Which Neighborhood-level Metric Is Most Appropriate for Pediatric Sports Medicine Disparities Research?

Abstract

Background

Neighborhood conditions are associated with access to care and outcomes. The Child Opportunity Index (COI) and Area Deprivation Index (ADI) are commonly used in pediatric research but without a clear rationale as to why one is chosen over the other. Despite an increasing volume of health equity research, there is little evidence that has directly compared the ADI and COI’s associations with previously established pediatric orthopaedic disparities. Thus, the most appropriate neighborhood-level measure for pediatric orthopaedic research remains unclear.

Questions/purposes

(1) Do COI and ADI correlate with each other? (2) How do COI and ADI compare in their associations with previously established pediatric orthopaedic disparities, including time to ACL reconstruction (ACLR) and the presence of concomitant meniscal and chondral pathology?

Methods

This is a retrospective, comparative study of patients aged 18 years or younger who underwent primary ACLR between 2010 and 2023 at one tertiary center. We excluded patients who were missing COI or ADI data. Patients who underwent multiligament reconstruction, revision ACLR, intentionally staged or delayed procedures, or previous surgery on either knee were also excluded. We initially considered 806 patients, of whom 9% (72) were excluded for prespecified reasons. Consequently, 734 patients were included in the study (mean ± SD age was 16 ± 2 years and 52% [382] were boys). The median (IQR) time between injury and surgery was 74 days (82). Fifty-five percent (401) of patients had public insurance (Medicaid), 42% (306) had private insurance, and 4% (27) had no insurance or other types of insurance. The COI and ADI scores were assigned by address at the time of surgery. The COI quantifies neighborhood resources with 44 indicators across three domains: education, health and environment, as well as social and economic. It is scored from 0 to 100, with 0 indicating the lowest level of neighborhood resources. The ADI quantifies socioeconomic deprivation with 17 indicators across four domains: education, income, employment, and housing quality. It is also scored from 1 to 100, with 100 indicating the highest level of deprivation. Only national-level COI and ADI scores were used, given that the ADI does not provide continuous data at the state and metropolitan levels. The mean ± SD COI and ADI scores for the study population were 50 ± 29 and 44 ± 22, respectively. Outcomes of interest included time to surgery and intraoperative concomitant pathology. The Pearson correlation coefficients (r) were calculated for the comparison of continuous variables. Negative r values indicate inverse relationships (for example, a negative value suggests that as the value of one variable increases, the other decreases). By convention, an r value of 1 is a perfect correlation, 0.7 < r < 1 is a strong correlation, 0.3 < r < 0.7 is a moderate correlation, and 0 < r < 0.3 is a weak correlation. Regression analyses, reported with regression coefficients or ORs and 95% confidence intervals (CIs), assessed the association of the COI or ADI with clinical outcomes (timing of surgery and concomitant meniscal or chondral pathology) while controlling for confounders.

Results

The COI and ADI demonstrated moderate correlation with each other (r = −0.69; p < 0.001), indicating that an increasing COI score (for example, more neighborhood opportunity) correlates with a decreasing ADI score (for instance, less neighborhood deprivation). After controlling for insurance and time to surgery, when applicable, a higher COI score (indicating higher level of neighborhood opportunity) was associated with a shorter time to ACLR (regression coefficient −0.56 [95% CI −0.95 to −0.16]; p = 0.006); lower odds of concomitant meniscectomy (OR 0.99 [95% CI 0.98 to 0.99]; p = 0.02); and surgery beyond 60 days (OR 0.99 [95% CI 0.98 to 0.99]; p < 0.001), beyond 90 days (OR 0.99 [95% CI 0.98 to 0.99]; p < 0.001), and beyond 180 days (OR 0.99 [95% CI 0.98 to 0.99]; p = 0.02) after injury. In separate multivariable models, a higher ADI score (indicating a higher level of deprivation) was associated with increased odds of surgery beyond 60 days (OR 1.02 [95% CI 1.01 to 1.03]; p < 0.001) and 90 days (OR 1.02 [95% CI 1.01 to 1.02]; p < 0.001) after injury, but no other outcomes that we assessed.

Conclusion

In the context of pediatric ACLR, the results suggest that the COI may be more appropriate than the ADI in identifying pediatric-specific disparities in sports medicine. This may be due to differences in the types and number of underlying indicators contributing to each index and the underlying methodologies used in each index’s development. Additionally, continuous ADI data are only available on the national scale.

Clinical Relevance

These findings highlight the importance of selecting appropriate neighborhood-level indices when conducting disparities research. Consideration should be given to the underlying components of these indices and their potential relevance to the population of interest and research question. Future research should focus on comparing the COI and ADI in other clinical contexts to assess the generalizability of these findings and better inform future research methodology.

Introduction

Associations between social determinants of health (SDOH) and healthcare access or clinical outcomes in pediatric patients are well established [5, 6, 15]. Previous health equity research in orthopaedics has commonly relied on insurance and median household income by ZIP code as proxies for socioeconomic status [7, 8, 14, 23, 30]. However, given the complexity and multidimensionality of SDOH, there is a need for indices that comprehensively account for multiple elements and better quantify socioeconomic resources [24]. The Child Opportunity Index (COI) quantifies neighborhood conditions across three domains: education, health and environment, as well as social and economic [9]. The Area Deprivation Index (ADI) is another neighborhood-level metric that quantifies socioeconomic deprivation across four major domains: education, income, employment, and housing quality [42]. Both have been used in a variety of research contexts including orthopaedic injury epidemiology, acute care utilization, access to care, and COVID-19 disparities research [3, 5, 12, 15, 16, 26–29, 34, 37–40, 43, 45, 48].

In children and adolescents undergoing ACL reconstruction (ACLR), neighborhood conditions and other SDOH are associated with timing of surgery [7, 31, 35, 36]. In young patients, delays to ACLR are associated with an increased risk of persistent instability, more severe chondral pathology, and irreparable meniscal injury [18, 20]. Prior studies have also demonstrated that delays in ACLR can be associated with worse postoperative outcomes and a higher risk of future reoperation [2, 7, 30, 32, 36, 46]. Both the COI and ADI have been used in pediatric ACLR research to examine the role of neighborhood-level factors in the timing of care and postoperative outcomes [20, 21, 30, 31, 35–37, 47]. Although several other similar metrics are used in research, these are among the two most common in pediatric orthopaedic research.

Despite the frequent use of both the COI and ADI, it remains unclear which metric is most appropriate for use in orthopaedic health equity research. Prior studies have used these metrics independently, but there is little orthopaedic research that has directly compared their associations with previously established disparities. Given the increasing volume of health equity research in orthopaedics [1, 11, 44], an understanding of how these metrics interrelate will aid in the interpretation of existing evidence and better inform future research design. Furthermore, since ACLR is a common procedure that is somewhat time-sensitive in children and adolescents, the reporting of accurate and meaningful data is critical in identifying disparities and ultimately designing interventions.

This study aims to answer the following questions: (1) Do COI and ADI correlate with each other? (2) How do COI and ADI compare in their associations with previously established pediatric orthopaedic disparities, including time to ACLR and the presence of concomitant meniscal and chondral pathology?

Patients and Methods

Study Design

This was a retrospective, comparative study of patients 18 years or younger who underwent primary ACLR at one urban, tertiary center between 2010 and 2023.

Participants

Patients were identified within the institution’s electronic health record system using Common Procedural Terminology (CPT) code 29888. Inclusion criteria were age 18 years or younger, primary ACLR, follow-up of at least 180 days, as well as documentation of date of injury and home address. Exclusion criteria included missing address, inability to calculate the COI or ADI score, multiligament reconstruction, revision ACLR, previous contralateral or ipsilateral ACL injury, history of surgery on the ACL-injured knee, and intentionally staged or delayed ACLR. Two board-certified, fellowship-trained pediatric orthopaedic surgeons (NMP, CF [not a study author]) performed all reconstructions.

Initially, 806 patients were identified. Of these, 9% (72) were excluded for the following reasons: inability to calculate the COI or ADI (46), multiligament reconstructions (12), previous ipsilateral surgery (5), intentionally staged ACLR (5), and skeletally immature patients initially treated non-operatively at another institution (4). This resulted in 734 patients included in the study. Patients who were excluded due to missing COI or ADI data had a younger mean age than those included in the study but were otherwise similar in terms of gender, race or ethnicity, insurance, and time to surgery (Supplemental Table 1; http://links.lww.com/CORR/B502).

Data Collection

We reviewed medical records for demographic, pre-operative, and intraoperative data. Pertinent data included age, gender, race and ethnicity (self-reported by the patient), insurance type, address at time of surgery, and date of injury. As mentioned previously, the COI and ADI were selected for comparison in this study because they are two of the most commonly used metrics in pediatric orthopaedic disparities research. Addresses were used to determine the COI and ADI scores. Operative reports were evaluated to collect information on surgical techniques, concomitant meniscal or chondral pathology, and additional procedures like meniscus repair or meniscectomy.

Child Opportunity Index

Initially published in 2014, the COI is a composite index that quantifies neighborhood resources related to child health and development, termed “neighborhood opportunity.” The most recent iteration, COI 3.0, measures 44 indicators that are grouped into the three main domains of education, health and environment, as well as social and economic [9]. The COI scores are calculated on a census tract level, which are small, relatively permanent geographic areas within a county used by the US Census Bureau. Census tracts typically contain 2500 to 8000 residents and are delineated with the intention of being maintained for a long period of time for statistical data collection purposes. Z-scores are calculated for each census tract on national, state, and metropolitan levels. These indicators are then aggregated across domains into a combined total COI score for each census tract. A percentile ranking is applied, creating a 0 to 100 scale, with 100 indicating the highest neighborhood opportunity. Scores are also organized into categories of very low, low, moderate, high, and very high based upon quintile rankings. To calculate COI scores in this study, patient addresses at the time of surgery were converted to 2010, 2015, or 2020 United States census tract identifiers. Scores from the COI iteration (1.0, 2.0, or 3.0) closest to the date of surgery were used for each patient. This was done to ensure the most relevant socioeconomic contexts for each patient at the time of surgery with the available census tract data. Regardless, according to developers, COI scores are comparable across time [9].

Area Deprivation Index

The ADI is a census-based index that measures the socioeconomic deprivation or disadvantage at the neighborhood level. It uses 17 indicators that are grouped into four domains: education, income, employment, and housing quality [22, 42]. The ADI scores are assigned on the census block level, which are smaller subdivisions of census tracts, much more variable in size, and typically contain 600 to 3000 residents. They range from 1 to 100, with a score of 100 indicating the highest level of deprivation/disadvantage. The ADI scores are determined by ranking all census block groups nationally and assigning a percentile value based on the national distribution. Each block group is placed in a 1% “bin” where a ranking of 1 indicates the lowest level of “disadvantage” and a ranking of 100 indicates the highest level of “disadvantage.” The ADI does not provide indicator-specific scores and only provides national-level rankings. Therefore, only national-level composite COI and ADI scores were used in this study to ensure standardized comparisons. In this study, the ADI scores were calculated using the 9-digit ZIP codes corresponding to the patient’s address at time of surgery. We used the ADI version that was closest to the patient’s date of surgery (2015, 2020, and 2022), with corresponding census data from the respective year.

Descriptive Data

In this study, the 734 included patients were a mean ± SD age of 16 ± 2 years, and 52% (382) were boys. The median (IQR) time between injury and surgery was 74 days (82). The mean COI and ADI scores for the study population were 50 ± 29 and 44 ± 22, respectively. Fifty-five percent (401) of patients had public insurance (Medicaid), 42% (306) had private insurance, and 4% (27) had no insurance or other types of insurance (no insurance, self-pay, and Tricare [military insurance]).

Primary and Secondary Outcomes of Interest

The primary focus of this study was to determine whether COI and ADI correlate with each other. The COI and ADI scores were obtained for each patient to conduct this analysis. The secondary focus of the study was to evaluate associations of the COI and ADI with time to ACLR, presence of a meniscus tear, usage of meniscectomy, and presence of chondral injury. We controlled for confounding variables (insurance type and time to surgery, when appropriate) in these analyses.

Ethical Approval

We obtained ethical approval for this study from Lurie Children’s Hospital Institutional Review Board (number 2019–2728).

Statistical Analysis

Statistical analysis was performed with SPSS for Macintosh, version 28.0 (IBM Corp). We used Kolmogorov-Smirnov testing to determine distribution normality, and we calculated Pearson correlation coefficients (r) for the comparison of continuous COI and ADI scores. As the r value increases, the parameters in question are more tightly correlated. Negative r values indicate inverse relationships (for example, a negative value suggests that as the value of one variable increases, the other decreases). By convention, an r value of 1 is a perfect correlation, 0.7 < r < 1 is a strong correlation, 0.3 < r < 0.7 is a moderate correlation, and 0 < r < 0.3 is a weak correlation. However, these are not absolute categorizations, as r is a continuous value. After confirming normal distribution of all continuous variables, we used independent-samples t-tests to compare means. We used chi-square or Fisher exact tests to compare categorical variables, as appropriate. Then, we conducted purposeful entry linear or logistic regression analyses to evaluate associations between dependent and independent variables while controlling for confounders. Variables were entered into regression models if they were statistically significant in the univariable analysis. The outcome variables of interest, as noted above, included time to ACLR, presence of meniscal or chondral injury, and usage of meniscectomy. Continuous COI or ADI scores and insurance were independent variables in separate models. Time to ACLR was included as an additional independent variable in the analyses focusing on concomitant meniscal or chondral injury and usage of meniscectomy, as this is a risk factor for such pathology [13, 17]. For each outcome of interest, separate models were created with the ADI exchanged for the COI. For linear regression, continuous variables were tested for linearity. The results of regression analyses are reported with regression coefficients (for linear regression) or ORs (for logistic regression) and 95% confidence intervals (CIs). Linear regression coefficients can be positive (the dependent variable increases as the independent variable increases) or negative (the dependent variable decreases as the independent variable increases). In logistic regression, an OR > 1 indicates increased odds and an OR < 1 indicates decreased odds. When interpreting the association of the COI or ADI (continuous variables) with categorical outcome variables in logistic regression, the OR reflects the change in odds for every 1-unit increase in the COI or ADI. Statistical significance was defined as p < 0.05.

Results

Correlation Between the COI and ADI

Although a broad dispersion of the data was noted visually (Fig. 1), the COI and ADI scores were found to correlate moderately with each other (r = −0.69; p < 0.001). This indicates that higher COI scores (greater neighborhood opportunity) are associated with lower ADI scores (less neighborhood deprivation) (Fig. 1).

Fig. 1.

Scatter plot demonstrating correlation between the COI and the ADI (r = −0.69; p < 0.001). Higher COI scores (greater neighborhood opportunity) are associated with lower ADI scores (indicating less deprivation).

Comparison of the COI and ADI in the Context of Previously Established Pediatric Orthopaedic Disparities

Time Between ACL Injury and Surgery

After controlling for insurance type, every 1-point increase in the COI score (indicating increased neighborhood opportunity) was associated with 0.56 fewer days between ACL injury to surgery (regression coefficient −0.56 [95% CI −0.95 to −0.16]; p = 0.006). In a separate model, every 1-point increase in the ADI score (indicating increased deprivation) was associated with 0.60 more days to ACLR (regression coefficient 0.60 [95% CI 0.03 to 1.0]; p = 0.04) (Table 1). Time to ACLR was then reorganized as a categorical variable. After controlling for insurance in multivariable analyses, higher COI values were associated with decreased odds of surgery beyond 60 days (OR 0.99 [95% CI 0.98 to 0.99]; p < 0.001), beyond 90 days (OR 0.99 [95% CI 0.98 to 0.99]; p < 0.001), and beyond 180 days (OR 0.99 [95% CI 0.98 to 0.99]; p = 0.02) after injury. For example, every 1-point increase in COI (increasing neighborhood opportunity) lowers the odds of surgery beyond 60 days 0.98 times. In similar multivariable models, higher ADI values were associated with increased odds of surgery beyond 60 days (OR 1.02 [95% CI 1.01 to 1.03]; p < 0.001) and 90 days (OR 1.02 [95% CI 1.01 to 1.02]; p < 0.001) after injury, but not 180 days (Table 2).

Table 1.

Factors associated with time to surgery (as a continuous variable)

Variable	Regression coefficient (95% CI)	p value
Regression model with the COI
COI	−0.56 (−0.95 to −0.16)	0.006
Public insurance	41 (21 to 62)	< 0.001
Regression model with the ADI
ADI	0.60 (0.03 to 1.0)	0.04
Public insurance	47 (25 to 68)	< 0.001

An increasing COI value (scored from 0–100) indicates greater neighborhood opportunity, whereas an increasing ADI value (scored from 1–100) indicates greater deprivation. Each 1-point increase in the COI or ADI is associated with a 1-day increase (if > 0.0) or decrease (if < 0.0) in the timing of surgery. Private insurance was the reference group to which public insurance was compared. Separate regression analyses were conducted, first with the COI, then with the ADI.

Table 2.

Factors associated with timing of surgery (as a categorical variable)

Variable	OR (95% CI)	p value
> 60 days after injury
COI	0.99 (0.98–0.99)	< 0.001
Public insurance	3.4 (2.4–4.8)	< 0.001
ADI	1.02 (1.01–1.03)	< 0.001
Public insurance	3.5 (2.4–4.9)	< 0.001
> 90 days after injury
COI	0.99 (0.98–0.99)	< 0.001
Public insurance	2.7 (1.9–3.9)	< 0.001
ADI	1.02 (1.01–1.02)	< 0.001
Public insurance	2.9 (2.0– 4.1)	< 0.001
> 180 days after injury
COI	0.99 (0.98–0.99)	0.02
Public insurance	2.3 (1.4–3.7)	0.001
ADI	1.0 (0.9–1.0)	0.3
Public insurance	2.7 (1.6–4.4)	< 0.001

An increasing COI score indicates greater neighborhood opportunity whereas an increasing ADI score indicates greater deprivation. Each 1-point increase in the COI or ADI is associated with the respective increase (if > 1.0) or decrease (if < 1.0) in odds of delayed surgery. Private insurance was the reference group to which public insurance was compared. Separate logistic regression analyses were conducted for each outcome of interest, first with the COI, then with the ADI.

Presence of a Concomitant Meniscus Tear

When controlling for insurance and time to surgery in the multivariable regression, higher COI values (greater neighborhood opportunity) were associated with decreased odds of the presence of a concomitant meniscus tear during ACLR (OR 0.99 [95% CI 0.98 to 0.99]; p = 0.009). In a similar model, no association was found with the ADI (Table 3).

Table 3.

Factors associated with a concomitant meniscus tear

Variable	OR (95% CI)	p value
Regression model with COI
COI	0.99 (0.98–0.99)	0.009
Public insurance	1.3 (0.9–1.9)	0.1
Time to surgery (for each 90-day increase)	1.2 (1.1–1.3)	0.02
Regression model with ADI
ADI	1.0 (0.9–1.01)	0.3
Public insurance	1.4 (1.0–2.1)	0.04
Time to surgery (for each 90-day increase)	1.2 (1.1–1.4)	0.007

An increasing COI score indicates greater neighborhood opportunity whereas an increasing ADI score indicates greater deprivation. Each 1-point increase in the COI or ADI and 1-day increase in time to surgery is associated with the respective increase (if > 1.0) or decrease (if < 1.0) in the odds of a concomitant meniscus tear. Private insurance was the reference group to which public insurance was compared. Separate logistic regression analyses were conducted, first with the COI, then with the ADI.

Concomitant Meniscectomy

After controlling for insurance and time to surgery, higher COI values were associated with decreased odds of a concomitant meniscectomy (OR 0.99 [95% CI 0.98 to 0.99]; p = 0.02), but ADI scores were not (Table 4).

Table 4.

Factors associated with concomitant meniscectomy

Variable	OR (95% CI)	p value
Regression model with COI
COI	0.99 (0.98–0.99)	0.02
Public insurance	1.1 (0.7–1.7)	0.7
Time to surgery (for each 90-day increase)	2.4 (1.1–14.3)	0.01
Regression model with ADI
ADI	1.0 (0.9–1.01)	0.3
Public insurance	1.3 (0.8–2.1)	0.2
Time to surgery (for each 90-day increase)	1.2 (1.1–1.3)	0.008

An increasing COI score indicates greater neighborhood opportunity whereas an increasing ADI score indicates greater deprivation. Each 1-point increase in the COI or ADI and 1-day increase in time to surgery is associated with the respective increase (if > 1.0) or decrease (if < 1.0) in odds of a concomitant meniscectomy. Private insurance was the reference group to which public insurance was compared. Separate logistic regression analyses were conducted, first with the COI, then with the ADI.

Presence of a Concomitant Chondral Injury

When controlling for insurance and time to ACLR in multivariable analyses, higher COI values were associated with decreased odds of the presence of chondral injury (OR 0.99 [95% CI 0.98 to 0.99]; p = 0.03), but the ADI was not (Table 5).

Table 5.

Factors associated with concomitant chondral injury

Variable	OR (95% CI)	p value
Regression model with COI
COI	0.99 (0.98–0.99)	0.03
Public insurance	1.4 (0.9–2.4)	0.1
Time to surgery (for each 90-day increase)	1.1 (1.0–1.2)	0.2
Regression model with ADI
ADI	1.0 (0.9–1.01)	0.2
Public insurance	1.6 (0.9–2.7)	0.06
Time to surgery (for each 90-day increase)	1.1 (0.8–1.2)	0.2

An increasing COI score indicates greater neighborhood opportunity whereas an increasing ADI score indicates greater deprivation. Each 1-point increase in the COI or ADI and 1-day increase in time to surgery is associated with the respective increase (if > 1.0) or decrease (if < 1.0) in odds of a concomitant chondral injury. Private insurance was the reference group to which public insurance was compared. Separate logistic regression analyses were conducted, first with the COI, then with the ADI.

Discussion

Previous studies have established that neighborhood conditions influence access to care and clinical outcomes in orthopaedic patients [8, 10, 19, 25, 29]. A single neighborhood-level measure, often the COI or ADI, is used in most studies. However, to our knowledge, no research has directly compared the COI and ADI in a clinical setting, especially in orthopaedics or sports medicine. Given the steady increase in the volume of disparities research in orthopaedic surgery and sports medicine [1, 11, 44], a refined understanding of these metrics may result in a more accurate interpretation of existing studies and an improvement of future study design. The present study directly compares the COI and ADI and examines their independent relationships with previously established disparities in pediatric ACLR. We found that the COI and ADI correlate moderately with each other. Even when controlling for covariates like insurance, higher COI values (increased neighborhood opportunity) were associated with decreased time to ACLR and decreased odds of the presence of a meniscus tear, meniscectomy, and presence of chondral injury. Meanwhile, the ADI was only found to be associated with surgical timing (but not surgery beyond 180 days after injury). This suggests that the COI may be more appropriate for disparities research in pediatric sports medicine, especially if investigators are interested in clinical outcomes. Understanding the relative strengths and weaknesses of the COI and ADI is valuable not only for interpreting previous pediatric disparities research but also for the selection of the most appropriate indices for future research and intervention implementation across a variety of clinical contexts. For example, when designing community-based research studies to address disparities in pediatric ACL care, our team now relies on the COI rather than the ADI as one of several factors to determine which neighborhoods to engage with. The results of this study may also prompt investigators in other specialties to consider more deeply the comparative characteristics of these and similar metrics to determine the most appropriate one for their patient populations and research questions.

Limitations

This study has several limitations. Given its retrospective design, selection bias is a potential concern as patients with greater socioeconomic resources may have more ability to return for follow-up, undergo surgical treatment, and thus be better captured in the electronic medical record. Conversely, patients within areas of greater deprivation may have faced unmeasured barriers that influenced their inclusion in this study. Additionally, this is a study conducted at a large, urban, single tertiary center. These findings may be most applicable to other large, health institutions in similar diverse metropolitan areas, and thus, its generalizability to other areas with differing geographic or socioeconomic composition may be limited. However, as the primary goal of this study was to compare the COI and ADI, the relationship of these metrics to each other is unlikely to change substantially. The COI 3.0 included 44 indicators, and the COI 1.0 and 2.0 comparatively used 29 indicators. These methodological differences across the COI versions could potentially introduce variability in our results, given that the COI 1.0 and 2.0 were used for most patients included in the study. However, the developers of the COI suggest that scores are comparable across the different iterations [9]. The methodology behind the ADI score calculations and composition also changed across the different iterations, possibly introducing similar variation. Additionally, the two indices may have different SDs, which may affect the interpretation of our results. However, we evaluated each metric in separate regression analyses. This study was limited to national-level continuous COI and ADI scores due to the lack of state- and metro-level ADI data. Still, this is offset, at least to some degree, by the sample size and use of the COI and ADI as continuous variables rather than using categorical values. Although numerous other neighborhood-level metrics exist, we selected the COI and ADI for this study because they are two of the most commonly used indices in pediatric orthopaedic research. As many of the reports on disparities in ACLR focus on the perioperative period [7, 8, 31, 33, 36, 37], this study had the same scope. Further research is needed to understand the associations between the COI and ADI with long-term clinical and patient-reported outcomes. Furthermore, even though this study describes several disparities, we were unable to determine the mechanisms underlying them. Additional qualitative research is necessary to investigate this. Notably, this study only examined the COI and ADI in the context of pediatric ACL injuries. This limits our ability to apply the conclusions to other pediatric orthopaedic conditions.

Correlation Between the COI and ADI

Although both the COI and ADI aim to quantify SDOH at the neighborhood-level, their moderate correlation indicates that they may capture meaningfully different aspects of social determinants. This could be explained by the methodological differences between the ADI and the COI. The most recent iteration of the COI incorporates 44 indicators, all of which were designed specifically to quantify opportunities that are crucial for child development and flourishment. In contrast, the ADI was primarily created to quantify economic disadvantage or deprivation using 17 indicators. It does not account for other factors like the COI does, such as pollution, healthcare resources, healthy environments, and other socioeconomic resources (Supplemental Tables 2 and 3; http://links.lww.com/CORR/B502). Although the ADI is based on a more granular geographic level (census block, 600 to 3000 people) than the COI (census tract, 1200 to 4000 people), this increased granularity did not improve its ability to identify disparities in pediatric ALCR care. Although this could vary in other contexts, the present study suggests that the COI may be sufficiently specific to neighborhoods for disparities research in pediatric sports medicine. Furthermore, the COI uses z-score transformation to standardize indicators measured on different scales and avoids disproportionate influence from certain variables and allows for scores to be compared over time [9]. Because of its rank-based scoring, the ADI does not possess these qualities. Finally, the COI provides data at the metropolitan, state, and national levels, offering flexibility in how neighborhoods are contextualized. For example, census tract 198.0, located in Harlem, NY, USA, is classified as a “moderate” COI score on the metropolitan level. Meanwhile, it is classified as a “very high” score on the national level. This difference could impact the results of research that is only able to utilize national-level ADI data.

Prior studies have also reported similarly modest correlation between multivariate indices such as the ADI, the Neighborhood Stress Score, and the Social Vulnerability Index. They highlight that each index reflects a different aspect of disadvantage rather than a single construct, even when attempting to quantify the same elements [4, 24]. Our findings support these conclusions, which have practical implications. Researchers and hospitals can use neighborhood-level indices to identify at-risk populations, guide community resource allocation, and design equityfocused interventions. Utilization of an inadequate metric can lead to under-recognition of clinically relevant disparities. In fact, this may even worsen health disparities if efforts are directed away from neighborhoods that were overlooked by use of an inappropriate metric. Our findings reiterate the importance of choosing an appropriate index based on its inherent characteristics as well as those of the research question and patient population of interest.

Comparison of the COI and ADI in the Context of Previously Established Pediatric Orthopaedic Disparities

Our study found that the COI and ADI had a similar magnitude of association with surgery beyond 60 and 90 days after injury, but only the COI remained associated with delays beyond 180 days. The COI was associated with the presence of a concomitant meniscus tear, meniscectomy, and chondral injury, whereas the ADI was not associated with any of these findings. This suggests that the COI may better capture the additional barriers to access that contribute to more prolonged surgical delays and more severe pathology, even when analyzing a smaller subset of outliers. Prior studies utilizing the COI reported that patients from low-COI neighborhoods experienced greater delays to surgery and a higher frequency of concomitant meniscus tears [36, 47]. However, neither found the COI score to be associated with the meniscectomy, possibly because they were underpowered to do so. Another study found no association between COI and time to orthopaedic evaluation [37]. However, these findings were limited by their homogenously low-COI study population that effectively resulted in a lack of a comparison group. The ability of the COI to capture disparities in higher-risk contexts has also been noted in nonorthopaedic research. A meta-analysis of 61 studies using the COI found that high-risk hospitalizations (medical, surgical/trauma, and longer-than-30-day hospitalizations) were associated with a low COI but not overall hospitalization [41]. These findings further suggest that the COI may be well suited to capture disparities in higher-acuity contexts such as concomitant surgical pathology. Multiple studies have reported that higher ADI scores are associated with longer times to surgery, but not with the presence of concomitant injuries or the performance of additional meniscal procedures in pediatric patients [8, 21, 33]. Further research is needed to elucidate the mechanisms behind the disparities identified in our research. Accurately identifying high-risk patients as well as the underlying etiology of the disparities they face can help clinicians and hospital systems anticipate patient care needs, target interventions, and allocate resources.

The results of this study can be used to guide future research design and intervention implementation. As the initial step in health equity research is to identify a disparity, utilization of the most appropriate neighborhood-level metrics will ensure that these are detected. Neighborhood-level disparities, more of which were detected in relation to the COI than the ADI, may benefit from neighborhood and multilevel interventions. A critical first step is forming relationships with community constituents to directly understand the underlying etiologies of these disparities. This requires qualitative research approaches (such as interviews or focus groups). Such data, in addition to genuine community relationships, can lead to the codesigning and implementation of interventions. The present study suggests that in the context of pediatric ACL injury, the COI may be a more appropriate measure than the ADI to identify neighborhoods and patients most at risk and thus more likely to benefit from further research and targeted interventions.

Conclusion

This study of 734 children undergoing ACLR found that the COI and ADI correlate moderately with each other. As COI scores increased (indicating greater neighborhood opportunity), the time to surgery and odds of concomitant intraarticular pathology decreased. As ADI scores increased (indicating greater neighborhood deprivation), time to surgery generally increased. However, the ADI was not associated with severity of intraarticular pathology or with surgical delays greater than 180 days. This study calls attention to possible differences in various geography-based SDOH metrics. Investigators and health systems should consider potential differences when designing research studies and interventions. Specifically, the composition and methodology of each metric should be evaluated critically in the context of the research question or patient population of interest. For example, based on the results of this study, we used COI scores as one of several criteria to identify the most appropriate neighborhoods to engage with for community-based research on pediatric ACL disparities research. Using the most appropriate metric will improve the quality of disparities research and the identification of at-risk populations and will better inform targeted community outreach programs.