<b>Use of record linkage in measuring HIV occurrence, time-to-event, and surveillance: a scoping review protocol</b>

According to the European Centre for Disease Prevention and Control (ECDC), in 2021, the overall rate of new HIV diagnoses for the European Union and European Economic Area (EU/EEA) was 3.9 per 100,000 population, showing a 36.1% decrease from 2016. Despite progress in the reduction of new infections, the HIV epidemic is still a public health concern, especially concentrated in key populations at higher risk, including gay, bisexual, and other men who have sex with men (GBMSM), transgender persons, people who inject drugs (PWID), sex workers, or migrants (Simões et al, 2021; UNAIDS, 2010).Testing is a cornerstone in the HIV response, especially in the context of an ongoing epidemic. Community-based testing plays an important role by offering anonymous testing targeted at key populations; therefore, collaborating to stop HIV transmission by improving early detection and immediate treatment (WHO, 2000; Sulat et al, 2018). This approach has also been shown to improve inclusivity in HIV prevention and control efforts (Sahay et al, 2011).Although all data collected, stored, and used as part of scaling-up HIV services must comply with the confidentiality and security requirements outlined by the Joint United Nations Programme on HIV/AIDS (UNAIDS), the anonymization of participants poses a challenge to assess HIV incidence and conduct longitudinal studies accurately. Additionally, linking an individual’s data across multiple databases becomes challenging when using anonymous tests or anonymous data, especially when institutional record procedures and software differ across settings, including the unique identifier composition. Nevertheless, health or research institutions could apply unique identifiers to databases entries to allow accounting persons/cases beyond HIV service/screening activity and track follow-up outcomes such as HIV occurrence. These unique identifiers can allow record linkage techniques and potentially track HIV outcomes across all settings of the HIV care continuum. The identifier code of anonymized study participants can be carried out manually or using specialized software designed for data extraction and deidentification (Traeger et al, 2024).WHO states that the record linkage techniques enable data integration and the tracking of individuals longitudinally while maintaining their anonymity. There are two main approaches: deterministic and probabilistic algorithms. Deterministic algorithms assess whether record pairs match or do not match based on a specified set of identifiers. In data-rich contexts where direct identifiers are available and of high quality, these algorithms are easier to implement, modify, and execute with minimal computational resources (Dusetzina et al, 2014). The deterministic approach fails to consider the differing discriminatory power of various identifiers or values. In contrast, probabilistic methods are designed to assess each identifier's discriminatory power and determine the likelihood that two records represent a true match based on their agreement or disagreement across multiple identifiers. In scenarios with limited information, where direct identifiers are absent and/or data quality is poor, and unique identifiers across settings only partially match, probabilistic methods consistently yield better results than deterministic methods (Dusetzina et al, 2014).Choosing the right data linkage algorithm depends on some factors such as resources, research objectives, and the quality and uniqueness of available variables (Dusetzina et al, 2014).