In surgical fields with high rates of technological innovation and in which new devices are constantly available, it is often challenging to obtain high-quality evidence, not only because of the intrinsic difficulties of running a surgical randomised controlled trial (RCT), but also because the timeframe and sample size required to assess clinically significant outcomes (especially harms) may hinder the conduct of meaningful clinical trials. RCTs might provide a higher quality level of evidence than registries do but they have limitations, such as very high costs, the time required, and the difficulty involved in conducting the study [14]. From a technical point of view, observational studies represent the vast majority of registry-based studies. If appropriately validated, registry data can provide statistical measures of real-world health status with valuable societal benefits [15]. They can also monitor surgical activities by keeping track of  outcomes and of continuous updates in techniques used for surgical procedures  [16]. Like any clinical study, registry studies must follow strict methodological rules to be valid and informative, since only a well-structured data collection allows solid population-based epidemiological and statistical analyses [17]. Implant registries for orthopaedic surgery aim to collect information on patients, implants, and procedures and to assess surgical methods as well as types of implants and their materials, design, and other features. Coverage and completeness are the main parameters used to evaluate the quality of registries and their ability to provide reliable feedback [18]. High levels of coverage and completeness imply the ability to detect the performance of a wide range of devices and surgical techniques used in the whole population in real-life situations, thus partly overcoming the selection bias that could play a significant role in the absence of randomisation. While differences in patient selection, surgeon ability, and organisational contexts can hinder the possibility of a fair comparison of devices in terms of effectiveness (i.e. determining which is the best implant) when considering a limited context, the large scale of registries allows quick detection of harms (i.e. the rapid identification of underperforming implants) [19].

A successful registry can promptly detect device failures [20] during the follow-up. Failure is considered the gold standard in joint replacement registries when both patient and surgeon have agreed that reoperation is needed because of a dramatic decrease in the patient’s quality of life. Spine surgery is a relatively new and ultra-specialised branch, with considerable evidence obtained on it in recent years. In spine surgery, however, reoperation does not necessarily indicate implant failure, as implant removal is sometimes the standard clinical procedure (e.g. in the case of fractures). Moreover, the challenge faced by spine surgery registries is further complicated by the anatomical and functional complexity of the spine, the nature of the different conditions treated with spine surgery, the high number of available implants, and the need to both collect more variables and consider other outcomes when assessing the patient’s quality-of-life improvement; these factors require a more global and complex approach than arthroplasty registries accomplish [21].

The inclusion criteria for our review, based on those of Malchau [9], were strictly established to select only national-based registries. This specific criterion caused the exclusion of registries that, although well founded, were multicentre and therefore did not cover a national population. Specifically, Kaiser Permanente (an American integrated managed care consortium), the National Neurosurgery Quality and Outcomes Database (N(2)QOD) (developed by the American Association of Neurological Surgeons), NeuroPoint (the NeuroPoint Alliance program to improve the quality of care), the National Surgical Quality Improvement Program (NSQIP) Registry and the American Spine Registry (a collaborative effort between the American Association of Neurological Surgeons and the American Academy of Orthopaedic Surgeons) are volunteer-based and only enrol patients from several areas of the United States. As for Spine Tango, although it is internationally recognised as a well-structured registry, it was excluded because it was conceived as a multinational state registry. Despite this, Spine Tango’s design and organisation could be worthwhile looking into, and we consider it a valuable source of inspiration for national registries, as has happened in Switzerland. In general, the scientific production of registries is hugely varied and diversely significant. It also includes reports that are usually available on the registries’ websites but are rarely indexed in peer-reviewed databases, so they can only be searched for manually. Thanks to a thorough web search, it was possible to identify three more active registries that were not otherwise detected by the scoping review.

Forty-six out of 108 papers based their main research question on pre-operative and post-operative patient questionnaires to evaluate the efficacy of surgery or carry out epidemiological studies on the population. Some of these articles have laid solid foundations upon which, even today, the effectiveness of spine surgery is based [22,23,24]. Many analysed the complication rate and compared the clinical efficacies of different surgical procedures. Many others compared other surgical procedures in terms of clinical effectiveness and analysed the complication rate, paving the way for clinical management [25]. PROMs represent the gold standard for spine surgery since they do not focus on the device’s technical effectiveness and safety but instead evaluate clinical outcomes regarding disability, pain, and quality-of-life improvement [11, 26]. The systematic collection of patient safety data and PROMs makes it possible to monitor surgical procedure impact and value and to research factors that might influence the outcomes and complication rates of different techniques [27]. Strömqvist et al. [16] reported that the spine registry in Sweden positively affected healthcare. It significantly reduced the national mean length of stay for microdiscectomy by maintaining the same clinical outcomes assessed using PROMs and considerably lowering the cost of the national health system. The clinical outcome of spine surgery is jeopardised by a high complication rate [28] and a narrow range of clinical improvement [29, 30]. It is based mainly on patient-reported outcome measures (PROMs) [31]. However, some procedures, such as spinal fusion surgery, are increasingly being used despite growing caution that high-quality studies are needed to support its clinical effectiveness and efficiency [32]. All the selected registries assess outcomes using different PROMs, but all consider the main domains of pain, function, and health-related quality of life (HRQoL). These outcomes are the basis on which the scientific evidence for spine registries is built [33]. Our scoping review showed that, over the years, a progressive alignment with International Consortium for Health OutcomesMeasurement (ICHOM) standards [34] has been reached, even if all the selected registries have yet to match those criteria fully.

The nine selected registries differ in structural and organisational characteristics and follow different modalities of data collection and outcome evaluation (Tables 2 and 3). As for the source of funding, many registries are funded and managed by clinician associations and receive public funding in a more or less stable way. For every case, SIRIS Spine gets a specific fee from hospitals, part of which is used to run the registry. Only the Australian and Canadian registries declare that they receive funds directly from industries. In these times of economic difficulties for many public health systems, this is good news, as it shows how spine device manufacturers are also interested in registry data. Following the introduction of the new European Medical Device Regulation (EU MDR) [35], this interest is rising, as registries can provide valuable data to accomplish post-market analysis requirements or device surveillance and traceability, assuming that rules for managing conflicts of interest are in place.

Eight of the nine selected registries have a website, but only four publish it in English. Seven of the nine selected registries publish a report, but only four are available in English. The lack of availability of information in English might represent an additional obstacle to the scientific dissemination of data. This problem might be overcome by making the information available in English by including a dedicated section on the website and by publishing reports or summaries in this language.

Seven of the nine registries analyse data by taking the number of enrolled patients as a statistical unit. Three of them also consider the total number of collected and monitored procedures. While this information is usually available in the reports, it is also published on the website in only some cases (Swespine and SIRIS Spine). There is high heterogeneity in how the data are presented and the patients/procedures are analysed. A common standard for the presentation of results is needed for readers to be able to find the information they seek. A first step to overcome this difficulty could be to define a standard form to present the principal data characterising the registry in English. An excellent example of this approach could be what has been done recently by the International Society of Arthroplasty Registries (ISAR), which has defined a form to collect such information for joint registries, sent it to all its members, and recommended putting it at the beginning of their reports or on their websites (Kajsa Erikson, ISAR administrator, personal communication to Marina Torre, email dated 01/06/2023).

Each of the selected registries collects personal, comorbidity, and diagnosis data. Two registries (German and Finnish) consider all the spinal interventions, while the others set enrolment criteria for specific anatomical districts, focusing only on some spinal disorders. Although most are based on the Glassman classification [36] or subsequent modifications, the classification systems are only sometimes declared. The results clearly show the need for a further joint classification effort, starting from the commonly used Glassman classification, which can partially catch the wide variety of clinical conditions and therapeutic approaches encountered in spine surgery. Indeed, the use of well-structured registries with data of high validity and representativeness allows for results that can be generalised and have an acceptable level of evidence, thus improving the quality and cost-effectiveness of care [37, 38]. Moreover, data collection in surgical practice allows for the documentation of pathologies and surgical approaches and the evaluation of the population in a standardised manner, thus creating a common language for benchmarking [39].

The selected registries markedly differ in their duration of follow-up. Most stop collecting data 1 or 2 years after surgery, while CSORN for over 5 years and DaneSpine, Swespine, FinSpine for over 10 years. 

Coverage and completeness are critical issues. Indeed, this information was unavailable in the published papers and needed to be requested during the survey. However, even surveys sent to the coordinators were sometimes returned with data missing. The proportion of departments involved out of the total number of departments performing spine surgery at a national level (i.e. coverage) is the first challenge: this data can easily be reached if spinal surgery centres are involved in data collection on a mandatory basis, as voluntary participation usually leads to low rates of coverage and completeness. Making data collection mandatory by law or linking data entry to reimbursements might overcome these issues [40, 41]. Coverage data are available for only six registries; among them, the Swedish and Norwegian ones show the best recruitment performance, reaching peaks of 100%. Six of the nine registries declared the completeness rate, which ranged from 3 to 86%. However, it still needs to be clarified whether all the registries managed data requests with a similar approach and whether the completeness data refer to only the registered procedures or to the follow-up as well. A completeness of 75% and over at follow-up is a reasonable threshold for accurate analyses [38, 42]. Completeness remains crucial when interpreting data from the registries and should be clearly stated in all related publications [11].

Spine registry data are gaining interest from all stakeholders in health systems. Indeed, the monitoring of the quality of a product used in spine surgery should be based on high-quality data, as required by the new European legislation [35]. Furthermore, these data are expected to become decisive in clinical decisions. For this reason, data entered into the registries must have a high level of completeness and accuracy. External registry monitoring would be helpful to certify the quality of the data; hence, the importance of considering systems in the registry design that monitor data quality and guarantee the accuracy of data entry. Prospective data collection, clinical data entry, quality control and data validation, study designs that control for confounding bias, and the coverage of the population enrolled and exposed to the surgery are well-defined features of an efficient registry [15].

Based on the results of our study, eight of the nine active national registries declared that they have implemented a quality control process based, for most of them, on their internal systems. None declared that they refer to external audit agencies.

Several difficulties emerged during our study to identify active registries in a national setting. The scoping review selected studies investigating spine surgery in a spine registry setting, though, in some cases, the authors did not cite the name of the registry on which their studies were based. In these cases, the paper was excluded to avoid bias due to misclassification, although this might have led to underestimating the real scientific impact of those registries. The same limitation applies to the possible alignment of the registries with ICHOM  standards. An intrinsic limit of our search strategy, which was based on Van Hooff et al.’s search strategy [11], is that it could only detect some of the existing active national registries because the inclusion/exclusion criteria were stringent; moreover, the original online search was performed 2 years before the analysis of results was completed. Therefore, the research process involved performing additional web searches and expert interviews to collect detailed and updated information. Finally, the aim of this study was limited to the collection of the organisational standards of the existing national registries, thus excluding the analysis of data collected in terms of their quality and comprehensiveness. Further studies might focus on this topic.