Increased Risk of Osteomyelitis and Septic Arthritis in Osteogenesis Imperfecta

Abstract

Objective:

We investigated the prevalence of osteomyelitis and septic arthritis in individuals with osteogenesis imperfecta (OI) as compared to the general population using a large, national database.

Methods:

We performed a retrospective cohort study utilizing the TriNetX Analytics Network, a federated health research network that aggregates de-identified electronic health record data from over 78 million patients across the U.S. We queried patients with OI, based on encounter diagnoses. Patients in this group with any occurrence of osteomyelitis or septic arthritis were recorded. We also established a control cohort to compare the prevalence in patients without OI.

Results:

Out of 8,444 individuals with osteogenesis imperfecta, 433 (5.13%, 95% CI: 4.68%–5.62%) had encounter diagnoses for osteomyelitis and 61 (0.72%, 95% CI: 0.56%–0.93%) had encounter diagnoses for septic arthritis. In comparison, of the 79,176,436 patients without OI, 352,009 (0.44%, 95% CI: 0.43%–0.46%) had encounter diagnoses for osteomyelitis while 106,647 (0.13%, 95% CI: 0.13%–0.14 had encounter diagnoses for septic arthritis. The relative risk for osteomyelitis in patients with OI was 11.53 (95% CI: 10.52–12.64), while the relative risk for septic arthritis was 5.36 (95% CI: 4.18–6.89). The relative risk for osteomyelitis in pediatric patients with OI was 30.55 (95% CI: 24.35–38.28).

Conclusion:

Compared to individuals without OI, the relative risk of recorded encounter diagnoses for osteomyelitis and septic arthritis in patients with OI was over 11.5 and 5 times higher, respectively. To our knowledge, this is the first study investigating the diagnosis of musculoskeletal infections in patients with OI, as well as the first to report the overall prevalence in the general population. Clinicians may benefit from an especially high index of suspicion for septic arthritis and osteomyelitis in patients with OI and corresponding symptoms. Further study is warranted to investigate if modifications to conventional diagnostic pathways and criteria are of value in this population given the significant increased relative risk.

Level of Evidence:

Retrospective Cohort Study - Level II

Introduction

Osteogenesis imperfecta (OI) is a heterogeneous group of inherited connective tissue disorders with mutations that affect the structure or quantity of functional collagen, resulting in cardiac, metabolic, musculoskeletal, and other abnormalities¹. OI represents a range of phenotypes that are often classified by severity: type I (mild), type II (perinatally lethal), type III (severe), and type IV (moderate)¹. Compared to other rare inherited diseases, OI is relatively common, affecting 1 in 15,000–20,000 births¹. In addition to an increased prevalence of fractures, OI patients are also affected with low bone mass, long bone bowing deformities, short stature, scoliosis, joint laxity, and osteoarthritis^{1, 2}.

Trauma resulting in bone and joint damage is a potential risk factor for musculoskeletal (MSK) infections³. Animal studies exploring the pathogenesis of osteomyelitis have demonstrated that trauma may increase the risk of hematogenous seeding during periods of transient bacteremia⁴. Despite the increased risk of joint damage and ease of fractures from minor trauma in patients with OI, to our knowledge, no studies have investigated the prevalence of septic arthritis and osteomyelitis in these patients. We sought to investigate the prevalence of osteomyelitis and septic arthritis in individuals with OI compared to a control population. Based on the aforementioned literature, we hypothesized that OI would carry an increased prevalence of septic arthritis and osteomyelitis compared to the general population.

Materials and Methods

We utilized a retrospective cohort design and the TriNetX Analytics Network (US Collaborative Network), a federated health research network that aggregates the electronic health record (EHR) data of 79,184,880 patients across 49 health care organizations (HCOs) within the United States. The platform contains only de-identified data and has been deemed exempt from the Western Institutional Review Board (IRB) by a qualified expert as defined in Section §164.514(b)(1) of the HIPAA Privacy Rule. All data used to create primary and secondary cohorts was queried and pulled from this database on June 30, 2022 (Figure 1).

Figure 1.

Flow diagram for patient inclusion.

Two primary patient groups were identified:

• 1Osteogenesis imperfecta group, n = 8,444
• 2Individuals without osteogenesis imperfecta (control group), n = 79,176,436

Patients in the OI group were defined as any patient with at least one encounter diagnosis of OI (ICD-10-CM code: Q78.0 “Osteogenesis Imperfecta”) Patients in the control group were defined as patients without any encounter diagnosis of OI. Each of these primary groups were further broken into secondary groups based on patient history of MSK surgery and orthopedic implant. Patients with positive histories were identified and excluded from these secondary groups.

History of musculoskeletal surgery:

• 3Osteogenesis imperfecta and no musculoskeletal surgery, n = 5,914
• 4Individuals without OI and no musculoskeletal surgery, n = 73,872,152

History of MSK surgery was defined as at least one surgical procedure primarily involving the musculoskeletal system (CPT: 20100– 29999 “Surgical Procedures on the Musculoskeletal System”). 2,530 individuals in the primary OI group and 5,304,284 individuals in the primary control group had positive histories of musculoskeletal surgery and were excluded.

History of orthopedic implants:

• 3Osteogenesis imperfecta and no orthopedic implants, n = 7,896
• 4Individuals without OI and no orthopedic implants, n = 78,033,510

History of orthopedic implant was defined as the presence of at least one orthopedic implant (ICD-10-CM: Z96.6 “Presence of orthopedic joint implants” and/or Z96.7 “Presence of other bone and tendon implants”). 548 individuals in the primary OI group and 1,142,926 individuals in the primary control group had positive histories of orthopedic implant and were excluded.

Statistical Analysis

All statistical analysis was performed using GraphPad Prism version 9.4.0 for Mac (GraphPad Software, San Diego, CA, USA). Using a 95% confidence interval, the relative risk was calculated based on the number of individuals with an encounter diagnosis of osteomyelitis (ICD-10-CM: M86 “Osteomyelitis” and/or M46.2 “Osteomyelitis of vertebra”) in each primary group. In the same manner, relative risk calculations were done for the encounter diagnosis of septic arthritis (ICD-10-CM: M00 “Pyogenic arthritis”).

After excluding individuals with any electronic health record CPT codes indicating a history of MSK surgery from the OI and control groups, the relative risk calculations for osteomyelitis and septic arthritis encounter diagnoses were completed. Similarly, after excluding individuals with any electronic health record ICD-10-CM encounter diagnosis codes indicating a history of orthopedic implants from the OI and control groups, additional relative risk calculations for osteomyelitis and septic arthritis encounter diagnoses were completed.

For the primary OI and control groups, stratification by age at the encounter diagnosis of our defined outcome was completed. For each 5-year interval, the number of patients with an encounter diagnosis of osteomyelitis that occurred within the defined age range was recorded.

Additional analysis was completed to compare pediatric patients with OI to pediatric control patients as well as adult patients with OI to adult patients without OI. Data used to create the pediatric and adult cohorts was queried and pulled from the database on December 21, 2022. The TriNetX Analytics Network is periodically updated with new EHR data. So, the cohorts created to complete this additional analysis differ slightly from those used to complete our primary analysis (queried for on June 30, 2022). Relative risks of encounter diagnoses for musculoskeletal infections between these OI and control groups were calculated in the same fashion as previously described.

Results

Out of the 79,184,880 patients in the TriNetX US Collaborative Network on June 30^th, 2022, 8,444 patients (0.0107%, 95% CI: 0.0104%–0.0109%) from 49 health care organizations across the United States had an osteogenesis imperfecta encounter diagnosis in their medical history. In this OI cohort, 57% of patients were female while 43% were male. Ages ranged from 0 to over 90 while the average age was 35. In the control cohort, 54% were female while 46% were male. Ages ranged from 0 to over 90 while the average age was 46. To protect potentially sensitive patient information, the exact upper limit of the age range has not been fully detailed.

Osteomyelitis

Within the osteogenesis imperfecta cohort, 433 of 8,444 individuals (5.13%, 95% CI: 4.68%–5.62%) had an ICD-10-CM code in their medical history indicating osteomyelitis. Within the control group, 352,009 of 79,176,436 patients had an ICD-10-CM code in their medical history indicating osteomyelitis (0.44%, 95% CI: 0.43%–0.46%). The relative risk for an osteomyelitis encounter diagnosis between the OI and control groups was 11.53 (95% CI: 10.52–12.64), meaning that the risk for an osteomyelitis encounter diagnosis in patients with an osteogenesis imperfecta encounter diagnosis was about 11.53 times greater than in patients without an OI encounter diagnosis (Table 1).

Table 1.

Prevalence and relative risk of an osteomyelitis encounter diagnosis in patients with at least one osteogenesis imperfecta encounter diagnosis.

Population	Total Patients	Patients with Osteomyelitis Diagnosis	Proportion with Osteomyelitis Diagnosis (%) (95% CI)	Relative Risk (95% CI)
Patients with OI	8,444	433	5.13 (4.68–5.62)	11.53 (10.52–12.64)
Patients without OI	79,176,436	352,009	0.44 (0.43–0.46)	1 (reference group)
Patients with OI and without musculoskeletal surgery	5,914	261	4.41 (3.92–4.97)	16.14 (14.22–18.16)
Patients without OI and musculoskeletal surgery	73,872,152	202,053	0.27 (0.27–0.28)	1 (reference group)
Patients with OI and without orthopedic implant	7,896	376	4.76 (4.31–5.25)	11.85 (10.74–13.08)
Patients without OI and orthopedic implant	78,033,510	313,495	0.402 (0.400–0.403)	1 (reference group)

OI=osteogenesis imperfecta; CI= confidence interval

After excluding patients with any electronic health record codes indicating a history of MSK surgery, there were 5,914 patients in the OI group (i.e., 2,530 were excluded due to history of MSK surgery) and 73,872,152 patients in the control group (i.e., 5,304,284 were excluded due to history of MSK surgery). Within this modified OI group, 261 of 5,914 individuals had an encounter diagnosis code for osteomyelitis in their medical history (4.41%, 95% CI: 3.92%–4.97%). Within the control group, 202,053 of 73,872,152 individuals had an encounter diagnosis code for osteomyelitis in their medical history (0.27%, 95% CI: 0.27–0.28). The relative risk of an osteomyelitis encounter diagnosis was found to be 16.14 (95% CI: 14.33–18.16).

After excluding patients with any electronic health record codes indicating a history of at least one orthopedic implant, there were 7,896 patients in the OI group (i.e., 548 were excluded due to history of an orthopedic implant) and 78,033,510 patients in the control group (i.e., 1,142,926 were excluded due to history of an orthopedic implant). Within this modified OI group, 376 of 7,896 individuals had an encounter diagnosis code for osteomyelitis in their medical history (4.76%, 95% CI: 4.31%–5.25%). Within the control group, 313,495 of 78,033,510 individuals had an encounter diagnosis code for osteomyelitis in their medical history (0.402%, 95% CI: 0.400%–0.403%). The relative risk of osteomyelitis encounter diagnosis was found to be 11.85 (95% CI: 10.74–13.08).

When stratifying OI patients with osteomyelitis by age, the age groups in which there were the most patients with at least one encounter diagnosis were 0–4, 5–9, and 10–14 with 138, 78, and 51 patients, respectively (Table 2, Figure 2). The numbers of OI patients with an osteomyelitis encounter diagnosis during each of the remaining age ranges were between 1 to 26 patients (Table 2, Figure 2). In contrast, the control group saw a steady number of affected individuals in childhood and early adulthood as well as a gradual increase beginning in the 30–34 age range (Table 2, Figure 3). The highest number of control patients with osteomyelitis as an encounter diagnosis were in their 50s and 60s (Table 2, Figure 3).

Table 2.

Number of patients with an OI encounter diagnosis with at least one osteomyelitis encounter diagnosis during 5-year age windows until age 90.

Age at Diagnosis of Osteomyelitis	Osteogenesis Imperfecta	Control
0–4	138	7071
5–9	78	6284
10–14	51	8401
15–19	16	7480
20–24	13	6988
25–29	20	9747
30–34	15	13234
35–39	16	17197
40–44	13	22007
45–49	18	29160
50–54	25	38640
55–59	26	46133
60–64	19	46490
65–69	19	41604
70–74	18	32243
75–79	10*	23296
80–84	0	13957
85–89	0	4937
90**	0	76

^*The number of patients diagnosed between ages 75–79 was between 1 and 10, but the exact number was rounded to 10 to maintain confidentiality.

^**Individuals over 90 years old were excluded from the age stratification to maintain confidentiality.

Figure 2.

Number of patients with at least one OI encounter diagnosis and an osteomyelitis encounter diagnosis stratified into 5-year age windows by age at event of interest. *75–79 range is rounded to 10 but is truly between 1–10.

Figure 3.

Number of control patients with at least one osteomyelitis encounter diagnosis stratified into 5-year age windows by age at event of interest.

After excluding patients aged 18 or older (i.e. only including patients currently 17 or younger), there were 2,495 patients in the OI group and 13,915,156 in the control group on December 21, 2022. The relative risk between these pediatric cohorts was found to be 30.55 (95% CI: 24.35–38.28). After excluding patients aged 17 or younger (i.e. only including patients currently 18 or older) there were 6,536 patients in the OI group and 77,254,811 in the control group on the same date. In this cohort, only cases of osteomyelitis that were diagnosed in adulthood were considered. The relative risk between these adult cohorts was found to be 6.27 (95% CI: 5.42–7.25).

Septic Arthritis

There were 61 patients in the OI cohort who had an ICD-10-CM code in their medical history indicating septic arthritis (61 of 8,444, 0.72%, 95% CI: 0.56%–0.93%). In the control cohort, 106,647 of 79,176,436 patients had an ICD-10-CM code in their medical history indicating septic arthritis (0.13%, 95% CI: 0.13%–0.14%). The relative risk for a septic arthritis encounter diagnosis between the OI and control groups was 5.36 (95% CI: 4.18–6.89), meaning that the risk for a septic arthritis encounter diagnosis in patients with an osteogenesis imperfecta encounter diagnosis was about 5.36 times greater than in patients without an OI encounter diagnosis (Table 3).

Table 3.

Prevalence and relative risk of a septic arthritis encounter diagnosis in patients with at least one osteogenesis imperfecta encounter diagnosis.

Population	Total Patients	Patients with Septic Arthritis Diagnosis	Proportion with Septic Arthritis Diagnosis (%) (95% CI)	Relative Risk (95% CI)
Patients with OI	8,444	61	0.72 (0.56–0.93)	5.36 (4.18–6.89
Patients without OI	79,176,436	106,647	0.13 (0.13–0.14)	1 (reference group)
Patients with OI and without musculoskeletal surgery	5,914	19	0.32 (0.21–0.50)	4.84 (3.10–7.555)
Patients without OI and musculoskeletal surgery	73,872,152	49,039	0.066 (0066–0.067)	1 (reference group)
Patients with OI and without orthopedic implant	7,896	44	0.56 (0.42–0.75)	5.40 (4.03–7.24)
Patients without OI and orthopedic implant	78,033,510	80,491	0.103 (0.102–0.104)	1 (reference group)

OI=osteogenesis imperfecta; CI= confidence interval

After excluding patients with any electronic health record codes indicating a history of MSK surgery, there were 5,914 patients in the OI group and 73,872,152 patients in the control group. Within this modified OI group, 19 of 5,914 individuals had a history of septic arthritis (0.32%, 95% CI: 0.21%–0.50%). Within the control group, 49,039 of 73,872,152 individuals had a history of septic arthritis (0.066%, 95% CI: 0.066%–0.067%). The relative risk of septic arthritis was found to be 4.84 (95% CI: 3.10 to 7.55).

After excluding patients with any electronic health record codes indicating a history of orthopedic implants, there were 7,896 patients in the OI group and 78,033,510 patients in the control group. Within this modified OI group, 44 of 7,896 individuals had an encounter diagnosis code for septic arthritis in their medical history (0.56%, 95% CI: 0.42%–0.75%). Within the control group, 80,491 of 78,033,510 individuals had an encounter diagnosis code for septic arthritis in their medical history (0.103%, 95% CI: 0.102–0.104%). The relative risk of septic arthritis was found to be 5.40 (95% CI: 4.03 to 7.24).

We attempted to perform age stratification for septic arthritis in the same fashion we did for osteomyelitis but could not because almost every age range contained only 1–10 patients. To maintain confidentiality, each of these values is automatically rounded to 10 by the TriNetX database. For the same reason, it was impossible to calculate relative risks after separating populations into adult and pediatric.

Discussion

In this retrospective cohort study of 8,444 patients with a diagnosis of OI, the prevalence of osteomyelitis and septic arthritis diagnoses was 5.13% (95% CI: 4.68%–5.62%) and 0.72% (95% Ci: 0.56%–0.93%), respectively. Patients with OI were significantly more likely to be diagnosed with osteomyelitis (over 11.5 times) and septic arthritis (over 5 times) compared to those without OI. These findings support the hypothesis that OI patients are at a significantly higher risk for these infections. Previous literature lacked information on the prevalence or incidence of septic arthritis and osteomyelitis in OI patients, likely due to challenges in obtaining a sufficient sample size. This study, with the largest sample size to date, provides robust results that are likely representative of the national population.

Both osteomyelitis and septic arthritis are important to detect early because prompt antibiotic treatment with or without aspiration has been shown to significantly decrease the risk of mortality and complications like permanent deformities and chronic pain^{3, 5, 6}. Septic arthritis specifically is an orthopedic urgency that can rapidly progress to total joint destruction if not promptly and properly treated^{1, 3, 6}. Because individuals with OI are predisposed to a variety of conditions that in many ways mimic osteomyelitis or septic arthritis, it is possible that their symptoms could be misinterpreted more frequently than those of individuals in the general population, which could then delay treatment. Understanding that these patients are at a higher risk of being diagnosed with these infectious complications should increase clinicians’ index of suspicion, inform ordering of the appropriate tests to allow definitive diagnosis, and optimize time to management. These results warrant consideration for changes to existing treatment and diagnostic protocol in patients with OI, employing a higher sensitivity for joint aspiration, labs, and urgent advanced imaging than is utilized in control populations. While studies have demonstrated that the Kocher criteria and modified versions of the Kocher criteria are beneficial for distinguishing between septic arthritis and transient synovitis in the hip in the general population, no one has investigated the efficacy of these criteria in OI patients^{3, 7}. Even in normal populations, the Kocher criteria is not universally applicable to all joints, with 52% of cases missed when applied to the knee joint rather than the hip⁸. Similarly, it is possible that certain sub-populations of patients may require separate criteria for maximizing sensitivity to the diagnosis of septic arthritis. Further research is needed to understand whether changes to these criteria are necessary when treating OI patients.

The age ranges in which there were the most OI patients with an osteomyelitis encounter diagnosis were 0–4, 5–9, and 10–14. The numbers of OI patients with an osteomyelitis encounter diagnosis during the remainder of the age ranges were comparatively very low. In contrast, the control group saw the highest number of patients with encounter diagnoses in their 50s and 60s. This age distribution is in line with what has been found in the available literature as the prevalence of important risk factors for osteomyelitis like type 2 diabetes mellitus increases significantly in older adults ^{9, 10}. Furthermore, the relative risk of an osteomyelitis encounter diagnosis between the pediatric OI and control groups was over 30 while the relative risk of osteomyelitis encounter diagnosis between the adult OI and control groups was only about 6. These findings suggest that the increase in risk of osteomyelitis due to OI is likely more pronounced in the pediatric population.

The difference in age distribution as well as the significantly increased relative risk in the pediatric OI cohort also support our proposed mechanism. Given that most cases of hematogenous spread occur during childhood and that repeated minor bone trauma in OI patients should theoretically increase the opportunity for hematogenous seeding, most of the increased risk of osteomyelitis should be most evident during childhood. However, given the complexity and constantly evolving understanding of OI, it is possible these findings result from a different, less understood mechanism altogether. While this project has demonstrated a significant correlation, it is not equipped to fully elucidate the underlying mechanism. This highlights a potential focus for future research.

The proportion of patients in this database with an OI encounter diagnosis was found to be 0.0107% (95% CI: 0.0104%–0.0109%), which matches the existing literature. The proportion of patients in the control group with an osteomyelitis encounter diagnosis was 0.44 % (95% CI: 0.43%–0.46%). After a thorough literature search, there is no clear evidence-based published reporting of the prevalence of osteomyelitis in the general population. The National Organization of Rare Diseases reports that 2–5 out of every 10,000 individuals experience acute hematogenous osteomyelitis specifically. This is significantly lower than our findings, which is logically consistent given that our finding included all types of osteomyelitis. Given our large sample size, our finding on the prevalence of osteomyelitis in the control group (0.44%, CI 95%: 0.43%–0.46%) is likely the best available representation of the overall prevalence in the U.S. general population. After another thorough literature review, there similarly does not seem to be a widely accepted prevalence proportion for septic arthritis. While the incidence of septic arthritis has been investigated, we believe we are the first to elucidate the overall prevalence in such a large population.

This large population-based study based on aggregated electronic health record data has several limitations. The accuracy of diagnostic and charting practices among clinicians in the contributing EHRs is crucial for the validity of the study. Some mild cases of osteogenesis imperfecta (OI) may not be diagnosed until later in life, leading to potential underrepresentation in the study. To mitigate selection bias, all patients with any encounter diagnoses of osteomyelitis and septic arthritis were included in the OI group, even if the diagnosis occurred after the event of interest. However, there is still a possibility of undiagnosed OI patients with milder cases affecting the results. Additionally, pediatric patients may have diagnoses for more severe forms of OI, explaining their increased risk of diagnoses for osteomyelitis and septic arthritis. These infections may be linked to comorbidities such as scoliosis, chest wall deformities, and decreased pulmonary function associated with severe OI increase^{1, 2}. Prospective surveillance of comparable groups is required to provide the most compelling evidence that OI patients have an increased risk of developing MSK infections. Another potential bias comes from the difference in mortality and lifespan between our control and test groups. Patients with milder OI generally have normal lifespans, but the overall median life expectancy for OI patients is lower than the general population, possibly resulting in an underestimation of the risk increase^{1, 2}. Furthermore, the higher proportion of female patients in the OI test group compared to the control group may underestimate the risk of MSK infection diagnoses in OI patients, as male children are more prone to such infections^{11, 12, 13, 14}. Additionally, orthopedic interventions were controlled for in the analysis. The modified groups showed either an increased risk or no significant change, suggesting that the increased risk of MSK infections in OI patients cannot be solely attributed to surgical sequelae.

Conclusion

The proportion of patients with osteomyelitis or septic arthritis documented and registered in their medical records was over 11.5 and 5 times higher respectively in patients with OI versus the control population. To our knowledge, the present study is the first to investigate the prevalence of osteomyelitis or septic arthritis in patients with OI. Compared to other studies that investigated different complications of OI, the present study has the largest sample size, thus providing reliable evidence for prevalence in this population. Considering the higher risk in the OI group, criteria and parameters that apply to the general population may not be accurate in OI patients. While caring for patients with OI, it may potentially benefit clinicians to have an especially high index of suspicion for osteomyelitis and septic arthritis as well as a low threshold for commencing the appropriate workup such that indicated interventions can be delivered in a timely fashion. However, more research is needed to justify such changes in approach. We advocate for further multicenter studies to better investigate underlying mechanisms, specific risk factors, laboratory and imaging findings, and clinical findings associated with septic arthritis and osteomyelitis in this patient population.

Highlights.

• Patients with osteogenesis imperfecta are at a significantly increased risk for osteomyelitis and septic arthritis

• The prevalence of osteomyelitis in the general population is likely higher than previously reported