Caught between profitability pressures and staffing shortages, many health care organizations in the United States struggle to deliver on their mission; some even struggle to continue to operate, eliminating services or closing.1 Hospital operating margins remain below pre-pandemic levels, particularly for nonprofit hospitals.2 Staffing shortages also persist: In a 2023 survey, two-thirds of hospital executives reported that staffing shortages caused their organizations to board patients and run at less than full capacity.3 Moreover, the shortages are not just among frontline clinicians; they are felt in the back office too, with a negative impact on the margins, teams, and care delivery.4 For example, staffing shortages in the revenue cycle can cause more denials, increased days in accounts receivable, and decreased productivity, all of which can negatively impact an organization’s operational and financial performance.5 In addition, there is a link between staffing shortages, burnout, and lack of trust in leadership.6
In our previous study, we outlined a path forward for addressing the shortages of frontline clinical workers that involves three components: reimagining care delivery and redesigning the work, investing in people, and restoring trust in organizational leadership. In this study, we discuss how health care organizations should redesign work for their entire workforce, using technological approaches. We discuss how such approaches can be used in any part of the organization, front or back, to address a big reason behind worker shortages: the decline of meaning, purpose, and joy in work.7 By redesigning work this way, health care organizations can gain efficiencies, lift profitability, and improve the experience of those involved in or touched by the work: consumers and patients who seek and receive care, clinicians who provide care, and nonclinical staff who support that care.
For example, we estimate that it is possible to free up to 50% of time for revenue cycle professionals and up to 20% for bedside nurses with the technologies available today. This extra time can empower revenue cycle professionals to take on more strategic and relationship-building responsibilities, while nurses can focus more on providing direct patient care. Whether organizations should start work redesign in the front office, back office, or both depends on their needs and strategic priorities. To help leaders with such decisions, we propose a strategic framework that involves three stages (more detail in figure 3 later in the article):
In some instances, the “stabilize” stage may not require major technological investments, but that is unlikely to be the case in the other stages.
The Deloitte Center for Health Solutions developed a model to estimate time savings achievable from the application of technologies available today such as automation, digital tools, artificial intelligence, and generative AI, so workers can do more fulfilling and valuable work, operate at the top of their skill set, and have new options for professional development and career growth. We illustrate these savings opportunities with two examples: one in the back office (revenue cycle) and one in the front office (bedside nursing). (For more details on our research approach, see the appendix at the end of this article.)
Revenue cycle, a critical organizational function for capturing revenue for health systems, has high worker turnover and high recruitment costs.8 In a recent survey, 90% of health care finance executives reported staffing shortages in their revenue cycle department.9 In another survey of financial leaders, the time to fill revenue cycle roles ranged from 84 days for entry-level positions to 207 days for senior roles.10
Many aspects of the revenue cycle—such as scheduling and registration, cost estimates, medical necessity, coding, billing, and collections—involve direct touch points with patients and providers, and almost all aspects of the work have a direct impact on consumer or provider experience. Yet revenue cycle staff often feel disconnected from the impact of their jobs because they are bogged down by administrative processes: hunting for information, organizing it, communicating it, and waiting for a response from one party (like provider) before contacting another party (like health plan).11 Streamlining these processes can improve the overall efficiency and effectiveness of the revenue cycle function.
For the nursing example in our estimation model, we focused on nurses working in general medical-surgical units who form a large share of clinical workers within health systems.12 Nurses play a special role at the center of human experience; they coordinate and drive patient care as well as the patients’ and their families’ perception of the organization and the overall system of care. Unfortunately, many nurses feel overworked and burned out.13
It is well-established that clinician burnout leads to low patient experience and poor outcomes.14 Moreover, there is a link between burnout and lack of trust in leadership, as we showed in our previous study, and the erosion of trust is alarming: Only 45% of frontline clinicians trust their organization’s leadership to do what’s right for the patients, and even fewer, 23%, trust them to do what’s right for the workers.15 Several studies have also shown that administrative burden significantly contributes to burnout.16 An estimated 15% to 28% of nurses’ work comprises low-value tasks, mostly to satisfy administrative requirements.17 Eliminating low-value work, reducing administrative burden, and enabling nurses to spend more time on direct patient care and valuable interactions with patients, families, and coworkers can begin to rebuild workers’ trust and should be a high priority for health care leaders.18
Our analysis reveals time-saving opportunities in both the revenue cycle function and nursing. When prioritizing jobs or processes for redesign, opportunities in the back office often fit the criteria for quick wins due to the low financial investment, reasonably quick return on investment, and simplicity of redesign solutions. (Refer to the section “What are quick wins?” and figure 3 for more information.)
We estimate that, on average, technologies can save 41% to 50% of revenue cycle professionals’ time (figure 1) across all three stages of the revenue cycle: patient access, or the “front end”; the clinical revenue cycle, or the “middle cycle”; and patient financial services, or the “back end.”
Revenue cycle workflow processes can be modernized with mature technologies because these processes are repeatable, frequent, and possible to standardize. Even when standardization is not practical, combining automation with AI allows for far more flexibility than robotic process automation.
In a recent survey, 74% of hospital chief financial officers and revenue cycle leaders said they already use automation or are in the process of implementing it in their revenue cycle operations.19 Successful implementation of these technologies, though, requires data modernization and interoperability.
In the front end (or patient access), revenue cycle staff spend a significant amount of time on scheduling and pre-registration. We estimate technologies can save 38% to 47% of time, or about 700 to 870 hours a year per scheduler. Despite the growing use of online scheduling tools, appointment scheduling often falls short of patients’ expectations. In a 2022 survey, 70% of patients had to finalize their appointment booking over the phone, even after using an online scheduling tool.20 As patients transition from hospital care into the ambulatory setting, or if they require multiple types of ambulatory services, there are friction points in the process. Even within the same organization, there may not be a common scheduling platform for ambulatory and other services. This means patients deal with multiple schedulers, wait on hold, repeat information, and have their calls disconnected. Implementing AI in this process can offer several benefits, including streamlining and standardizing scheduling tasks, confirming appointments, integrating appointments into patients’ digital calendars, providing liability estimates, facilitating payments, conducting real-time eligibility checks, and combing medical policy for authorization and medical necessity requirements.
In the middle revenue cycle, we see significant time-saving opportunities (61% to 70%) in health information management and coding. This work domain includes medical records management, such as collecting and digitizing documents related to patient care; facilitating access, retrieval, and sharing of patient information; as well as dictation and transcription. AI-driven workflow modifications could include real-time prompts to providers with the most appropriate current procedural terminology code based on electronic health record (EHR) data or on ambient clinical documentation. This would ensure accurate coding of a patient’s severity of illness and acute clinical manifestations and sufficient specificity in the clinical documentation to justify the use of the code. Consequently, this could decrease coding errors and reduce the back and forth between providers and coders.
Similar opportunities for automation exist in the back-end revenue cycle, or patient financial services. Today, claims processing is one of the most time-consuming revenue cycle activities, prone to errors and rework. We estimate that 44% to 53% of time, or about 810 to 980 hours, of an average claims-processing professional can be saved annually through the implementation of technologies that include automation and generative AI. For instance, when an insurance claim is denied, claims-processing professionals face the tedious task of digging up the applicable medical policy, reviewing the patient’s record, and matching the policy requirements to the information in the record. Today, although more than 60% of denied claims are recoverable, only a fraction are appealed due to vague denial reasons and a time-intensive verification process.21 A generative AI retrieval model can comb through volumes of medical policies to locate the relevant information for a claims appeal, simplifying this task for the claims-processing professional. The ability to automate aspects of claims authorization and appeals could lead to significant time and cost savings and improved revenue capture.22
Revenue cycle systems could also be a source of operational and financial insight.23 For instance, applying AI to data from multiple systems (such as individual patients’ medical records, payer-specific patterns of processing claims, and patient journey patterns from aggregated longitudinal EHR data) could help predict whether a patient account will be complex to resolve. This could inform staff allocation and flagging for the accounts receivable teams. The evolution of the revenue cycle organization to one based on insights could open up new career options for revenue cycle professionals.
For organizations experiencing revenue cycle staffing shortages, work redesign may not necessitate head count reduction. Gradual implementation of work redesign along with change management could allow for a smooth transition where existing staff can take on more strategic and relationship-building responsibilities.
Technology can be a valuable tool in streamlining revenue cycle processes, as the following examples illustrate:
We estimate that technology can free up between 13% and 21% of nurses’ time, which translates to 240 to 400 hours of a single staff nurse’s time in a year (figure 2). Even modest time savings can be a game changer for nurses facing long hours and heavy workloads. This can allow nurses to have more time for meeting patient needs and coordinating with care team members, enhancing the dynamic between the care team, patients, and families and leading to greater patient experience and loyalty. These time savings could be even greater when combined with nontech solutions like team restructuring and redesign of the workflow and physical environment.
The greatest opportunity by far is in streamlining documentation activities, around 95 to 134 hours in a year for a nurse (figure 2). Other areas where savings could be realized include patient assessment, medication-related activities, administrative work, and care planning and coordination.
Multiple studies have pointed to documentation as an area of need and opportunity.27 In the Deloitte 2022 Survey of US Frontline Clinicians, when nurses were asked to name one thing they do in their job that has minimal clinical value and could be eliminated, the No. 1 response was documentation for administrative purposes (17%), followed by charting (12%). Today, nurses often feel they spend more time in front of computers than with patients.28
Within documentation, the largest opportunity according to our model is around composing nursing progress notes, which constitute the largest share of time spent on documentation. Different technological approaches can alleviate different types of frictions in this activity.
From 2019 to 2020, a large academic health system redesigned the assessment section of nursing documentation in EHRs. As a result, the time spent in nursing documentation decreased by 30 minutes per 12-hour shift, and the time savings continued post implementation, resulting in reduced stress associated with the EHR, improved efficiencies, and decreased duplicative documentation. Nurses in the participating departments said they were pleased that the process change allowed them to devote more time to direct patient care.29 Changes like this can begin to restore the humanity of care that many are concerned is lost in our system of care.30 Clinicians can spend more time caring for patients, and patients can feel cared for.
While this demonstrates EHR optimization for improving documentation, other possibilities include:
Another opportunity is administrative work. According to our analysis, while administrative processes take approximately 5% of nurses’ time, infusing technology in these processes can save about half of this time. Examples include digitized scheduling and reminders to facilitate admission, transfer, or discharge of patients; auto-population of consents and other forms for procedures; and digital facilitation of advance directives.
Under care planning and coordination, tele-nursing31 or even chatbots32 can be used to provide and reinforce patient education during discharge, including confirming a patient’s understanding of what they should expect, what is concerning versus normal in their healing process, when to call 911, or when to follow up with their care team. In the patient assessment domain, technology could help piece together information from multiple sources (such as patient history, medical record, and vitals) and suggest symptoms or vitals to pay special attention to when assessing a patient for conditions that could result in rapid deterioration.
The domain of patient care is what we traditionally think of as nursing work: for example, managing patients’ pain relief, monitoring their responses to treatment, obtaining samples for laboratory work, keeping patients comfortable, or caring for wounds. Almost none of these activities are good candidates for redesigning with technology. This is because they involve the use of most senses (touch, smell, sound); the assessment of patients’ and families’ mental, emotional, and social well-being; or because technology could reduce or interfere with valuable nurse-patient interactions (for example, patient-care activities like learning more about the patient as a person or removing their glasses and hearing aid before they go to sleep).
However, one area where technological solutions could come into play is to help nurses stay in the flow, reducing workflow fragmentation, distractions, and the cognitive burden of switching between tasks.
Tele-nursing (also known as virtual nursing) can ease the burden on bedside nurses across a range of activities, as the following examples illustrate.
We expect the biggest challenge for leaders in thinking about redesigning work with technology is deciding where to start, how to define and measure success, defining and measuring the ROI, and striking the balance between the business and the humanity of care.
Another consideration is to understand that while technology can help and augment health care workers in their jobs, it can also lead to distrust and anxiety.35 Having a transparent and honest two-way conversation with workers about difficult choices and trade-offs needed to redesign work will be an essential step for organizations to maintain trust. Stabilizing the workforce, redesigning work, and using technology effectively to give workers time to do more engaging and meaningful work will require creativity, innovation, and leadership.
To equip leaders to navigate these complex conversations, we propose a strategic framework (figure 3) that involves three stages:
As leaders consider implementing technological solutions during any of these stages, it will be important for them to assess whether to build capabilities in house, buy relevant technologies, or enter strategic partnerships to gain access to these capabilities. These decisions could be informed by the assessment of current resources and capabilities, and the speed at which organizations are prepared to move.
Quick wins are projects or initiatives that require low financial investment and produce savings or new revenue in a short period of time, usually within six to 18 months.
The criteria for quick wins are a combination of several factors:
Across different hospital functions, we have found some workflow processes are consistently good candidates for modernizing with technology, and others can be improved with non-technological approaches. Often, these opportunities are at the intersection of processes across different functions, such as supply chain, different types of scheduling (surgeon blocks, nurse staffing), or documentation and document management systems.
For most health care organizations today, the first order of business is to stabilize the workforce and operations now and position for growth and sustainability. Redesigning work in the long term will involve a comprehensive approach that includes redesigning teams, work processes, and the workplace. While comprehensive redesign is critical, strategically implementing relevant technologies now can significantly accelerate progress toward a sustainable future.
The Deloitte Center for Health Solutions developed an estimation model to illustrate what time savings can be achieved over the next one to three years with the technologies available today, such as automation, AI, generative AI, telehealth, and a combination of these technologies. We modeled these savings for two jobs: revenue cycle and bedside nursing.
To inform model inputs, we drew upon secondary data available publicly and by subscription, including the Occupational Information Network by the US Department of Labor and a database that pools operational data from hospital organizations that are members of one of the largest group purchasing organizations in the country (which we refer to as “hospital database” in this section).
Additionally, we employed the Delphi method36 to review and establish consensus on model inputs, assumptions, and outputs. Delphi panels for each role were composed of functional experts with recent professional or consulting experience in the role, as well as technical experts knowledgeable about the workflow and technology that could be used to optimize that workflow. This included individuals with expertise in clinical informatics, robotic process automation, machine learning, and AI.
The key stages are listed below:
In our estimation model, we chose to illustrate opportunities within two core hospital functions: revenue cycle and nursing. We chose these functions to demonstrate the potential impact of technology on both clinical and nonclinical work. The modeling assumed the conditions and workforce distribution at an average community hospital in the United States.
To estimate the time savings from technology for both revenue cycle and nursing, we took a four-step approach:
For the revenue cycle, our level of analysis was the revenue cycle department. Based on literature review, and with help from the Delphi panel of Deloitte revenue cycle specialists, we created a detailed workflow of each of the three major revenue cycle stages: patient access, the middle revenue cycle, and patient financial services. We then used operational data from the hospital database on full-time employee (FTE) hours worked in the revenue cycle function to estimate the share of time spent on individual workflow activities for an average community hospital.
For nursing, our level of analysis was an individual nurse. We referred to the Occupational Information Network by the US Department of Labor to obtain a list of job tasks, including frequency, for acute care nurses.
For both functions, we used the Delphi method to review and achieve consensus around model inputs and assumptions. We had one Delphi panel of revenue cycle experts and two separate Delphi panels of functional and technical nursing experts.
For each function, our functional Delphi panelists evaluated each task or activity to estimate what share of time spent today on this task could be saved by employing one or more technologies. Delphi panelists were specifically instructed to think about technologies that are mature enough to be implemented in the next one to three years.
Based on literature review, we assumed that an average nurse works a total time of 48 weeks (around 1,920 hours) in a year.37 We used FTE benchmarks from the hospital database to estimate that an average revenue cycle professional works a total of around 1,850 hours in a year.
We averaged the estimates from the Delphi panelists to arrive at the total time savings for each role or function. We then expressed these time savings as percentages of total worker time as well as in FTE hours per year.
To assess the sensitivity of our results to extreme assumptions, we calculated three sets of estimates: conservative, normal, and liberal. Normal estimates are the average of all panelists’ inputs. For conservative and liberal estimates, we excluded outliers on the high end and low end, respectively. When the variance was 20 points or greater between an individual panelist’s estimate and the average (mean) across all panelists, we treated those individual estimates as outliers. For the revenue cycle function, conservative and liberal range estimates differed from the normal estimate by eight to 11 percentage points. For the nursing function, conservative and liberal range estimates differed from the normal estimate by one to three percentage points.
We recognize that our findings are illustrative and provide directional insights.
The model assumes an existing process and suggests how each element in this process can be improved with technology. The model does not assume or suggest how the process itself can be optimized. For this reason, the real savings could be greater as a result of process redesign. On the other hand, applying technology to every single element of the process does not make practical or financial sense, and for this reason the model overstates the savings. On balance, though, these two limitations should cancel each other out.
We also recognize that hospitals organize themselves differently and have different levels of comfort and experience with technology, which can influence the extent to which they can leverage its benefits.
Given the multiple ways to approach and solve a problem, our panelists relied on their domain knowledge, leading to variability in their responses, which may have impacted our estimates.
Our analysis did not examine the impact on other jobs or adjacent workflows. Our modeling assumes the entire workflow stays the same. In reality though, for many jobs, departments, functions, and entire organizations, work redesign would result in entirely new processes. This would eliminate certain tasks altogether and create new tasks and jobs.
Our modeling does not account for variability between organizations due to different processes and patient mix, as well as variability within an organization due to a lack of process standardization and different patient complexity.
Lastly, financial and operational indicators for hospitals and health systems in the hospital database are based on organizations that are members of a large group purchasing organization. As such, they may not be representative of hospital organizations that are not members.