Femoral head collapse after hip intra-articular corticosteroid injection: An institutional response to improve practice and increase patient safety

Introduction Femoral head collapse (FHC) is a rarely reported complication of hip intra-articular corticosteroid injection (IACSI). Upon observing a high rate of FHC after hip IACSI, we sought to (1) describe how we addressed the problem through a systematic evaluation of clinical data and institutional care practices followed by process improvement efforts; and (2) report our results. Methods Patients receiving hip IACSI during a 27-month period underwent retrospective review to determine the rate of FHC and to identify associated patient factors or practice shortfalls. Findings led to institution-wide interventions: (1) to improve patient/provider awareness of this association; and (2) to develop/implement practice guidelines. Rates of FHC after hip IACSI and practice patterns among providers before and after intervention were compared. Results Initial FHC rate after hip IACSI was 20.4%. Patient-related factors included body mass index (p = 0.025), history of cancer therapy (p = 0.012), Vitamin D level (p = 0.030), and multiple injections (p = 0.004). Volume/dose of injectate and post-injection surveillance methods varied widely. Quality improvement (QI) intervention resulted in fewer treatment referrals (from 851 to 436), fewer repeat injections (mean = 1.61 to 1.37; p = 0.0006), and a 5% lower FHC rate (p = 0.1292). Variation in practice patterns persisted, so a systems-based Clinical Pathway was established. Discussion When a high rate of FHC after hip IACSI was found to be associated with certain patient and practice factors, introduction of education materials and treatment guidelines decreased number of referrals, number of injections per patient, and FHC rate. In the absence of the systems-based Pathway, the type, dose, and volume of injectate and post-procedure follow-up remained variable.


Introduction
Femoral head collapse (FHC) is a rarely reported complication of hip intra-articular corticosteroid injection (IACSI). Upon observing a high rate of FHC after hip IACSI, we sought to (1) describe how we addressed the problem through a systematic evaluation of clinical data and institutional care practices followed by process improvement efforts; and (2) report our results.

Methods
Patients receiving hip IACSI during a 27-month period underwent retrospective review to determine the rate of FHC and to identify associated patient factors or practice shortfalls. Findings led to institution-wide interventions: (1) to improve patient/provider awareness of this association; and (2) to develop/implement practice guidelines. Rates of FHC after hip IACSI and practice patterns among providers before and after intervention were compared.
From 2015 to 2017, we observed several patients with hip OA treated with IACSI who developed FHC. Given limited information on risk factors for this adverse outcome and lack of best-practices for prevention, a quality improvement (QI) investigation was undertaken to address the problem. Initial correlation with patient-related factors led to an institution-wide campaign to educate patients and providers. Discovery of variation in procedural and periprocedural practices led to the development and introduction of treatment guidelines for providers.
We conducted a retrospective study of patients undergoing hip IACSI in our medical center before and after QI intervention. Goals of this study were (1) to identify patient risk factors or practice shortfalls associated with FHC after IACSI; and (2) to assess the impact of a QI intervention on institutional care practices and patient outcomes.

Materials and methods
This was a single center, before/after quasi-experimental study designed to bring about immediate quality improvements in health delivery in our institution. The Institutional Review Board provided exemption for this investigation and there was no external funding.

Quality improvement inception
After witnessing multiple cases of FHC after hip IACSI, we reviewed and compared our experience and institutional care practices with evidence-based literature. Although evidence supported IACSI for treatment of hip OA, no strong evidence of association with FHC existed. FHC was not acknowledged as a potential complication by the American Academy of Orthopaedic Surgeons (AAOS) appropriate use criteria [1]. There was no consensus as to best/safest injectate, and limited/no consensus on post-injection follow-up practices. A QI investigation was undertaken to inform risk management and practice improvement. obstructive pulmonary disease, prior hip trauma, HIV/AIDS, hip IACSI prior to study period, and number of hip IACSIs during the study period. If obtained during the study period, hemoglobin A1c, albumin, and Vitamin D levels were tabulated.
Anterior-posterior pelvis and/or hip radiographs were reviewed using the Picture Archiving and Communication System (PACS) (Visage version 7.1, Richmond, Victoria, Australia). All patients studied had pre-injection radiographs. If a patient did not have a pelvis or hip radiograph available at least one-month post-injection, the patient was excluded from FHC analysis, but was included in analysis of post-injection radiographic follow-up. All radiographs were obtained using our institution's standardized Radiology department protocol. No images were excluded for poor quality. Timing of pre-and post-injection radiographs was non-standardized. Each pre-and post-injection radiograph was compared side-by-side for interval changes. Post-injection FHC was defined radiographically as new loss of femoral head sphericity and was measured by comparing the contour of the weightbearing portion of the femoral head on each radiograph. For each hip image, a circle tool was used to outline the contour of the subchondral bone-edge in the weightbearing portion of the femoral head. Images were initially compared and interpreted by two orthopedic surgery residents and later independently verified by an attending orthopedic staff. In all cases, FHC was dramatic, easily detected, and the interim change in femoral head sphericity secondary to collapse was confirmed by official radiology reports (Fig 1).

Quality improvement intervention
After the initial investigation, a workgroup was formed comprised of stakeholders from Orthopedics, Radiology, Physical Medicine/Rehabilitation, and Pharmacy. The work-group reviewed preliminary institutional data and evidence-based literature to develop treatment guidelines addressing: 1) indications/contraindications; 2) medications; 3) injection technique; and 4) post-injection surveillance (Fig 2). Education materials were also created (Fig 3). Next, a high-profile, well-coordinated, institution-wide, campaign to raise awareness of work-group findings and recommendations was undertaken to educate providers in Radiology, Orthopedics, Rheumatology, Physical Medicine/Rehabilitation, and Primary Care. Guidelines and patient-education materials were then distributed hospital-wide for their use in clinical practice. Work-group findings and practice recommendations were also highlighted at our institution's annual research symposium. Over the next 27 months, the number of hip IACSI referrals, number of injections per hip, FHC rate, and post-injection radiograph rate were tabulated and compared to patients treated pre-intervention.

Statistical analysis
Descriptive statistics were used to summarize patient characteristics (Table 1). To examine bivariate relationships between modifiable and non-modifiable patient-predictors and FHC, comparisons between patients with and without FHC were made. Pearson's Chi-Square test was used for discrete variables, while two sample t-test was used for continuous variables. Statistical analysis was performed using SAS version 9.4 (Cary, North Carolina). P-values were considered statistically significant if < 0.05.     Table 2).

Initial quality improvement findings
While the rate of FHC after IACSI for patients excluded from our pre-intervention analysis is unknown, a best-case scenario would be that none developed collapse during the study period based on lack of need for radiographic follow-up. If this were the case, the initial FHC rate would be 15.8% (84/531).
Practice-related factors. Specialties that referred patients for hip IACSI included Orthopedics, Rheumatology, Physical Medicine/Rehabilitation, and Primary Care. Procedure-specific education materials did not exist. All hips were injected under fluoroscopy with 0.5% ropivacaine hydrochloride (range 3.0-6.0 mL), triamcinolone acetonide (range 1-2 mL; 40 mg/mL), and iopromide contrast (1 mL). Medication volume/dose were at discretion of the radiologist. No significant association was identified between FHC and local anesthetic or corticosteroid dosage, volume, or medication type. Post-procedure, patient follow-up with the referring provider varied. In most cases, decision for follow-up was left to the patient. Post-IACSI radiographs were ordered/obtained for only 77.6% of patients. Patients frequently underwent repeat injections without interval office visits or new radiographs.

Quality improvement intervention impact
In the 27 months after QI investigation and implementation of practice recommendations (Fig 2), reduction in number of procedures was 48.8% (851 to 436), reduction in number of hips injected was 39.6% (531 to 321), number of injections per hip reduced from 1.61 to 1.37 (p = 0.0006), reduction in post-IACSI FHC was 67.9% (84 to 27), and rate of FHC decreased

Development of a clinical pathway for hip IACSI
During the investigation/intervention, the work-group was developing a systems-based Clinical Pathway for hip IACSI (Fig 4) to expand on the initial informal recommendations (Fig 2). Based on lessons-learned, the work-group created a flowchart to guide peri-procedural care. Through a series of meetings and reviews, this flowchart was revised to incorporate as many evidence-based care practices as possible. When evidence-based practices were not available, practices were adopted on the basis of community standards and/or consensus among workgroup members. In addition to mapping out the sequence of patient care, the Pathway incorporated use of patient-education materials, standardized physician order-sets for medications and follow-up, and pharmacy involvement/oversight. Significance was set at p < 0.05. � Results are reported as the mean and standard deviation for continuous variables and the count and percentage for discrete variables.
�� HIV/AIDS was not a significant variable in our analysis and was omitted from final data reporting in order to protect sensitive patient information. https://doi.org/10.1371/journal.pone.0259242.t002

Discussion
FHC after hip IACSI has been described in case reports and small case series. Suggested etiologies include toxicity from injectate, volume-related increased intracapsular pressure restricting femoral head blood supply, and occult subchondral fracture or AVN progressing to collapse with medication-related inhibition of joint pain [2][3][4][5][6][7]. While no cause-effect relationship is known, the association between such a common treatment and such a pathogenic process is concerning.
At the inception of this investigation, we had already managed several patients with FHC occurring shortly after hip IACSI. Cases were characterized by acute increase in hip pain and disability, radiographically-confirmed articular collapse, and adjustment in care plan to allow earliest possible total hip arthroplasty (THA). A small number of patients with FHC less than 3 months post-IACSI required delay in THA to decrease risk of prosthetic joint infection [8]. Consequently, their care was further complicated by an increase in sedentary behavior and presentation for surgery in a deconditioned state (Fig 5).
Initial QI analysis of FHC after IACSI suggested certain patient factors were associated (elevated BMI, history of cancer therapy, low Vitamin D level). Our analysis also revealed lack of consistency in volume/dose of injectate, time-interval between treatments, and post-injection follow-up. The practice inconsistencies permitted many patients to receive hip injections "on demand", often without pre-injection clinical or radiographic evaluation. As our analysis also suggested that multiple IACSIs within a narrow time-frame (27 months) were associated with FHC, we recognized the need for improved organization/oversight.
To improve safety and consistency of hip IACSI, we created procedure-specific education materials. We also developed guidelines for procedural and peri-procedural practices and distributed to all referring and treating providers. Importantly, given the urgent need to improve patient safety, our initial intervention relied entirely on individual providers' awareness and responsibility. The systems-based Clinical Pathway was in development (Fig 4), but not yet in place.
In the two years that followed the initial intervention, hip IACSIs decreased (approximately half as many) along with fewer repeat treatments. While it is not certain that these referral patterns resulted directly from providers following our recommendations, the rate of FHC after IACSI decreased by 5%. However, in the absence of the systems-based Clinical Pathway, volume/dose of injectate and post-injection follow-up remained variable.

Patient factors
Obesity. Patients who developed FHC after IACSI had higher mean BMIs (33.00 kg/m 2 vs. 31.09 kg/m 2 , p = 0.025). Only one other study has analyzed BMI as a variable for FHC, and no association was noted [5]. Yet, obesity is associated with AVN and subchondral insufficiency fracture [9][10][11]. Such conditions may account for some cases of FHC in our analysis. Ultimately, as obese patients have increased risk for hip conditions characterized by unhealthy bone, treatments that decrease joint pain and may promote increased joint stress (e.g. IACSI) may also increase risk for FHC. For our obese patients, the decision for IACSI is always informed by complete clinical history and recent hip imaging.
History of treatment for malignancy. Patients developing FHC after IACSI had a higher percentage history of cancer therapy (14.3% vs. 6.1%, p = 0.012). Radiation and chemotherapy are linked to bone damage/loss and increased fracture risk [12,13]. This association has important implications for palliative care of hip pain. For patients undergoing cancer therapy, THA may be contraindicated and hip IACSI preferred for symptomatic relief. We conclude that such patients should be selected carefully and followed closely if IACSI performed.
Low serum Vitamin D levels. Patients developing FHC after IACSI had lower serum Vitamin D levels. However, our finding must be interpreted with caution as these levels were not tested at standardized intervals around the time of each hip procedure. Furthermore, the role of Vitamin D levels in the evaluation and treatment of orthopaedic patients remains controversial, and supplementation therapy has not been found to prevent osteoporosis-related fractures [14,15]. Ultimately, while the association is logical, no firm conclusion can be drawn from our analysis. Avascular necrosis. The natural history of avascular necrosis of the femoral head is FHC. In our analysis there were only 8 hips (1.51%) that had a diagnosis of AVN (7 ultimately included in analysis). Of the patients with AVN there was a 28.57% FHC rate (2/7) after IACSI. While there is a possibility AVN was underdiagnosed in our patient population, we were limited by the retrospective nature of the review and the information contained in the medical record at the time of retrieval. Due to the slightly higher FHC rate seen with this diagnosis and the natural history of the disease, we consider AVN as a relative contraindication for hip IACSI.

Practice-factors
Volume/dose of medications injected. Initial investigation demonstrated variation in dose of triamcinolone acetonide (range, 40-80 mg/mL), volume of ropivacaine hydrochloride (range 3.0-6.0 mL), and total volume of injectate (range, 5-9 mL). We endeavored to research and recommend the best/safest medication mixture for our patients. This was challenging as there is no consensus as to what constitutes "best practice" for hip injectate.
Final recommendations for medications were derived through collaboration with all clinical stakeholders. However, at the time of this study these recommendations were not yet part of the formal Clinical Pathway with standardized order-sets and pharmacy involvement/oversight. Despite efforts to encourage more consistent practice, without the system-based Pathway yet in place, the type, dose, and volume of medications injected remained highly variable.

Patients receiving multiple injections / lack of consistent follow-up after procedure
Patients treated with greater than one hip IACSI during the initial 27-month study period were more likely to develop FHC. While the number and frequency of hip IACSIs has never been reported as a risk factor for FHC, it is logical that repeated injections of medications with dose-and time-dependent toxicity in a narrow timeframe might increase risk of joint damage [16][17][18][19][20][21][22][23].
Initial investigation revealed repeat IACSIs were commonly ordered at the patient's request, oftentimes without pre-procedure clinical or radiographic evaluation. A small number of patients were undergoing regularly scheduled hip IACSIs (i.e. every three or six months). While all patients who underwent first-time hip IACSI had pre-procedure hip radiographs, there was no consistent practice for post-IACSI imaging or imaging between procedures. Initial findings were that post-injection radiographs were obtained for 77.6% of patients. Without strong evidence to inform the ideal frequency of hip IACSIs and/or best follow-up practices, work-group consensus was to adopt a very conservative approach. Specifically, prior to referral for initial or repeat IACSI, guidelines recommended in-person evaluation with new radiographs to assess the hip and to review/discuss treatment alternatives.
After QI intervention, we observed a significant decrease in number of repeat injections per patient, mean = 1.61 vs. 1.37, (p = 0.0006). However, variation persisted in follow-up practices and the number of patients with post-injection radiographs decreased to 55.8% from 77.6%. Many patients continued to receive one or more injections between hip radiographs. Again, this persistent variation is attributed to lack of the systems-based Pathway during the study period and reliance on individual providers to follow work-group guidelines.

Strengths
To our knowledge, our experience with identifying and managing FHC after hip IACSI is the largest series ever reported. Our investigation was performed at a single institution in a single health system. The electronic medical record and integrated PACS facilitated complete and consistent data collection. Our study demonstrates how a QI investigation and intervention may improve patient safety. Our study also demonstrates the formidable challenges associated with implementing sustainable institution-wide change.

Limitations
The primary study limitation is its retrospective observational design comparing nonrandomized groups. A randomized controlled trial would better define the relationship between hip IACSI and FHC and/or establish any causality related to patient, provider, or environment. Although our initial QI investigation identified potential patient-related risk factors for FHC after IACSI and made recommendations for/against treatment based on these, the impact of our QI intervention on number of patients referred for hip IACSI with potential risk factors was not tracked or tabulated. Also, the unique population studied (predominantly older males) may limit the generalizability of findings and recommendations. It is also possible that all or some of the FHC observed in this study occurred as part of the natural progression of patients' hip OA. However, to our knowledge, there is no natural history study that reports a similar or accepted rate of FHC in patients with hip OA.
Due to the retrospective nature of this work, the radiographic follow-up for pre-and postintervention study groups was non-standardized. Additionally, the time to chart/radiographic review was shorter for the post-intervention group, thereby introducing the potential for a falsely decreased FHC rate in the post-intervention group. While the shorter length of radiographic follow up for the post-intervention group and the possibility of a falsely low FHC rate in this group is a weakness of our study, it strengthens our rationale for instituting the Clinical Pathway to standardize hip IACSI practices so to allow for further prospective study.
Finally, a large number of patients (22.4% pre-intervention and 44.2% post-intervention) were excluded from final analysis of occurrence of FHC due to lack of post-injection imaging. However, even if all patients initially excluded due to incomplete radiographic follow-up were included in the final analysis with the assumption of "no collapse", FHC rates after IACSI remain high (i.e. pre-intervention = 15.8%; post-intervention = 8.4%). At our institution, even this "best case scenario" was sufficiently concerning to motivate further investigation and intervention.

Conclusion
When a QI investigation suggested FHC after hip IACSI was associated with certain patient and practice factors, we rapidly intervened with introduction of new education materials and treatment guidelines. This intervention decreased the number of hip injection referrals, number of repeat injections, and the rate of post-injection FHC. However, the intervention failed to improve the consistency of medications injected and post-procedure follow-up practices. The failure to change key practice patterns is attributed to the absence of the systems-based Clinical Pathway during the study period. Although the Pathway was in development during this time, progress was slow due to the need to involve multiple stakeholders in decision-making, time required to create standardized order-sets in the electronic medical record, and the lengthy process of gaining acceptance of a new workflow. As we were confronted with an immediate need to address a patient safety issue, our initial intervention that relied on patient/ provider awareness and responsibility was all that was possible.
As a next step, to further improve quality, safety, and consistency of the practice of hip IACSI at our institution, we have introduced a formal systems-based Clinical Pathway for hip IACSI. At a minimum, standardization of this care practice should simplify future risk management. Future research will show if this Pathway improves clinical outcomes and/or further decreases rates of FHC.