key: cord-1024026-2esde0ds
authors: Mont, Michael A.; Pivec, Robert; Banerjee, Samik; Issa, Kimona; Elmallah, Randa K.; Jones, Lynne C.
title: High-Dose Corticosteroid Use and Risk of Hip Osteonecrosis: Meta-Analysis and Systematic Literature Review
date: 2015-04-08
journal: J Arthroplasty
DOI: 10.1016/j.arth.2015.03.036
sha: a8a3cc7f31b73040096c07b52a382cf0b2b40cdc
doc_id: 1024026
cord_uid: 2esde0ds

The effect of varying corticosteroid regimens on hip osteonecrosis incidence remains unclear. We performed a meta-analysis and systematic literature review to determine osteonecrosis occurrences in patients taking corticosteroids at varying mean and cumulative doses and treatment durations, and whether medical diagnoses affected osteonecrosis incidence. Fifty-seven studies (23,561 patients) were reviewed. Regression analysis determined significance between corticosteroid usage and osteonecrosis incidence. Osteonecrosis incidence was 6.7% with corticosteroid treatment of >2 g (prednisone-equivalent). Systemic lupus erythematosus patients had positive correlations between dose and osteonecrosis incidence. Each 10 mg/d increase was associated with a 3.6% increase in osteonecrosis rate, and >20 mg/d resulted in a higher osteonecrosis incidence. Clinicians must be wary of osteonecrosis in patients on high corticosteroid regimens, particularly in systematic lupus erythematosus.

(1947-to-present); SCOPUS (1966-to-present) ; and Web-of-Science (1945-to-present Data were independently extracted and recorded by two authors (RP and SB) into spreadsheets (Excel; Microsoft Corporation, Redmond, Washington). For inconsistencies in numerical values, a third author (KI) reviewed manuscripts and corrected potential errors. Each study was evaluated sequentially. Two authors reviewed manuscripts and if inconsistencies were identified, these were clarified before the next manuscript was reviewed. Inconsistencies recognized were typographical in nature (incorrect number accidentally inputted) and had minimal impacts on reporting quality.

Extracted data included study level-of-evidences, patient demographics, medical diagnoses, corticosteroid types, time-to-diagnoses, mean corticosteroid doses, maximal daily doses, cumulative doses, and treatment durations. Studies that reported differing corticosteroid agents and doses were normalized to relative potencies in prednisoneequivalent doses in milligrams [13] . For analyses of osteonecrosis incidences between high-and low-dose corticosteroids, a 10,000 mg prednisone-equivalents cut-off was utilized because several studies reported doses at or below this level, or substantially above this level (e.g. N 15 g). Thus, this level represented cut-offs in published literature. A third investigator (KI) independently reviewed data accuracy and inter-reviewer consistencies to aid in standardization of pooled data.

The literature search yielded 372 articles between 1960 and 2011, of which twenty-one were review articles. Following assessment of abstracts, 319 in vitro studies on histological changes associated with corticosteroids or non-clinical data (e.g. review articles) were excluded, leaving fifty-three studies for review. Reference list examinations identified four additional reports, for a total of fifty-seven studies that were included in the systematic review. There were two level-I, seven level-II studies, and forty eight level-III studies (see Appendix 1) [10, 12, .

We evaluated the association of certain disease entities for which systemic corticosteroids were used with the development of osteonecrosis. Specifically, we assessed the association in patients with cardiac, liver, or renal transplants; myeloproliferative diseases (multiple myeloma, acute lymphoblastic leukemia); systemic lupus erythematosus (SLE); or severe acute respiratory syndrome (SARS) . No studies assessed corticosteroid effects in other potentially at-risk populations, such as alcohol-users or smokers.

For the meta-analysis, our review resulted in the exclusion of 50 studies due to incomplete outcome reporting, such as inadequate information on mean daily corticosteroid intake, cumulative doses, and duration of treatment, which prevented estimation of odds ratios (i.e. odds that an outcome occurs, given a particular exposure, compared to odds of the outcome occurring in absence of the exposure). Seven studies remained for the meta-analysis [19, 20, 23, 34, 52, 55, 57, 69] , and these consisted of 1 high-level prospective cohort study, as well as 5 case-control studies (Level III) and 1 case series (Level IV), from which odds ratios could be deduced (Fig. 1) . Studies not included in the meta-analysis were used for multivariate analyses to answer secondary questions. The seven studies included 1515 patients undergoing treatment with corticosteroids for SLE (n = 140), renal transplantation (n = 774), or bone marrow transplantation (n = 601; see Table 1 for further details). The mean age of patients in these studies was thirtythree years (range, 15 to 60 years). Studies were performed in the United States (n = 4), the United Kingdom (n = 1), Canada (n = 1), and Denmark (n = 1).

Validity assessment of RCTs was conducted independently by two authors (RP and SB) utilizing the Detsky scale cutoff score of 75% [70, 71] . Extracted data were pooled from studies meeting inclusion criteria to calculate treatment effects (odds ratio and 95% confidence intervals) and weights utilizing statistical software (Comprehensive Meta-Analysis v2; Biostat, Englewood, New Jersey) to calculate effects of corticosteroid treatments and doses on incidences. Linear regressions and correlation statistics determined osteonecrosis incidences relative to corticosteroid doses (mean, cumulative, and duration; refer to Appendix for breakdown of included studies for each diagnosis category). Regression analysis was performed to find the quantitative relationship and its significance between the variables, corticosteroid daily dose, cumulative dose, and duration of treatment, and the incidence of osteonecrosis as a dependent variable. Due to the lack of complete dosage data in several studies, we were only able to assess this in SLE and renal transplant patients. The reported R-and R-square results were un-adjusted and reported only for simple linear regressions, rather than for multiple linear regressions to avoid distortion by strong associations between average doses, cumulative doses, and durations. A random effects model was used in the meta-analysis, with publication bias assessed using Orwin's fail-safe N and Duval and Tweedies trim and fill statistics. Furthermore, heterogeneity was assessed using Cochrane's Q and I 2 statistics. Statistical analyses were performed utilizing SSPS 17.0 statistical software (IBM, Armonk, New York). Significance was defined as P values ≤ 0.05 (Appendix 2 Statistics).

There were no external sources of funding for this meta-analysis and systematic literature review.

The systematic review demonstrated overall osteonecrosis incidence of 6.7% (range, 0.3% to 52%) in patients taking high-dose corticosteroids. Two level I studies proved a significant positive correlation between cumulative dose and the incidence of osteonecrosis, whereas, five level II studies failed to show it.

Osteonecrosis incidence for SARS was 21.8%, SLE 15.7%, renal transplant 14.7%, and BMT 6.6% (Fig. 2) . Across all diagnoses, we observed positive associations between mean corticosteroid doses and osteonecrosis (Fig. 3) . This was irrespective of underlying disease, as analysis of variance of osteonecrosis incidence between patients with different medical diagnoses (SLE, severe acute respiratory syndrome, bone marrow transplantation, renal transplantation) demonstrated no differences between diagnostic categories (P = 0.16). The regression analysis demonstrated a significant positive correlation in SLE patients (r = 0.81; R 2 = 0.67; P b 0.05), however, this was not significant in renal transplant recipients (r = 0.32; R 2 = 0.09; P N 0.05). It was also noted that renal transplant recipients and SLE patients were more likely to develop osteonecrosis if they were younger than 35 years compared to those who were older (22 versus 13%; P = 0.04, and 33 versus 7%; P = 0.02, respectively).

Meta-analysis of osteonecrosis in patients treated with greater than 20 mg per day demonstrated significantly higher odds than less than 20 mg per day corticosteroid users (OR 9.1; 95% confidence interval, 4.6 to 19.8) (Fig. 4A ). For patients treated with high cumulative corticosteroid doses (greater than 10 g), the odds ratio for developing osteonecrosis was 2.4 (95% CI. 0.8 to 6.4), and lower dosing regimens were associated with a lower osteonecrosis incidence (OR 0.4; 95%, confidence interval 0.25 to 0.54) (Fig. 4B) . Additionally, we observed that 10 mg per day dose increases resulted in a 3.6% increase in the rate of osteonecrosis. Due to the lack of data and control groups, the metaanalysis could not be performed comparing other dosing regimens or treatment duration effects on osteonecrosis risks.

Cumulative doses and treatment durations had negative associations with incidence for both SLE and renal transplant patients. In SLE patients, cumulative dose and the duration of treatment showed negative trends (r = −0.85, R 2 = 0.65 and r = −0.53, R 2 = 0.29, respectively) with the incidence of osteonecrosis, but these were not significant (P N 0.05) and may not represent a true trend. In renal transplant recipients, there was no evidence of a significant correlation between cumulative doses and incidence of osteonecrosis (r = 0.31; R 2 = 0.42; P N 0.05).

Mean osteonecrosis incidence was 33% in twenty studies evaluating the effects of pulsed corticosteroids based on data from the systematic review.

Our aim was to evaluate the available literature and to assess the association between corticosteroids and hip osteonecrosis, utilizing statistical methodologies. In particular, we assessed the effect of dosing regimens and treatment durations, as well as the role of different disease entities. This study builds upon individual reports that have demonstrated independent risk factors for osteonecrosis, which were not observed in earlier studies [2, 20, 66, 72] . Our results showed that osteonecrosis incidences were affected by treatment with corticosteroids, corticosteroid doses, and patient age. Specifically, patients treated with high-dose corticosteroids may be up to ten times as likely to develop osteonecrosis, and cumulative doses greater than 10 g may increase the likelihood of developing osteonecrosis by two-fold, compared to cumulative doses less than 10 g. In addition, it was observed in the regression analysis that the correlation between corticosteroid dose and osteonecrosis incidence was most evident in SLE patients. However, no differences between diagnoses were noted using the analysis of variance, and further study is needed before a stronger conclusion can be drawn.

Multiple studies have demonstrated that corticosteroids are independent risk factors for osteonecrosis. Shibatani et al, in a study of 150 patients, noted a significant association between the total dose of corticosteroids and osteonecrosis incidence in patients during the first two months following renal transplantation (OR = 4, P = 0.02) [60] . Nakamura et al reported a 10.3 odds ratio of developing osteonecrosis in SLE patients, which compared similarly with the results of the present study (OR = 9.1) [73] . We also observed strong correlations (R 2 N 0.8) between mean daily corticosteroid doses, cumulative doses, and treatment durations and osteonecrosis incidences. However, no single factor predicted variability in osteonecrosis incidences, which pointed to possible synergistic effects between all three factors.

The underlying diagnoses may potentially affect the risk for developing osteonecrosis. However, it is unclear which plays the dominant role, the underlying disease or the effects of corticosteroids, which may have stronger negative synergistic effects for some disorders compared to others. A prospective magnetic resonance imaging (MRI) study by Shigemura et al demonstrated that SLE patients had significantly higher risk (RR 2.1) of osteonecrosis than non-SLE patients (37 versus 21%; P = 0.001) [66] . However, they excluded organ transplant recipients due to higher mortality rates, and relied primarily on other systemic inflammatory diseases (e.g. inflammatory bowel diseases, vasculitides, dermatologic autoimmune diseases) as comparison groups. Furthermore, Leiberman et al reported low incidences (3%) of osteonecrosis diagnosed with MRI at a mean of 31 months post-liver transplantation [47] . Presently, the true effect of underlying disease on osteonecrosis incidence remains to be determined, and additional prospective studies are needed.

In addition to differences in diagnosis, it is important to highlight the effects of demographic factors and race on the development of osteonecrosis. Although it is difficult to specify osteonecrosis rates in a whole population, some studies have evaluated osteonecrosis in Asian populations and have shown higher disease incidence, particularly in those who have corticosteroid-dependent conditions. For example, Yamaguchi et al demonstrated a rising trend in the incidence of nontraumatic ON in Japanese patients, particularly in those who had corticosteroid-dependent conditions, with rates of up to 44% in SLE patients who were on high-dose corticosteroids [74] . However, it was inconclusive whether this was a true trend or just an improvement in diagnostic capabilities. In addition, Fukushima et al observed that according to the Research Committee on Idiopathic Osteonecrosis of the Femoral Head in Japan, roughly 2200 new patients per year were diagnosed with ON. Comparatively, United States-based studies have shown much lower ON rates, with roughly a 15% ON prevalence in SLE patients [75] . Race-specific differences in gene mutations, such as factor V Leiden and prothrombin, have been implicated, however, definitive causes for disparities in incidence remain unclear, and may also be idiopathic or attributed to lifestyle [76] . Nevertheless, it is still important to highlight that there may be fundamental differences in the way certain patients respond to corticosteroids. Several studies published in the last two decades have demonstrated associations between high corticosteroid doses and osteonecrosis incidences, although others did not have this trend [12, 42, 50] . Similar to results observed in this meta-analysis, a prior systematic review found a 4.6% increase in the rate of osteonecrosis for every 10 mg per day increase in corticosteroid doses ( Fig. 2 ; present study had 10% increase per 10 mg/day) [67] . In addition, mean daily doses of greater than 40 mg per day (prednisone-equivalent) have been associated with a higher risk for developing osteonecrosis. One prospective study demonstrated 4-fold higher incidences of osteonecrosis in patients treated with mean daily corticosteroid doses of greater than 40 mg (P b 0.05) [66] . Nagasawa et al, in a study of 45 patients diagnosed with SLE and treated with oral prednisolone (40 mg/day), observed dose-response relationships with osteonecrosis over five years [77] . Based on yearly MRI scans, they found that administration of high corticosteroid doses (N1000 mg/day) was more frequently found in patients with early stage osteonecrosis than those without osteonecrosis (87% versus 37%; P b 0.01). It is possible that discrepancies between studies may be attributed to recent increases in MRI use for the diagnosis of earlystage osteonecrosis, which may have otherwise been undetectable on radiographic assessment.

Multiple studies have also reported on the association between cumulative corticosteroid doses and osteonecrosis. For example, previous reports suggest that patients receiving greater than 2 g of cumulative corticosteroid doses are at a higher risk for developing osteonecrosis [1, 3] . However, it is unclear whether this represents a ceiling above which patients develop osteonecrosis at the same rates irrespective of total doses, or a floor, above which risk for osteonecrosis increases with higher cumulative doses. Additionally, Nakamura et al evaluated 201 patients with SLE over 13 years and observed that osteonecrosis risk was associated with increasing cumulative corticosteroid doses, with 15% of patients requiring increased corticosteroid doses developing the disease [39] . Shibatani and colleagues evaluated renal transplant recipients who underwent multiple rejection cycles, and they demonstrated associations between osteonecrosis and cumulative corticosteroid doses (OR 4.2; P = 0.008), but not with rejection episode numbers, which allude to the likelihood that higher cumulative doses, rather than systemic effects of host-versus-graft disease, may be implicated in osteonecrosis pathogenesis [60] . However, accurate differentiation between effects of mean daily corticosteroid doses and cumulative doses on the osteonecrosis risk is difficult, since patients with the highest cumulative doses often receive higher mean daily doses and/or are treated for longer durations.

Treatment durations have also been implicated as independent risk factors for developing osteonecrosis. Nakamura et al evaluated 201 patients (537 joints) with SLE who were treated with prednisone doses of greater than 40 mg per day [39] . Of the 537 joints, 238 (44%) developed osteonecrosis. They concluded that progression of osteonecrosis was associated with higher doses of corticosteroid treatment for longer durations.

The association between pulsed corticosteroid therapy and osteonecrosis has been reported to be variable. Oinuma et al, studying 72 patients with SLE, found no differences in osteonecrosis incidences in patients treated with pulsed methylprednisolone therapy in combination with minimum daily corticosteroid doses of 40 mg/day [38] . Of the 32 patients who developed ON, 17 were treated with pulsed corticosteroids while of the 40 patients who did not develop osteonecrosis, 18 were treated with pulsed corticosteroids. However, this study did not specify pulsed corticosteroid dosages. However, in a study by Ce et al, 60 multiple sclerosis patients, who did not have any risk factors for osteonecrosis, were treated with pulsed corticosteroids and were compared to a matched group of patients who did not receive corticosteroids [72] . Cumulative pulsed corticosteroid doses received by the treatment group was greater than 10 g, and they only received pulsed-corticosteroids and were not treated with any maintenance doses between pulses. It was observed that treatment patients had significantly higher incidences of femoral head osteonecrosis as diagnosed on MRI (15.5%; P b 0.05) compared to the non-corticosteroid group (0%). Similar results were observed in a study of 498 renal transplant patients, which demonstrated a significantly greater incidence of osteonecrosis in those receiving pulsed therapies (11%) compared to patients receiving non-pulsed therapy (3%; P b 0.01) [55] .

There were several limitations of this study. The lack of prospective randomized-controlled trials may have contributed to both studydesign and reporting bias in individual reports, which may have skewed the observed outcomes. Although double-blinded, prospective randomized-controlled trials are gold standards for evidence-based medicine, in certain situations, such as renal transplantation or SLE, where the first-line therapy is corticosteroids, it would have been unethical to design studies denying treatment. Thus, we relied primarily on case-control studies. There was also a lack of consistency in osteonecrosis diagnostic methods. Studies published after 1990 utilized MRI, while earlier studies relied on patient symptoms, radiographic findings, biopsies, and bone scanning, which due to low diagnostic sensitivity [78] may have underestimated the true osteonecrosis incidence. In addition, lower quality studies on medical diagnoses other than renal transplantation or SLE prevented multivariate analysis due to insufficient degrees of freedom in the statistical analyses. This was evident in studies on bone marrow transplant recipients, cardiac transplants, and SARS. However, we were able to analyze two common patient groups (SLE and renal transplantation). Furthermore, almost all studies reported using high cumulative corticosteroid doses. This study did not consider incidental corticosteroid doses, such as the use of dose-packs or corticosteroids in minimal doses that are typically not associated with osteonecrosis. We have focused on patients who received minimum cumulative doses of 2 g or 30 mg daily doses for less than 2 months. Thus, these results may not be indicative of patients given low cumulative doses (b 2 g) of corticosteroids for medical conditions that were not directly associated with osteonecrosis.

These meta-analysis and assessment of the available literature demonstrated that high-dose corticosteroid treatments may increase risk for developing osteonecrosis up to ten-fold. Patients treated with daily doses N40 mg were at higher risk, with 3.6% increase in incidences for every 10 mg increase in doses. Effects of cumulative corticosteroid doses and treatment durations are less clear, but are likely to have synergistic relationships with daily doses and underlying diagnoses. Pulsed-therapy has effects on increasing osteonecrosis risk. As molecular and genetic bases for this disease evolves, further knowledge of risk factors for osteonecrosis with prospective MRI-based studies will assist medical practitioners in educating patients. Presently, we would advise erring on the side of caution and using the lowest possible corticosteroid doses, while still maintaining clinical efficacy and minimizing risks. 

Non-traumatic avascular necrosis of the femoral head

Osteonecrosis: etiology, diagnosis, and treatment

Nontraumatic osteonecrosis of the femoral head: ten years later

Prevalence of osteonecrosis of the femoral head: a nationwide epidemiologic analysis in Korea

Nationwide epidemiologic survey of idiopathic osteonecrosis of the femoral head

Treatment of femoral head osteonecrosis in the United States: 16-year analysis of the Nationwide Inpatient Sample

The pathogenesis of osteonecrosis

Current concepts on the pathogenesis and natural history of steroid-induced osteonecrosis

Osteogenic gene expression decreases in stromal cells of patients with osteonecrosis

Genetic association of angiogenesis-and hypoxiarelated gene polymorphisms with osteonecrosis of the femoral head

Avascular necrosis of the femoral and humeral heads after high-dosage corticosteroid therapy

Corticosteroid therapy associated with ischemic necrosis of bone in systemic lupus erythematosus

Potency and duration of action of glucocorticoids. Effects of hydrocortisone, prednisone and dexamethasone on human pituitary-adrenal function

Osteonecrosis as a complication of treating acute lymphoblastic leukemia in children: a report from the Children's Cancer Group

Avascular necrosis of bone-a complication of aggressive therapy for acute lymphoblastic leukemia

Avascular osteonecrosis in patients treated for Hodgkin's disease

Granulomatous osteonecrosis in Crohn's disease

Inflammatory bowel disease related osteonecrosis: report of a large series with a review of the literature

Avascular necrosis following bone marrow transplantation: a case-control study

Corticosteroid dose as a risk factor for avascular necrosis of the bone after hematopoietic cell transplantation

Avascular necrosis of the femoral head secondary to corticosteroid therapy for graft-versus-host disease after marrow transplantation: effective therapy with hip arthroplasty

Avascular necrosis of bone: a common serious complication of allogeneic bone marrow transplantation

Age and total-body irradiation in addition to corticosteroid dose are important risk factors for avascular necrosis of the bone

Avascular necrosis of bone after allogeneic bone marrow transplantation: analysis of risk factors for 4388 patients by the Societe Francaise de Greffe de Moelle (SFGM)

Avascular necrosis of bone after allogeneic bone marrow transplantation: clinical findings, incidence and risk factors

Osteonecrosis of the femoral head after allogeneic bone marrow transplantation

Avascular necrosis of bone following allogeneic stem cell transplantation: MR screening and therapeutic options

Osteonecrosis of hip and knee in patients with severe acute respiratory syndrome treated with steroids

Horizontal error bars represent the 95% confidence interval of the odds ratio for each individual study. The diamond represents the pooled odds ratio for all the studies with a corresponding horizontal bar representing the 95% confidence interval. A pooled odds ratio N1 represents higher-odds of having osteonecrosis if treated with greater than 2 g of corticosteroid (A) while an odds ratio of less than 1 is associated with lower odds of

Steroid-induced osteonecrosis in severe acute respiratory syndrome: a retrospective analysis of biochemical markers of bone metabolism and corticosteroid therapy

Factors of avascular necrosis of femoral head and osteoporosis in SARS patients' convalescence

Aseptic necrosis of bone in systemic lupus erythematosus. Relationship to corticosteroid therapy

Computer analysis of factors influencing the appearance of aseptic necrosis in patients with SLE

Current therapy of systemic lupus erythematosus. A comparative evaluation of corticosteroids and their side-effects with emphasis on fifty patients treated with dexamethasone

Predictive factors for symptomatic osteonecrosis in patients with systemic lupus erythematosus

High-dose intravenous methylprednisolone therapy associated with osteonecrosis in patients with systemic lupus erythematosus

Avascular osteonecrosis in patients with SLE: relation to corticosteroid therapy and anticardiolipin antibodies

Risk factors for osteonecrosis in systemic lupus erythematosus

Osteonecrosis in patients with systemic lupus erythematosus develops very early after starting high dose corticosteroid treatment

Spontaneous repair of asymptomatic osteonecrosis associated with corticosteroid therapy in systemic lupus erythematosus: 10-year minimum follow-up with MRI

Risk factors of avascular necrosis of the femoral head in patients with systemic lupus erythematosus under high-dose corticosteroid therapy

Osteonecrosis in systemic lupus erythematosus, steroid-induced or a lupus-dependent manifestation?

Avascular necrosis in SLE. An apparent predilection for young patients

Factors associated with osteonecrosis in Thai lupus patients: a case control study

Aseptic necrosis and glucocorticosteroids in systemic lupus erythematosus: a reevaluation

Damage in systemic lupus erythematosus and its association with corticosteroids

Symptomatic osteonecrosis of the hip and knee after cardiac transplantation

Symptomatic osteonecrosis of the hip after orthotopic liver transplantation

Avascular necrosis of bone after cardiac transplantation. Prevalence and relationship to administration and dosage of steroids

Hyperparathyroidism and avascular necrosis of bone after kidney transplantation. A retrospective study of 170 transplant recipients

Avascular bone necrosis in the renal transplant patient: a discriminant analysis of 144 cases

Steroid treatment and aseptic necrosis of the femoral head in renal transplant recipients

Avascular necrosis of bone after renal transplantation

Avascular osteonecrosis after renal transplantation

Aseptic necrosis of bone following renal transplantation: experience in 194 transplant recipients and review of the literature

Necrosis of the femoral head after renal transplantation

Incidence of osteonecrosis after renal transplantation

Low dose oral prednisolone in renal transplantation

Aseptic bone necrosis after renal transplantation

Avascular necrosis of bone following renal transplantation

Degree of corticosteroid treatment within the first 2 months of renal transplantation has a strong influence on the incidence of osteonecrosis of the femoral head

Avascular necrosis of bone after renal transplantation

Risk factors for avascular bone necrosis after renal transplantation

Clinically occult avascular necrosis of the hip: prevalence in an asymptomatic population at risk

Bone necrosis after renal transplantation

Epiphyseal osteonecrosis in transplanted patients: effect of surgical treatment

Incidence of osteonecrosis associated with corticosteroid therapy among different underlying diseases: prospective MRI study

Across-study evaluation of association between steroid dose and bolus steroids and avascular necrosis of bone

Avascular necrosis of bone in severe acute respiratory syndrome

Avascular Necrosis (AVN) after allogeneic hematopoietic stem cell transplantation (allo-HSCT): implications for pre-emptive therapy in high risk populations

Incorporating variations in the quality of individual randomized trials into meta-analysis

The quality of reporting of randomized trials in the

Avascular necrosis of the bones: an overlooked complication of pulse steroid treatment of multiple sclerosis

Age at time of corticosteroid administration is a risk factor for osteonecrosis in pediatric patients with systemic lupus erythematosus: a prospective magnetic resonance imaging study

Incidence of nontraumatic osteonecrosis of the femoral head in the Japanese population

Musculoskeletal complications of systemic lupus erythematosus in the Hopkins Lupus Cohort: an update

Prevalence of genetic risk factors related with thrombophilia and hypofibrinolysis in patients with osteonecrosis of the femoral head in Poland

Very early development of steroid-associated osteonecrosis of femoral head in systemic lupus erythematosus: prospective study by MRI

Bone scanning of limited value for diagnosis of symptomatic oligofocal and multifocal osteonecrosis

Statistical Appendix (2) Random effects model was used in the meta-analysis due to variations in studies with respect to age, diagnoses, and comorbidities as it was assumed that it was unlikely that single effect sizes would be consistent across all studies. Orwin's fail-safe N and Duval and Tweedie's trim and fill statistics were used to investigate publication bias. The adjusted odds-ratio for developing osteonecrosis with corticosteroid use was 6.4 (95% CI, 3.3 to 12.3). Using the criteria for trivial odds (i.e. not clinically important) ratio as 1.1 and mean odds ratio of 1.0 in missing studies, we found that 89 additional studies were necessary to reduce odds ratio to 1.1. Similarly, adjusted odds-ratios for developing osteonecrosis with high corticosteroid doses compared to low-dose corticosteroids were 2.4 (95% CI, 0.8 to 6.4). An additional 42 studies with mean odds ratio of 1.0 would be necessary to reduce odds-ratios of developing osteonecrosis with high-dose corticosteroids to under 1.1 suggesting that there exists nominal publication bias.Heterogeneity was investigated by conducting subgroup analyses and the extent evaluated using Cochrane Q and I 2 statistics with higher values suggestive of greater heterogeneity. There were substantial heterogeneities among 4 studies evaluating effects of low-and high-dose corticosteroids in osteonecrosis development (Cochrane Q = 7.8; I 2 = 61.4%; P = 0.05). Substantially less heterogeneities were found in subgroup analyses involving 3 reports on renal transplant recipients (Cochrane Q = 2.3, I 2 = 12.2%; P = 0.32). Marked homogeneity was found between 4 reports evaluating corticosteroid effects on osteonecrosis development (Cochrane Q = 2.1; I 2 = 0%; P = 0.32). In addition, substantial homogeneity was found in the subgroup of 3 reports investigating effects of corticosteroids on osteonecrosis incidences in hematopoietic stem cell transplant recipients (Cochrane Q = 1.8; I 2 = 0%; P = 0.4).1512.e5