Data Resource Profile: The Multiple Sclerosis Documentation System 3D and AOK PLUS Linked Database (MSDS-AOK PLUS)

Real-world evidence in multiple sclerosis (MS) is limited by the availability of data elements in individual real-world datasets. We introduce a novel, growing database which links administrative claims and medical records from an MS patient management system, allowing for the complete capture of patient profiles. Using the AOK PLUS sickness fund and the Multiple Sclerosis Documentation System MSDS3D from the Center of Clinical Neuroscience (ZKN) in Germany, a linked MS-specific database was developed (MSDS-AOK PLUS). Patients treated at ZKN and insured by AOK PLUS were recruited and asked for informed consent. For linkage, insurance IDs were mapped to registry IDs. After the deletion of insurance IDs, an anonymized dataset was provided to a university-affiliate, IPAM e.V., for further research applications. The dataset combines a complete record of patient diagnoses, treatment, healthcare resource use, and costs (AOK PLUS), with detailed clinical parameters including functional performance and patient-reported outcomes (MSDS3D). The dataset currently captures 500 patients; however, is actively expanding. To demonstrate its potential, we present a use case describing characteristics, treatment, resource use, and costs of a patient subsample. By linking administrative claims to clinical information in medical charts, the novel MSDS-AOK PLUS database can increase the quality and scope of real-world studies in MS.


Introduction
Real-world data (RWD) are used to address an increasing number of questions related to multiple sclerosis' (MS) course of disease, prognosis, and treatment [1]. Real-world evidence (RWE) of this type can also be important to payers and healthcare decision makers to inform pricing, reimbursement, and labels for new therapies [2]. For such studies, RWD on MS patients can be collected from a large variety of sources, including product and disease registries, electronic medical records (EMRs), and administrative claims databases [1][2][3][4]. However, each data source individually presents specific limitations that restrict the number of questions that can be addressed and hinder the accuracy of the results, thereby introducing bias [1]. For example, while claims data offer a comprehensive picture of patients' healthcare resource use and costs, as well as comorbidities, they lack information on disease severity and duration, laboratory results, and important prognostic measures such as the Expanded Disability Status Scale (EDSS) [5,6]. On the other hand, data sources such as registries and EMRs, which are richer in terms of documenting patients' clinical characteristics and disease severity, lack information on the economic burden of the disease and/or the broader comorbidity profile of the patient.

Data Resource Basics
MSDS-AOK PLUS is a growing linked database collecting information from administrative claims data provided by AOK PLUS and EMRs from the MSDS 3D patient management system.

MSDS 3D
MSDS was developed to improve the documentation and management of patients with MS, and has been made available in different versions over the years (MSDS clinic, MSDS practice, MSDS 3D ) [16][17][18][19]. The latest development, MSDS 3D was created at the beginning of 2010 by the MSDS project group in Dresden, Saxony [14,15]. The software allows the input of patients' personal data, a structured history and clinical examination, clinical scores, and treatment details. The highly specialized system has been used for the Multiple Sclerosis Partners Advancing Technology and Health Solutions (MS PATHS), an initiative that integrates routine clinical care with both standardized patient reported outcomes (PROs) and standardized magnetic resonance imaging (MRI) sequences to better understand MS and subsequently approaches to personalized medicine [20]. With various therapy-specific modules, MSDS 3D serves as a key platform for post-marketing effectiveness, comparative, or post-authorization safety studies [15,[21][22][23][24][25][26]. Specifically, MSDS 3D consists of data on patient MS clinical history and routine care, including treatment of MS symptoms and disease-modifying therapy (DMT) use. Moreover, the system allows for the documentation of functional performance, including cognitive testing, gait analysis, MS performance testing (MSPT), EDSS, and imaging procedures (MRI and OCT) over time [27]. Notably, MSDS 3D includes patient reported outcome (PROs) questionnaires, among which are the MS-Health Resource Survey (MS-HRS), Quality of Life in Neurological Disorders (NeuroQoL), Early Mobility Impairment Questionnaire (EMIQ), and the Multiple Sclerosis Walking Scale (MSWS-12) [28][29][30][31]. Currently, the connection to a patient portal is being established, which, in addition to communication and data exchange between doctor and patient, enables patients to feed in documents and information themselves.

AOK PLUS
Approximately 85% of the German population is insured by statutory health insurance (SHI) funds [32]. Specifically, the AOK PLUS sickness fund consists of data on 3.4 million patients insured by AOK PLUS in the regions of Saxony and Thuringia. The dataset provides nearly a full coverage of healthcare-related services, including a record of all inpatient and outpatient diagnoses and procedures, inpatient hospital admissions, drug use, as well as outpatient general practitioner (GP) and specialist visits. In addition to healthcare resource utilization, the sickness fund allows for the determination of patient costs in various settings, including primary care, hospitalizations, rehabilitation, prescriptions, and indirect costs from inability to work. In healthcare research, the dataset is used for a multitude of studies, including but not limited to epidemiological assessments, description of treatment patterns, investigating disease burden, comparative effectiveness, and evaluation of healthcare resource use and costs.

Linkage: MSDS-AOK PLUS
To generate the linked database, patients within the MSDS 3D system insured by AOK PLUS were recruited during regular visits to ZKN and asked to provide informed consent (IC) for their inclusion. ZKN then transferred a list of pseudonymized registry IDs and AOK PLUS insurance numbers pertaining to the patient to AOK PLUS for linkage ( Figure 1). Using the insurance number, the registry ID was mapped to a pseudonymized claims data ID. After subsequent deletion of patient insurance numbers, a linked complementary dataset was generated and is currently accessible for subsequent analysis through IPAM e.V., a university-affiliated research institute. As of February 2023, the database consisted of 500 patients; however, recruitment and subsequent linkage is ongoing and expected to increase at a rate of 200-250 patients per year. A total of approximately 900 AOK PLUS insured patients are currently included in the MSDS 3D system. As a future directive, patients insured by AOK PLUS and treated at other neurological practices which employ MSDS 3D can also be included in the linked dataset, expanding on the existing cohort.

Ethical Considerations
Prior to linking data between MSDS 3D and AOK PLUS, the physicians ascertain that the patient has comprehended the information and ZKN obtains a written IC form from each patient. All records identifying the subject are kept confidential and will not be made publicly available. Approval from the Ethics Committee at the Technical University of Dresden and the Ministry of Saxony (SGB § 75) was obtained to generate and maintain the database. No further ethical approval is required on a study-specific basis.

Ethical Considerations
Prior to linking data between MSDS 3D and AOK PLUS, the physicians ascertain that the patient has comprehended the information and ZKN obtains a written IC form from each patient. All records identifying the subject are kept confidential and will not be made publicly available. Approval from the Ethics Committee at the Technical University of Dresden and the Ministry of Saxony (SGB § 75) was obtained to generate and maintain the database. No further ethical approval is required on a study-specific basis.

Setting
The MSDS-AOK PLUS database provides regional coverage of MS patients visiting ZKN in Dresden and insured by AOK PLUS in Saxony and Thuringia (Table 1). Administrative claims data are provided retrospectively starting from 1st January 2014, with continuous updates with approximately a six-month data gap, whereas MSDS 3D documents patient data prospectively from initial MS diagnosis until the patient's last visit at the center. To use the MSDS-AOK PLUS database for research purposes, ZKN and AOK PLUS provide approval after the review of a study protocol outlining objectives, methods, and planned analyses. Upon approval, IPAM e.V. will extract the data based on a data extraction manual, defining inclusion and exclusion criteria and all variables and format, and conduct the analyses. Indirect access to the database by external parties is not possible.

Data Contents
A detailed overview of the data parameters available in the MSDS-AOK PLUS database is presented in Table 2. Notably, some data parameters may be sourced from both AOK PLUS and MSDS 3D . For research purposes, recommended use of data from each source will be based on the strengths of each dataset. Clinical information, such as MS subtype or date of first MS diagnosis is most reliable as documented in MSDS 3D , which more accurately captures clinical parameters as a result of documentation by the treating MS center. In contrast, parameters such as healthcare resource utilization, prescription medications, or the broader patient comorbidity profile of the patient are captured more completely in AOK PLUS, due to their direct relevance for reimbursement purposes. In the case of prescriptions, while MSDS 3D captures medications prescribed by the physician at the center, whether the prescription was filled in the pharmacy is captured by AOK PLUS, serving as a better indication, although not full confirmation, of compliance to medications. Administrative information such as date of birth or sex is expected to be consistent between datasets due to the high-quality documentation systems in place. In the event of significant inconsistencies in such variables, there is a possibility to quality control data parameters with the patients at the center. In the event of remaining discrepancies or unresolved quality checks, it is recommended that the data parameter is transformed to missing or the patient is excluded from the analysis sample, conditional on the significance of the data point with respect to patient selection or outcomes.    Patient-reported MSWS-12 scores based on 12 item questionnaire to determine how much MS has limited walking ability in the past two weeks X Documentation yearly a Visits will be based on counted dates of invoiced EBM codes; b rehabilitation refers to inpatient rehabilitation stays paid by the sickness fund. Note that the majority of rehabilitation services are covered by the pension insurance; c documentation of all relevant aids and remedies in a structured format is planned as a future development, to be documented on each patient visit (every 3-6 months or once a year); d in the outpatient setting, services are invoiced

Purpose & Applications
RWD can provide sample populations for MS research to complement clinical trial evidence, which is often collected among narrowly defined and controlled patient populations [1,3]. In contrast to randomized controlled trials (RCTs), RWE can help inform risk stratification, outcome definitions, and guide treatment choices or sequences among heterogeneous groups of patients in clinical practice. Furthermore, RWE allows patient follow up in the long term, which is advantageous in evaluating the risks and benefits of therapies [1]. In addition to its contribution to understanding disease behavior and clinical effectiveness, RWE increasingly serves an important role in health economic decision making [2]. While there are various applications of RWD in MS, the evaluation of many research questions holistically has been a challenge due to variability in data parameters among different RWD sources. The MSDS-AOK PLUS linked database is a step towards closing this data gap, by providing a complete picture of the patient's clinical state, capturing disease severity indicators documented regularly at MS related visits at a specialized neurology center, as well as all-cause resource use and diagnoses based on the patients' continued encounter with the healthcare system ( Table 2).

The MSDS-AOK PLUS Mission
The MSDS-AOK PLUS database was generated for the purpose of expanding realworld evidence capabilities in MS, providing data to support various research objectives to ultimately guide clinical decision making. The database can be an asset to determine efficacy and safety in routine clinical care, understand treatment patterns and disease behavior as a step towards personalized medicine, and inform on the generalizability of RCT results. Detailed applications of the MSDS-AOK PLUS database are described in Table 3. Table 3. Applications of the MSDS-AOK PLUS database.

MSDS-AOK PLUS Applications
Comparative effectiveness

Use Case: Baseline Characteristics and Resource Consumption of a Selected MSDS-AOK PLUS Patient Cohort
To highlight the data capture of the MSDS-AOK PLUS database, we present a descriptive use case of a selected cohort of patients. From the initial cohort of enrolled patients, we selected those with an existing diagnosis for MS, who were continuously insured for at least three years, starting from 30 June 2018 to 30 June 2021 (study period). Using 1 January 2019 as an arbitrary index date, we described patient baseline characteristics in the six-month period prior to index and select healthcare resource use, costs, and DMT use in the 2.5-year period from index until the end of the study period.
Overall, 186 patients (76.34% female) were selected (Table 4). At baseline, patients had a mean age of 49.14 years and a median time from the first MS diagnosis to the index of 12.07 years. The majority of patients (77.42%) had relapse-remitting MS (RRMS) at a timepoint closest to index, whereas 9.68% had secondary progressive MS (SPMS), with 7.53% of patients included in the sample missing information on MS subtype in the inclusion period. Closest to index, MS patients had a mean EDSS of 3.35, ranging from 1 to 8. One third of patients suffered from hypertension (33.33%) and depression or anxiety (32.80%) each at baseline. From January 2019 to June 2021, a mean of 12.99 outpatient visits per patient year (PY) were observed, approximately half of which were MS-related (mean 6.69 visits per PY). In the follow-up period, 71 (38.71%) and 61 (32.80%) of patients had at least one all-cause hospitalization or MS-related hospitalization, with a mean 0.40 and 0.28 hospital admissions per PY, respectively. Related to their disease, mean MS-associated costs were highest for DMT use at €12,718.29 per PY, followed by €1,699.37 per PY for all-cause inpatient admissions, and €431.00 per PY for outpatient visits, independent of specialty ( Table 5). The most commonly used DMTs in the follow-up period were ocrelizumab and fingolimod (18.28% each) (Figure 2). Moreover, a similar proportion of patients received glatiramer acetate and dimethyl fumarate, respectively (11.83% vs. 11.29%).

Significance of MSDS-AOK PLUS
A strong evidence base is essential for understanding MS behavior and treatment in the real world. There are various clinical and patient MS registries available in Europe, such as the Big MS Data (BMSD) network combining the Danish, Italian, and Swedish MS registries with MSBase and the French Observatory of Multiple Sclerosis (OFSEP), as well as alternative national registries including the German and UK MS Registries [34][35][36]. A detailed overview of MS registries for observational research is maintained by the Multiple Sclerosis Data Alliance (MSDA) Catalogue [37]. Furthermore, MS patients can be analyzed for a subset of research questions from a number of national and regional administrative claims databases covering a wide population of patient profiles. However, none of these data sources alone can provide a full clinical picture of patients together with infor-

Significance of MSDS-AOK PLUS
A strong evidence base is essential for understanding MS behavior and treatment in the real world. There are various clinical and patient MS registries available in Europe, such as the Big MS Data (BMSD) network combining the Danish, Italian, and Swedish MS registries with MSBase and the French Observatory of Multiple Sclerosis (OFSEP), as well as alternative national registries including the German and UK MS Registries [34][35][36]. A detailed overview of MS registries for observational research is maintained by the Multiple Sclerosis Data Alliance (MSDA) Catalogue [37]. Furthermore, MS patients can be analyzed for a subset of research questions from a number of national and regional administrative claims databases covering a wide population of patient profiles. However, none of these data sources alone can provide a full clinical picture of patients together with information on the overall comorbidity status, concomitant non-MS related therapies and general health care resource utilization and costs. While a few real-world studies in MS have made use of linkage between data sources, often to supplement clinical information and/or HCRU and costs, standalone linked databases of claims and EMRs of patients with MS remain scarce. This is especially true in Europe, where to the best of our knowledge there is no database linking claims to EMRs for MS patients, comparable to the MSDS-AOK PLUS database. In the UK, the Clinical Practice Research Datalink (CPRD) offers linkage opportunities for the general population, but in MS only references linkage of primary care data from GPs with hospital data, therefore lacking MS specific data parameters [38]. Moreover, as CPRD is a primary care-driven EMR database, a broad capture of data as seen in the AOK PLUS claims a dataset covering all healthcare settings (GP, specialties, inpatient) is not available without linkage to further sources. In the US, prior attempts to link claims data for MS patients to EMRs were hindered by high levels of missingness on the clinical indicators, such as the EDSS score, an issue that is not confronted in our data thanks to the high-quality assessment of the EDSS in the MSDS 3D system [11]. As such, with limited data sources available to address most research questions in MS, the MSDS-AOK PLUS linked database is a step towards improving the data gap and expanding the capabilities of real-world evidence in the field.
By bringing together the strengths of an MS patient management system, documenting disease specific characteristics and relevant PROs and in-depth comorbidity records, healthcare resource use, and costs documented via a regional claims database, the MSDS-AOK PLUS database will serve a crucial role in health economic outcome studies. The importance of both clinical and economic data in such studies can be highlighted by recent observational studies which have utilized the MS-HRS to quantify the economic impact of MS [28,39,40]. Research using the MS-HRS has shown a similar economic burden among men and women affected by MS, and an increased economic burden with relapses and disease severity [41]. Overall, these studies point to the importance of considering patient disease characteristics to appropriately assess costs in economic evaluations. While the MS-HRS adds a societal perspective considering resources which may not be relevant for reimbursement purposes, it is limited by reliable recall periods and patient discretion. As such, claims data supplemented by AOK PLUS in the linked MSDS-AOK PLUS database can be used to provide an accurate overview of resource consumption among total and subpopulations of MS patients across longer follow-up periods.

Limitations
Despite its advantages, the MSDS-AOK PLUS database is accompanied by several limitations. MS patients included in the database are selected based on their health insurance, AOK PLUS, based particularly in the regions of Saxony and Thuringia, potentially presenting selection and regional bias. In addition, all patients included in the database attend ZKN, which is a specialized center of excellence and therefore may raise questions with regards to the representativeness of the MSDS-AOK PLUS cohort compared to the wider MS population in Germany. To minimize any bias, all eligible patients were recruited without pre-screening at the time of their appearance to ZKN. Moreover, due to uniform healthcare regulations and standard practices across Germany, patients included are expected to be generally representative of the broader MS population. Furthermore, the combined dataset will consist of data derived from routine medical practice and, as such, exhibit a degree of missing data, coding error, or unstructured data with regards to outpatient recording. Despite this, claims data are considered to be valid and of high quality, [42][43][44] and various algorithms for data structuring and extraction in MSDS 3D are in place. In MSDS 3D , data fields are predefined for the different areas of clinical routine and stored with reference values, e.g., for laboratory data or vital signs. Thus, the entered data is not only structured but also directly checked for plausibility, and error messages are displayed in case of incorrect entries. Edit checks ensure that missing data are avoided, and data entries are further checked for validity. This form of data entry and recording has been established in the field of clinical trials for several years but has additionally been expanded for documentation in the context of clinical routine.

Data Access and Future Perspectives
Future studies with the MSDS-AOK PLUS linked dataset must be approved by the data providers (AOK PLUS, ZKN) and carried out through the university-affiliated research institute IPAM e.V. To date, the database has been used to investigate the validation of a claims data-based proxy for EDSS, assessment of hospitalization costs by EDSS score in an alternative use case analysis, the investigation of MS subtype coding in claims data, and validating efficacy of propensity score methodologies for generating balanced comparator populations based on confounders observed in claims data (publications to follow) [45][46][47]. Additional validation purposes planned include the comparison of congruency between claims data and patient medical records or validation of outcome tools such as the MS-HRS questionnaire.
As of February 2023, the dataset captures 500 linked MS patients. Patient recruitment is ongoing in an effort to expand the current sample and set the stage for future RWE studies. Long term future goals include the expansion of the database into further regions covered by multiple insurance providers or additional specialized MS centers.

Conclusions
The MSDS-AOK PLUS linked database presents a unique opportunity in MS, expanding the current landscape and potentials of RWE. By combining the strengths of administrative claims databases and detailed clinical information provided in MS patient charts, MSDS-AOK PLUS offers a more complete picture of the patients' history and clinical state. The database is actively growing, with the overall purpose of expanding the capabilities of MS research in the real-world.