Beta-blocker use in inflammatory bowel
disease (IBD)
A retrospective cohort study
Master Thesis
Tinka Bakker
August 15, 2016
Student Tinka Bakker Master of Medical Informatics University of Amsterdam Student number: 10667083 E-mail: tinka.bakker@gmail.com Mentor R.A. Willemze, MD PhD student Tytgat Institute for Liver and Intestinal Research Academic Medical Centre Prof. dr. W.J. de Jonge Tytgat Institute for Liver and Intestinal Research Academic Medical Centre Tutor M. Pippias, MBChB PhD student Department of Medical Informatics Academic Medical Centre, University of Amsterdam SRP Duration September 2015 – August 2016 Location of SRP Tytgat Institute for Liver and Intestinal Research Meibergdreef 69, 1105 BK Amsterdam
Table of Contents
Glossary & Abbreviations 4 Glossary 4 Abbreviations 5 Summary 6 Nederlandse samenvatting 7 Introduction 8 Beta-blockers & Inflammation 9 Research Question & Relevance 12 Methods 13 Data sources 13 Study population 14 Assessment of primary outcome 14 Sensitivity Analysis 16 Statistical Analysis 16 Results 19 Study Population 19 Patient characteristics 21 Clinical characteristics 23 Frequency of relapses 23 Sensitivity Analysis 25 Cox Proportional Hazards models 26 Stratified CoxPH Models 26 Discussion 28 Main findings 28 Comparison to other studies 28 Strengths & Limitations 29 Conclusion & Implications 31 References 32
Glossary & Abbreviations
GLOSSARY
Autonomous nervous system Part of the nervous system that unconsciously regulates body functions, such as respiratory rate and heart rate. β-adrenergic receptors A set of receptors that are targeted by catecholamines. There are three subtypes. Beta-blocker A group of drugs mainly used to treat cardiovascular diseases. Beta-blockers work by blocking the β-adrenergic receptors and therefore the effects of binding catecholamines. Catecholamines A class of neurotransmitters including epinephrine, norepinephrine and dopamine. Cox regression A type of survival models in statistics that models time-to-event data. Cytokine Small protein that plays an important role in immunological cell signalling. Epinephrine Neurotransmitter, also known as adrenaline, that binds to adrenergic receptors. It plays an important role in the physiological response to events perceived as a threat to survival, i.e. the fight-or-flight response, by for example increasing blood flow to muscles or output for the heart. Hazard ratio The proportion of two hazard rates. A hazard rate is the rate at which events happen. Inflammatory bowel disease Chronic inflammatory disease affecting the gastrointestinal tract. The primary subtypes are ulcerative colitis and Crohn’s disease. Intra-subject correlation Correlation between multiple events that occur within one subject. Relapse A flare of symptoms in a patient with established inflammatory bowel disease who is in clini-cal remission.Remission The resolution of symptoms and healing of the gastrointestinal tract in a patient with inflam-matory bowel disease. Shared frailty A random component one can add to a Cox proportional hazards model that describes the in- trinsic susceptibility that an individual has to experience an event, by taking into account un-measured heterogeneity. Sympathetic nervous system One of the two main branches of the autonomic nervous system. The sympathetic nervous system is responsible for the body functions in an active state, for example it increases heart rate.
ABBREVIATIONS
AMC – Academic Medical Centre CoxPH – Cox proportional hazards CD – Crohn’s disease ECCO – European Crohn’s and Colitis Organisation HBI – Harvey-Bradshaw index IBD – inflammatory bowel disease IBD-u – inflammatory bowel disease of unclassified type SCCAI – Simple Clinical Colitis Activity Index SNS – sympathetic nervous system TNF-α – tumor-necrosis-factor-α UC – ulcerative colitis 5-ASA – 5-amino-salicylic-acidSummary
Introduction: Inflammatory bowel disease (IBD) has a high disease burden and disease relapses are unpredictable. It is necessary to study factors that may be influencing the disease course. The widely used beta-blockers, prescribed to treat cardiovascular disease, could be such a fac- tor. Beta-blockers block the adrenergic receptors and thereby block the effect of catechola- mines, which possibly have an anti-inflammatory effect. The research question of this retro-spective cohort study was: Is there an association between the use of beta-blockers and the number of disease relapses in patients with inflammatory bowel disease?Methods: The study population was constructed from the Pobasic cohort. Relapses were iden-tified by using drug prescriptions as a proxy. The number of relapses per 100 person-years was calculated and compared between the group of patients using beta-blockers and the group of patients not using beta-blockers. Cox proportional hazards (CoxPH) models with shared frailty were created to assess whether the risk of relapse was the same between the two groups. Results: After reviewing the drug history, 250 patients were included in the study of which 12% used a beta-blocker and 88% did not. In the beta-blocker group patients experienced 21 re-lapses per 100 persons-year, compared to 29 relapses per 100 person-years. The unadjusted CoxPH model with shared frailty showed a 26% increased risk of relapse in the beta-blocker group compared to the non-beta-blocker group (hazards ratio [HR]: 1.26, 95% CI: 0.88-1.80, p: 0.21). According to the model adjusted for age and gender, beta-blocker users had a 54% higher risk of relapse than non-users (HR: 1.54, 95% CI: 1.05-2.25, p: 0.03). Discussion: Our retrospective cohort study is the first study to evaluate the association between beta-blocker use and relapses in patients with IBD. Patients with IBD taking beta-blockers have a higher risk of relapses. However, a large prospective trial with a more reliable method of identifying relapses is recommended to examine the relation between beta-blocker use and relapses in IBD. If the use of beta-blockers increases the risk of a relapse in patient with IBD this could have important consequences for the management of IBD.
Nederlandse samenvatting
Introductie: Inflammatoire darmziekten (IBD) veroorzaken een hoge ziektelast en opvlammin-gen van IBD zijn onvoorspelbaar. Het is daarom van belang om factoren te onderzoeken die het ziekteverloop kunnen beïnvloeden. Het gebruik van bètablokkers, een veel gebruikt medicijn, zou van invloed kunnen zijn. Bètablokkers blokkeren de adrenerge receptoren en daarmee het effect van catecholamines, die mogelijk een anti-inflammatoire eigenschappen hebben. De on-derzoeksvraag van deze retrospectieve cohortstudie was: Is er een associatie tussen gebruik van bètablokkers en het aantal opvlammingen in patiënten met IBD? Methode: De onderzoekspopulatie bestaat uit patiënten uit het Pobasic cohort waarvan de medicatiehistorie beschikbaar was. Opvlammingen werden opgespoord met behulp van medi- catievoorschriften. Het aantal opvlammingen per 100 person-years werd berekend en vergele- ken tussen de patiënten die bètablokkers gebruikten en patiënten die geen bètablokker ge-bruikten. Om het risico op een opvlamming te beoordelen werden Cox proportional hazards (CoxPH) modellen met shared frailty gecreëerd. Resultaten: 250 patiënten werden geïncludeerd waarvan 12% een bètablokker gebruikte en 88% geen bètablokker gebruikte. In de bètablokker groep hadden patiënten gemiddeld 21 op- vlammingen per 100 person-years ten opzichte van 29 per 100 person-years in de groep niet-gebruikers. Volgens het niet gecorrigeerde CoxPH model met shared frailty hadden bètablokker gebruikers een 26% hoger risico op een opvlamming dan niet gebruikers (hazard ratio [HR]: 1.26, 95% betrouwbaarheidsinterval [BI]: 0.88-1.80, p-waarde: 0.21). Volgens het model gecor-rigeerd voor leeftijd en geslacht hadden bètablokker gebruikers een 54% hoger risico op een opvlamming dan niet gebruikers. (HR: 1.54, 95% BI: 1.05-2.25, p-waarde :0.03). Discussie: Dit is de eerste studie die de associatie tussen bètablokker gebruik en opvlammingen in patiënten met IBD onderzoekt. Patiënten die een bètablokker gebruikten hadden een hoger risico op een opvlamming. Echter, een grote prospectieve studie is noodzakelijk om de relatie tussen bètablokkers en opvlammingen beter te onderzoeken. Mocht de in deze studie gevon- den associatie gevalideerd worden, dan kan dit belangrijke consequenties hebben voor de be-handeling van patiënten met IBD.Introduction
Inflammatory bowel disease (IBD) involves chronic inflammation of the gastrointestinal tract. IBD primarily includes ulcerative colitis (UC), Crohn’s disease (CD) and IBD of an unclassified type (IBD-u). UC only affects the colon, while CD can affect any part of the gastrointestinal tract. The incidence and prevalence of IBD are highest in North America and Europe, and incidence is still increasing.1,2 In the Netherlands, the incidence is approximately 29.2 per 100,000 inhab-itants per year, and prevalence is about 432.1 per 100,000 inhabitants.3 Both CD and UC are mostly diagnosed in early adulthood, between the ages of 15 and 30 years. The age distribution of IBD diagnosis is bimodal, with a second smaller peak between the ages of 50 and 70 years.4 Symptoms are similar in CD and UC, mainly consisting of abdominal pain, diarrhoea, weight loss and fatigue.5 The disease course of IBD is chronic and characterized by periodic relapses and remissions. A relapse is a flare of symptoms and remission is defined as the resolution of symp-toms and healing of the gastrointestinal tract.2 The pathophysiology of IBD is unclear, according to current theories multiple factors contribute to the development of IBD, including genetic, environmental and immunological factors.1 It is hypothesized that IBD results from a dysregulated mucosal immune response (immunological factor) to the commensal intestinal microbiota (environmental factor) in genetically susceptible individuals (genetic factor).2 The course of IBD remains unpredictable and currently there is no medical cure.4,6 Treatment of IBD is focused on inducing remission, maintaining remission and preventing complications.2,4 Medication used to treat IBD consists of anti-inflammatory drugs, i.e. 5-aminosalicylic acid (5- ASA), local and systemic steroids, immunosuppressive medication such as thiopurines and cy- closporine, and biological agents such as tumor-necrosis-factor-α (TNF-α) blockers, e.g. inflixi-mab and adalimumab.7,8 Response to medication differs among patients and sometimes sur-gery is necessary. Eventually, up to 75% of the patients with a diagnosis of CD and 25% to 33% of the patients with a diagnosis of UC require surgery.4IBD has a high impact both on the patient and on society. Due to the early onset of IBD and the fact that there is no cure, patients usually require a lifetime of medical follow-up. The majority of the patients experiences a relapsing disease course or have continuous active disease and a significant number of patients require surgery.2,4 Although mortality from IBD is low, morbidity is high and the quality of life is impaired. Considering the impact on society, IBD accounts for substantial costs, concerning both healthcare costs and societal costs due to missed workdays. In Europe, the direct healthcare costs are on average 2,000 to 3,000 euros per patient per year.2 Furthermore, 10% of the patients with IBD are unemployed, permanent work disability* is twice as likely in patients with IBD and on average patients spend 3-6 weeks per year on sick leave.2 Altogether, the disease burden of IBD can be considered as high. In summary, IBD has no cure, the incidence of IBD is increasing and the disease burden is high. Therefore it is of great clinical interest to study the disease course of IBD and factors that may be influencing the disease course.3,6 Drugs prescribed for diseases other than IBD might be such a factor. For this reason, we are interested in the influence of beta-blockers on the disease course of IBD.
BETA-BLOCKERS & INFLAMMATION
Beta-blockers are widely used and play a major role in the treatment of cardiovascular dis-eases.9 They are primarily prescribed for hypertension, angina pectoris, heart rhythm disorders, and chronic heart failure, but can also be prescribed for migraine and thyrotoxicosis†.9,10 Beta- blockers work by blocking β-adrenergic receptors, which are the main receptors for the neuro-transmitters of the sympathetic nervous system (SNS). The SNS is one of the two main branches of the autonomic nervous system, the other being the parasympathetic nervous system. The most important neurotransmitters of the SNS are epinephrine, norepinephrine and dopamine, also known as catecholamines. Activation of the SNS happens through the binding of catechol-amines to the adrenergic receptors, including β-adrenergic receptors. Beta-blockers block the * Permanent work disability here means that one is not able to work at all for permanent time. † Thyrotoxicosis a condition due to excess production of thyroid hormone.
β-adrenergic receptors, preventing catecholamines from binding to the receptors resulting in partial blockade of the SNS. It has been well established that the autonomic nervous system modulates the immune and inflammatory response.11,12 The SNS has also been studied in relation to intestinal inflamma-tion. Straub et al.13 showed that ablation of the SNS increased inflammation in experimental colitis, suggesting that the SNS has an anti-inflammatory influence. Nijhuis et al.14 showed that stimulating the sympathetic nervous system increased production of anti-inflammatory cyto-kines and inhibits pro-inflammatory cytokines in murine immune cells. These studies suggest that stimulating the SNS decreases inflammation and blocking the SNS increases inflammation. Beta-blockers work by inhibiting the SNS, and this could influence the inflammatory response in immune mediated diseases like IBD. To our knowledge, no studies examining the effect of beta-blockers on IBD in humans have been reported. However, there are animal studies that report an anti-inflammatory effect mediated by adrenergic receptors.15,16 Kang et al.15 studied the effect of electroacupuncture on experimental colitis in rats. They stated that electroacupuncture activates the SNS. They observed an anti-inflammatory effect of electroacupuncture on colitis, suggesting that activation of the SNS decreased inflam- mation. When the rats were given a beta-blocker, this effect was supressed. Kang et al. sug-gested that the anti-inflammatory effect of electroacupuncture is mediated via β-adrenergic receptors.15 Their results are in line with the idea that the SNS has an anti-inflammatory effect and that blocking the SNS with beta-blockers increases inflammation. When looking at the relation between beta-blockers and inflammation in a broader sense, sev- eral studies investigated this relation in the area of systemic inflammation, arthritis, heart fail-ure and stroke-induced bacterial infections.16,17,18,19,20,21,22,23,24 A study evaluating the effect of electroacupuncture, i.e. stimulating the SNS, on inflammation found similar results as the study by Kim et al.16 They found that in a murine model for inflam-mation electroacupuncture activated the SNS, and this decreased inflammation. Deactivating the SNS, either by chemical denervation or the administration of a beta-blocker, was found to suppress the anti-inflammatory effect of electroacupuncture. These results suggest the in-volvement of β-adrenergic receptors in the anti-inflammatory effect of electroacupuncture.16
This is line with the idea that the SNS has an anti-inflammatory effect and blocking the SNS with beta-blockers increases inflammation. Two mouse based studies investigated the effect of treatment with propanolol, a non-selective beta-blocker targeting all three subtypes of the adrenergic receptors. Results showed that administration of propanolol deteriorated clinical parameters and increased mortality in mice with systemic inflammation.17,18 This suggests that blocking the SNS with a beta-blocker increases inflammation. Another murine study19 reported that epinephrine, an important neu-rotransmitter of the SNS, reduced the production of pro-inflammatory cytokines.19 This is in line with the idea that the SNS has an anti-inflammatory influence. A study by Straub et al.20 also measured production of pro-inflammatory cytokines, in mice with arthritis. In accordance with the results of Deng et al.19, they found that stimulating the SNS inhibited the production of pro-inflammatory cytokines, i.e. that the SNS decreased inflammation.20 Unfortunately, not all evidence points in one direction and studies with contrasting results have also been published. For example, Nessler et al.21 observed that patients with heart failure treated with carvedilol, a non-selective beta-blocker, had significantly increased levels of pro-inflammatory cytokines. This suggests that blocking β-adrenergic receptors, i.e. blocking the SNS, has an anti-inflammatory effect.21 Pauschinger et al.22 reported similar results in mice and found a decrease in pro-inflammatory cytokines upon treatment with carvedilol in experi-mental inflammation of the heart, suggesting that blocking the SNS decreases inflammation.22 The relation between the non-selective beta-blocker propanolol and stroke-induced bacterial inflammation has been reported in two studies by Prass et al.23,24 Both studies found that treat- ment with the beta-blocker propanolol in mice with acute stroke prevented progression of bac-terial pneumonia, which also suggested that blocking β-adrenergic receptors, i.e. blocking the SNS, had an anti-inflammatory properties. In summary, there is evidence for a relation between beta-blockers and inflammation from studies investigating other inflammatory diseases in both humans and mice. Whether beta-blockers exert a pro-inflammatory or anti-inflammatory effect is not agreed upon, evidence is reported for both effects. No studies investigating the effect of beta-blockers on IBD have been reported, although it would be interesting to see whether the use of beta-blockers, which block the SNS, induces or reduces inflammation in patients with IBD. The frequency of relapses in IBD
can be a clinical measurement of the activity of the inflammatory disease. We used a retro-spective cohort to examine if patients with IBD using a beta-blocker had the same risk of a relapse as patients with IBD not using a beta-blocker. We hypothesised that patients using a beta-blocker are more likely to experience a relapse than patients not using a beta-blocker.
RESEARCH QUESTION & RELEVANCE
IBD has a high disease burden and disease relapses are unpredictable. It is necessary to inves-tigate factors that could influence disease course. Blocking adrenergic receptors could be such a factor and therefore we investigate the possible association between beta-blocker use and the number of relapses in patients with IBD. The research question of this thesis is as follows:
Is there an association between the use of beta-blockers and the number of disease re-lapses in patients with inflammatory bowel disease?
Relapses of IBD symptoms reflect increased inflammation. If an association between beta-blocker use and relapses is identified, this could be input for further investigation of the relation between beta-blockers and inflammation in IBD. If an actual relation exists, clinicians could an- ticipate on the effect of beta-blockers on the risk of a relapse. When beta-blockers are pre-scribed, clinicians might consider to prescribing alternative drugs if possible.
Methods
DATA SOURCES
The Pobasic cohort is a population based-cohort set up in 2010 by the department of Gastro-enterology at the Academic Medical Centre (AMC) in Amsterdam. The cohort was originally set up to determine the incidence and prevalence of IBD in the Netherlands. The Pobasic cohort includes 1461 adult patients who were under the care of a gastroenterologist or internal med-icine specialist at the AMC, Tergooiziekenhuizen or Flevoziekenhuis between 1 January 2004 and 1 January 2012. The AMC is a tertiary referral hospital, Tergooiziekenhuizen and Flevoziekenhuis are peripheral hospitals. Other inclusion criteria for entry into the Pobasic co-hort were a minimum age of 18 years and a diagnosis of IBD according to the Lennard Jones criteria (i.e. compatible endoscopic picture and clinical signs in absence of an infectious colitis, supported by histological examination of mucosal biopsy or surgical specimen). The Pobasic cohort contains patient data such as date of birth, gender, ethnicity, age at diagnosis, IBD type, prior IBD-related surgery‡ , smoking habits, prior IBD medication, disease activity and IBD phe-notype. These variables were measured at the time of patient inclusion. The Pobasic cohort has been previously described by Groof et al.3 For this study, we used patient data from the Pobasic cohort. In addition, we collected drug history data from the outpatient pharmacy. In 2011, drug history overviews were requested from the outpatient pharmacy for all patients in the Pobasic cohort that had provided prior informed consent. Drug history overviews were received in pa-per format for 372 patients. The drug history overviews consisted of a list of prescriptions with the prescription date, name of the drug, and dosage. The data from the Pobasic cohort and the drug history overviews were combined in order to evaluate disease relapses in patients using beta-blockers and patients not using beta- ‡ Ileocaecal resection, small bowel surgery, large bowel surgery, proctocolectomy, stricture plasty, perianal sur-gery and balloon dilatationblockers. Our primary outcome was the number of relapses, we used medication prescriptions as a proxy for disease relapses, this is described in the section “Assessment of primary out-come”.
STUDY POPULATION
The study population for this study was constructed from the Pobasic cohort, including all pa-tients for whom a drug history was available. Inclusion criteria were the availability of at least six months of a drug history and the presence of at least one prescription for the treatment of IBD, henceforth called ‘IBD medication’. The following medications were considered as IBD medications: 5-aminosalicylic acid (5-ASA), corticosteroids, thiopurines, methotrexate, ciclo-sporin, tacrolimus and anti-TNF agents.A patient was considered a beta-blocker user if there were prescriptions available for a beta-blocker in the drug history for at least three consecutive months, including all systemically acting non-selective and selective beta-blockers. Drug histories were manually reviewed to identify prescriptions for beta-blockers. The index date for patients not using beta-blockers was defined as the date of the earliest prescription available, on the condition that the patient was ≥18 years and had a diagnosis of IBD. The index date for patients using a beta-blocker was defined as the date of the earliest prescription after three months of beta-blocker use, on the condition that the patient was ≥18 years and had a diagnosis of IBD. Patients were censored at the date of their last available prescription or at end of the study, 1 January 2012. Reasons for exclusion were an unclear or incomplete drug history or suspicion of rheu-matoid arthritis based on drug prescriptions for rheumatoid arthritis by a rheumatologist. We considered it necessary to exclude those patients since IBD and rheumatoid arthritis are treated with similar medication, which could lead to an overestimation of relapses.
ASSESSMENT OF PRIMARY OUTCOME
The primary outcome was the occurrence of a relapse between the index date and end date. Relapses were identified by using drug prescriptions as a proxy. A relapse was defined as anincrease in medication level or the start of steroids, after a period of stable medication use of at least four months. Stable medication use was defined as no changes in medication level. The levels were defined as shown in Table 1. The order of the levels was based on the order in which IBD medication is prescribed according to the guidelines of the European Crohn’s and Colitis Organisation (ECCO).25,26 The ECCO guidelines state that the goal of the treatment of IBD is to induce remission and subsequently remain in a state of steroid-free remission. Therefore, ster- oid use can be considered merely as remission induction therapy and its prescription is indica-tive of a relapse.25,26 In order to assess the relapses for every patient, their drug history between the index date and the end of follow-up was reviewed manually. Within this period, every prescription involving IBD medication was assessed and a timeline was constructed in order to identify re-lapses. In Figure 1 an example of a timeline is presented. In this example, the index date was 1 January 2006 and the end date was 1 August 2007. 5-ASA is used chronically during the whole follow-up time, steroids were started three times and thiopurines were started on one occa-sion. A relapse is identified at the sixth month of 2006, since steroids are started and there were no changes in medication in the four prior months. Another relapse is identified in the first month of 2007, when steroids are started along with thiopurines. In the four months prior, no changes in medication were observed. In the sixth month of 2007, steroids are also started, but this is not considered a relapse, since there in the second month of 2007 steroids are stopped, and therefore there are changes in medication in the four months prior to the start of steroids.
Level 0 no IBD medication Level 1 5-ASA Level 2 thiopurines Level 3 methotrexate Level 4 Ciclosporin, tacrolimus, anti-TNF agents
Medication 2006 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 2007 5-ASA Steroids Thiopurines Relapse ↑ ↑
Figure 1 – Example of a timeline constructed from a patient’s drug history. Relapses are displayed with an arrow.
SENSITIVITY ANALYSIS
A sensitivity analysis was performed to evaluate the sensitivity of determining relapses using medication increase as a proxy, as described above. Eighteen patients were randomly selected. However, all patients were treated in the AMC, since their electronic patient records were avail-able to us. Notes and letters in the electronic patient record were analysed, and based on these notes and letters the number of relapses within the study period was determined. This number was compared to the number of relapses found with medication increase as a proxy.STATISTICAL ANALYSIS
To evaluate the similarity of the beta-blocker group and the non-beta-blocker group, a number of variables were compared between the groups. Baseline demographics that were compared included age, gender and ethnicity (i.e. Caucasian or other, including: Southeast Asian, Sub-Saharan African, North African, Chinese, Latin American). Disease-related variables included treatment centre, age at diagnosis, IBD type, IBD phenotype, disease activity at inclusion, prior IBD-related surgeries and the prior use of IBD medication. IBD phenotype was defined accord-ing to the Montreal classification.27 The Simple Clinical Colitis Activity Index (SCCAI) for ulcera-tive colitis and IBD-u and the Modified Harvey-Bradshaw index (HBI) for Crohn’s disease were used to assess disease activity. In addition, some variables that are known to influence disease activity were also compared between the two groups. Firstly, the number of prescriptions forantibiotics during the study period were assessed, since it has been hypothesised that admin-istration of antibiotics might provoke relapses.29 Secondly, undergoing an appendectomy is hy-pothesised to have a beneficial effect on the disease course in ulcerative colitis.30 Therefore we also compared the number of patients that had undergone an appendectomy in the beta-block-ers group and non-beta-blocker group. Thirdly, whether or not the patients had smoked in the last 20 years before inclusion in the Pobasic cohort was compared between the two groups. Smoking is known to be related to cardiovascular disease, and smoking also affects inflamma-tion in both UC and CD. Patients with UC who smoke have a less severe disease course, while in patients with CD smoking worsens the disease course.4 For categorical variables the Chi-squared test was used, or Fisher’s exact test in case of small sample sizes. For numerical variables, the Student’s t-test or Mann-Whitney test was used. To answer the question of whether patients with IBD using beta-blockers have the same risk of a relapse as patients with IBD not using beta-blockers, we used a regression model and the absolute risk of a relapse was calculated for both groups and expressed as the number of re-lapses per 100 person-years. The aim of survival analysis is to model and analyse time-to-event data.31 In this study, the event is a relapse, and the time to a relapse is modelled with a Cox proportional hazards (CoxPH) model. For the regression analysis, an extension of the CoxPH model was used which is called the shared frailty Cox proportional hazards model. Usually in analysis of time-to-event data only time until the first event is modelled. However, there are many clinical outcomes that recur in the same patient, such as sports injuries, cancer recurrences, or in our case relapses of IBD.32 A common characteristic among those recurrent events is the correlation between events oc-curring within one subject.32 When one uses the traditional CoxPH model to analyse recurrent events, the correlation between events within one subject is ignored, and this could generate misleading results.32,33 To avoid this problem, we used the shared frailty extension of the CoxPH model. In this model, a random component - or frailty term - is added to the model to adjust for the within-subject correlation. The idea is that the random component describes the intrin- sic susceptibility an individual has to experience an event, taking into account unmeasured het-erogeneity.33
Using the addition of a shared frailty component, multivariate CoxPH models were cre- ated. To prevent potential confounding, the model was adjusted for age and gender. We cal-culated both unadjusted and adjusted hazard ratios. Furthermore, models stratified for IBD type were created. Patients with IBD-u were excluded from this analysis. All statistical analyses were performed with R (version 3.1.2) with the use of the coxme package. Data management was via Excel 2011 and SPSS version 23.
Results
STUDY POPULATION
A total of 361 patients within the Pobasic cohort fulfilled the inclusion criteria and had an avail- able drug history. Of this group, 43 patients used a beta-blocker and 318 did not use a beta- blocker. After reviewing the drug history, in total 111 patients from the initial 361 were ex-cluded, leaving 250 patients in the analyses. 65 patients were excluded because they had less than six months of drug history available. A further 16 patients were excluded as there were no prescriptions of IBD medication available in their drug histories. Furthermore, 22 patients whose drug history was unclear or incomplete were excluded and seven patients were ex- cluded because of suspicion of a diagnosis of rheumatoid arthritis. Finally, one patient was ex-cluded because it was not clear when this patient was diagnosed with IBD and therefore the index date could not be determined. Of the 250 patients included in the study, 12.0% (30/250) patients used a beta-blocker and 88.0% (220/250) did not use a beta-blocker. Beta-blocker use was equal in the included and excluded patient group (11.9% vs 12.0%). The inclusion flow chart is presented in Figure 2. Prescriptions for five different beta-blockers or combinations of these were found in the drug histories. Most frequently a selective β1-blocker was prescribed, of which metoprolol was most frequently prescribed. In Table 2 the beta-blockers and frequency of prescriptions is summa-rized.
Beta-blocker Type Number of patients with
prescription Metoprolol Selective β1-blocker 17 (57%)
Atenolol Selective β1-blocker 4 (13%) Propanolol Non-selective β-blocker 3 (10%) Sotalol Non-selective β-blocker 2 (7%) Bisoprolol Selective β1-blocker 1 (3%)
Combination Both 3 (10%)
PATIENT CHARACTERISTICS
The patient characteristics are presented in Table 3. Of the 30 patients in the beta-blocker group 43.3% were male (13/30), in the non-beta-blocker group 41.4% were male (91/220). There was a significant difference in the mean age of the two groups (p<0.001). The beta- blocker group had a mean age of 64 (95% confidence interval (CI): 60.0-68.3), while in the non-beta-blocker group the mean age was 49 (95% CI: 46.6-50.5). The average duration of a drug history was 57 months for the beta-blocker group and 55 months for the non-beta-blocker group. The earliest drug history dated back to July 1991.
In both groups, most patients were treated in a peripheral hospital, i.e. Tergooiz-iekenhuizen. No patients in our study group were treated in Flevoziekenhuis, because at the time the drug histories were requested, no patients treated in Flevoziekenhuis had already pro-vided informed consent. The majority of the patients in the beta-blocker group had a diagnosis of ulcerative colitis (63.3%), followed by Crohn’s disease (33.3%) and IBD-u (3.3%). In the non-beta-blocker group the proportions of IBD type were similar; 55.0% UC, 36.4% CD and 8.6% IBD-u. Smoking habits and use of antibiotics were similar in patients using beta-blockers and patients not using beta-blockers. Conversely, the median age at diagnosis was significantly higher in the blocker group compared to the non-blocker group (median beta-blocker group: 51, median non-beta-blocker group: 35, p<0.001). The proportion of patients that had an appendectomy was 13.3% in the beta-blocker group versus 8.2% in the non-beta-blocker group (p: 0.05). Patients using beta-blockers were diagnosed with IBD later in life and more often underwent an appendectomy. This difference was due to the difference in age of the patient groups, when compared between patient groups of the same age, no significant differences remain in the age at diagnosis or the proportion of patients that underwent an appendectomy (results not shown).
Variable Beta-blocker No beta-blocker Significance
(n=30) (n=220) p=
Age [mean (standard deviation)] 64 (11.6) 49 (15.0) <0.001*
Male [n (%)] 13 (43.3) 91(41.4) 0.99
Ethnicity [n (%)] 0.43
Caucasian 26 (86.6) 204 (92.3)
Other 4 (13.3) 16 (7.3)
Follow-up in months [median (Q1-Q3)] 57 (42-75) 55 (23-105) 0.66
Hospital [n (%)] 0.68 AMC 8 (26.6) 71 (32.3) Tergooiziekenhuizen 22 (73.3) 149 (67.7) IBD type [n (%)] 0.60 Ulcerative colitis 19 (63.3) 121 (55.0) Crohn’s disease 10 (33.3) 80 (36.4) IBD-u 1 (3.3) 19 (8.6)
Age at diagnosis [median (Q1-Q3)] 51 (37-63) 35 (25-47) <0.001*
Prior IBD related surgeries [n (%)] 0.77
0 25 (83.3) 173 (78.6) 1 2 (6.6) 28 (12.7) >1 3 (10) 19 (8.7) Smoking† [n (%)] 0.95 Yes 14 (46.6) 105 (47.7) No 16 (53.3) 109 (49.5) Missing 0 (0) 6 (3) Appendectomy [n (%)] 6 (13.3) 18 (8.2) 0.05
Antibiotics prescriptions per year [n (%)] 0.40
0 23 (76.6) 183 (83.2)
1 5 (16.7) 23 (10.5)
>1 2 (6.7) 14 (6.3)
Table 3. Patient characteristics of patients using beta-blockers versus not using beta-blockers.
*Significant with α=0.05
† Smoking is defined as being a smoker in the last 20 years.
Abbreviations: AMC=Academic Medical Centre, IBD=inflammatory bowel disease, UC=ulcerative colitis, CD=Crohn’s disease, IBD-u= inflammatory bowel disease unclassified.
CLINICAL CHARACTERISTICS
Table 4 details the clinical characteristics of the two groups. Considering the IBD phenotype (Montreal classification), disease activity and medical therapy, no significant differences were observed between the beta-blocker and non-beta-blocker groups. With regard to the disease activity, 55% of the patients with UC and IBD-u in the beta- blocker group were in remission at inclusion in the Pobasic cohort compared to 65% in the non-beta-blocker group. 60% of the patients with CD, in the beta-blocker group were in remission compared to 58.8% of the patients in the non-beta-blocker group. Medical therapy was similar in patients using beta-blockers and patients not using beta- blockers. Over 80% of the patients had used 5-ASA therapy before inclusion in the Pobasic co-hort. Systemic steroids and local steroids were prescribed at some point during the follow-up to 43.3% and 33.3% of the patients in the beta-blocker group and to 50.5% and 30.9% of the patients in the non-beta-blocker group respectively. Anti-TNF was prescribed to 6.7% in the beta-blocker group and to 7.7% in the non-beta-blocker group.FREQUENCY OF RELAPSES
Overall, 380 relapses were observed in 250 patients, varying from zero to nine relapses per patient. Out of 250 patients, 62.8% of the patients experienced at least one relapse (157/250) and 63.1% of those patients experienced subsequent relapses (99/157). In the beta-blocker group, 60.0% of the patients experienced at least one relapse (18/30), compared to 63.2% in the non-beta-blocker group (139/220). The follow up of the 250 patients included in this study, corresponded to 1338 person-years of follow up. This translates to 28 relapses per 100 person- years (95% CI: 25.5-31.3). In the beta-blocker group, 33 relapses were observed in 155 person-years, corresponding to 21 per 100 person-years (95% CI: 14.0-28.6). In the non-beta-blocker group in total 347 relapses were observed in 1184 person-years, corresponding to 29 relapses per person-years (95% CI: 26.2-32.4).
Variable [n (%)] Beta-blocker No beta-blocker Significance
(n=30) (n=220) p=
Montreal classification UC/IBD-ua n=20 n=140
Extent 0.25 E1: proctitis 4 (20) 24 (17.1) E2: left-sided UC 12 (60) 60 (42.9) E3: extensive UC 4 (20) 54 (38.6) Missing 0 (0) 2 (1.4) Severity 0.72 S1: mild 4 (20) 30 (21.4) S2: moderate 10 (50) 76 (54.3) S3: severe 4 (20) 19 (13.6) Missing 2 (10) 15 (10.7) Montreal classification CD n=10 n=80 Age at diagnosis 0.35
A1: below 16 years 0 (0) 9 (11.3)
A2: between 17 and 40 years 5 (50) 49 (61.3)
A3: above 40 years 5 (50) 22 (27.5)
Missing 0 (0) 0 (0) Localization 0.13 L1: ileal 2 (20) 23 (28.8) L2: colonic 3 (30) 25 (31.3) L3: ileocolonic 2 (20) 28 (35) L4: upper GI 2 (20) 2 (2.5) Missing 1 (10) 2 (2.5) Behaviour 0.91 B1: non-stricturing, non-penetrating 4 (40) 32 (40.0) B2: stricturing 3 (30) 20 (25.0) B3: penetrating 2 (20) 22 (27.5) Missing 1 (10) 6 (7.5)
Disease activity UC/IBD-ub at inclusion n=20 n=140
SCCAIc = 0.79
<3 (remission) 11 (55.0) 91 (65.0) >3 (active disease) 9 (45.0) 48 (34.3)
Missing 0 (0.0) 1 (0.7)
Disease activity CD at inclusion n=10 n=80
HBId = 0.55 <5 (remission) 6 (60) 47 (58.8) 5-7 (mild) 1 (10) 20 (25.0) 8-16 (moderate) 3 (30) 11 (13.8) >16 (severe) 0 (0) 1 (1.3) Missing 0 (0) 1 (1.3) Medical therapy n=30 n=220 5-ASA 25 (83.3) 184 (83.6) 1.00 Steroids systemic 13 (43.3) 111 (50.5) 0.59 Steroids local 10 (33.3) 68 (30.9) 0.95 Thiopurines 8 (26.7) 97 (44.1) 0.11 Methotrexate 1 (3.3) 8 (3.6) 1.00 Ciclosporin 0 (0) 4 (1.8) 1.00 Anti-TNF 2 (6.7) 17 (7.7) 1.00
Table 4. Clinical characteristics in patients using beta-blockers versus not using beta-blockers.
a The Montreal Classification for Ulcerative Colitis was also used for patients with IBD-u. b The Simple Clinical Colitis Activity Index (SCCAI) was also used for patient with IBD-u. c Cut-off value for remission was based on a publication by Higgins et al34 d Levels of HBI (remission, mild, moderate and severe) were based on the original publication of the HBI35 Abbreviations: UC=ulcerative colitis, CD=Crohn’s disease, IBD-u= inflammatory bowel disease unclassified, SCCAI=simple clinical colitis activity index, HBI=Harvey Bradshaw index, 5-ASA=5-amino-salicylic-acid, TNF=tumor necrosis factor, GI=gas-trointestinal.
SENSITIVITY ANALYSIS
To determine the accuracy of the method used to identify relapses by means of increases in IBD medication, a sensitivity analysis was performed. Notes and letters from the patient records of eighteen randomly selected patients were reviewed, and relapses mentioned in the patient record were listed. The number of relapses were compared to the number of relapses detected by means of medication increases, i.e. the method used in this study. Results of this analysis are shown in Table 5. In total, 24 relapses were identified by our method and 25 relapses were identified in the notes. In 55.6% (10/18 patients), the number of relapses calculated from the drug history corresponded exactly to the notes. In 27.8% (5/18 patients), our method overes-timated the number of relapses whereas in 22.2% (4/18 patients), our method underestimated the number of relapses. Patient Relapses in patient record Relapses by medication use Difference 1 2 2 = 2 3 2 -1 3 0 0 = 4 3 3 = 5 0 1 +1 6 1 2 +1 7 0 0 = 8 2 2 = 9 1 1 = 10 2 0 -2 11 2 5 +3 12 0 1 +1 13 3 0 -3 14 2 2 = 15 0 1 +1 16 0 0 = 17 1 1 = 18 2 2 = Total 24 25 +1Table 5. Sensitivity analysis results: detection of relapses with
medication use as a proxy versus relapses in the patient record.
COX PROPORTIONAL HAZARDS MODELS
Results from the shared frailty CoxPH models are presented in Table 6. The unadjusted model showed a 26% increase in the risk of a relapse in the beta-blocker group compared to the non-beta-blocker group (HR: 1.26, 95% CI: 0.88-1.80, p: 0.21) over a median follow-up time of 55 months. The adjusted risk of a relapse in the beta-blocker group was 54% higher than in the non-beta-blocker group (adjusted HR: 1.54, 95% CI: 1.05-2.25, p: 0.03) over a median follow-up time of 55 months. Both the unadjusted model and the model adjusted for age and sex suggest that patients using a beta-blocker are more likely to have a relapse than patients not using a beta-blocker. Also, in both models, the frailty effect was found to be significant with a p-value smaller than 0.001. This means that there was a significant correlation between events observed within the same subject, underscoring the need for extension of the CoxPH model with shared frailty. Variable Estimates
Hazard ratio (HR) 95% CI p-value
Unadjusted Beta-blocker use 1.26 0.88-1.80 0.21
Frailty effect - - <0.001**
Adjusted* Beta-blocker use 1.54 1.05-2.25 0.03**
Frailty effect - - <0.001**
* Adjusted for age and gender
Table 6. Output for the unadjusted and unadjusted shared frailty Cox proportional hazards model.
** Significant with α=0.05
STRATIFIED COXPH MODELS
To determine whether there was a difference in the hazard ratio between patients using beta-blockers or not within the group of patients with UC or patients with CD, shared frailty CoxPH models stratified for IBD type were calculated. The results of the stratified models are shown in Table 7.For patients with UC in the beta-blocker group, the risk of a relapse was 33% higher than in the non-beta-blocker group (adjusted HR: 1.33, 95% CI: 0.83-2.15, p: 0.24) over a me-dian follow-up time of 54 months. For patients with CD in the beta-blocker group there was a 59% increase in the risk of a relapse compared to the non-beta-blocker group (adjusted HR: 1.59, 95% CI: 0.78-3.22, p: 0.20) over a median follow-up time of 55 months. The frailty effect was significant in both models, implying a significant correlation between intra-subject events. Variable Estimates
Hazard ratio (HR) 95% CI p-value
Ulcerative colitis
Adjusted* Beta-blocker use 1.33 0.83-2.15 0.240
Frailty effect - - <0.001
Crohn’s disease
Adjusted* Beta-blocker use 1.59 0.78-3.22 0.200
Frailty effect - - <0.001
* Adjusted for age and gender
Table 7. Output for Cox regression model stratified for IBD type
Altogether, all four models suggest that the use of beta-blockers is associated with an increase in the risk of relapse, varying from a 33% increase to a 59% increase. In fact, the adjusted model including both UC and CD reports a significant hazard ratio of 1.54, implying a 54% increased risk of relapse in the beta-blocker group over a median follow-up time of 55 months.
Discussion
MAIN FINDINGS
IBD has a high disease burden and disease relapses are unpredictable, therefore it is necessary to identify factors which could influence the disease course and provoke or protect patients from relapses. In this retrospective cohort study we evaluated the possible association between beta-blocker use and relapses in patients with IBD. We found on average 21 relapses per 100 person-years in the beta-blocker group and 29 relapses per 100 person-years in the non-beta-blocker group. Patients in the beta-blocker group were found to have an 54% higher adjusted risk of relapse compared to patients in the non-beta-blocker group. The models stratified for IBD type also showed a trend towards an increased risk of relapse in patients in the beta-blocker group. However, these results did not reach statistical significance, this is probably due to the small sample size in the stratified analysis. When stratified by disease type, we found a 33% higher adjusted risk of relapse in the patients with UC taking beta-blockers compared to pa-tients with UC not taking beta-blockers. A 59% higher adjusted risk of relapse was found for patients with CD in the beta-blocker group compared to patients with CD in the non-beta-blocker group.COMPARISON TO OTHER STUDIES
To our knowledge, this is the first study to examine the association between beta-blocker use and relapses in patients with IBD. There have been studies reporting on the inflammatory ef-fects mediated by adrenergic receptors.13,14,36 Elsenbruch et al.36 showed that the adrenergicagonist isoproterenol, i.e. stimulating adrenergic receptors, suppresses TNF-α production in patients with ulcerative colitis. TNF-α is a pro-inflammatory cytokine that plays an important
role in the inflammatory process seen in IBD. The finding that stimulating the sympathetic nerv-ous system reduces TNF-α production in patients with UC would imply that stimulating the SNS has an anti-inflammatory effect in patients with UC. In contrast with these findings, a number of studies23,24,37 supporting the idea that acti- vation of the sympathetic nervous system worsens inflammation have been published. For ex-ample, Chi et al.37 report that epinephrine, i.e. an agonist of the SNS activating adrenergic re-ceptors, upregulates the pro-inflammatory cytokine production of human mast cells, and this effect is reduced by the beta-blocker propanolol. These findings suggest that sympathetic ac-tivity increases inflammation and beta-blockers reverse this effect. We found that patients with IBD using beta-blockers have an increased risk of relapse compared to patients not using beta-blockers. The association of beta-blockers with relapses of IBD in humans has not been studied before, although there is sufficient evidence that ma-nipulating sympathetic activity via adrenergic receptors has its – either pro-inflammatory or anti-inflammatory – effects on inflammation. The results from this study support the hypothe-sis that blocking sympathetic activity via adrenergic receptors increases inflammation.
STRENGTHS & LIMITATIONS
This study has several strengths. The choice of the Cox proportional hazards model with shared frailty allowed us to improve the precision of the estimate of effect and the related confidence interval. The Cox proportional hazards model is the most appropriate choice of statistical tech- niques for analysing recurrent time to event data. The statistical challenge to analysing recur-rent events is to take into account correlation between events occurring within one subject. Ignoring the correlated nature of the data could lead to artificially narrow confidence intervals and possible biasing away from the null32,33. Extending the CoxPH model with shared frailty allowed us to model recurrent events while taking into account the intra-subject correlation. A number of limitations also need to be considered when interpreting the findings of this study. First, the results found in this study might be biased by the method of selecting patients, which might lead to a sample not representative of the general population of patients with IBD. When patients were selected for the Pobasic cohort, recruitment of patients treated in the AMC washandled differently than for patients treated in Tergooiziekenhuizen, leading to the risk of se-lecting patients with a higher disease activity in the group of AMC patients. However, Groof et al.3 report that there was no difference in disease activity score for patients treated in the AMC
versus the patients treated in Tergooiziekenhuizen. In addition, in our patient sample we did not observe a difference in the proportion of patients treated in the AMC between the beta-blocker group and the non-beta-not observe a difference in the proportion of patients treated in the AMC between the beta-blocker group, therefore potential selection bias should equally affect both groups. Second, the results found in this study might be confounded by differences between the beta-blocker group and the non-beta-blocker group. We found differences in age, age at diagnosis and the proportion of patients that underwent an appendectomy. A difference in age was expected, since patients with an indication for a beta-blocker are typically older. The dif-ferences in age at diagnosis and the proportion of patients that underwent an appendectomy. These differences are most likely caused by the difference in age in the two groups, which we adjusted for. Although we adjusted for age and gender, there may be residual confounding. Furthermore, there are potential confounders we were not able to adjust for due to the lack of data. For example, we did not adjust for body mass index, which is known to be associated with both cardiovascular diseases and IBD.37 In addition, we did not adjust for smoking, due to the lack of detailed data on smoking behaviour. Third, the results found in this study might be biased due to observation bias. While reviewing the drug histories the reviewer was aware of the hypothesis which might have led to systematic differences in identifying relapses. Fourth, the method of identifying relapses remains a point for discussion. There are several reasons why this method would not be adequate to identify actual relapses. Firstly, prescriptions handled by the outpatient pharmacy do not necessarily reflect medication use. Medication can be prescribed but not used, conversely, medication that is used is not always prescribed via the outpatient pharmacy. For example, patients can get their medication via the inpatient pharmacy of a hospital, or via another outpatient pharmacy. Furthermore, cortico-steroids can be administered intravenously, this will not be stated in the outpatient pharmacy drug history. This also holds for the anti-TNF agents, which are primarily administered intrave-nously. Missing this information could lead to an underestimation of the number of relapses. Secondly, changing medication type or starting corticosteroids treatment does not necessarily mean there is a relapse. For example, it could be that the patient experiences side effects from
his or her medication and therefore changes to another type of medication. Thirdly, the results of the sensitivity analysis show that our method for identifying relapses is not perfect. Identify-ing relapses with notes and letters in the patient record might also not reflect the true number and moment of relapses, but we assumed that this method was quite reliable, and the discrep-ancy in the results from the two methods shows that in some cases we underestimated the number of relapses, and in some cases overestimated the number of relapses. Nevertheless, miscalculating the number of relapses is expected to happen equally in the beta-blocker group and the non-beta-blocker group.
CONCLUSION & IMPLICATIONS
In this retrospective cohort study, use of beta-blockers in patients with IBD was associated with a 54% increased risk of relapse in a median follow-up time of 55 months. While the findings of this study suggest that beta-blockers are a risk factor for relapses in patients with IBD, the ob-servational nature of this study means that it is hypothesis generating at best. Our study is the first to evaluate the association between beta-blocker use and relapses in patients with IBD and the findings highlight the need to further investigate the association between beta-block-ers and disease relapses in patients with IBD. Large prospective trials to examine the relation of beta-blocker use and relapses in IBD are necessary before potential guideline responses could be generated. Also, it would be interest-ing to determine whether the association between beta-blockers and relapses can be at-tributed to the medical indication for the beta-blocker or the beta-blocker itself. However, if it is apparent that beta-blockers affect the relapse rate in patients with IBD, this would have im-portant consequences, especially as beta-blockers are widely prescribed. Management of IBD could be improved, by anticipating the effect of beta-blockers on the risk of relapse. Clinicians could consider to prescribe alternative drugs to patients with IBD if possible.References
1. Loftus, E. V. (2004). Clinical epidemiology of inflammatory bowel disease: incidence, prevalence, and environmental influences. Gastroenterology, 126(6), 1504-1517. 2. Burisch, J., Jess, T., Martinato, M., & Lakatos, P. L. (2013). The burden of inflammatory
bowel disease in Europe. Journal of Crohn's and Colitis, 7(4), 322-337.
3. de Groof, E. J., Rossen, N. G., van Rhijn, B. D., Karregat, E. P., Boonstra, K., Hageman, I., ... & Mallant-Hent, R. C. (2016). Burden of disease and increasing prevalence of inflammatory bowel disease in a population-based cohort in the Netherlands.
European Journal of Gastroenterology & Hepatology.
4. Hanauer, S. B. (2006). Inflammatory bowel disease: epidemiology, pathogenesis, and therapeutic opportunities. Inflammatory bowel diseases, 12(5), S3-S9.
5. Uptodate.com, Clinical manifestations, diagnosis, and prognosis of ulcerative colitis in adults, Topic 4067 Version 26.0. Last visited August 10th, 2016
6. Bernstein, C. N., Singh, S., Graff, L. A., Walker, J. R., Miller, N., & Cheang, M. (2010). A prospective population-based study of triggers of symptomatic flares in IBD. The
American journal of gastroenterology, 105(9), 1994-2002. G. S. Seo, and S. C. Chae, 'Biological Therapy for Ulcerative Colitis: An Update', World J Gastroenterol, 20 (2014), 13234-8.
7. Kalla, R., Ventham, N. T., Satsangi, J., & Arnott, I. D. (2014). Crohn’s disease. Bmj,
349(7984), 27-31.
8. López-Sendó, J., Swedberg, K., McMurray, J., Tamargo, J., Maggioni, A. P., Dargie, H., ... & Pedersen, C. T. (2004). Expert consensus document on β-adrenergic receptor block ers. European Heart Journal, 25(15), 1341-1362. 9. https://www.farmacotherapeutischkompas.nl/bladeren-volgens-boek/inleidingen/inl- beta-receptorblokkerende-sympathicolytica Last consulted on July 15th 2016. 10. Flierl, M. A., Rittirsch, D., Huber-Lang, M., Sarma, J. V., & Ward, P. A. (2008). Catechola- mines--Crafty Weapons in the Inflammatory Arsenal of Immune/Inflammatory Cells or Opening Pandora's Box?. Molecular medicine, 14. 11. Madden, K. S., Sanders, V. M., & Felten, D. L. (1995). Catecholamine influences and sym- pathetic neural modulation of immune responsiveness. Annual review of pharmacol- ogy and toxicology, 35(1), 417-448. 12. Straub, R. H., Grum, F., Strauch, U., Capellino, S., Bataille, F., Bleich, A., ... & Obermeier, F. (2008). Anti-inflammatory role of sympathetic nerves in chronic intestinal inflamma- tion. Gut, 57(7), 911-921.
13. Nijhuis, L. E., Olivier, B. J., Dhawan, S., Hilbers, F. W., Boon, L., Wolkers, M. C., ... & de Jonge, W. J. (2014). Adrenergic β2 receptor activation stimulates anti-inflammatory properties of dendritic cells in vitro. PloS one, 9(1), e85086.
14. Kang, J. W., Kim, T. W., La, J. H., Sung, T. S., Kim, H. J., Kwon, Y. B., ... & Yang, I. S. (2004). Electroacupuncture ameliorates experimental colitis induced by acetic acid in rat.
Journal of veterinary science, 5(3), 189-196.
15. Kim, H. W., Kang, S. Y., Yoon, S. Y., Roh, D. H., Kwon, Y. B., Han, H. J., ... & Lee, J. H. (2007). Low-frequency electroacupuncture suppresses zymosan-induced peripheral inflame mation via activation of sympathetic post-ganglionic neurons. Brain research, 1148, 69-75.
16. Schmitz, D., Wilsenack, K., Lendemanns, S., Schedlowski, M., & Oberbeck, R. (2007). β- Adrenergic blockade during systemic inflammation: impact on cellular immune func- tions and survival in a murine model of sepsis. Resuscitation, 72(2), 286-294. 17. Oberbeck, R., Schmitz, D., Wilsenack, K., Schüler, M., Pehle, B., Schedlowski, M., & Exton, M. S. (2004). Adrenergic modulation of survival and cellular immune functions during polymicrobial sepsis. Neuroimmunomodulation, 11(4), 214-223. 18. Deng, J., Muthu, K., Gamelli, R., Shankar, R., & Jones, S. B. (2004). Adrenergic modulation of splenic macrophage cytokine release in polymicrobial sepsis. American Journal of
Physiology-Cell Physiology, 287(3), C730-C736.
19. Straub, R. H., Rauch, L., Rauh, L., & Pongratz, G. (2011). Sympathetic inhibition of IL-6, IFN- γ, and KC/CXCL1 and sympathetic stimulation of TGF-β in spleen of early arthritic mice.
Brain, behavior, and immunity, 25(8), 1708-1715.
20. Nessler, J., Nessler, B., Kitliński, M., Gackowski, A., Piwowarska, W., & Stepniewski, M. (2008). Concentration of BNP, endothelin 1, pro-inflammatory cytokines (TNF-alpha,
IL-6) and exercise capacity in patients with heart failure treated with carvedilol. Kardi- ologia polska, 66(2), 144-51.
21. Pauschinger, M., Rutschow, S., Chandrasekharan, K., Westermann, D., Weitz, A., Schwimmbeck, L. P., ... & Schultheiss, H. P. (2005). Carvedilol improves left ventricular function in murine coxsackievirus-induced acute myocarditis Association with reduced myocardial interleukin-1β and MMP-8 expression and a modulated immune response.
European journal of heart failure, 7(4), 444-452.
22. Prass, K., Meisel, C., Höflich, C., Braun, J., Halle, E., Wolf, T., ... & Volk, H. D. (2003). Stroke- induced immunodeficiency promotes spontaneous bacterial infections and is medi- ated by sympathetic activation reversal by poststroke T helper cell type 1–like im- munostimulation. The Journal of experimental medicine, 198(5), 725-736.
23. Prass, K., Braun, J. S., Dirnagl, U., Meisel, C., & Meisel, A. (2006). Stroke propagates bacte- rial aspiration to pneumonia in a model of cerebral ischemia. Stroke, 37(10), 2607- 2612.
24. Dignass, A., Lindsay, J. O., Sturm, A., Windsor, A., Colombel, J. F., Allez, M., ... & Öresland, T. (2012). Second European evidence-based consensus on the diagnosis and management of ulcerative colitis part 2: current management. Journal of Crohn's and
Colitis, 6(10), 991-1030.
25. Dignass, A., Van Assche, G., Lindsay, J. O., Lémann, M., Söderholm, J., Colombel, J. F., ... & Hommes, D. W. (2010). The second European evidence-based consensus on the diagnosis and management of Crohn's disease: current management. Journal of
Crohn's and Colitis, 4(1), 28-62
26. Satsangi, J., Silverberg, M. S., Vermeire, S., & Colombel, J. (2006). The Montreal classi- fication of inflammatory bowel disease: controversies, consensus, and implications. Gut, 55(6), 749-753.
27. Meyer, A. M., Ramzan, N. N., Heigh, R. I., & Leighton, J. A. (2006). Relapse of inflammatory bowel disease associated with use of nonsteroidal anti-inflammatory drugs. Digestive
diseases and sciences, 51(1), 168-172.
28. Bernstein, C. N., Singh, S., Graff, L. A., Walker, J. R., Miller, N., & Cheang, M. (2010). A prospective population-based study of triggers of symptomatic flares in IBD. The
American journal of gastroenterology, 105(9), 1994-2002.
29. Gardenbroek, T. J., Eshuis, E. J., Ponsioen, C. I. J., Ubbink, D. T., D’Haens, G. R. A. M., & Bemelman, W. A. (2012). The effect of appendectomy on the course of ulcerative colitis: a systematic review. Colorectal Disease, 14(5), 545-553.
30. Jager, K. J., van Dijk, P. C., Zoccali, C., & Dekker, F. W. (2008). The analysis of survival data: the Kaplan–Meier method. Kidney international, 74(5), 560-565.
31. Amorim, L. D., & Cai, J. (2015). Modelling recurrent events: a tutorial for analysis in epidemiology. International journal of epidemiology, 44(1), 324-333.
32. Ullah, S., Gabbett, T. J., & Finch, C. F. (2012). Statistical modelling for recurrent events: an application to sports injuries. British journal of sports medicine, bjsports-2011.
33. Higgins, P. D. R., Schwartz, M., Mapili, J., Krokos, I., Leung, J., & Zimmermann, E. M. (2005). Patient defined dichotomous end points for remission and clinical improve- ment in ulcerative colitis. Gut, 54(6), 782-788.
34. Harvey, R. F., & Bradshaw, J. M. (1980). A simple index of Crohn's-disease activity. The
Lancet, 315(8167), 514.
35. Elsenbruch, S., Langhorst, J., Popkirowa, K., Müller, T., Luedtke, R., Franken, U., ... & Schedlowski, M. (2005). Effects of mind-body therapy on quality of life and neuroen-
docrine and cellular immune functions in patients with ulcerative colitis. Psychother- apy and psychosomatics, 74(5), 277-287.
36. Chi, D. S., Fitzgerald, S. M., Pitts, S., Cantor, K., King, E., Lee, S. A., ... & Krishnaswamy, G. (2004). MAPK-dependent regulation of IL-1-and β-adrenoreceptor-induced inflamma tory cytokine production from mast cells: Implications for the stress response. BMC
immunology, 5(1), 1.
37. Dong, J., Chen, Y., Tang, Y., Xu, F., Yu, C., Li, Y., ... & Dai, N. (2015). Body Mass Index Is Associated with Inflammatory Bowel Disease: A Systematic Review and Meta-Analysis.
