Advertisement

Spinal cord injury in Italy: A multicenter retrospective study

  • M.Grazia Celani
    Affiliations
    Servizio per le Malattie Cerebrovascolari, Perugia (Celani, Ricci), Dipartimento di Geriatria e Riabilitazione, Parma (Zampolini, Franceschini), Dipartimento della Prevenzione, Ministero della Sanità, Roma (Spizzichino); and U.O. di Riabilitazione Intensiva Neuromotoria, Trevi, (Zampolini, Franceschini), Italy
    Search for articles by this author
  • Lorenzo Spizzichino
    Affiliations
    Servizio per le Malattie Cerebrovascolari, Perugia (Celani, Ricci), Dipartimento di Geriatria e Riabilitazione, Parma (Zampolini, Franceschini), Dipartimento della Prevenzione, Ministero della Sanità, Roma (Spizzichino); and U.O. di Riabilitazione Intensiva Neuromotoria, Trevi, (Zampolini, Franceschini), Italy
    Search for articles by this author
  • Stefano Ricci
    Affiliations
    Servizio per le Malattie Cerebrovascolari, Perugia (Celani, Ricci), Dipartimento di Geriatria e Riabilitazione, Parma (Zampolini, Franceschini), Dipartimento della Prevenzione, Ministero della Sanità, Roma (Spizzichino); and U.O. di Riabilitazione Intensiva Neuromotoria, Trevi, (Zampolini, Franceschini), Italy
    Search for articles by this author
  • Mauro Zampolini
    Affiliations
    Servizio per le Malattie Cerebrovascolari, Perugia (Celani, Ricci), Dipartimento di Geriatria e Riabilitazione, Parma (Zampolini, Franceschini), Dipartimento della Prevenzione, Ministero della Sanità, Roma (Spizzichino); and U.O. di Riabilitazione Intensiva Neuromotoria, Trevi, (Zampolini, Franceschini), Italy
    Search for articles by this author
  • Marco Franceschini
    Affiliations
    Servizio per le Malattie Cerebrovascolari, Perugia (Celani, Ricci), Dipartimento di Geriatria e Riabilitazione, Parma (Zampolini, Franceschini), Dipartimento della Prevenzione, Ministero della Sanità, Roma (Spizzichino); and U.O. di Riabilitazione Intensiva Neuromotoria, Trevi, (Zampolini, Franceschini), Italy
    Search for articles by this author
  • for the Retrospective Study Group on SCI
  • Author Footnotes
    NO LABEL a. BMA House, Tavistock Square, London WCIH 9JR, UK.
    NO LABEL b. SPSS Inc, 233 S Wacker Dr, 11th Fl, Chicago, IL 60606.

      Abstract

      Celani MG, Spizzichino L, Ricci S, Zampolini M, Franceschini M, Retrospective Study Group on SCI. Spinal cord injury in Italy: a multicenter retrospective study. Arch Phys Med Rehabil 2001;82:589-96. Objective: To investigate certain factors influencing the length of stay (LOS) in a rehabilitation center, the incidence of pressure ulcers, and the neurologic improvement of patients with traumatic (T/SCI) and nontraumatic spinal cord injury (NT/SCI). Design: A multicenter retrospective study of patients with SCI admitted to rehabilitation centers between 1 January 1989 and 31 December 1994 (only first admissions). Setting: Seven Italian rehabilitation centers. Patients: A total of 859 consecutively admitted adult patients with SCI. Intervention: Examined medical records of patients admitted to rehabilitation centers. Main Outcome Measures: Pressure ulcers on admission as an indicator of nursing care in acute phase, LOS in rehabilitation centers, and neurologic improvement on discharge (using the Frankel classification system). Other measures included level of lesion, associated lesions (if T/SCI), surgical stabilization (if T/SCI), and time from the event to admission to a rehabilitation center. Results: In all cases, the time from event to admission to a rehabilitation center exceeded 30 days (average ± standard deviation: T/SCI, 54.6 ± 43.7d; NT/SCI, 166.9 ± 574d); pressure ulcers on admission were present in 34.1% of T/SCI and 17.1% of NT/SCI patients. The average LOS in a rehabilitation center was 143.1 ± 89.1 days for T/SCI and 91.7 ± 78.9 days for NT/SCI; Frankel grades improved by 1 or more in 34.4% of T/SCI and 34.1% of NT/SCI patients. The presence of pressure ulcers on admission, rehabilitation LOS, and neurologic improvement on discharge correlated highly with severe neurologic damage on admission in both T/SCI and NT/SCI patients as well as with management of the patient immediately before admission to a rehabilitation center, mainly in NT/SCI patients. Conclusions: Severe neurologic damage is the major determining factor in predicting neurologic recovery. Pressure ulcer prevention is statistically associated with neurologic improvement and the shortening of rehabilitation LOS. Patient management immediately before admission to rehabilitation has a statistical correlation with neurologic improvement in all patients studied and on both rehabilitation LOS and incidence of pressure ulcers in the NT/SCI patients. © 2001 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation

      Keywords

      SPINAL CORD INJURY (SCI) is a devastating condition that bears high rates of morbidity and mortality, though the latter has decreased somewhat owing to improved resuscitation techniques. Those affected are mostly young and middle-age adults, and recovery difficulties for these individuals are determined by the type and severity of lesion, the quality of rehabilitation programs, and possibilities of performing everyday activities.
      • Cawley MF
      • Yarkony GM
      • Bergman SB.
      Spinal cord injury rehabilitation. 5. Through the lifespan.
      • Westgren N
      • Levi R.
      Quality of life and traumatic spinal cord injury.
      • Waters RL
      • Yakura JS
      • Adkins RH
      • Sie I.
      Recovery following complete paraplegia.
      • Tator CH
      • Duncan EG
      • Edmonds VE
      • Lapczak LI
      • Andrews DF.
      Neurological recovery, mortality and length of stay after acute spinal cord injury associated with changes in management.
      • Schönherr MC
      • Groothoff JW
      • Mulder GA
      • Eisma WH
      Functional outcome of patients with spinal cord injury: rehabilitation outcome study.
      The incidence of SCI varies widely from country to country according to cause, study methodology, and source of data. Furthermore, the occurrence of traumatic SCI (T/SCI) depends heavily on the local policies adopted to prevent trauma, as defined by systematic epidemiologic and economics studies.
      • Blumer CE
      • Quine S.
      Prevalence of spinal cord injury: an international comparison.
      Nontraumatic lesions have received less attention, mainly because they call for a well-concerted multidisciplinary diagnosis.
      This situation is particularly true in Italy because no large-scale surveys on the incidence and prevalence of SCI have been conducted to date. The few retrospective studies performed are not comprehensive enough to allow for any reliable conclusions to be drawn, whereas prospective studies involved only 1 hospital and therefore lacked case mix.
      • Caldana L
      • Lucca L.
      Epidemiological remarks on traumatic spinal cord injuries and non traumatic spinal cord diseases in Veneto 1994-1995.
      • Di Carlo S
      • Taricco M
      • Adone R.
      Studio epidemiologico sulle lesioni midollari. Considerazioni su 262 pazienti ricoverati presso l'ospedale di Passirana di Rho negli anni 1977-1988.
      Even less is known about the rehabilitative management of SCI patients, in particular, when the rehabilitation program started, the type of patient admitted to a rehabilitation center, any clinical complications, and the degree of neurologic recovery.
      Among the different variables studied, 3 can be considered particularly significant to draw a basic clinical understanding of these patients. First, the length of stay (LOS) is considered an indicator of the costs and the efficiency of health care provided.
      • Eastwood EA
      • Hagglund KJ
      • Ragnarsson KT
      • Gordon WA
      • Marino RJ
      Medical rehabilitation length of stay and outcomes for persons with traumatic spinal cord injury—1990-1997.
      Second, though pressure ulcers that develop during the acute phase of care before admission to a rehabilitation center highlight the need to pay attention to the identification and management of this condition, it is necessary to prevent development or worsening of pressure ulcers with adequate nursing and medical care.
      • Chen D
      • Apple Jr, DF
      • Hudson LM
      • Bode R
      Medical complications during acute rehabilitation following spinal cord injury—current experience of the Model Systems.
      Finally, neurologic improvement on discharge is considered an indicator of residual capacity to minimize impairment.
      • Tator CH
      • Duncan EG
      • Edmonds VE
      • Lapczak LI
      • Andrews DF.
      Neurological recovery, mortality and length of stay after acute spinal cord injury associated with changes in management.
      • Kiwerski J
      • Weiss M.
      Neurological improvement in traumatic injuries of cervical spinal cord.
      A retrospective study was performed to estimate the cause of trauma, etiopathology of the lesion, gender, age, and other risk factors. Considering the need for more data, 7 Italian rehabilitation centers collaborated to gather the information. Certain factors influencing the LOS in a rehabilitation center, the incidence of pressure ulcers, and the neurologic recovery of patients with both T/SCI and nontraumatic SCI (NT/SCI) were investigated.

      Methods

      Seven rehabilitation centers situated in central and northern Italy participated in this retrospective study, which included 859 patients with SCI. The medical records of all patients with a diagnosis of T/SCI or NT/SCI admitted to these rehabilitation centers between 1 January 1989 and 31 December 1994 were examined; those on their first admission after the event were included in the study. A first admission to a rehabilitation center may have occurred years after the actual event. Whenever a patient was discharged or transferred for a period of no longer than 3 weeks, this was considered a single admission. Any readmission after the 3-week period was considered a second admission, and the patient was excluded from the study.
      A standard form of simple and consistent items that could be easily extracted from clinical records
      • Franceschini M
      • Baratta S
      • Zampolini M
      • Lotta S
      • Loria D
      • Di Benedetto P.
      Studio multicentrico sulle lesioni midollari traumatiche: implicazioni per la prevenzione e la programmazione sanitaria.
      was prepared to minimize major biases that can lead to error in systematic data collection. To reduce further selection bias, hospital discharge records were analyzed and selected if they matched the specific code of International Classification of Diseases, 9th revision, on the discharge form. The likelihood is minimal that any appropriate case was missed or that any case initially identified before record was considered. Data were collected from each center and input into a single database designed by one of us (MZ).
      Pressure ulcer on admission, LOS, and any neurologic improvement according to the Frankel classification system
      • Frankel HL
      • Hancock DO
      • Hyslop G
      • Melzak J
      • Michaelis LS
      • Ungar GH
      • et al.
      The value of postural reduction in the initial management of closed injuries of the spine with paraplegia and tetraplegia.
      (table 1) were analyzed as main outcome variables.
      Table 1The Frankel Classification System
      GradeClinical Condition
      AComplete motor and sensory paralysis below the lesion
      BComplete motor paralysis but with sensory sparing
      CBoth sensory and motor sparing but with motor power useless
      DSensory and motor sparing with useful power, with or without appliances
      ENeurologically intact including normal bowel and bladder function
      The presence or absence of a pressure ulcer was recorded alone, regardless of the extent of skin loss or the presence of tissue (grade 2 or more), or bone or tendon necrosis. LOS was the time (in days) from the first admission to the last discharge. Any evolution toward a better Frankel grade was considered an improvement.
      Correlation among some other important variables was performed, including cause, age, gender, level of lesion, management immediately before rehabilitation (ie, type of ward the patient was admitted to before transfer to the rehabilitation center), associated lesions, surgical stabilization, time lapse from the event to admission, any of the rehabilitation centers included in this study, or whether the patient was registered in the same health administration area as the rehabilitation center.
      Two clinical groups were established on the basis of the cause of the lesion, T/SCI and NT/SCI, to ascertain whether the 3 predictive factors were similar. A descriptive analysis was performed by determining the frequency of the 3 variables examined in the 2 groups of patients. Confidence limits of the proportions were calculated with the CIA
      a. BMA House, Tavistock Square, London WCIH 9JR, UK.
      program.
      • Gardner S
      • Winter P
      • Gardner M.
      CIA.
      Association between potential risk factors and the major outcome measurements were determined with chi-square statistics for categoric variables and t test or F test for continuous variables. Odds ratios (ORs) and Cornfield confidence limits were calculated when appropriate. The clinically relevant factors that were statistically associated with the 3 main outcome variables and those that did not reach the standard level of significance were analyzed with multiple logistic regression by using a forward stepwise procedure, whereas for LOS analysis of covariance (ANCOVA) models were performed. A diagnostic analysis of the models was also performed.
      To estimate the effects on LOS of neurologic impairment grade at time of admission to the rehabilitation center, prognostic models were developed by using the linear regression model, G(y) = Dx1 + Cx2 + Bx3 + Ax4, where G = expectation of y; y = number of days (LOS); and x1...x4 are dummy variables that assume value 1 in the presence of any Frankel grade (A, B, C, D), and 0 elsewhere. All statistical analyses were performed by using SPSS software.
      b. SPSS Inc, 233 S Wacker Dr, 11th Fl, Chicago, IL 60606.
      This study was approved by the participant centers' ethics committees.

      Results

      During the 5-year period, 859 patients (630 men, 229 women; age range, 2-84yr; mean, 37.9yr) were admitted to the 7 rehabilitation centers participating in the study (table 2).
      Table 2Rehabilitation Centers
      Center (n = 859)T/SCI (n = 642)NT/SCI (n = 217)
      n%n%n%
      Ferrara364.2193177.8
      Trevi10912.7609.34922.6
      Trieste303.5162.5146.5
      Turin14516.911718.22812.9
      Udine12814.99915.42913.4
      Vicenza22526.218128.34420.3
      Villanova18621.715023.43616.6
      The characteristics of the population, divided into T/SCI and NT/SCI groups, are listed in tables 3 and 4. Further details of the 2 populations have been published elsewhere.
      • Franceschini M
      • Baratta S
      • Zampolini M
      • Lotta S
      • Loria D
      • Di Benedetto P.
      Studio multicentrico sulle lesioni midollari traumatiche: implicazioni per la prevenzione e la programmazione sanitaria.
      Table 3Characteristics of T/SCI Group (n = 642)
      Age (yr) mean ± SD (median)34.3 ± 15.5 (30)
      Gender n (%)Men509 (79.3)
      Health administration area n (%)Local430 (67)
      Cause n (%)Motor vehicle crash375 (58.4)
      Weapon19 (3)
      Home accident33 (5.1)
      Work accident98 (15.3)
      Sport53 (8.3)
      Suicide27 (4.2)
      Other37 (5.8)
      Level of lesion n (%)Paraplegia413 (64.3)
      Management immediately before rehabilitation n (%)Surgery12 (1.9)
      Internal medicine39 (6.1)
      Neurosurgery310 (48.3)
      Neurology17 (2.6)
      Orthopedics186 (29)
      Intensive care58 (9)
      Home14 (2.2)
      Other6 (0.9)
      Associated lesions n (%)Yes315 (49.1)
      Surgical stabilization n (%)Yes521 (81.2)
      Frankel grade (on admission) n (%)A351 (54.7)
      B63 (9.8)
      C179 (27.9)
      D45 (7)
      Not known4 (0.6)
      Time from injury to admission (d) mean ± SD (median)54.6 ± 43.7 (42.5)
      LOS (d) mean ± SD (median)143.1 ± 89.1 (131)
      Neurologic improvement n (%)Yes221 (34.4)
      Pressure ulcers n (%)Yes219 (34.1)
      Table 4Characteristics of NT/SCI Group (n = 217)
      Age (yr) mean ± SD (median)48.2 ± 18.1 (52)
      Gender n (%)Men121 (55.8)
      Health administration area n (%)Local159 (73.3)
      Cause n (%)Infective30 (13.8)
      Neoplasia79 (36.4)
      Vascular55 (25.3)
      Other53 (24.4)
      Level of lesion n (%)Paraplegia168 (77.4)
      Management immediately before rehabilitation n (%)Surgery5 (2.3)
      Internal medicine22 (10.1)
      Neurosurgery84 (38.7)
      Neurology39 (18)
      Orthopedics17 (7.8)
      Intensive care7 (3.2)
      Home37 (17.1)
      Other6 (2.8)
      Frankel score (on admission) n (%)A55 (25.3)
      B27 (12.4)
      C77 (35.5)
      D43 (19.8)
      E2 (0.9)
      Not known13 (6)
      Time from injury to admission (d) mean ± SD (median)166.9 ± 574 (57)
      LOS (d) mean ± SD (median)91.7 ± 78.9 (80)
      Neurologic improvement n (%)Yes76 (34.1)
      Pressure ulcers n (%)Yes37 (17.1)

      T/SCI group

      Pressure ulcers

      In this group of patients, the prevalence of pressure ulcers was 34.1% (table 3). This indicator of the quality of preventive nursing care is, in the univariate analysis, significantly associated with longer time lapse from the event (p <.000), admission to 1 of the 7 rehabilitation centers in the study—pressure ulcer percentage at the centers ranged from 12.5% to 52% (p <.00002), patients coming from a different health administration area (p <.00001), the worst Frankel grade evaluated on admission (p <.00000), and marginally correlated with higher injury level (cervical vs dorsolumbar, p <.058) (table 5).
      Table 5Analysis of Pressure Ulcers
      T/SCI GroupNT/SCI Group
      ModalityStatistical Test95% CLStatistical Test95% CL
      Aget = 1.58t = 1.55
      GenderMenOR = 1.360.8-2.1OR = 2.47*1.1-5.8
      Health administration areaLocalOR =.46*0.3-0.6OR =.980.4-2.4
      Rehabilitation centerχ2 = 31.6*χ2 = 7.17
      Causeχ2 = 10.1χ2 = 21.15*
      Level of lesionParaplegiaOR = 1.380.9-1.9OR = 2.750.9-11.2
      Management before rehabilitationχ2 = 3.3χ2 = 56.08*
      Associated lesionsYesOR = 1.83*1.3-2.6
      Surgical stabilizationYesOR = 1.280.8-2
      Frankel gradeχ2 = 59.8*χ2 = 22.77*
      Time from injury to admissiont = 6.2*−0.5
      Multivariate Analysis
      T/SCI Group (χ2 = 136.55, p >.00001)NT/SCI Group (χ2 = 65.89, p >.00001)
      FactorpFactorp
      Frankel on admission.0000Management immediately before rehabilitation.076
      Time from injury to admission.0000Frankel on admission.037
      Rehabilitation center.0005
      Age.0067
      Level of lesion.0221
      * Statistically significant association (p <.05 and confidence limits [CLs] of OR noncomprehensive of A unit).
      Multiple logistic regression identified 5 statistically significant factors that were independent predictors of pressure ulcers (p >.000): the worst Frankel grade on admission, longer time lapse from the event to admission, admission to 1 of the 7 rehabilitation centers, older age, and higher injury level (table 5).
      In the univariate analysis, the health administration area the injured subject came from was statistically significant because of the high correlation between delayed admission and the distance of the patient's home from the rehabilitation center, whereas in the multiple regression model, age and injury level were also important (table 5).

      Neurologic improvement

      Neurologic impairment level improved in approximately one third (34.4%) of the patients. In the univariate analysis, this result was statistically associated with age (p <.000), older patients seemed to have a better recovery (mean age in the group that showed an improvement, 37.2 vs 32.7yr), shorter time from the event (mean time lapse in the group that showed an improvement, 49.2 vs 57.6d, p <.020), and shorter LOS at the rehabilitation center (mean LOS in the group with improvement, 115.7 vs 157.7d, p <.000). Neurologic improvement was also associated with admission to the particular rehabilitation center to which the patient was admitted (p <.0004), with whether that rehabilitation center was in the patient's local health administration area or in another health district (p <.0176), with management immediately before rehabilitation (type of hospital department or home) (p <.02), with cause of the trauma (p <.01), with surgical stabilization of the spinal column (p <.036), and with presence of pressure ulcers on admission (p <.0000) (table 6).
      Table 6Analysis of Neurologic Improvement
      T/SCI GroupNT/SCI Group
      ModalityStatistical Test95% CLStatistical Test95% CL
      Aget = −3.57*t = −.56
      GenderMenOR =.740.5-1.1OR =.760.4-1.4
      Health administration areaLocalOR = 1.54*1.02-2.3OR =.680.4-1.3
      Pressure ulcersYesOR =.31*0.2-0.5OR =.35*0.1-0.9
      Rehabilitation centerχ2 = 24.6*χ2 = 7.8
      Causeχ2 = 16.3*χ2 = 10.2*
      Level of lesionParaplegiaOR =.860.6-1.2OR =.640.3-1.8
      Management before rehabilitationχ2 = 15.1*χ2 = 12.55
      Associated lesionsYesOR =.820.6-1.2
      Surgical stabilizationYesOR =.65*0.4-1.0
      Frankel gradeχ2 = 276.2*χ2 = 51.3*
      Time from injury to admissiont = 2.33*t = −.87
      LOSt = 6.29*t = 1.45
      Multivariate Analysis
      T/SCI Group (χ2 = 235.48, p >.00001)NT/SCI Group (χ2 = 74.4, p >.00001)
      FactorpFactorp
      Frankel on admission.0000Frankel on admission.0000
      Rehabilitation center.0008Local health authority or other.0176
      Pressure ulcers.0128Management immediately before rehabilitation.0502
      Management immediately before rehabilitation.0435Cause.0547
      * Statistically significant association (p <.05 and CLs of ORs noncomprehensive of A unit).
      In the present study, it would seem that patients with grade C (76%) and B (67%) of the Frankel classification system had a greater possibility of improvement (fig 1).
      Figure thumbnail gr1
      Fig. 1The percentage of patients who improved at least 1 grade on the Frankel classification system from admission to a rehabilitation center to discharge. ■ represents T/SCI patients; ▴ represents NT/SCI patients.
      In the multiple logistic regression analysis, the model that best predicted clinical improvement (p >.000) included a better Frankel grade on admission, admission to 1 of the 7 rehabilitation centers, absence of pressure ulcers, and quality of management immediately before rehabilitation (table 6). Other variables, such as age, were excluded from the model because of the lack of significant correlation in predicting this independent variable.

      Length of stay

      The mean LOS was 143.1 ± 89.1 days (median, 131d; range, 3-922d) (table 3). A longer LOS was statistically associated with admission to 1 of the 7 rehabilitation centers: mean LOS in the 7 rehabilitation centers ranged from 95 to 202 days (p <.0000), if there was surgical stabilization of the patient's spinal column (p <.01), presence of associated lesions (p <.0002), pressure ulcer on admission (p <.0000), worst Frankel grade on admission (p <.0000), neurologic improvement on discharge (p <.0000), and age (p <.0029). The older the patient was, the shorter the LOS proved to be (table 7).
      Table 7Analysis of LOS
      T/SCI GroupNT/SCI Group
      ModalityStatistical TestStatistical Test
      Aget = −2.99*t = −1.75
      GenderMenF = 1.7F =.02
      Health administration areaLocalF = 1F =.03
      Rehabilitation centerF = 7.06*F = 8.95*
      Pressure ulcersYesF = 29.26*F = 4.98*
      CauseF = 1.66F = 3.44*
      Level of lesion (para vs quad)ParaF = 3.1F = 9.99*
      Management before rehabilitationF = 1.5F = 7.1*
      Associated lesionsYesF = 14.38*
      Surgical stabilizationYesF = 6.65*
      Frankel gradeF = 26.48*F = 5.45*
      Time from injury to admissiont =.36t = −1.49
      ImprovementYesF = 33.6*t = 1.37
      Multiple Analysis
      T/SCI Group (R2 = 13%, p >.00001)NT/SCI Group (R2 = 14.4%, p >.00001)
      FactorpFactorp
      Frankel on admission.000Management immediately before rehabilitation.0011
      Pressure ulcers.0049Level of lesion.002
      Associated lesions.0069Frankel on admission.017
      Level of lesion.013Cause.0286
      Age.03
      * Statistically significant association (p <.05).
      Abbreviations: para, paraplegia; quad, quadriplegia.
      To improve LOS prediction, an ANCOVA model was devised to control the age effect. Six independent prognostic factors were included in the final model: the worst Frankel grade on admission, which rehabilitation center, pressure ulcers on admission, higher lesion level, associated lesions, and older age as a covariate variable (table 7).
      The Frankel grade was the principal factor associated, both clinically and statistically, with the 3 outcome variables considered in these analyses. For this reason, neurologic impairment was chosen as an explicative variable in a simple predictive model with LOS: number of days = 71.3 (D) + 117.9 (C) + 130.2 (B) + 167.6 (A).
      This model is highly significant (p >.0001) and has an R2 of 75% (with 4 degrees of freedom [df]) of A, B, C, D, representing the Frankel grades in the score system; the presence of 1 of the Frankel grades produces the average LOS expressed by the correspondent parameter, which is highly significant.

      NT/SCI group

      Similar analyses were performed for the NT/SCI group to investigate any differences. Of the 217 NT/SCI patients, more than one third had a neoplasm, one quarter suffered from vascular disease, one quarter from a poorly defined disease (eg, osteodegenerative disease), and the remaining from an infective or inflammatory disease (table 4).
      This NT/SCI population had a significantly higher mean age than the T/SCI group (48.2 ± 18.1yr vs 34.3 ± 15.5yr, p >.0001). The mean time lapse from the event to admission to the rehabilitation center was even longer for the NT/SCI patients (166.9 ± 574d vs 54.6 ± 43.7d, p >.0001). Moreover, the analysis of patient management immediately before rehabilitation showed that as many as 17.1% of these patients had been sent home after the acute diagnostic phase. LOS was significantly shorter for the NT/SCI group than the T/SCI group (91.7 ± 78.9d vs 143.1 + 89.1d, p >.0001), irrespective of similar improvement (Table 3, Table 4).

      Pressure ulcers

      Similar to the T/SCI group, the worst Frankel grade for pressure ulcer evaluated on admission was highly significant (p <.0003). Both prerehabilitation management and type of etiopathology proved to be significant (p <.0000, p <.0001, respectively). No correlation was found with age (p <.123), and a marginal correlation was observed for lower lesion levels (p <.06) (table 5). In the multiple logistic regression, only 191 cases were included because of missing data. The 2 variables independently associated with pressure ulcers were worst Frankel grade on admission and, marginally, prerehabilitation management (table 5).

      Neurologic improvement

      Neurologic improvement according to the Frankel classification system from admission to discharge was evaluated for 204 cases, because data for 13 were missing. The univariate analysis identified a significant association between different cause (p <.0169), presence of pressure ulcers on admission (p <.0148), and the worst Frankel grade on admission (p <.0000) (table 6). Again it seems that patients with grade C (64%) and B (44%) of the Frankel classification system had a greater probability of improvement (fig 1).
      The variables that independently predicted neurologic improvement on discharge, evaluated with a multiple logistic regression, were again the worst Frankel grade on admission, rehabilitation center in the local health administration area rather than another health district, and, marginally, management immediately before rehabilitation and etiopathology (table 6).

      Length of stay

      LOS was positively associated with admission to a particular rehabilitation center, mean LOS ranged from 52.5 to 193.2 days (p <.0000), presence of a pressure ulcer on admission (p <.0267), worst Frankel grade on admission (p <.0004), management immediately before rehabilitation (p <.0000), etiopathology of the lesion (p <.0176), and higher lesion level (p <.0018) (table 7).
      The model that independently predicted a longer stay at the rehabilitation center included 4 prognostic variables, 2 of which were identical to the T/SCI group—worst Frankel grade on admission and higher level of the lesion—whereas 2 were different yet consistent with the other predicting models: etiopathology of the lesion and management immediately before rehabilitation (table 7).
      As with the T/SCI group, a predictive model for the LOS was fitted and associated with worse neurologic damage on admission: number of days = 51.3 (D) + 93.1 (C) + 126.1 (B) + 103.9 (A).
      This model was highly significant (p >.0001) and had an R2 of 63% (with 4 df) of A, B, C, D, representing the grades in the Frankel classification system; the presence of 1 of the Frankel grades produced the average LOS expressed by the correspondent parameter, which was also highly significant and could be interpreted as in the T/SCI group.

      Discussion

      An important objective of clinicians who currently manage SCI patients is to define crucial parameters capable of predicting outcome, in terms of both good recovery and quality of life once back in society. Economic analysis is also important for providing decision-makers with sound evidence of the relative cost effectiveness of different strategies for the individual patient.
      This study set out to verify, at a low cost and in a short time, the effectiveness of a multicenter retrospective collaboration of specialists, who collaborated for the first time in Italy to evaluate the clinical correlation of the management of these patients and to increase knowledge of neurologic recovery. They aimed to design more successful rehabilitation programs based on realistic goals. This intent led to the extraction of basic information with minimal bias from clinical records as far back as 7 years, with the clear understanding that the findings would be of limited value but that the collaborative experience could be repeated and improved in a larger prospective study.

      Pressure ulcers

      Pressure ulcers are generally considered the most common dermatologic complication of SCI, accounting for 25% of the total cost of caring for SCI patients, as well as being a prominent cause of morbidity and mortality.
      • Byrne DW
      • Salzberg CA.
      Major risk factors for pressure ulcers in the spinal cord disabled: a literature review.
      It has been estimated that between 7% and 8% of SCI patients die because of this complication and more than 70% suffer from multiple pressure ulcers.
      • Byrne DW
      • Salzberg CA.
      Major risk factors for pressure ulcers in the spinal cord disabled: a literature review.
      In this study, the presence of a pressure ulcer on admission was remarkably common—one third of the patients, regardless of the pathogenesis of the lesion, had at least 1 pressure ulcer. Generally, the incidence of pressure ulcers on admission and during the stay at the rehabilitation center is reported as being between 11% and 21%.
      • Hart C
      • Williams E.
      Epidemiology of spinal cord injuries: a reflection of changes in South African society.
      • Maharaj JC.
      Epidemiology of spinal cord paralysis in Fiji: 1985-1994.
      • Soopramanien A.
      Epidemiology of spinal injuries in Romania.
      However, in 1 study
      • da Paz AC
      • Beraldo PS
      • Almeida MC
      • Neves EG
      • Alves CM
      • Khan P.
      Traumatic injury to the spinal cord. Prevalence in Brazilian hospitals.
      it was as high as 54%, and correlated with the severity of the neurologic impairment (worst grade of Frankel classification system), confirming that severity, rather than lesion level, appears to be the major risk factor determining pressure sores.
      • Byrne DW
      • Salzberg CA.
      Major risk factors for pressure ulcers in the spinal cord disabled: a literature review.
      In fact, the former indicates the grade of motor and sensory impairment could facilitate the appearance of a pressure ulcer through confinement to bed or wheelchair, absence of adequate preventive mobilization, and proper devices. Indeed, in this multiple model, the level of the lesion showed only a marginal statistical significance for patients with T/SCI. Other related variables included delay from the event to admission and to which rehabilitation center the patient was admitted, confirming the findings of Yarkony and Heinemann
      • Yarkony GM
      • Heinemann AW.
      Pressure ulcers.
      in their large-scale, long-term study of pressure ulcers, where a prompt admission to a Spinal Cord Injury Model Systems center was likely to reduce the incidence of pressure ulcers significantly. Delayed admission because of pressure ulcers is not caused by their management before transfer to the rehabilitation center, because pressure ulcers are not usually dealt with by acute wards in Italy, but by a comprehensive rehabilitation program.
      Though it would seem obvious that immobilization plays an important role in increasing the risk of pressure ulcers, the delay variable confirms that the general approach to SCI lesions in Italy is unacceptable for this high-risk group. Currently, patients spend too much time in wards that are not suitable for their particular needs in terms of both nursing care and equipment. Moreover, a cost-benefit evaluation has estimated that pressure ulcer prevention would cost less than one tenth of the amount spent on treatment.
      • Byrne DW
      • Salzberg CA.
      Major risk factors for pressure ulcers in the spinal cord disabled: a literature review.
      It can be speculated that the lower prevalence of pressure ulcers in other Western countries is in part related to the multidisciplinary team that is activated as soon as the injury occurs, and works together to prevent such complications as additional pressure ulcers, deep venous thrombosis, urinary tract infections, and respiratory tract infections, and to optimize the treatment of neurologic damage and the psychologic support for both patients and their relatives.
      • DeVivo MJ
      • Kartus PL
      • Stover SL
      • Fine PR.
      Benefits of early admission to an organised spinal cord injury care system.
      The correlation between pressure ulcers and admission to 1 of the 7 rehabilitation centers could be attributed to inconsistencies in the quality of preventive care and local policies regarding pressure ulcers. Indeed, to contain medical costs, some rehabilitation centers give a low priority to admission of patients with pressure ulcers.

      Neurologic improvement

      Thirty-four percent of the patients admitted to rehabilitation centers showed neurologic improvement before discharge according to the Frankel grade, which was equally distributed between the T/SCI and NT/SCI groups. This finding is in agreement with other studies
      • Maharaj JC.
      Epidemiology of spinal cord paralysis in Fiji: 1985-1994.
      • Biering-Sørensen E
      • Pedersen V
      • Clausen S
      Epidemiology of spinal cord lesions in Denmark.
      regarding the overall estimate, however, our study does not confirm the significant improvement reported for cervical lesion compared with dorsolumbar lesion (38.3% vs 33.5%) or for the NT/SCI group compared with the T/SCI group (34.1% vs 34.4%).
      • Hart C
      • Williams E.
      Epidemiology of spinal cord injuries: a reflection of changes in South African society.
      Other studies
      • Silberstein B
      • Rabinovich S.
      Epidemiology of spinal cord injuries in Novosibirsk, Russia.
      on the acute phase report a globally greater improvement. The variable that best predicted improvement was neurologic impairment on admission (with a quadratic relationship), and greater improvement occurred if the patient had a midlevel Frankel grade (ie, C or B) (fig 1).
      • Soopramanien A.
      Epidemiology of spinal injuries in Romania.
      In fact, incomplete lesions had a 17.2 OR (95% confidence limit, 11.3N26.4), indicating a 17 times higher chance of improvement than complete lesions (Frankel grade A). In terms of absolute risk, 60.3% of patients with incomplete lesions showed improvement compared with only 8.1% of those with a complete lesion, equal to a 50% absolute reduction in the likelihood of improvement for complete lesions. These findings are supported by results reported in literature.
      • Maharaj JC.
      Epidemiology of spinal cord paralysis in Fiji: 1985-1994.
      • Biering-Sørensen E
      • Pedersen V
      • Clausen S
      Epidemiology of spinal cord lesions in Denmark.
      • Bravo P
      • Labarta C
      • Alcaraz MA
      • Mendoza J
      • Verdú A
      An assessment of factors affecting neurological recovery after spinal cord injury with vertebral fracture.
      Prevention of complications and management immediately before rehabilitation also played a part in the improvement of SCI patients in general, as did the rehabilitation center for the T/SCI group (whether the rehabilitation center was or was not in the local health administration area) and cause for the NT/SCI group.
      Bearing in mind that neurologic improvement is only 1 aspect of the total rehabilitation program, these results confirm the urgent need for evidence-based guidelines for prevention and treatment in Italy.

      Length of stay

      The mean hospitalization time was 130 days (range 1N922d) with a median of 121 days. This is comparable with data from rehabilitation centers in Brazil
      • da Paz AC
      • Beraldo PS
      • Almeida MC
      • Neves EG
      • Alves CM
      • Khan P.
      Traumatic injury to the spinal cord. Prevalence in Brazilian hospitals.
      (126d, which probably included the acute phase), Spain,
      • Garcìa-Reneses J
      • Herruzo-Cabrera R
      • Martinez-Moreno M
      Epidemiological study of S.C.I. in Spain 1984-85.
      and Portugal
      • Martins F
      • Freitas F
      • Martins L
      • Dartigues JF
      • Barat M
      Spinal cord injuries—epidemiology in Portugal's central region.
      (140d); in the United States, the mean LOS since 1990 has been as low as 68 days.
      • Yarkony GM
      • Roth EJ
      • Meyer PRJ
      • Lovell L
      • Heinemann AW
      • Betts HB.
      Spinal cord injury care system: fifteen-year experience at the Rehabilitation Institute of Chicago.
      In our study, the most relevant predictor of a greater mean LOS was the severity of the lesion, as found in Denmark
      • Biering-Sørensen E
      • Pedersen V
      • Clausen S
      Epidemiology of spinal cord lesions in Denmark.
      and the Netherlands.
      • Schönherr MC
      • Groothoff JW
      • Mulder GA
      • Eisma WH
      Rehabilitation of patients with spinal cord lesions in The Netherlands: an epidemiological study.
      The high statistical difference in LOS between the T/SCI and NT/SCI groups can be explained by the worse clinical (eg, associated injuries, pressure ulcers) and neurologic condition of the former group, as confirmed by the multiple analysis. A similar difference in LOS was found by Schönherr et al (205 vs 85d).
      • Schönherr MC
      • Groothoff JW
      • Mulder GA
      • Eisma WH
      Rehabilitation of patients with spinal cord lesions in The Netherlands: an epidemiological study.
      In the NT/SCI group, management immediately before rehabilitation was found to be relevant, probably because of the high percentage of patients discharged after the acute phase and sent home before being admitted to a specific rehabilitation center.
      The positive correlation between LOS and pressure ulcers found in T/SCI patients confirms that pressure ulcers determine a direct cost to the national health system and that cost-efficient and timely prevention programs are called for to contain them.
      • Byrne DW
      • Salzberg CA.
      Major risk factors for pressure ulcers in the spinal cord disabled: a literature review.
      It should be emphasized that it was not possible to quantify other important factors that are likely to be related to LOS, such as lack of equipment in the acute setting, unsuitable patient living conditions, or nursing homes.

      Conclusion

      From our study, it appears that rehabilitation should no longer be considered separate from management of the acute phase of SCI patients. A multispecialist team, including neuro-orthopedic surgeons, intensive care specialists, and physiatrists, should commence work as soon as the patient is admitted to hospital after the SCI to design a well-coordinated medical and social program. Although there are sporadic episodes of this in Italy, the holistic approach needs to become general practice to enhance prompter intervention, to shorten LOS, to reduce complications, to identify the best evidence-based treatment, and to develop the resources needed to guarantee a better quality of life. This would also reduce overall costs to the national health system.
      This study confirms the need for a heterogeneous clinical approach to SCI patients. The wide range of patients admitted to the rehabilitation centers with pressure ulcers and the wide range in LOS is an indirect confirmation of this. The analysis of clinical risk factors has proved that pressure ulcers are important in improving clinical recovery and shortening the rehabilitation LOS. Patient management before rehabilitation should be shortened as soon as possible, and if this is not possible owing to organizational problems, management should at least be performed with reproducible protocols. The lapse of time before admission to a rehabilitation center determines the length of rehabilitative hospitalization; unfortunately, this study is not able to provide further information on this topic. Currently, a prospective multicenter study on a number of the earlier-mentioned aspects is in progress.

      Acknowledgements

      The authors thank Michele Kildea and Hilary Ann Giles, MA, for their English language advice and assistance.
      The Retrospective Study Group on SCI includes the following institutions and individuals: Centro di Riabilitazione, Villanova d'Arda-Piacenza: Domenico Nicolotti, Sergio Lotta; D.R.R.F.-Unità Spinale, Vicenza: Alessandro Giovannini, Luigi Caldana; Fisiatria Nord-Osp. Gervasutta, Udine: Marsilio Saccavini, Luigi Lovati; S.R.R.F.-Unità Spinale, Turin: Maria Luisa Macchia, Valeria Chiapatti, Daniel Loria; S.R.R.F. Osp. S. Anna, Ferrara: Maurizio Zambelli, Nino Basaglia; S.R.R.F. Osp. Santorio, Trieste: Stefano Mosetti, Paolo Di Benedetto; and U.O. di Riabilitazione Intensiva Neuromotoria, Trevi-Perugia: Silvano Baratta, Mauro Zampolini, Marco Franceschini.

      References

        • Cawley MF
        • Yarkony GM
        • Bergman SB.
        Spinal cord injury rehabilitation. 5. Through the lifespan.
        Arch Phys Med Rehabil. 1997; 78: S73-S78
        • Westgren N
        • Levi R.
        Quality of life and traumatic spinal cord injury.
        Arch Phys Med Rehabil. 1998; 79: 1433-1439
        • Waters RL
        • Yakura JS
        • Adkins RH
        • Sie I.
        Recovery following complete paraplegia.
        Arch Phys Med Rehabil. 1992; 73: 784-789
        • Tator CH
        • Duncan EG
        • Edmonds VE
        • Lapczak LI
        • Andrews DF.
        Neurological recovery, mortality and length of stay after acute spinal cord injury associated with changes in management.
        Paraplegia. 1995; 33: 254-262
        • Schönherr MC
        • Groothoff JW
        • Mulder GA
        • Eisma WH
        Functional outcome of patients with spinal cord injury: rehabilitation outcome study.
        Clin Rehabil. 1999; 13: 457-463
        • Blumer CE
        • Quine S.
        Prevalence of spinal cord injury: an international comparison.
        Neuroepidemiology. 1995; 14: 258-268
        • Caldana L
        • Lucca L.
        Epidemiological remarks on traumatic spinal cord injuries and non traumatic spinal cord diseases in Veneto 1994-1995.
        Eur Med Phys. 1998; 34: 159-168
        • Di Carlo S
        • Taricco M
        • Adone R.
        Studio epidemiologico sulle lesioni midollari. Considerazioni su 262 pazienti ricoverati presso l'ospedale di Passirana di Rho negli anni 1977-1988.
        in: 2nd National Congress of the Italian Paraplegia Medical Society; Milan1990: 417-420
        • Eastwood EA
        • Hagglund KJ
        • Ragnarsson KT
        • Gordon WA
        • Marino RJ
        Medical rehabilitation length of stay and outcomes for persons with traumatic spinal cord injury—1990-1997.
        Arch Phys Med Rehabil. 1999; 80: 1457-1463
        • Chen D
        • Apple Jr, DF
        • Hudson LM
        • Bode R
        Medical complications during acute rehabilitation following spinal cord injury—current experience of the Model Systems.
        Arch Phys Med Rehabil. 1999; 80: 1397-1401
        • Kiwerski J
        • Weiss M.
        Neurological improvement in traumatic injuries of cervical spinal cord.
        Paraplegia. 1981; 19: 31-37
        • Franceschini M
        • Baratta S
        • Zampolini M
        • Lotta S
        • Loria D
        • Di Benedetto P.
        Studio multicentrico sulle lesioni midollari traumatiche: implicazioni per la prevenzione e la programmazione sanitaria.
        Eur Med Phys. 1996; 32: 207-213
        • Frankel HL
        • Hancock DO
        • Hyslop G
        • Melzak J
        • Michaelis LS
        • Ungar GH
        • et al.
        The value of postural reduction in the initial management of closed injuries of the spine with paraplegia and tetraplegia.
        I. Paraplegia. 1969; 7: 179-192
        • Gardner S
        • Winter P
        • Gardner M.
        CIA.
        1.0 ed. : British Medical Journal, London1989
      1. Statistics with confidence. : British Medical Journal, London1989
        • Byrne DW
        • Salzberg CA.
        Major risk factors for pressure ulcers in the spinal cord disabled: a literature review.
        Spinal Cord. 1996; 34: 255-263
        • Hart C
        • Williams E.
        Epidemiology of spinal cord injuries: a reflection of changes in South African society.
        Paraplegia. 1994; 32: 709-714
        • Maharaj JC.
        Epidemiology of spinal cord paralysis in Fiji: 1985-1994.
        Spinal Cord. 1996; 34: 549-559
        • Soopramanien A.
        Epidemiology of spinal injuries in Romania.
        Paraplegia. 1994; 32: 715-722
        • da Paz AC
        • Beraldo PS
        • Almeida MC
        • Neves EG
        • Alves CM
        • Khan P.
        Traumatic injury to the spinal cord. Prevalence in Brazilian hospitals.
        Paraplegia. 1992; 30: 636-640
        • Yarkony GM
        • Heinemann AW.
        Pressure ulcers.
        in: Spinal cord injury. Clinical outcomes from the model systems. : Aspen, Gaithersburg (MD)1995: 100-116
        • DeVivo MJ
        • Kartus PL
        • Stover SL
        • Fine PR.
        Benefits of early admission to an organised spinal cord injury care system.
        Paraplegia. 1990; 28: 545-555
        • Biering-Sørensen E
        • Pedersen V
        • Clausen S
        Epidemiology of spinal cord lesions in Denmark.
        Paraplegia. 1990; 28: 105-118
        • Silberstein B
        • Rabinovich S.
        Epidemiology of spinal cord injuries in Novosibirsk, Russia.
        Paraplegia. 1995; 33: 322-325
        • Bravo P
        • Labarta C
        • Alcaraz MA
        • Mendoza J
        • Verdú A
        An assessment of factors affecting neurological recovery after spinal cord injury with vertebral fracture.
        Paraplegia. 1996; 34: 164-166
        • Garcìa-Reneses J
        • Herruzo-Cabrera R
        • Martinez-Moreno M
        Epidemiological study of S.C.I. in Spain 1984-85.
        Paraplegia. 1991; 28: 180-190
        • Martins F
        • Freitas F
        • Martins L
        • Dartigues JF
        • Barat M
        Spinal cord injuries—epidemiology in Portugal's central region.
        Spinal Cord. 1998; 36: 574-578
        • Yarkony GM
        • Roth EJ
        • Meyer PRJ
        • Lovell L
        • Heinemann AW
        • Betts HB.
        Spinal cord injury care system: fifteen-year experience at the Rehabilitation Institute of Chicago.
        Paraplegia. 1990; 28: 321-329
        • Schönherr MC
        • Groothoff JW
        • Mulder GA
        • Eisma WH
        Rehabilitation of patients with spinal cord lesions in The Netherlands: an epidemiological study.
        Spinal Cord. 1996; 34: 679-683