Scandinavian health registries, like the Swedish Fracture Register (SFR), exist to profoundly improve public health understanding and medical research. Their purpose is to provide comprehensive insights into health trends, disease patterns, and the effectiveness of treatments, thereby enhancing healthcare policies and practices.
These registries leverage a national personal identity number (PIN) system, enabling extensive and diverse health data collection across the population. The SFR exemplifies this approach by tracking a wide range of data on fractures, including patient-specific, event-specific, fracture-specific variables, and patient-reported outcome measures (PROMs) over time. The process involves large-scale data and real-world evidence collection. The outcome of this systematic approach is a rich database with detailed information on over 890,000 fractures.
These registries provide valuable resources for healthcare research, offering insights essential for making informed public health decisions and improving medical care. Despite data quality and generalizability challenges, the SFR is a valuable research tool that is continuously evolving through ongoing adaptations.
Key takeaways
- Extensive coverage of health registries in Scandinavia plays a vital role in epidemiological research. They benefit from unique personal identity numbers (PINs) and are supported by publicly funded, fairness-focused social systems.
- Strengths and Challenges of National Health Registries: They offer valuable resources for healthcare research due to their large sample sizes, comprehensive data, and support for longitudinal studies. However, they can face data quality and logistical constraints.
- Case study: The Swedish Fracture Register (SFR) is a comprehensive registry with data on 890,000 fractures. It exemplifies the use of registries in tracking patient outcomes and supporting research.
- Patient-reported outcome measures (PROM) include the EQ-5D survey and the Short Musculoskeletal Functional Assessment. They offer insights into patient health outcomes following fractures. These surveys help in tracking health outcomes over time.
- Strengths of the SFR lie in its comprehensive data collection, ability to conduct longitudinal studies, large sample sizes, reduced selection bias, potential for studying rare conditions, and provision of real-world evidence.
- Limitations and Validity Concerns of the SFR. While the SFR is a valuable tool, it faces limitations in data quality and completeness, generalizability issues, and potential for coding errors. Studies assessing the SFR’s validity indicate variations in reliability depending on the fracture type.
Unlocking medical insights: the power and challenges of Nordic health registries in research
Health registries are essential to epidemiological research looking at prominent population health trends.
In particular, Scandinavian countries are known for their large-scale population registers. They stem from a long history of every resident having a unique personal identity number (PIN), publicly funded and fairness-focused social systems that depend on those systems. The different countries have similar social systems and health registries.
National health registries in Nordic countries provide potent tools for healthcare research due to their large sample sizes, comprehensive data, and ability to support longitudinal and real-world studies. These strengths facilitate a deeper understanding of health trends, disease patterns, and treatment effectiveness, contributing significantly to medical knowledge and public health policy. National health registries are valuable resources for medical research, but they come with challenges. These challenges relate to data quality, generalizability, privacy, ethical considerations, and logistical constraints. These limitations must be carefully managed to ensure the integrity and relevance of research findings derived from these data sources.1–7
Case study: The Swedish Fracture Register (SFR)
The Swedish Fracture Register (SFR) was developed between 2009 and 2010 and was launched in 2011 at Salgrenska University Hospital in Gothenburg, Sweden. As of 2012, more departments and clinics were invited. Today, all orthopedic trauma centers in Sweden have joined. We will make a case study of the SFR.8–11
In-depth look at the Swedish Fracture Register’s data and patient-reported outcomes
As of 2023, the SFR contains data on 890,000 fractures in Sweden. Variables are patient-specific, event-specific, fracture-specific, and patient-reported variables. Each entry is a separate fracture; if an individual has multiple fractures, these are separate entries even if they occurred from the same trauma. A total of 268 variables are tracked, and more are planned. Not all variables are possible at the same time. A comprehensive list of variables can be found here. More are planned, e.g., implant tracking.
Patient-reported outcome measures
Patient-reported outcome measures (PROM) are self-reported surveys that patients answer. After being registered in the SFR, they respond to a pre-injury survey and, one year later, to a 1-year survey. The first survey asks about the patient’s function just before the fracture, and the second asks after the current function one year later.
Each PROM survey consists of two parts. The first part is the EQ-5D survey, and the second is the Short Musculoskeletal Functional Assessment (SFMA). In 2019, the EQ-5D was changed from three levels for each answer (EQ-5D-3L) to the five-level version (EQ-5D-5L). Unlike fractures, PROM surveys reflect events, and PROM is not requested for every fracture. If the patient answers both surveys, tracking their health outcomes after a fracture is possible.
Advantages
- Comprehensive data collection: The SFR tracks many variables and collects data from all over Sweden.
- Longitudinal studies are readily facilitated since the SFR launched 2011 and track patient data over time (PROM before and one year after injury).
- Large sample sizes as the SFR covers all hospitals in Sweden managing trauma patients. Since the start, somewhat uniquely, it traced data on all fractures, even those that did not undergo surgical treatment. As of 2023, there are over 890,000 fractures.
- Linkage across different data sources: The PIN allows connection to other national registries, such as the population register (for death dates) and the Swedish Arthroplasty Register.
- Reduced selection bias, as all fractures are recorded.
- Potential for studying rare conditions due to the large sample size.
- Real-world evidence instead of controlled clinical trial environments. Patients and clinicians report and update the register daily.
- Cost-effectiveness for the researcher, as the data is already collected.
- Ethical research advantages: Using existing registry data for research can sometimes bypass ethical concerns associated with primary data collection, as long as the use adheres to privacy and ethical guidelines.
Challenges
- Data quality and completeness: The quality and completeness of data in the SFR varies. The response rate to PROM surveys varies greatly between locations. It also differs significantly for the initial PROM and 1-year PROM. During 2019, the SFR had a 6-month hiatus in the PROM collection. During this time, they switched from EQ5D-3L to EQ5D-5L. These are not comparable, though there are conversions.13,14
- Generalizability issues: As all fractures in patients with a PIN are registered, generalizability to Sweden and a similar population is probably acceptable. However, EQ-5D is country-specific.
- Limited control over data collection: The typical SFR user is not an orthopedic surgeon but a junior doctor or resident.
- Potential for coding errors: Several studies examine the validity of the SFR with varying results, as we will see.
- Privacy and ethical concerns: Large-scale personal health data raises privacy and ethical concerns. Ethical approval for access and study has previously varied between locations but is now centralized to the Swedish Ethical Review Authority. Get more insights about medical ethics and health data in our “Medical ethics in artificial intelligence applications” post.
- Lack of detailed clinical information: The SFR has no information on comorbidities, medical history, or other information.
- Dependency on healthcare system structure: The SFR collects data from hospitals with orthopedic trauma departments. Fractures detected in primary care might not be recorded.
- Challenges in data interpretation: Interpreting data from registries requires careful consideration of the context in which the data were collected, including healthcare practices, coding systems, and population characteristics.
- Time lag in data availability: Fractures are entered daily, and the register is updated constantly, but fracture registration can lag.
- Limited to existing variables: Research is constrained to the variables collected in the registry. It can only be studied if specific information was collected or coded.
- Cost and resource requirements for data management: It is fully web-based and is funded by public healthcare providers and the government.
- The risk of over-reliance on registry data is a general concern. There is a risk that the ease of access to large datasets might lead to an over-reliance on registry data, potentially neglecting other crucial research methods and data sources.
Validity and reliability of the SFR
As of November 2023, the SFR reports 64 published studies, and more are underway. Nine studies deal directly or indirectly with the validity of fracture classification in the SFR. Additional studies deal with the validity or completeness of other data (patient-reported outcome measures and register comprehensiveness for humeral fractures.)
Interobserver reliability is the difference in measurements between different observers. Intraobserver reliability is the difference between the same observer at other times. “Observer” can be “reviewer” or “rater” in various studies, but the meaning does not change.
Fracture classification
The SFR uses the AO/OTA classification system for most fracture types.11,16–18 The AO/OTA system builds on other fracture classification systems. For example, the Danis-Weber classification for malleolar fractures, Judet-Letournel for acetabular fractures, and Salter-Harris for pediatric distal radius fractures.
Humerus fractures. A study of 116 humerus fracture classifications (AO/OTA) found a moderate agreement between reviewers and the SFR classification. They noted that “errors” were usually adjacent groups. Applying a modification, the agreement substantially improved, indicating a high level of consistency in the classification of fractures among observers.
Tibia fractures. A study of 114 tibia fractures demonstrated that the SFR was a reliable resource for classifying tibia fractures, especially at the AO/OTA type level. The agreement with the expert reviewers was moderate for groups.
Femoral fractures. A study of 118 femoral fractures found that the SFR had almost perfect agreement with the expert reviewers for type and substantial agreement for the AO/OTA group. This suggests that the fracture classification in a national quality register can be accurate enough to evaluate fracture treatment in specific groups of fractures.
Clavicle fractures. Clavicle fractures in the SFR are classified according to the Robinsons’ classification. Studying 132 clavicle fractures in the SFR, the accuracy of the classification of clavicle fractures in the SFR was only fair.
The inter-observer agreement between the expert reviewers was good in the four studies mentioned.
Distal radius fractures. One study found moderate accuracy in classifying 128 distal radius fractures from the SFR. They also found varying inter-reviewer (between the expert reviewers) and inter-reviewer (the same reviewer at different times) agreements. This was in line with other studies that find that the AO/OTA classification of distal radius fractures is difficult to agree on.
So far, the studies have collected data from the institution that founded the SFR. They can have an increased focus on registering and how to register. The same clinicians who did the initial registration may be reviewing the examination. We might get better agreement due to testing intra-observer rather than intra-observer reliability. This does not limit the validity but is a study limitation. It can make them less representative of the national SFR.
The following studies assess national data collected from different clinics in the SFR.
Ankle fractures. One-hundred and fifty-two ankle fractures were randomly selected from the SFR. This study collected data from 17 clinics. SFR-reviewer agreement was substantial for both AO/OTA type and group level. As usual, agreement was higher for the fracture type than for the fracture group.
Courtesy of Nazzar Tellisi, I S Abdulkareem, P G Giannoudis, CC BY 4.0 via Wikimedia Commons
Pipkin fractures. A study used data from the SFR to analyze Pipkin fractures in adult patients over a specified period. They collected studies on all registered Pipkin fractures in the SFR, including CT scans. They reported that 26 out of 73 (36%) fractures registered as Pipkin fractures and had to be excluded because of incorrect classification. However, classification was not a studied outcome.
Acetabular fractures. Assessing the validity of acetabular fracture classifications, one study of 124 fractures from 24 hospitals found moderate agreement between experts and the SFR. They also found that agreement was highly variable between fracture groups and observers. They highlighted the importance of considering variation in accuracy across different fracture groups when using SFR data for analysis.
Atypical fractures. Imaging of every patient in the SFR registered as having an AFF was collected (i.e., 53 possible hospitals), and 178 patients were reviewed. 58% held up as AFF, though there are reasons to believe this was overly pessimistic.
Patient-reported outcome measures (PROM)
A study suggested that non-responders in the SFR had similar reported functions compared to initial responders. This was true both for the pre-injury and 1-year surveys. The age and gender of patients influenced the response rate. This study was published in 2017, and with the switch to EQ5D-5L, the continued validity of these results is unclear.
Conclusion
The SFR is a generally reliable and valid registry for fracture classification and patient outcomes. However, there are variations in reliability depending on the type of fracture. Some studies have limitations due to potential institutional biases and challenges in classifying certain fracture types. The ongoing research and adaptation to findings (like modifications in classification for better agreement) demonstrate a commitment to improving the registry’s accuracy and usefulness.
Sources
Numbered sources
1. Bergdahl C, Nilsson F, Wennergren D, Ekholm C, Möller M. Completeness in the Swedish Fracture Register and the Swedish National Patient Register: An Assessment of Humeral Fracture Registrations. CLEP. 2021 May 21;13:325–33.
2. Cnudde P, Rolfson O, Nemes S, Kärrholm J, Rehnberg C, Rogmark C, et al. Linking Swedish health data registers to establish a research database and a shared decision-making tool in hip replacement. BMC Musculoskeletal Disorders. 2016 Oct 4;17(1):414.
3. Flaten O. Nordic Health Registers as a Source for Value-Based Evidence. Pharm Med. 2014 Jun 1;28(3):119–22.
4. Laugesen K, Ludvigsson JF, Schmidt M, Gissler M, Valdimarsdottir UA, Lunde A, et al. Nordic Health Registry-Based Research: A Review of Health Care Systems and Key Registries. CLEP. 2021 Jul 19;13:533–54.
5. Reigstad MM, Larsen IK, Myklebust TÅ, Robsahm TE, Oldereid NB, Omland AK, et al. The Nordic Health Registries: an important part of modern medical research. Human Reproduction. 2016 Jan 1;31(1):216–7.
6. Rosén M. National Health Data Registers: a Nordic heritage to public health. Scand J Public Health. 2002;30(2):81–5.
7. Smith Jervelund S, De Montgomery CJ. Nordic registry data: value, validity and future. Scand J Public Health. 2020 Feb 1;48(1):1–4.
8. Wennergren D, Möller M. Implementation of the Swedish Fracture Register. Unfallchirurg. 2018 Dec 1;121(12):949–55.
9. Wennergren D, Ekholm C, Sandelin A, Möller M. The Swedish fracture register: 103,000 fractures registered. BMC Musculoskeletal Disorders. 2015 Nov 6;16(1):338.
10. Möller M, Wolf O, Bergdahl C, Mukka S, Rydberg EM, Hailer NP, et al. The Swedish Fracture Register – ten years of experience and 600,000 fractures collected in a National Quality Register. BMC Musculoskeletal Disorders. 2022 Feb 11;23(1):141.
11. Svenska Frakturregistret [Internet]. [cited 2023 Nov 24]. Available from: https://sfr.registercentrum.se/
12. van Hout B, Janssen MF, Feng YS, Kohlmann T, Busschbach J, Golicki D, et al. Interim scoring for the EQ-5D-5L: mapping the EQ-5D-5L to EQ-5D-3L value sets. Value Health. 2012;15(5):708–15.
13. Hout BA van, Shaw JW. Mapping EQ-5D-3L to EQ-5D-5L. Value in Health. 2021 Sep 1;24(9):1285–93.
14. Marsh J, Slongo T, Agel J, Broderick J, Creevey W, DeCoster T, et al. Fracture and Dislocation Classification Compendium – 2007. Journal of Orthopaedic Trauma [Internet]. 2007 Nov 1 [cited 2019 Apr 16];21(10). Available from: insights.ovid.com
15. Meinberg EG, Agel J, Roberts CS, Karam MD, Kellam JF. Fracture and Dislocation Classification Compendium-2018. J Orthop Trauma. 2018;32 Suppl 1:S1–170.
16. Slongo T, Audigé L. Fracture and Dislocation Classification Compendium for Children. Journal of Orthopaedic Trauma [Internet]. 2007 Nov 1 [cited 2019 Apr 16];21(10). Available from: insights.ovid.com