Hospital records contain patient information routinely collected through hospital facilities as part of the patient care process. Information available through these records includes patient demographic data, general health data, details surrounding patient presentation to hospital, treatment provided, length of stay and observed health outcomes. Hospital records specifically documenting drowning cases can be extracted and analysed. Here, the sociodemographic characteristics of patients and information on the context of drowning events can be used to identify factors and activities associated with an increased risk of drowning.
If sufficiently detailed data is available, further analysis can track patterns of drowning hospitalisations within different population subgroups, and investigate patient outcomes, such as length of hospital stay or secondary complications resulting from the initial drowning event. Hospital records may be hand written notes, paper forms, part of an electronic database or a combination of all three.
The methods and coding structures used to document health data within hospital records may differ between hospitals, health districts and countries. An example of a standardised international coding system for health data is the World Health Organisation’s International Statistical Classification of Diseases and Related Health Problems (ICD). Through this system, it is possible to assign predefined codes to describe the cause and context of a drowning event. Examples include ‘W67’, indicating accidental drowning and submersion while in swimming-pool, or ‘V92’, indicating drowning and submersion due to fall off watercraft. If a non-standardised coding system is used, it may be necessary to develop collaborative relationships with hospital stakeholders to interpret their record keeping systems.
Hospitals act as data custodians, or gatekeepers, of hospital record data. Therefore, permission to review hospital records will likely need to be sought from the hospital itself as well as an external ethics committee. If the data is de-identified or non-identifiable and no linking of data will occur, the request process may be simpler and faster.
Police records contain information on crimes and incidents that have been reported to the police. Police may be called when an accident or death occurs, or may approach a crime scene to collect evidence and assist individuals involved. Primarily, police reports will capture drowning cases that are anticipated to be homicidal or suicidal in nature, with many accidental drowning cases not necessitating police investigation. Content from police records may borrow from other registries, such as a cause of death registry, to provide more detailed information in reports. In Australia for example, a complete coronial record available on the online National Coronial Information System to approved users, will feature a coronial finding, an autopsy report, a toxicology report and a police report.
Police records are usually highly sensitive. Access to, and extraction of data from them, may be difficult and will most likely require high-level authorisations and external ethical approval. Similar to hospital records, primary data collection will likely be recorded in a number of formats including hand written notes, paper forms and electronic databases. The depth and type of information collected by police can vary greatly between areas.
- Population-level data is available.
- Data is routinely collected through both hospital and police records.
- A cost-effective alternative to performing primary data collection.
- In many settings, information is available from long periods of time, allowing for retrospective review to assess patterns or trends.
- Many variables are included in both hospital records and police records, making them a rich source of information.
- Hospital data can be used to examine fatal and non-fatal drowning.
- There is potential for large volumes of un-related data to be incidentally included in review and analysis.
- It can be time consuming to review and analyse, particularly if records are paper-based.
- Extraction of data will require authorisation by data custodians and external ethical approval.
- It is difficult to analyse a combination of handwritten notes, paper forms and electronic records.
- Hospitals and police stations may use different data collection and coding systems, making data difficult to compare between sites.
- A drowning event will not be recorded if an individual does not present to hospital. This can happen if the drowning injury is not considered to be severe enough to require treatment, if a drowning death occurs in a community setting at the time of the drowning event, or if an individual is unable to access to a hospital for financial reasons. This leads to an underreporting of drowning cases in hospitalisation records.
- A drowning event will not be recorded if it is not reported to the police, leading to an underreporting of cases in police records.
- Information captured as ‘open text’ through data sources is likely to be non-standardized and difficult to analyse.
- Ethics processes may take time and be costly.
- If outcome is not reported, it may be difficult to count both fatal and non-fatal drowning. It can be difficult to quantify re-admissions and deaths from conditions related to an initial non-fatal drowning incident (for example, infections).
Hospital data can be used to analyse the incidence, trends and outcomes of drowning, yet only captures severe drowning cases which require treatment. Police records will likely contain information on when and how a drowning incident has occurred, particularly if the circumstances of death are unclear. Although both hospital and police databases are likely to be large, it is not guaranteed that the data collected will be of high quality or will accurately represent the local population. Quality and coverage of hospital and police data from low income settings may be particularly low due to lower investment into data collection systems.
If the volume of information seems overwhelming, consider selecting a smaller number of hospital sites or police record systems to review. Ensure to first search the existing literature prior to conducting an analysis as previous, large-scale analyses of similar records may have already been completed.
Example of a patient hospital record (US)
Examples of police record databases (US)
Report from the National Health Survey (Australia) (PDF 2,179KB)
Report from the National Crime Records Bureau (India) (PDF 31,634KB)
Assess the situation