Statewide Investigation of Medically Attended Adverse Health Conditions of Persons With Spinal Cord Injury

Abstract

Background/Objective:

To report over a 10-year period the statewide prevalence and incidence of medically attended adverse health conditions in people with new traumatic spinal cord injury (TSCI).

Design:

Retrospective cohort study.

Methods:

(a) Identified all new TSCI cases discharged alive from statewide acute care hospitals, 1996 to 2000, using ICD-9-CM methodology. (b) Followed cases from 1996 to 2005 to quantify medically attended health conditions documented during emergency department visits, acute care hospitalizations, and outpatient hospital visits. (c) Used the life table method to calculate the prevalence and incidence of health conditions. (d) Examined Cox proportional hazard ratio of mortality by gender controlling for age and TSCI severity.

Results:

Nine hundred eighty-eight residents (257 women, 731 men) with TSCI were alive 90 days after discharge from acute care hospitalization from 1996 to 2000. Nine hundred twenty-three (251 female, 672 male) (93.4%) residents had an observed medically attended adverse health condition in the 10-year follow-up period. The most prevalent classes of diseases and disorders were (a) muscle and connective tissue, (b) renal and urinary, (c) digestive, (d) circulatory, (e) respiratory, (f) endocrine/nutritional/metabolic, and (g) infectious. Incidence of new injury was 29.0% for males and 26.9% for females. During the follow-up period, 49 women (19%) and 104 men (14%) died.

Conclusions:

People with TSCI experience diverse adverse health conditions in the 10 years after initial injury. An interdisciplinary health care provider team approach to allocating resources and implementing countermeasures to prevent or limit occurrence of these conditions is vital to these patients' continuum of care.

INTRODUCTION

People with traumatic spinal cord injury (TSCI) are at increased risk of developing on-going adverse health conditions that require hospital care (1–3) and physician office and emergency department (ED) visits (4). During the first year after injury, hospitalizations are commonly related to the original injury (1,5,6). Primary reasons for hospitalizations after year 1 include cardiopulmonary complications; urinary complications; dermatologic conditions (7); digestive, musculoskeletal, and nervous system complications; and secondary fractures of the lower limbs (8,9). Conditions considered secondary to SCI can lead to death (10,11). Although higher-level and neurologic completeness of spinal cord lesion (1,12,13) significantly increases the likelihood of hospitalization and further disability, (14) many TSCI-related secondary conditions are modifiable or preventable (6,7).

Gender differences in terms of adverse health conditions have been reported among people with TSCI (15–17). However, age comparisons between genders are potentially confounded by age at onset, duration of disability, and neurologic level of injury (2).

Morbidity and survival of people with TSCI have mainly been reported for cross-sectional samples (9,18–22) or for limited ages or degrees of injury severity (1,5,23). Thus, the ability to generalize findings to all people with TSCI is limited. To fill this void, this study reports the prevalence and incidence of medically attended health conditions across all ages and TSCI severity over a 10-year follow-up period in a statewide population. To our knowledge, this is the first article to report these findings.

METHODS

Data Sources

The Institutional Review Board of the Medical University of South Carolina approved the study. This study examined 2 statewide datasets, the State of South Carolina Acute Care and Outpatient Hospital Discharge Data Set and the South Carolina Emergency Department Visit Data Set. State law mandates that all 62 EDs and 62 nonfederal hospitals in the state report data abstracted from the discharge uniform billing system to the Office of Research and Statistics of the State Budget and Control Board. The data set contains demographics, dates of admission and discharge, up to 10 diagnosis codes documented according to the International Classification of Diseases, 9th Revision, Clinical Modifications (ICD-9-CM) (24), primary and secondary external causes of injury codes, total charges, discharge dispositions, source of admission, principal payers, types of care, and patient identifiers. Inclusion of personal identifiers prevented duplication of repeat visits for the same event. The reliability and utility of administrative health care data have been established (18,25–28). The advantages of using health care billing data are that they provide documentation of individual encounters and allow assessment of adverse health conditions in a large and complete population (25). In addition, the use of ICD-9-CM methodology to identify morbidities has been determined to be a reliable alternative to chart review (29).

Study Population

ICD-9-CM methodology identified 1,008 state residents who sustained a TSCI that resulted in an ED visit or an acute care hospital discharge in the state from January 1, 1996, through December 31, 2000. Patients (n  =  20) who died within 89 days of initial ED visit or hospital discharge were considered to have trauma-related deaths and were excluded. A case of TSCI was defined as any hospital discharge or ED visit with a primary or secondary diagnosis of TSCI in accordance with the Centers for Disease Control and Prevention case definition of incident TSCI (30). The nature of injury codes (N-codes) used were 806.0–806.9 (fracture of vertebral column with spinal cord lesion) or 952.0–952.9 (spinal cord lesion without evidence of spinal bone injury). The fourth and fifth digits of the 806 and 952 N-codes were further used to identify level and completeness of injury. Spinal cord injury type was defined as “unclassified” when these digits were missing. Other researchers have used ICD-9-CM–based methodology to identify TSCI cases (31–34) and to identify secondary conditions in people with disabilities (18). State legislative authority (§44-38-60) helped the study obtain mortality status through linkage of a unique combination of name, social security number, gender, date of birth, and last known address with the state vital records dataset.

Follow Up

For follow up, ICD-9-CM methodology, as identified in Table 1, was further used to identify morbidities recorded in the medical record during visits in 1996 to 2005 to an ED, an acute care hospital, or an outpatient hospital clinic in the state. Diagnosis codes for late effects of TSCI, aftercare, plegias, phantom limb, and complications or conditions secondary to a procedure, surgery, graft, or prevalent disease were excluded. We further excluded diagnosis codes for conditions related to pregnancy, childbirth, and puerperi/perinatal, and dental status; congenital anomalies, hereditary disorders, genetic disorders and conditions, mental retardation, vision and auditory conditions, screening visits, and status codes. Remaining health classes and conditions were classified according to ICD-9-CM methodology. Each condition was counted at each occurrence.

Table 1.

Classification of Diseases and Disorders

For each case, TSCI severity was determined by translating the ICD-9-CM diagnosis codes into an ASIA impairment scale score for the spine region using ICDMAP-90 software (35). If the fifth digit of the diagnosis code was missing, the least severe rating was assigned. Traumatic SCI severity was then categorized based on the ASIA impairment scale score as “critical” (5–6), “severe” (4), or “moderate” (2–3).

Gender differences in demographics and injury characteristics were examined using chi-square, Fisher exact, or Wilcoxon rank sum tests, where appropriate. Prevalence and incidence of disease classes and major health conditions were calculated by the time-to-event life table method and are reported separately by gender. Date of first observed diagnosis code for a specific condition was used as the time of first event. Conditions treated within 89 days of initial hospitalization were considered baseline conditions. The at-risk population at time of subsequent medically attended events was calculated as people alive at the beginning of the interval minus half of the censored observations during the observation period. Total or prevalent health conditions include baseline conditions. New or incident conditions refer to those observed 90 or more days after date of first hospital admission.

Detailed prevalence and incidence analyses of specific ICD-9-CM classes were conducted if the umbrella category had a large proportion of cases with at least 1 medically attended condition after TSCI and if the specific medical condition was potentially TSCI related. In addition, a descriptive analysis was conducted of the frequency of injuries and acute conditions per person of those who had a repeat occurrence after baseline. Injuries or acute conditions were considered new if dates of service appeared at intervals greater than 7 days. Baseline injuries are not discussed, because they are implicit in the inclusion criteria. A Cox regression survival analysis, using the Efron method (36) for ties, examined the hazard rates in men and women controlling for age and TSCI severity. Collinearity, concordance statistics, and the proportional hazard assumption were examined during model building. SAS 9.1 (37) and STATA9 (38) software were used to conduct the statistical analyses.

RESULTS

Table 2 shows the characteristics of the 731 men and 257 women who met the inclusion criteria. Overall, they differed significantly in age at TSCI, race, and type of SCI. No significant difference in the proportion of concurrent traumatic brain injury by gender was observed.

Table 2.

Patient Characteristics by Gender

In total, 923 patients (93.4%) had 1 or more postinjury ED visits, hospitalizations, or hospital-associated outpatient visits in South Carolina during the 10-year follow-up period. More specifically, 56.6% were treated and released from an ED, 62.5% were admitted to an acute care hospital, and 26.7% received care in a hospital-associated outpatient clinic. Table 3 shows frequencies of medically attended adverse disease classes and injuries classified using ICD-9-CM methodology that were observed within the first 89 days after injury, 90 days to 1 year, 1 to 5 years, and 5 to 10 years. Overall, the 7 most prevalent disease classes were muscle/connective tissue, genitourinary, renal and urinary, digestive, circulatory, respiratory, endocrine/nutritional/metabolic, and infectious.

Table 3.

Prevalence and Incidence of Health Classes and Selected Conditions in Women (W) and Men (M)

For men, the 5 most prevalent disease or disorder classes were muscle/connective tissue, genitourinary, digestive, circulatory, and respiratory. All classes had prevalence greater than 63%. For women, the 5 most prevalent classes respectively were genitourinary, muscle/connective tissue, endocrine/nutritional/metabolic, circulatory, and digestive. All classes had a prevalence rate greater than 70%. New or incident cases were observed at least 90 days after initial injury. For men, the 5 classes of diseases and disorders with highest incidence were muscle/connective tissue, digestive, genitourinary, endocrine/nutritional/metabolic, and respiratory. All classes had an incidence rate greater than 46%. For women, the 5 classes for new cases were the same as the prevalent classes, with an observed incidence of 50% or greater.

Incidence of each injury class observed 90 days to 10 years from the incident TSCI is listed by gender on Table 3. The at-risk population for incidence calculations is based on time after TSCI to first event. The occurrence of extremity fracture, dislocation, sprain, and amputation was 29% for men and 26.9% for women. Also, 13.8% of women and 11.1% of men had an observed case of traumatic brain injury; TSCI; or fracture of the head, neck, face, or spine after initial injury. The other subclassification includes a high prevalence of injury caused by complications from surgery.

Specific conditions within each class that have been reported to have a high incidence rate among people with TSCI (5,18). Table 4 reflects a breakdown of the most prevalent conditions within classes requiring medical intervention during the follow-up period at an ED, hospital, or outpatient hospital–affiliated physician office. Approximately 20% of men and women received medical attention for septicemia or bacteremia at least once. Nearly 50% of women and 41% of men had a new urinary tract infection. In addition, 24% of women and almost 31% of men were treated for chronic skin ulcers. Other new medical conditions observed in more than 20% of the population were fluid, electrolyte, or acid/base disorders; hypertension; ischemic heart disorder or chest pain; pneumonia; chronic obstructive pulmonary disease or allied condition; gastrointestinal ulcer; intestinal obstruction or constipation; pain; and disk disorders.

Table 4.

Most Prevalent Health Conditions Within Each Disease Class in Women (W) and Men (M)

Table 5 reflects descriptive statistics on repeat injuries or acute medical conditions. Of those cases with at least 1 medical condition or injury, 33.71% of women and 48.33% of men had 3 or more episodes of urinary tract infection at least 7 days apart. Nearly 50% of women and 60% of men had 3 or more episodes of skin ulcers during the follow-up period. Additionally, 28% of women and 29% of men had 3 or more episodes of dehydration and/or potassium or sodium deficiency.

Table 5.

Repeat Injuries and Repeat Acute Medical Conditions Observed During the Follow-up Period in Women (W) and Men

Mortality

During the 10-year follow-up period, 49 women (19.1%) and 104 men (14.2%) died. Using a Cox survival regression model, we found no difference between genders for mortality after controlling for age and TSCI severity (HR  =  1.19; 95% CI: 0.82, 1.74). The type 3 χ² P value for gender was 0.36. The Supremum test for proportional hazards assumption held that the 2 curves based on gender did not differ (P  =  0.61).

DISCUSSION

This study used administrative billing data and ICD-9-CM methodology to identify incident TSCI cases over a 5-year period. Overall, 988 patients met the inclusion criteria. Gender differences were noted for age, race, and type of TSCI. ICD-9-CM methodology was further used to identify medically attended adverse health conditions documented during patient admissions to acute care hospitals, outpatient hospital clinics, and EDs statewide over a subsequent 10-year follow-up period. It is of interest that, based on the life table method, the highest incidences of conditions considered secondary to TSCI were recorded in the first 5 years after injury, a finding also reported elsewhere (1,12,23,39).

Of the 988 patients, 153 (15.5%) had died over the follow-up period. Three studies reporting mortality over a 10-year period after TSCI found occurrences of 7% and 9% (21,23,40). The finding that a higher proportion of patients in our study had died is puzzling, because our study includes patients with all severities and not just those with the most severe injuries who would be discharged from acute care rehabilitation facilities. Moreover, the comparison studies did not include an 89-day exclusion period after acute care hospital discharge to determine the cohorts. However, caution should be used when comparing findings from this study with other studies, because the methods of determining mortality status for this study may have been more complete and accurate than used elsewhere.

Most Prevalent Adverse Health Classes and Conditions

Adverse health classes with a prevalence rate greater than 60% in both men and women were noted (muscle/connective tissue, genitourinary, renal/urinary, digestive, circulatory, respiratory, and endocrine/nutritional/metabolic). All disease classes have been reported as possibly secondary to TSCI (1,3,5,8,9,41–43). When examining specific conditions, high incidences of urinary tract infection, ischemic heart disease, hypertension, disc disorders, chronic skin ulcers, bowel problems, nutritional and fluid disorders, respiratory complications, and pain were observed. These findings compare with adverse health conditions reported in other studies of people with TSCI (3,44–48) and comparable disabilities (49–52). For instance, the most commonly reported adverse health conditions that required rehospitalization of patients in the Spinal Cord Injury Model System were genitourinary complications, pressure ulcers, and respiratory complications (5).

Injury and Acute Medical Conditions

Incidence (Table 3), prevalence (Table 4), and repeat occurrence (Table 5) of medically attended acute medical conditions and injury after TSCI is disturbing, particularly considering that the majority of these conditions are preventable. Of note is that the incidence of injury in this study is greater in comparison with another study (53) that reported that nearly 20% of 1,328 people who had experienced TSCI for an average of 10 years had at least 1 injury per year that required medical attention, with approximately 5% requiring hospitalization. These findings suggest that further research is needed into circumstances that contribute to preventable medical conditions and injuries after TSCI so that prevention strategies can be implemented to prevent their occurrence. In particular, pneumonia, pulmonary emboli, and septicemia would be of great concern to control or modify, because they have been reported as causes of death that appear to have the greatest impact on reduced life expectancy in the Spinal Cord Injury Model System population (54).

Strengths and Limitations

The major strength of this study centers on the utilization of a prospective cohort design with a population-based statewide sampling of all people with new TSCI who during a 10-year period received medical attention in an ED, hospital, or hospital-associated outpatient clinic. The cohort design reduces selection bias, and the prospective follow-up and assessment of medical conditions by medical record eliminates recall bias. The population-based sampling increases the ability to generalize the study's findings to all people with new TSCI, not just those with the most severe injuries who receive medical rehabilitation. Additionally, the findings can inform public policy of the importance and usefulness of medical surveillance of TSCI to inform health care providers of conditions in which to modify and control long-term TSCI. Finally, the statistical method described in this manuscript can be applied to future investigations for deeper probing into a specific condition or set of conditions.

This study has a few limitations. First, the study does not include data on health care received out of state or in nonhospital-associated outpatient clinics. Second, incidence of TSCI could be inflated, because using ICD-9-CM methodology to identify TSCI without confirming with medical record review could overestimate incidence (33). Third, the current study is reliant on coding that is normally associated with billing, and any new or “trendy” syndromes or codes may not have been consistently applied throughout the follow-up period, especially if they do not affect billing. For example, metabolic syndrome for which code 277.7 was approved in 2001 is rarely, if ever, recorded as a diagnosis in clinical practice (55). Fourth, the aim of this paper is to provide prevalence and incidence estimates in the 10 years after injury for observed medical conditions in a population with TSCI. Thus, the relative occurrence of these conditions in other populations cannot be estimated. Therefore, further research is needed to determine whether these conditions are more prevalent or occur at an increased rate compared with other injured or able-bodied populations. Fifth, this research covers a 10-year period. A longer follow-up period would allow for conditions attributable to aging with TSCI to be noted.

CONCLUSIONS

Not surprisingly, this study revealed that people with TSCI encounter a variety of adverse health conditions that require medical attention in the 10 years after injury. Yet these conditions are not solely those generally considered secondary to TSCI. These findings have explicit implications for the long-term health management of people with TSCI. We recommend that a health care team with diverse specialties, not solely rehabilitation specialists, collaborate early after injury to develop a health management plan that provides for the continuum of care for each patient with TSCI. The ultimate goal is to prevent, limit, or control modifiable adverse health conditions so that people with TSCI have the best possible health outcome, enhanced quality of life, and increased life expectancy.