Reproduction of the Porto criteria in the registry
The registry fully reflects the guideline’s recommendations for the initial diagnosis of PIBD. The disease activity indices PCDAI for CD and PUCAI for UC were not included as variables in the CEDATA-GPGE registry during the study period. However, the PCDAI could be calculated in more than 75% of CD cases, while the PUCAI could be calculated in less than half of the UC cases. Through the variable ‘Physician overall assessment’, disease activity could be subjectively assessed by the treating physician as an alternative basis for decision making for further treatment. In 2016, the algorithms for calculating disease activity indices and thus the variables PCDAI and PUCAI could be implemented in the registry.
Practical application of the guideline recommendations documented in the registry
The observed symptoms in our study sample corresponds to the typical triad of symptoms of PIBD reported in the literature (15, 24). The proportion of documented diagnostic examinations varied within the 12 diagnostic categories and also between the three diagnoses. Some categories have a high level of documentation across the three diagnoses, e.g. ‘Case history’ and ‘Physical examination’. Some of the frequencies of the instrumental diagnosis categories, such as the ‘Imaging of the small bowel’, differ between documented diseases. ‘Imaging of the small bowel’ and ‘Ileocolonoscopy with multiple biopsies’ are categories in which examinations are documented poorly for all three diseases. Furthermore, the very low documentation of ‘screening for immunodeficiencies’ in children and adolescents under 6 years of age is conspicuously. The supplementary survey indicated that while the majority of screening is performed, documentation in the registry is often missing. This suggests an underreporting, e.g., due to lack of capacity. Another possible reason is the effort involved in documentation. The documentation needs to be minimal and performed by dedicated personnel. This is further underlined by three phenomena: First, categories that are less relevant in the disease context tend to be less complete, e.g., growth retardation and pubertal delay are far more common in CD than in UC. This is also applicable for upper GI endoscopy and small bowel imaging, although the later may be delayed or at another institution depending on access to MRI or video capsule endoscopy. The second category involves high detail data input, e.g. laboratory values. This needs to be simplified by offering various entry methods (SI units or common other units) or automated transfer of data from hospital information systems (HIS), like evaluated in current and future projects. The last category is data with search effort in the registry, e.g. less common lab investigations like trough levels, vaccination status, or screening for immunodeficiencies. Our survey revealed, that far more of these investigations are performed than reported. This underreporting can only be addressed by improving data entry mechanisms and active query management for missing data and plausibility checks, as also addressed in current projects in CEDATA-GPGE. The registry has been intensely improved from 2016 on, to address these issues, including live plausibility checks and improvement in data entry mechanisms. The increase in data documentation from 2016 on may be due to the Online-Tool 2.0. Automated or semiautomated data transfer from HIS is still not possible, but will be implemented in future projects in concurrence with rollout of the German Telematic infrastructure initiative.
The increase in instrumental diagnostics in all three diagnoses between 2014 and 2018 is in line with the observed increase in ileocolonoscopy and esophagogastroduodenoscopy by Buderus et al. for the period 2004 to 2014 (24). Since 2011, the German Society for Gastroenterology, Digestive and Metabolic Diseases (DGVS) recommend esophagogastroduodenoscopy not only in CD but also in UC (33). The Porto criteria recommend esophagogastroduodenoscopy in all suspected cases of IBD, too (17). During our observation period, a lower rate of ileocolonoscopy and esophagogastroduodenoscopy was performed compared to Buderus et al. (24). The low rates of some instrumental diagnostic measures, such as endoscopy, must be viewed with caution, considering input errors or incorrect reporting procedures or a diagnosis including endoscopy that took place outside the reporting center. The documentation characteristics in the registry have changed in 2016, as described above, which may lead to better documentation. This will be evaluated in future studies.
Which examination technique was used may depend on advances in medical technology and availability at a clinic or practice (34). For example, X-ray was a common examination for imaging of the small bowel for a long time. However, because radiography has undesirable side effects (radiation exposure) for young patients, gentler examinations such as MRI or video capsule endoscopy should be preferred (17). The data of Buderus et al (24) as well as our study show that this recommendation finds acceptance among documenting centers.
The recommendation ‘Diagnosis by gastroenterologists with pediatric expertise’ is not in itself a diagnostic measure in the strict sense, but it is the most frequently documented variable from the recommended measures for diagnosis. This is due to the fact that mainly pediatric gastroenterologists document in the registry. Patients with IBD-U were diagnosed by a pediatric gastroenterologist more frequently than patients with CD or UC. It is conceivable that patients present to a specialist more often when findings are unclear, that is typical for IBD-U. IBD-U patients are younger and thus more prone to be treated by pediatric gastroenterological specialists.
The accurate documentation in the registry according to current guidelines is an important challenge for the participating centers but also for the administration of the CEDATA-GPGE registry. Missing documentation may influence the determination of important parameters, such as disease activity indices. An accurate documentation leads to improved process and outcome measures in children and adolescents with IBD, as Crandall et al. showed (23).
It should be considered that in some cases deviations from guidelines are justifiable (35). Reasons may be disease type, age group, previous diseases, comorbidities, but also acceptance and reliability of parents. In practice, local availability is also likely to play a role. For example, by far not all outpatient clinics have access to video capsule endoscopy or easy access to MRI. The documentation of structural data of the reporting centers should be implemented in CEDATA-GPGE in the future in order to expand the possibilities of the registry for scientific questions of health services research. This kind of data allows comparisons between hospitals (benchmarking) (36) but also between the levels of care, staff and their qualification. Furthermore, they can provide insights into the use of certain equipment in the clinics (16, 34).
There may be reasons why some measures were not applied or documented in the registry, but these reasons were not documented in the registry during the observation period. Automated plausibility checks offer the opportunity to facilitate manual data input and to minimize missing data (26). Since 2016, a new feature in the registry is that practitioners can use comments to inform about their decision on a specific issue. To increase data quality, regular data monitoring will be conducted so that individualized feedback can be provided to the recruiting centers (37). Financing of dedicated personnel capacity for external quality assurance, as is mandatory by German law (§ 135a SGB V German Social Law), should be strictly implemented to ensure adequate participation and representative data extraction.
Limitations
Generalizations or transfer of the results to the entire patient population with PIBD in German-speaking countries are not possible, because the documenting centers are selective compared to the overall treatment of PIBD. Only patients who were diagnosed in specialized centers and gave written consent were included in the registry. Other reasons why patients are not included in the registry may be time restrictions during data collection or in the centers; presentation in the emergency department or other clinics; change of clinic with loss to patient’s first reports; changes in the nursing, medical and scientific staff of an outpatient clinic. Four of the five centers of the additional survey stated, that they report between 50% and 70% of patients in the registry. Lack of staff and time were frequently mentioned as reasons for not reporting all patients. Therefore, the number of patients documented in the registry is smaller than the number who actually were diagnosed at the participating centers. Consequently, there is a risk of overinterpretation of the results (selection bias).
What implications can the results provide for health care research based on patient registries?
The aim of health care research is to generate knowledge about disease-specific diagnostic and therapeutic care, to evaluate treatment systems scientifically, and to derive recommendations for improvement that are relevant to patients or populations (38). On the basis of patient-registries, the diagnostic of defined patient populations can be observed in a real healthcare setting (39). Depending on the number of variables implemented, disease-related patient registries offer a very good overview of guideline-based care provided by participating centers. However, the data do not represent the entire reality of health care of PIBD. Disease-specific patient registries such as CEDATA-GPGE provide important insights into the practice of highly specialized care. Nonetheless, the majority of care for patients with PIBD in Germany and Austria is managed outside of the centers that participate in the registry. Combination with other data sources, such as comparison with reimbursement data from health insurance companies, can provide additional insights into care in the different sectors and levels of care in the healthcare system.
The German Network for Health Services Research (DNVF) named criteria for data quality in patient registries in 2010 (updated in 2019). The network emphasized that due to the predominantly missing legislative basis, sufficient registry quality (completeness and validity of the data) can only be achieved by a high acceptance of the registry by patients and documenting institutions (39, 40). Incomplete and non-valid data are a major problem of patient registries. They often do not occur randomly and may contribute to result bias or false conclusions. Data quality highly depends on the administration of the registry in addition to the compliance of the documenting institutions (40).
Registries need a high administrative, temporal and financial effort (39). The financing of the CEDATA-GPGE registry was realized by donations during the study period only. The longer a registry exists, the more experience is generated over time and implemented in the registry. Consequently, time itself is a quality-enhancing factor for registries in the sense of the quality improvement cycle (PDCA) according to Shewhart (36).
Patient registries can help health care communities to deal with three aspects of change over time. For once, there is mostly continuous but sometimes disruptive change in recommendations and therapeutic strategies with development of new treatments or better understanding of disease behavior. Besides capturing adherence to diagnostic and treatment recommendations, it may be valuable to obtain detailed case data to compare treatment strategies and adjust for captured confounders, e.g., with advanced models like propensity score matching instead of unreflective historical control groups that would lead to overestimation of treatment effects in current therapies (41).
Secondly, there is development of disease classification systems over time, which leads to differentiated therapeutic stratification. In registries, one can support this by applying new classification systems to previously documented cases with known follow-up and outcome. In PIBD, this is currently reflected by the definition of atypical UC but also the role of isolated Crohn’s colitis (42).
Finally, technological innovation leads to new investigation and data acquisition methods. Data acquisition may lead to improvement of data quality and especially completeness, but it does not do so by itself or automatically, as shown above. However, it can increase usability. It can incorporate other data sources, as currently observed with patient-reported outcomes through the CEDMO-app, that links patient diaries with CEDATA-GPGE registry data and help adolescents with IBD in everyday life (43). It can also lead to the need for biobanking with patient registries to obtain biological samples that may be reexamined with new technologies later.
Disease-related patient registries can support the recruitment of patients for future studies in rare diseases, which in turn can provide important insights into guideline-based care. Clinical trials in comparison are much more restricted by inclusion and exclusion criteria, creating a homogenous dataset, that does not reflect real world application in most cases. In addition, registries have a longer observation interval than clinical trials, allowing the acceptance and application of current guidelines to be observed over a very long time. The additional implementation of patient related outcomes, e.g. quality of life, provides the opportunity to evaluate the effect of adherence to medical guidelines as implemented in CEDATA-GPGE recently (40).