There are no formal requirements for the conduct of managing a clinical registry although there are operational documents referenced in peer-reviewed articles. An example is a 2010 editorial [23], that references (citation number [15] of the editorial) the Australian Commission on Safety and Quality in Health Care, Operating principles and Technical standards for Australian clinical quality registries. The registry imperative is evolving and perhaps this is indicated by the comprehensive policy document published in 2011 by the American Heart Association, in the journal Circulation [24]. This is in contrast to the long-standing existence of documents that clearly outline how to design, execute and report the results of a RCT [45, 46].

Registries aim to have complete, or almost complete, capture of all eligible procedures, thereby minimizing selection and enrolment bias [23]. Capturing a complete, or near complete, patient population with sequential enrolment is the goal. Registries operate as “business as usual”, with no cherry picking of good results and include both good and bad outcomes. Other challenges include providing timely feedback to collaborators, privacy issues, management of a large dataset, and access to individuals with appropriate statistical or epidemiological expertise, funding and intellectual rights in the case of multicentre involvement .

The data elements that registries collect need to be carefully considered and should be epidemiologically sound, meaning that the data should be simple, objective and reproducible. Examples of appropriate data to collect include patient demographics, surgical characteristics and anaesthetic type and dosage, other practice patterns and clinical effectiveness outcomes. The data elements need not be static, but rather change according to the important clinical questions that need to be addressed. The dataset for a registry should be simple and only data that are required to address the question or issue of interest should be collected. Logistically the data should be simple enough that physicians can efficiently enter information in a database in the context of a busy clinical practice.

For ARAC , data were recorded relating to the performance and effectiveness of PNB , adverse effects and complications. These data included a unique patient code, date of procedure, operation type, needle bevel type, local anaesthetic and dose, level of sedation and block success. PNB type was recorded: interscalene, periclavicular, axillary, distal humeral/forearm, femoral/fascia iliaca, sciatic, other peripheral lower limb nerves and trunk blocks. The technology used to locate plexus/nerves was recorded: ultrasound alone, nerve stimulator alone, ultrasound and nerve stimulator and other. The definitions used for this project were available online at www.​regional.​anaesthesia.​org.​au. The timing of follow-up for potential neurologic complications was either at 7–10 days or 6 weeks postoperatively, depending on practice location and time period. Patients were not considered to be uncontactable by phone until four attempts had been made at different times and using alternative phone numbers, including a mobile number if available. To detect potential neurologic complications patients were asked a standardized set of questions: Do you have any numbness? Do you have any tingling? Do you have any abnormal sensations? Do you have any pain? Do you have any weakness? These questions were asked in relation to the operative limb, and if the patient responded with “yes” to any of the questions, then further queries were made taking into account the anatomy relevant to the surgery and the PNB . Symptoms that were immediately adjacent to the wound, consistent with normal tissue healing or the initial trauma were not considered relevant in terms of anaesthesia being a causal factor. Symptoms that clearly were not related to the PNB were not considered significant. For patients with ambiguous symptoms or complaints, repeat contact was made with the patient. Triggers for referral to a neurologist were new onset of motor and/or sensory deficit; non-resolving paraesthesia; pain; allodynia; or dysaesthesia and any concern expressed by the surgical team regarding the potential for a PNB-related neurologic deficit. Assessment by the neurologist included history, examination, documentation and investigation. Investigations included electrodiagnostic tests [nerve conduction studies and/or an electromyogram], imaging [computed tomography, magnetic resonance imaging] and blood tests.

Data quality control was enhanced by the following methods:

When ARAC was completed, there was no other large-scale investigation into the safety of ultrasound-guided PNB [47]. The web-based interface utilized in ARAC facilitated ease of data entry, multicentre collaboration and collection of data from a large patient cohort. This project set up the foundation to develop a larger, more comprehensive clinical registry.

To further develop the registry a comprehensive new online interface (www.​anesthesiaregist​ry.​org) and secure remote database were created. The new online interface had features that enhanced its quality including improved security, ease of use and functionality, improved reporting, rules enforcement, audit trail of changes to data and improved data quality control (for example, context specific drop-down menus and criteria based entry). Because the new interface was more complex, specific training was implemented for data collectors and collaborators so as to familiarize them with the new interface at www.​anaesthesiaregis​try.​org. A broader range of outcomes was incorporated including clinical effectiveness outcomes, patient-rated outcomes, wrong-site block and respiratory complications. Development of the registry included expanding the patient cohort available for analysis and refinement of outcomes measuring clinical effectiveness. Additional patient and block related fields and patient-rated outcomes were included in the new online interface, introduced into practice on June 1st 2011.