Original Article

Prediction of low bone mineral density in patients with inflammatory bowel diseases

United European Gastroenterology Journal 2016, Vol. 4(5) 669–676 ! Author(s) 2016 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/2050640616658224 ueg.sagepub.com

Solvey Schu¨le1, Jean-Benoıˆt Rossel2, Diana Frey3, Luc Biedermann1, Michael Scharl1, Jonas Zeitz1, Nata´lia Freitas-Queiroz1, Vale´rie Pittet2, Stephan R Vavricka1,4, Gerhard Rogler1 and Benjamin Misselwitz1 for the Swiss IBD cohort study

Abstract Background: Low bone mineral density (BMD) remains a frequent problem in patients with inflammatory bowel diseases (IBD). There is no general agreement regarding osteoporosis screening in IBD patients. Methods: Cases of low BMD and disease characteristics were retrieved from 3172 patients of the Swiss IBD cohort study. Multivariate logistic regression analysis was conducted for predictive modeling. In a subgroup of 877 patients, 253 dualenergy X-ray absorptiometry (DXA) scans were available for validation. Results: Low BMD was prevalent in 19% of patients. We identified seven predictive factors: type of IBD, age, recent steroid usage, low body mass index, perianal disease, recent high disease activity and malabsorption syndrome. Low BMD could be predicted with a sensitivity of 79% and a specificity of 64%, a positive predictive value (PPV) of 35% and a negative predictive value (NPV) of 93%. The area under the curve of the receiver operating characteristics was 0.78. In the validation cohort we calculated a PPV of 26% and an NPV of 88%. Conclusion: We provide a comprehensive analysis of risk factors for low BMD and propose a predictive model with seven clinical variables. The high NPV of models such as ours might help in excluding low BMD to prevent futile investigations.

Keywords Osteoporosis, bone mineral density, inflammatory bowel diseases, predictive model Received: 28 February 2016; accepted: 12 June 2016

MTWAI OR ROC SIBDC UC

Abbreviations ACG AGA BMD BMI BSG CD CDAI CI CRP DXA ECCO EIM IBD IC

American College of Gastroenterology American Gastroenterological Association bone mineral density body mass index British Society of Gastroenterology Crohn’s disease Crohn’s Disease Activity Index confidence interval C-reactive protein dual-energy x-ray absorptiometry European Crohn’s and Colitis Organization extra-intestinal disease manifestation inflammatory bowel disease indeterminate colitis

Modified Truelove and Witts Activity Index odds ratio receiver operating characteristic Swiss IBD cohort Ulcerative colitis

1

Division of Gastroenterology and Hepatology, University Hospital Zurich and Zurich University, Switzerland 2 Institut de Me´decine Sociale et Pre´ventive Unite´ d’Evaluation des Soins Bio2/02/185, Switzerland 3 Division of Rheumatology, University Hospital Zurich and Zurich University, Switzerland 4 Department of Medicine, Division of Gastroenterology, Triemli Hospital, Zurich, Switzerland Corresponding author: Benjamin Misselwitz, Division of Gastroenterology and Hepatology, University Hospital Zurich, Ra¨mistr. 100, 8091 Zu¨rich, Switzerland. Email: [email protected]

670

United European Gastroenterology Journal 4(5)

Introduction Osteoporosis is an important clinical problem in patients with inflammatory bowel diseases (IBD).1–3 The fracture risk for IBD patients is increased by approximately 40%60% compared to controls.4,5 Risk factors for decreased bone mineral density (BMD) in IBD patients include activity and severity of gut inflammation, systemic steroid usage, and intestinal malabsorption leading to calcium and vitamin D deficiency.1,6–13 In addition general risk factors for low BMD such as low body mass index (BMI), advanced age, smoking, alcohol intake and reduced physical activity also apply to IBD patients.6,8,11,14,15 However, not all risk factors have been consistently identified in all investigations and considerable controversy remains. Low BMD is a powerful predictor of fracture risk, and a BMD one standard deviation below the age adjusted mean increases the relative fracture risk by 1.6–2.6.16 BMD is usually determined using dualenergy X-ray absorptiometry (DXA). For the T score, measurements are compared to a healthy, young, adult, sex-matched reference population while the Z score in addition controls for age and ethnicity.17 A T score 2.5 defines osteoporosis in postmenopausal women and in men 50 years while osteopenia is diagnosed with a T score 3 months in the presence of  2 additional risk factors: continuing active disease, weight loss > 10%, BMI < 20 kg/m2, age > 70 years - Persistently/ continuing active disease - Patients in high-risk situations such as age > 70 years, with very active disease, with disease responding poorly to treatment, with poor nutrition (and other features)

European Crohn’s and Colitis Organization (ECCO) 2010 (CD) 2013 (UC)21,22

- Repeatedly exposed to corticosteroid - Persistently/ continuing active disease - Long disease duration

American College of Gastroenterology (ACG) 2009 (CD),24 2010 (UC)23 and American Gastroenterological Association (AGA) 200325

- Steroid usage > 3 months (consecutively or recurrent users) - Persistently/ continuing active disease - ACG: DXA scan should be considered in IBD patients with risk factors for osteoporosis such as smoking, low BMI, sedentary lifestyle, hypogonadism, family history, and nutritional deficiencies; over age 60 - AGA: DXA scan should be considered in IBD patients with low trauma fracture, postmenopausal female or male aged > 50, hypogonadism

IBD: inflammatory bowel diseases; DXA: dual-energy X-ray absorptiometry; CD: Crohn’s disease; UC: ulcerative colitis; BMI: body mass index.

Schu¨le et al. Table 2. Epidemiological characteristics of our patients with IBD Epidemiology, n ¼ 3172 CD: 1794 (56.6%) UC/IC: 1378 (43.4%) Gender Male: 1611 (50.8%) Female: 1561 (49.2%) Age Mean: 41.7 years, range: 1.3–91.5 years BMI Mean: 24.0 kg/m2 , range 12.5–62.5 kg/m2 Current smoker Total: 661 (20.8%) Duration of disease Mean: 12.0 years, range: 0.1–58.2 years Initial extent of Proctitis: 251 (18.2%) disease (UC) Left-sided colitis: 424 (30.8%) Pancolitis: 566 (41.1%) Unknown: 137 (9.9%) Initial disease L1 (ileal): 406 (22.6%) location (CD) L2 (colonic): 362 (20.2%) L3 (ileo-colonic): 837 (46.7%) L4 (upper GI): 17 (0.9%) Unknown: 172 (9.6%) Past intestinal surgery 48 (3.5%) (UC only) Past intestinal surgery 667 (37.2%) (CD only) Treatment with Infliximab: 1255 (39.5%) anti-TNF drug ever Adalimumab: 487 (15.3%) Certolizumab: 186 (5.9%) Any anti-TNF: 1415 (44.6%)

671 patients); (3) diagnosis of osteoporosis/osteopenia in patient chart with clear reference to a DXA scan but no scores available (18 patients.)

Diagnosis

IBD: inflammatory bowel diseases; IC: indeterminate colitis; UC: ulcerative colitis; CD: Crohn’s disease; GI: gastrointestinal; anti-TNF: anti-tumor necrosis factor.

Truelove and Witts activity index (MTWAI, for UC/ IC patients) and the Crohn’s disease activity index (CDAI, for CD patients) was linearly normalized to a score ranging from 0 (no disease activity) to 100 (highest disease activity). The assessment of malabsorption was based on the diagnosis of the responsible physician using criteria including presence of steatorrhea, continuing weight loss, B12/folate or other specific deficiencies and/or a requirement for nutritional supplements. In six centers a chart review of all IBD patients was performed and 877 patient charts were reviewed for evidence of past DXA scans. All T scores and Z scores were retrieved. For the diagnosis of osteoporosis, cut-offs of 2.5 and 2 were chosen for the T score and Z score, respectively. For osteopenia, cut-offs of 1 > T score >2.5 and 1 > Z score > 2 were used. The lowest scores for hip or spine were used. For diagnosis or exclusion of osteoporosis and osteopenia during chart review, the following hierarchy was used: (1) T score (available for 229 out of 253 patients); (2) Z score (if no T score available, six

Ethical considerations The SIBDC study protocol has been approved by all ethics committees where patients were included. Data analysis was performed according to the Declaration of Helsinki.

Data analysis Univariate and multivariate logistic regression analyses were used to determine the association with various potential risk factors for osteoporosis in IBD patients. For the multivariate models individual variables with a p value >0.157 were excluded in a step-wise approach.26 For the predictive model of osteoporosis the multivariate analysis was performed to minimize the Akaike information criterion. The cut-off for a positive test was chosen to maximize the sum of the corresponding sensitivity and specificity (Youden’s criterion). For this analysis Stata software was used (StataCorp., 2015. Stata Statistical Software: Release 14. College Station, TX, USA: StatCorp LP).

Results Risk factors for osteoporosis For our analysis we used data from SIBDC, a large prospective cohort study of well-characterized Swiss IBD patients. Altogether, data for 3172 IBD patients could be retrieved; basic characteristics of our cohort patients can be found in Table 2. Our cohort thus represents a typical IBD population with respect to age, gender and distribution of disease severity. Presence of osteoporosis/osteopenia was mentioned for 616 patients (19.4%). IBD characteristics including IBD diagnosis, disease activity, steroid usage, stenosis, perianal disease (including fistulae), extra-intestinal manifestations (EIM), malabsorption syndrome and disease duration were significantly associated with low BMD in a univariate regression analysis (Table 3). Additional significant parameters were body mass index (BMI), smoking and lack of physical activity. Multivariate analysis for risk factors of low BMD. Multivariate logistic regression analysis restricted risk factors for low BMD to seven parameters: IBD diagnosis, age, BMI, disease activity, steroids at last measurement, perianal disease including fistulae and malabsorption syndrome (Table 4).

672

United European Gastroenterology Journal 4(5)

Table 3. Univariate analysis of risk factors for low BMD in IBD patients Univariate logistic regressions Diagnosis CD (n ¼ 1793) UC or IC (n ¼ 1377) Gender Men (n ¼ 1610) Women (n ¼ 1561) Last BMI (n ¼ 3110) Last smoking status No (n ¼ 2511) Yes (n ¼ 661) Steroids at last measurement No (n ¼ 1785) Yes (n ¼ 1387) Stenosis No (n ¼ 2440) Yes (n ¼ 732) Perianal disease (fistula, fissure, abscess) No (n ¼ 2277) Yes (n ¼ 895) Intestinal surgery No (n ¼ 2270) Yes (n ¼ 902) Malabsorption syndrome No (n ¼ 3046) Yes (n ¼ 126) Extra-intestinal manifestations No (n ¼ 1706) Yes (n ¼ 1466) Age at last measurement, in years (n ¼ 3170) Childhood diagnosis of IBD No (n ¼ 2528) Yes (n ¼ 635) Disease duration, in years (n ¼ 3163) Family history of IBD No (n ¼ 1859) Yes (n ¼ 415)

Table 3. Continued

Odds ratio (95% CI; p value)

Drinking alcohol more than once a day No (n ¼ 2111) Yes (n ¼ 175)

1 (ref) 0.789 (0.659–0.944; 0.010)

Sport at least once a week No (n ¼ 878) Yes (n ¼ 1354)

1 (ref) 1.155 (0.968–1.377; 0.109) 0.952 (0.932– 0.972; < 0.001)

Last activity indexa (n ¼ 3170)

1 (ref) 1.274 (1.034–1.569; 0.023) 1 (ref) 4.731 (3.890– 5.753; < 0.001) 1 (ref) 1.530 (1.256– 1.863; < 0.001) 1 (ref) 1.635 (1.357– 1.969; < 0.001) 1 (ref) 1.641 (1.363– 1.976; < 0.001) 1 (ref) 2.498 (1.719– 3.630; < 0.001) 1 (ref) 1.442 (1.209– 1.721; < 0.001) 1.022 (1.016– 1.027; < 0.001) 1 (ref) 0.549 (0.426– 0.706; < 0.001) 1.025 (1.016– 1.034; < 0.001) 1 (ref) 1.168 (0.907–1.504; 0.228) (continued)

1 (ref) 1.133 (0.786–1.631; 0.504) 1 (ref) 0.775 (0.631–0.952; 0.015) 1.037 (1.031– 1.043; < 0.001)

a For the calculation of the activity index, the last measurement of Crohn’s disease activity index was divided by 500 or the last measurement of the modified Truelove and Witts severity index was divided by 21, respectively, and multiplied by 100. BMD: bone mineral density; IBD: inflammatory bowel diseases; IC: indeterminate colitis; CD: Crohn’s disease; UC: ulcerative colitis; CI: confidence interval; BMI: body mass index.

Table 4. Predictive model/ multivariate logistic regression analysis of risk factors for low BMD in IBD patients Multivariate logistic regression (n ¼ 3108)

Odds ratio (95% CI; p value)

Diagnosis CD 1 (ref) UC 0.814 (0.650–1.019; 0.073) Age at last measurement, in years 1.032 (1.026–1.039; < 0.001) Last BMI 0.929 (0.906–0.952; < 0.001) a Last activity index 1.029 (1.022–1.036; < 0.001) Steroids at last measurement No 1 (ref) Yes 4.353 (3.528–5.370; < 0.001) Perianal disease (fistula, fissure, abscess) No 1 (ref) Yes 1.589 (1.266–1.993; < 0.001) Malabsorption syndrome No 1 (ref) Yes 1.984 (1.288–3.055; 0.002) The model selection is based on the minimization of the Akaike Information criterion. To calculate the likelihood p of low BMD the following calculations need to be performed: p ¼ exp(s)/(1 þ exp(s)); s ¼ –2.299 (0.310) – 0.206 (0.115) * UC þ 0.032 (0.003) * age – 0.074 (0.013) * BMI þ 0.028 (0.004) * activity index þ 1.471 (0.107) * steroid usage þ 0.463(0.116) * perianal disease þ 0.685 (0.220) * malabsorption. Values in brackets are the standard error. A value p > 0.16 is indicative of low BMD according to our model. a For the calculation of the Activity Index, the last measurement of Crohn’s disease activity index was divided by 500 or the last measurement of the modified Truelove and Witts severity index was divided by 21, respectively, and multiplied by 100. BMD: bone mineral density; IBD: inflammatory bowel diseases; CD: Crohn’s disease; UC: ulcerative colitis; CI: confidence interval; BMI: body mass index.

Schu¨le et al.

673

To exclude confounding effects of preexisting osteoporosis/osteopenia on our model, we performed a sensitivity analysis, restricting our analysis to patients with new-onset of low BMD. For this secondary analysis 435 out of 616 patients (71%) with evidence of osteoporosis/osteopenia at enrollment were excluded. The resulting model was highly similar to the original model and six risk factors for low BMD were confirmed: BMI, steroids, perianal disease, malabsorption, age and last activity index. EIM showed significance only in our sensitivity analysis and IBD diagnosis lost significance (data not shown). In another sensitivity analysis CD and UC/IC patients were considered separately: For CD identical risk factors were identified compared to the whole cohort (not shown). Risk factors for low BMD in UC/IC patients were also highly similar, only malabsorption syndrome was not significant anymore while presence of EIM was a positive risk factor for low BMD (p ¼ 0.03, not shown).

Prediction of low BMD in IBD patients We tested the power of our multivariate logistic model (Table 4) for the prediction of low BMD. The receiver operating characteristic (ROC) curve of our model is shown in Figure 1 and yields an area under the curve of 0.78. Using an optimal threshold (see Methods), we calculated a sensitivity of 79% and a specificity of 64% for low BMD in our cohort. In our patients our model had a positive predictive value of 35% and a negative predictive value of 93%. Globally, our model was able to classify 2074 out of 3108 patients correctly (67%). No cross-validation was performed. To validate our predictive model, we performed a chart review of 877 IBD patients and received data of individual DXA scans. Information regarding BMD was found for 253 patients. For these individuals a

1

Sensitivity

0.75

0.50

0.25 Area under ROC curve = 0.78

0 0

0.25

0.50 1- Specificity

0.75

1

Figure 1. The receiver operating characteristic (ROC) curve of our predictive model (compare with Table 4).

high degree of agreement with the SIBDC database was noted: In 132 patients with low BMD according to the SIBDC database information regarding DXA scans were available and the diagnosis was confirmed in 122 (92%). For five out of 10 patients with discordant results (i.e. low BMD according to the SIBDC database but a formally normal measurement) the discrepancy was likely due to T scores at the lower limit of normal (between 0.8 and 1.0). Similarly, normal BMD in the SIBDC database was confirmed in 122 of 169 patients (72%) by our chart review. In 47 discordant cases 10 showed borderline osteopenia (T score 1.1 to 1.3), four were performed before 2006, one showed normalization of osteoporosis in the latest DXA scan and one DXA scan was recent and possibly not covered by the latest data entry. No possible explanation for a discrepancy was noted in 31 cases (18%). When the results of DXA scans were used to define low BMD, our model had a sensitivity of 71%, a specificity of 51%, a positive predictive value of 25% and a negative predictive value of 88%.

Discussion In our study we provide a comprehensive analysis of decreased BMD in patients with IBD. We use diseaseand patient-specific risk factors to provide a model for prediction of low BMD. Our analysis identifies high disease activity, perianal disease including fistulae, steroid usage, low BMI, malabsorption syndrome and age as risk factors for low BMD. We find a correlation between classical disease activity scales such as CDAI and MTWAI and osteoporosis/ osteopenia. In most previous studies no association between disease activity indices and low BMD was detected and only one publication described the Rachmilewitz Colitis activity index as a predictor for osteoporosis in a small number of UC patients.9 Severe or persistently active disease has been established as a risk factor for osteoporosis/osteopenia;13,27 however, up to now neither CDAI nor MTWAI but different definitions for ‘‘severe’’ and ‘‘active’’ disease have been applied including biochemical criteria such as levels of C-reactive protein (CRP)13,27 or albumin,28 need for ileum resection27 and patient symptom questionnaires.29 Our model also identifies perianal disease with fistulae, abscesses or fissures as a risk factor for low BMD. Patients with perianal disease differ from other IBD patients by having a younger age at onset of disease,30 a more aggressive disease course,31 a higher likelihood of EIM32 and higher rates of recurrences after surgery.33 Perianal disease therefore might be a strong marker for a long, severe and recurring disease course, which in turn would predispose to low BMD.

674 In IBD patients, a correlation of BMI and BMD6–8,14,15,29 has been noted in our analysis as well as in previous studies. Thereby, positive effects of a high BMI for stimulation of bone formation cannot be distinguished from a low BMI as a marker of severe inflammation or disease activity. In line with previous studies,1,6,10–13,27,34 we also identified malabsorption as a risk factor for low BMD. Malabsorption might lead to low levels of calcium and vitamin D as well as other nutrients, resulting in preferential bone resorption. One study even demonstrated a direct relationship between low vitamin D levels and osteoporosis in IBD patients.12 Our study also confirmed steroid treatment as a risk factor for low BMD, in line with previous analyses.6–9,29,35 Several international guidelines regarding osteoporosis screening in IBD patients exist with only partially overlapping recommendations (Table 1). Furthermore, most guidelines leave ambiguity and neither the European Crohn’s and Colitis Organization (ECCO)21,22 nor the British Society of Gastroenterology (BSG)19,20 guidelines provide an objective definition for ‘‘persistent’’ or ‘‘continuous’’ activity that should trigger a DXA scan. Guidelines of the American Gastroenterological Association (AGA) and the American College of Gastroenterology (ACG) provide objective criteria23–25; however, no criterion specifically applies to IBD. Overutilization of DXA scans would clearly limit cost-efficiency of osteoporosis screening. Therefore, accurate and easy applicable recommendations would be desirable. All seven clinical parameters in our model are available in the office. While the positive predictive power of our model was modest at best (35%), the negative predictive power was considerable (93%), which would be clinically very useful. Our model could therefore potentially be used for exclusion of low BMD to limit screening to the fraction of patients with a positive predictive test. However, our model first needs to be validated in an independent cohort of patients. In addition, development of applications for in-office risk calculations might increase usability of our model. Future studies might even improve the predictive power by inclusion of additional variables (menopause state, male hypogonadism, family history of hip fracture, previous low trauma fracture and other inflammatory conditions). The size of our study of 3172 patients compares favorably with previous studies that included up to 1230 patients. In addition, the high level of detail provided by the SIBDC database, which included detailed disease characteristics such as EIM and disease activity scores but also lifestyle factors such as physical activity and alcohol consumption, has not been provided by

United European Gastroenterology Journal 4(5) most previous studies. Since our results remained robust in a sensitivity analysis excluding all patients with low BMD at enrollment, risk factors of our model likely marks a disease dynamic directly leading to osteoporosis rather than identifying preexisting patient differences at baseline. Our study has several limitations: (i) Our model needs validation in an independent cohort of IBD patients.(ii) No systematic testing of all cohort patients by DXA scans has been performed and our analysis might miss osteoporosis/osteopenia cases. However, chart review and validation of our analyses using available DXA data largely confirmed our results. (iii) Since the SIBDC database does not distinguish between osteoporosis and osteopenia, both conditions cannot be distinguished in our predictive model. (iv) No control group of non-IBD patients is available and our study cannot determine the increase in osteoporosis risk attributable to IBD. (v) Even though the SIBDC study represents patients treated in tertiary referral centers as well as in private practice, the cohort study is biased toward large referral centers and our data might not be representative for the Swiss population affected by IBD. In summary, in a comprehensive analysis we propose a model for the prediction of low BMD using parameters available in the office, which after validation might facilitate the decision regarding osteoporosis screening. A model with a high negative predictive value might prevent unnecessary DXA scans. Acknowledgments We acknowledge all IBD patients participating in the SIBDC study for providing their personal medical history and current clinical symptoms, and regularly responding to the cohort questionnaires. Furthermore, we would like to acknowledge everybody from the data center of the SIBDC study for their tremendous work in data collection. SRV, GR and BM designed the study, SS performed the chart review of IBD patients, JBR performed the statistical analysis, SS and BM contributed to the statistical analysis and wrote the paper, JBR, DF, LB, MS, JZ, NFQ, VP, SRV and GR reviewed and edited the paper for important intellectual content. All authors have seen and approved the final version of the manuscript.

Funding This work was supported by research grants from the Swiss National Science Foundation to BM (Grant No. 32473B_156525), and the Swiss IBD Cohort (Grant No. 3347CO-108792).

Declaration of conflicting interests None declared.

Schu¨le et al. References 1. Rothfuss KS, Stange EF and Herrlinger KR. Extraintestinal manifestations and complications in inflammatory bowel diseases. World J Gastroenterol 2006; 12: 4819–4831. 2. Larsen S, Bendtzen K and Nielsen OH. Extraintestinal manifestations of inflammatory bowel disease: Epidemiology, diagnosis, and management. Ann Med 2010; 42: 97–114. 3. Levine JS and Burakoff R. Extraintestinal manifestations of inflammatory bowel disease. Gastroenterol Hepatol (N Y) 2011; 7: 235–241. 4. Bernstein CN, Blanchard JF, Leslie W, et al. The incidence of fracture among patients with inflammatory bowel disease. A population-based cohort study. Ann Intern Med 2000; 133: 795–799. 5. Card T, West J, Hubbard R, et al. Hip fractures in patients with inflammatory bowel disease and their relationship to corticosteroid use: A population based cohort study. Gut 2004; 53: 251–255. 6. Azzopardi N and Ellul P. Risk factors for osteoporosis in Crohn’s disease: Infliximab, corticosteroids, body mass index, and age of onset. Inflamm Bowel Dis 2013; 19: 1173–1178. 7. Frei P, Fried M, Hungerbuhler V, et al. Analysis of risk factors for low bone mineral density in inflammatory bowel disease. Digestion 2006; 73: 40–46. 8. Targownik LE, Bernstein CN, Nugent Z, et al. Inflammatory bowel disease has a small effect on bone mineral density and risk for osteoporosis. Clin Gastroenterol Hepatol 2013; 11: 278–285. 9. Schulte C, Dignass AU, Mann K, et al. Bone loss in patients with inflammatory bowel disease is less than expected: A follow-up study. Scand J Gastroenterol 1999; 34: 696–702. 10. Ali T, Lam D, Bronze MS, et al. Osteoporosis in inflammatory bowel disease. Am J Med 2009; 122: 599–604. 11. Katz S and Weinerman S. Osteoporosis and gastrointestinal disease. Gastroenterol Hepatol (N Y) 2010; 6: 506–517. 12. Miznerova E, Hlavaty T, Koller T, et al. The prevalence and risk factors for osteoporosis in patients with inflammatory bowel disease. Bratisl Lek Listy 2013; 114: 439–445. 13. Jahnsen J, Falch JA, Mowinckel P, et al. Bone mineral density in patients with inflammatory bowel disease: A population-based prospective two-year follow-up study. Scand J Gastroenterol 2004; 39: 145–153. 14. Leslie WD, Miller N, Rogala L, et al. Body mass and composition affect bone density in recently diagnosed inflammatory bowel disease: The Manitoba IBD Cohort Study. Inflamm Bowel Dis 2009; 15: 39–46. 15. Atreja A, Aggarwal A, Licata AA, et al. Low body mass index can identify majority of osteoporotic inflammatory bowel disease patients missed by current guidelines. ScientificWorldJournal 2012; 2012: 807438.

675 16. Marshall D, Johnell O and Wedel H. Meta-analysis of how well measures of bone mineral density predict occurrence of osteoporotic fractures. BMJ 1996; 312: 1254–1259. 17. Cosman F, de Beur SJ, LeBoff MS, et al. Clinician’s guide to prevention and treatment of osteoporosis. Osteoporos Int 2014; 25: 2359–2381. 18. Schousboe JT, Shepherd JA, Bilezikian JP, et al. Executive summary of the 2013 International Society for Clinical Densitometry Position Development Conference on bone densitometry. J Clin Densitom 2013; 16: 455–466. 19. Scott EM, Gaywood I and Scott BB. Guidelines for osteoporosis in coeliac disease and inflammatory bowel disease. British Society of Gastroenterology. Gut 2000; 46(Suppl 1): i1–i8. 20. Lewis NR and Scott BB. Guidelines for osteoporosis in inflammatory bowel disease and coeliac disease. London: British Society of Gastroenterology, 2007. 21. Van Assche G, Dignass A, Bokemeyer B, et al. Second European evidence-based consensus on the diagnosis and management of ulcerative colitis part 3: Special situations. J Crohns Colitis 2013; 7: 1–33. 22. Van Assche G, Dignass A, Reinisch W, et al. The second European evidence-based consensus on the diagnosis and management of Crohn’s disease: Special situations. J Crohns Colitis 2010; 4: 63–101. 23. Kornbluth A and Sachar DB; Practice Parameters Committee of the American College of Gastroenterology. Ulcerative colitis practice guidelines in adults: American College of Gastroenterology, Practice Parameters Committee. Am J Gastroenterol 2010; 105: 501–523; quiz 524. 24. Lichtenstein GR, Hanauer SB, Sandborn WJ, et al. Management of Crohn’s disease in adults. Am J Gastroenterol 2009; 104: 465–483; quiz 464, 484. 25. Bernstein CN, Leslie WD and Leboff MS. AGA technical review on osteoporosis in gastrointestinal diseases. Gastroenterology 2003; 124: 795–841. 26. Rousson V. Statistique applique´e aux sciences de la vie. Paris: Springer, 2013. 27. van Hogezand RA, Ba¨nffer D, Zwinderman AH, et al. Ileum resection is the most predictive factor for osteoporosis in patients with Crohn’s disease. Osteoporos Int 2006; 17: 535–542. 28. Compston JE, Judd D, Crawley EO, et al. Osteoporosis in patients with inflammatory bowel disease. Gut 1987; 28: 410–415. 29. Targownik LE, Leslie WD, Carr R, et al. Longitudinal change in bone mineral density in a population-based cohort of patients with inflammatory bowel disease. Calcif Tissue Int 2012; 91: 356–363. 30. Choi JH, Kim ES, Cho KB, et al. Old age at diagnosis is associated with favorable outcomes in Korean patients with inflammatory bowel disease. Intest Res 2015; 13: 60–67. 31. Beaugerie L, Seksik P, Nion-Larmurier I, et al. Predictors of Crohn’s disease. Gastroenterology 2006; 130: 650–656.

676 32. Ardizzone S, Puttini PS, Cassinotti A, et al. Extraintestinal manifestations of inflammatory bowel disease. Dig Liver Dis 2008; 40(Suppl 2): S253–S259. 33. Bernell O, Lapidus A and Hellers G. Risk factors for surgery and postoperative recurrence in Crohn’s disease. Ann Surg 2000; 231: 38–45.

United European Gastroenterology Journal 4(5) 34. Bernstein CN and Leslie WD. The pathophysiology of bone disease in gastrointestinal disease. Eur J Gastroenterol Hepatol 2003; 15: 857–864. 35. Bernstein CN, Seeger LL, Sayre JW, et al. Decreased bone density in inflammatory bowel disease is related to corticosteroid use and not disease diagnosis. J Bone Miner Res 1995; 10: 250–256.