Introduction

In Australia, aged care and disability service providers are required to maintain comprehensive administrative and clinical records for clients [1, 2]. Aged care services include both residential (facility-based) and home-based care for non-Indigenous Australians aged over 65 years and Indigenous Australians aged over 50 years [3]. Disability services include residential and home-based care for all Australians under 65 years with long-term intellectual, mental, cognitive, physical and sensory impairments [4]. In 2022, an estimated 178,000 Australians received aged care services, while an estimated 519,000 received disability services through the National Disabilities Insurance Scheme [5], constituting an essential part of Australia’s healthcare sector.

To ensure quality and safety of aged care and disability services, enhanced information reporting requirements have been introduced in recent years as part of legal, regulatory and funding obligations. Care providers are required to collect data on, but not limited to, financial and prudential matters, clinical quality indicators, care delivery, and serious incidents [6]. Reporting requirements vary according to service type within both aged care and disability based on location (residential vs. home-based care) and duration of care (long-term vs. transitional care). Service providers are therefore challenged to collect extensive clinical information to meet multiple reporting requirements.

To facilitate reporting, providers rely on routine health information collected at the point-of-care, primarily through electronic health record (EHR) systems or clinical information systems (CIS) [7]. Clinical data is defined as any information documented within the client’s EHR that supports the workforce to complete clinical and care activities for the clients, including demographic information, clinical history, clinical assessments, care notes, and medication records. These clinical data also serve as crucial inputs which enable advanced real-time analytics to support clinical decision-making and business decisions [8]. Poor data quality is a major barrier to the effective use of CIS [9] and can lead to healthcare errors [10, 11], flawed business decisions, and increased operating costs [12]. Reliable clinical data is therefore imperative for health service performance and has made data quality an essential part of information systems research and practice in recent years [13].

Optimising data quality within health services is expected to contribute to improved patient outcomes [4], support reporting requirements, and reduce operating costs. However, there is limited guidance available to direct aged care and disability service providers in what constitutes optimal documentation [14]. This is in contrast to hospital and primary care settings where standardised national guidelines for health information documentation exist [15, 16]. Further, studies aimed at understanding and improving data quality in CIS within healthcare have also primarily been conducted in hospital settings or within specialist care settings such as dementia care [13, 17,18,19,20,21,22,23,24,25,26,27].

Few studies addressing improving clinical data quality in aged care and disability services currently exist. Some studies have examined the impact of EHR implementation on workflow and resident health outcomes in Australian aged care facilities [28,29,30,31,32,33] and highlighted the potential to transform routinely collected aged care data for research and sector innovation [8]. These studies show that while there is widespread agreement of the benefits of CIS for improving workforce efficiency [22, 34] and resident health outcomes [35], there is a need to improve data quality to enhance care delivery [36]. However, the specific causes of data quality concerns within aged care and disability services and solutions for optimisation have yet to be explored.

While the definition of data quality can differ from the perspectives of data producers, consumers and custodians [37], ‘fitness for use’ [38] is widely used in the literature to describe data quality. This term focusses on consumer or ‘user’ requirements of the data being captured which can encompass multiple dimensions [37]. Data-user involvement in quality improvement research is a well-established method to understand underlying problems and develop effective and sustainable quality improvement solutions from those who are directly involved in the process of interest [39, 40].

As part of the OPTIMISE study [41], this study aimed to explore aged care and disability workforce (referred to as ‘data-users’) experiences and perceived root causes of clinical data quality concerns at a large aged care and disability services provider in Western Australia (WA). The specific objectives of the study were to: (i) understand day-to-day clinical documentation workflow across different data-user groups, and (ii) identify root causes of clinical data quality issues, to inform optimisation solutions across care services.

Methods

Study design and setting

This study forms part of the larger OPTIMISE study [41], which is a prospective pre-post optimisation study using an integrated Agile Lean Six Sigma DMAIC (Define, Measure, Analyse, Improve, Control) framework [42] aimed at identifying opportunities for optimising the quality of clinical data across aged care and disability services operated by Brightwater Care Group (‘Brightwater’) in WA. Brightwater provides residential and home-based care for approximately n = 2,000 aged care and n = 500 disability services clients across eight different programs (Table 1) [41]. Two standalone vendor-based systems or CIS are used across care services: Clinical Manager [43] (for residential care facilities) and CareLink+ [44] (for home-based care). Details of the full study design are presented in our protocol paper [41].

Table 1 Overview of aged care and disability services
Full size table

In the Measure stage, a baseline audit of clinical data quality was undertaken using EHRs for n = 2,404 clients active in aged care and disability services between 1 July to 31 December 2021 [45]. Baseline data quality assessments used a combination of data warehouse audits and manual clinician review, to assess the quality of EHRs across six data-user-identified metrics: Completeness, Currentness, Accuracy, Clarity, Compliance, and Usability. In total, 18 key data quality issues were identified across services [45]. Baseline assessments were followed by the Analyse stage, which used qualitative research methodology to understand data-user experiences of Clinical Manager and CareLink+, and to explore potential root causes of clinical data quality issues in these systems.

Participants and recruitment

Purposive sampling was used to select services for consultation. A total of 10 residential care facilities were selected for face-to-face or online interviews (5 aged care, 5 disability), while home-based care staff were consulted through an online survey. For programs with multiple facilities such as Residential Aged Care (RAC) and Supported Independent Living (SIL), we purposively selected the 3 facilities with the highest, median, and lowest data quality scores from the baseline audit [45]. Based on prior research, carrying out usability testing with 8–10 participants should identify 80% of usability problems [46].

Participant inclusion criteria were staff members who used: (1) Clinical Manager and/or CareLink + for clinical documentation (primary data-users), and/or (2) clinical data for reporting, analytics or clinical/service decision making (secondary data-users). Staff from five functional groups were included: Service managers, clinical staff, care staff, administration staff, and corporate staff.

Data collection

Interviews

A semi-structured interview guide was developed, piloted and revised with clinical and research team members prior to use. Interview questions were designed to understand the underlying or ‘root causes’ of key data quality issues identified in the Measure stage. The final interview guide consisted of eight structured and semi-structured questions [47] (Appendix 1).

In-person visits to residential care facilities were conducted by two researchers (GT, RW) between September and October 2022. The first interviewer (GT) is an Information Systems Analyst, and the second (RW), a clinician. During each visit, convenience sampling was used to recruit eligible staff members who were available at the time for interviews. After verbal informed consent was obtained, one researcher interviewed participants, while the other recorded participant responses into a spreadsheet. Interviews were not audio recorded. Recruitment continued until all eligible staff were approached or saturation of responses was reached [48]. Online interviews were conducted over Microsoft Teams for key participants (e.g., site managers, clinical leaders) who were unavailable during the day of the visit.

Surveys

Additionally, an online survey, consisting of a written version of the interview questions, was created using Survey Monkey [49] and emailed to all staff providing home-based care. Surveys were used as home-based staff work directly within client homes without a central office location, thus presenting logistical challenges for conducting face-to-face interviews.

Data analysis

Inductive thematic analysis was used to analyse interview and survey responses. This involved identifying themes from the data without depending on an established theoretical framework [50]. Inductive analysis was selected to reduce potential bias caused by one author completing the initial coding. The researchers used a combination of consensus and split coding, due to the small nature of the research team and the desire to maintain transparency and transferability in the analysis [51]. One researcher (GT) read through all participant responses to establish familiarity with the data, assign initial codes then generate subthemes and themes. Two researchers (RW, LT) independently analysed a random sample of responses (30% of all responses) to assign codes and generate themes. A hierarchical coding frame was used, creating top-level and second-level codes, which became the themes and sub-themes generated by the analysis [52]. All three researchers met following initial analysis to discuss where coding and themes did not match or where extra codes and themes were required. A consensus method was used for discrepancies. Where researchers could not agree on a theme, a fourth researcher (AM), with expertise in qualitative research and analysis, was consulted to resolve discrepancies. Once initial themes and subthemes were determined, two researchers (GT, RW) checked the findings to ensure there was adequate evidence to support the themes and subthemes. A final consensus meeting was held to ensure all researchers agreed with the developed themes and subthemes. During this meeting, quotes that strongly captured the essence of the theme were selected from a list of relevant quotes, based on consensus.

Results

Participant characteristics

A total of n = 135 staff members participated in the interviews (n = 84, 62%) and surveys (n = 51, 38%), representing approximately 6.3% of the total organisational workforce. There was a balanced composition of participants from aged care (n = 41, 30%), disability services (n = 27, 20%), community-based care (n = 51, 38%), and corporate staff (n = 16, 12%). Most participants had been employed at the organisation for less than four years (Table 2).

Table 2 User Consultation Participants by Functional Groups (n = 135)
Full size table

Eight overarching themes and 29 subthemes were identified as potential causes of data quality issues in CISs (Table 3).

Table 3 Summary of Themes and Subthemes Representing Factors Influencing Data Quality in iCare and CareLink+
Full size table

Theme 1: staff-related challenges

Staff behaviour and practices

Staff non-compliance and a poor attitude to documentation tasks were viewed as a main challenge, “Some staff just don’t care [about documenting properly]” (1). Second-hand documentation was also a common practice where staff would document client notes on behalf of others, which resulted in the omission of details, “I request others to document on my behalf. So, I write only brief observation on a paper” (16). Verbal handover without updating information in clinical systems was also common. Although verbal handover is a standard health practice, it should be complimented by written documentation. One participant stated, “I wish everything was [documented] in Clinical Manager for us, as things get forgotten at handover” (66). Unintentional human error, such as forgetting to complete documentation and scanning, or entering data in incorrect forms and charts, was also acknowledged as a cause of data quality issues, “We have a lot of documents to upload…the timeliness of info going into Clinical Manager….transfer of information from admission documents – human error can happen sometimes” (72).

Cultural diversity

A majority of participants reported that English language proficiency challenges among the aged care and disability services workforce contributed to poor quality of information documentation. “There are so many free text fields where [the] quality of notes largely depends on the level of English of staff completing those fields” (84).

Diversity in technical skills

There was consensus that the care workforce varied in terms of technical skills and that existing CIS do not “appropriately engage the data-user at their level of knowledge” (32). As explained by a direct care worker, “A lot of us were not used to computers. We learned from bits and pieces and picking other people’s brain.” (64).

Diversity in clinical knowledge

A majority of participants also reported that clinical knowledge varied across the workforce, including knowledge of clinical terminology and abbreviations, which impacted the quality of clinical documentation. A non-clinical participant stated, “[A lot of] information is entered in the system which are summary text with acronyms and short-form [and] are difficult to understand. Information should be entered in a language that can be understood by other data-users” (81).

Theme 2: education and training

Formal training in system use

A lack of structured and standardised training was reported to affect the quality of clinical data. Many participants reported learning to use the CIS informally, often through co-workers, “Usually the person who is new…is shown how to use Clinical Manager through someone on site” (26), self-discovery, “It wasn’t until I was thrown into this role that I just clicked on everything” (63), or previous training, “[I was] trained in Clinical Manager elsewhere” (8).

Role-specific training

Participants also reported a lack of role-specific education and training on how systems should be used by different data-user groups (e.g., nursing, allied health, care workers) and which groups are responsible for inputting specific information. “People understanding their roles and responsibilities [impacts how up to date data and documentation are]” (25).

Theme 3: external barriers

Governance of external providers

External providers and agency staff are commonly employed to fill workforce gaps in aged care. Participants reported that external providers are not required to adhere to internal documentation policies which can impact data quality. “External providers have their own book for notes… so we don’t know what happens” (41).

Errors in health/medical documentation

Errors in external clinical documentation such as incorrect hospital discharge summaries were also common and are imported into CIS on admission. “[There are often] incorrect information from hospital and [we] need to redo assessment” (5).

Families as information source

Participants acknowledged the quality of initial data and documentation is often dependent on the quality of information provided by family members at admission. “Sometimes family members don’t understand what we are trying to get out of them as admission information, which goes into better supporting clients, [so this information is missing]” (54).

Sector-wide staff shortages

Participants frequently cited staff shortages as a cause of data quality issues. “[Sites are] short staffed and therefore time is an issue” (1), and “Staff are in a hurry, [there is a] shortage of staff [and] clients need attention” (40).

Theme 4: operational guidelines and procedures

Data and documentation requirements

Requirements for information documentation were unclear across the organisation and this impacted data quality. “This is not just a systems problem;… [the organisation] needs to know [the data] requirements clearly to make [the] system provide what is needed instead of [collecting] all unnecessary details.” (29).

Staff roles and responsibilities for data collection

Participants also reported a lack of clarity of roles and responsibilities for data collection across the organisation. “Expectations for demographic data and accountability for roles [are] not clearly outlined or monitored” (32).

Staff roles and responsibilities for monitoring data collection

Lack of clarity around staff roles and responsibilities for monitoring data collection and quality was also reported. “No one is monitoring [the] update of information [to ensure data is of high quality]” (63). A second participant added, “who would monitor it anyway, who would start it, and who does training on it?” (34).

Theme 5: organizational practice and culture

‘One size fits all’ approach

Participants reported that Clinical Manager is used across all residential sites and CareLink + for at-home services. However, the two systems are not able to meet the unique data and reporting requirements across all services. “[Clinical Manager] may be working well for RAC but not for TCP” (1) as operational requirements differ in these two service areas. Similarly, a participant from disability said, “Clinical Manager is RAC focused, the care plans populate automatically in a way that is not appropriate for disability” (71) as disability clients have functional impairments that are not captured through care plans focussed for aged care.

Workload and burden of documentation

Participants reported that the amount of documentation to be completed affected quality. “If you have to write too many things, people copy and paste and it’s not always right” (40). Relatedly, at larger sites, the volume of documentation required was often overwhelming, “[There are] hundreds of progress notes… [It’s] too much” (59). Direct care workload was also prioritised over documentation tasks, “[Documentation] get completed at the end of the shift as we are busy, and often, [details] are forgotten at the end of the shift” (50).

Communication of system or procedure changes to workforce

Participants also reported that changes to systems and documentation processes were not always effectively communicated to sites. “[The organisation] doesn’t communicate changes when they make changes to processes.” (56).

Duplication of effort

Duplication of effort in documentation was commonly reported and impacted data quality. Examples included doubling up of electronic and paper-based documentation, “Some staff still use paper-based forms which takes time to be entered [electronically]” (82), storing documents in multiple locations (37), and needing to enter the same information twice in different areas of the system, “Forms don’t speak to each other and thus create duplication” (28).

Theme 6: Technological infrastructure challenges

System speed, connectivity, and reliability

System speed, connectivity and reliability at sites were important factors impacting the ability to document client notes in real-time. “The system is slow. It plays a big part in me putting information into the system and there is a lag during busy times” (41).

Technological support

Insufficient technical support was also identified as an important barrier to EHR use, “[We] don’t know who to contact to fix bugs” (41).

Equipment and resources

Participants expressed insufficient hardware at sites to support uninterrupted documentation at the point-of-care. “Staff queue at the end of shift to document notes” (31). Another participant stated, “Technology has been the biggest issue, with constant glitches. One day iPads aren’t working [but] desktops are working, but there are not enough computers” (81).

Theme 7: systems design limitations

Data extraction and reporting

Participants reported major difficulties generating reports from the front-end systems, “The function to pull out data [from Clinical Manager] is difficult, and it is difficult to work with generated reports” (49). Similar difficulties were expressed by back-end data-users. “The modelling of the data and tables at the back-end makes it difficult to pull data out for reporting. I think when these systems were created, they looked at it from function, and didn’t think about pulling data out for reporting” (79).

Search and filter

Participants also expressed frustrations with the lack of inbuilt search functions to efficiently filter information. “There is no search function to look for documents by keyword” (30).

Single view of information

Participants reported frustrations with the lack of such single view of information, adding that they had to navigate through multiple screens to access information. “Everything should be in one place rather than having to navigate through many places” (49).

Change history

Participants expressed that timestamps and detailed change history are lacking which made it difficult to know whether client information was recent, “There is no visible history of data being up-to-date” (31). This made it difficult to keep track of who made changes to client information. “I don’t think it’s possible to tell which field was updated by which person on this date” (81).

Theme 8: systems configuration

Dropdown values

Participants also reported frustrations with the configuration of dropdown menus, indicating that most values do not capture adequate clinical detail. “[A client is] blind in the right eye for example – but the dropdown only has ‘sight/vision [issue]’. Dropdowns can limit and prevent person-centred care [by not allowing a sufficient level of clinical detail] (4)”. Frustrations with choosing from long dropdown lists were also reported, “It is difficult and time-consuming to understand what to choose from dropdown list (31)”.

Free-text fields

Participants reported an overrepresentation of free-text fields which contributed to variation in data quality. “There are so many free-text fields where the quality of notes largely depends on the level of English of staff” (84) and “Information is put in free-text fields, so it’s hard to be able to pull information out” (26).

Automated information mapping

Participants expressed dissatisfaction with limited automation in CIS. “Assessments don’t create alerts automatically and that’s why the reviews [are] forgotten when [staff] are busy” (2). Where automation features were enabled, staff reported several errors in the configuration. “[Adding a] second wound entry overwrites [the] active wound reminder” (60).

Unused capabilities

Participants also believed that not all system capabilities were being used. “There are more data in Clinical Manager that we could use but we don’t have training to be able to use all of its features” (51).

Integration of multiple systems and information transfer

Finally, many participants reported frustrations with information transfer between systems. “If we document client contact details in Clinical Manager, it disappears, but if we update in e10, it will populate in Clinical Manager” (80). In addition to integration within internal systems, participants also commented on a lack of integration with external health data systems such as the Australian Government’s My Health Record system. “Linking with the government health system will help share information with doctors and hospitals” (29).

Discussion

This study identified eight major causes of data quality issues in CIS at a large Australian aged care and disability services provider. Causes described were multi-faceted, combining both internal and external factors, and together highlight the complex environment surrounding the quality of clinical data and documentation in aged care and disability services. A coordinated and collaborative effort is required between service providers and the wider sector, to identify optimisation solutions which support safe and high-quality person-centred care and improved regulatory reporting.

The majority of causes identified in our study parallel barriers and facilitators previously identified in hospital settings, indicating that data quality challenges are prevalent across different healthcare environments. These include equipment and resource challenges, staffing and workload, role clarity for data collection, extraction, and monitoring, duplication of effort, standardisation of free-text documentation, communication of change, systems support, and staff training [7, 21, 27, 53, 54]. These challenges can be addressed through effective and clear clinical data governance policies and procedures and organisational leadership with a strong digital strategy. Adequate information technology infrastructure is the foundation of effective clinical documentation [54], which needs to be supported with formal staff education and training on the proper use of CIS in their day-to-day work [7, 25,26,27, 55]. Additionally, role clarity with clear workflows is essential to enhance performance and data quality [56].

Additional causes specific to aged care and disability services were identified which require solutions from a sector and/or policy level. Many of the identified causes related to systems design and configuration can be attributed to reliance on vendor-based “shrink-wrapped” or off-the-shelf systems modelled on designs from other healthcare environments. In particular, previous research has shown that it is not common practice for software vendors to engage clinician end-users in shaping the development of health information systems according to their preferences which contributes to system design limitations [57]. Moreover, as CIS systems are expensive to buy and maintain [58, 59], a common practice within the sector is to use a single CIS to meet multiple organizational requirements. Although cost-effective, a one-size fits all approach may not adequately support the workflows of multiple business areas [60]. Configuration of dropdown menus, automated alerts, information mapping, integration with other information systems, and building a single view of crucial information to meet the specific workflows and health information requirements of aged care and disability services is essential.

While systems can be configured to a degree at the organisation-level, a sector-level solution involving collaboration between service providers, vendors and regulatory bodies would promote standardised and systematic information capture and improved regulatory reporting across the sector. Standardisation would also facilitate data linkage [8] and information transfer with external CIS such as hospital and primary care and increase continuity of care. Effective integration of systems and information transfer is recognised as essential requirement of modern systems development [61].

National standardised CIS for aged care and disability services in Australia could also alleviate some of the burden associated with clinical documentation and regulatory reporting from individual service providers, especially amidst workforce shortages and increased information requirements with the expansion of the National Quality Indicator Program [62]. In 2022, a shortage of almost 100,000 aged care workers was estimated in Australia [63]. Workforce shortages underlie several causes of data quality issues described by participants in our study including reliance on external agency staff, and deprioritisation of documentation tasks, second-hand documentation, and verbal handover practices due to workload.

Effective CIS design for aged care and disability services in Australia also needs to be tailored to the cultural and technical diversity of the workforce. Almost a quarter (23%) of direct care workers in aged care have English as a second language [64]. The aged care workforce is also older than the overall workforce in Australia [64] and are likely to have lower levels of computer literacy [65]. Despite this, current CIS in aged care and disability services consist primarily of unstructured free-text fields, with the quality of documentation ultimately dependent on the English language and technical skills of the workforce. Unstructured free-text data is also difficult to analyse on a large-scale to derive meaningful insights [66].

Finally, prior research suggests that a combination of solutions that address both behavioural and technical factors are most effective in improving health information systems [20]. Well-designed CIS can support higher-quality clinical data by guiding the workforce to input the required information more efficiently but may not address all process-related barriers, which requires strong organisational leadership with clear and effective clinical data governance policies.

Limitations

Study findings are limited to data-user experiences at a single service provider in WA and may not be generalisable to all aged care and disability services in Australia, particularly services which use different CIS than Clinical Manager and Carelink+. Moreover, although a balanced purposive sample of participants from all eight service areas were included, not all residential care facilities were consulted due to time and cost constraints. Likewise, convenience sampling was used which limited participation in this study to only a small sample of staff who were available at care sites at the time of visits. Finally, coder agreement measurement was not assessed.

Conclusion

The quality of clinical data and documentation within aged care and disability services is influenced by a complex interplay of internal and external factors. While internal factors can be addressed to a degree by strong organisational leadership and governance, a coordinated and collaborative effort is ultimately required between service providers and the wider sector to identify optimisation solutions. Such solutions should aim to standardise the quality of clinical data to support safe and high-quality person-centred care and improved regulatory reporting.