Metastatic spinal cord compression (MSCC) is one of the most dreaded complications of cancer. A Nationwide Inpatient Sample (NIS) from 2000 to 2011 was used to extract data for all in-hospital stays of patients with MSCC using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes. The usage and timing of radiation therapy and surgical interventions were identified using ICD-9-M codes. These interventions were defined as ‘early intervention’ if they were provided within the first 48 hours of hospitalization. Multivariable logistic regression was used to examine the factors associated with delays in ‘early intervention’. We also examined whether delays in treatment led to worse outcomes in terms of mortality and morbidity. 13 457 patients were admitted with MSCC from 2000 to 2011 who received one or more modalities of treatment. Of these, 5035 (37%) received early intervention. Female gender, private-for-profit hospitals and higher comorbidity index were associated with lower rate of early intervention. In-hospital mortality was lower in the early intervention group (5.0% vs 6.9%, p=0.04). Patients receiving early intervention were discharged home more often (44.2% vs 36.4%, p<0.001) along with lower need of home health services (14.6% vs 18.8%, p=0.004). The length of stay (LOS) was significantly shorter and hospital charges lower in those who received early intervention (LOS: median 6 vs 11 days, p<0.001; charges: $34 354 vs $50 062; p<0.001). The rate of care delivery by early intervention in MSCC is suboptimal. Early treatment results in lower mortality, shorter LOS and higher rates of discharges to home along with lower hospital charges and decreased usage of home healthcare.
Significance of this study
What is already known about this subject?
Malignant involvement of spinal cord is an emergency and needs urgent intervention.
Spinal cord involvement in patients with cancer may be misdiagnosed in the setting of chronic severe back pain.
Recovery from metastatic spinal cord involvement is dependent on the extent of loss of neurological function prior to intervention.
What are the new findings?
Early intervention, that is, within 48 hours of symptomatic presentation results in favorable outcomes.
Female gender, private-for-profit hospitals and higher comorbidity index were associated with lower rates of early intervention.
Patients who received early intervention have a shorter period of hospital stay, lesser cost of hospitalization and have better disposition outcomes.
How might these results change the focus of research or clinical practice?
The current rate of healthcare delivery by early intervention in patients with cancer with spinal cord involvement is suboptimal. Emphasis on intervention within the first 48 hours will result in lower mortality, shorter length of stay and higher rates of discharges to home.
Metastatic involvement of the spinal cord may occur from any primary site. Multiple myeloma, lung, breast and prostate cancer are the most common malignancies associated with metastatic spinal cord compression (MSCC) in adults while neuroblastomas and sarcomas and germ cell tumors are more common in the pediatric population.1–5 The incidence of MSCC has been estimated to vary between 3.4% and 6% based ion epidemiological studies on hospitalized patients and autopsy reports.2 ,4 Most cases of MSCC are diagnosed when further workup is being done to investigate the extent of newly discovered vertebral metastasis. However, almost 20% of MSCC cases may present as an initial manifestation of an underlying malignancy.6
The clinical presentation of MSCC is dependent on its location. A majority (60%) of MSCCs occur in the thoracic spine followed by lumbosacral and cervical spine with 30% and 10% frequencies, respectively.1 Arterial seeding is the most common mechanism for tumor dissemination to the thecal sac. Growth along the thecal sac leads to rapid increase in pressure on the spinal cord, due to constraints in volume within the spinal canal. Even small differences in growth contribute to rapid increase in pressure which is initially manifested as spinal cord edema. Any additional worsening of spinal cord compression will lead to infarction and subsequent neurological manifestations of motor weakness, sensory deficits and pain resulting in permanent neurological deficits. Urgent intervention is imperative to prevent irreversible neurological sequelae.
Many reasons have been invoked to explain the delay in treatment of patients with MSCC.
Most patients present with gradually worsening back pain, a symptom that often precedes other neurological manifestation by several weeks. The pain is non-specific initially and may mimic symptoms of spinal stenosis, degenerative joint disease, spinal stenosis and other less ominous disorders of the spine.7 Pain may present in a localized radicular or funicular pattern.2 ,7 ,8 Therefore, a high index of suspicion and radiological investigation of back pain in patients with cancer is extremely important to detect early stages of thecal involvement. Loss of motor function, onset of sensory deficits, loss of bowel and bladder control and ataxia occur as MSCC worsens.7 ,9 Initial workup involves MRI or CT myelography. Although both of them have approximately the same sensitivity and specificity, CT myelography may be preferred in patients who are not suitable for MRI (metallic implants, inability to stay in supine position for long time periods, lesions difficult to visualize on MRI, etc).10 Both of these imaging techniques also help in distinguishing MSCC from radiation myelopathy, an entity which may have a similar clinical presentation.11 ,12 Other entities which may mimic MSCC are infiltrative disease (sarcoidosis), gout, arthritis of the spine, extramedullary hematopoiesis, meningiomas, hemangiomas, epidural hematomas, etc.13–18
Patients often present with a history of back pain which starts ∼2 months earlier which in the setting of generalized weakness, presence of prior neurological deficits and concomitant severe cancer pain may confound the presentation of MSCC.19 Dexamethasone is often used in the acute setting to help alleviate vasogenic edema.20 ,21 Subsequent surgical decompression and radiotherapy (RT) are based on a careful assessment of the stability of the spine by spine surgeons and radiation oncologists. When surgery plus RT versus surgery alone for MSCC were compared, the rates of motor function improvement, post-treatment ambulatory rates, local control rates were similar between the treatments—ranging from 20% to 27%.22 A recent large meta-analysis showed better ambulatory outcomes (relative risk 1.43; 95% CI 1.14 to 1.78) when direct decompression surgery with radiation therapy was used to treat MSCC. However, this may reflect a surgical selection bias with more frail patients and less functional patients being treated with less-invasive treatments such as radiation therapy alone.23 Studies which have compared the role of surgical decompression followed by RT and RT alone have usually excluded patients with paraplegia for more than 48 hours and patients with primary spinal tumors.24
Few studies have compared the clinical outcomes in patients with MSCC who received early versus delayed intervention. Significant delay in the initiation of therapy (14 days) since the initial diagnosis was reported to cause adverse outcomes.19 The delay in intervention correlated to the gap in knowledge about adverse outcomes in patients with MSCC in whom intervention in the form of surgery or RT was provided after the first 48 hours of hospitalization. Our study compares outcomes of early versus delayed treatment and related healthcare costs. A nationally representative database developed by the Agency of Healthcare Research and Quality (AHRQ) was used to ensure sufficient sample size and enhance generalizability and validity of the study's results.
Patient population and study design
In this study, the Nationwide Inpatient Sample (NIS) Database was used to obtain demographic and outcomes data for patients aged ≥18 years with metastatic cancer for the years 2000–2011. NIS is the largest publicly available inpatient healthcare database administered by the Healthcare Cost and Utilization Project of the AHRQ of the US government. This database has been compiled from inpatient data obtained from 1000 hospitals sampled to approximate a 20% stratified sample of US community hospitals. National estimates are produced based on the provided hospital and discharge weights. This data set has no patient identifiers and has been used extensively in healthcare policy design.
The International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes were used to identify adult patients (age ≥18 years) with metastatic cancer (196–199) who were discharged with a primary diagnosis of MSCC (336.9). The usage and timing of interventions by RT or surgery was also identified using ICD-9 codes (appendix). Intervention was classified as ‘early’ when provided within the first 48 hours and ‘late’ when provided after 48 hours of hospitalization.
Individual comorbidities were identified using ICD-9-CM codes (appendix) and were used in the construction of the Charlson-Deyo's comorbidity index to account for the burden of comorbid disease. Rates of invasive mechanical ventilation were identified using the ICD-9-CM code: 96.70–96.72. The patient population was divided into those with hematological and non-hematological malignancies. Lung, breast and colon cancer were included in non-hematological malignancies. NIS classification and variables were used to identify patients' demographical characteristics and hospital characteristics such as ownership, size, teaching status, urban location and geographical region.
The primary outcome of interest was the all-cause in-hospital mortality in the early and late intervention categories. Other outcomes studied were discharge dispositions, length of hospital stay and cost of hospitalization. Since spinal cord compression is a condition which often requires extensive physical therapy, discharge disposition was specifically studied under the groups: home, home with healthcare, transfers to other healthcare facilities (under which NIS includes skilled nursing facilities, intermediate care, inpatient rehabilitation, psychiatric hospitals, inpatient hospice, and others) and other disposition (includes against medical advice, unknown, and missing).
Statistical analysis was performed with STATA IC V.11.0 (STATA-Corp, College Station, Texas, USA) using the strata weights and survey commands to generate national estimates of patients with metastatic cancer. Statistical comparisons were performed using Pearson's χ2 test and Wilcoxon rank test for discrete and continuous variables, respectively. In-hospital mortality was examined with a multivariate logistic regression model to study whether early intervention was independently associated with better outcomes. A univariate model was initially used to identify variables associated with risk of mortality. Variables that were found to be significant in the univariate analysis at p<0.10 were included in the final multivariate model. For the variables included in our final model, both the tolerance and the variation inflation factor were close to unity, indicating minimal collinearity. We adjusted the final model for age, gender, race, insurance, Charlson's comorbidity index, hospital characteristics, individual cancers, use of mechanical ventilation and year of admission. The techniques of data collection and analysis performed in this study have been discussed in previous publications.25
Between the years 2000 and 2011, 13 457 patients were discharged with a diagnosis of MSCC in whom at least one kind of intervention was provided. Of these, 5035 (37%) patients received early intervention (within 48 hours of hospitalization) and 8422 (63%) patients received late intervention (after 48 hours).
Demographical and clinical characteristics
Patients with MSCC who received early or late interventions were equally distributed among all age groups (table 1). A slight male preponderance was noted in patients who received early intervention. More white patients received early intervention compared with other races. While the patient payer mix was similar in both groups, these interventions are predominantly performed in large, urban, teaching, private non-profit hospitals. Notably, the mean Charlson's comorbidity index was lower in those who received early intervention. The proportion of lung, breast and colon cancer was marginally lower in the group that received early intervention (table 2).
Variables associated with late intervention in patients with metastatic cord compression
Female patients with MSCC had 0.77 times lower odds (95% CI 0.66 to 0.91) for receiving early intervention (table 3). Similarly private-for-profit hospitals had 0.53 times lower odds (95% CI 0.32 to 0.88). Charlson's comorbidity index was inversely related to odds of receiving early intervention (OR 0.92; 95% CI 0.89 to 0.95) (figure 1). Notably, through the years 2000–2011, the odds of receiving early intervention have not changed. Race, payer status, hospital teaching status and hospital size did not appear to be associated with provision of early intervention.
The in-hospital mortality in patients with metastatic disease with spinal cord compression was significantly higher in those who receive late intervention compared with early intervention (6.9% vs 5%, p<0.001; table 4). Analysis of yearly trends shows persistent differences in provision of early and late interventions for patients with MSCC (figure 2) but no significant change was noted from 2000 to 2011 (p=0.15). A higher proportion of MSCC patients who received early intervention were discharged to home with therapy while a higher proportion of patients with MSCC who received late intervention were discharged to specialized centers (nursing home, etc) including hospice care. The median length of stay (LOS) in the early intervention group was also significantly shorter compared with the late intervention group (6 vs 11 days, p<0.001). The hospital charges for the late intervention group were significantly higher than that of the early intervention group (table 4). A significant correlation between timing of intervention, Medicaid payers and discharge to home was observed (table 5).
The study presented here is the first to describe an analysis of clinical outcomes of spinal cord compression in patients with metastatic cancer over a 12-year period on nationwide scale. While the overall survival of the patient is dependent on stage and type of the underlying malignancy, involvement of the spinal cord has significant implications for the quality of life for the patient. RT is commonly used intervention in patients with a stable spine and is very effective in local pain control.26 However, the most important predictor for postintervention neurological function is the extent of pretreatment neurological function.2 ,26 This clinical correlate signifies the importance of preserving neurological function by rapid intervention. Tumors which are more radiosensitive lead to favorable outcomes and lower rates of relapse compared with radioresistant tumors.7 ,27 A scoring system has been developed to predict post-RT outcomes in patents with spinal cord compression secondary to malignancy. This scoring system is based on the type of primary tumor, interval time gap between cancer diagnosis to MSCC, presence of visceral metastases, extent of motor function before RT and the time involved in the development of motor deficits before RT.28
This study also shows that only 37% of the patients who receive any intervention receive it within the first 48 hours of hospitalization. The percentage of patients receiving early intervention has not changed significantly from 2000 to 2011 (p=0.15). Although this time frame represents an improvement compared with past studies which recorded a treatment delay of 8–16 days,29 ,30 the lack of any significant improvement in the proportion of patients who received early treatment over the studied 12-year period is concerning. Since, most of these patients are seen in large urban tertiary care hospitals, it is unlikely that the lack of availability of adequate infrastructure is the underlying reason. Improvement in healthcare delivery based on implementation of a pathway by multidisciplinary teams was demonstrated in a small study involving 17 patients.31 Recognizing the possibility that early recognition and referral is key to early intervention leading to favorable outcomes, the National Institute for Health and Care Excellence (NICE) issued guidance in terms for pathway and quality standards in 2008.32 These guidelines were adapted by AHRQ and reaffirmed by the National Collaborating Centre for Cancer in 2012. The extent to which medical practitioners follow these guidelines has not been studied.
Age and race were not found to be associated with early intervention in this study. These trends are surprising since older age has been associated with less aggressive and timely interventions in patients with cancer.33–35 Similarly, racial bias has been studied and demonstrated in healthcare delivery for minority patients with cancer and without cancer.36–38 This study demonstrated that the in-hospital mortality was lower in the early intervention group (5.0% vs 6.9%, p=0.04). Since patient populations with different types of malignancies (lung, breast, colon and hematological) are roughly equal within both the early and late intervention groups, it is unlikely that the differences in the nature of the malignancies accounted for the survival advantage. A plausible explanation is better cancer control due to timely intervention. Unfortunately, this explanation cannot be verified by variables available in the NIS database.
Referral to larger, tertiary care, urban hospitals may explain the higher representation of both interventions when compared with smaller to medium sized hospitals and rural hospitals. Female gender, private-for-profit hospitals and higher comorbidity index were associated with lower use of early intervention. Gender has been shown to affect both the time to diagnose and healthcare preferences in patients with cancer.39 ,40 On a similar note, patients with a higher comorbidity index may represent a patient population which is significantly moribund and therefore less amenable to surgical or RT-based interventions. Notably, almost twice the number of patients in the delayed intervention group was referred to hospice care. It should be noted that the use of Charlson's comorbidity index in patients with cancer has been controversial since progression of a chronic disease may confound the morbidity caused by the progression of malignancy.41
The differences in the discharge disposition of the early and delayed intervention groups are very different. Patients receiving early intervention were discharged home more often (44.2% vs 36.4%, p<0.001) along with lower need of home health services (14.6% vs 18.8%, p=0.004). This likely reflects lower neurological impairment in the early intervention group, although this clinical aspect could not be conclusively deduced from the available parameters of the NIS database. The LOS was significantly shorter and hospital charges lower in those who received early intervention (LOS: median 6 vs 11 days, p<0.001; charges: $34 354 vs $50 062; p<0.001). Since it is not possible to track the survival of these patients after hospital discharge, further cost-benefit analysis could not be performed.
There are very few studies in the available medical literature until date which have studied factors associated with early and delayed interventions in patients with MSCC. Despite these interesting results, there are several notable limitations to the current study. Owing to the administrative nature of the database, differences in coding practices may affect the quality of the obtained data. The current coding practice also places limits in determining the severity of the disease, accurate stage of the malignancy and also does not provide information about changes in neurological status during hospitalization. Multiple hospitalizations of the same patient may artificially increase the sample size of the database. This is of lesser concern as these patents have a limited life expectancy and are often referred to home hospice when significant worsening of the disease has occurred.
This study is the first nationwide analysis of the outcomes comparing early and delayed interventions in patients with MSCC. From a healthcare delivery point of view, it shows that the rate of early intervention in patients with MSCC is suboptimal. It shows that intervention when provided within the first 48 hours is associated with lower mortality, shorter LOS, lower hospitalization cost and a better discharge disposition.
This work was presented in the form of an abstract and a poster at the ASCO 2016 Annual Meeting in Chicago, USA
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement All data used in this study are based on a publicly available database of the US government which uses de-identified patient data.