Colorectal cancer mortality after spinal cord injury

Abstract

Context: Colorectal cancer (CRC) is the second leading cause of cancer related deaths in the US. There is paucity of data regarding CRC and the spinal cord injury (SCI) community. Persons with SCI have suboptimal rates of colonoscopies and face extensive barriers to care. The aim of our study was to compare CRC mortality in persons with SCI to CRC mortality in the general population.

Design: A prospective follow-up study.

Setting: Analysis of the National SCI database.

Participants: 54,965 persons with SCI.

Interventions: Not applicable.

Outcome Measures: Current survival status and causes of death were determined. The expected number of CRC deaths was calculated for the general US population, using ICD-10 codes. Standardized mortality ratios (SMR) were calculated as the ratio of observed to expected CRC deaths stratified by current age, sex, race, time post-injury and neurologic group.

Results: The CRC mortality was 146 persons out of 54,965 persons with SCI. The overall SMR was determined to be 1.11 (95% CI [0.94, 1.31]). Among subgroups, one finding was significant and this was for patients with injury level C1-4 with an American Spinal Injury Association Impairment Scale Grade of A, B or C with an SMR of 1.68 ([95% CI [1.03–2.61]).

Conclusion: Although persons with SCI receive suboptimal rates of preventative care screenings and report extensive barriers to care, overall, they are not at an increased risk of CRC mortality. The current recommendations for CRC screening should be continued for these individuals while reducing barriers to care.

Introduction

According to the National Spinal Cord Injury Statistical Center, there are approximately 291,000 persons living with a spinal cord injury (SCI) in the United States with about 17,730 new SCI cases per year.^1–3 The average age of injury has been increasing and the majority of new SCI cases are male.³ In persons with SCI, overall cancer mortality rates are not increased compared to the general population, which may be because SCI should not have a bearing on most cancers.^4,5 However, it has been demonstrated that persons with SCI have an increased risk of bladder cancer mortality attributable to SCI related circumstances.⁶ However, beyond urogenital morbidity, persons with SCI also have gastrointestinal alterations in comparison to the general population and it has been demonstrated that the majority of persons with SCI lack normal bowel function.⁷ It is unknown if these gastrointestinal alterations have an effect on gastrointestinal cancer mortality, specifically colorectal cancer (CRC).

In the United States, CRC is the second leading cause of cancer related deaths and the third most commonly diagnosed cancer.⁸ The overall incidence of CRC is decreasing, and this can be in part attributed to colonoscopy screening.⁹ Furthermore, colonoscopies have been reported to potentially decrease left-sided CRC mortality as well.^10,11 Even though these benefits of colonoscopy exist, it has been reported that approximately one-third of adults have not been screened with a colonoscopy as recommended.¹²

When looking at SCI, cancer is the third leading cause of death in the SCI community with CRC being the third most common cancer.¹³ Despite data demonstrating colonoscopies in persons with SCI are safe and therapeutically beneficial, many SCI patients have not undergone appropriate screening.¹⁴ In one study by Stillman et al., over one-third of persons with SCI greater than or equal to 50 years old had not undergone a screening colonoscopy.¹⁵ This may be attributed to the general barriers listed above as well as difficulties with more colonoscopy-specific barriers. For example, administering bowel prep can lead to frequent bathroom transfers, prolonged toilet sitting placing skin at injury risk, and abdominal discomfort which may trigger autonomic dysreflexia.¹⁶ Cancer screening is crucial in this population, especially as SCI patients have at least an equivalent CRC risk and have been found to have presentation at diagnosis with more advanced disease.¹⁷

It is important to note that the SCI population faces extensive barriers to care including inaccessible examination tables and lack of transfer aids and, in one study, nearly two-thirds of persons with SCI felt they had been provided incomplete care by a provider as a result of their SCI.¹⁵

The aim of our study was to compare CRC mortality in SCI patients to CRC mortality in the general population to determine if any disparities exist, as increasing age at SCI, low CRC screening rates and presentation of more advanced disease may be risk factors.

Methods

Institutional review board review

Institutional review board approval was obtained from each participating facility as well as the National SCI Statistical Center. The review board for the National Death Index (NDI) also approved this study, as did review boards for state vital statistics departments whenever required for release of death certificates with unredacted cause of death information.

Study population

This study utilized the same overall methods as a previous study conducted by authors M.D. and Y.C. analyzing bladder cancer in the SCI population.⁶ Three data sources were utilized to obtain participant information and SCI data for this study. The first was the National SCI Database (NSCID).¹⁸ Since its inception, 28 federally funded Spinal Cord Injury Model Systems (SCIMS) centers throughout the United States have provided data to the database.

To be qualified for the NSCID, patients must have 1) sustained SCI due to a traumatic event, 2) had a clinically discernible degree of neurological deficit and 3) received initial hospital care from one of the SCIMS within 1 year of injury.⁶

The earliest enrolled patients were injured in 1973, with ongoing enrollment to this day.

To increase sample size and to supplement mortality information contained in the NSCID, a series of collaborative SCI survival studies were conducted between 1981 and 1999, adding participants from several SCIMS who were not registered in the NSCID, primarily due to being admitted to the SCIMS more than one year after injury.¹⁹ While approximately 99.6% of participants were injured after 1970, there are participants injured as long ago as 1936 and admitted prior to official SCIMS designation as early as 1960. Using data from these patients increases representativeness of the study population and extends length of follow-up beyond what would be possible with the NSCID.

The third source of data was the National Shriners SCI Database (NSSCID) that enrolls children with SCI who received care from the three SCI units of the Shriners Hospital for Children in California, Illinois and Pennsylvania.²⁰ Patients were enrolled since the inception of the database in 1987 using the same eligibility criteria as the NSCID, except that the NSSCID includes patients admitted to the Shriners Hospital beyond one year of injury in accordance with the expanded eligibility criteria of the collaborative survival studies.^18–22 Using data from these patients increases representativeness and sample size by including young children who are not often treated at SCIMS.

A total of 54,965 individuals injured between 1936 and 2017 were enrolled in this study, including 47,644 from the NSCID, 5,199 from the collaborative survival study and 2,122 from the NSSCID.

Data collection

Demographic data and injury characteristics (sex, race/ethnicity, age at injury, neurologic group, ventilator status) for the NSCID and NSSCID were collected prospectively prior to discharge by trained personnel using an identical data collection protocol. Data for the collaborative survival studies were collected retrospectively by the same personnel responsible for NSCID data collection using the same variable definitions. Current survival status was determined by routine follow-up conducted in person, by phone, or mail, at each SCIMS or Shriners Hospital, searches of the Social Security Death Index (SSDI) and National Death Index (NDI), online state vital statistics files, online newspaper obituaries, and other online sources such as find a grave, tributes.com, legacy.com, mylife.com, etc. Searches of the NDI, autopsy reports, online obituaries, death certificates, and hospital discharge summaries, were used to determine causes of death. Patient follow-up was terminated in 2018 when data analysis was initiated.

Statistical analysis

A person-year data set was created in which each person followed for each year was treated as a separate observation.²³ Thus, a person who died during the sixth year after injury would contribute six observations to the data file. For each year, a binary outcome measure was created indicating whether the person survived or died during the year. Therefore, for this example, the person would be considered alive for the first five observations and dead for the sixth observation. Years between injury and admission to the SCIMS or Shriners Hospital were excluded to avoid bias resulting from the absence of recorded deaths prior to admission. The final data set included 815,577 observations.

For this specific analysis, the variables utilized included sex, race/ethnicity, current age, years since injury, and neurologic group. Since 1982, non-ventilator dependent patients were grouped in accordance with the concurrent version of the International Standards for Neurological Classification using the American Spinal Injury Association Impairment Scale (AIS).²⁴ Prior to 1982, Frankel’s classification scale was used to grade completeness of injury.²⁵ Current age and years since injury were determined using the age at injury and date of injury as baseline values.

Next, the expected and actual number of CRC deaths was calculated from this person-year data set based on the sex, race, and current age during each follow-up year. Age-sex-race-cause-specific CRC mortality rates for the 2001 calendar year (the mid-year of patient follow-up) for the general US population were used to calculate expected numbers of CRC deaths using ICD-10 codes C18–C20.²⁶ Standardized mortality ratios (SMR) along with their 95% confidence limits were calculated as the ratio of observed to expected CRC deaths stratified by current age, sex, race, time post-injury, and neurologic group in accordance with pre-specified analytical plans.

Results

The patient demographics were recorded for the total participants (n=54,965) and participants with colon cancer deaths (n=146) (Table 1). The SMRs were determined for the different patient demographics (Table 2). The total CRC mortality in the population with SCI was 146 persons out of 54,965 persons, which is greater than the expected 131.2 deaths. The overall SMR using these values was determined to be 1.11 (95% CI [0.94, 1.31]) indicating an 11% increase in mortality due to CRC compared to the general population of comparable age, sex, race, and length of follow-up; however, this was not statistically significant.

Table 1. Demographics of participants.

Characteristic	Total Participants		Colon Cancer Deaths
	n	%	n	%
Sex
Male	43,632	79.4	118	80.8
Female	11,333	20.6	28	19.2
Race/Ethnicity
White	38,384	69.8	105	71.9
African American	10,801	19.7	35	24.0
Other Race/Ethnicity	5,480	10.0	6	4.1
Unknown	300	0.6	0	0.0
Age at Injury
0–29	27,838	50.7	30	20.6
30–59	21,092	38.4	83	56.9
60+	6,035	11.0	33	22.6
Neurologic Group
Ventilator-Dependent	1,379	2.5	1	0.7
C1-4 AIS ABC	7,201	13.1	18	12.3
C5-8 AIS ABC	10,634	19.4	16	11.0
T1-S3 AIS ABC	17,847	32.5	44	30.1
All Levels AIS D	13,286	24.2	52	35.6
Unknown	4,618	8.4	15	10.3
Year of Injury
1936–1970	327	0.6	2	1.4
1971–1980	8,276	15.1	50	34.3
1981–1990	12,777	23.3	47	32.2
1991–2000	14,119	25.7	35	24.0
2001–2010	12,260	22.3	12	8.2
2011–2017	7,206	13.1	0	0.0

American Spinal Injury Association Impairment Scale Grade of A, B or C (AIS ABC).

Table 2. Results including SMR values.

Factor	Actual Deaths	Expected Deaths	SMR	95% Lower Limit	95% Upper Limit
Overall	146	131.2	1.11	0.94	1.31
Sex
Male	118	107.2	1.10	0.92	1.31
Female	28	24.1	1.16	0.79	1.66
Race/Ethnicity
White	105	88.3	1.19	0.98	1.43
African American	35	34.8	1.01	0.71	1.38
Other Race/Ethnicity	6	7.7	0.78	0.32	1.62
Unknown	0	0.3
Current Age
0–29	1	0.8	1.25	0.06	6.17
30–59	57	48.0	1.19	0.91	1.53
60+	88	82.4	1.07	0.86	1.31
Years Since Injury
<1	4	7.6	0.53	0.17	1.27
1–9	33	33.6	0.98	0.69	1.36
10–19	47	36.0	1.31	0.97	1.72
20+	62	54.0	1.15	0.89	1.46
Neurologic Group
Ventilator-Dependent	1	1.1	0.91	0.05	4.48
C1-4 AIS ABC	18	10.7	1.68	1.03	2.61
C5-8 AIS ABC	16	20.2	0.79	0.47	1.26
T1-S3 AIS ABC	44	36.3	1.21	0.89	1.61
All Levels AIS D	52	48.8	1.07	0.80	1.39
Unknown	15	14.2	1.06	0.61	1.70

American Spinal Injury Association Impairment Scale Grade of A, B or C (AIS ABC).

There were some variables with relatively higher SMR and near significant 95% CI. For the white population, there were 105 CRC deaths with an SMR of 1.19 (95% CI [0.98, 1.43]). For the current age group of 30–59 years-old, there were 57 CRC deaths with an SMR of 1.19 (95% CI [0.91, 1.53]) and for the current age group of 60+ there were 88 deaths with an SMR of 1.07 (95% CI [0.86, 1.31). For years since injury, the 10–19 years post-injury group had 47 CRC deaths with an SMR of 1.31 (95% CI [0.97, 1.72]) and for the 20+ years since injury group there were 62 CRC deaths and an SMR of 1.15 (95% CI [0.89, 1.46]). For the neurologic group T1-S3 AIS ABC, there were 44 CRC deaths with an SMR of 1.21 ([95% CI [0.89–1.61])

There was one statistically significant subgroup found and this was for patients with injury level C1-4 with an American Spinal Injury Association Impairment Scale Grade of A, B or C (AIS ABC) with 18 deaths out of 7,201 persons and an SMR of 1.68 ([95% CI [1.03–2.61]).

Discussion

In the United States, colorectal cancer (CRC) is the third most commonly diagnosed cancer and the second leading cause of cancer related deaths.⁸ In addition, as CRC is the third most common cancer-related mortality in the SCI population, it is important to analyze the relationship between CRC and SCI.¹³ Advanced age is a risk factor for CRC and when looking at the SCI population: the average age at SCI increased to 43 in recent years and there was a recent increase in SCI incidence rates in the elderly population (≥65 years).^27–29 Furthermore, it has been reported that CRC lesions found in SCI patients are often far advanced (presenting with Stage III or IV tumors).³⁰ Our objective was to see if there were currently any disparities in CRC mortality between persons with SCI and the general population.

There was no significant difference in overall CRC mortality between the general population and the population with SCI, perhaps because CRC screening rates are similar between the two groups. The CDC reports approximately one-third of adults have not been screened as recommended, and one study by Stillman et al. found over one-third of participants with SCI 50 years old or greater had not had a screening colonoscopy.^12,15 However, larger studies/analyses of the screening rates in the SCI population should be completed. Even with the CRC screening rates being equivalent between the two groups, the rates are suboptimal. Ideally 100% of individuals would undergo appropriate CRC screening; however, the National Colorectal Cancer Roundtable had put out an “80% by 2018” initiative to increase CRC screening rates to 80% by the year 2018.³¹ This initiative hoped to increase screening rates as CRC can be detected at an early stage when treatment is more likely to be successful and screening can even prevent CRC by detecting and removing precancerous lesions.³¹ While it is unknown if the goal had been met, in 2016 no states had screening rates of ≥80%.¹² It is important to endorse screening in every population, but it is especially important to encourage appropriate screening in the SCI population, as the complexity of their care can potentially overshadow preventative care.

Furthermore, in addition to similar screening rates, the overall mortality rates may be comparable between the two groups as treatment options are presumably equivalent between both populations; although there is paucity of data. One 1996 study by Stratton et al. reported that postoperative chemotherapy and radiation therapy were well-tolerated by SCI patients and should not be withheld from appropriate patients because of their SCI.³⁰ The study does mention a 34% incidence of postoperative complications related to SCI (pulmonary, cutaneous, urinary tract morbidity), but encourages aggressive measures to be used for complication avoidance and does not discourage patients with SCI with CRC from undergoing operative treatment.³⁰ Depending on the patient with SCI, surgical resection with permanent end colostomy or ileostomy versus resection with primary anastomosis are all safe and effective treatments.³⁰ More recent studies are needed to reevaluate CRC treatments for patients with SCI, especially with recent advances in care.

There was one statistically significant subgroup found (patients with injury level C1-4 with an American Spinal Injury Association Impairment Scale Grade of A, B or C) and because of the large number of subgroups analyzed, one being statistically significant is expected by chance. However, it is of interest that this increase in mortality was discovered in the group with the highest level of injury. If this is a true increase in CRC mortality in the C1-4 AIS ABC group, speculation about why this disparity exists can be given. First, it is likely that those with higher injuries have greater barriers to care (transfer difficulty, accommodation) that could be even more prohibitive to CRC screening in this group. This could result in an increased mortality as CRC screening reduces both incidence and mortality from the disease.^32–34 Furthermore, risk factors for CRC include obesity and inactivity and this may be a factor in increased mortality for high levels of injury.^19,22 Persons with SCI have been found to have 40% of the activity level of able-bodied peers and are high risk for inactive lifestyle and furthermore, individuals with SCI are at risk of developing obesity because of their reliance on wheelchairs for mobility.^35,36 Physical inactivity may be even more prevalent in individuals with high-levels of injury as that leads to more deficits and greater immobility. Potential variables such as late diagnosis and poor bowel function/management also warrant evaluation. CRC mortality trends should continue to be evaluated in this specific group moving forward.

To the authors’ knowledge this is the first study looking at CRC mortality in persons with SCI versus the general population. Strengths of this study include a large sample size and long-term follow-up of participants.

This study has several limitations. First, cancer incidence, timing of diagnosis, response to treatment and cancer pathologies were variables that were not available in the database for further assessment. Next, the NSCID is not representative of all SCI in the United States, although representativeness of the overall study population is somewhat enhanced by the inclusion of the collaborative survival study and Shriners Hospital patients. More severely injured patients and racial minorities are slightly overrepresented. The racial disparity is accounted for by using race-specific rates in the calculation of expected numbers of deaths. However, if patients with C1-4 AIS ABC injuries are at higher CRC risk than other patients as this study suggests, the overall SMR will be slightly overestimated due to the overrepresentation of these patients in the study. However, risk factor-specific SMRs should not be affected.

Alternatively, all reported SMRs should be viewed as slightly conservative because 8% of all cancer deaths in the study population did not specify primary organ site. Apportioning these cancer deaths of unknown location based on the proportions of known cancer deaths due to CRC would increase the overall number of CRC deaths from 146 to 157 and would increase the estimated CRC SMR to 1.20 (95% CI [1.02, 1.40]). This would reflect a very small but marginally significant increased risk of CRC mortality after SCI. In addition, the SSDI and NDI are not 100% sensitive and specific. A small number of persons who died might be misclassified as alive, and a small percentage of those might have died as a result of CRC. Finally, cause of death is not always accurately reflected on death certificates.

Conclusion

Although persons with SCI receive suboptimal rates of preventative care screenings and report extensive barriers to care, they do not appear to be at meaningfully increased risk of CRC mortality. The current recommendations for CRC screening should be continued for these individuals while reducing barriers to care. Future research should evaluate CRC incidence, screening rates and risk factors in persons with SCI and varying neurological groups.

Disclaimer statements

Contributors Jenna Koblinski – Conception, literature review, writing the article, critical revision of the article. Dr. Michael DeVivo – Design, data collection, analysis and interpretation, writing the article, critical revision of the article. Dr. Yuying Chen – Design, data collection, writing the article, critical revision of the article. Dr. Valentine Nfonsam – Conception, writing the article, critical revision of the article.

Funding This work was supported by National Institute on Disability, Independent Living, and Rehabilitation Research: [Grant Number 90DP0083]; Dr. Michael DeVivo and Dr. Yuying Chen provided the data from the National Spinal Cord Injury Database, which was provided support through grant funding (National Institute on Disability, Independent Living, and Rehabilitation Research: [Grant Number 90DP0083]).

Conflicts of interest Dr. Valentine Nfonsam and Miss Jenna Koblinski have no conflicts of interest to declare. Dr. Michael DeVivo and Dr. Yuying Chen provided the data through the National Spinal Cord Injury Database. This database was provided support through grant funding (National Institute on Disability, Independent Living and Rehabilitation Research [grant number: 90DP0083]). Dr. Michael DeVivo gives paid expert witness testimony on life expectancy after spinal cord injury.