Abstract
Background
The published reports examining socioeconomic factors and their relationship to osteosarcoma presentation and treatment suggest an association between lower socioeconomic status and a worse response to chemotherapy and lower survivorship. However, the driving factors behind these disparities are unclear. The Child Opportunity Index was developed by diversitydatakids.org (https://www.diversitydatakids.org/) in 2014 to cumulatively quantify social determinants of health in an index specifically tailored toward a pediatric population and organized by census tract. The Childhood Opportunity Index can be used to explore the relationship between a patient’s socioeconomic background and disparities in osteosarcoma presentation, treatment, and outcomes.
Question/purposes
Are differences in a child’s Childhood Opportunity Index score associated with differences in (1) time from symptom onset to first office visit for osteosarcoma, (2) timing of chemotherapy or timing and type of surgical resection, or (3) initial disease severity, development of metastatic disease, or overall survival?
Methods
A retrospective therapeutic study was conducted using data drawn from the institutional records of a large pediatric tertiary cancer center located in the Mid-Atlantic region of the United States from the years 2006 to 2022. Our main site is in an urban setting, with ample access to public transit. Patients were excluded from analysis if they were seeking a second opinion or our institution was not the main point of orthopaedic care (20% [45 of 223]). Of the remaining patients, those with incomplete electronic medical records (24% [43 of 178]), resided in an international country (5% [9 of 178]), presented after relapse (4% [8 of 178]), or lacked 2 years of follow-up at our institution (3% [5 of 178]) were excluded as well. A total of 113 pediatric patients (children younger than 18 years) met the inclusion criteria. The Child Opportunity Index is a composite index derived from three domains (education, health and environment, and social and economic) and 29 indicators within the domains that serve to capture the cumulative effect of disparities on child well-being. National Childhood Opportunity scores were collected and scored from 1 to 100. Each score represents an equal proportion of the US population of children 18 years of age or younger. A higher number indicates higher levels of socioeconomic opportunity. The overall Childhood Opportunity Index score was then broken down into three groups representative of the child’s relative socioeconomic opportunity: lowest tertile for scores < 34, middle tertile for scores between 34 and 66, and highest tertile for scores > 66. Means, ranges, medians, IQRs, and percentages were used to describe the study sample. Data analysis was conducted across the three groups (lowest tertile, middle, and highest), assessing differences in time to presentation, treatment variations, disease severity, and overall survivorship. Chi-square and Fisher exact tests were applied to compare categorical variables. Mann-Whitney U tests compared continuous data. Kaplan-Meier survival analysis, stratified by Childhood Opportunity Index tertile, was performed for a 5-year period to evaluate the development of metastatic disease and overall survivorship. A log-rank test was applied to evaluate statistical significance. Due to the small sample size, we were unable to control potential confounders such as race and insurance. However, the three domains (education, health and environment, and social and economic) encapsulated by the Childhood Opportunity Index data indirectly account for disparities related to race and insurance status.
Results
There was no association between lower levels of socioeconomic opportunity, as expressed by the lack of difference between the Childhood Opportunity Index tertiles for the interval between symptom onset and first office visit (mean ± SD lowest tertile 77 ± 67 days [95% confidence interval (CI) 60 to 94], middle tertile 69 ± 94 days [95% CI 50 to 89], and highest tertile 56 ± 58 days [95% CI 41 to 71]; p = 0.3). Similarly, we found no association between lower levels of socioeconomic opportunity, as expressed by the lack of difference between the Childhood Opportunity Index tertiles and the time elapsed from the first office visit to the first chemotherapy session (lowest tertile 19 ± 12 days [95% CI 12 to 26], middle 19 ± 14 days [95% CI 11 to 26], and highest 15 ± 9.7 days [95% CI 8.4 to 21]; p = 0.31), the time to surgical resection (lowest tertile 99 ± 35 days [95% CI 87 to 111], middle 88 ± 28 days [95% CI 77 to 99], and highest 102 ± 64 days [95% CI 86 to 118]; p = 0.24), or the type of surgical resection (limb-sparing versus amputation: 84% [21 of 25] in lowest tertile, 83% [24 of 29] in the middle tertile, and 81% [48 of 59] in the highest tertile received limb-sparing surgery; p = 0.52). Finally, we found no differences in terms of disease-free survival at 5 years (lowest tertile 27% [95% CI 7.8% to 43%], middle 44% [95% CI 23% to 59%], and highest 56% [95% CI 40% to 67%]; p = 0.22), overall survival (lowest 74% [95% CI 58% to 95%], middle 82% [95% CI 68% to 98%], and highest 64% [95% CI 52% to 78%]; p = 0.27), or in terms of survivorship of the cohort, excluding patients who presented with metastatic disease (lowest 84% [95% CI 68% to 100%], middle 91% [95% CI 80% to 100%], and highest 68% [95% CI 55% to 83%]; p = 0.10).
Conclusion
In our single-center retrospective study of 113 children who presented with osteosarcoma, we did not find an association between a patient’s national socioeconomic opportunity and their time to presentation, chemotherapy treatment, time to and type of surgical resection, or disease-free and overall survival. Prior work has shown an association between socioeconomic background and disparities in osteosarcoma treatment. It is possible that these findings will be similar to those from other hospitals and geographic areas, but based on our findings, we believe that proximity to providers, access to public transit, and regional insurance policies may help diminish these disparities. Future multicenter studies are needed to further explore the role that regional variations and the aforementioned factors may play in osteosarcoma treatment to help inform the direction of public policy.
Level of Evidence
Level III, therapeutic study.
Introduction
Osteosarcoma is the most common pediatric primary malignant bone tumor, with an estimated 5-year survival between 60% to 70% [8, 9]. Prior studies examining the relationship between delays to presentation in osteosarcoma treatment are associated with larger primary tumor size and an increased rate of metastasis, but not necessarily overall survival [10, 11, 14, 25-27, 33, 34]. It has been theorized that the patients with prolonged symptom duration include both patients with neglected and misdiagnosed disease, as well as those whose osteosarcoma has a less-aggressive biology [3]. Research into disparities in treatment based on socioeconomic background may help distinguish between delays in care and biological factors, as lower socioeconomic opportunity is often associated with treatment delays, disparities in care, and differences in outcomes [4, 5, 12, 15, 16, 18, 23, 28]. Two studies using national databases found associations between patients from a more disadvantaged socioeconomic background and worse overall survival [19, 20], although Richardson et al. [25] noted an association between worse histologic response to chemotherapy but not overall survival. It is important to better understand the impact of healthcare system shortcomings that may disproportionately impact patients with less socioeconomic opportunity.
The Child Opportunity Index was developed by diversitydatakids.org (https://www.diversitydatakids.org/) in 2014 to cumulatively quantify social determinants of health in an index specifically tailored toward a pediatric population and organized by census tract [22]. The Child Opportunity Index is a composite index derived from three domains (education, health and environment, and social and economic) and 29 indicators within the domains that serve to capture the cumulative effect of disparities on child well-being. These indicators were then standardized and applied to > 27,000 census tracts. Census tracts were distributed into quintiles containing exactly 20% of the US child population. These designations were generated using the US Census.gov geocoder [31]. GeoIDs were then matched with the Child Opportunity Index data provided by diversitydatakids.org. This emphasis on child-relevant neighborhood features affords an unprecedented level of specificity and granularity to the study of malignant bone tumor severity at presentation and social determinants of health.
Multiple studies have found that worse social determinants of health are associated with disparities in cancer treatment, including delays and access to care, complications, and higher mortality [4, 5, 12, 15, 16, 18, 23, 28]. However, it is currently unclear why there is a negative association between socioeconomic background and treatment outcomes in patients with osteosarcoma [19, 20]. If varying levels of socioeconomic opportunity are associated with disparities in pediatric osteosarcoma treatment, it may be of use to offer appropriate support and educational opportunities to these families to help mitigate these disparities.
We therefore asked: Are the differences in a child’s Childhood Opportunity Index score associated with differences in (1) time from symptom onset to first office visit for osteosarcoma, (2) timing of chemotherapy or timing and type of surgical resection, or (3) initial disease severity, development of metastatic disease, or overall survival?
Materials and Methods
Study Design and Setting
This is a retrospective therapeutic study drawn from the institutional records of a large pediatric tertiary cancer center located in the Mid-Atlantic region of the United States from the years 2006 to 2022. Our main site is in an urban setting with ample access to public transit. Multiple bus lines, as well as a train station, are within a block of this site. Additionally, starting in 2017, patients had the option to be seen at a satellite site, 30 minutes outside the city. Our cohort’s demographic characteristics are reflective of the population in the tristate area that our institution predominately cares for (Pennsylvania, New Jersey, and Delaware). All surgeries were conducted by the senior author (AA), a fellowship-trained orthopaedic oncology surgeon, or his predecessors who specialize in orthopaedic oncology. Furthermore, our hospital employs a dedicated oncologic social work service. This team offers psychological assessments and supportive counseling and helps families obtain state-funded medical assistance, federal supplemental income, transportation assistance, and more.
Child Opportunity Index
The Childhood Opportunity Index collected data on neighborhood factors that shape children’s access to components of healthy development: education, health and environment, and social and economic opportunity. The Childhood Opportunity Index’s multivariable database allows for granular quantitative comparison of socioeconomic background across the cohort to better understand how a child’s circumstances may influence his or her disease course. A detailed breakdown of the variables collected has been provided (Supplemental Tables 1-3; http://links.lww.com/CORR/B352). Using the Childhood Opportunity Index’s publicly available database, the children’s addresses were converted into the GeoID of their corresponding census tract as per the US Census [31]. The GeoID was then converted to the patient’s overall Child Opportunity Scores, which were scored from 1 to 100. Each score represents an equal proportion of the United States population of children. A higher number indicates higher levels of socioeconomic opportunity. The overall Childhood Opportunity Index score was then broken down into three groups representative of the child’s relative socioeconomic opportunity on a national level: lowest tertile for scores < 34, middle tertile for scores between 34 and 66, and highest tertile for scores > 66.
Patients
We considered potentially eligible patients with osteosarcoma identified using a combination of case log review and filtering by ICD-9 (170.x) and ICD-10 codes (C40.xx and C41.xx). Inclusion criteria consisted of patients with complete, accessible, and sufficient follow-up at our single center as documented in the electronic medical record (EMR). We excluded patients who presented with recurrent osteosarcoma, those with incomplete EMR data, and patients seeking a second opinion. Between January 2006 and October 2022, a total of 223 children 18 years of age or younger presented to our institution for osteosarcoma. Among these patients, approximately 20% (45 of 223) were seeking a second opinion or our institution was not the main point of orthopaedic care. Of the remaining patients, 24% (43 of 178) were excluded due to missing presentation or incomplete EMR data and 5% (9 of 178) were children predominantly residing in an international country. Four percent (8 of 178) presented after relapsed disease. Finally, we excluded 3% (5 of 178) of patients who lacked 2 years of follow-up at our institution to ensure comprehensive data on their care. We conducted analysis on the remaining 113 patients. Of these 113 patients, five lacked documentation on the onset of symptoms related to their osteosarcoma in our retrospective review. Two children with parosteal osteosarcoma were not treated with chemotherapy. One child did not undergo a tumor resection, and the family elected for palliative care due to the severe disease progression during their initial chemotherapy course.
Variables
Data collected included patient age, gender, race based on the self-reported data available in each patients’ EMR, insurance status (public, private, or other), date of patients’ first reported symptom, date of initial orthopaedic visit, location of primary tumor, date of chemotherapy treatment initiation, surgical treatment date, type of surgical resection, metastatic disease (at presentation, during, or after treatment), overall survival, and Childhood Opportunity Index score. The “other” insurance status represented patients with coverage from both private and public plans. The number of days elapsed between onset of initial symptoms to first orthopaedic evaluation, first orthopaedic evaluation to chemotherapy initiation, and surgical resection were calculated. Surgical resection was categorized as an amputation, limb-sparing surgery, rotationplasty, and other. Limb-sparing surgery included the wide intercalary resection with allograft reconstruction (with or without vascularized fibula) or wide resection with endoprosthetic reconstruction.
Descriptive Data
Fifty-one percent (58 of 113) of patients were boys, and the mean ± age was 13.0 ± 4.0 years (range 3.6 to 18.8 years [95% confidence interval (CI) 9.1 to 16.9]) at the time of presentation (Table 1). Most patients (59% [67 of 113]) were White, while 18% of patients were Black (20 of 113) and 5% were Asian (6 of 113). Sixteen percent (18 of 113) of patients did not fit into one of these categories, and 2% (2 of 113) refused to answer. Private insurance was the most common coverage for patients (56% [63 of 113]), followed by public insurance (34% [38 of 113]). The most common primary site of osteosarcoma was the femur (53% [60 of 113]), followed by the tibia (23% [26 of 113]) and the humerus (17% [19 of 113]) (Table 2). Other less common primary sites of osteosarcoma included the fibula (3% [3 of 113]), rib (2% [2 of 113]), ulna/radius (2% [2 of 113]), and pelvis (1% [1 of 113]). Forty-seven percent (53 of 113) of patients presented with metastatic disease or developed metastases during or after their chemotherapy treatment. Of the patients with metastatic disease, 19% (22 of 113) presented with metastases, 25% (28 of 113) developed metastatic disease after the completion of their initial chemotherapy, and 3% (3 of 113) developed metastasis during their initial chemotherapy course.
Table 1.
Demographic and follow-up data for entire cohort and cohort stratified by Childhood Opportunity Index tertiles
| Characteristic | Overall (n = 113) | Lowest (n = 25) | Middle (n = 29) | Highest (n = 59) | p valuea |
| Age at initial orthopaedic evaluation | 14.3 (10.0-16.0) | 13.5 (10.0-14.7) | 13.4 (9.0-16.3) | 14.4 (12.6-16.2) | 0.41 |
| Boys | 51 (58) | 56 (14) | 28 (8) | 61 (36) | 0.011 |
| Raceb | < 0.001 | ||||
| White | 59 (67) | 24 (6) | 62 (18) | 73 (43) | |
| Black | 18 (20) | 48 (12) | 21 (6) | 3 (2) | |
| Asian | 5 (6) | 0 (0) | 3 (1) | 8 (5) | |
| Other/refused | 18 (20) | 28 (7) | 14 (4) | 15 (9) | |
| Insurance | < 0.001 | ||||
| Private | 56 (63) | 24 (6) | 55 (16) | 69 (41) | |
| Public | 34 (38) | 68 (17) | 34 (10) | 19 (11) | |
| Otherc | 9 (10) | 8 (2) | 10 (3) | 8 (5) | |
| Uninsured | 1.8 (2) | 0 (0) | 0 (0) | 3 (2) |
Data presented as median (IQR) or % (n).
Kruskal-Wallis rank sum test, Pearson chi-square test, and Fisher exact test.
Race was identified via self-reported data in the patient EMR.
Other insurance indicates a combination of both public and private insurance plans.
Table 2.
Diagnosis, presentation, treatment, and outcome data for the entire cohort and stratified by Childhood Opportunity Index tertiles
| Characteristic | Overall (n = 113) | Lowest (n = 25) | Middle (n = 29) | Highest (n = 59) | p valuea |
| Primary site of osteosarcoma | 0.71 | ||||
| Femur | 53 (60) | 52 (13) | 62 (18) | 49 (29) | |
| Tibia | 23 (26) | 24 (6) | 21 (6) | 24 (14) | |
| Humerus | 17 (19) | 12 (3) | 10 (3) | 22 (13) | |
| Other | 7 (8) | 12 (3) | 7 (2) | 5 (3) | |
| Metastatic disease | 0.33 | ||||
| At presentation | 19 (22) | 16 (4) | 21 (6) | 20 (12) | |
| Developed during treatment | 3 (3) | 0 (0) | 7 (2) | 2 (1) | |
| Developed after completion of chemotherapy | 25 (28) | 16 (4) | 17 (5) | 32 (19) | |
| None | 53 (60) | 68 (17) | 55 (16) | 46 (27) | |
| Surgical resection | 0.52 | ||||
| Limb-sparing surgery | 74 (84) | 80 (20) | 72 (21) | 73 (43) | |
| Endoprosthetic or megaprosthetic | 59 (67) | 60 (15) | 62 (18) | 58 (34) | |
| Allograft with vascularized fibula | 12 (14) | 16 (4) | 7 (2) | 14 (8) | |
| Allograft without vascularized fibula | 2 (2) | 4 (1) | 0 (0) | 2 (1) | |
| Wide resection | 1 (1) | 0 (0) | 3 (1) | 0 (0) | |
| Amputation | 17 (19) | 16 (4) | 14 (4) | 19 (11) | |
| Rotationplasty | 7 (8) | 0 (0) | 14 (4) | 7 (4) | |
| Otherb | 2 (2) | 4 (1) | 0 (0) | 2 (1) | |
| Days elapsed from initial report of symptoms to initial orthopaedic evaluation (n = 109) | 35 (17-88) | 61 (26-107) | 35 (14-73) | 31 (16-81) | 0.30 |
| Days elapsed from date of diagnosis to first chemotherapy session (n = 110) | 14 (9-20) | 16 (10-26) | 14 (11-16) | 13 (8-20) | 0.24 |
| Days elapsed between initial orthopaedic visit and surgical resection (n = 112) | 91 (84-104) | 103 (85-106) | 89 (85-101) | 91 (83-100) | 0.31 |
| Patient survivorship | 0.24 | ||||
| Alive with disease | 2 (2) | 0 (0) | 0 (0) | 3 (2) | |
| Dead from disease | 28 (32) | 24 (6) | 17 (5) | 36 (21) | |
| No evidence of disease | 70 (79) | 76 (19) | 83 (24) | 61 (36) | |
| Length of follow-up in years | 4.5 (2.4-7.2) | 4.7 (2.4-6.7) | 4.8 (2.7-7.2) | 4.2 (2.4-7.3) | 0.95 |
Data presented as % (n) or median (IQR).
Fisher exact test and Kruskal-Wallis rank sum test.
Other resection type includes no resection (n = 1) and ankle fusion (n = 1).
The mean ± SD age at time of resection was 13.3 ± 3.9 years (95% CI 9.4 to 17.3). The most frequent method of local control was limb-sparing surgery (74%). Seventeen percent (19 of 113) of patients underwent an amputation and 7% (8 of 113) had a rotationplasty. The mean ± SD follow-up time was 5.4 ± 3.7 years (95% CI 1.6 to 9.3). The mean ± SD time to metastasis, excluding patients who presented with metastatic disease, was 1.0 ± 1.2 years (range 0 to 5). Based on Kaplan-Meier analysis of the entire cohort 5 years after initial evaluation, survivorship for the entire cohort was estimated at 70% (95% CI 62% to 80%). When excluding from analysis patients who presented with metastatic disease, survivorship for this cohort was estimated at 77% (95% CI 68% to 87%).
Twenty-two percent (25 of 113) of patients were categorized into the lowest tertile, 26% (29 of 113) were in the middle tertile, and 52% (59 of 113) were in the highest. There was no relationship in the age at time of presentation across the Childhood Opportunity Index group (lowest tertile 12.4 ± 3.6 years [95% CI 8.4 to 16.3], middle tertile 13.0 ± 4.4 [95% CI 8.5 to 17.1], and highest 13.8 ± 3.9 [95% CI 9.5 to 17.3]; p = 0.41). There was a relationship between the proportion of male patients across the Childhood Opportunity group, with the lowest and highest tertile having a larger proportion of male patients compared with the middle tertile (lowest 56% [95% CI 35% to 75%], middle 28% [95% CI 13% to 47%], and highest 61% [95% CI 47% to 73%]; p = 0.011). Finally, there was a relationship between race and insurance status across tertiles, as expressed by a higher proportion of patients identifying as White and between those with private insurance in the highest tertile compared with the lowest and middle tertiles. Twenty-four percent of the lowest tertile were White (95% CI 10% to 46%) and 24% (6 of 25) were covered by private insurance (95% CI 10% to 46%). Sixty-two percent (18 of 29) of the middle tertile were White (95% CI 42% to 79%) and 55% (16 of 29) were covered by private insurance (95% CI 36% to 73%). Seventy-three percent (43 of 59) of the highest tertile were White (95% CI 60% to 83%; p < 0.001) and 69% (41 of 59) were covered by private insurance (95% CI 56% to 80%; p < 0.001).
Primary and Secondary Study Outcomes
Our primary study goal was to assess whether there were differences in overall and disease-free survival based on the Childhood Opportunity Index tertile. To achieve this, we ran a Kaplan-Meier survival analysis across the lowest, middle, and highest Childhood Opportunity Index tertiles based on 5-year survival with and without patients with metastatic disease at presentation, as well as on the development of metastatic disease.
Our secondary study goal was to assess whether there were differences in time or type of treatment based on the Childhood Opportunity Index tertile. This was assessed via bivariate analysis on the time to initial presentation from the onset of symptoms, time from initial presentation to chemotherapy and surgical resection, and the type of surgical resection.
Ethical Approval
Ethical approval for this study was waived by the institutional review board of The Children’s Hospital of Philadelphia (IRB 17-014774).
Statistical Analysis
We extracted data from the chart review, and results were collected, managed, and stored using REDCap electronic data capture tools [13]. Means, ranges, medians, IQRs, and percentages were used to describe the study sample. Data analysis was conducted across the three groups (lowest tertile, middle, and highest) assessing associations with time to presentation, treatment variations, disease severity, and overall survivorship. Chi-square and Fisher exact tests were applied to compare categorical variables. Mann-Whitney U tests compared continuous data across Childhood Opportunity Index tertiles. Follow-up time was defined as the time from diagnosis to the date of last visit according to EMR. For those who died, their date of death marked the end of the follow-up period. Time to metastasis and survivorship at 5 years of follow-up were reported using cumulative incidence and Kaplan-Meier survival analysis. A log-rank test was applied to evaluate differences in disease-free and overall survivorship. Due to the small sample size, we were unable to control for potential confounders such as race and insurance. However, the three domains (education, health and environment, and social and economic) encapsulated by the Childhood Opportunity Index data indirectly accounted for disparities related to race and insurance status (Supplemental Tables 1-3; http://links.lww.com/CORR/B352). A p value < 0.05 was used as the statistical significance threshold, and all p values were two-sided. All analyses were completed using R version 4.2.2 (R Foundation for Statistical Computing).
Results
Relationship Between Child Opportunity Index and Time to Presentation
There was no association between lower levels of socioeconomic opportunity, as expressed by the lack of difference between the Childhood Opportunity Index tertiles for the interval between symptom onset and first office visit (mean ± SD lowest tertile 77 ± 67 days [95% CI 60 to 94], middle tertile 69 ± 94 days [95% CI 50 to 89], and highest tertile 56 ± 58 days [95% CI 41 to 71]; p = 0.30).
Relationship of Child Opportunity Index and Variations in Patient Treatment
We found no association between lower levels of socioeconomic opportunity, as expressed by the lack of difference between the Childhood Opportunity Index tertiles and the time elapsed from the first office visit to the first chemotherapy session (mean ± SD lowest tertile 19 ± 12 days [95% CI 12 to 26], middle tertile 19 ± 14 days [95% CI 11 to 26], and highest tertile 15 ± 10 days [95% CI 8.4 to 21]; p = 0.31).
Likewise, no association was found in the timing of surgical resection based on Childhood Opportunity Index tertiles (mean ± SD lowest tertile 99 ± 35 days [95% CI 87 to 111], middle tertile 88 ± 28 days [95% CI 77 to 99], and highest tertile 102 ± 64 days [95% CI 86 to 118]; p = 0.24) or in terms of the type of surgical resection (limb-sparing versus amputation: 84% [21 of 25] in the lowest tertile, 83% [24 of 29] in the middle tertile, and 81% [48 of 59] in the highest tertile received limb-sparing surgery; p = 0.52).
Relationship Between Child Opportunity Index and Disease Severity and Overall Survival
We found no association in terms of disease-free survival nor overall survivorship across Child Opportunity Index groups. Twenty-seven percent of the lowest tertile developed metastatic disease at 5 years (95% CI 7.8% to 43%) compared with 44% of the middle tertile (95% CI 23% to 59%) and 56% of the highest tertile (95% CI 40% to 67%; p = 0.22) (Fig. 1). The lowest tertile’s survivorship was 74% (95% CI 58% to 95%), the middle tertile’s survivorship was 82% (95% CI 68% to 98%), and the highest tertile’s survivorship was 64% (95% CI 52% to 78%; p = 0.27). No association between survivorship during Kaplan-Meier analysis and the Childhood Opportunity Groups was found even after excluding patients who presented with metastatic disease (lowest tertile 84% [95% CI 68% to 100%], middle tertile 91% [95% CI 80% to 100%], and highest tertile 68% [95% CI 55% to 83%]; p = 0.10) (Fig. 2).
Fig. 1.
This graph shows the cumulative incidence of the development of metastatic disease over 5 years for the entire cohort stratified by Childhood Opportunity Index group (p = 0.22).
Fig. 2.
These graphs show the Kaplan-Meier curves stratified by Childhood Opportunity Index group for patient survivorship (A) after diagnosis of the entire cohort (p = 0.27) and (B) without patients who presented with metastatic disease (p = 0.10).
Discussion
Prior work has found a relationship between a more disadvantaged socioeconomic background and worse survivorship in patients with osteosarcoma [19, 20]. Yet, there is also evidence of a worse histologic response to chemotherapy in patients with less socioeconomic opportunity without differences in overall survivorship [25]. It remains unclear whether delays to care, difficulty accessing care, or other unknown factors relate to these findings. We aimed to compare the presentation, treatment, and outcomes of pediatric patients with osteosarcoma presenting to our institution based on differences in socioeconomic opportunity. Our study found no association between a child’s socioeconomic opportunity and delays to presentation or treatment, frequency of patients presenting with metastatic disease, variations in treatment, or worse overall survivorship at our large, urban institution located in the northeast region of the United States. Based on our findings, we believe that the lack of association between socioeconomic background and disparities of care in our study, which was found to be present in prior studies, may be related to structural factors outside the direct control of physicians or patients, such as greater access to physicians and public transit.
Limitations
First, the design of our study was retrospective, which introduced a potential selection bias. However, we included all patients who presented with osteosarcoma, making the risk for selection bias low. Most patients who were excluded were seeking a second opinion, or the initiation of their treatment predated storing patient data in EMRs. These exclusion criteria impacted all Childhood Opportunity Index groups equally, which kept our risk for selection bias low. Additionally, because the study was conducted at a single, large, pediatric tertiary care referral center, generalizability may be limited. While our cohort’s demographic characteristics were not reflective of the US pediatric population, they were reflective of the population in the tristate area that our institution cares for (Pennsylvania, New Jersey, and Delaware). This study is most applicable to large, urban, academic pediatric hospitals across the United States that offer similar support and access to public transit. While generalizability is limited, our study has offered the novel finding that disparities across socioeconomic backgrounds in osteosarcoma treatment were not seen at our institution even though they were found to be present in prior work. The lack of an association in our study may suggest that the resources and accommodations common to our region or Children’s hospitals (such as insurance access, public transit, comprehensive assistance from social work) help children to be seen more quickly, but future work on this topic is necessary. Second, few patients were lost to follow-up or had missing data. All included patients with missing data were only missing one variable. These excluded patients and missing data were evenly distributed across the Childhood Opportunity Index tertiles, which indicated that the risk of transfer bias impacting our results was low.
Third, due to the limited sample size and distribution of race and insurance in our study, we were unable to account for these confounding variables in our analysis of Childhood Opportunity Index tertiles. Doing so would create the risk of sparse data bias. Furthermore, the Childhood Opportunity Index encompasses quality of education, healthcare resources, and neighborhood conditions. Controlling for race and insurance would not substantially add to this study, as many of the variables in the Childhood Opportunity Index may indirectly account for disparities related to race and insurance status. Our goal in this study was not to suggest a causal relationship, which would require controlling for race and insurance. Instead, we hoped to highlight the apparent lack of association between socioeconomic status and the presentation, treatment, and outcomes for patients presenting to our institution and to generate hypotheses. There was also an associated lower frequency of male patients in the middle tertile compared with the lowest and highest. It is established that male patients with osteosarcoma often have a worse prognosis [32]. However, given that the lowest and highest tertiles saw no difference in the development of metastatic disease and overall survivorship at 5 years, it did not appear to affect our findings in the expected manner (more severe prognosis in the lowest and highest tertile). Fourth, our institution follows international standards for osteosarcoma treatment typical of those found across the United States. The MAP (methotrexate, doxorubicin, and cisplatin) regimen has been well established since 2005 and was the predominant form of chemotherapy for the duration of this study [35]. Similarly, the proportion of patients treated with amputation versus limb-salvage surgery (limb sparing, rotationplasty, wide resection) appears to be reflective of that typically seen in osteosarcoma treatment at pediatric hospitals for a young and active patient population. Therefore, we believe that the likelihood of cotreatment bias is low. Finally, although the Childhood Opportunity Index provides depth into many aspects of social determinants of health, it does not capture all aspects of the social determinants of health that may contribute to health outcomes. Despite the limitations of the Childhood Opportunity Index as a proxy for social determinants of health, it still remains one of the most holistic and granular approaches for this study design.
Discussion of Key Findings
We found no association between differing scores on the Child Opportunity Index, a validated metric for socioeconomic disadvantage, and key metrics related to timing of presentation; in other words, patients from more distressed socioeconomic circumstances did not present later with their bone sarcomas. While in our study we did not find this relationship, the reason as to why is unclear. Prior research has established lower socioeconomic circumstances as a risk factor for delays to care [4, 5, 12, 15, 16, 18, 23, 28]. Richardson et al. [25] theorized that the socioeconomic factors negatively associated with histologic response may be representative of patient factors that increase difficulties in accessing care. However, even though our results did not find an association between delays to care and socioeconomic background, it is reasonable to believe that these discrepancies likely exist in other patient populations. One possible explanation for our findings may relate to the less strict Medicaid eligibility guidelines for children and adolescents and fewer discrepancies in cost and healthcare access when children are on public versus private insurance [6, 7, 30]. With fewer disparities and increased access to care in pediatrics, when compared with children who are uninsured, children with Medicaid are four times more likely to have consistent healthcare visits and two times more likely to utilize preventative care services [1]. Physicians treating osteosarcoma should be aware of the risks for delays in presentation (such as insufficient insurance coverage, transportation, or language barriers) and whether these factors are creating barriers to care in their patient population.
Likewise, we did not find any apparent relationship between the Child Opportunity Index scores and time to treatment or type of resection used by surgeons at our urban academic tumor center. After presentation, no associations were found in the time elapsed from initial evaluation to chemotherapy and surgical resection or in the type of surgical resection performed, which suggests that treatment was relatively uniform regardless of socioeconomic opportunity. Given our institution’s access to public transit, ranked sixth in the nation by US News and World Report [21], we believe it is plausible that it helps diminish the effects that lower socioeconomic opportunity has on medical treatment. Similarly, the more compact population density of the northeast region in the United States, with a higher density of physicians, may also help promote the accessibility of care. Prior research utilizing the Surveillance, Epidemiology, and End Results (SEER) database has found an association between states with lower physician density and worse overall survival following a cancer diagnosis [29]. Urban centers are associated with a higher density of providers. More specifically, the distance to a pediatric specialist is less, and the relative supply is higher [2, 17]. Prior work has found that regions with less geographic access to providers are correlated with higher rates of unplanned readmissions [24]. Given this information, physicians practicing in areas with lower population density should consider whether their patient population experiences longer delays to care and see if there are offerings in their area that promote equitable access to care.
Finally, and surprisingly, we found no relationship between Child Opportunity Index scores and survivorship free from metastatic disease or overall survivorship. In the context of the lack of association between presentation and treatment and socioeconomic opportunity, our findings on survivorship indicated no association in treatment outcomes as well. Prior studies utilizing national databases found evidence to suggest that socioeconomic factors are associated with a worse histologic response and survival outcomes [19,20,25]. Interestingly, Nathan and Healy [20] found that the survival outcomes in less affluent states in the United States remained the same over the 3 decades of their study, while the more affluent states saw improvements. However, these studies included both pediatric and adult patients. The differences in insurance eligibility, increased access to and more affordable treatment for children, and the accommodations offered at our institution may all have contributed to dampening the effect that lower socioeconomic opportunity had on osteosarcoma treatment. Perhaps our dedicated social work team and other beneficial resources more prevalent in Children’s hospitals compared with general hospitals help patients to be seen more quickly. These explanations are not factors inherent to our institution but rather structural factors outside of the direct control of providers or patients. To inform future directions of public policy, it is important to further elucidate the relationship between a patient’s proximity to a provider, access to public transit, and regional insurance eligibility criteria and disparities in healthcare treatment public policy.
Conclusion
In our single-center retrospective study of 113 children who presented with osteosarcoma, we did not find an association between a patient’s national socioeconomic opportunity and their time to presentation, chemotherapy treatment, time to and type of surgical resection, and disease-free and overall survival. Prior work has found an association between socioeconomic background and disparities in osteosarcoma treatment, but based on our findings, we believe that support services more common in Children’s hospitals, proximity to providers, access to public transit, and regional insurance policies may help diminish these disparities. Future multicenter studies are needed to further explore the role that regional variations and the aforementioned factors may play in osteosarcoma treatment to help inform the direction of public policy.
Acknowledgment
We thank Divya Talwar MPH, PhD for her review of the statistical analysis conducted during this study.
Footnotes
Each author certifies that there are no funding or commercial associations (consultancies, stock ownership, equity interest, patent/licensing arrangements, etc.) that might pose a conflict of interest in connection with the submitted article related to the author or any immediate family members.
All ICMJE Conflict of Interest Forms for authors and Clinical Orthopaedics and Related Research® editors and board members are on file with the publication and can be viewed on request.
Ethical approval for this study was waived by the institutional review board of The Children’s Hospital of Philadelphia (IRB 17-014774).
Contributor Information
David Kell, Email: kelld1@chop.edu.
Daniel Yang, Email: danielzyang@yahoo.com.
Juliana Lee, Email: leej35@chop.edu.
Kevin Orellana, Email: kevin.orellana01@utrgv.edu.
Sarah Wetzel, Email: wetzels1@chop.edu.
References
- 1.AHIP. The value of Medicaid: providing access to care and preventive health services. Available at: https://www.ahip.org/resources/the-value-of-medicaid-providing-access-to-care-and-preventive-health-services. Accessed February 14, 2024.
- 2.Bell N, Lòpez-De Fede A, Cai B, Brooks J. Geographic proximity to primary care providers as a risk-assessment criterion for quality performance measures. PLoS One. 2022;17:e0273805. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Bielack SS, Kempf-Bielack B, Delling G, et al. Prognostic factors in high-grade osteosarcoma of the extremities or trunk: an analysis of 1,702 patients treated on neoadjuvant cooperative osteosarcoma study group protocols. J Clin Oncol. 2002;20:776-790. [DOI] [PubMed] [Google Scholar]
- 4.Bouchard L, Albertini M, Batista R. Research on health inequalities: a bibliometric analysis (1966-2014). Soc Sci Med. 2015;141:100-108. [DOI] [PubMed] [Google Scholar]
- 5.Chen J, Vargas-Bustamante A, Mortensen K, Ortega AN. Racial and ethnic disparities in health care access and utilization under the Affordable Care Act. Med Care. 2016;54:140-146. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Dubay L, Kenney GM. Health care access and use among low-income children: who fares best? Health Aff . 2001;20:112-121. [DOI] [PubMed] [Google Scholar]
- 7.Duderstadt KG, Hughes DC, Soobader MJ, Newacheck PW. The impact of public insurance expansions on children’s access and use of care. Pediatrics. 2006;118:1676-1682. [DOI] [PubMed] [Google Scholar]
- 8.Fletcher CD, Bja PCW, Hogendoorn F. WHO Classification Tumors Soft Tissue Bone. International Agency Research Cancer; 2013:282-288. [Google Scholar]
- 9.Friebele JC, Peck J, Pan X, Abdel-Rasoul M, Mayerson JL. Osteosarcoma: a meta-analysis and review of the literature. Am J Orthop (Belle Mead NJ). 2015;44:547-553. [PubMed] [Google Scholar]
- 10.Goedhart LM, Gerbers JG, Ploegmakers JJW, Jutte PC. Delay in diagnosis and its effect on clinical outcome in high-grade sarcoma of bone: a referral oncological centre study. Orthop Surg. 2016;8:122-128. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Goyal S, Roscoe J, Ryder WDJ, Gattamaneni HR, Eden TOB. Symptom interval in young people with bone cancer. Eur J Cancer. 2004;40:2280-2286. [DOI] [PubMed] [Google Scholar]
- 12.Hanna TP, King WD, Thibodeau S, et al. Mortality due to cancer treatment delay: systematic review and meta-analysis. BMJ. 2020;371:m4087. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Harris PA, Taylor R, Minor BL, et al. The REDCap consortium: building an international community of software platform partners. J Biomed Inform. 2019;95:103208. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Lawrenz JM, Styron JF, Parry M, Grimer RJ, Mesko NW. Longer duration of symptoms at the time of presentation is not associated with worse survival in primary bone sarcoma. Bone Joint J. 2018;100-B:652-661. [DOI] [PubMed] [Google Scholar]
- 15.Li X, Galvin JW, Li C, Agrawal R, Curry EJ. The impact of socioeconomic status on outcomes in orthopaedic surgery. J Bone Joint Surg Am. 2020;102:428-444. [DOI] [PubMed] [Google Scholar]
- 16.Lin SW, Anisa KN. Effects of socioeconomic status on cancer patient survival: counterfactual event-based mediation analysis. Cancer Causes Control. 2021;32:83-93. [DOI] [PubMed] [Google Scholar]
- 17.Mayer ML, Beil HA, von Allmen D. Distance to care and relative supply among pediatric surgical subspecialties. J Pediatr Surg. 2009;44:483-495. [DOI] [PubMed] [Google Scholar]
- 18.McVicker L, Cardwell CR, Edge L, et al. Survival outcomes in endometrial cancer patients according to diabetes: a systematic review and meta-analysis. BMC Cancer. 2022;22:427. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Miller BJ, Gao Y, Duchman KR. Socioeconomic measures influence survival in osteosarcoma: an analysis of the National Cancer Data Base. Cancer Epidemiol. 2017;49:112-117. [DOI] [PubMed] [Google Scholar]
- 20.Nathan SS, Healey JH. Making a case for the socioeconomic determinacy of survival in osteosarcoma. Clin Orthop Relat Res. 2013;471:784-791. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Navarre B. The 10 best cities for public transportation. US News and World Report. 2024. Available at: https://www.usnews.com/news/cities/slideshows/10-best-cities-for-transportation?slide=6. Accessed August 28, 2024. [Google Scholar]
- 22.Noelke C, McArdle N, Baek M, et al. Child Opportunity Index 2.0 technical documentation. 2020. Available at: http://new.diversitydatakids.org/sites/default/files/2020-01/ddk_coi2.0_technical_documentation_20200115_2.pdf. Accessed February 14, 2024.
- 23.Pinheiro LC, Reshetnyak E, Akinyemiju T, Phillips E, Safford MM. Social determinants of health and cancer mortality in the Reasons for Geographic and Racial Differences in Stroke (REGARDS) cohort study. Cancer. 2022;128:122-130. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Pinto NM, LuAnn Minich L, Yoo M, et al. Influence of geographic access on surgical center readmissions after index congenital heart surgery. J Pediatr. 2021;234:195-204.e3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Richardson SM, Wurtz LD, Collier CD. Ninety percent or greater tumor necrosis is associated with survival and social determinants of health in patients with osteosarcoma in the National Cancer Database. Clin Orthop Relat Res. 2023;481:512-522. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Rougraff BT, Davis K, Lawrence J. Does length of symptoms before diagnosis of sarcoma affect patient survival? Clin Orthop Relat Res. 2007;462:181-189. [DOI] [PubMed] [Google Scholar]
- 27.Sasi A, Ganguly S, Biswas B, et al. Determinants and impact of diagnostic interval in bone sarcomas: a retrospective cohort study. Pediatr Blood Cancer. 2023;70:e30135. [DOI] [PubMed] [Google Scholar]
- 28.Sharrocks K, Spicer J, Camidge DR, Papa S. The impact of socioeconomic status on access to cancer clinical trials. Br J Cancer. 2014;111:1684-1687. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Siddappa Malleshappa SK, Giri S, Patel S, et al. Overall survival based on oncologist density in the United States: a retrospective cohort study. PLoS One. 2021;16:e0250894. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Tolbert J, Drake P, Damico A. Key facts about the uninsured population. Available at: https://www.kff.org/uninsured/issue-brief/key-facts-about-the-uninsured-population/. Accessed February 15, 2024.
- 31.US Census Bureau. Understanding geographic identifiers (GEOIDs). 2020. Available at: https://www.census.gov/programs-surveys/geography/guidance/geo-identifiers.html. Accessed February 14, 2024.
- 32.Williams LA, Barragan S, Lu Z, Weigel BJ, Spector LG. Sex differences in osteosarcoma survival across the age spectrum: a National Cancer Database analysis (2004-2016). Cancer Epidemiol. 2024:102565. [DOI] [PubMed] [Google Scholar]
- 33.Wurtz LD, Peabody TD, Simon MA. Delay in the diagnosis and treatment of primary bone sarcoma of the pelvis. J Bone Joint Surg Am. 1999;81:317-325. [DOI] [PubMed] [Google Scholar]
- 34.Yang JYK, Cheng FWT, Wong KC, et al. Initial presentation and management of osteosarcoma, and its impact on disease outcome. Hong Kong Med J. 2009;15:434-439. [PubMed] [Google Scholar]
- 35.Yu D, Zhang S, Feng A, et al. Methotrexate, doxorubicin, and cisplatinum regimen is still the preferred option for osteosarcoma chemotherapy. Medicine (Baltimore). 2019;98:e15582. [DOI] [PMC free article] [PubMed] [Google Scholar]