Health services in Australia and the impact of antiquated rostering practices on medical scientists: a case for HR analytics and evidenced-based human resource management Free

The work of medical scientists in Australia is critical to the healthcare system, and therefore providing HR analytics data on these workers, work processes and related HR outcomes should be considered by management as “a vital part of the continuum of patient care” (Church and Naugler, 2020, p. 324). The nature of medical scientists' work has changed drastically in recent decades through technological advancement, automation, increased work volume, budget constraints, efficiency drives and diversified work hours (DHHS Victoria, 2018; Plebani et al., 2019; Willis and Weekes, 2005). In this study, we examine the challenges faced by medical scientists and in particular the impact of sub-standard rostering practices that impact on their daily work. We argue that incorporating appropriate HR analytics around rostering practices and evidence-based human resource management (EBHRM) into decision-making is critical to mitigate various negative influences on medical scientists and their erratic work schedules. In the case of medical scientists, we have selected a broader definition of HR analytics by Collins (cited in Lalwani (2019, para.25):

In Australia, there are 701 public hospitals, and most have pathology services, employing over 16,000 medical scientists (AIHW, 2019). However, there are systemic shortages of key medical scientists in hospitals due to rostering challenges that potentially put patients' well-being at risk (Knust and Xie, 2019). Difficulties associated with the medical science workforce in Australia have been of on-going concern to the sector and to both Federal and State Governments (Badrick and St John, 2012; DHHS Victoria, 2018; Weekes, 2002). Work intensification has become a major challenge, and further exacerbated by poor rostering practices within pathology services (Badrick and St John, 2012; Legg and Associates, 2008; Weekes, 2002). Many healthcare institutions continue to use manual rostering processes (Knust and Xie, 2019; Petrovic and Berghe, 2012), which compromises HRM outcomes such as recruitment of medical scientists, absenteeism and their intention to leave (Silvestro and Silvestro, 2008). Given the complexity and essential nature of 24-hour medical science work, issues around employee burnout and absenteeism (Isouard, 2012), we examine ways to improve the rostering practices of medical scientists.

We contend that EBHRM may be important to improve rostering practices as it provides a detailed and systematic approach to decision-making on HRM policies and practices. EBHRM is defined as “making decisions, promoting practices and advising the organisation's leadership” by drawing on “the best available scientific evidence; reliable and valid organisational facts, metrics and assessments; practitioner reflection and judgement; and the concerns of affected stakeholders” (Rousseau and Barends, 2011, pp. 222–223). We examine the role of medical scientists and the significant challenges they face because of the erratic rostering practices at each of the health services. We do this by focussing on the core workforce issue of rostering, where current practices are typically low-tech, informal and unstructured (Knust and Xie, 2019; Petrovic and Berghe, 2012). This study aims to understand how to improve rostering practices of medical scientists through EBHRM underpinned by the LAMP model (Boudreau and Ramstad, 2007) and HR analytics. The LAMP model incorporates logic, analytics, measurement and process to offer a detailed approach to diagnostics and measurement of management problems. Our study is informed by two research questions:

This paper draws on interviews and focus groups with medical scientists and managers/line managers at four Australian public hospitals, as well as rostering samples from a three-month period for pathology services at each health service. We contribute to the literature in two ways. First, we unpack the views and working relationships between line management and medical scientists relative to rostering practices. We contribute to the dearth of literature on the relationship between HRM, line management and rostering. Second, we propose the integration of the LAMP model (Boudreau and Ramstad, 2007) and HR analytics into an evidence-based HRM approach to examine and inform HR-line management decision-making.

The medical science profession constantly experiences work intensification through longer hours, increased shift-work, and the need for larger workloads (Willis and Weekes, 2005). A key challenge has been operation of rostering systems to meet increased demand by implementing non-standard work hours (i.e. shift work) and round-the-clock work (Weekes, 2002). Meanwhile, a range of problems associated with rostering practices in this profession have been identified, including poor shift arrangements, inadequate breaks, excessive and unpaid overtime and inadequate opportunities for training and development (AIMS, 2009; DHHS Victoria, 2018). This has coincided with reports of higher levels of stress, fatigue and burnout amongst medical scientists (DHHS Victoria, 2018). Silvestro and Silvestro (2008, pp. 102–103) note that: “hospital performance relies critically on the competence and effectiveness of roster planning activities … these activities are therefore of strategic significance”. However, most healthcare institutions, including pathology services, continue to use manual rostering with inconsistent and unclear processes (Drake, 2014; Knust and Xie, 2019; Petrovic and Berghe, 2012; Silvestro and Silvestro, 2008).

Poor rostering practices can impact organisational service delivery, budgeting, compliance and performance (Petrovic and Berghe, 2012; Silvestro and Silvestro, 2008; Van den Bergh et al., 2013). A current lack of data analytic decision-support around rostering therefore has significant implications for the performance of medical scientists and pathology services. In public healthcare, performance outcomes have been classified as human resource outcomes (such as job satisfaction, retention and social climate) and quality of patient care (such as reduction in errors, patient satisfaction and clinical outcomes) (Leggat et al., 2011, p. 283). For example, rostering can impact staff outcomes for medical scientists such as perceptions of fairness and manageability of workloads.

Information sharing and transparency around organisational performance and operational systems, such as rostering are important to signals trust in employees (Bartram et al., 2014; Pfeffer, 1998). Trust can be undermined with uncertain processes that can impact staff satisfaction, morale and retention (Drake, 2014). An important part of rostering is determining the balance of workload between employees and staffing levels needed to complete the workload, including when employees are unavailable (Van den Bergh et al., 2013). High workloads requiring prolonged exertion and inability to take timely breaks are associated with fatigue, impaired performance, errors rates, job dissatisfaction and turnover (Dall’ Ora et al., 2016). Rostering practices must accommodate these considerations to ensure human resource and quality of patient care outcomes (such as job satisfaction, minimisation of errors and clinical outcomes).

EBHRM is underpinned by “evidence-based practice … … for making decisions that integrate the best available research evidence with decision maker expertise and client/customer preferences to guide practice toward more desirable results” (Rousseau, 2006: 258). EBHRM involves decision-making and management practices informed by the best available scientific evidence (Rousseau and Barends, 2011; Rousseau and McCarthy, 2007). According to Rousseau (2006) evidence-based practice is characterised in the following way: (1) understanding causal relationships in professional work; (2) isolating factors that demonstrably impact the attainment of strategic outcomes; (3) creating a culture around evidence-based decision making; (4) using information technology and other data gathering systems to minimise the overuse, underuse and misuse of work practices; (5) developing decision making structures to promote work practices that are supported by evidence, along with other organisational policies and practices that support decision implementation; and (6) promoting access to knowledge and evidence-based management protocols amongst individual, organisational and institutional actors.

EBHRM is important for line managers to make informed decisions about their workers to minimise workplace risks, improve their performance and work experiences (Marler and Boudreau, 2017). HR and line managers need to make decisions that are founded on human capital data and any other work-related evidence by applying “the best available scientific evidence; reliable and valid organisational facts, metrics and assessments; practitioner reflection and judgement; and the concerns of affected stakeholders” (Rousseau and Barends, 2011, pp. 222–223). In many healthcare organisations, archaic HRM practices and technological systems are used (Tursunbayeva, 2019; Andersen, 2017). Lawler (2007) suggests that despite the growth in the use of HR information systems by many HR departments, most HR managers and line managers do not practice EBHRM in their day-to-day work activities, due in part to skill deficiencies and understanding of EBHRM protocols. McIver et al. (2018, p. 398) contend that for most organisations “HR is usually on the bottom of the technology resource allocation plans” and data analytics programs are disregarded. Meanwhile, there is growing evidence that EBHRM in rostering can be used to manage workflow and capture HR information to improve feedback and evaluative mechanisms and inform improved management systems (Hasson et al., 2018). Quality HR data are the key to enact EBHRM for effective decision-making (McCartney and Fu, 2022). Therefore, when detailed information is available on the negative impacts of antiquated rostering systems on medical scientists, management should be able to apply evidence to make decisions that enhance employee and organisational outcomes.

The growing availability and access to HR technologies such as human resource information systems (HRIS), mobile applications and cloud platforms offers HR and line managers greater scope to collect, manage and analyse employee data (McCartney and Fu, 2022). Such ability to interrogate high quality data is a key element in the development of organisational HR analytics (McIver et al., 2018; Minbaeva, 2018) to inform managerial decision-making (Dulebohn and Johnson, 2013). HR analytics involves sophisticated analysis, using “data related to HR processes, human capital, organisational performance, and external economic benchmarks to establish business impact and enable data-driven decision-making” (Marler and Boudreau, 2017, p. 15). According to McCartney and Fu (2022) HR analytics not only involves examining and enhancing functional aspects of HRM but also applying analytical techniques and HR data to improve organisational strategy and performance. This includes data related to rostering and such analysis requires appropriate software, resourcing, competencies and analytic and strategic capabilities (Andersen, 2017; Minbaeva, 2018). Andersen (2017, p. 134) observes that HR analytics is still in its infancy, due to factors such as “lack of the implementation of good software solutions, bad data/lack of proper data, too few resources as well as lack of organisational wide buy-in”. Data and analytics can help to provide evidence for operational control (effective and efficient completion of day-to-day HR activities in the organisation, such as “employee record keeping, wage and salary administration, and benefit enrolment” (Dulebohn and Johnson, 2013, p. 76)). HR analytics may well enhance the performance and wellbeing of medical scientists by providing line managers with data-driven insights into HR processes, human capital and employee performance to improve rostering practices (Marler and Boudreau, 2017). In the case of pathology services HR analytics is not considered a priority and hence antiquated rostering practices continue (Andersen, 2017).

The acronym LAMP (Boudreau and Ramstad, 2007) represents logic, analytics, measurement and process and we draw on this theoretical framework to examine current decision-making around the rostering of medical scientists and their work experiences.

Boudreau and Ramstad (2007) postulate each one of the elements of the LAMP model represents one element of measurement. Logic is about searching for understandings and establishing what associations need to be made in relation to what is happening within a specific context such as in health services to inform decision-making about how best to support medical scientists through rostering in pathology work. Jones et al. (2019) advocate the use of logic models to first develop analysis tools and then to implement, analyse and evaluate data for effective outcomes. Therefore, logic in relation to medical scientists and rostering practices at work could be about unpacking job design, communications and relationships between management and medical scientists and related employee and work outcomes.

Analysis refers to how to provide the information needed to change and transform data into knowledge to support decision making (Boudreau and Ramstad, 2007). In this study, we focus on rostering data that impact on medical scientists that often fail to meet the requirements of the enterprise agreement. One of the critical issues of analysis is knowing whether you have someone in the organisation who has the expert skills to draw conclusions from the data that are accurate (Lawler et al., 2004). This includes quantitative and qualitative data and interpreting such data to be useful and presenting it in a rigorous and cohesive way that will support management and employee decision making around rostering.

Measure is about knowing what to quantify or understand how to recognise levels of performance, engagement, absences and turnover (Boudreau and Ramstad, 2007). For instance, there are numerous technical solutions to determine turnover and the costs associated with losing staff, but the measurement of their performance and engagement should be included to present a more holistic perspective. Therefore, in the case of medical scientists it is important to measure their workdays, hours, how work rostering is determined, the level of absenteeism and possible reasons for turnover.

Process refers to the change management needed to create different procedures and processes based on the analysed data (Boudreau and Ramstad, 2007). Process, measures, decisions and behaviours are impacted by social and cultural norms and levels of employee knowledge and skills (Boudreau and Ramstad, 2007). Ben-Gal (2019) argues that most cases studies in the HR analytics literature have failed to use the LAMP model. Therefore, we take up the challenge to present a thorough examination of data by not only applying the LAMP model but integrating it with HR analytics and EBHRM to support line management decision-making as it relates to the rostering practices for medical scientists.

We argue for the integration of the LAMP model, HR analytics and EBHRM. EBHRM can be viewed as the architecture to support informed decision-making processes for HR and line managers. The LAMP model is useful to identify and understand a critical problem or phenomenon, its cause(s) and impact(s) (e.g. rostering processes and outcomes). Moreover, data are gathered through HR analytics systems (HRIS) and purpose-built rostering applications that are used to gather, measure, analyse and present data informed by the LAMP model. HR analytics requires EBHRM architecture so that line managers can make strategically informed decisions for the benefit of various stakeholders. Using EBHRM, we argue that the application of HR analytics guided by the LAMP model is useful to produce detailed understandings of rostering related challenges which then inform HR and line management decision-making.

This study took a qualitative case study approach (in lieu of a survey) which was a collective decision of the four Australian public hospitals and stakeholders. The study was commissioned and funded by the Victorian Department of Health and Human Services and supported by the Medical Scientists Association of Victoria which is part of the Health and Community Services Union of Australia Number 4 Branch. A condition of the project, and to protect the identity of the health services and the medical scientists, we do not provide descriptions that could reveal the locations or the workers.

Gatekeeper access was assured, and ethical clearance was obtained from RMIT University – Human Research Ethics Committee (HREC) (project no. 21648). The study focused on medical scientists working in pathology departments comprising Anatomical, Biochemistry, Haematology and Microbiology and the Chief Investigator contacted each Health Service and obtained organisational consent. At each of the health services there was a pathology workforce comprising 107 medical scientists at health service A, 138 at health service B, 135 at health service C and 120 at health service D. Each health service had at least two managers/directors in each department and in there were between 8- and 10-line managers across pathology. The human resources manager was contacted at each of the health services and managers/line managers and every medical scientist in pathology received a group email. Prospective participants then contacted the HR manager and registered their interest in attending focus groups to support the study.

The research team then approached prospective participants via email to attend interviews and focus groups; we outlined the purpose of the research and assured the participants confidentiality (Schensul, 1999). Participant information statements and voluntary consent forms were then provided to each participant and signed consent obtained prior to the commencement of data collection. Data were collected from semi-structured interviews with 21 managers/line managers, each for approximately 45 min. Interview questions were around how the roster system was managed, how managers consult with staff about the rosters and challenges of the roster system that impact on medical scientists. A total of nine focus groups were conducted across the four Health Services with a total of 53 medical scientists and each focus group lasted for approximately 1 h. We asked questions about how flexible the roster system was, if and how rosters are changed with or without consultation and how the rosters impacted on their work–life balance.

The medical scientists who participated were predominantly female with approximately 5% males. Of the 53 medical scientists there were 30 part-time scientists and 23 full-time scientists. The participants held university qualifications, and most were multi-skilled in pathology services but assigned to anatomical, biochemistry, haematology, or microbiology departments. Participants ranged from entry medical scientists to senior scientists with over 20 years-experience.

To protect the identity of the participants, in our reporting we refer to the Director or Manager at each of the health services as manager in accordance with a letter of the alphabet, for example Manager_HSA or Manager_HSB and so on. We refer to medical scientists according to the Health Service, for example MS_HSA or MS_HSB and so on. To ensure reliability and validity (Creswell and Miller, 2000) the researchers searched for the convergence of different sources of information to formulate categories and themes within the data.

At each of the medical services we were provided with computerised spreadsheets covering a three-month period that provided work allocations and absentee data, but the spreadsheets were not easy to follow. Data were changed with handwritten notes which made it difficult to arrive at an accurate number of staff who were absent in any one week or month.

The interviews with managers/line managers were initially transcribed and then analysed using NVivo, following the steps of content analysis outlined by Weber (1990). NVivo is a tool that aids in categorising and coding the raw data (Bazeley and Jackson, 2013) and was employed for the first stage analysis of the research data to identify themes in the data (Weber, 1990) through searching for patterns in participants' responses (Yin, 2013). The transcript of each recorded interview, handwritten interview notes and focus groups, were also coded independently by two coders until saturation. The coding of thematic data were defined by words extracted from the interview and focus group questions such as “rostering”, “rostering practices”, “workload”, “absenteeism” and phrases such as “operational decision making” and “application of the enterprise agreement”. Each coder read the transcripts of raw data and inter-rater reliability was determined by their frequency of agreement thus ensuring the reliability of the coding framework (Yin, 2013). Where there was disagreement between the coders, a third rater helped finalise the coding. The results of NVivo and the thematic analysis of the coders were combined to reach agreement on the main themes in the data.

In presenting the findings, we apply the LAMP model (Boudreau and Ramstad, 2007) to provide insights into the stories of managers and medical scientists who work in pathology across four health services. We argue that it will be the in-depth narratives of the participants who will provide us with clearer understandings of what happens in pathology at the health services and what needs to change. Moreover, we suggest the findings will highlight the need for HR analytics and EBHRM.

Managers and line managers from the four health services who were interviewed concurred that it was difficult to ensure effective and efficient management of rostering for medical scientists. Managers explained that rostering and decisions made about rostering are fundamentally impacted by budget constraints. One manager stated, “every manager is held back by the budget ….and we struggle to manage staff and meet the demands of the Enterprise Agreement (EA)” (Manager, HSA). From the managers' perspectives there is not a rostering system in existence that accommodates all the restrictive provisions of the legally-binding enterprise agreement (EA): “No one has come up with a roster system that meets the EA” (Manager_HSA). Managers explained how unstructured and complex rostering is:

Managers corresponded that “the workload for Medical Scientists is almost criminal” (Manager, HSA) and another claimed, “medical scientists are so overworked, but they push themselves to the limit to get the work done ….patients are always at the forefront” (Manager, HSB). Managers told us it was procedural and logical that rosters are produced on Excel spreadsheets before different types of rosters (such as annual rosters, fortnightly rosters, quarterly rosters and availability rosters) are printed and pinned to corkboards or on walls within each department. Managers agreed the rosters were reasonable to this point but “within hours numerous changes are made and then the rosters are difficult to interpret” (Manager HSA). Managers explained that each department has a different method of presenting roster data, and the legends, symbols and acronyms for each roster document were inconsistent and not evident to a reader.

Due to managers (in interviews) and medical scientists (in focus groups) often referring to the EA we examined two sections of Clause 58 that were claimed to be problematic in terms of how management should ensure medical scientist do not engage in overtime:

1. When overtime work is necessary it shall, wherever reasonably practicable, be so arranged that employees have at least 10 consecutive hours off duty between the works of successive shifts. 


2. An employee who works so much overtime between the termination of their previous rostered ordinary hours of duty and the commencement of their next succeeding period of duty such that he/she would not have had at least 10 consecutive hours off duty between those times, shall, subject to this clause be released after completion of such overtime/recall worked until he/she has had 10 consecutive hours off duty without loss of pay for rostered ordinary hours occurring during such absence.

Whilst management should ensure medical scientists have 10 consecutive hours off duty after overtime medical scientists told us “we do not get ten hours off after overtime” (MS_HSA), “some nights, after I've done overtime, I get as little as four hours sleep before returning to my next shift” (MS_HSB). We were told consistently that medical scientists do not receive adequate breaks in accordance with the EA.

When we attempted to analyse the spreadsheets in each department, they were a conglomeration of pencil, pen and text indecipherable writings which made each spreadsheet appear chaotic. We could not make sense of the changes and could not determine if they were due to swapped shifts, cancelled shifts, sick leave, carer's leave or any other type of leave. All managers agreed that the information on the rostering documents should represent accurate information for the work schedules of each medical scientist. However, the fact rostering spreadsheets were changed manually, most acronyms did not make sense, codes were not easily understood from one department to another, was identified as a major issue for managers/directors.

The Director at HSA explained he was not able to interpret every aspect of the rosters across Anatomical, Biochemistry, Haematology or Microbiology. “There are some parts that are obvious” but others are “very confusing in parts”. He articulated a basic understanding of what each of the rosters are meant to provide but when we asked for details, he could not explain most of the acronyms or provide explanations for why some staff appeared to work excessive hours that were not compliant with the EA. The manager of HSB noted a major issue when reading the rosters as rotating shifts which meant if staff work a night shift then according to the EA they have to have two days leave. The manager told us he could not possibly give a medical scientist two days off before a weekend because “it means four days off”. The change was immediately made to a roster spreadsheet, and he was open in terms of admitting he had no intention of complying with the EA because “the work has to be done”. All managers had issues with rostering practices and expressed their concerns about making decisions for medical scientists' workloads and often outside the EA:

Managers explained that planned rosters, drawn up on computerised spreadsheets, were very different to the rosters that are “actually worked”. Shift work, including night shift and on-call arrangements and irregular start and finishing times are initially rostered on a spreadsheet and then reverted to a manual system. The changes made the roster spreadsheets very difficult to understand.

In the focus groups discussion was animated with frustration when medical scientists talked about how to interpret rostering spreadsheets. Medical scientists told us that the rosters were driven by management to get the work done rather than listening to them about the challenges around rostering. Medical scientists were in agreement that the decision-making around rostering did not take into consideration worker needs and did not meet the EA and this was reflected in the erratic spreadsheets. Medical scientists' views on how to analyse roster documents focused on inequity in terms of part-time medical scientists and perceived favouritism in the decision-making of managers. They gave the following examples of managerial decision-making and inequities that clearly did not demonstrate concern for fair rostering practices:

In each of the interviews, the managers expressed how rostering practices are “unpredictable”, “erratic” and “dominated by sick leave” and “absenteeism”. Managers concurred that the one thing they are certain of is that the rosters can be measured by days that are “plagued with absenteeism”. The Director at HSA clarified “staff members on the floor are not coping” and there is one certain factor “every day there will be up to five medical scientists in any one department on sick leave”. Medical scientists find themselves in situations where they “might have partners who can't mind the children” and rather than finding alternate arrangements for childcare and other personal commitments they take sick leave.

Managers concurred that high levels of absenteeism and sick leave put pressure on those remaining in the workplace and led not only to extra stress in terms of getting the work done but also the fear of making mistakes which can potentially impact on patients' lives.

In the course of their daily work medical scientists described an average day at work as “chaotic”, “highly pressured” and “stressful” because of the excessive “workload” created by rostering practices that are certain to have a number of staff on “sick leave”. This is a clear expression of the impact of current practices on quality of work, and in particular employee workloads. Medical scientists expressed their frustrations around the operational structures:

Full-time medical scientists across all four health services also told us that the rosters can be measured by the number of part-time medical scientists and that they find it “challenging to work around part-time staff”. A contributing factor to the failure of the rostering system was viewed by the medical scientists as a lack of management transparency and a system influenced by management favouritism for part-time medical scientists:

Managers concurred that the current rostering systems across the four health services do not work and in fact they “create more work”. They suggested that part of the operations around a lack of effective communications present issues for managers and impacts on the daily work of medical scientists. The managers at each of the health services were very open in their views about how communications do not work well and need to change:

Medical scientists agreed that change needs to be around not just data analytics but examining data that may inform improved rostering systems. Medical scientists expressed their views on “poor operational communication systems” and what needs to change in each of their workplaces to support them. Workers told us that the rosters should be communicated to them monthly but “that never happens in a systematic way” or “with accurate information”. The focus groups participants talked about rostering changes that happen on a daily and often on an hourly basis which is very challenging:

The medical scientists also reflected on practices outside the EA that need to change and how they are impacted by unfair management rostering practices. One issue discussed in great detail was the fact “part-timers (medical scientists) do not work night shift”. This causes a multitude of problems because first full-time medical scientists are basically “forced” to work night shift; second, increasing pressure on night shift workers causes “stress and burnout” and their health and wellbeing “is not a management consideration”; third, if “we work night shift we do not get two days off”; management is constantly reminding us “they do not want the union involved” and we have to consider “how we will be treated later if we do get the union involved”. We talked about the rostering system and any other communication that may capture the concerns of medical scientists, but it was unanimous “there's nothing”, “our voices are not heard”, “no one cares”.

Nevertheless, medical scientists and their concern for patients lead them to make extra efforts to cover and support each other and maintain quality of patient care often to their own detriment which leads to cycles of sickness and absenteeism. They told us “we work at least 20 min extra every day”. Medical scientists exclaimed in frustration that they want change around staffing numbers, sick leave, workload, potential risks for patients and how they are treated by doctors and nurses:

To synthesise, the decision-making around rostering, the lack of operational structures and the lack of support from data analytic systems, we heard from managers that there is an increasing pressure to seek data gathering, and interpretation and analysis of data to better support medical scientists. Managers confirmed there were large gaps in the rostering systems due to manual processes to organise rostering. One manager (HSA) expressed a view that was consistent with all managers, “the way we have to manage is reactive rather than proactive …. we're constantly applying survival strategies”. The allocation of attention and time from pressing managerial responsibilities highlight the main issues for managers around meeting the EA, working to budgetary restraints that impact staff numbers and managing absenteeism and unplanned leave.

Medical scientists are challenged by the intensity of work and a constant workflow without adequate breaks. Data around work intensification are not currently being collated or communicated to determine human resource needs. Data on the strengths and deficiencies of current human capital and associated issues (such as absenteeism, actual time worked and understaffing), are not being effectively captured and analysed to inform planning decisions around rostering to meet staff requirements.

Given the breadth and complexity of rostering practices for medical scientists it is clear there are deficiencies in current systematic approaches to manage this cohort of workers. At the operational level, rostering practices in each of these pathology services were manual and lacked structured support. In these circumstances, we found evidence of ineffective rostering practices due to the fact none of the health services in this study maintain formalised, structured rostering systems. The roster systems across each medical service are inconsistent; no two departments within each medical service maintain an identical roster system. Our findings indicate that uptake of information around rostering is limited in each of the pathology services which impacts medical scientists' stress and burnout and their overall wellbeing. The diversion of management attention from decision-making priorities, compromises organisational objectives (such as high absenteeism and litigation risk through increased potential for testing errors) and a lack of recognition of the nature of medical science work in strategic planning (such as expectations around testing times, workload and staffing levels). These deficiencies undermine performance outcomes in public healthcare, including those relating to human resource outcomes (such as job satisfaction) and potentially compromise quality of patient care (such as minimisation of error risk) (Leggat et al., 2011).

Current operational-level decision-making around rostering was found to generally be low-tech, manual, informal, often unclear and not conducive to operational transparency. Due to a lack of data-gathering and analytics related to these outcomes means that there are insufficient data to support evidence-based strategic-level decision-making (e.g. adequate staffing levels). We found no evidence of an EBHRM approach to using HR analytics to inform decision-making. By recognising where HR decision-making can best draw on information systems and decision support, the objective is to fulfil the requirements of this kind of task “as efficiently and accurately as possible” (Dulebohn and Johnson, 2013).

It is clear from our findings that staffing levels need to be closely monitored and increased in the pathology services to alleviate any rostering shortcomings (Badrick and St John, 2012; Isouard, 2012). Currently, there is insufficient consideration of these issues at strategic planning levels in these organisations and by government. This is compounded by a lack of respect and recognition of medical scientists, with their work being mostly hidden from the public, hospital senior management and other clinicians, leading to misconceptions about the complexity and skilled nature of their work and the excessive demands being placed on medical scientists in pathology.

Our research demonstrates that HR analytics have been neglected as a key tool in building EBHRM in healthcare organisations. We argue that health services need to invest in examining the stories of medical scientists around rostering and generate data to design systems that will meet the needs and autonomy of these critical workers. Moreover, we argue that HR analytics informed by the LAMP model and underpinned EBHRM architecture should be used to capture quality data on work issues (e.g. rostering challenges and related outcomes) identified to substantiate and guide managerial decision-making.

We contribute to HRM theory by integrating the LAMP model, HR analytics and EBHRM to understand the processes through which line manages can improve their decision-making to achieve strategic outcomes which positively affect various stakeholders (e.g. medical scientists). Critical to integrating the three approaches is to establish an EBHRM architecture to guide and implement HRM policies and practices to inform decision-making processes. Importantly, EBHRM protocols should incorporate the LAMP model in which line managers' decision-making take logic, analysis, measurement and process to identify and understand an HR problem (e.g. rostering). Moreover, an important part of this process is to gather stakeholder data through HR analytics systems and analyse, present and interpret such data. To enhance line manager decision-making, it is critical that they understand strategic goals and associated outcomes, challenges and their root causes inhibiting their achievement, as well as the effects on key stakeholders (e.g. employees). An important part of EBHRM is creating a culture of evidence-based decision making underpinned by HRIS and data analytics software, HRM policies and practices (Rousseau, 2006).

We contribute to the HRM literature about how current HRM rostering line-management decisions are reactive or as Boxall et al. (2007) suggest temporal decisions founded on no feedback and for the short-term. Our study demonstrates that absenteeism is unrestrained at each of the health services. We argue, HRM decisions should be underpinned by EBHRM and based on data analytics, guided by the LAMP approach to consider the causes and consequences of absenteeism and to inform long-term outcomes such as enhanced wellbeing and retention of medical scientists. Applying the elements of the LAMP model is critical to assess the logic, analysis, measure and process related to absenteeism. We collected data from managers and medical scientists and every participant provided examples of how absenteeism is driven by rostering (e.g. understaffing, low budget, excessive workloads, stress and burnout). Therefore, by examining HR analytics, data decisions can be made through EBHRM to inform managers, and benefit medical scientists and the health services.

Our study highlights organisational issues in rostering practices in hospitals where HR analytics underpinned by EBHRM are not used by managers. Current rostering practices are inconsistent and use antiquated practices that are dominated by manual changes. We would encourage EBHRM architecture be put in place within the medical science departments to develop improve decision making to promote work practices that are supported by evidence. This may include the implementation of management processes to understand causal relationships in professional work and how they are related to strategic goals and stakeholder outcomes, creation of a culture around evidence-based decision making, and promotion of access to knowledge and evidence-based management protocols which use information technology and other data gathering systems (Rousseau, 2006).

Data could be collected to inform strategic decision-making around staffing, such as employee actual time worked (incomplete breaks, unpaid overtime), unplanned absences, staff turnover, rostering adjustments and unfilled rosters (not filled with current staffing). Significantly, this would provide data on where current rostering rules and staffing are inadequate to ensure organisational performance. We also recommend hospital-wide strategies to enhance the status and recognition of the importance of medical science to the work of other clinicians (e.g. doctors) to enhance the respect for medical scientists (e.g. a medical scientist on hospital board).

At an operational level, rostering systems could apply individual contracts (part-time, full-time and casual) and preferences embed decision rules to enforce additional requirements such as the EA and contingencies due to unplanned leave, and so on. This task could be consolidated through managerial discretion where necessary (for example, when there are unforeseen staff absences). This could also help to build workforce skills and worker motivation in relation to professional development in two ways; first, by embedding formal communications into rostering and second, by reducing the time spent by senior medical scientists on administrative matters so that more time can be dedicated to mentoring junior scientists (DHHS Victoria, 2018).

This study is not without limitations. We undertook interviews and focus groups at four health services in only one state of Australia with managers and medical scientists. We did not interview doctors or nurses who use pathology services. For privacy reasons, we were unable to access more objective performance information, such as medical error rates and adverse events. The study did not include a survey of medical scientists state-wide or across the country.

A research project that examines error rates and adverse events could possibly provide deeper understandings of the issues that emanate from outdated and poor rostering practices. We encourage future research that includes a survey to examine health services across the country. Such research could adopt quantitative methods (e.g. hierarchical linear modelling) to unpack the relationship between employee and management perceptions of rostering and employee participation mechanisms and their effects on employee burnout, health and wellbeing and objective measure of quality of patient care.

In sum, the findings indicate managers do not challenge the dedication of the medical scientists, the work they perform or the personal sacrifices they make through unpaid overtime, shortened or missed breaks and compromised wellbeing, to meet organisational goals. Managers agree that the current operational issues are “systemic” and relate to “a lack of funding” and “understaffing”. It is evident that work-related data are not gathered or considered in strategic decision-making around staffing needs in any of the four health services of this study. We argue that building an EBHRM architecture that uses HR analytics and the LAMP model may improve HR and line managers' decision-making around rostering and may also apply to other work practices. Medical science work is a unique occupation, not respected by other clinicians, hospital management or the public. Whilst medical scientists are highly skilled experts, their work is repetitive, stressful and hidden. Our findings demonstrate the potential benefits of integrating the LAMP model, HR analytics and EBHRM to improve rostering practices and mitigate the decline in the wellbeing of medical scientists.