Authors: Andrew J. Schoenfeld (1Department of Orthopaedic Surgery, Center for Surgery and Public Health, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts;), Lingwei Xiang (2Department of Surgery, Center for Surgery and Public Health, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts), Rachel R. Adler (2Department of Surgery, Center for Surgery and Public Health, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts), Alyssa L. Schoenfeld (3Medfield Public Schools, Medfield, Massachusetts), James D. Kang (1Department of Orthopaedic Surgery, Center for Surgery and Public Health, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts;), Joel S. Weissman (2Department of Surgery, Center for Surgery and Public Health, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts)
Categories: Article
Source: The Journal of bone and joint surgery. American volume
Authors: Andrew J. Schoenfeld, Lingwei Xiang, Rachel R. Adler, Alyssa L. Schoenfeld, James D. Kang, Joel S. Weissman
The incidence of odontoid fractures among the elderly population has been increasing in recent years. Elderly individuals with dementia may be at increased risk for inferior outcomes following such fractures. Although surgical intervention has been maintained to optimize survival and recovery, it is unclear if this benefit extends to patients with dementia. We hypothesized that patients with dementia who were treated operatively for odontoid fractures would experience improved survival and lower rates of hospice admission but higher rates of delirium and of intensive interventions.
We used Medicare claims data (2017 to 2018) to identify community-dwelling individuals with dementia who sustained type-II odontoid fractures. We considered treatment strategy (operative or nonoperative) as the primary predictor and survival as the primary outcome. The secondary outcomes consisted of post-treatment delirium, hospice admission, post-treatment intensive intervention, and post-discharge admission to a nursing home or a skilled nursing facility. In all models, we controlled for age, biological sex, race, Elixhauser Comorbidity Index, Frailty Index, admission source, treating hospital, and dual eligibility. Adjusted analyses for survival were conducted using Cox proportional hazards regression. Adjusted analyses for secondary outcomes were performed using generalized estimating equations. To address confounding by indication, we performed confirmatory analyses using inverse probability of treatment weighting.
In this study, we included 1,030 patients. The median age of the cohort was 86.5 years (interquartile range, 80.9 to 90.8 years), 60.7% of the patients were female, and 90% of the patients were White. A surgical procedure was performed in 19.8% of the cohort. Following an adjusted analysis, patients treated surgically had a 28% lower hazard of mortality (hazard ratio, 0.72 [95% confidence interval (CI), 0.53 to 0.98]), but higher odds of delirium (odds ratio, 1.64 [95% CI, 1.10 to 2.44]). These findings were preserved in the inverse probability weighted analysis.
We found that, among individuals with dementia who sustain a type-II odontoid fracture, surgical intervention may confer a survival benefit. A surgical procedure may be an appropriate treatment strategy for individuals with dementia whose life-care goals include life prolongation and maximizing quality of life in the short term following an injury.
Therapeutic Level III. See Instructions for Authors for a complete description of levels of evidence.
Along with hip fractures, fractures of the odontoid process are increasingly recognized as fragility fractures that can alter the life arc of elderly individuals, accelerating functional deterioration, loss of independence, and mortality^1–7^. The incidence of these injuries among individuals ≥65 years of age has been increasing recently^2,5^, as has the enthusiasm for surgical intervention^1,7,8^. For example, Alluri et al. reported that the rate of surgical procedures for elderly patients with odontoid fractures nearly doubled from 2003 to 2017^1^. Some studies have documented that elderly patients treated operatively have superior outcomes in terms of quality of life and physical function^8^, whereas others have maintained that there is no difference in functional outcomes^3,4,9,10^. Several studies have substantiated improved survival for patients treated with operative intervention, although this may have been due to selection bias^4,8^, and such findings are generally confined to individuals <80 years of age.
At the same time, surgical intervention has a non-negligible complication profile and may accelerate neurologic dysfunction or mortality in the event of surgical site infection, sepsis, aspiration, or other perioperative adverse events^2,11–13^. This risk may be even higher among elderly individuals with dementia, who represent both a growing subgroup in most industrialized nations and a population at an elevated risk for falls and other injuries^14,15^ that may lead to odontoid fractures. Elevated risks in this population may be linked to the extent of cognitive impairment, comorbidities, and increased physiologic frailty^14,15^. Prior studies in other areas of musculoskeletal injury have found that patients with dementia were more likely to experience postoperative delirium and other complications, greater loss of mobility, and higher rates of mortality following a surgical procedure than counterparts without cognitive impairment^15^. For many individuals with dementia, operative intervention may also not align with predefined goals of care or may lead to events such as prolonged intubation, tracheotomy, and insertion of feeding tubes that contravene advanced directives^14^.
At present, there has been limited information to support robust shared decision-making with regard to clinical results and the care trajectory of patients with dementia who sustain odontoid fractures. In this context, we sought to use recent Medicare claims data to characterize the influence of operative or nonoperative management on clinical and patient-centered outcomes including survival, the development of post-treatment delirium, admission to hospice, and post-treatment intensive interventions, such as cardiopulmonary resuscitation, mechanical ventilation, or feeding tube insertion. Based on prior work in the area of hip fractures^14,15^, we hypothesized that patients with dementia who were treated operatively for odontoid fractures would experience improved survival and lower rates of hospice admission but higher rates of delirium and of intensive interventions.
This study received institutional review board approval prior to commencement. We used national Medicare data files for individuals undergoing treatment from January 1, 2017, to June 30, 2018, including the Master Beneficiary Summary File (MBSF), Carrier File, Inpatient and Outpatient claims data, Skilled Nursing Facility and Home Health claims, Minimum Data Set (MDS), and Hospice files.
The sample consisted of community-dwelling patients with dementia, identified using a previously validated approach that considered beneficiaries without an MDS assessment in the 180 days before an initial incident code for odontoid fracture. Dementia was identified using a previously validated approach that surveyed for associated International Classification of Diseases, Tenth Revision (ICD-10) codes in Inpatient, Outpatient, Carrier, Hospice, Home Health, and Skilled Nursing Facility files^14^. We only evaluated beneficiaries who had 19 months of continuous fee-for-service coverage, allowing for a gap of up to 1 month in the year prior. Patients were identified as having sustained an odontoid fracture based on ICD-10 inpatient diagnosis codes (S12110A, S12112A, S12111A, S12120A, S12112B). We limited consideration to patients with type-II odontoid fractures^16^ to reduce heterogeneity and increase the prospect of surgical intervention. Surgical intervention was identified on the basis of the ICD-10 procedure codes for a relevant cervical spine surgical procedure that occurred in association with the index admission and was associated with 1 of the indicated diagnosis codes.
We used the MBSF and Inpatient files to assess survival at 30, 90, and 180 days following the initial diagnosis of odontoid fracture. The occurrence of in-hospital delirium following the initial fracture diagnosis was determined on the basis of the presence of an established ICD-10 code for this condition^17^. Hospice admission was determined via the presence of a new hospice claim within 180 days of discharge after the index admission^14^. New nursing home admissions within 180 days after discharge were determined via previously validated approaches for claims-based data^14,18,19^. Intensive interventions were identified using surveys for appropriate codes in Inpatient, Outpatient, Carrier, Hospice, Home Health, and Skilled Nursing Facilities files for cardiopulmonary resuscitation, intubation, mechanical ventilation, feeding tube insertion, or new dialysis at any time point within 180 days following the odontoid fracture diagnosis^20^.
We considered the treatment strategy (operative or nonoperative) to be the primary predictor and survival to be the primary outcome. Other patient-centered and clinical outcomes were considered secondarily and consisted of post-treatment delirium, hospice admission, post-treatment intensive intervention, and post-treatment nursing home admission. In all models, we controlled for patient characteristics that may have influenced the decision for operative or nonoperative management, including age, biological sex, race (categorized as White compared with non-White due to sample-size limitations among racial minority groups), Elixhauser Comorbidity Index (see Appendix 1)^21,22^, Frailty Index, admission source (categorized as home or not home [for example, skilled nursing facility, hospice, or other inpatient facility]), and dual Medicare and Medicaid eligibility as a marker for socioeconomic status. A Cox proportional hazard model and generalized estimating equations (GEEs) were used to account for hospital clustering, as certain facilities may be more likely to surgically treat patients with spine conditions than others^23^.
We presented continuous data using medians and interquartile ranges (IQRs), with categorical data described using frequencies and percentages. Wilcoxon 2-sample tests and Pearson chi-square tests were conducted to compare the patient characteristics between operative and nonoperative treatment groups. Raw, unadjusted comparisons were made across secondary outcomes using GEEs. Survival was evaluated using Kaplan-Meier curves, which were also used to assess the proportionality assumption. We also evaluated the rate of survival at 30, 90, and 180 days following the fracture diagnosis. Adjusted analyses for survival were conducted using Cox proportional hazards regression, with output presented using hazard ratios (HRs), 95% confidence intervals (CIs), and p values. All abstracted variables were included as covariates in adjusted models. Adjusted analyses for all secondary outcomes were performed using GEEs accounting for hospital clustering, with output presented using odds ratios (ORs), 95% CIs, and p values. In order to address confounding by indication and selection bias, we performed confirmatory testing using inverse probability of treatment weighting via propensity scores. We further assessed the impact of the surgical procedure by time frame, creating a cut-point at 45 days to assess whether the benefit of the surgical procedure was realized only in the first 6 weeks following the procedure or over a longer time period. Significance was established, a priori, for all variables as HR and OR point estimates and 95% CIs exclusive of 1.0 and p values of <0.05. All data were analyzed using SAS version 9.4 (SAS Institute). All coding algorithms used in this study are available from the authors by request.
We identified 1,030 community-dwelling patients with dementia who sustained odontoid fractures and met full inclusion criteria. The median age of the cohort was 86.5 years (IQR, 80.9 to 90.8 years), and 60.7% of the population was female. Ninety percent of the population was White, and the median Elixhauser Comorbidity Index was 10.0. Seventy-three percent of the cohort was admitted from home, and 17.8% were dual-eligible Medicare recipients. Among all patients with an odontoid fracture, 19.8% (204) were treated operatively and 80.2% (826) underwent nonoperative management. Because of the size of the sample, there were several modest but significant differences in demographic and clinical characteristics between the operatively and nonoperatively treated cohorts that were not considered to be clinically meaningful (Table I).
Patients with dementia who underwent surgical intervention were significantly less likely to die within 30 days (10.8% in the operatively treated group compared with 23.6% in the nonoperatively treated group; p < 0.001), 90 days (18.6% in the operatively treated group compared with 29.9% in the nonoperatively treated group; p < 0.002), and 180 days (24.0% in the operatively treated group compared with 39.1% in the nonoperatively treated group; p < 0.001) following the diagnosis of an odontoid fracture (Fig. 1). In the adjusted analysis, patients treated surgically were found to have a 28% lower hazard of mortality (HR, 0.72 [95% CI, 0.53 to 0.98]; p = 0.04) (Table II).
However, individuals who underwent surgical intervention had a higher likelihood of experiencing delirium (25.5% in the operatively treated group compared with 19.4% in the nonoperatively treated group; p = 0.05). In the adjusted analysis, the surgical cohort demonstrated a 64% increase in the odds of delirium (OR, 1.64 [95% CI, 1.10 to 2.44]; p = 0.02) (Table III).
In the adjusted analysis, the odds of hospice admission were not significantly different in the surgical group compared with those managed nonoperatively (OR, 0.67 [95% CI, 0.42 to 1.05]; p = 0.08) (Table IV). No significant differences were noted between the operative and nonoperative cohorts in terms of nursing home admissions (OR, 0.89 [95% CI, 0.59 to 1.37]; p = 0.62) or intensive interventions (OR, 1.07 [95% CI, 0.69 to 1.65]; p = 0.77).
These results were largely preserved following inverse probability weighting, including a lower hazard of mortality (HR, 0.66 [95% CI, 0.48 to 0.91]; p = 0.01) and higher odds of delirium (OR, 1.48 [95% CI, 1.01 to 2.18]; p = 0.046) for patients treated surgically. Following inverse probability weighting, patients treated surgically also had lower odds of hospice admission (OR, 0.54 [95% CI, 0.33 to 0.87]; p = 0.01). When considered by time period, a significant survival benefit for undergoing a surgical procedure was seen within the first 45 days (HR, 0.65 [95% CI, 0.43 to 0.97]; p = 0.003), but not for the time frame beyond 45 days (HR, 0.86 [95% CI, 0.52 to 1.44]; p = 0.57).
Akin to hip fractures, odontoid fractures have been shown to dramatically alter the clinical trajectory of an elderly individual’s life, often portending rapid functional declination, loss of independence, and near-term mortality^1–7^. The risk of such adverse events is likely even higher in elderly individuals with dementia who, because of cognitive impairment and physiological frailty, are not only at higher risk for developing these cervical fractures but also have a higher likelihood of suboptimal outcomes regardless of treatment. This presents a daunting clinical challenge, as surgeons must balance the goals of care and the risk of surgical procedures against the prospects of accelerated functional deterioration and lower likelihood of survival in both the near term and the long term. If a surgical procedure can meaningfully improve survival and functional independence, then this would be of value to patients. However, if the utilization of a surgical procedure only accelerates the spiral of clinical deterioration, while simultaneously driving the use of intensive interventions (for example, tracheotomy, a feeding tube, and mechanical ventilation) and institutionalization, this would be a suboptimal treatment method.
Our analysis suggests that, among community-dwelling elderly patients with dementia, surgical intervention not only improves survival but also reduces the potential for admission to hospice without simultaneously increasing the potential for other intensive interventions or nursing home admission. In terms of mean age and clinical and demographic characteristics, our cohort is similar to those in other investigations of fracture management in patients with dementia^14^. The rate of surgical intervention and the incidence of adverse events, such as mortality^6,11–13^ and post-treatment delirium^24^, in our cohort are aligned with estimates from prior series, which we believe endorses the external validity of our findings. This work is also strengthened by the use of Medicare data, which allowed for the identification of a large sample of patients treated in a variety of different health-care settings across the United States^14,23^. The use of Medicare claims also allowed us to follow patient trajectories longitudinally and characterize the post-treatment course, including survival, the environment of post-treatment institutional care, and the need for additional interventions. We believe that these characteristics support the generalizability of our findings and the prospects for ready clinical application.
The survival benefits for elderly patients with dementia who underwent operative intervention for type-II odontoid fractures are similar to those in previous work that has been performed among elderly individuals as a whole^4,6,8,25–28^. For example, Vaccaro et al. reported that, at 1 year following treatment, mortality was higher in elderly individuals treated nonoperatively for odontoid fractures and Neck Disability Index and Short Form (SF)-36v2 Bodily Pain scores were inferior among survivors^8^. Similarly, Robinson et al. encountered superior survival among individuals with odontoid fractures who were 70 to 88 years of age and underwent surgical management in a multinational registry^4^. Comparable results were also found in the report of the AOSpine group in 2013^28^. Barlow et al. previously reported that operative treatment for odontoid fractures was cost-effective in elderly patients <85 years of age^27^. Our results are also concordant with hip fracture treatment among patients with dementia, where it has been shown that surgical intervention was associated with a lower risk of mortality, as well as improved quality of life^14^. Surgical intervention, when properly indicated, may lead to the obviation of cervical immobilization that can precipitate aspiration, functional restrictions, and increased risk of falls^6,12,13^. A surgical procedure may also reduce pain and contribute to a lower need for pain medications that may further impact sensory and cognitive function^14,15^. Based on our subset analysis, these benefits appear to be most robust in the first few months following the surgical procedure. Although our results do indicate a short-term increase in the likelihood of delirium that may be tied to the use of anesthesia^24^, we believe that the risk of this adverse event is modest and it did not seem to exert an adverse influence on other post-treatment outcomes.
We believe that our findings are important to orthopaedic fracture services, spine surgeons, hospitalist geriatric liaison teams, policymakers, health-care institutions, patients, and families. Our work corroborates and expands on prior literature that highlights the notable risk of mortality and functional deterioration that follow in the wake of an odontoid process fracture among individuals with dementia. At the same time, surgical intervention does appear to provide a survival advantage and reduces the potential for hospice admission and confinement to environments of care that are outside the patient’s home. Therefore, surgical fixation of odontoid process fractures would appear to be an appropriate strategy for individuals with dementia whose life-care goals include life prolongation, maximizing time in the home of residence, and quality of life.
The results of our work should be interpreted in light of this study’s limitations. Foremost, this was a retrospective study that used Medicare claims data. The nature of the substrate data meant that we were dependent on coding algorithms to identify patients who sustained odontoid fractures, as well as patients who had dementia. Although we used previously accepted coding algorithms to establish these cohorts, we could not rule out the potential for coding errors to introduce confounding. Further, we did not have patient-reported outcome measures, or objective scores, to quantify physical function or quality of life. We were solely reliant on proxy measures such as the environment of care, care intensity, and number of post-treatment health-care encounters. Finally, we lacked clinically granular details to help to delineate treatment approaches and management strategies, including time from admission to the surgical procedure. We also lacked details with regard to fracture stability and displacement. Surgeon discretion and intuition remain important factors in deciding how patients will be managed following an odontoid fracture. Although our statistical approaches included many characteristics that would likely impact such decisions for a surgical procedure and our use of inverse probability weighting via propensity scores addresses selection bias to the extent possible, we do recognize the prospect for residual confounding to impact our results. We therefore caution that the results presented here should not be used as definitive recommendations with regard to surgical procedures in this population.
In conclusion, our results convey that, among individuals with dementia who sustain a type-II odontoid process fracture, surgical intervention may confer a survival benefit and optimize the chance for functional recovery and return to preinjury environments of care, including the patient’s home. At the same time, the injury itself exerts a notable effect on the prospect of survival, with non-negligible levels of mortality regardless of treatment. Although patients treated surgically had a modestly increased likelihood of delirium following treatment, they also had a lower rate of hospice admission without a concomitant increase in intensive interventions. These findings can prove useful when discussing care goals and the optimal treatment strategy with patients and their families, and they may also aid in shared decision-making. The ultimate treatment method should be tempered to the clinical context and informed by patient and family care goals. That said, in those individuals with life-care goals that include life prolongation and maximizing quality of life, our results would indicate that a surgical procedure should not be denied solely because of the presence of dementia.