Influence of Psychiatric Comorbidities on Postoperative Outcomes Following Pediatric Medial Patellofemoral Ligament Reconstruction

Abstract

Purpose

To compare postoperative complications, rehabilitation outcomes, and health care utilization in children undergoing primary medial patellofemoral ligament reconstruction for patellofemoral instability with and without a pre‐existing mental illness.

Methods

The TriNetX U.S. Collaborative Network was queried to identify patients younger than 18 years old undergoing medial patellofemoral ligament reconstruction for patellar instability in the United States between 2000 and 2020. Patients were then separated into those with a pre‐existing mental illness (PMI Group) and those without (Control Group). After propensity score matching for patient demographics, chronic pain, and obesity, postoperative complication rates, rehabilitation outcomes, and health care utilization of both groups were compared across 1 year.

Results

After propensity score matching, each group consisted of 223 patients for a total cohort of 446 patients. The PMI Group showed significantly greater rates of postoperative knee pain (51.5% vs 41.0%, P = .004, risk ratios (RR): 1.40), opioid analgesic use (42.2% vs 31.4%, P = .018, RR: 1.34), nonopioid analgesic use (34.5% vs 22.0%, P = .006, RR: 1.27), emergency department visits (22.0% vs 4.9%, P < .001, RR: 4.46), and physical therapy treatment sessions (61.4% vs 51.1%, P = .028, RR: 1.20).

Conclusions

Children with pre‐existing mental illness exhibit significantly greater rates of persistent knee pain, higher opioid consumption, and increased health care utilization following medial patellofemoral ligament reconstruction.

Level of Evidence

Level III, retrospective comparative study.

Medial patellofemoral ligament (MPFL) reconstruction is a reliable treatment modality in the setting of recurrent chronic patellofemoral instability (PFI). ¹ , ² , ³ Although MPFL reconstruction generally yields favorable outcomes, complication rates reported in the literature show substantial heterogeneity across individual studies, with reported complications including patellar fracture, residual instability, redislocation, decreased knee flexion, and anterior knee pain. ¹ , ⁴

Pre‐existing mental illness (PMI) has shown considerable deleterious effects on postoperative outcomes following numerous knee procedures including anterior cruciate ligament reconstruction. ⁵ , ⁶ , ⁷ Similar trends of increased complication rates have been reported in patients with depression or anxiety undergoing skeletal trauma fixation. ⁸ In the setting of patellofemoral pathology, patients with PFI have been shown to show higher rates of anxiety and depression compared with age‐matched controls, along with greater pain sensitivity and reduced quality of life. ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ Although psychological factors have been increasingly recognized as important contributors to pain perception, function, and return‐to‐sport readiness after surgical treatment of PFI, ¹⁰ , ¹¹ , ¹⁵ the specific influence of PMI on postoperative recovery following MPFL reconstruction in children remains incompletely understood.

Children with anxiety and depression report more severe pain symptoms and higher rates of chronic pain compared with their peers, ¹⁶ , ¹⁷ and psychosocial factors such as baseline pain burden, depressive symptoms, and sleep disturbance have been shown to predict both acute and chronic postoperative pain following major musculoskeletal procedures in adolescents. ¹⁸ , ¹⁹ Depression is associated with reduced pain thresholds, whereas conditions such as post‐traumatic stress disorder may show paradoxical pain modulation, including both hypoalgesia and hyperalgesia. ¹⁶ , ²⁰ , ²¹ Given these developmental vulnerabilities and the paucity of pediatric orthopaedic data, ²² , ²³ the influence of PMI on postoperative recovery following MPFL reconstruction warrants focused investigation. Although the relationship between mental health and functional outcomes is well established in adult knee surgery populations, the high prevalence of mental health concerns during adolescence ¹³ , ²⁴ highlights a critical and understudied gap in the pediatric PFI literature. The purpose of this study was to compare postoperative complications, rehabilitation outcomes and health care utilization in children undergoing primary MPFL reconstruction for PFI with and without a PMI. We hypothesized that patients with a mental illness would experience a greater frequency of rehabilitation complications and health care utilization rates with comparable rates of postoperative complications.

METHODS

Data Source

This retrospective cohort study utilized the TriNetX U.S. Collaborative Network (TriNetX, LLC, Cambridge, MA) to investigate the influence of mental illness on postoperative outcomes following MPFL reconstruction for PFI in the pediatric patient population. The TriNetX US Collaborative Network aggregates deidentified electronic health record data of over 115 million patients from a geographically diverse group of 71 academic medical centers, community hospitals, and integrated health care systems across the United States. Contributing organizations primarily represent insured patient populations receiving care within participating health systems; uninsured or self‐pay encounters may therefore be underrepresented, which may limit generalizability. The database includes information regarding diagnoses, procedures, laboratory values, medications, and health care utilization. Data reporting is validated by TriNetX to promote consistency and reliability in coding and billing practices. TriNetX has been used extensively in orthopaedic research. ⁵ , ⁶ , ²⁵ , ²⁶ , ²⁷ International Classification of Diseases, 10th Revision (ICD10) and Current Procedural Terminology (CPT) codes were used to search the database for patients of interest. Data collection followed the Strengthening the Reporting of Observational studies in Epidemiology guidelines with additional consideration of the Reporting of Studies Conducted Using Routinely Collected Health Data statement for database‐specific reporting elements. Data provided by TriNetX undergo rigorous anonymization and is compliant with Health Insurance Portability and Accountability Act and applicable privacy regulations. As such, full Institutional Review Board approval was not required for this study. Methodology for this study was modeled after that as reported by Hinton et al. and Moore et al. ⁶ , ²⁸ with the additional consideration of multiple mental health diagnoses.

Cohort Creation

Inclusion criteria were defined as all patients under the age of 18 with a diagnosis of PFI (ICD10: M23.5, M22.0, M22.1, or S83.0) undergoing MPFL reconstruction (CPT: 27420, 27422, and 27427) with and without a concurrent pre‐existing diagnosis of any of the following mental illnesses: mood disorders (ICD10: F30‐F39), post‐traumatic stress disorder (ICD10: F43.1), or anxiety (ICD10: F41). Patients undergoing MPFL reconstruction across the United States between 2000 and 2020 were considered for inclusion. Patients were categorized into two groups: those with PMI (PMI group) and those without a PMI (control group).

Outcomes of Interest

The primary outcome were postoperative complications and rehabilitation outcomes within 1 year following MPFL reconstruction. Postoperative complications included postoperative pain, reoperation, subsequent patellar dislocation, residual patellar instability, surgical site infection, knee effusion, and knee aspiration. Rehabilitation outcomes included knee pain, knee stiffness, gait or mobility abnormalities, opioid and nonopioid analgesic use, lower extremity fracture, and intentional self‐harm. Health care utilization, including frequency of physical therapy (PT) evaluation/treatment sessions, emergency department visits, and inpatient hospital admissions was a secondary outcome. Patients lacking data for a given outcome were excluded only from analyses of that specific outcome and remained included in all other outcome analyses.

Demographic differences between the PMI and control groups were also investigated as seen in Table 1. A list of relevant ICD10 and CPT codes can be found in Table S1.

TABLE 1.

Cohort Demographics Before and After 1:1 Propensity Score Matching

Demographic	Unmatched Cohort				Matched Cohort
	PMI Group N = 235 (%)	Control N = 1158 (%)	P  Value	SMD	PMI Group N = 223 (%)	Control N = 223 (%)	P  Value	SMD
Age at Index Surgery (Mean ± SD)	14.6 ± 1.88	14.0 ± 2.31	<.001	0.274	14.5 ± 1.88	14.6 ± 1.71	.599	0.050
Sex
Male	60 (25.5)	428 (37.0)	<.001	0.249	60 (26.9)	57 (25.6)	.747	0.031
Female	175 (74.5)	730 (63.0)	<.001	0.249	163 (73.1)	166 (74.4)	.747	0.031
Race
White	191 (81.3)	798 (68.9)	<.001	0.289	180 (80.7)	183 (82.1)	.715	0.035
African American	26 (11.2)	126 (9.7)	.935	0.006	25 (10.9)	24 (8.3)	.880	0.010
Asian	12 (5.1)	41 (3.5)	.042	0.457	11 (4.9)	10 (4.5)	.880	0.010
Ethnicity
Hispanic or Latino	32 (13.6)	165 (14.2)	.800	0.018	30 (13.5)	22 (9.9)	.238	0.091
Not Hispanic or Latino	167 (71.1)	816 (70.4)	.855	0.013	159 (71.3)	168 (75.3)	.338	0.091
Obesity	54 (23.0)	95 (8.2)	<.001	0.416	43 (19.3)	34 (15.2)	.260	0.097
Chronic Pain	70 (29.8)	159 (13.7)	<.001	0.397	58 (26.0)	62 (27.8)	.669	0.041

Note: Bold indicates significance a P < .05.

PMI, pre‐existing mental illness; SD, standard deviation; SMD, standard mean difference.

Statistical Analysis

The TriNetX platform was utilized for all statistical analyses with all data presented as aggregate counts/means as applicable. In accordance with TriNetX privacy and data use policies, patient counts fewer than 10 were reported as ≤10 despite the use of deidentified data. The PMI and control groups were propensity score matched (PSM) on a 1:1 basis using a nearest‐neighbor matching algorithm with a caliper of 0.1. Covariates included patient demographics including age at index surgery, biological sex, race, and ethnicity, the presence of chronic pain, ¹⁹ and obesity. Sex was defined according to the designation recorded in the electronic health record. Race and ethnicity were defined using patient‐reported or administratively assigned categories available within the database and were analyzed as recorded. TriNetX uses Java, R, and Python scripts to compute statistical calculations, performing Mann‐Whitney U, chi‐squared, independent sample t‐test, and Shapiro‐Wilk testing when appropriate. Importantly, the built‐in comparisons for continuous variables rely on parametric methods, even after propensity score matching. Statistical comparisons are presented as risk ratios (RR), 95% confidence intervals (CI) and P values with statistical significance set as P < .05.

RESULTS

A total of 1393 patients met the inclusion criteria, 235 (16.9%) of which were categorized into the PMI group and 1158 (83.1%) into the control group. After propensity score matching, both groups consisted of 223 patients for a total matched cohort of 446 patients. Figure 1 outlines the cohort creation process. Demographic breakdown before and after 1:1 propensity score matching can be found in Table 1.

FIGURE 1.

Cohort creation flow diagram. (MPFL, medial patellofemoral ligament; PFI, patellofemoral instability; PSM, propensity score matched.)

One Year Surgical and Rehabilitation Outcome Comparison in the PSM Cohort

Table 2 reports the postoperative outcome comparison between the two groups at one year following MPFL reconstruction. The rates of postoperative complications were comparable between both groups; however, patients in the PMI group showed significantly greater rates of several rehabilitation complications. Namely, patients in the PMI group showed a 10.1% increased absolute risk of persistent knee pain (51.1% vs 41.0%, P = .004, RR: 1.40), 10.8% increased absolute risk of opioid analgesic use (42.2% vs 31.4%, P = .018, RR: 1.27), and 11.6% increased absolute risk of nonopioid analgesic use (34.5% vs 22.9%, P = .006, RR: 1.27). No significant differences were observed in the rates of knee stiffness, gait or mobility abnormality, or lower extremity fracture. Notably, zero patients in either group showed intentional self‐harm.

TABLE 2.

One‐Year Outcome Comparison in the Propensity Score Matched Cohort

Outcome	PMI Group N (%)	Control N (%)	P Value	Risk Ratio	95% CI
Surgical Complications
Acute Postoperative Pain	27 (12.1)	21 (9.4)	.359	1.29	(0.75, 2.21)
Reoperation	18 (8.1)	22 (9.9)	.507	0.82	(0.45, 1.48)
Subsequent Patellar Dislocation	73 (32.7)	69 (30.9)	.684	1.06	(0.81, 1.39)
Residual Patellar Instability	87 (39.0)	78 (35.0)	.378	1.12	(0.88, 1.42)
Surgical Site Infection	≤10 (4.5)	≤10 (4.5)	‐	‐	‐
Knee Effusion	44 (19.7)	38 (17.0)	.463	1.16	(0.78, 1.71)
Knee Aspiration	≤10 (4.5)	≤10 (4.5)	‐	‐	‐
Rehabilitation Complications
Knee Pain	114 (51.1)	91 (41.0)	.004	1.40	(0.11, 1.33)
Knee Stiffness	19 (8.5)	21 (9.4)	.740	0.91	(0.50, 1.64)
Gait or Mobility Abnormality	24 (10.8)	25 (11.2)	.880	0.96	(0.57, 1.63)
Opioid Analgesic Use	94 (42.2)	70 (31.4)	.018	1.34	(1.05, 1.72)
Nonopioid Analgesic Use	77 (34.5)	51 (22.9)	.006	1.27	(1.12, 2.04)
Lower Extremity Fracture	≤10 (4.5)	11 (4.9)	.823	0.97	(0.39, 2.10)
Intentional Self Harm	0 (0.0)	0 (0.0)	‐	‐	‐

Note: Bold indicates significance a P < .05.

CI, confidence intervals; PMI, pre‐existing mental illness.

One‐Year Health care Utilization Comparison in the PSM Cohort

Patients in the PMI group had a 17.1% higher absolute risk of at least one emergency department visit within 1 year postoperatively compared with controls (22.0% vs 4.9%, P < .001; RR, 4.46) (Table 3). In addition, patients in the PMI group had a higher mean number of emergency department visits than controls (1.67 vs 1.08, P = .028). Similarly, these patients had a 10.3% higher absolute risk of requiring at least one PT evaluation session (61.4% vs 51.1%, P = .028, RR: 1.20). However, the mean number of PT sessions did not differ significantly between groups (15.47 ± 11.7 vs 14.74 ± 14.5, P = .657). No inpatient hospital admissions were recorded for any patient in the propensity matched cohort.

TABLE 3.

One‐Year Health care Utilization Comparison in the Propensity Score Matched Cohort

Health care Encounter	PMI Group N (%)	Control N (%)	P Value	Risk Ratio	95% CI
Emergency Department
Number of Patients with ≥ 1 Visit	49 (22.0)	11 (4.9)	<.001	4.46	(2.38, 8.35)
Average Number of Visits (mean ± SD)	1.7 ± 0.9	1.1 ± 0.3	.028	‐	‐
Physical Therapy
Number of Patients with ≥ 1 Evaluation	137 (61.4)	114 (51.1)	.028	1.20	(1.02, 1.42)
Average Number of Sessions (mean ± SD)	15.5 ± 11.7	14.7 ± 14.5	.657	‐	‐
Inpatient Hospitalization
Number of Patients with ≥ 1 Admission	0 (0.0)	0 (0.0)	‐	‐	‐
Average Number of Admissions	‐	‐	‐	‐	‐

Note: Bold indicates significance a P < .05.

PMI, pre‐existing mental illness; SD, standard deviation.

DISCUSSION

In this propensity‐matched database study, children with PMI showed greater rates of persistent knee pain (51.1% vs 41.0%, RR: 1.40), higher rates of opioid (42.2% vs 31.1%, RR: 1.34) and nonopioid (34.5% vs 22.9%, RR: 1.27) prescriptions, and increased health care utilization following MPFL reconstruction compared with a matched control group without PMI. Notably, these differences occurred despite similar rates of mechanical complications such as acute postoperative pain, reoperation, redislocation, or stiffness.

The influence of PMI remains an important topic in pediatric care. Current figures suggest that approximately 15% of children aged 3 to 17 years have a diagnosis of anxiety or depression. ²⁹ Previous research has identified female sex, obesity, advanced age, prior tibial tubercle osteotomy, and psychiatric history as among several risk factors for prolonged opioid use across all age groups undergoing MPFL reconstruction. ³⁰ , ³¹ , ³² The present study identifies pediatric patients with PMI as an additional at‐risk population, showing greater utilization of both opioid and nonopioid analgesics compared with their age‐matched peers without PMI. Given the widespread use of opioids in orthopaedic practice, ³³ these data underscore the importance of tailoring perioperative pain management strategies to account for mental health comorbidities.

Although patients with PMI required greater analgesia and reported more persistent knee pain, these findings were not accompanied by higher rates of objective complications such as reoperation or redislocation. Although adult studies also link mood disorders to increased health care utilization and opioid use after knee surgery, associations with reoperation risk vary by procedure, including decreased rates following anterior cruciate ligament reconstruction and increased rates after isolated primary arthroscopic partial meniscectomy. ⁵ , ⁷ , ³⁴ In the present study, reoperation rates were comparable regardless of the presence of PMI, contrasting with pediatric‐specific literature in which Hinton et al. reported higher reoperation rates among adolescents with depression undergoing anterior cruciate ligament reconstruction. ⁶ A systematic review by Lawal et al. further showed that depression, anxiety, and mood disorders independently predict prolonged postoperative opioid use (between 90 and 180 days) across multiple surgical specialties. ³⁵ Our results extend these associations to the pediatric population, emphasizing the relevance of mental health in postoperative pain management to orthopaedic surgeons caring for pediatric patients.

The mechanisms behind these associations are likely multifactorial. In pediatric populations, neurodevelopmental differences in central pain processing, emotional regulation, and cognitive control may amplify the influence of psychiatric comorbidity on postoperative recovery. Prior work has shown that mood disorders are associated with altered central pain modulation through heightened nociceptive sensitivity and neuroinflammatory pathways. ³⁶ , ³⁷ , ³⁸ These effects may be more pronounced during adolescence, when neural circuits involved in affective pain processing are still maturing, potentially explaining the increased pain seen in the PMI group. Psychological factors such as fear avoidance, catastrophizing, and reduced resilience associated with anxiety and mood disorders ¹⁶ , ¹⁹ may also contribute to the functional limitation and decreased return to sport. ¹⁵ , ³⁹ Similarly, higher resilience, as measured by the Grit Scale, has been correlated with faster recovery of motion after pediatric knee surgery and improved patient‐reported outcomes. ¹⁵ , ⁴⁰ , ⁴¹ Therefore, it is possible that psychological rather than structural barriers are responsible for the increased need for PT treatment sessions observed in the PMI group. Although the PMI group had a higher likelihood of completing at least one PT evaluation (absolute risk difference 10.3%; RR: 1.20), the mean number of PT sessions was similar, suggesting that the difference may reflect earlier engagement with care rather than greater rehabilitation intensity.

The higher rate of emergency department visits among patients with PMI (absolute difference, 17.1%; RR, 4.46) reflects a substantial difference in postoperative health care utilization. This increased utilization may reflect lower symptom tolerance or a need for additional assessment and management of underlying mental health conditions rather than true surgical complications. Similar trends have been reported in pediatric fracture literature, where depression and anxiety predict greater unplanned health care utilization, ⁴² further reinforcing the need for proactive psychosocial assessment and expectation management in this patient population. Collectively, these data suggest that standard postoperative pain management pathways may not adequately address the unique needs of pediatric patients with mental health comorbidities.

Orthopaedic surgeons must remain aware of the complex interplay between mental and physical health in the pediatric patient population. ¹⁸ , ⁴³ It is possible that incorporating multimodal, opioid‐sparing analgesic regimens may compound on the benefits provided by offering behavioral health consultation or cognitive‐behavioral therapy for pediatric anxiety disorders. ⁴⁴ Additionally, routine preoperative screening using validated tools such as the Patient Health Questionnaire‐9 or Generalized Anxiety Disorder‐7 could identify at‐risk patients who would benefit from anticipatory counseling or closer follow‐up. ⁴⁵ , ⁴⁶ Prospective studies incorporating validated mental health screening tools are warranted to better quantify the impact of psychiatric illness on surgical outcomes.

The use of a large, multi‐institutional dataset provides a diverse, representative sample and allows for robust matching to minimize demographic confounding. The findings of this study have important implications for pediatric orthopaedic and sports medicine surgeons. Mental health comorbidities should be recognized as risk factors for greater postoperative pain and health care utilization, warranting individualized counseling, close follow‐up, and multidisciplinary management. Future studies incorporating prospective data and validated psychiatric scales would help clarify how symptom burden, treatment response, and resilience modulate surgical recovery.

Limitations

This study has several limitations. Reliance on administrative data introduces challenges inherent to coding‐based research. ICD‐10 and CPT codes lack sufficient granularity to definitively distinguish MPFL reconstruction from other extra‐articular knee ligament procedures and cannot reliably differentiate isolated MPFL reconstructions from combined procedures, such as tibial tubercle osteotomy or lateral release, ⁴⁷ which may influence rehabilitation protocols and postoperative pain trajectories. This introduces a potential misclassification consistent with prior reports. ²⁷ Psychiatric diagnoses were identified through ICD‐10 coding without granularity regarding severity, chronicity/duration of symptoms, or treatment status. As such, the PMI group encompasses a spectrum ranging from mild anxiety to major depressive or post‐traumatic stress disorders which likely influences the outcomes assessed in this study. Differences in medication adherence or therapy participation could also influence postoperative outcomes but were not quantifiable within the TriNetX platform. Sensitivity analyses evaluating the robustness of the propensity score–matching procedure were not performed, which may limit assessment of the stability of the matched results. Limitations in the statistical analysis prevent robust clinical interpretation of outcomes such as mean number of emergency department visits or PT sessions. Baseline equivalence of outcome measures could not be assessed, as TriNetX does not provide pre‐index measurements for postoperative outcomes, limiting the ability to evaluate preoperative differences between groups. Additionally, patient deidentification practices prevent disaggregation of data by sex, further limiting generalizability of the results discussed in this study. Similarly, we were unable to identify whether bilateral MPFL reconstruction procedures were performed on the same patient. Because TriNetX does not provide user‐level missingness metrics or support data imputation, outcomes were analyzed using outcome‐specific exclusions, limiting the ability to characterize or adjust for patterns of missing data. Finally, TriNetX lacks patient‐reported outcome measures, radiographic findings, perioperative management strategies, and detailed activity levels.

CONCLUSIONS

Pediatric patients with PMI exhibit significantly greater rates of persistent knee pain, higher opioid consumption, and increased health care utilization following MPFL reconstruction.

SUPPORTING INFORMATION

Additional supporting information can be found online in the Supporting Information section.

DISCLOSURES

The authors (H.A.A., P.L., S.T.E., A.M., K.A.W.) declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article.

Supporting information

Supplementary Material