired via the authors’ electronic medical record system and undoubtedly contained inaccuracies as is inevitable with medical documentation. Second, this study was limited in the factors assessed. The authors did not include all previously established PI risk factors in this analysis, including higher level of SCI, SES as determined by factors other than MHI by ZIP code, development of pneumonia, need for and duration of mechanical ventilation, or duration of SCI beyond 15 years. Furthermore, the authors only had trauma-specific AIS data and not American Spinal Cord Injury Impairment Scale data which provides more defined information on retained sensation and function and may have offered more clarity on the physiologic impact of the injuries.^2,5,6,8 Third, the current study was limited to patients who sought follow-up at the authors’ institution, which likely skewed the results. Fourth, data regarding patient’s living environment were not collected as the authors could not verify the length of time spent at the discharge location noted in a patient’s chart (ie, rehabilitation facility vs home setting). However, it is certainly possible that the significance of the current findings would vary depending on this as the systems for PI prevention at a rehabilitation facility are likely different than a home environment. The definition of violent etiology, which considered mechanism of injury but not intent, was also limiting. For example, gunshot wounds were defined as violent and motor vehicle collisions as non-violent, but a gunshot wound could be accidental, and a motor vehicle collision could be intentional. Determining the intention of the injury was not possible utilizing the records used. Lastly, the authors did not track changes in function and sensation amongst patients with incomplete SCIs, which would likely have an impact on PI risk.

Ultimately, there are clearly a multitude of risk factors for PI development in the patient with SCI, including clinical, physiologic, behavioral, and social support factors. The current study primarily evaluated mechanism of injury and sociodemographic factors at the time of injury. These findings can be added to existing risk assessment tools, such as the Braden scale, to improve their applicability to those with traumatic SCI. The authors’ next steps will be to develop and prospectively test the efficacy of such a risk assessment tool. There is precedent for using scoring systems and risk assessment tools in this context. Ideally, a risk stratification tool could initially be utilized in the index hospitalization and incorporated at subsequent follow-up appointments to provide ongoing reassessment for any of the risk factors, such as physical medicine and rehabilitation or primary care appointments during monthly or 6-month follow-up. It would be important to continuously re-assess risk level as social support and resource availability may change for patients over time and could impact PI risk which were suggested in the current univariate but not multivariate analysis. As such, further investigation into the link between clinical, demographic, and socioeconomic variables in the context of PI risk is warranted to target intensive interventions and resources for PI prevention.

Violent mechanism of SCI, male sex, and complete SCI are predictive of long-term PI risk in patients with a traumatic SCI. Additionally, violent mechanism of SCI is an independent predictor of PI development and recurrence. These patients tend to develop higher severity PIs and undergo a higher rate of flap surgery. Resources and interventions targeting PI prevention in male patients who had a violent traumatic complete SCI may be of additional benefit in preventing these debilitating and cost-intensive complications. Further research utilizing a larger multi-institutional validation study is warranted to further explore these sociodemographic variables.