Introduction

Several authors stated that coronary artery disease (CAD) and hypertension (HT) are more prevalent in patients after spinal cord injuries (SCI) than in the general population.1 Yekutiel et al.2 reported a significantly increased prevalence of HT and ischemic heart disease among persons with SCI compared with controls.2 Groah et al.3 reported 1.44 relative risk for cardiovascular disease in SCI patients with A, B, C Frankel or AIS grades compared with grade D patients.3 Cragg et al.4 found that SCI, after adjusting for age and gender, was associated with a significantly increased odds of reported heart disease (adjusted odds ratio 2.72), in a cross-sectional Canadian Community Health Survey.4 Bauman et al.5 reported silent cardiac ischemia revealed by cardiac stress testing with thallium-201 imaging, in 13 of 20 middle-aged subjects with paraplegia (65%), 5 of whom also had ECG ischemic changes during stress.5 Lee et al. reported that a stress test showed evidence for silent CAD in 64% of patients with SCI.6 Zhu et al. showed that 39–60% of veterans with SCI had a diagnosis of HT.7 In a group with SCI, Orakzai et al. identified a significant increase in coronary artery calcium score and concluded that patients with SCI have greater atherosclerotic burden than able-bodied controls.8 Miyatani et al. found abnormal aortic pulse wave velocity, which indicates elevated aortic arterial stiffness, an independent CAD predictor, in 36% of patients with paraplegia.9

These reports are supported by plenty of evidence for the presence of known risk factors for atherosclerosis in SCI patients, attributed to the sedentary character of these patients. Tharion et al. observed fasting hyperglycemia in 19%, glucose intolerance in 23%, and low-level high-density lipoprotein (HDL) in 58% of subjects with SCI. They concluded that glucose intolerance and dyslipidaemias are common among paraplegic and tetraplegic individuals, and that these metabolic derangements may contribute to increased cardiovascular morbidity.10 Krum et al. reported low-HDL cholesterol levels >10 years after SCI.11 Lee et al. showed that in individuals with SCI, C-reactive protein was significantly associated with the presence of other well-known cardiovascular disease risk factors, including multiple lipid abnormalities, metabolic syndrome, insulin resistance and elevated Framingham risk score.12 Finnie et al. reported that high-sensitivity C-reactive protein values indicated 36.7% of chronic SCI patients as being at high-coronary heart disease (CHD) risk.13 Bauman and Spungen noted that in patients with SCI, there is an inverse relationship between serum HDL cholesterol values and abdominal circumference, and a direct relationship between serum triglycerides levels and abdominal circumference. They described lower serum HDL levels, and a higher prevalence of insulin resistance and diabetes mellitus, comparing SCI patients with able-bodied controls. They concluded that individuals with SCI have an increased prevalence of abnormalities in carbohydrate and lipid metabolism because of immobilization, muscle atrophy and relative adiposity.14 Libin et al. found in 76.9% of the SCI population, clustering of cardio-metabolic risk factors, the most common of which were overweight or obesity, high levels of low-density lipoprotein and low levels of HDL.15

Along with these reports, however, some conflicting findings were also reported. Groah et al., comparing SCI patients with A, B, C Frankel or AIS grades with grade D patients, found a relative risk for CAD that tended to be lower in the more severe lesions (0.8).3 Tharion et al. found that total cholesterol was abnormal in 2% only.10 Krum et al. demonstrated a low-cardiovascular risk in SCI population, with relatively low blood pressure levels and frequencies of risk factors, such as total cholesterol and smoking, which are comparable with those of the general population.11 LaVela et al. showed that the odds for CHD were significantly lower for veterans with SCI aged 65 years or older, when compared with other veterans or with persons from the general population. They concluded that SCI appeared to be protective of CHD.16

Barry et al. showed that veterans with traumatic SCI have a significantly lower prevalence of HT than matched non-injured control subjects.17 The Finnie et al. group found that the prevalence of metabolic syndrome was up to 5.4 times lower in SCI participants compared with general population, and noted that the Framingham risk score categorized only 3.1% of participants as being at high 10-year CHD risk.13 Finally, Saunders et al. showed that the relation between injury level and ambulatory status and CAD, and that between years post-injury and HT were non-significant when controlling for the other factors.18

This short review reflects the fact that despite the well-documented risk factors, not all risk-factor studies indicate an increased risk for CAD in SCI patients, and direct evidence for increased prevalence of CAD after SCI is scant. Moreover, direct evidence for increased CAD after SCI was found by comparing selected SCI patients (for example, patients wounded in wars) with specific subpopulations (for example, healthy employed persons). The patients did not represent the entire Israeli SCI patients population, and the controls, who came for a screening test, did not represent the general Israeli population, which includes people with various health problems and levels of education.2 The argument that the CAD prevalence is underestimated because of diagnostic difficulties in patients with SCI1, 13 can be contradicted by the claim that the tight medical follow-up in many SCI units, with relatively frequent routine ECG tracings and blood enzymes examination, increases the probability of CAD diagnosis after SCI, in comparison with the general population. In addition, the sympathetic disruption in patients with tetraplegia or high paraplegia may be a factor that protects these patients from prolonged HT and CAD.11 It seems, therefore, that the customary belief that the risk for CAD and HT is increased after SCI requires further evaluation. If the protecting effect of SCI supersedes the effect of risk factors, SCI patients could have a CAD and HT prevalence which is lower than that of the general population.

In order to further assess the risk of atherosclerotic pathologies after SCI, we studied the prevalence of CAD, myocardial infarction (MI) and HT, in a non-selected population with traumatic and non-traumatic SCI. We also studied the effect of specific patient characteristics and of atherosclerosis risk factors on the prevalence of these conditions. Unlike previous studies, which enrolled selected SCI sub-groups and specific control subjects, the present study examined patients from diverse population categories, from across the country, and compared them with people from the general population who have similar characteristics.

Subjects and methods

The records of the SCI patients admitted consecutively for rehabilitation or check-up examinations between 1967 and 2011 to Loewenstein Rehabilitation Hospital, in Raanana, Israel were screened for inclusion in the study. Inclusion criteria were Frankel grade A or B after injury, age over 35 years and a time interval of at least 5 years between SCI onset and the last documented follow-up. SCI onset in patients with non-traumatic injuries was defined as the earliest event related to the SCI in the patient’s hospital records. The initial screening process produced records of 249 patients whose characteristics did not conflict with the inclusion criteria. For some of the 249 patients, however, certain data were missing or the follow-up was too short for inclusion. For these patients, we sought complementary data in two additional sources of information: the ‘Ofec’ computerized medical data system of Clalit Health Services, and telephone interviews. Ofec provided data on 112 patients, and 99 patients were interviewed (there was some overlap between Ofec and the interviews). After all the data (hospital records, Ofec and interviews) were reviewed, 95 of the initial 249 patients were excluded from the analysis because of missing data or because the complementary data were inconsistent with the inclusion criteria, leaving 154 participants (Figure 1).

Figure 1
figure 1

Flow chart of patient selection.

Full size image

Patient characteristics were described by their distribution in the patient population or by their mean value ±s.d. (mean and median values were very close). The list of diagnoses of the patients who were considered as having CAD included, in addition to CAD itself, ischemic heart disease, MI, angina pectoris and status post percutaneus transluminal coronary angioplasty. Patients were considered as having HT if the diagnoses in their medical records included HT, if they reported HT in the interview, if their medication included antihypertensive drugs without an indication that they were needed for another medical problem, or if blood pressure values of 140/90 or above were found in their records more than once.

Among the affecting factors we examined were age, gender, traumatic or non-traumatic etiology of the SCI, Frankel grade, the level of injury, hypercholesterolemia, smoking, body mass index (BMI), years of education, having a steady partner and having diabetes mellitus (DM). DM was diagnosed if fasting blood sugar of 126 mg dl−1 or above was recorded more than once, if use of anti-diabetic medication was recorded, or if the patient reported DM at interview.

The prevalence of atherosclerotic pathologies and factors affecting them in the study population were compared with published data of general populations from Israel and the US. We used a χ2-test to statistically compare the prevalence in the study population with that in samples of similar size, in which the atherosclerotic pathologies and factors affecting them were distributed similarly to the general population. The comparison was performed after adjusting the prevalence in the general population for the age, gender and years of education of the patients included in the current study, because the data published on the general population show that these factors tend to affect the prevalence of the dependent variables examined.19, 20, 21, 22, 23 We used linear interpolation across age and education strata in the general population to find comparison values for the averages observed in our data. Because data concerning the general population were available for each gender separately, they were combined according to the gender mix in our data. Data on years of education in the general Israeli population were missing for patients with CAD, and the prevalence of patients with CAD in this population was adjusted, therefore, for the years of education of patients with MI, which is available.

We considered BMI>30 to be obesity for the present study to allow comparison with data published on the US general population.20 Hypercholesterolemia was defined for Israel general population and for this study, as reported total cholesterol levels >200 mg dl−1 for men and >220 mg dl−1 for women, and for the US general population as >240 mg dl−1.21, 23 The data available and presented for hypercholesterolemia in the Israeli general population are of cholesterol and/or triglycerides.21 The hypercholesterolemia rate in the general population was adjusted for age and gender only, using linear interpolation across age strata and data combination according to gender mix, and may have been higher if it had also been adjusted for years of education. The data available for a steady partner in the general population are not adjusted for age, gender or education.20, 21

P-values <0.05 were considered significant. Association between categorical variables was tested using Fisher’s exact test or χ2-tests. We used stepwise Cox regression models to find the variables associated with developing CAD, MI and HT. We examined the effect of injury severity and level on the atherosclerotic pathologies and factors affecting them using both a stepwise Cox regression model and a chi-square test. The statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS version 21.0, Chicago, IL, USA).

Results

Patient characteristics

The characteristics of the 154 included patients are shown in Table 1. The follow-up range after SCI onset was 5–55 years for the entire sample, as well as for 120 patients who were alive at the last documented follow-up. For about a quarter of these (n=31), the follow-up range was 5–10 years. Of the 154 patients who were included in the study, about 5% were diagnosed as having CAD or HT before the SCI onset (2 CAD, 2 HT and 2 CAD and HT.

Table 1 Patient characteristics
Full size table

Primary outcomes

Evidence for CAD, MI and HT, was found in 21, 12 and 47 of all the included patients, and in 14, 8 and 35 of the patients who were still alive at the last documented follow-up (11.7, 6.7 and 29.2%). Corresponding data, reported by living persons of similar age, gender and education groups in Israel (2007–2010) and the US (2011) are presented in Table 2.19, 20, 21 Differences between study outcomes and corresponding data in the general population were not statistically significant, except for HT prevalence, which was significantly higher in the corresponding US general population.

Table 2 Prevalence of atherosclerotic pathologies (percentage of persons alive with CAD, MI or HT) in patients with spinal cord injury (the number of the patients in brackets) and in the Israeli and US general population, after adjustment for age, gender and years of education
Full size table

Risk factors

Risk factors evident among the persons still alive at the last documented follow-up in this study, as well as corresponding data, reported by living persons of similar age, gender and education in the general population in Israel (2007–2010) and the US (2007–2012) are presented in Table 3.19, 20, 21, 22, 23

Table 3 Prevalence (%) of persons with factors affecting coronary artery disease and hypertension among the patients with spinal cord injury of this study who were alive at the end of follow-up (the number of the patients in brackets), and in Israeli and US general populations, after adjustment for age, gender and years of education
Full size table

When controlling for all the examined variables, in the entire patient group included in this study a significant relative risk for CAD was found only for BMI>30. DM and traumatic SCI tended to increase the risk of CAD, but the increase in risk they caused was not statistically significant. A steady partner decreased the risk of CAD and MI. Hypercholesterolemia and older age at injury were found to increase the risk for HT. The values of the hazard imposed by these risk factors are shown in Table 4. Male gender, current and past smoking, and fewer years of education were not found to be a significant hazard.

Table 4 The relative risk (hazard) for CAD, MI or HT imposed by some of the examined risk factors, when controlling for all the examined variables in the entire patient group included in the study
Full size table

Effect of injury severity and level

The effect of injury severity (Frankel grade A or B) or the level C1–T3 vs T4–T6 or C1–T6 vs T7–L5 on the odds of acquiring CAD or MI was non-significant, although Frankel grade A tended to increase the risk of CAD (hazard 5, confidence interval 0.9–28.2, P=0.69). The effect of injury severity or the level C1–T6 vs T7–L5 on the odds of acquiring HT was also non-significant. The level T4–T6 vs C1–T3, however, had a significant effect on the odds of acquiring HT (52 vs 23.3%, P=0.02).

Injury level affected some of the risk factors for CAD and HT. DM was found in 29% in patients with T7–L5 SCI, and only in 12.9% of the patients with C1–T6 SCI (P=0.016). Past smoking was found in 26.1% in patients with T7–L5 SCI, and only in 11.8% of the patients with C1–T6 SCI (P=0.034), and hypercholesterolemia was found in 47.8% of the patients with T7–L5 SCI, and only in 30.6% of the patients with C1–T6 SCI (P=0.032). Patients with C1–T6 SCI had 11.65 years of education, and patients with T7–L5 SCI had 10.25 years of education (P=0.016).

Discussion

The findings of the present study hardly support the common belief and statements that SCI patients are prone to developing CAD and HT because of their sedentary lifestyle. These statements lack strong direct evidence, and most of them cite earlier statements made over a period of more than two decades. In the present paper we re-evaluate these statements based on a study involving an SCI population that is likely to represent the SCI population of Israel, as well as general populations with comparable characteristics (age, gender and education).

After adjusting for age, gender and education, the prevalence of CAD and MI among our patients was slightly higher (but not statistically significantly) than in the Israeli and US general populations. The same was true with regard to HT, compared with the general population in Israel (Table 2).19, 20, 21 These differences between SCI patients and the general population are much smaller than those demonstrated previously between SCI patients and controls,2 and the prevalence of HT was significantly lower in our patients when compared with the corresponding US general population (Table 2).20

The implications of these results are related to risk and protecting factors of the SCI and general population, to the effects of the SCI severity and level, and to the demographic characteristics of SCI populations.

Prevalence of risk and protecting factors

Consistent with previous publications,10, 11, 14, 15, 18 certain findings in this study support the possible contribution of risk factors to acquiring atherosclerotic pathology. Our study showed that hypercholesterolemia increased the risk of HT, and BMI>30 increased the risk of acquiring CAD in our patients. The study also showed that a more severe SCI tended to increase the risk of CAD and that DM, which also tended to increase this risk, was somewhat more prevalent in our SCI patients than in the general population (Table 3).19, 20

But other findings in the current study indicate that risk factors may not promote acquiring such pathology. The prevalence of BMI>30 in our sample was lower, and the prevalence of hypercholesterolemia was quite similar when compared with the calculated value of the prevalence in a general population with similar age, gender and education in Israel or the US (Table 3).20, 23 Current and past smoking were not found to be significant risk factors in our sample, and their prevalence in our sample was also lower than the calculated value for the prevalence of corresponding Israeli or US general population (Table 3).19, 20, 22 Finally, most of the examined known risk factors were not found to be a significant hazard for HT in this sample, and most of the risk factors were less prevalent in our SCI patients than in the corresponding general population (Table 2).

The effect of injury severity and level

Regarding the effect of injury severity and level on outcomes, there are some discrepancies between our and previous findings. Whereas Groah et al. found that a more rostral level of SCI was associated with a lower risk of CAD or MI, and injury severity did not affect risk,3 our results indicate that the effects of neither injury level nor severity affect risk significantly, but Frankel grade A tends to increase the risk. Whereas Libin et al. found that systolic HT was comparable in C3–C8 tetraplegia and T1–T6 paraplegia, but was twice as prevalent in T7–L2 paraplegia,15 our results indicate that the risk of HT is similar with SCI below T6 and above T7, but higher with SCI of T4–T6 than in SCI of C1–T3. Our last finding is compatible, however, with the observation of Libin et al.15 that patients with tetraplegia tend to be hypo to normotensive. This means that the low prevalence of HT in patients with SCI above T4, is probably responsible for the relatively low prevalence of HT in our SCI sample, which may also reduce the CAD prevalence in the entire SCI population.

The effect of injury level and severity on risk factors also differs between our and previous findings. Whereas Libin et al. concluded that patients with tetraplegia tend to be more sedentary, resulting in lower HDL and a greater propensity toward impaired carbohydrate metabolism,15 our results indicate that most of the risk factors for atherosclerotic pathology, including DM, hypercholesterolemia and past smoking were not influenced by the SCI severity, and were more prevalent below T6 than above T7. The lower prevalence of such factors above T7 could contribute to lowering the prevalence of CAD and HT, and may explain the lower risk for CAD and HT at more rostral levels, as described by Groah et al.3 The variability among studies regarding the effect of injury severity and level on the risk for atherosclerotic morbidity may indicate that clustering of risk factors after SCI is not necessarily related to the SCI itself.

Demographic characteristics of SCI populations

When comparing SCI patients with control groups, several authors matched the groups for age and gender, but not for years of education. If not controlled for the years of education, the prevalence of CAD and HT in the control groups may be underestimated because, in the general population, years of education may affect the prevalence of CAD and HT even more than age or gender.20 This could cause a relative overestimation of the prevalence of CAD and HT in the compared SCI population. In this study we addressed years of education for the comparison with the general population. Our patients had, however, more years of education with SCI above T7. This could be a reason contributing to reduced CAD in patients with higher spinal cord lesions, who are prone to develop this complication, and to the lack of statistical significance of the overall effect of education on CAD prevalence in our sample.

Over 60% of our SCI patients had a steady partner, which was found to be a factor protecting from CAD. A comparison of the prevalence of this factor with the general population is difficult, however, because adjusted corresponding values in the general population are not available (Table 3).20, 21

The relative contribution of SCI to the prevalence of CAD and HT

Although SCI itself may affect the prevalence of CAD and HT, the prevalence is determined by combination of various factors. Some of the factors, such as neuropathic pain and depression,24 may be related to the SCI, whereas many others may not. The contribution of SCI to CAD because of decreased mobility becomes questionable in light of recent findings that after controlling for age, years post-injury, gender and race, mobility status became non-significant in relation to CAD.18 The overall effect of these factors depends on the specific clustering in a given population, which may be either additive or nullifying. It seems that looking at our sample, and at the relative prevalence of risk factors in it, the contribution of the SCI to acquiring CAD or HT is minor.

The clinical meaning and consequences of the findings

Our findings and the above discussion provide a new perspective on the need for physical activity and elimination of atherosclerosis risk factors in SCI patients. Contrary to previous publications, which advised or implied that such measures are required to prevent the increased risk of atherosclerotic pathologies,4, 14, 15, 18, 24, 25, 26 the present study suggests that these measures can also be used to take advantage of the protective features of SCI. Combining our outcomes with those reported earlier, we believe that physical training, glycemic and lipemic control, weight reduction, cessation of smoking and improving social support and education are imperative for reducing the CAD and HT prevalence in SCI patients to a level below that of the general population. This is the challenge, and the results of the present study indicate that it is probably achievable.

Limitations of the study

The present study has several methodological limitations. It was retrospective, a significant number of patients were excluded from the analysis because of missing data and the follow-up was relatively short in some of the studied patients (5–10 years in 26% of those alive at the end of the follow-up). In addition, the precision of the comparison with published data on the general population was limited by the need to adjust the data, by the difference in size of the compared populations and by minor differences in the definitions of the compared variables.

Conclusions

The findings of this study show that the prevalence of CAD, HT and some of their risk factors following SCI is generally, but not consistently and not statistically significantly, slightly higher than in corresponding general population, and the contribution of the SCI to acquiring CAD or HT seems minor. These findings bring up the possibility and the challenge to reduce the prevalence of atherosclerotic morbidity after SCI below that in the general population.

Data archiving

There were no data to deposit.