In Australia, aged care and disability service providers are legally required to maintain comprehensive and accurate documentation of the care provided to each client [1,2]. This clinical documentation is crucial to support safe and high-quality care delivery, maintain professional accreditation, and acquire government funding [3-5]. However, widespread clinical documentation and data quality issues have been identified across the sector, with evidence suggesting that most existing clinical data in the aged care and disability services sector are of substandard quality and lack consistency within and across individual organizations [3,6,7].

This problem has been partly because of the absence of national data standards and minimum requirements to guide information collection and documentation processes [3,7]. Therefore, service providers are challenged to collect extensive clinical information to meet multiple funding and regulatory reporting requirements across different and often segregated information systems [3,7]. Existing clinical information systems have also largely used “shrink-wrapped” or off-the-shelf systems owned by external vendors that are modeled on designs from other health care environments, with limited customization to account for the specific environment, workflows, and information requirements of aged care and disability services [8-10].

Investigations of workflow in Australian residential aged care facilities (RACFs) have shown that clinical documentation and information processing are major time-intensive staff activities [9,11]. A study involving 6 RACF sites in New South Wales and Victoria estimated that registered nurses spent a median of 60 minutes on documentation per shift, whereas service managers spent a median of 360 minutes on documentation activities, including filling forms, progress notes, incident reporting, and medication management [9]. Similarly, a second study estimated that RACF support workers spent a mean 14.5% of the total working time per shift on documentation tasks, which is equivalent to approximately 60 minutes for a standard 7.5-hour shift [11].

Although documentation is an essential activity of health care provision, excess time spent on documentation directly reduces the amount of care time spent with residents, which can compromise the quality of care delivery [12]. A survey in the United Kingdom found that 81% of the nurses believed they spent a disproportionate amount of time on record keeping and documentation tasks, which prevented them from providing direct care [13]. Other research has shown that the clinical documentation burden is a major driver of burnout among clinical and care staff [14,15], who believe that time could be better spent attending to residents [16-18], and can negatively impact job satisfaction [17].

Health IT (HIT) has a great capacity to support efficient information processing in aged care and disability services and facilitate high-quality person-centered clinical decision-making at the point of care [19,20]. Research has shown that the overall workforce and management perception of the use of electronic health records (EHRs) in aged care is positive, with widespread agreement that EHRs are beneficial for improving workforce efficiency [21]. However, there is a need to optimize HIT systems based on a detailed understanding of the workflow and requirements specific to aged care and disability services [9]. Several pre-post implementation studies in Australian RACFs have shown that the implementation of electronic systems alone does not automatically lead to greater efficiency [11,12,19] or data quality improvement [22] compared with paper-based systems. Instead, the implementation of suboptimal HIT systems can lead to unintended adverse consequences, including increased documentation time, difficulty in data entry and information retrieval, increased complexity of information management, and increased documentation burden and business costs [23].

Several common issues with EHR systems used in aged care and disability services have been described in the literature. End users have reported that existing systems lack clarity and contain inconsistent data fields, definitions, and terminology that do not match the specific information requirements for aged care and disability services [3,19]. When relevant input fields are unavailable, information is either omitted or recorded elsewhere, meaning that the EHR is often incomplete and does not contain all essential information necessary for the delivery of safe care [24]. Systems have also been reported to be difficult to navigate [25] and lack the structure and organization to guide the workforce to record the required clinical information efficiently [7,20]. A clinical documentation audit study in 7 Australian RACFs showed that although EHRs contained a greater quantity of information than paper-based records, information recorded in EHRs had a lower total mean quality score [22].

Other research has shown that end users can become reluctant to use EHR systems if they do not easily integrate into their workflow and often revert to using paper-based documentation [3]. This creates further inefficiency with respect to double documentation and the duplication of effort, contributing to a “vicious cycle” [19] of increased documentation burden and poorer data quality [3,9]. One study found that RACF nurses reported an excessive amount of time spent entering the same information across multiple incompatible systems and that double documentation resulted in the omission of important information, inaccuracy, and potential safety concerns [24].

In addition to EHR system–related issues, other complex process-related issues have been described that can contribute to missing, incomplete, inaccurate, or not up-to-date clinical data. In the primary care setting, for example, workload and time constraints [26], workforce attitudes toward documentation tasks, and prioritization of direct care tasks [27] have all been shown to affect the quality of nursing documentation. Well-designed EHR systems 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 related to clinical documentation.

Ultimately, as organizations become more data mature, important business decisions are increasingly reliant on analytics using administrative, clinical, and other service data [28]. Poor data quality can result in flawed business decisions and increased business costs [29]. Therefore, there is a clear need to optimize clinical documentation processes and clinical information systems to support efficient, safe, and quality care delivery and effective business decision-making in aged care and disability services [9].

However, any form of organizational change in systems or processes represents a major cost for service providers, which can be a major barrier to implementation. Therefore, it is important to be able to clearly demonstrate the added value and cost benefits of transformation initiatives to organizational leadership [19]. Currently, few aged care and disability service providers collect benchmarks to routinely evaluate and monitor data quality or its impact on care delivery and organizational efficiency [30]. Implementation without routine evaluation or a control plan can substantially increase business costs, while adding little value to care delivery or operational efficiency.

The overarching goal of the Optimizing the Quality of Clinical Data in an Australian Aged Care and Disability Service to Improve Care Delivery and Clinical Outcomes (OPTIMISE) study is to use an integrated Agile Lean Six Sigma (LSS) framework [31,32] to (1) identify opportunities for the optimization of clinical documentation processes and clinical information systems, (2) implement and test optimization solutions, and (3) evaluate postoptimization outcomes in a large postacute community-based health service providing aged care and disability services in Western Australia.

Specifically, the 3-stage study will include the following:

Table 1 presents the specific study objectives mapped to the Agile LSS phases (see the Study Design and Framework section).

The OPTIMISE study will be a prospective optimization study using an integrated Agile LSS Define, Measure, Analyze, Improve, Control (DMAIC) framework.

Six Sigma is a statistical measurement-based method for process optimization and quality improvement that aims to reduce the number of defects in a process to <3.4 defects per 1 million opportunities [33]. Lean methodology is focused on reducing waste in a process, which is defined as any unnecessary or suboptimal items, actions, tasks, components, materials, systems, or human resources that increase costs and time spent on a process [34]. Therefore, the Lean Six Sigma framework aims to reduce both defects and waste to improve the quality of products and services, improve efficiency, and reduce business costs [35,36]. Originally conceptualized in the manufacturing industry, LSS has been increasingly applied in health care to control rising costs and improve the quality of care delivery [30,36-38].

The strength of LSS is its rigorously structured framework for quality improvement based on statistical measurement. However, this structure can lack flexibility to accommodate changes across the project cycle [32]. Agile methodology can complement LSS to increase flexibility and adaptability [39]. Agile is a popular methodology used in software development that involves the iterative development of a product over short-term incremental cycles called sprints [31]. It is a highly flexible and responsive methodology that allows for product changes during development based on customer feedback. However, this method neglects decision-making based on objective and measurable data, which can compromise quality and efficiency. Therefore, the integration of Agile and LSS frameworks provides a structured methodology for product development, evaluation, and improvement that is simultaneously flexible and responsive to changes over iterative cycles. Hybrid Agile LSS frameworks have been increasingly used in the literature as a more powerful method for quality improvement than single frameworks [39-42].

The study will be undertaken at Brightwater Care Group (“Brightwater”), a large postacute, residential, and community-based aged care and disability service in Perth, Western Australia. Brightwater provides residential and home-based care for approximately 1882 aged care and 522 disability services clients across 10 different programs (Table 2).

Patients or the public were not involved in the design, conduct, reporting, or dissemination plans of our research, as the end product (clinical information system) is not directly used by patients or the public.

The study protocol was approved by Brightwater Care Group as a Level 2: Low Risk Study (reference: 2021/BCG2110). This study is classified as an internal service evaluation and does not require ethics approval for research in Australia.

Client EHR data will be deidentified before the analysis. Clients provided written consent for their deidentified clinical data to be used for research and service evaluation as part of the conditions of entry into the service. Staff member participation in user surveys and interviews will be voluntary, and nonparticipation will not impact their usual role. All participant data will be deidentified by assigning a randomized numbered code. Participant names will be changed to pseudonyms when referring to qualitative data. When transcribing interviews verbatim, words or statements that could be used to identify participants will be removed to ensure privacy. Audio recordings will be deleted once the transcriptions are complete.

The primary study outcome is data quality measured using DPU, DPO, DPMO and Sigma level. Improvement in data quality from T₀ to 12 months after optimization will be determined by a reduction in DPU, DPO, and DPMO and an increase in Sigma level. A well-performing process should operate at a Sigma level of 6 and have ≤3.4 DPMO [43].

Quantitative analyses will be conducted using STATA (version 16.0; StataCorp LLC) [55]. Multilevel mixed effects regression models will be used to evaluate any change in primary and secondary outcomes from T₀ to 12 months after optimization. For clustered longitudinal data, multilevel modeling recognizes that a change in outcome is affected by a fixed effect (ie, implemented changes) and random effects at both individual and group levels and can explicitly account for this multilevel random variation [56].

An a priori power calculation was performed to determine the minimum number of EHRs required to be evaluated to detect a significant change in our primary outcome (ie, data quality measured using DPU, DPO, DPMO, and Sigma level) from T₀ to 12 months after optimization. First, we used G*Power [57] to compute the required sample size to detect a medium difference (f=0.15) at a Cronbach α level of .05 and a power level of .80 using a linear multiple regression analysis with k=3 fixed predictors (time, program, and the time × program interaction), which returned a required sample size of 77.

This was multiplied by the anticipated design effect [58], which is an adjustment factor for clustering in multilevel models. The design effect is calculated as:

1 + [(n − 1) × ICC]                       (1)

where n is the expected number of subjects per cluster and “ICC” is the intracluster coefficient or the expected correlation within clusters [58]. We defined a cluster as a facility (ie, site) with a median cluster size of 30 clients per facility. With an anticipated moderate intracluster coefficient of 0.10, the design effect is equal to 3.9, giving a required sample size of 77 × 3.9 = 300. On the basis of the occupancy rates and client population size (n=2200) as of January 2022, our primary outcome analysis will have sufficient statistical power.

Qualitative data from user interviews and surveys will be coded and analyzed using NVivo 12 (QSR International) [59]. An inductive thematic analysis approach will be used to identify and understand key themes, specifically to classify, order, and reassemble data to identify converging and diverging perspectives. A total of 50 interviews are anticipated to reach saturation.