An introduction to Random Forests for modeling tasks - theory and practice in python with public health case studies

RFs are a powerful, volatile, and easy to use method, making it an excellent benchmark method for different types of analysis, notably in the public health domain. RFs offer outstanding performance with minimal effort from the user (given its low sensitivity to its hyperparameters) and can be used for many different modeling tasks (classification, regression, clustering, outlier detection). In practice, RFs are easily applicable through libraries in Python and R languages, allowing users to benefit from their capabilities with minimal coding knowledge. In addition to their pure performance abilities, RFs have some practical advantages when compared to many classical statistical models: it does not require any normalization of the data, handles very large datasets (in population or variables) and all kinds of data types (e.g., binary, categorical, continuous); it handles outliers and is insensitive to multicollinearity within the input variables. Their main limitation is their less straightforward interpretation of the final model they build. However, they offer additional tools, such as variable importance and proximity metrics, that improve the understanding of their results and potentially provide insights that traditional models cannot. Finally, its construction based on decision trees grants it additional capabilities, notably the possibility to extract prediction intervals and to handle efficiently imbalanced data problems with a variant of the algorithm called Balanced Random Forests.

Introduction and purpose

Dashboards are health information tools, made to illustrate specific indicator sets. They may demonstrate developments over time, geographic distribution, social inequalities or correlations between different factors, use a wide variety of infographics, and are generally updated very frequently. Dashboards can illustrate data for users without statistical knowledge, and are also useful data repositories for more proficient users. Dashboards are a modern approach to an old problem: data visualization. A well-known historical example is John Snow's mapping of cholera cases in London (1854). Current computer technology and the existence of the internet have made it possible to visualize health indicators in close to real time, and dashboards, like Dr Snow's map, are meant to, store, report and illustrate health indicators in order to affect policy. The indicators selected should be the ones amenable to change because their unsurpassed ability to monitor changing conditions and effects of interventions. Indicator selection is vital to set a baseline for such monitoring, and the indicators selected will of necessity be the ones where measures are taken in order to affect change.

Policy will influence the choice of indicators, but indicators will also influence the choice of policies, because these two are intimately intertwined. Dashboards, be they national or regional, open to the public or just open to professional groups, are important implements to reach the policy-making audience and to signal what is defined as important. Dashboards can help to formulate policies that will improve public health for all groups.

An overview of AI and Machine Learning methods: motivations, concepts, and examples

What is artificial intelligence? Big Data? Machine learning? It may seem very difficult to define these terms and identify their use in real life, aside from the hype surrounding them over the past ten years. All of these terms are interconnected - sometimes they intersect, often they are included in each other. The main focus of this presentation is to present ‘Machine learning' methods, which can be defined as algorithms that learn patterns (a function) about a particular phenomenon (an output variable), based on examples (training data), often without any a priori on the shape of these patterns. Therefore, they are a broad family of methods that include traditional statistical methods such as linear regression models, but which often offer more flexibility in terms of hypothesis and complex data settings. Multiple frameworks exist, based on how many examples are available (size of the training data) and the kind of modeling application one wants to perform (e.g., classification, regression, clustering). Many different families of methods exist for each modeling application, which offers a tremendous amount of different algorithms, each having advantages and drawbacks. In practice, easy-to-use libraries were developed in popular programming languages such as Python and R, offering implementations of the majority of essential algorithms.

Impact of the early use of immunomodulators or TNF antagonists on bowel damage and surgery in Crohn's disease

BACKGROUND

The impact of early treatment with immunomodulators (IM) and/or TNF antagonists on bowel damage in Crohn's disease (CD) patients is unknown.

AIM

To assess whether 'early treatment' with IM and/or TNF antagonists, defined as treatment within a 2-year period from the date of CD diagnosis, was associated with development of lesser number of disease complications when compared to 'late treatment', which was defined as treatment initiation after >2 years from the time of CD diagnosis.

METHODS

Data from the Swiss IBD Cohort Study were analysed. The following outcomes were assessed using Cox proportional hazard modelling: bowel strictures, perianal fistulas, internal fistulas, intestinal surgery, perianal surgery and any of the aforementioned complications.

RESULTS

The 'early treatment' group of 292 CD patients was compared to the 'late treatment' group of 248 CD patients. We found that 'early treatment' with IM or TNF antagonists alone was associated with reduced risk of bowel strictures [hazard ratio (HR) 0.496, P = 0.004 for IM; HR 0.276, P = 0.018 for TNF antagonists]. Furthermore, 'early treatment' with IM was associated with reduced risk of undergoing intestinal surgery (HR 0.322, P = 0.005), and perianal surgery (HR 0.361, P = 0.042), as well as developing any complication (HR 0.567, P = 0.006).

CONCLUSIONS

Treatment with immunomodulators or TNF antagonists within the first 2 years of CD diagnosis was associated with reduced risk of developing bowel strictures, when compared to initiating these drugs >2 years after diagnosis. Furthermore, early immunomodulators treatment was associated with reduced risk of intestinal surgery, perianal surgery and any complication.

The gap between scientific evidence and clinical practice: 5-aminosalicylates are frequently used for the treatment of Crohn's disease

BACKGROUND

There is uncertain evidence of effectiveness of 5-aminosalicylates (5-ASA) to induce and maintain response and remission of active Crohn's disease (CD), and weak evidence to support their use in post-operative CD.

AIM

To assess the frequency and determinants of 5-ASA use in CD patients and to evaluate the physicians' perception of clinical response and side effects to 5-ASA.

METHODS

Data from the Swiss Inflammatory Bowel Disease Cohort, which collects data since 2006 on a large sample of IBD patients, were analysed. Information from questionnaires regarding utilisation of treatments and perception of response to 5-ASA were evaluated. Logistic regression modelling was performed to identify factors associated with 5-ASA use.

RESULTS

Of 1420 CD patients, 835 (59%) were ever treated with 5-ASA from diagnosis to latest follow-up. Disease duration >10 years and colonic location were both significantly associated with 5-ASA use. 5-ASA treatment was judged to be successful in 46% (378/825) of treatment episodes (physician global assessment). Side effects prompting stop of therapy were found in 12% (98/825) episodes in which 5-ASA had been stopped.

CONCLUSIONS

5-Aminosalicylates were frequently prescribed in patients with Crohn's disease in the Swiss IBD cohort. This observation stands in contrast to the scientific evidence demonstrating a very limited role of 5-ASA compounds in the treatment of Crohn's disease.

A better sequence-read simulator program for metagenomics

Background

There are many programs available for generating simulated whole-genome shotgun sequence reads. The data generated by many of these programs follow predefined models, which limits their use to the authors' original intentions. For example, many models assume that read lengths follow a uniform or normal distribution. Other programs generate models from actual sequencing data, but are limited to reads from single-genome studies. To our knowledge, there are no programs that allow a user to generate simulated data following non-parametric read-length distributions and quality profiles based on empirically-derived information from metagenomics sequencing data.

Results

We present BEAR (Better Emulation for Artificial Reads), a program that uses a machine-learning approach to generate reads with lengths and quality values that closely match empirically-derived distributions. BEAR can emulate reads from various sequencing platforms, including Illumina, 454, and Ion Torrent. BEAR requires minimal user input, as it automatically determines appropriate parameter settings from user-supplied data. BEAR also uses a unique method for deriving run-specific error rates, and extracts useful statistics from the metagenomic data itself, such as quality-error models. Many existing simulators are specific to a particular sequencing technology; however, BEAR is not restricted in this way. Because of its flexibility, BEAR is particularly useful for emulating the behaviour of technologies like Ion Torrent, for which no dedicated sequencing simulators are currently available. BEAR is also the first metagenomic sequencing simulator program that automates the process of generating abundances, which can be an arduous task.

Conclusions

BEAR is useful for evaluating data processing tools in genomics. It has many advantages over existing comparable software, such as generating more realistic reads and being independent of sequencing technology, and has features particularly useful for metagenomics work.

Equivalent input produces different output in the UniFrac significance test

Background

UniFrac is a well-known tool for comparing microbial communities and assessing statistically significant differences between communities. In this paper we identify a discrepancy in the UniFrac methodology that causes semantically equivalent inputs to produce different outputs in tests of statistical significance.

Results

The phylogenetic trees that are input into UniFrac may or may not contain abundance counts. An isomorphic transform can be defined that will convert trees between these two formats without altering the semantic meaning of the trees. UniFrac produces different outputs for these equivalent forms of the same input tree. This is illustrated using metagenomics data from a lake sediment study.

Conclusions

Results from the UniFrac tool can vary greatly for the same input depending on the arbitrary choice of input format. Practitioners should be aware of this issue and use the tool with caution to ensure consistency and validity in their analyses. We provide a script to transform inputs between equivalent formats to help researchers achieve this consistency.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2105-15-278) contains supplementary material, which is available to authorized users.

Topical therapy is underused in patients with ulcerative colitis

The availability of new topical preparations for the treatment of left sided ulcerative colitis offers a therapy optimization for many patients. Rectal application of steroids and 5-aminosalicylic acid (5-ASA) is associated with fewer side effects and has a higher therapeutic efficacy in left-sided colitis as compared to a systemic therapy. Therefore, we were interested in the use of topical therapy in patients with ulcerative colitis. The key question was whether topical treatment is more frequently used than oral therapy in patients with proctitis and left sided colitis. Data of 800 patients of the Swiss IBD cohort study were analyzed. Sixteen percent of patients of the cohort had proctitis, 21% proctosigmoiditis and 41% pancolitis. Topical therapy with 5-ASA or corticosteroids was given in 26% of patients with proctitis, a combined systemic and topical treatment was given in 13%, whereas systemic treatment with 5-ASA without topical treatment was given in 29%. Proportion of topical drug use decreased with respect to disease extension from 39% for proctitis to 13.1% for pancolitis (P=0.001). Patients with severe colitis received a significantly higher dose of topical 5-ASA than patients in remission. Side effects of topical or systemic 5-ASA or budesonide treatment were less frequently seen compared to other medications. Topical treatment was frequently stopped over time. The quality of life was the same in patients with limited disease compared to patients with pancolitis. Topical treatment in proctitis patients was underused in Switzerland. Since topical treatment is safe and effective it should be used to a larger extend.

Transcriptome sequence and plasmid copy number analysis of the brewery isolate Pediococcus claussenii ATCC BAA-344 T during growth in beer

Growth of specific lactic acid bacteria in beer leads to spoiled product and economic loss for the brewing industry. Microbial growth is typically inhibited by the combined stresses found in beer (e.g., ethanol, hops, low pH, minimal nutrients); however, certain bacteria have adapted to grow in this harsh environment. Considering little is known about the mechanisms used by bacteria to grow in and spoil beer, transcriptome sequencing was performed on a variant of the beer-spoilage organism Pediococcusclaussenii ATCC BAA-344^T (Pc344-358). Illumina sequencing was used to compare the transcript levels in Pc344-358 growing mid-exponentially in beer to those in nutrient-rich MRS broth. Various operons demonstrated high gene expression in beer, several of which are involved in nutrient acquisition and overcoming the inhibitory effects of hop compounds. As well, genes functioning in cell membrane modification and biosynthesis demonstrated significantly higher transcript levels in Pc344-358 growing in beer. Three plasmids had the majority of their genes showing increased transcript levels in beer, whereas the two cryptic plasmids showed slightly decreased gene expression. Follow-up analysis of plasmid copy number in both growth environments revealed similar trends, where more copies of the three non-cryptic plasmids were found in Pc344-358 growing in beer. Transcriptome sequencing also enabled the addition of several genes to the P. claussenii ATCC BAA-344^T genome annotation, some of which are putatively transcribed as non-coding RNAs. The sequencing results not only provide the first transcriptome description of a beer-spoilage organism while growing in beer, but they also highlight several targets for future exploration, including genes that may have a role in the general stress response of lactic acid bacteria.

Reclassification of Paralactobacillus selangorensis Leisner et al. 2000 as Lactobacillus selangorensis comb. nov

The taxonomic status of Paralactobacillus selangorensis is described and, based on evidence presented, transfer of the species to the genus Lactobacillus with the name Lactobacillus selangorensis comb. nov. is proposed. This reclassification is supported by multilocus sequence analysis of the 16S rRNA gene and portions of the cpn60, pheS and rpoA genes. Mode of cell division and existing phenotypic information also show that P. selangorensis cannot be differentiated from the genus Lactobacillus. The type strain of Lactobacillus selangorensis comb. nov. is ATCC BAA-66T ( = LMG 17710T  = CIP 106482T).

Analysis and comparison of the pan-genomic properties of sixteen well-characterized bacterial genera

Background

The increasing availability of whole genome sequences allows the gene or protein content of different organisms to be compared, leading to burgeoning interest in the relatively new subfield of pan-genomics. However, while several studies have analyzed protein content relationships in specific groups of bacteria, there has yet to be a study that provides a general characterization of protein content relationships in a broad range of bacteria.

Results

A variation on reciprocal BLAST hits was used to infer relationships among proteins in several groups of bacteria, and data regarding protein conservation and uniqueness in different bacterial genera are reported in terms of "core proteomes", "unique proteomes", and "singlets". We also analyzed the relationship between protein content similarity and the percent identity of the 16S rRNA gene in pairs of bacterial isolates from the same genus, and found that the strength of this relationship varied substantially depending on the genus, perhaps reflecting different rates of genome evolution and/or horizontal gene transfer. Finally, core proteomes and unique proteomes were used to study the proteomic cohesiveness of several bacterial species, revealing that some bacterial species had little cohesiveness in their protein content, with some having fewer proteins unique to that species than randomly-chosen sets of isolates from the same genus.

Conclusions

The results described in this study aid our understanding of protein content relationships in different bacterial groups, allowing us to make further inferences regarding genome-environment relationships, genome evolution, and the soundness of existing taxonomic classifications.

Appropriateness of therapy for fistulizing Crohn's disease: findings from a national inflammatory bowel disease cohort

BACKGROUND

About 30-50% of patients with Crohn's disease (CD) develop fistulae, implying significant disease burden and complicated clinical management.

AIM

To assess appropriate use of therapy for fistulizing CD patients enrolled in the Swiss Inflammatory Bowel Disease Cohort using criteria developed by the European Panel on the Appropriateness of Crohn's disease Therapy.

METHODS

Specific questionnaires were used to gather information on disease and its management. We assessed appropriateness of therapy at enrolment for adult CD patients with one or several fistulae.

RESULTS

Two hundred and eighty-eight CD patients had fistulizing disease, of which 80% had complex fistulae and 32% currently had active draining fistulae. Mean age (s.d.) at diagnosis was 27 years (11), 51% males. Of the patients, 78% were judged as having globally an appropriate therapy, which was more often given for complex fistulae (87%) than for simple fistulae (67%). Antibiotics, azathioprine/MP, methotrexate and conservative surgery were almost always appropriate. Anti-tumor necrosis factor α was considered globally appropriate (91%), although most often with an uncertain indication. The 5ASA compounds, steroids and aggressive surgery were most often inappropriate (84%, 58% and 86% respectively).

CONCLUSIONS

Formal appropriateness criteria for CD therapy were applied to a national cohort of IBD patients. For more than three-quarters of the patients with fistulizing CD, therapy was globally appropriate.

The occurrence of intra-operative hypotension varies between hospitals: observational analysis of more than 147,000 anaesthesia

BACKGROUND

Hypotension, a common intra-operative incident, bears an important potential for morbidity. It is most often manageable and sometimes preventable, which renders its study important. Therefore, we aimed at examining hospital variations in the occurrence of intra-operative hypotension and its predictors. As secondary endpoints, we determined to what extent hypotension relates to the risk of post-operative incidents and death.

METHODS

We used the Anaesthesia Databank Switzerland, built on routinely and prospectively collected data on all anaesthesias in 21 hospitals. The three outcomes were assessed using multi-level logistic regression models.

RESULTS

Among 147,573 anaesthesias, hypotension ranged from 0.6% to 5.2% in participating hospitals, and from 0.3% up to 12% in different surgical specialties. Most (73.4%) were minor single events. Age, ASA status, combined general and regional anaesthesia techniques, duration of surgery and hospitalization were significantly associated with hypotension. Although significantly associated, the emergency status of the surgery had a weaker effect. Hospitals' odds ratios for hypotension varied between 0.12 and 2.50 (P < or = 0.001), even after adjusting for patient and anaesthesia factors, and for type of surgery. At least one post-operative incident occurred in 9.7% of the procedures, including 0.03% deaths. Intra-operative hypotension was associated with a higher risk of post-operative incidents and death.

CONCLUSION

Wide variations remain in the occurrence of hypotension among hospitals after adjustment for risk factors. Although differential reporting from hospitals may exist, variations in anaesthesia techniques and blood pressure maintenance may also have contributed. Intra-operative hypotension is associated with morbidities and sometimes death, and constant vigilance must thus be advocated.

Appropriateness of colonoscopy in Europe (EPAGE II). Presentation of methodology, general results, and analysis of complications

BACKGROUND AND STUDY AIMS

Appropriate use of colonoscopy is a key component of quality management in gastrointestinal endoscopy. In an update of a 1998 publication, the 2008 European Panel on the Appropriateness of Gastrointestinal Endoscopy (EPAGE II) defined appropriateness criteria for various colonoscopy indications. This introductory paper therefore deals with methodology, general appropriateness, and a review of colonoscopy complications.

METHODS

The RAND/UCLA Appropriateness Method was used to evaluate the appropriateness of various diagnostic colonoscopy indications, with 14 multidisciplinary experts using a scale from 1 (extremely inappropriate) to 9 (extremely appropriate). Evidence reported in a comprehensive updated literature review was used for these decisions. Consolidation of the ratings into three appropriateness categories (appropriate, uncertain, inappropriate) was based on the median and the heterogeneity of the votes. The experts then met to discuss areas of disagreement in the light of existing evidence, followed by a second rating round, with a subsequent third voting round on necessity criteria, using much more stringent criteria (i. e. colonoscopy is deemed mandatory).

RESULTS

Overall, 463 indications were rated, with 55 %, 16 % and 29 % of them being judged appropriate, uncertain and inappropriate, respectively. Perforation and hemorrhage rates, as reported in 39 studies, were in general < 0.1 % and < 0.3 %, respectively

CONCLUSIONS

The updated EPAGE II criteria constitute an aid to clinical decision-making but should in no way replace individual judgment. Detailed panel results are freely available on the internet (www.epage.ch) and will thus constitute a reference source of information for clinicians.

Appropriateness of colonoscopy in Europe (EPAGE II). Screening for colorectal cancer

BACKGROUND AND STUDY AIMS

To summarize the published literature on assessment of appropriateness of colonoscopy for screening for colorectal cancer (CRC) in asymptomatic individuals without personal history of CRC or polyps, and report appropriateness criteria developed by an expert panel, the 2008 European Panel on the Appropriateness of Gastrointestinal Endoscopy, EPAGE II.

METHODS

A systematic search of guidelines, systematic reviews, and primary studies regarding colonoscopy for screening for colorectal cancer was performed. The RAND/UCLA Appropriateness Method was applied to develop appropriateness criteria for colonoscopy in these circumstances.

RESULTS

Available evidence for CRC screening comes from small case-controlled studies, with heterogeneous results, and from indirect evidence from randomized controlled trials (RCTs) on fecal occult blood test (FOBT) screening and studies on flexible sigmoidoscopy screening. Most guidelines recommend screening colonoscopy every 10 years starting at age 50 in average-risk individuals. In individuals with a higher risk of CRC due to family history, there is a consensus that it is appropriate to offer screening colonoscopy at < 50 years. EPAGE II considered screening colonoscopy appropriate above 50 years in average-risk individuals. Panelists deemed screening colonoscopy appropriate for younger patients, with shorter surveillance intervals, where family or personal risk of colorectal cancer is higher. A positive FOBT or the discovery of adenomas at sigmoidoscopy are considered appropriate indications.

CONCLUSIONS

Despite the lack of evidence based on randomized controlled trials (RCTs), colonoscopy is recommended by most published guidelines and EPAGE II criteria available online (http://www.epage.ch), as a screening option for CRC in individuals at average risk of CRC, and undisputedly as the main screening tool for CRC in individuals at moderate and high risk of CRC.

Appropriateness of colonoscopy in Europe (EPAGE II). Chronic diarrhea and known inflammatory bowel disease

BACKGROUND AND STUDY AIMS

To summarize the published literature on assessment of appropriateness of colonoscopy for investigation of chronic diarrhea, management of patients with known inflammatory bowel disease (IBD), and for colorectal cancer (CRC) surveillance in such patients, and to report report appropriateness criteria developed by an expert panel, the 2008 European Panel on the Appropriateness of Gastrointestinal Endoscopy, EPAGE II.

METHODS

A systematic search of guidelines, systematic reviews, and primary studies regarding the evaluation of chronic diarrhea, the management of IBD, and colorectal cancer surveillance in IBD was performed. The RAND/UCLA Appropriateness Method was applied to develop appropriateness criteria for colonoscopy for these conditions.

RESULTS

According to the literature, colonoscopic evaluation may be justified for patients aged > 50 years with recent-onset chronic diarrhea or with alarm symptoms. Surveillance colonoscopy for CRC should be offered to all patients with extensive ulcerative colitis or colonic Crohn's disease of 8 years' duration, and to all patients with less extensive disease of 15 years' duration. Intervals for surveillance colonoscopy depend on duration of evolution, initial diagnosis, and histological findings. The EPAGE II criteria also confirmed the appropriateness of diagnostic colonoscopy for diarrhea of > 4 weeks' duration. They also suggest that, in addition to assessing extent of IBD by colonoscopy, further colonoscopic examination is appropriate in the face of persistent or worsening symptoms. Surveillance colonoscopy in IBD patients was generally appropriate after a lapse of 2 years. In the presence of dysplasia at previous colonoscopy, it was not only appropriate but necessary.

CONCLUSIONS

Despite or perhaps because of the limitations of the available published studies, the panel-based EPAGE II (http://www.epage.ch) criteria can help guide appropriate colonoscopy use in the absence of strong evidence from the literature.

Appropriateness of colonoscopy in Europe (EPAGE II). Iron-deficiency anemia and hematochezia

BACKGROUND AND STUDY AIMS

To summarize the published literature on assessment of appropriateness of colonoscopy for the investigation of iron-deficiency anemia (IDA) and hematochezia, and report appropriateness criteria developed by an expert panel, the 2008 European Panel on the Appropriateness of Gastrointestinal Endoscopy, EPAGE II.

METHODS

A systematic search of guidelines, systematic reviews and primary studies regarding the evaluation and management of IDA and hematochezia was performed. The RAND/UCLA Appropriateness Method was applied to develop appropriateness criteria for colonoscopy for these conditions.

RESULTS

IDA occurs in 2 %-5 % of adult men and postmenopausal women. Examination of both the upper and lower gastrointestinal tract is recommended in patients with iron deficiency. Colonoscopy for IDA yields one colorectal cancer (CRC) in every 9-13 colonoscopies. Hematochezia is a well-recognized alarm symptom and such patients are likely to be referred for colonoscopy. Colonoscopy is unanimously recommended in patients aged > or = 50. Diverticulosis, vascular ectasias, and ischemic colitis are common causes of acute lower gastrointestinal bleeding (LGIB); CRC is found in 0.2 %-11 % of the colonoscopies performed for LGIB. Most patients with scant hematochezia have an anorectal or a distal source of bleeding. The expert panel considered most clinical indications for colonoscopy as appropriate in the presence of IDA (58 %) or hematochezia (83 %).

CONCLUSION

Despite the limitations of the published studies, guidelines unanimously recommend colonoscopy for the investigation of IDA and hematochezia in patients aged > or = 50 years. These indications were also considered appropriate by EPAGE II, as were indications in patients at low risk for CRC with no obvious cause of bleeding found during adequate previous investigations.

Appropriateness of colonoscopy in Europe (EPAGE II). Functional bowel disorders: pain, constipation and bloating

BACKGROUND AND STUDY AIMS

To summarize the published literature on assessment of appropriateness of colonoscopy for the investigation of functional bowel symptoms, and report appropriateness criteria developed by an expert panel, the 2008 European Panel on the Appropriateness of Gastrointestinal Endoscopy, EPAGE II.

METHODS

A systematic search of guidelines, systematic reviews and primary studies regarding the evaluation and management of functional bowel symptoms was performed. The RAND/UCLA Appropriateness Method was applied to develop appropriateness criteria for colonoscopy for these conditions.

RESULTS

Much of the evidence for use of colonoscopy in evaluation of chronic abdominal pain, and/or constipation and/or abdominal bloating is modest. Major limitations include small numbers of patients and lack of adequate characterization of these patients. Large community-based follow-up studies are needed to enable better definition of the natural history of patients with functional bowel disorders. Guidelines stress that alarm features ("red flags"), such as rectal bleeding, anemia, weight loss, nocturnal symptoms, family history of colon cancer, age of onset > 50 years, and recent onset of symptoms should all lead to careful evaluation before a diagnosis of functional bowel disorder is made. EPAGE II assessed these symptoms by means of 12 clinical scenarios, rating colonoscopy as appropriate, uncertain and inappropriate in 42 % (5/12), 25 % (3/12), and 33 % (4/12) of these, respectively.

CONCLUSIONS

Evidence to support the use of colonoscopy in the evaluation of patients with functional bowel disorders and no alarm features is lacking. These patients have no increased risk of colon cancer and thus advice on screening for this is not different from that for the general population. EPAGE II criteria, available online (http://www.epage.ch), consider colonoscopy appropriate in patients of > 50 years with chronic or new-onset bowel disturbances, but not in patients with isolated chronic abdominal pain.

Appropriateness of colonoscopy in Europe (EPAGE II). Surveillance after polypectomy and after resection of colorectal cancer

BACKGROUND AND STUDY AIMS

To summarize the published literature on assessment of appropriateness of colonoscopy for surveillance after polypectomy and after curative-intent resection of colorectal cancer (CRC), and report appropriateness criteria developed by an expert panel, the 2008 European Panel on the Appropriateness of Gastrointestinal Endoscopy, EPAGE II.

METHODS

A systematic search of guidelines, systematic reviews and primary studies regarding the evaluation and management of surveillance colonoscopy after polypectomy and after resection of CRC was performed. The RAND/UCLA Appropriateness Method was applied to develop appropriateness criteria for colonoscopy for these conditions.

RESULTS

Most CRCs arise from adenomatous polyps. The characteristics of removed polyps, especially the distinction between low-risk adenomas (1 or 2, small [< 1 cm], tubular, no high-grade dysplasia) vs. high-risk adenomas (large [> or = 1 cm], multiple [> 3], high-grade dysplasia or villous features), have an impact on advanced adenoma recurrence. Most guidelines recommend a 3-year follow-up colonoscopy for high-risk adenomas and a 5-year colonoscopy for low-risk adenomas. Despite the lack of evidence to support or refute any survival benefit for follow-up colonoscopy after curative-intent CRC resection, surveillance colonoscopy is recommended by most guidelines. The timing of the first surveillance colonoscopy differs. The expert panel considered that 56 % of the clinical indications for colonoscopy for surveillance after polypectomy were appropriate. For surveillance after CRC resection, it considered colonoscopy appropriate 1 year after resection.

CONCLUSIONS

Colonoscopy is recommended as a first-choice procedure for surveillance after polypectomy by all published guidelines and by the EPAGE II criteria. Despite the limitations of the published studies, colonoscopy is also recommended by most of the guidelines and by EPAGE II criteria for surveillance after curative-intent CRC resection.

Appropriate treatment for Crohn's disease: methodology and summary results of a multidisciplinary international expert panel approach--EPACT

BACKGROUND/AIMS

For many therapeutic decisions in Crohn's disease (CD), high-grade evidence is lacking. To assist clinical decision-making, explicit panel-based appropriateness criteria were developed by an international, multidisciplinary expert panel.

METHODS

10 gastroenterologists, 3 surgeons and 2 general practitioners from 12 European countries assessed the appropriateness of therapy for CD using the RAND Appropriateness Method. Their assessment was based on the study of a recent literature review of the subject, combined with their own expert clinical judgment. Panelists rated clinical indications and treatment options using a 9-point scale (1 = extremely inappropriate; 9 = extremely appropriate). These scenarios were then discussed in detail at the panel meeting and re-rated. Median ratings and disagreement were used to aggregate ratings into three assessment categories: appropriate (A), uncertain (U) and inappropriate (I).

RESULTS

569 specific indications were rated, dealing with 9 clinical presentations: mild/moderate luminal CD (n = 104), severe CD (n = 126), steroid-dependent CD (n = 25), steroid-refractory CD (n = 37), fistulizing CD (n = 49), fibrostenotic CD (n = 35), maintenance of medical remission of CD (n = 84), maintenance of surgical remission (n = 78), drug safety in pregnancy (n = 24) and use of infliximab (n = 7). Overall, 146 indications (26%) were judged appropriate, 129 (23%) uncertain and 294 (52%) inappropriate. Frank disagreement was low (14% overall) with the greatest disagreement (54% of scenarios) being observed for treatment of steroid-refractory disease.

CONCLUSIONS

Detailed explicit appropriateness criteria for the appropriate use of therapy for CD were developed for the first time by a European expert panel. Disease location, severity and previous treatments were the main factors taken into account. User-friendly access to EPACT criteria is available via an Internet site, www.epact.ch, allowing prospective evaluation and improvement of appropriateness of current CD therapy.