Hospitalists who teach in the clinical environment face challenges that include increased workload,[1] perception among trainees that there is less time to teach,[2] and competition with electronic devices for teaching engagement.[3, 4] In view of these and other challenges, we believe there is potentially much to gain from considering and adapting educational techniques that have been successful in nonhospital and even nonmedical domains. Innovative teaching methods include those designed for the grade‐school classroom (Courage to Teach,[5] Teaching With Love and Logic[6]), and the business world (Teaching Smart People How to Learn,[7] The Back of the Napkin[8]), among other nonmedical professions. Within medicine, we can also re‐examine strategies long utilized in the ambulatory setting. Pascoe and colleagues offer an important example of this in their review of one‐minute preceptor (OMP) and SNAPPS, techniques developed by our colleagues in the outpatient setting but with great potential for framing discussion of clinical reasoning in the inpatient space.[9]

Applying OMP and SNAPPS to inpatient teaching presents some challenges but also genuine opportunities not found in traditional outpatient teaching. As noted by the authors, unlike the solitary learner typical of the outpatient setting, in the inpatient setting the attending is more commonly working with a group of learners of multiple levels and sometimes multiple disciplines. Furthermore, the supervising resident typical of inpatient teams is a learner who inhabits the roles of both trainee and teacher. One can imagine that if OMP and SNAPPs are applied with absolute fidelity to the inpatient setting, without reflection on venue, the teaching encounter might be overly focused on the presenting learner, leaving the rest of the team unattended to, disengaged, and not benefitting from the models. Therefore, attention to group engagement in the process is necessary for successful adaptation. Both models have the potential to help organize the group dynamic during rounds to promote broad participation. The authors describe some examples of how to engage various group members in different steps. It is worth highlighting a few key themes that enable successful use of these models in the inpatient setting.

One key theme is to teach the model to the supervising resident at the beginning of the rotation and agree, before rounds, how the attending and resident will interact as coleaders of the discussion. Because these models offer a stepwise approach to going through a case with a learner, they have the potential to demystify the teaching process, offering an accessible framework for supervising residents to learn teaching both by practicing and by comprehending what their attending is doing to lead a team through a case discussion. With attending support, the supervising resident can be encouraged to manage the team discussion, leading the team using either approach. It can be helpful to touch base briefly before rounds each day to define the teaching roles, giving the resident progressively more responsibility leading the discussion as the rotation progresses.

Another key theme is to use graduated participation. As the authors note, the group must be engaged in the discussion, and the example scenarios illustrate each step of the models being applied to the group. To ensure that the entire group remains eager to partake, the leader must maintain a nonthreatening teaching atmosphere, organizing participation in a way that does not shame learners or undermine the roles people inhabit. To this end, it can be helpful to direct questions to particular members or levels of the group at a time. When expanding participation around a specific question or concept, always work from junior members to senior members, never imposing the reverse. This principle is clearly not exclusive to using these models, but is requisite to successful adaptation of these traditionally dyadic models, in which there is no particular attention to group dynamics within the framework.

A third key theme is to utilize the unique expertise of the other health professionals on the team in steps 4, 5, and 6 of SNAPPS and step 3 of OMP. In step 4 and 5 of SNAPPS, when the teaching attending introduces the team to the model, it is important to encourage them to probe not just the teacher but other disciplines on the team for input. In the inpatient setting, these steps provide an organized point in the discussion in which to involve the other members of the professional team, modeling collaborative interdisciplinary practice.

As Pascoe et al. point out, there are limited studies of OMP and SNAPPS as teaching models in the inpatient environment. This should stimulate academic hospitalists with interest in medical education research to consider how these models might be studied. For example, in comparison to traditional inpatient teaching rounds, do these approaches provide equivalent content coverage? How do they impact the efficiency of teaching rounds? Are attendings who consistently apply these models more effective in providing feedback or assessing training milestones? How much training and practice is required to incorporate these teaching models in the inpatient environment?

Given the time pressure and increasing complexity of medical care in the hospital, coupled with the evolving needs and resources of our learners, we must seek innovative educational practices from sources outside our hospitals to provide the best possible training in hospital medicine. An outstanding recent review by Martin et al. provided an overview of other strategies for teaching in today's environment.[10] We also have much to learn from our colleagues in outpatient medicine, not only in clinical care, but also in medical education. And we have much that we have learned about teaching as hospitalists that needs to be more broadly disseminated.

Hospitalists who teach in the clinical environment face challenges that include increased workload,[1] perception among trainees that there is less time to teach,[2] and competition with electronic devices for teaching engagement.[3, 4] In view of these and other challenges, we believe there is potentially much to gain from considering and adapting educational techniques that have been successful in nonhospital and even nonmedical domains. Innovative teaching methods include those designed for the grade‐school classroom (Courage to Teach,[5] Teaching With Love and Logic[6]), and the business world (Teaching Smart People How to Learn,[7] The Back of the Napkin[8]), among other nonmedical professions. Within medicine, we can also re‐examine strategies long utilized in the ambulatory setting. Pascoe and colleagues offer an important example of this in their review of one‐minute preceptor (OMP) and SNAPPS, techniques developed by our colleagues in the outpatient setting but with great potential for framing discussion of clinical reasoning in the inpatient space.[9]

Applying OMP and SNAPPS to inpatient teaching presents some challenges but also genuine opportunities not found in traditional outpatient teaching. As noted by the authors, unlike the solitary learner typical of the outpatient setting, in the inpatient setting the attending is more commonly working with a group of learners of multiple levels and sometimes multiple disciplines. Furthermore, the supervising resident typical of inpatient teams is a learner who inhabits the roles of both trainee and teacher. One can imagine that if OMP and SNAPPs are applied with absolute fidelity to the inpatient setting, without reflection on venue, the teaching encounter might be overly focused on the presenting learner, leaving the rest of the team unattended to, disengaged, and not benefitting from the models. Therefore, attention to group engagement in the process is necessary for successful adaptation. Both models have the potential to help organize the group dynamic during rounds to promote broad participation. The authors describe some examples of how to engage various group members in different steps. It is worth highlighting a few key themes that enable successful use of these models in the inpatient setting.

One key theme is to teach the model to the supervising resident at the beginning of the rotation and agree, before rounds, how the attending and resident will interact as coleaders of the discussion. Because these models offer a stepwise approach to going through a case with a learner, they have the potential to demystify the teaching process, offering an accessible framework for supervising residents to learn teaching both by practicing and by comprehending what their attending is doing to lead a team through a case discussion. With attending support, the supervising resident can be encouraged to manage the team discussion, leading the team using either approach. It can be helpful to touch base briefly before rounds each day to define the teaching roles, giving the resident progressively more responsibility leading the discussion as the rotation progresses.

Another key theme is to use graduated participation. As the authors note, the group must be engaged in the discussion, and the example scenarios illustrate each step of the models being applied to the group. To ensure that the entire group remains eager to partake, the leader must maintain a nonthreatening teaching atmosphere, organizing participation in a way that does not shame learners or undermine the roles people inhabit. To this end, it can be helpful to direct questions to particular members or levels of the group at a time. When expanding participation around a specific question or concept, always work from junior members to senior members, never imposing the reverse. This principle is clearly not exclusive to using these models, but is requisite to successful adaptation of these traditionally dyadic models, in which there is no particular attention to group dynamics within the framework.

A third key theme is to utilize the unique expertise of the other health professionals on the team in steps 4, 5, and 6 of SNAPPS and step 3 of OMP. In step 4 and 5 of SNAPPS, when the teaching attending introduces the team to the model, it is important to encourage them to probe not just the teacher but other disciplines on the team for input. In the inpatient setting, these steps provide an organized point in the discussion in which to involve the other members of the professional team, modeling collaborative interdisciplinary practice.

As Pascoe et al. point out, there are limited studies of OMP and SNAPPS as teaching models in the inpatient environment. This should stimulate academic hospitalists with interest in medical education research to consider how these models might be studied. For example, in comparison to traditional inpatient teaching rounds, do these approaches provide equivalent content coverage? How do they impact the efficiency of teaching rounds? Are attendings who consistently apply these models more effective in providing feedback or assessing training milestones? How much training and practice is required to incorporate these teaching models in the inpatient environment?

Given the time pressure and increasing complexity of medical care in the hospital, coupled with the evolving needs and resources of our learners, we must seek innovative educational practices from sources outside our hospitals to provide the best possible training in hospital medicine. An outstanding recent review by Martin et al. provided an overview of other strategies for teaching in today's environment.[10] We also have much to learn from our colleagues in outpatient medicine, not only in clinical care, but also in medical education. And we have much that we have learned about teaching as hospitalists that needs to be more broadly disseminated.

Hospitalists who teach in the clinical environment face challenges that include increased workload,[1] perception among trainees that there is less time to teach,[2] and competition with electronic devices for teaching engagement.[3, 4] In view of these and other challenges, we believe there is potentially much to gain from considering and adapting educational techniques that have been successful in nonhospital and even nonmedical domains. Innovative teaching methods include those designed for the grade‐school classroom (Courage to Teach,[5] Teaching With Love and Logic[6]), and the business world (Teaching Smart People How to Learn,[7] The Back of the Napkin[8]), among other nonmedical professions. Within medicine, we can also re‐examine strategies long utilized in the ambulatory setting. Pascoe and colleagues offer an important example of this in their review of one‐minute preceptor (OMP) and SNAPPS, techniques developed by our colleagues in the outpatient setting but with great potential for framing discussion of clinical reasoning in the inpatient space.[9]

Applying OMP and SNAPPS to inpatient teaching presents some challenges but also genuine opportunities not found in traditional outpatient teaching. As noted by the authors, unlike the solitary learner typical of the outpatient setting, in the inpatient setting the attending is more commonly working with a group of learners of multiple levels and sometimes multiple disciplines. Furthermore, the supervising resident typical of inpatient teams is a learner who inhabits the roles of both trainee and teacher. One can imagine that if OMP and SNAPPs are applied with absolute fidelity to the inpatient setting, without reflection on venue, the teaching encounter might be overly focused on the presenting learner, leaving the rest of the team unattended to, disengaged, and not benefitting from the models. Therefore, attention to group engagement in the process is necessary for successful adaptation. Both models have the potential to help organize the group dynamic during rounds to promote broad participation. The authors describe some examples of how to engage various group members in different steps. It is worth highlighting a few key themes that enable successful use of these models in the inpatient setting.

One key theme is to teach the model to the supervising resident at the beginning of the rotation and agree, before rounds, how the attending and resident will interact as coleaders of the discussion. Because these models offer a stepwise approach to going through a case with a learner, they have the potential to demystify the teaching process, offering an accessible framework for supervising residents to learn teaching both by practicing and by comprehending what their attending is doing to lead a team through a case discussion. With attending support, the supervising resident can be encouraged to manage the team discussion, leading the team using either approach. It can be helpful to touch base briefly before rounds each day to define the teaching roles, giving the resident progressively more responsibility leading the discussion as the rotation progresses.

Another key theme is to use graduated participation. As the authors note, the group must be engaged in the discussion, and the example scenarios illustrate each step of the models being applied to the group. To ensure that the entire group remains eager to partake, the leader must maintain a nonthreatening teaching atmosphere, organizing participation in a way that does not shame learners or undermine the roles people inhabit. To this end, it can be helpful to direct questions to particular members or levels of the group at a time. When expanding participation around a specific question or concept, always work from junior members to senior members, never imposing the reverse. This principle is clearly not exclusive to using these models, but is requisite to successful adaptation of these traditionally dyadic models, in which there is no particular attention to group dynamics within the framework.

A third key theme is to utilize the unique expertise of the other health professionals on the team in steps 4, 5, and 6 of SNAPPS and step 3 of OMP. In step 4 and 5 of SNAPPS, when the teaching attending introduces the team to the model, it is important to encourage them to probe not just the teacher but other disciplines on the team for input. In the inpatient setting, these steps provide an organized point in the discussion in which to involve the other members of the professional team, modeling collaborative interdisciplinary practice.

As Pascoe et al. point out, there are limited studies of OMP and SNAPPS as teaching models in the inpatient environment. This should stimulate academic hospitalists with interest in medical education research to consider how these models might be studied. For example, in comparison to traditional inpatient teaching rounds, do these approaches provide equivalent content coverage? How do they impact the efficiency of teaching rounds? Are attendings who consistently apply these models more effective in providing feedback or assessing training milestones? How much training and practice is required to incorporate these teaching models in the inpatient environment?

Given the time pressure and increasing complexity of medical care in the hospital, coupled with the evolving needs and resources of our learners, we must seek innovative educational practices from sources outside our hospitals to provide the best possible training in hospital medicine. An outstanding recent review by Martin et al. provided an overview of other strategies for teaching in today's environment.[10] We also have much to learn from our colleagues in outpatient medicine, not only in clinical care, but also in medical education. And we have much that we have learned about teaching as hospitalists that needs to be more broadly disseminated.

Although primary care physicians (PCPs) have traditionally treated patients in both ambulatory and hospital settings, many relinquished inpatient duties to hospitalists in recent decades.[1] Little is known about the PCPs who relinquished inpatient care duties or how the transition to the hospitalist model occurred. For example, what are the characteristics of PCPs who change? Do PCPs adopt the hospitalist model enthusiastically or cautiously? Characterizing PCPs who adopted the hospitalist model can help hospitalists understand their specialty's history and also inform health services research.

Much of the interest in the hospitalist model has been generated by studies reporting improved outcomes and lower hospital lengths of stay associated with hospitalist care.[2, 3, 4, 5] Conversely, detractors of the model point to reports of higher postacute care utilization among hospitalist patients.[6] Although these studies usually adjusted for differences among patients and hospitals, they did not account for PCP characteristics. As patients' access to PCPs and their PCP's capabilities are both plausible factors that could influence hospital length of stay (eg, decisions to complete more or less of a workup in the hospital), quality of care transitions, and postdischarge utilization, it is important to determine if PCPs who use hospitalists differ systematically from those who do not to correctly interpret health system utilization patterns that currently are attributed only to hospitalists.[7, 8]

We conducted this study to determine if observable PCP factors are associated with patients' use of hospitalists and to describe the trajectory by which PCPs referred their patients to hospitalists over time.

METHODS

RESULTS

During the 2001 through 2009 study period, between 2252 and 2848 PCPs were associated with at least 20 hospitalized beneficiaries in any single year. Among these, 1172 PCPs were associated with at least 20 hospitalized beneficiaries in every year of the study period. These 1172 PCPs were associated with 608,686 hospitalizations over the 9 years.

Table 1 presents the characteristics of the PCPs who contributed to the cross‐sectional analyses in 2001 (N=2252) and 2009 (N=2387), as well as the 1172 PCPs for whom we had data for all 9 years for the longitudinal analyses. Most PCPs were male, trained in the United States, and were board certified. The average number of Medicare patients seen by these PCPs and number of outpatient Medicare visits went up about 7% between 2001 and 2009.

Figure 1 graphs the percentage of PCPs as a function of what percent of their hospitalized patients received care from hospitalists, and how that changed from 2001 to 2009. For 70.9% of PCPs, fewer than 5% of their hospitalized patients received hospitalist care in 2001. By 2009, the percent of PCPs in this category had decreased to 15.2%. In contrast, in 2001, more than half of the patients for 2.1% of PCPs received hospitalist care, and the percent of PCPs in this category increased to 26.3% by 2009.

Figure 1

The pattern in Figure 1 shows that PCPs' use of hospitalists changed continuously and gradually over time. However, this pattern describes the PCPs as a group. When examined at the individual PCP level, different patterns emerge. Figure 2, which presents selected individual PCP's use of hospitalists over time, shows several distinct subpatterns of PCP practice behaviors. First, there are PCPs whose use of hospitalists was high in 2001 and stayed high or increased over time (eg, PCP A). There also were PCPs whose use of hospitalists stayed low over the entire study period (eg, PCP B). Finally, there were PCPs whose use of hospitalists was low in 2001 but high in 2009 (eg, PCP C). For this last group, the pattern of change in hospitalist utilization over time was discontinuous; that is, most of the increase occurred over a 1‐ or 2‐year period, instead of increasing gradually over time.

Figure 2

Among the 1172 PCPs associated with 20 hospitalized beneficiaries each year in all 9 years of the study period, group‐based trajectory modeling classified their practice patterns into 4 distinct trajectories (Figure 3). Among PCPs in group 1, more than one‐third of their hospitalized patients were cared for by hospitalists in 2001, and this increased to 60% by 2009. PCPs in groups 2 and 3 rarely used hospitalist care in 2001 but increased their use over time. The increase started early in the period for PCPs in group 2 and later for those in group 3. PCPs in group 4 were associated with little hospitalist use throughout the study period.

Figure 3

We constructed a model to describe the odds of a patient receiving care from hospitalists during the study period using patients associated with these 1172 PCPs. After adjusting for patient characteristics, the residual intraclass correlation coefficient for PCP level was 0.334, which indicates that 33.4% of the variance in whether a hospitalized patient received care from a hospitalist is explained by which PCP the patient saw. When adjusting for both patient and PCP characteristics, the overall odds of a patient receiving hospitalist care increased by 30% (95% confidence interval [CI]: 1.29‐1.30) per year from 2001 through 2009.

There were also significant interactions between year of hospitalization and several PCP characteristics. These interactions are illustrated in Table 2, which stratifies each of those PCP characteristics by 3 time periods: 2001 to 2003, 2004 to 2006, and 2007 to 2009. In all time periods, patients were more likely to receive hospitalist care if their PCP was US trained (US vs international medical graduate: odds ratio [OR]: 1.42, 95% CI: 1.19‐1.69 in 20012003; OR: 1.46, 95% CI: 1.23‐1.73 in 20072009), or specialized in family medicine (family medicine vs internal medicine: OR: 1.46, 95% CI: 1.25‐1.72 in 20012003; OR: 1.46, 95% CI: 1.25‐1.70 in 20072009). Over time, the relative odds of a patient receiving care from hospitalists decreased if their PCP was female (female vs male: OR: 1.91, 95% CI: 1.46‐2.50 in 20012003 vs OR: 1.50, 95% CI: 1.15‐1.95 in 20072009) or practiced in an urban area (largest vs smallest MSA: OR: 3.34, 95% CI: 2.72‐4.09 in 20012003; OR: 2.22, 95% CI: 1.82‐2.71 in 20072009). Although the longest‐practicing PCPs were most likely to use hospitalists in the early 2000s, this effect disappeared by 2007 to 2009 (most vs least years in practice: OR: 1.35, 95% CI: 1.06‐1.72 in 20012003 vs OR: 0.92, 95% CI: 0.73‐1.17 in 20072009).

In terms of PCP workload, patients of PCPs with high outpatient activity were more likely to receive hospitalists care throughout the study period, although the association had decreased by 2007 to 2009 (highest vs lowest outpatient volume: OR: 1.46, 95% CI: 1.30‐1.63 in 20012003 vs OR: 1.32, 95% CI: 1.20‐1.44 in 20072009). In contrast, PCPs with the lowest inpatient volumes became more likely to use hospitalists by the end of the study period (highest vs lowest inpatient volume: OR: 1.05, 95% CI: 0.95‐1.18 in 20012003 vs OR: 0.55, 95% CI: 0.51‐0.60 in 20072009).

The characteristics of PCPs' practice panels also were associated with patients' likelihood of receiving care from hospitalists. PCPs whose practice panels consisted of patients who were predominantly male, white, or with more outpatient comorbidities were consistently more likely to use hospitalists throughout the study period. PCPs with older patient panels were less likely to use hospitalists in 2001 to 2003, but by 2007 to 2009, they were slightly more likely to do so (oldest vs youngest average outpatient panel age: OR: 0.72, 95% CI: 0.64‐0.81 in 20012003 vs OR: 1.15, 95% CI: 1.05‐1.26 in 20072009).

CONCLUSIONS

Prior studies of the hospitalist model have shown that the likelihood of a patient receiving inpatient care from hospitalists is associated with patient characteristics, hospital characteristics, geographic region, and type of admission.[1, 16, 17] We found that PCP characteristics also predict whether patients receive care from hospitalists and that their use of hospitalists developed dynamically between 2001 to 2009. Although many factors (such as whether patients were admitted to a hospital where their PCP had admitting privileges) can influence the decision to use hospitalists, we found that over one‐third of the variance in whether a hospitalized patient received care from a hospitalist is explained by which PCP the patient saw. In showing that systemic differences exist among PCPs who use hospitalists and those who do not, our study suggests that future research on the hospitalist model should, if possible, adjust for PCP characteristics in addition to hospital and patient factors.

Although this study identifies the existence and magnitude of differences in whether or not PCPs use hospitalists, it cannot explain why the differences exist. We only can offer hypotheses. For example, our finding that PCPs with the most years of practice experience were more likely to use hospitalists in the early 2000s but not in more recent years suggests that in hospital medicine's early years, long‐practicing generalist physicians were choosing between practicing traditionalist medicine and adopting the hospitalists model, but by 2009, experienced generalist physicians had already specialized to either inpatient or outpatient settings earlier in their careers. On the other hand, the decreasing odds of urban PCPs using hospitalists may reflect a relative growth in hospitalist use in less populated areas rather than a change in urban PCPs' practice patterns.

PCPs trained in family medicine have reported less inpatient training and less comfort with providing hospital care,[18, 19] thus it is unsurprising that family physicians were more likely to refer patients to hospitalists. Although a recent study reported that family physicians' inpatient volumes remained constant, whereas those of outpatient internists declined between 2003 and 2012, the analysis used University Health Consortium data and thus reflects practice patterns in academic medical centers.[20] Our data suggest that outside of academia, family physicians have embraced the hospitalists as clinical partners.

Meltzer and Chung had previously proposed an economic model to describe the growing use of hospitalists in the United States. They posited that decisions to adopt the hospitalist model are governed by trade‐offs between coordination costs (eg, time and effort spent coordinating multiple providers across different settings) and switching costs (eg, time spent traveling between the office and the hospital or the effort of adjusting to different work settings).[16] The authors hypothesized that empirical testing of this model would show PCPs are more likely to use hospitalists if they have less available professional time (ie, work fewer hours per week), are female (due to competing demands from domestic responsibilities), have relatively few hospitalized patients, or live in areas with high traffic congestion. Our findings provide empirical evidence to support their division‐of‐labor model in showing that patients were more likely to receive hospitalist care if their PCP was female, practiced in an urban location, had higher outpatient practice volumes, or had lower inpatient volumes.

At first glance, some of our findings appear to contradict our earlier study, which showed that younger, black, male patients are more likely to receive inpatient care from hospitalists.[1] However, that study included patients regardless of whether they had a PCP. This study shows that when patients have a PCP, their PCPs are more likely to refer them to hospitalists if they are older, white, male, and have more comorbid conditions. A potential explanation for this finding is that PCPs may preferentially use hospitalists when caring for older and sicker hospitalized patients. For example, commentators often cite hospitalists' constant availability in the hospital as a valuable resource when caring for acutely ill patients.[21, 22]

Another potential explanation is that despite their preferences, PCPs who care for younger, minority patients lack access to hospitalist services. One large study of Medicare beneficiaries reported that physicians who care for black patients are less well‐trained clinically and often lack access to important clinical resources such as diagnostic imaging and nonemergency hospital admissions.[23] Similarly, international medical graduates are more likely than their US‐trained counterparts to care for underserved patients and to practice in small, independent offices.[24, 25, 26] As hospitalist groups often rely on cross‐subsidization from sources within a large healthcare organization, independent PCPs may have less access to their services when compared with PCPs in managed care organizations or large integrated groups. Viewed in this context, our findings imply that although hospitalists often care for socioeconomically vulnerable patients (eg, younger, uninsured, black men) who lack access to primary care services,[1] they also appear to share care responsibilities for more complex hospitalized patients with PCPs in more affluent communities. Further research may determine if the availability of hospitalists influences racial disparities in hospital care.

Our study has limitations. It is an observational study and thus subject to bias and confounding. As our cohort was formed using fee‐for‐service Medicare data in a single, large state, it may not be generalizable to PCPs who practice in other states, who care for a younger population, or who do not accept Medicare. Our findings also may not reflect the practice patterns of physicians‐in‐training, PCP populations with high board‐certification rates, those employed in temporary positions, or those who interrupt their practices for personal reasons, as we restricted our study to established PCPs who had been in practice long and consistently enough to be associated with 20 hospitalized patients during every year of the study. For example, the lower proportion of female PCPs in our cohort (15.6% in our study in 2009 vs 27.5% reported in a nationally representative 2008 survey[27]) may be explained by our exclusion of women who take prolonged time off for childcare duties. We also did not establish whether patient outcomes or healthcare costs differ between PCPs who adopted the hospitalist model and traditionalists. Finally, we could not examine the effect of a number of PCP factors that could plausibly influence whether or not PCPs relinquish inpatient care to hospitalists, such as their comfort with providing inpatient care, having hospital admitting privileges, having office‐based access to hospitals' electronic medical records, or the distance between their office and the hospital. However, this study lays the groundwork for future studies to explore these factors.

In summary, this study is the first, to our knowledge, to characterize PCPs who relinquished inpatient responsibilities to hospitalists. Our findings suggest that some groups of PCPs are more likely to refer patient to hospitalists, that the relationship between hospitalists and PCPs has evolved over time, and that the hospitalist model still has ample room to grow.

Although primary care physicians (PCPs) have traditionally treated patients in both ambulatory and hospital settings, many relinquished inpatient duties to hospitalists in recent decades.[1] Little is known about the PCPs who relinquished inpatient care duties or how the transition to the hospitalist model occurred. For example, what are the characteristics of PCPs who change? Do PCPs adopt the hospitalist model enthusiastically or cautiously? Characterizing PCPs who adopted the hospitalist model can help hospitalists understand their specialty's history and also inform health services research.

Much of the interest in the hospitalist model has been generated by studies reporting improved outcomes and lower hospital lengths of stay associated with hospitalist care.[2, 3, 4, 5] Conversely, detractors of the model point to reports of higher postacute care utilization among hospitalist patients.[6] Although these studies usually adjusted for differences among patients and hospitals, they did not account for PCP characteristics. As patients' access to PCPs and their PCP's capabilities are both plausible factors that could influence hospital length of stay (eg, decisions to complete more or less of a workup in the hospital), quality of care transitions, and postdischarge utilization, it is important to determine if PCPs who use hospitalists differ systematically from those who do not to correctly interpret health system utilization patterns that currently are attributed only to hospitalists.[7, 8]

We conducted this study to determine if observable PCP factors are associated with patients' use of hospitalists and to describe the trajectory by which PCPs referred their patients to hospitalists over time.

METHODS

RESULTS

During the 2001 through 2009 study period, between 2252 and 2848 PCPs were associated with at least 20 hospitalized beneficiaries in any single year. Among these, 1172 PCPs were associated with at least 20 hospitalized beneficiaries in every year of the study period. These 1172 PCPs were associated with 608,686 hospitalizations over the 9 years.

Table 1 presents the characteristics of the PCPs who contributed to the cross‐sectional analyses in 2001 (N=2252) and 2009 (N=2387), as well as the 1172 PCPs for whom we had data for all 9 years for the longitudinal analyses. Most PCPs were male, trained in the United States, and were board certified. The average number of Medicare patients seen by these PCPs and number of outpatient Medicare visits went up about 7% between 2001 and 2009.

Figure 1 graphs the percentage of PCPs as a function of what percent of their hospitalized patients received care from hospitalists, and how that changed from 2001 to 2009. For 70.9% of PCPs, fewer than 5% of their hospitalized patients received hospitalist care in 2001. By 2009, the percent of PCPs in this category had decreased to 15.2%. In contrast, in 2001, more than half of the patients for 2.1% of PCPs received hospitalist care, and the percent of PCPs in this category increased to 26.3% by 2009.

Figure 1

The pattern in Figure 1 shows that PCPs' use of hospitalists changed continuously and gradually over time. However, this pattern describes the PCPs as a group. When examined at the individual PCP level, different patterns emerge. Figure 2, which presents selected individual PCP's use of hospitalists over time, shows several distinct subpatterns of PCP practice behaviors. First, there are PCPs whose use of hospitalists was high in 2001 and stayed high or increased over time (eg, PCP A). There also were PCPs whose use of hospitalists stayed low over the entire study period (eg, PCP B). Finally, there were PCPs whose use of hospitalists was low in 2001 but high in 2009 (eg, PCP C). For this last group, the pattern of change in hospitalist utilization over time was discontinuous; that is, most of the increase occurred over a 1‐ or 2‐year period, instead of increasing gradually over time.

Figure 2

Among the 1172 PCPs associated with 20 hospitalized beneficiaries each year in all 9 years of the study period, group‐based trajectory modeling classified their practice patterns into 4 distinct trajectories (Figure 3). Among PCPs in group 1, more than one‐third of their hospitalized patients were cared for by hospitalists in 2001, and this increased to 60% by 2009. PCPs in groups 2 and 3 rarely used hospitalist care in 2001 but increased their use over time. The increase started early in the period for PCPs in group 2 and later for those in group 3. PCPs in group 4 were associated with little hospitalist use throughout the study period.

Figure 3

We constructed a model to describe the odds of a patient receiving care from hospitalists during the study period using patients associated with these 1172 PCPs. After adjusting for patient characteristics, the residual intraclass correlation coefficient for PCP level was 0.334, which indicates that 33.4% of the variance in whether a hospitalized patient received care from a hospitalist is explained by which PCP the patient saw. When adjusting for both patient and PCP characteristics, the overall odds of a patient receiving hospitalist care increased by 30% (95% confidence interval [CI]: 1.29‐1.30) per year from 2001 through 2009.

There were also significant interactions between year of hospitalization and several PCP characteristics. These interactions are illustrated in Table 2, which stratifies each of those PCP characteristics by 3 time periods: 2001 to 2003, 2004 to 2006, and 2007 to 2009. In all time periods, patients were more likely to receive hospitalist care if their PCP was US trained (US vs international medical graduate: odds ratio [OR]: 1.42, 95% CI: 1.19‐1.69 in 20012003; OR: 1.46, 95% CI: 1.23‐1.73 in 20072009), or specialized in family medicine (family medicine vs internal medicine: OR: 1.46, 95% CI: 1.25‐1.72 in 20012003; OR: 1.46, 95% CI: 1.25‐1.70 in 20072009). Over time, the relative odds of a patient receiving care from hospitalists decreased if their PCP was female (female vs male: OR: 1.91, 95% CI: 1.46‐2.50 in 20012003 vs OR: 1.50, 95% CI: 1.15‐1.95 in 20072009) or practiced in an urban area (largest vs smallest MSA: OR: 3.34, 95% CI: 2.72‐4.09 in 20012003; OR: 2.22, 95% CI: 1.82‐2.71 in 20072009). Although the longest‐practicing PCPs were most likely to use hospitalists in the early 2000s, this effect disappeared by 2007 to 2009 (most vs least years in practice: OR: 1.35, 95% CI: 1.06‐1.72 in 20012003 vs OR: 0.92, 95% CI: 0.73‐1.17 in 20072009).

In terms of PCP workload, patients of PCPs with high outpatient activity were more likely to receive hospitalists care throughout the study period, although the association had decreased by 2007 to 2009 (highest vs lowest outpatient volume: OR: 1.46, 95% CI: 1.30‐1.63 in 20012003 vs OR: 1.32, 95% CI: 1.20‐1.44 in 20072009). In contrast, PCPs with the lowest inpatient volumes became more likely to use hospitalists by the end of the study period (highest vs lowest inpatient volume: OR: 1.05, 95% CI: 0.95‐1.18 in 20012003 vs OR: 0.55, 95% CI: 0.51‐0.60 in 20072009).

The characteristics of PCPs' practice panels also were associated with patients' likelihood of receiving care from hospitalists. PCPs whose practice panels consisted of patients who were predominantly male, white, or with more outpatient comorbidities were consistently more likely to use hospitalists throughout the study period. PCPs with older patient panels were less likely to use hospitalists in 2001 to 2003, but by 2007 to 2009, they were slightly more likely to do so (oldest vs youngest average outpatient panel age: OR: 0.72, 95% CI: 0.64‐0.81 in 20012003 vs OR: 1.15, 95% CI: 1.05‐1.26 in 20072009).

CONCLUSIONS

Prior studies of the hospitalist model have shown that the likelihood of a patient receiving inpatient care from hospitalists is associated with patient characteristics, hospital characteristics, geographic region, and type of admission.[1, 16, 17] We found that PCP characteristics also predict whether patients receive care from hospitalists and that their use of hospitalists developed dynamically between 2001 to 2009. Although many factors (such as whether patients were admitted to a hospital where their PCP had admitting privileges) can influence the decision to use hospitalists, we found that over one‐third of the variance in whether a hospitalized patient received care from a hospitalist is explained by which PCP the patient saw. In showing that systemic differences exist among PCPs who use hospitalists and those who do not, our study suggests that future research on the hospitalist model should, if possible, adjust for PCP characteristics in addition to hospital and patient factors.

Although this study identifies the existence and magnitude of differences in whether or not PCPs use hospitalists, it cannot explain why the differences exist. We only can offer hypotheses. For example, our finding that PCPs with the most years of practice experience were more likely to use hospitalists in the early 2000s but not in more recent years suggests that in hospital medicine's early years, long‐practicing generalist physicians were choosing between practicing traditionalist medicine and adopting the hospitalists model, but by 2009, experienced generalist physicians had already specialized to either inpatient or outpatient settings earlier in their careers. On the other hand, the decreasing odds of urban PCPs using hospitalists may reflect a relative growth in hospitalist use in less populated areas rather than a change in urban PCPs' practice patterns.

PCPs trained in family medicine have reported less inpatient training and less comfort with providing hospital care,[18, 19] thus it is unsurprising that family physicians were more likely to refer patients to hospitalists. Although a recent study reported that family physicians' inpatient volumes remained constant, whereas those of outpatient internists declined between 2003 and 2012, the analysis used University Health Consortium data and thus reflects practice patterns in academic medical centers.[20] Our data suggest that outside of academia, family physicians have embraced the hospitalists as clinical partners.

Meltzer and Chung had previously proposed an economic model to describe the growing use of hospitalists in the United States. They posited that decisions to adopt the hospitalist model are governed by trade‐offs between coordination costs (eg, time and effort spent coordinating multiple providers across different settings) and switching costs (eg, time spent traveling between the office and the hospital or the effort of adjusting to different work settings).[16] The authors hypothesized that empirical testing of this model would show PCPs are more likely to use hospitalists if they have less available professional time (ie, work fewer hours per week), are female (due to competing demands from domestic responsibilities), have relatively few hospitalized patients, or live in areas with high traffic congestion. Our findings provide empirical evidence to support their division‐of‐labor model in showing that patients were more likely to receive hospitalist care if their PCP was female, practiced in an urban location, had higher outpatient practice volumes, or had lower inpatient volumes.

At first glance, some of our findings appear to contradict our earlier study, which showed that younger, black, male patients are more likely to receive inpatient care from hospitalists.[1] However, that study included patients regardless of whether they had a PCP. This study shows that when patients have a PCP, their PCPs are more likely to refer them to hospitalists if they are older, white, male, and have more comorbid conditions. A potential explanation for this finding is that PCPs may preferentially use hospitalists when caring for older and sicker hospitalized patients. For example, commentators often cite hospitalists' constant availability in the hospital as a valuable resource when caring for acutely ill patients.[21, 22]

Another potential explanation is that despite their preferences, PCPs who care for younger, minority patients lack access to hospitalist services. One large study of Medicare beneficiaries reported that physicians who care for black patients are less well‐trained clinically and often lack access to important clinical resources such as diagnostic imaging and nonemergency hospital admissions.[23] Similarly, international medical graduates are more likely than their US‐trained counterparts to care for underserved patients and to practice in small, independent offices.[24, 25, 26] As hospitalist groups often rely on cross‐subsidization from sources within a large healthcare organization, independent PCPs may have less access to their services when compared with PCPs in managed care organizations or large integrated groups. Viewed in this context, our findings imply that although hospitalists often care for socioeconomically vulnerable patients (eg, younger, uninsured, black men) who lack access to primary care services,[1] they also appear to share care responsibilities for more complex hospitalized patients with PCPs in more affluent communities. Further research may determine if the availability of hospitalists influences racial disparities in hospital care.

Our study has limitations. It is an observational study and thus subject to bias and confounding. As our cohort was formed using fee‐for‐service Medicare data in a single, large state, it may not be generalizable to PCPs who practice in other states, who care for a younger population, or who do not accept Medicare. Our findings also may not reflect the practice patterns of physicians‐in‐training, PCP populations with high board‐certification rates, those employed in temporary positions, or those who interrupt their practices for personal reasons, as we restricted our study to established PCPs who had been in practice long and consistently enough to be associated with 20 hospitalized patients during every year of the study. For example, the lower proportion of female PCPs in our cohort (15.6% in our study in 2009 vs 27.5% reported in a nationally representative 2008 survey[27]) may be explained by our exclusion of women who take prolonged time off for childcare duties. We also did not establish whether patient outcomes or healthcare costs differ between PCPs who adopted the hospitalist model and traditionalists. Finally, we could not examine the effect of a number of PCP factors that could plausibly influence whether or not PCPs relinquish inpatient care to hospitalists, such as their comfort with providing inpatient care, having hospital admitting privileges, having office‐based access to hospitals' electronic medical records, or the distance between their office and the hospital. However, this study lays the groundwork for future studies to explore these factors.

In summary, this study is the first, to our knowledge, to characterize PCPs who relinquished inpatient responsibilities to hospitalists. Our findings suggest that some groups of PCPs are more likely to refer patient to hospitalists, that the relationship between hospitalists and PCPs has evolved over time, and that the hospitalist model still has ample room to grow.

Although primary care physicians (PCPs) have traditionally treated patients in both ambulatory and hospital settings, many relinquished inpatient duties to hospitalists in recent decades.[1] Little is known about the PCPs who relinquished inpatient care duties or how the transition to the hospitalist model occurred. For example, what are the characteristics of PCPs who change? Do PCPs adopt the hospitalist model enthusiastically or cautiously? Characterizing PCPs who adopted the hospitalist model can help hospitalists understand their specialty's history and also inform health services research.

Much of the interest in the hospitalist model has been generated by studies reporting improved outcomes and lower hospital lengths of stay associated with hospitalist care.[2, 3, 4, 5] Conversely, detractors of the model point to reports of higher postacute care utilization among hospitalist patients.[6] Although these studies usually adjusted for differences among patients and hospitals, they did not account for PCP characteristics. As patients' access to PCPs and their PCP's capabilities are both plausible factors that could influence hospital length of stay (eg, decisions to complete more or less of a workup in the hospital), quality of care transitions, and postdischarge utilization, it is important to determine if PCPs who use hospitalists differ systematically from those who do not to correctly interpret health system utilization patterns that currently are attributed only to hospitalists.[7, 8]

We conducted this study to determine if observable PCP factors are associated with patients' use of hospitalists and to describe the trajectory by which PCPs referred their patients to hospitalists over time.

METHODS

RESULTS

During the 2001 through 2009 study period, between 2252 and 2848 PCPs were associated with at least 20 hospitalized beneficiaries in any single year. Among these, 1172 PCPs were associated with at least 20 hospitalized beneficiaries in every year of the study period. These 1172 PCPs were associated with 608,686 hospitalizations over the 9 years.

Table 1 presents the characteristics of the PCPs who contributed to the cross‐sectional analyses in 2001 (N=2252) and 2009 (N=2387), as well as the 1172 PCPs for whom we had data for all 9 years for the longitudinal analyses. Most PCPs were male, trained in the United States, and were board certified. The average number of Medicare patients seen by these PCPs and number of outpatient Medicare visits went up about 7% between 2001 and 2009.

Figure 1 graphs the percentage of PCPs as a function of what percent of their hospitalized patients received care from hospitalists, and how that changed from 2001 to 2009. For 70.9% of PCPs, fewer than 5% of their hospitalized patients received hospitalist care in 2001. By 2009, the percent of PCPs in this category had decreased to 15.2%. In contrast, in 2001, more than half of the patients for 2.1% of PCPs received hospitalist care, and the percent of PCPs in this category increased to 26.3% by 2009.

Figure 1

The pattern in Figure 1 shows that PCPs' use of hospitalists changed continuously and gradually over time. However, this pattern describes the PCPs as a group. When examined at the individual PCP level, different patterns emerge. Figure 2, which presents selected individual PCP's use of hospitalists over time, shows several distinct subpatterns of PCP practice behaviors. First, there are PCPs whose use of hospitalists was high in 2001 and stayed high or increased over time (eg, PCP A). There also were PCPs whose use of hospitalists stayed low over the entire study period (eg, PCP B). Finally, there were PCPs whose use of hospitalists was low in 2001 but high in 2009 (eg, PCP C). For this last group, the pattern of change in hospitalist utilization over time was discontinuous; that is, most of the increase occurred over a 1‐ or 2‐year period, instead of increasing gradually over time.

Figure 2

Among the 1172 PCPs associated with 20 hospitalized beneficiaries each year in all 9 years of the study period, group‐based trajectory modeling classified their practice patterns into 4 distinct trajectories (Figure 3). Among PCPs in group 1, more than one‐third of their hospitalized patients were cared for by hospitalists in 2001, and this increased to 60% by 2009. PCPs in groups 2 and 3 rarely used hospitalist care in 2001 but increased their use over time. The increase started early in the period for PCPs in group 2 and later for those in group 3. PCPs in group 4 were associated with little hospitalist use throughout the study period.

Figure 3

We constructed a model to describe the odds of a patient receiving care from hospitalists during the study period using patients associated with these 1172 PCPs. After adjusting for patient characteristics, the residual intraclass correlation coefficient for PCP level was 0.334, which indicates that 33.4% of the variance in whether a hospitalized patient received care from a hospitalist is explained by which PCP the patient saw. When adjusting for both patient and PCP characteristics, the overall odds of a patient receiving hospitalist care increased by 30% (95% confidence interval [CI]: 1.29‐1.30) per year from 2001 through 2009.

There were also significant interactions between year of hospitalization and several PCP characteristics. These interactions are illustrated in Table 2, which stratifies each of those PCP characteristics by 3 time periods: 2001 to 2003, 2004 to 2006, and 2007 to 2009. In all time periods, patients were more likely to receive hospitalist care if their PCP was US trained (US vs international medical graduate: odds ratio [OR]: 1.42, 95% CI: 1.19‐1.69 in 20012003; OR: 1.46, 95% CI: 1.23‐1.73 in 20072009), or specialized in family medicine (family medicine vs internal medicine: OR: 1.46, 95% CI: 1.25‐1.72 in 20012003; OR: 1.46, 95% CI: 1.25‐1.70 in 20072009). Over time, the relative odds of a patient receiving care from hospitalists decreased if their PCP was female (female vs male: OR: 1.91, 95% CI: 1.46‐2.50 in 20012003 vs OR: 1.50, 95% CI: 1.15‐1.95 in 20072009) or practiced in an urban area (largest vs smallest MSA: OR: 3.34, 95% CI: 2.72‐4.09 in 20012003; OR: 2.22, 95% CI: 1.82‐2.71 in 20072009). Although the longest‐practicing PCPs were most likely to use hospitalists in the early 2000s, this effect disappeared by 2007 to 2009 (most vs least years in practice: OR: 1.35, 95% CI: 1.06‐1.72 in 20012003 vs OR: 0.92, 95% CI: 0.73‐1.17 in 20072009).

In terms of PCP workload, patients of PCPs with high outpatient activity were more likely to receive hospitalists care throughout the study period, although the association had decreased by 2007 to 2009 (highest vs lowest outpatient volume: OR: 1.46, 95% CI: 1.30‐1.63 in 20012003 vs OR: 1.32, 95% CI: 1.20‐1.44 in 20072009). In contrast, PCPs with the lowest inpatient volumes became more likely to use hospitalists by the end of the study period (highest vs lowest inpatient volume: OR: 1.05, 95% CI: 0.95‐1.18 in 20012003 vs OR: 0.55, 95% CI: 0.51‐0.60 in 20072009).

The characteristics of PCPs' practice panels also were associated with patients' likelihood of receiving care from hospitalists. PCPs whose practice panels consisted of patients who were predominantly male, white, or with more outpatient comorbidities were consistently more likely to use hospitalists throughout the study period. PCPs with older patient panels were less likely to use hospitalists in 2001 to 2003, but by 2007 to 2009, they were slightly more likely to do so (oldest vs youngest average outpatient panel age: OR: 0.72, 95% CI: 0.64‐0.81 in 20012003 vs OR: 1.15, 95% CI: 1.05‐1.26 in 20072009).

CONCLUSIONS

Prior studies of the hospitalist model have shown that the likelihood of a patient receiving inpatient care from hospitalists is associated with patient characteristics, hospital characteristics, geographic region, and type of admission.[1, 16, 17] We found that PCP characteristics also predict whether patients receive care from hospitalists and that their use of hospitalists developed dynamically between 2001 to 2009. Although many factors (such as whether patients were admitted to a hospital where their PCP had admitting privileges) can influence the decision to use hospitalists, we found that over one‐third of the variance in whether a hospitalized patient received care from a hospitalist is explained by which PCP the patient saw. In showing that systemic differences exist among PCPs who use hospitalists and those who do not, our study suggests that future research on the hospitalist model should, if possible, adjust for PCP characteristics in addition to hospital and patient factors.

Although this study identifies the existence and magnitude of differences in whether or not PCPs use hospitalists, it cannot explain why the differences exist. We only can offer hypotheses. For example, our finding that PCPs with the most years of practice experience were more likely to use hospitalists in the early 2000s but not in more recent years suggests that in hospital medicine's early years, long‐practicing generalist physicians were choosing between practicing traditionalist medicine and adopting the hospitalists model, but by 2009, experienced generalist physicians had already specialized to either inpatient or outpatient settings earlier in their careers. On the other hand, the decreasing odds of urban PCPs using hospitalists may reflect a relative growth in hospitalist use in less populated areas rather than a change in urban PCPs' practice patterns.

PCPs trained in family medicine have reported less inpatient training and less comfort with providing hospital care,[18, 19] thus it is unsurprising that family physicians were more likely to refer patients to hospitalists. Although a recent study reported that family physicians' inpatient volumes remained constant, whereas those of outpatient internists declined between 2003 and 2012, the analysis used University Health Consortium data and thus reflects practice patterns in academic medical centers.[20] Our data suggest that outside of academia, family physicians have embraced the hospitalists as clinical partners.

Meltzer and Chung had previously proposed an economic model to describe the growing use of hospitalists in the United States. They posited that decisions to adopt the hospitalist model are governed by trade‐offs between coordination costs (eg, time and effort spent coordinating multiple providers across different settings) and switching costs (eg, time spent traveling between the office and the hospital or the effort of adjusting to different work settings).[16] The authors hypothesized that empirical testing of this model would show PCPs are more likely to use hospitalists if they have less available professional time (ie, work fewer hours per week), are female (due to competing demands from domestic responsibilities), have relatively few hospitalized patients, or live in areas with high traffic congestion. Our findings provide empirical evidence to support their division‐of‐labor model in showing that patients were more likely to receive hospitalist care if their PCP was female, practiced in an urban location, had higher outpatient practice volumes, or had lower inpatient volumes.

At first glance, some of our findings appear to contradict our earlier study, which showed that younger, black, male patients are more likely to receive inpatient care from hospitalists.[1] However, that study included patients regardless of whether they had a PCP. This study shows that when patients have a PCP, their PCPs are more likely to refer them to hospitalists if they are older, white, male, and have more comorbid conditions. A potential explanation for this finding is that PCPs may preferentially use hospitalists when caring for older and sicker hospitalized patients. For example, commentators often cite hospitalists' constant availability in the hospital as a valuable resource when caring for acutely ill patients.[21, 22]

Another potential explanation is that despite their preferences, PCPs who care for younger, minority patients lack access to hospitalist services. One large study of Medicare beneficiaries reported that physicians who care for black patients are less well‐trained clinically and often lack access to important clinical resources such as diagnostic imaging and nonemergency hospital admissions.[23] Similarly, international medical graduates are more likely than their US‐trained counterparts to care for underserved patients and to practice in small, independent offices.[24, 25, 26] As hospitalist groups often rely on cross‐subsidization from sources within a large healthcare organization, independent PCPs may have less access to their services when compared with PCPs in managed care organizations or large integrated groups. Viewed in this context, our findings imply that although hospitalists often care for socioeconomically vulnerable patients (eg, younger, uninsured, black men) who lack access to primary care services,[1] they also appear to share care responsibilities for more complex hospitalized patients with PCPs in more affluent communities. Further research may determine if the availability of hospitalists influences racial disparities in hospital care.

Our study has limitations. It is an observational study and thus subject to bias and confounding. As our cohort was formed using fee‐for‐service Medicare data in a single, large state, it may not be generalizable to PCPs who practice in other states, who care for a younger population, or who do not accept Medicare. Our findings also may not reflect the practice patterns of physicians‐in‐training, PCP populations with high board‐certification rates, those employed in temporary positions, or those who interrupt their practices for personal reasons, as we restricted our study to established PCPs who had been in practice long and consistently enough to be associated with 20 hospitalized patients during every year of the study. For example, the lower proportion of female PCPs in our cohort (15.6% in our study in 2009 vs 27.5% reported in a nationally representative 2008 survey[27]) may be explained by our exclusion of women who take prolonged time off for childcare duties. We also did not establish whether patient outcomes or healthcare costs differ between PCPs who adopted the hospitalist model and traditionalists. Finally, we could not examine the effect of a number of PCP factors that could plausibly influence whether or not PCPs relinquish inpatient care to hospitalists, such as their comfort with providing inpatient care, having hospital admitting privileges, having office‐based access to hospitals' electronic medical records, or the distance between their office and the hospital. However, this study lays the groundwork for future studies to explore these factors.

In summary, this study is the first, to our knowledge, to characterize PCPs who relinquished inpatient responsibilities to hospitalists. Our findings suggest that some groups of PCPs are more likely to refer patient to hospitalists, that the relationship between hospitalists and PCPs has evolved over time, and that the hospitalist model still has ample room to grow.

An important role of the hospitalist educator is to teach residents and medical students how to diagnose and manage acute medical problems. However, clinical reasoning is complex and nuanced, and there are many challenges to teaching this important process. Medical inpatients are increasingly complex, older, and more seriously ill.[1] Documentation requirements and productivity obligations compete with teaching time. Hospitalists must adjust their teaching for learners from different professions and at various levels of training. In addition, hospitalists tend to be less experienced, and must balance the need to learn their roles as clinicians with developing their own skills as educators.[2]

Despite the challenges inherent to the setting, inpatient rotations provide tremendous teaching and learning opportunities. Patients with undifferentiated complaints or known diagnoses in need of management decisions are available to stimulate discussion. Hospitalist educators have the opportunity to assess residents' progress along the developmental milestones, which residency programs are now required to report for accreditation,[3] and provide role modeling for residents who are developing their own teaching skills.

To maximize these opportunities, attendings must engage trainees to practice clinical reasoning and identify their own knowledge gaps. Various strategies for facilitating the clinical reasoning discussion exist, but two frameworksthe One‐Minute Preceptor (OMP) and SNAPPShave been well studied, albeit mainly in the outpatient setting. Both models offer ways to maximize teaching and assess clinical reasoning, but they have different methods and strengths. This article provides a narrative review of the two frameworks and discusses how they can be applied to the inpatient teaching environment. Hospitalists can utilize these models or components of each framework to facilitate teaching on inpatient teams and enhance their roles as educators.

ONE‐MINUTE PRECEPTOR

The OMP was first described in 1992 by Neher and colleagues as an alternative to the traditional model of precepting.[4] It gives preceptors a method to facilitate learners presentation of their thought process and then for the preceptor to provide targeted teaching points.[4] The OMP helps diagnose both learner and patient, whereas the traditional model focuses on diagnosing the patient.[5] In the traditional model, the attending questions the learner to diagnose the patient, which does not often make clear the learner's thinking process. Thus, there may be a mismatch between the teaching points the preceptor makes and what the learner really needs to know.[5] There are several key benefits to the OMP compared to the traditional model; broadly, these relate to improved ability to assess the learner and provide targeted teaching,[4, 5, 6, 7] improved integration of feedback,[4, 8, 9, 10] learner preference,[11] and ease with which it is learned by faculty members.[4]

The OMP model consists of five steps outlined in Table 1. Step 1, getting a commitment, can involve any aspect of the casediagnosis, treatment, or follow‐upand learners should be challenged to make intellectual commitments just beyond their level of comfort.[12] Steps 1 and 2 bring to light the learner's individual learning needs,[11] then the preceptor follows up with personalized teaching. The OMP is efficient; no increase in time was needed to precept a case in an outpatient study.[9] In a separate outpatient study, the OMP led preceptors to be more likely to teach about disease‐specific points and differential diagnosis, as compared to generic items such as history taking and presentation skills with the traditional model.[5]

Faculty feel better prepared to assess learners and provide feedback with the OMP model.[6, 9] Aagaard and colleagues provided 116 mostly ambulatory preceptors with scripted, videotaped encounters of the OMP and traditional models. The OMP improved preceptors' confidence at rating students' presentation skills, clinical reasoning, and fund of knowledge. It was rated more efficient and effective, and preceptors were able to diagnose the patient with the same or improved accuracy compared to the traditional model.[6] In a pre‐post study assessing the efficacy of a faculty development workshop, students rated ambulatory teaching encounters incorporating the OMP model as having increased quantity and quality of feedback. Furthermore, faculty reported improved ability to evaluate students and were more likely to let students reach their own conclusions and create their own postencounter learning plans.[9]

The OMP is also well‐received by trainees. Teherani and colleagues analyzed medical students' responses to videotaped teaching encounters of the OMP and traditional models. Students gave higher mean ratings for all studied items (including feedback, involving the student in decision‐making, and overall effectiveness) to the OMP model, and preferred it over the traditional model.[11]

Several studies have evaluated the OMP for use by residents as teachers,[10, 13, 14] and it is one of the most common models taught to residents.[13] One study evaluated the impact of a one‐day workshop for 276 residents that included the five‐step microskills model (also known as the OMP).[10] Residents felt more prepared to teach, set expectations, and provide feedback.[10] The OMP model, despite brief training, is effective in improving residents' teaching effectiveness and confidence.[13]

The only study we found that exclusively evaluated the OMP in the inpatient setting was a randomized trial[8] involving 57 internal medicine residents. Interns and students rated OMP‐trained residents more highly in 4 of 5 behaviors. The behavior that showed no difference from the control group was teaching general rules.[8] However, there was no difference in ratings of overall teaching effectiveness between groups.[8]

Our review of the literature on the OMP shows it is a quickly learned, easily implemented framework for teaching clinical reasoning. It has been used across specialties and settings, provides a built‐in mechanism for feedback, and allows educators to assess trainees' reasoning while extracting the clinical information needed to work efficiently.

SNAPPS

SNAPPS was first described in 2003 by Wolpaw and colleagues. It is a six‐step learner‐centered model as outlined in Table 2.[15] Unlike the OMP, SNAPPS requires both trainee and teacher to learn the framework. In doing so, the responsibility for directing the teaching encounter is shifted toward the learner.[15] Consequently, this model may be best suited to advanced or motivated learners. Like the OMP, SNAPPS was originally described for the ambulatory environment. However, it has been studied in the inpatient setting as well.

With SNAPPS, the teaching encounter is learner driven. The trainee presents the case and directs the discussion of differential diagnosis. The educator does not have an active role until the fourth step, where the learner asks questions or identifies areas of uncertainty. But even at this stage, the discussion is learner driven. Step 5, planning management, is collaborative, with trainees suggesting management plans with appropriate attending guidance. Depending on learner skill level or case difficulty, the preceptor may need to play more or less of an active role. The final step, picking a case‐related issue to examine, extends the learning beyond the initial encounter, and ensures that it is individualized and relevant. This step also encourages learner progression toward the Accreditation Council for Graduate Medical Education (ACGME) competency of practice‐based learning and improvement.[3]

A handful of studies have evaluated the SNAPPS model. A randomized comparison group trial found that SNAPPS‐trained students outperformed students trained to elicit feedback and students who received the usual and customary preparation.[16] Notably, SNAPPS students expressed more than twice as many differential diagnoses, justified their reasoning more than five times as often, and expressed more questions and uncertainties. The SNAPPS students' presentations were no longer than in the usual and customary group, and were just one minute longer than in the group trained to elicit feedback.[16] A follow‐up analysis found that 100% of the SNAPPS students expressed an uncertainty (i.e. step 4) compared with 54% of the comparison group, and that most of these uncertainties related to diagnostic reasoning.[17]

A study of medicine clerkship students evaluated the impact of extending SNAPPS to the inpatient setting and including educational prescriptions.[18] The goal was to facilitate the formulation and answering of clinical questions by using the patient, intervention, comparison, outcome (PICO) format for step 6 (selecting a case‐based issue to learn about). Dubbing this SNAPPS‐Plus, the authors found that 99% of cases included a question, and 93% of those were answered. Most questions related to therapeutics, and there was a positive correlation between questions more closely corresponding to the PICO format and higher quality answers.[18]

As with the OMP, SNAPPS does not require additional time for case presentations compared to the usual method.[16] From the perspective of a busy hospitalist, this model takes some responsibility for education away from faculty and places it on the learner. This is an important process for fostering self‐directed learning. As with the OMP, SNAPPS appears easily translatable from the outpatient to inpatient setting. Its main downside is the training time required for both parties to implement it.

TRANSLATING THE MODELS TO THE INPATIENT SETTING

The OMP and SNAPPS have largely been used in the outpatient setting. However, we propose that hospitalists can adapt either model for teaching on ward rotations, as the steps of each framework are not exclusive to one clinical setting.

Although the OMP is typically used between a preceptor and single trainee, it is well suited to engaging the entire group on inpatient rounds (Table 3). For example, a student could commit to and support a diagnosis (steps 1 and 2), whereas the intern could commit to and provide evidence for a treatment or management option. Attendings can repeat steps 1 and 2 for patients' secondary problems, encouraging learners to commit to other items on the problem list.

While teaching general rules (step 3) in the group setting, hospitalists should emphasize basic principles for students (which will serve as reinforcement for residents) as well as discuss more complex rules for the edification of all team members. Hospitalists should encourage senior residents to speak up during this step and share their knowledge with the group. This is an opportunity for residents to practice their role as teachers, and for faculty to assess their clinical acumen. However, residents struggled with teaching general rules in Furney and colleagues' randomized trial.[8] Successful clinical teachers use a mix of improvisational teaching and curriculum scripts developed through years of experience.[19] Hospitalists can model this method of instruction for residents who are learning to teach. For more junior hospitalists who may still be developing their own teaching scripts, the OMP provides an opportunity to regularly integrate these scripts into rounds.

The OMP teaching encounter ends with feedback. Providing real‐time feedback to an individual in the group setting could feel awkward. Reassuringly, in Furney and colleagues' study, some of the greatest gains were in the realm of feedback, as reported by both the senior residents providing the feedback and the interns and students on the receiving end.[8] Although the OMP builds in a space for feedback, it does not teach one how to give feedback. Although it is possible that not all feedback is beneficial, trainees are eager to receive constructive input, and hospitalists should not fear providing this in front of the group. Thoughtful critique of one trainee can provide learning opportunities for others listening in.

SNAPPS is also well suited to inpatient education (Table 4). Because it emphasizes a discussion of differential diagnosis, it works well for new admissions. Because hospitalized patients usually have multiple problems, learners may repeat steps 2 and 3 for each problem, or just for the primary issue. On subsequent days, a standard presentation may work better, but if new problems arise (e.g. fever), hospitalists can ask learners to go through the SNAPPS steps for the new issue.

Step 6 of SNAPPS provides trainees an opportunity to search for and present relevant information to guide patient management. To incorporate more formal teaching time each day, set aside 10 minutes before rounds for learners to present their answers to the team. Also, because SNAPPS has the learner ask about uncertainties, faculty can use their on‐the‐fly teaching time to answer questions for which trainees do not know the answer. In the era of problem‐based learning (PBL) and medical school curricula that foster self‐directed learning from day one, many students should find SNAPPS a natural extension of PBL‐style learning from the preclinical into the clinical years.

Unlike the OMP, SNAPPS does not build in a step for feedback. Therefore, preceptors should focus on step 4 as an opportunity for this. Because feedback is paired with discussion of an uncertainty, it focuses on a trainee's immediate needs and can maximize learning opportunities.[17]

Clinical educators must simultaneously diagnose and manage patients as well as assess learners' abilities.[20] Workplace‐based assessment is particularly important for residents, and hospitalists play a pivotal role in determining their progression along the developmental milestones for achieving the ACGME competencies in medical knowledge, patient care, and practice‐based learning and improvement.[3] Both the OMP and SNAPPS frameworks encourage trainees to think out loud, providing some transparency to their thought process and enabling faculty to more accurately assess their clinical reasoning.

CONCLUSION

Many hospitalists may already use a teaching approach resembling the OMP. It has a familiar, back‐and‐forth rhythm. By explicitly following its steps, however, attendings can ensure they are providing feedback and individualized teaching with each case. SNAPPS, on the other hand, relieves faculty of their familiar role of leading the thought process and imparting teaching points. Instead, the trainee directs the encounter, leaving the attending in the role of guide.[15] SNAPPS aims to help students and residents take charge of their education and develop lifelong learning skills.

Both frameworks can be transferred from the ambulatory to inpatient setting with little modification. The OMP is older and better studied. It is easy to learn, and can be utilized by attendings and residents as teachers. In contrast, SNAPPS requires both teacher and trainee to learn the framework. Typically, this means that SNAPPS needs to be implemented systematically, via a clerkship or residency program. However, if a team was motivated, they could learn and apply it for their time together on service. Though it requires more effort to put in place, SNAPPS provides a novel approach to teaching clinical reasoning. Finally, hospitalists need not implement all steps of either framework for every teaching encounter, but can use components of either model, depending on the individual learners, team composition, time available, or clinical case.

Additional studies examining both frameworks' use for inpatient teaching and assessment would be helpful. Potential questions to address include how the team structure of inpatient rotations impacts the effectiveness of either model (e.g. which trainees benefit when committing to diagnoses or getting feedback in front of a group?), whether either model improves senior residents' ability to lead rounds and teach, whether written faculty assessments of residents are more specific and accurate with either model, and the impact of not following all steps of either model. Higher level outcomes for both models would be another area for investigation, including change in clinical performance, exam performance of students and residents, or patient outcomes, such as length of stay, cost per case, or need for rapid response/emntensive care unit transfer.

An important role of the hospitalist educator is to teach residents and medical students how to diagnose and manage acute medical problems. However, clinical reasoning is complex and nuanced, and there are many challenges to teaching this important process. Medical inpatients are increasingly complex, older, and more seriously ill.[1] Documentation requirements and productivity obligations compete with teaching time. Hospitalists must adjust their teaching for learners from different professions and at various levels of training. In addition, hospitalists tend to be less experienced, and must balance the need to learn their roles as clinicians with developing their own skills as educators.[2]

Despite the challenges inherent to the setting, inpatient rotations provide tremendous teaching and learning opportunities. Patients with undifferentiated complaints or known diagnoses in need of management decisions are available to stimulate discussion. Hospitalist educators have the opportunity to assess residents' progress along the developmental milestones, which residency programs are now required to report for accreditation,[3] and provide role modeling for residents who are developing their own teaching skills.

To maximize these opportunities, attendings must engage trainees to practice clinical reasoning and identify their own knowledge gaps. Various strategies for facilitating the clinical reasoning discussion exist, but two frameworksthe One‐Minute Preceptor (OMP) and SNAPPShave been well studied, albeit mainly in the outpatient setting. Both models offer ways to maximize teaching and assess clinical reasoning, but they have different methods and strengths. This article provides a narrative review of the two frameworks and discusses how they can be applied to the inpatient teaching environment. Hospitalists can utilize these models or components of each framework to facilitate teaching on inpatient teams and enhance their roles as educators.

ONE‐MINUTE PRECEPTOR

The OMP was first described in 1992 by Neher and colleagues as an alternative to the traditional model of precepting.[4] It gives preceptors a method to facilitate learners presentation of their thought process and then for the preceptor to provide targeted teaching points.[4] The OMP helps diagnose both learner and patient, whereas the traditional model focuses on diagnosing the patient.[5] In the traditional model, the attending questions the learner to diagnose the patient, which does not often make clear the learner's thinking process. Thus, there may be a mismatch between the teaching points the preceptor makes and what the learner really needs to know.[5] There are several key benefits to the OMP compared to the traditional model; broadly, these relate to improved ability to assess the learner and provide targeted teaching,[4, 5, 6, 7] improved integration of feedback,[4, 8, 9, 10] learner preference,[11] and ease with which it is learned by faculty members.[4]

The OMP model consists of five steps outlined in Table 1. Step 1, getting a commitment, can involve any aspect of the casediagnosis, treatment, or follow‐upand learners should be challenged to make intellectual commitments just beyond their level of comfort.[12] Steps 1 and 2 bring to light the learner's individual learning needs,[11] then the preceptor follows up with personalized teaching. The OMP is efficient; no increase in time was needed to precept a case in an outpatient study.[9] In a separate outpatient study, the OMP led preceptors to be more likely to teach about disease‐specific points and differential diagnosis, as compared to generic items such as history taking and presentation skills with the traditional model.[5]

Faculty feel better prepared to assess learners and provide feedback with the OMP model.[6, 9] Aagaard and colleagues provided 116 mostly ambulatory preceptors with scripted, videotaped encounters of the OMP and traditional models. The OMP improved preceptors' confidence at rating students' presentation skills, clinical reasoning, and fund of knowledge. It was rated more efficient and effective, and preceptors were able to diagnose the patient with the same or improved accuracy compared to the traditional model.[6] In a pre‐post study assessing the efficacy of a faculty development workshop, students rated ambulatory teaching encounters incorporating the OMP model as having increased quantity and quality of feedback. Furthermore, faculty reported improved ability to evaluate students and were more likely to let students reach their own conclusions and create their own postencounter learning plans.[9]

The OMP is also well‐received by trainees. Teherani and colleagues analyzed medical students' responses to videotaped teaching encounters of the OMP and traditional models. Students gave higher mean ratings for all studied items (including feedback, involving the student in decision‐making, and overall effectiveness) to the OMP model, and preferred it over the traditional model.[11]

Several studies have evaluated the OMP for use by residents as teachers,[10, 13, 14] and it is one of the most common models taught to residents.[13] One study evaluated the impact of a one‐day workshop for 276 residents that included the five‐step microskills model (also known as the OMP).[10] Residents felt more prepared to teach, set expectations, and provide feedback.[10] The OMP model, despite brief training, is effective in improving residents' teaching effectiveness and confidence.[13]

The only study we found that exclusively evaluated the OMP in the inpatient setting was a randomized trial[8] involving 57 internal medicine residents. Interns and students rated OMP‐trained residents more highly in 4 of 5 behaviors. The behavior that showed no difference from the control group was teaching general rules.[8] However, there was no difference in ratings of overall teaching effectiveness between groups.[8]

Our review of the literature on the OMP shows it is a quickly learned, easily implemented framework for teaching clinical reasoning. It has been used across specialties and settings, provides a built‐in mechanism for feedback, and allows educators to assess trainees' reasoning while extracting the clinical information needed to work efficiently.

SNAPPS

SNAPPS was first described in 2003 by Wolpaw and colleagues. It is a six‐step learner‐centered model as outlined in Table 2.[15] Unlike the OMP, SNAPPS requires both trainee and teacher to learn the framework. In doing so, the responsibility for directing the teaching encounter is shifted toward the learner.[15] Consequently, this model may be best suited to advanced or motivated learners. Like the OMP, SNAPPS was originally described for the ambulatory environment. However, it has been studied in the inpatient setting as well.

With SNAPPS, the teaching encounter is learner driven. The trainee presents the case and directs the discussion of differential diagnosis. The educator does not have an active role until the fourth step, where the learner asks questions or identifies areas of uncertainty. But even at this stage, the discussion is learner driven. Step 5, planning management, is collaborative, with trainees suggesting management plans with appropriate attending guidance. Depending on learner skill level or case difficulty, the preceptor may need to play more or less of an active role. The final step, picking a case‐related issue to examine, extends the learning beyond the initial encounter, and ensures that it is individualized and relevant. This step also encourages learner progression toward the Accreditation Council for Graduate Medical Education (ACGME) competency of practice‐based learning and improvement.[3]

A handful of studies have evaluated the SNAPPS model. A randomized comparison group trial found that SNAPPS‐trained students outperformed students trained to elicit feedback and students who received the usual and customary preparation.[16] Notably, SNAPPS students expressed more than twice as many differential diagnoses, justified their reasoning more than five times as often, and expressed more questions and uncertainties. The SNAPPS students' presentations were no longer than in the usual and customary group, and were just one minute longer than in the group trained to elicit feedback.[16] A follow‐up analysis found that 100% of the SNAPPS students expressed an uncertainty (i.e. step 4) compared with 54% of the comparison group, and that most of these uncertainties related to diagnostic reasoning.[17]

A study of medicine clerkship students evaluated the impact of extending SNAPPS to the inpatient setting and including educational prescriptions.[18] The goal was to facilitate the formulation and answering of clinical questions by using the patient, intervention, comparison, outcome (PICO) format for step 6 (selecting a case‐based issue to learn about). Dubbing this SNAPPS‐Plus, the authors found that 99% of cases included a question, and 93% of those were answered. Most questions related to therapeutics, and there was a positive correlation between questions more closely corresponding to the PICO format and higher quality answers.[18]

As with the OMP, SNAPPS does not require additional time for case presentations compared to the usual method.[16] From the perspective of a busy hospitalist, this model takes some responsibility for education away from faculty and places it on the learner. This is an important process for fostering self‐directed learning. As with the OMP, SNAPPS appears easily translatable from the outpatient to inpatient setting. Its main downside is the training time required for both parties to implement it.

TRANSLATING THE MODELS TO THE INPATIENT SETTING

The OMP and SNAPPS have largely been used in the outpatient setting. However, we propose that hospitalists can adapt either model for teaching on ward rotations, as the steps of each framework are not exclusive to one clinical setting.

Although the OMP is typically used between a preceptor and single trainee, it is well suited to engaging the entire group on inpatient rounds (Table 3). For example, a student could commit to and support a diagnosis (steps 1 and 2), whereas the intern could commit to and provide evidence for a treatment or management option. Attendings can repeat steps 1 and 2 for patients' secondary problems, encouraging learners to commit to other items on the problem list.

While teaching general rules (step 3) in the group setting, hospitalists should emphasize basic principles for students (which will serve as reinforcement for residents) as well as discuss more complex rules for the edification of all team members. Hospitalists should encourage senior residents to speak up during this step and share their knowledge with the group. This is an opportunity for residents to practice their role as teachers, and for faculty to assess their clinical acumen. However, residents struggled with teaching general rules in Furney and colleagues' randomized trial.[8] Successful clinical teachers use a mix of improvisational teaching and curriculum scripts developed through years of experience.[19] Hospitalists can model this method of instruction for residents who are learning to teach. For more junior hospitalists who may still be developing their own teaching scripts, the OMP provides an opportunity to regularly integrate these scripts into rounds.

The OMP teaching encounter ends with feedback. Providing real‐time feedback to an individual in the group setting could feel awkward. Reassuringly, in Furney and colleagues' study, some of the greatest gains were in the realm of feedback, as reported by both the senior residents providing the feedback and the interns and students on the receiving end.[8] Although the OMP builds in a space for feedback, it does not teach one how to give feedback. Although it is possible that not all feedback is beneficial, trainees are eager to receive constructive input, and hospitalists should not fear providing this in front of the group. Thoughtful critique of one trainee can provide learning opportunities for others listening in.

SNAPPS is also well suited to inpatient education (Table 4). Because it emphasizes a discussion of differential diagnosis, it works well for new admissions. Because hospitalized patients usually have multiple problems, learners may repeat steps 2 and 3 for each problem, or just for the primary issue. On subsequent days, a standard presentation may work better, but if new problems arise (e.g. fever), hospitalists can ask learners to go through the SNAPPS steps for the new issue.

Step 6 of SNAPPS provides trainees an opportunity to search for and present relevant information to guide patient management. To incorporate more formal teaching time each day, set aside 10 minutes before rounds for learners to present their answers to the team. Also, because SNAPPS has the learner ask about uncertainties, faculty can use their on‐the‐fly teaching time to answer questions for which trainees do not know the answer. In the era of problem‐based learning (PBL) and medical school curricula that foster self‐directed learning from day one, many students should find SNAPPS a natural extension of PBL‐style learning from the preclinical into the clinical years.

Unlike the OMP, SNAPPS does not build in a step for feedback. Therefore, preceptors should focus on step 4 as an opportunity for this. Because feedback is paired with discussion of an uncertainty, it focuses on a trainee's immediate needs and can maximize learning opportunities.[17]

Clinical educators must simultaneously diagnose and manage patients as well as assess learners' abilities.[20] Workplace‐based assessment is particularly important for residents, and hospitalists play a pivotal role in determining their progression along the developmental milestones for achieving the ACGME competencies in medical knowledge, patient care, and practice‐based learning and improvement.[3] Both the OMP and SNAPPS frameworks encourage trainees to think out loud, providing some transparency to their thought process and enabling faculty to more accurately assess their clinical reasoning.

CONCLUSION

Many hospitalists may already use a teaching approach resembling the OMP. It has a familiar, back‐and‐forth rhythm. By explicitly following its steps, however, attendings can ensure they are providing feedback and individualized teaching with each case. SNAPPS, on the other hand, relieves faculty of their familiar role of leading the thought process and imparting teaching points. Instead, the trainee directs the encounter, leaving the attending in the role of guide.[15] SNAPPS aims to help students and residents take charge of their education and develop lifelong learning skills.

Both frameworks can be transferred from the ambulatory to inpatient setting with little modification. The OMP is older and better studied. It is easy to learn, and can be utilized by attendings and residents as teachers. In contrast, SNAPPS requires both teacher and trainee to learn the framework. Typically, this means that SNAPPS needs to be implemented systematically, via a clerkship or residency program. However, if a team was motivated, they could learn and apply it for their time together on service. Though it requires more effort to put in place, SNAPPS provides a novel approach to teaching clinical reasoning. Finally, hospitalists need not implement all steps of either framework for every teaching encounter, but can use components of either model, depending on the individual learners, team composition, time available, or clinical case.

Additional studies examining both frameworks' use for inpatient teaching and assessment would be helpful. Potential questions to address include how the team structure of inpatient rotations impacts the effectiveness of either model (e.g. which trainees benefit when committing to diagnoses or getting feedback in front of a group?), whether either model improves senior residents' ability to lead rounds and teach, whether written faculty assessments of residents are more specific and accurate with either model, and the impact of not following all steps of either model. Higher level outcomes for both models would be another area for investigation, including change in clinical performance, exam performance of students and residents, or patient outcomes, such as length of stay, cost per case, or need for rapid response/emntensive care unit transfer.

An important role of the hospitalist educator is to teach residents and medical students how to diagnose and manage acute medical problems. However, clinical reasoning is complex and nuanced, and there are many challenges to teaching this important process. Medical inpatients are increasingly complex, older, and more seriously ill.[1] Documentation requirements and productivity obligations compete with teaching time. Hospitalists must adjust their teaching for learners from different professions and at various levels of training. In addition, hospitalists tend to be less experienced, and must balance the need to learn their roles as clinicians with developing their own skills as educators.[2]

Despite the challenges inherent to the setting, inpatient rotations provide tremendous teaching and learning opportunities. Patients with undifferentiated complaints or known diagnoses in need of management decisions are available to stimulate discussion. Hospitalist educators have the opportunity to assess residents' progress along the developmental milestones, which residency programs are now required to report for accreditation,[3] and provide role modeling for residents who are developing their own teaching skills.

To maximize these opportunities, attendings must engage trainees to practice clinical reasoning and identify their own knowledge gaps. Various strategies for facilitating the clinical reasoning discussion exist, but two frameworksthe One‐Minute Preceptor (OMP) and SNAPPShave been well studied, albeit mainly in the outpatient setting. Both models offer ways to maximize teaching and assess clinical reasoning, but they have different methods and strengths. This article provides a narrative review of the two frameworks and discusses how they can be applied to the inpatient teaching environment. Hospitalists can utilize these models or components of each framework to facilitate teaching on inpatient teams and enhance their roles as educators.

ONE‐MINUTE PRECEPTOR

The OMP was first described in 1992 by Neher and colleagues as an alternative to the traditional model of precepting.[4] It gives preceptors a method to facilitate learners presentation of their thought process and then for the preceptor to provide targeted teaching points.[4] The OMP helps diagnose both learner and patient, whereas the traditional model focuses on diagnosing the patient.[5] In the traditional model, the attending questions the learner to diagnose the patient, which does not often make clear the learner's thinking process. Thus, there may be a mismatch between the teaching points the preceptor makes and what the learner really needs to know.[5] There are several key benefits to the OMP compared to the traditional model; broadly, these relate to improved ability to assess the learner and provide targeted teaching,[4, 5, 6, 7] improved integration of feedback,[4, 8, 9, 10] learner preference,[11] and ease with which it is learned by faculty members.[4]

The OMP model consists of five steps outlined in Table 1. Step 1, getting a commitment, can involve any aspect of the casediagnosis, treatment, or follow‐upand learners should be challenged to make intellectual commitments just beyond their level of comfort.[12] Steps 1 and 2 bring to light the learner's individual learning needs,[11] then the preceptor follows up with personalized teaching. The OMP is efficient; no increase in time was needed to precept a case in an outpatient study.[9] In a separate outpatient study, the OMP led preceptors to be more likely to teach about disease‐specific points and differential diagnosis, as compared to generic items such as history taking and presentation skills with the traditional model.[5]

Faculty feel better prepared to assess learners and provide feedback with the OMP model.[6, 9] Aagaard and colleagues provided 116 mostly ambulatory preceptors with scripted, videotaped encounters of the OMP and traditional models. The OMP improved preceptors' confidence at rating students' presentation skills, clinical reasoning, and fund of knowledge. It was rated more efficient and effective, and preceptors were able to diagnose the patient with the same or improved accuracy compared to the traditional model.[6] In a pre‐post study assessing the efficacy of a faculty development workshop, students rated ambulatory teaching encounters incorporating the OMP model as having increased quantity and quality of feedback. Furthermore, faculty reported improved ability to evaluate students and were more likely to let students reach their own conclusions and create their own postencounter learning plans.[9]

The OMP is also well‐received by trainees. Teherani and colleagues analyzed medical students' responses to videotaped teaching encounters of the OMP and traditional models. Students gave higher mean ratings for all studied items (including feedback, involving the student in decision‐making, and overall effectiveness) to the OMP model, and preferred it over the traditional model.[11]

Several studies have evaluated the OMP for use by residents as teachers,[10, 13, 14] and it is one of the most common models taught to residents.[13] One study evaluated the impact of a one‐day workshop for 276 residents that included the five‐step microskills model (also known as the OMP).[10] Residents felt more prepared to teach, set expectations, and provide feedback.[10] The OMP model, despite brief training, is effective in improving residents' teaching effectiveness and confidence.[13]

The only study we found that exclusively evaluated the OMP in the inpatient setting was a randomized trial[8] involving 57 internal medicine residents. Interns and students rated OMP‐trained residents more highly in 4 of 5 behaviors. The behavior that showed no difference from the control group was teaching general rules.[8] However, there was no difference in ratings of overall teaching effectiveness between groups.[8]

Our review of the literature on the OMP shows it is a quickly learned, easily implemented framework for teaching clinical reasoning. It has been used across specialties and settings, provides a built‐in mechanism for feedback, and allows educators to assess trainees' reasoning while extracting the clinical information needed to work efficiently.

SNAPPS

SNAPPS was first described in 2003 by Wolpaw and colleagues. It is a six‐step learner‐centered model as outlined in Table 2.[15] Unlike the OMP, SNAPPS requires both trainee and teacher to learn the framework. In doing so, the responsibility for directing the teaching encounter is shifted toward the learner.[15] Consequently, this model may be best suited to advanced or motivated learners. Like the OMP, SNAPPS was originally described for the ambulatory environment. However, it has been studied in the inpatient setting as well.

With SNAPPS, the teaching encounter is learner driven. The trainee presents the case and directs the discussion of differential diagnosis. The educator does not have an active role until the fourth step, where the learner asks questions or identifies areas of uncertainty. But even at this stage, the discussion is learner driven. Step 5, planning management, is collaborative, with trainees suggesting management plans with appropriate attending guidance. Depending on learner skill level or case difficulty, the preceptor may need to play more or less of an active role. The final step, picking a case‐related issue to examine, extends the learning beyond the initial encounter, and ensures that it is individualized and relevant. This step also encourages learner progression toward the Accreditation Council for Graduate Medical Education (ACGME) competency of practice‐based learning and improvement.[3]

A handful of studies have evaluated the SNAPPS model. A randomized comparison group trial found that SNAPPS‐trained students outperformed students trained to elicit feedback and students who received the usual and customary preparation.[16] Notably, SNAPPS students expressed more than twice as many differential diagnoses, justified their reasoning more than five times as often, and expressed more questions and uncertainties. The SNAPPS students' presentations were no longer than in the usual and customary group, and were just one minute longer than in the group trained to elicit feedback.[16] A follow‐up analysis found that 100% of the SNAPPS students expressed an uncertainty (i.e. step 4) compared with 54% of the comparison group, and that most of these uncertainties related to diagnostic reasoning.[17]

A study of medicine clerkship students evaluated the impact of extending SNAPPS to the inpatient setting and including educational prescriptions.[18] The goal was to facilitate the formulation and answering of clinical questions by using the patient, intervention, comparison, outcome (PICO) format for step 6 (selecting a case‐based issue to learn about). Dubbing this SNAPPS‐Plus, the authors found that 99% of cases included a question, and 93% of those were answered. Most questions related to therapeutics, and there was a positive correlation between questions more closely corresponding to the PICO format and higher quality answers.[18]

As with the OMP, SNAPPS does not require additional time for case presentations compared to the usual method.[16] From the perspective of a busy hospitalist, this model takes some responsibility for education away from faculty and places it on the learner. This is an important process for fostering self‐directed learning. As with the OMP, SNAPPS appears easily translatable from the outpatient to inpatient setting. Its main downside is the training time required for both parties to implement it.

TRANSLATING THE MODELS TO THE INPATIENT SETTING

The OMP and SNAPPS have largely been used in the outpatient setting. However, we propose that hospitalists can adapt either model for teaching on ward rotations, as the steps of each framework are not exclusive to one clinical setting.

Although the OMP is typically used between a preceptor and single trainee, it is well suited to engaging the entire group on inpatient rounds (Table 3). For example, a student could commit to and support a diagnosis (steps 1 and 2), whereas the intern could commit to and provide evidence for a treatment or management option. Attendings can repeat steps 1 and 2 for patients' secondary problems, encouraging learners to commit to other items on the problem list.

While teaching general rules (step 3) in the group setting, hospitalists should emphasize basic principles for students (which will serve as reinforcement for residents) as well as discuss more complex rules for the edification of all team members. Hospitalists should encourage senior residents to speak up during this step and share their knowledge with the group. This is an opportunity for residents to practice their role as teachers, and for faculty to assess their clinical acumen. However, residents struggled with teaching general rules in Furney and colleagues' randomized trial.[8] Successful clinical teachers use a mix of improvisational teaching and curriculum scripts developed through years of experience.[19] Hospitalists can model this method of instruction for residents who are learning to teach. For more junior hospitalists who may still be developing their own teaching scripts, the OMP provides an opportunity to regularly integrate these scripts into rounds.

The OMP teaching encounter ends with feedback. Providing real‐time feedback to an individual in the group setting could feel awkward. Reassuringly, in Furney and colleagues' study, some of the greatest gains were in the realm of feedback, as reported by both the senior residents providing the feedback and the interns and students on the receiving end.[8] Although the OMP builds in a space for feedback, it does not teach one how to give feedback. Although it is possible that not all feedback is beneficial, trainees are eager to receive constructive input, and hospitalists should not fear providing this in front of the group. Thoughtful critique of one trainee can provide learning opportunities for others listening in.

SNAPPS is also well suited to inpatient education (Table 4). Because it emphasizes a discussion of differential diagnosis, it works well for new admissions. Because hospitalized patients usually have multiple problems, learners may repeat steps 2 and 3 for each problem, or just for the primary issue. On subsequent days, a standard presentation may work better, but if new problems arise (e.g. fever), hospitalists can ask learners to go through the SNAPPS steps for the new issue.

Step 6 of SNAPPS provides trainees an opportunity to search for and present relevant information to guide patient management. To incorporate more formal teaching time each day, set aside 10 minutes before rounds for learners to present their answers to the team. Also, because SNAPPS has the learner ask about uncertainties, faculty can use their on‐the‐fly teaching time to answer questions for which trainees do not know the answer. In the era of problem‐based learning (PBL) and medical school curricula that foster self‐directed learning from day one, many students should find SNAPPS a natural extension of PBL‐style learning from the preclinical into the clinical years.

Unlike the OMP, SNAPPS does not build in a step for feedback. Therefore, preceptors should focus on step 4 as an opportunity for this. Because feedback is paired with discussion of an uncertainty, it focuses on a trainee's immediate needs and can maximize learning opportunities.[17]

Clinical educators must simultaneously diagnose and manage patients as well as assess learners' abilities.[20] Workplace‐based assessment is particularly important for residents, and hospitalists play a pivotal role in determining their progression along the developmental milestones for achieving the ACGME competencies in medical knowledge, patient care, and practice‐based learning and improvement.[3] Both the OMP and SNAPPS frameworks encourage trainees to think out loud, providing some transparency to their thought process and enabling faculty to more accurately assess their clinical reasoning.

CONCLUSION

Many hospitalists may already use a teaching approach resembling the OMP. It has a familiar, back‐and‐forth rhythm. By explicitly following its steps, however, attendings can ensure they are providing feedback and individualized teaching with each case. SNAPPS, on the other hand, relieves faculty of their familiar role of leading the thought process and imparting teaching points. Instead, the trainee directs the encounter, leaving the attending in the role of guide.[15] SNAPPS aims to help students and residents take charge of their education and develop lifelong learning skills.

Both frameworks can be transferred from the ambulatory to inpatient setting with little modification. The OMP is older and better studied. It is easy to learn, and can be utilized by attendings and residents as teachers. In contrast, SNAPPS requires both teacher and trainee to learn the framework. Typically, this means that SNAPPS needs to be implemented systematically, via a clerkship or residency program. However, if a team was motivated, they could learn and apply it for their time together on service. Though it requires more effort to put in place, SNAPPS provides a novel approach to teaching clinical reasoning. Finally, hospitalists need not implement all steps of either framework for every teaching encounter, but can use components of either model, depending on the individual learners, team composition, time available, or clinical case.

Additional studies examining both frameworks' use for inpatient teaching and assessment would be helpful. Potential questions to address include how the team structure of inpatient rotations impacts the effectiveness of either model (e.g. which trainees benefit when committing to diagnoses or getting feedback in front of a group?), whether either model improves senior residents' ability to lead rounds and teach, whether written faculty assessments of residents are more specific and accurate with either model, and the impact of not following all steps of either model. Higher level outcomes for both models would be another area for investigation, including change in clinical performance, exam performance of students and residents, or patient outcomes, such as length of stay, cost per case, or need for rapid response/emntensive care unit transfer.

Acute kidney injury (AKI) is a clinical syndrome broadly defined as an abrupt decline in renal function occurring over a period of hours to days resulting in the retention of nitrogenous and metabolic waste products. Although the initial clinical manifestation of AKI may be oliguria, urine volume can remain normal or even increase. Patients may be asymptomatic, especially early in the course of AKI. The diagnosis is often made in hospitalized patients when biochemical screening reveals a recent increase in serum creatinine and/or blood urea nitrogen concentrations, or when there is a dramatic decrease in urine output.

Older studies looking at the incidence of AKI in hospitalized patients are difficult to interpret due to variable definitions of AKI. Those based on administrative databases were limited by lack of clinical context and/or variation in coding for AKI.[1]

There is no universally accepted operational definition of AKI, and more than 30 different criteria have been employed in various clinical studies. Difficulty in defining AKI lies in the lag time in the rise and fall of the serum creatinine concentration with injury and recovery, the variability of oliguria, and in the heterogeneity of patterns of renal injury. Two classification systems that attempt to capture the spectrum of AKI are the RIFLE (Risk, Injury, Failure, Loss, End Stage) criteria and the AKIN (Acute Kidney Injury Network) criteria.[2, 3] The AKIN criteria parallel the risk, injury, and failure stages of the RIFLE criteria and are the most applicable to characterizing AKI in the hospital (Table 1). AKI is commonly classified by daily urine output as anuric (50 mL/day), oliguric (500 mL/day), or nonoliguric.

With a move toward standardized definitions, recent studies have shown a rising incidence of AKI in hospitalized patients.[4, 5, 6] According to these series, AKI develops in up to 7% of hospitalized patients and in about 30% of those admitted to intensive care units. In one study of consecutive hospital admissions, patients classified by the RIFLE criteria had a sharp rise in the rate of in‐hospital mortality whether they had no change or improvement in creatinine (4.4%), or fell into a risk (15.1%), injury (29.2%), or failure (41.1%) class.[7] The in‐hospital mortality of critically ill patients with AKI is higher than 50%. AKI increases length of stay and hospital costs, and affects the clinical course after discharge.[8, 9] Small increases in serum creatinine during an intensive care unit stay predict increased 10‐year mortality above a critical illness alone.[10]

Risk factors for AKI include advanced age, male gender, African American ethnicity, and diabetes mellitus.[11] The most important risk factor, however, is preexisting chronic kidney disease (CKD).[12] AKI and CKD are tightly linked, each increasing the risk of the other.[13, 14, 15] Preexisting renal insufficiency is a key predictor of postoperative AKI and poor surgical outcomes.[16, 17]

AKI AND CLINICAL CONTEXT

The causes of AKI can be broadly divided into 3 categories: prerenal azotemia (a disorder characterized by renal hypoperfusion in which renal parenchymal tissue integrity is preserved), intrinsic kidney injury with parenchymal tissue injury, and postrenal AKI (dysfunction due to acute obstruction of the urinary tract). Table 2 lists several clinical scenarios sorted into these 3 categories.[18] The general epidemiology of AKI varies based on whether it was acquired in the community or in a hospital setting. Prerenal azotemia accounts for the bulk of community‐acquired AKI, followed in lesser frequency by postrenal and intrinsic etiologies. Prerenal azotemia continues to be the major cause of hospital‐acquired AKI, but intrinsic kidney injury becomes more common.[5, 19]

MEDICAL HISTORY

The initial goal of history taking is to establish whether the patient has AKI rather than the acute discovery of a more chronic process. A recent serum creatinine measurement can be valuable in this regard. In some cases the clinician must make a presumptive diagnosis of AKI while simultaneously reviewing past medical history and family history to assess for underlying CKD. A diagnosis of AKI is more readily established when it occurs during a hospitalization through review of urine output and serial laboratory values.

Symptoms of poor oral intake as well as salt and fluid losses from diarrhea or vomiting suggest a prerenal etiology. Subjective symptoms of lightheadedness, visual clouding, and near‐syncope with standing also suggest volume depletion. Patients should be asked about recent nonsteroidal anti‐inflammatory drug (NSAID) use, as these agents can exacerbate renal hypoperfusion through loss of prostaglandin‐mediated afferent arteriole dilatation. Angiotensin‐converting enzyme inhibitors and angiotensin receptor blockers, especially when combined with diuretics, can generate a hypoperfusion state. Heart failure and liver disease regularly result in an expanded extracellular fluid (ECF) compartment yet a reduced effective arterial volume and predispose to renal hypoperfusion.

A history of decreased urine output or anuria suggests postrenal AKI, but its absence does not rule out urinary tract obstruction. Voiding symptoms such as urinary frequency, hesitancy, or incontinence also raise the possibility of obstructive uropathy. Flank pain and hematuria often accompany obstruction from nephrolithiasis.

Symptoms of fever, skin rash, arthralgias, sinusitis, and/or hemoptysis raise the possibility of glomerulonephritis from infection, collagen vascular disease, or vasculitis. Risk factors for viral hepatitis and human immunodeficiency virus are important to clarify as are systemic symptoms of autoimmunity (eg, dry eyes, dry mouth, eye pain/emnflammation, or visual changes). The recent start of any new medication, including NSAIDs, antibiotics, or proton‐pump inhibitors, raises the possibility of a drug‐induced interstitial nephritis.[20] Statins are direct myotoxins, and the risk of rhabdomyolysis with renal injury increases with dose. Patients may not associate intravenous (IV) contrast or phosphate‐containing bowel preparations (eg, Fleet Phospho Soda), with the development of AKI, thus the clinician must carefully review for recent exposures that could result in intrinsic renal injury.[21]

PHYSICAL EXAMINATION

Estimation of the ECF volume and effective arterial volume are central to assessing the likelihood of renal hypoperfusion. Overt hypotension is the strongest indicator of hypoperfusion, and a careful review of initial blood pressure prior to worsening of renal function can provide significant information. Normal blood pressure does not exclude renal hypoperfusion, as acute tubular necrosis (ATN) may develop in chronically hypertensive patients whose blood pressures are acutely reduced.[22] Less‐severe volume depletion is suggested by an orthostatic pulse increase of more than 30 beats/minute, measured 1 minute after standing. Orthostatic hypotension, defined as a drop in systolic pressure of more than 20 mm Hg after standing, is less helpful, as it occurs in 10% of normovolemic subjects.[23] Dry axillae and mucous membranes with a furrowed tongue are useful signs of volume depletion. Poor skin turgor and slow capillary refill have not been shown to be reliable signs of hypovolemia in adults. The neck veins are usually flat when volume contraction exists, though engorged neck veins in the setting of elevated right‐sided pressures from heart failure or pulmonary hypertension may obscure this sign. Similarly, pulmonary rales, ascites, and peripheral edema may confound the exam in patients with underlying heart failure and/or cirrhosis.

Flank tenderness or a bladder palpable or percussable above the pelvic brim suggests possible urinary tract obstruction. Prostate exam should be performed on all men with AKI and a bimanual pelvic exam considered in women with changes in usual voiding pattern or with suspected gynecologic disease. Postvoid residual can be assessed at the bedside with either straight catheterization or bladder scan where available.

Signs of systemic disease associated with intrinsic AKI include fever, skin and joint findings of connective tissue disease, a new or changing heart murmur, purpura, and petechiae. Cholesterol emboli, disrupted by interarterial catheterization (eg, cardiac catheterization, angiography), cardiac or aortic surgeries, or, rarely, by systemic anticoagulation can shower throughout the vasculature, causing organ dysfunction and local inflammation. Kidney injury due to atheroemboli often has a stuttering course and may be separated in time from the vascular procedure by days to weeks. Physical exam findings of atheroembolic disease include livedo reticularis, blue toes, purpura, painful skin nodules, and gangrene. Retinal examination may reveal atheroembolic emboli (Hollenhorst plaques).[24, 25]

LABORATORY TESTING

Initial testing in AKI aims to assess the severity of injury as well as the likely mechanism of the injury. Estimation of glomerular filtration rate (GFR) gives an approximate measure of the number of functioning nephrons and hence an overall measure of renal function. Mathematical estimates of GFR, however, assume a steady state, and AKI, by definition, is not a steady state. This makes GFR estimates based on plasma creatinine unreliable. A rising serum creatinine concentration indicates that the renal injury is persistent or worsening, whereas a stable or falling creatinine concentration suggests recovery. Interventions that expand the ECF (eg, volume resuscitation with normal saline) will dilute the plasma creatinine concentration and must be considered when interpreting a falling creatinine concentration. A daily rise in the serum creatinine concentration of more than 1 mg/dL nearly always implies a GFR of 10 mL/min. Any change in serum creatinine must be interpreted with the nonlinear relationship of GFR and serum creatinine in mind (Figure 1).[26]

Figure 1

The fractional excretion of sodium (FENa) has been used to differentiate prerenal azotemia from intrinsic renal injury in patients with oligoanuria. Specifically, an FENa of 1% implies a prerenal cause for the oliguric AKI, whereas if it is >1%, then intrinsic renal injury is more likely. Unfortunately, there are significant limitations to this laboratory measure.[27] The FENa may be low (1%) in any intrinsic process that causes tissue ischemia, such as vasculitis, acute glomerulonephritis, atheroembolic disease, or from intense vasoconstriction such as after IV contrast administration. Patients with severe heart failure or portal hypertension often have avid sodium retention, and can have a FENa 1% even in the setting of ATN. Alternatively, the FENa may be elevated (>1%) in prerenal patients on diuretics, with osmotic diuresis, or in the setting of aldosterone deficiency.

Examination of the urinalysis and urine sediment provides valuable information about the etiology of the AKI. Prerenal and postrenal AKI typically present with a bland urine, without evidence of blood, protein, or leukocyte esterase on urinalysis and few cells or hyaline casts in the sediment. The urinalysis typically has a high specific gravity in prerenal AKI, reflecting intact tubules producing a concentrated urine. An active urinary sediment suggests intrinsic renal injury that is either the mechanism of the current AKI or indicative of underlying CKD. ATN, the most common cause of intrinsic renal injury, often produces a dirty urinalysis with many epithelial cells and muddy brown granular and epithelial cell casts. The urine is generally isosthenuric (ie, specific gravity of 1.010) due to loss of tubular function. A urinalysis positive for heme pigment but without red cells on microscopic analysis suggests the presence of either myoglobin from rhabdomyolysis or hemoglobin from hemolysis. Acute glomerulonephritis disrupts the usual glomerular barrier to large proteins and red cells and results in proteinuria and hematuria. Red cells that weather the journey from the glomerulus through the nephron often become dysmorphic with Mickey Mouse ear blebs in their membrane or are bound together by Tamm‐Horsfall protein into red cell casts. Acute interstitial nephritis results in pyuria, proteinuria, and white cell casts. Urinary eosinophils are neither sensitive nor specific for interstitial nephritis and have little utility in its diagnosis.[28, 29]

Given the limitations of serum creatinine as a marker of renal injury, a number of new urinary biomarkers have been recognized over the past decade.[30, 31, 32] These molecules are normal constituents of renal tubular cells that are upregulated and released into the urine in response to renal injury. Early measurement of these biomarkers might allow for detection of AKI within hours of the insult. The 2 biomarkers with the most promise include kidney injury molecule‐1 (KIM‐1) and neutrophil gelatinase‐associated lipocalin (NGAL). KIM‐1 is expressed by proximal tubular cells, and its production is sharply upregulated in response to ischemic injury. NGAL is a protein expressed primarily in immune cells, but also by renal tubular cells. Urinary NGAL levels rapidly rise in response to renal ischemia, and return to baseline following resolution of the injury. Although these urinary biomarkers are promising, they have a relatively low (70%75%) sensitivity and specificity, and have not yet been adopted into routine clinical practice.[33]

IMAGING

Renal ultrasound is useful both in the assessment of AKI as well as in the investigation for underlying CKD. Patients with long‐standing kidney disease frequently have small, echogenic kidneys consistent with fibrosis and nephron loss, or markedly distorted renal architecture in cystic diseases. Hydronephrosis and/or hydroureter suggest an acute or chronic urinary tract obstruction. However, this may not be present in the setting of early obstruction or ureteric encasement. Doppler ultrasonography of the renal vasculature can assess patency when vascular obstruction is suspected. The use of computerized tomography, magnetic resonance imaging, or angiography may be helpful in selected clinical circumstances, but their use is often limited due to the potential risk of contrast nephrotoxicity. Nuclear renal scans use less radiation than computerized tomography and are a preferred imaging modality for pediatric patients. When volume status is uncertain, echocardiography to assess both inferior vena cava volume and change in volume with respiration may be helpful.

MANAGEMENT

The general principles for management of AKI are to limit further injury and prevent systemic complications. Management of the patient with AKI greatly depends on which category of AKI is suspected, namely prerenal, intrinsic renal injury, or a postrenal (obstructive) cause. If a prerenal etiology due to true ECF volume depletion is suspected, volume resuscitation to replace baseline and ongoing losses is imperative. Careful attention to intake and output as well as serial volume assessment should dictate the strategy for resuscitation. Hyperchloremic acidosis is an expected consequence of normal saline resuscitation but is irrelevant to clinical outcomes.[34] NSAIDs, antihypertensives, especially those that affect the angiotensin/aldosterone system, and diuretics should be discontinued. Ongoing hypotension despite volume resuscitation suggests the possibility of blood loss, infection, or autonomic nervous system dysfunction. If this occurs, the patient may need to be transferred to an intensive care unit for pressor support to keep the mean arterial pressure >70 mm Hg. When prerenal AKI from reduced effective circulating volume is suspected, as in decompensated heart failure or cirrhosis, management must be tailored to the underlying pathophysiology.

If judicious volume resuscitation produces no improvement in renal function or if oliguria develops, repeat urinalysis and urine microscopy should be considered to assess for intrinsic renal injury. Aggressive volume resuscitation in the face of oliguria will not speed recovery from the intrinsic injury and may cause signs or symptoms of volume overload. This could also potentially necessitate renal replacement therapy earlier than anticipated.

In patients where an obstructive etiology for the AKI is identified, the obstruction must be relieved as soon and as safely as possible. In this regard, a timely urologic consultation may be helpful in assuring that urethral and/or ureteral conduits are placed rapidly. Interventional radiology can also assist in those patients who need percutaneous nephrostomies for the relief of the obstruction. In many patients with obstructive nephropathy, a timely intervention will avoid the need for renal replacement therapy.

The suspected mechanism of injury influences the management of intrinsic AKI. The management of ATN is primarily supportive, paying close attention to optimizing volume status, correcting electrolyte abnormalities, avoiding further nephrotoxic agents, and adjusting medication doses to the low GFR present. Over the last several decades, multiple studies have explored treatment strategies for established ATN using various drugs and biologic agents. All have been uniformly disappointing.

When the trajectory of AKI is uncertain and the creatinine continues to rise, all medication dosing should be adjusted for GFR 10 mL/min. Antibiotics routinely will require dose reduction, but all current medications should be reviewed for risk of accumulation in renal failure. Because the half‐life of oral hypoglycemic medications is unpredictable in AKI, these medications should be discontinued and replaced with insulin. Vigilance for hypoglycemia is necessary, as renal clearance of insulin is also reduced. Narcotics such as morphine and oxycodone, which are renally cleared, can produce unwanted sedation and respiratory depression if not discontinued. Fentanyl, methadone, and hydromorphone are safer choices for controlling pain in a patient with AKI.[35] Gabapentin is regularly used to treat symptoms of neuropathic pain, but can produce encephalopathy and myoclonus if not dose reduced in renal failure.[36] Clinicians should weigh the risk of overdose with underdose for each medication, namely antibiotics in critically ill patients.

TIMING OF NEPHROLOGY CONSULTATION

The optimal timing for nephrology consultation in hospital‐acquired AKI is uncertain, though several studies have suggested better outcomes, including shorter length of stay and reduced mortality, with early consultation.[37, 38, 39] A renal consult is indicated when intrinsic ATN does not reverse in a timely fashion. Renal replacement therapy should be instituted to limit the systemic complications of prolonged AKI and to allow time for the renal injury to improve or resolve over time. If acute glomerulonephritis or interstitial nephritis is suspected, an urgent consultation may be required for consideration of biopsy, immunosuppression, and guidance for further management. Early consultation may help limit drug toxicities and volume overload in the setting of decreased renal clearance. Guidance on vascular access (eg, peripherally inserted central catheter placement) may prevent future complications with hemodialysis access if the patient ultimately develops end‐stage renal disease (ESRD).[40]

PREVENTION OF AKI

Most studies of AKI prevention have focused on clinical scenarios where the likelihood of ATN was substantial such as in vascular or open heart surgery, or with the use of intravenous contrast agents.[41, 42] This topic remains controversial, though generally supported strategies include judicious volume expansion, avoidance of hypotension, and, when using contrast, limiting the volume of contrast and using iso‐osmolar formulations. As recent studies have shown uncertain benefit, the role for pretreatment with n‐acetylcysteine remains uncertain. Many clinicians, however, continue to use it as a preventive strategy as there are few side effects with this medication.

TAKE HOME POINTS

• AKI is common in hospitalized patients, with pre‐renal azotemia being the dominant etiology in both community‐acquired and hospital‐acquired AKI.
• CKD is an important risk factor for AKI. AKI increases the long‐term risk of developing CKD and ESRD.
• The diagnosis of AKI hinges on detailed medical history, careful physical exam, and key laboratory parameters including the urinalysis and urinary sediment.
• The management of AKI is tailored to the likely mechanism of injury. Reconsideration of the likely etiology is imperative if AKI fails to respond to initial attempts to reverse or limit injury.
• Early renal consultation for AKI is indicated when the etiology remains uncertain, AKI persists despite initial management, or acute glomerulonephritis or interstitial nephritis are suspected.

Acute kidney injury (AKI) is a clinical syndrome broadly defined as an abrupt decline in renal function occurring over a period of hours to days resulting in the retention of nitrogenous and metabolic waste products. Although the initial clinical manifestation of AKI may be oliguria, urine volume can remain normal or even increase. Patients may be asymptomatic, especially early in the course of AKI. The diagnosis is often made in hospitalized patients when biochemical screening reveals a recent increase in serum creatinine and/or blood urea nitrogen concentrations, or when there is a dramatic decrease in urine output.

Older studies looking at the incidence of AKI in hospitalized patients are difficult to interpret due to variable definitions of AKI. Those based on administrative databases were limited by lack of clinical context and/or variation in coding for AKI.[1]

There is no universally accepted operational definition of AKI, and more than 30 different criteria have been employed in various clinical studies. Difficulty in defining AKI lies in the lag time in the rise and fall of the serum creatinine concentration with injury and recovery, the variability of oliguria, and in the heterogeneity of patterns of renal injury. Two classification systems that attempt to capture the spectrum of AKI are the RIFLE (Risk, Injury, Failure, Loss, End Stage) criteria and the AKIN (Acute Kidney Injury Network) criteria.[2, 3] The AKIN criteria parallel the risk, injury, and failure stages of the RIFLE criteria and are the most applicable to characterizing AKI in the hospital (Table 1). AKI is commonly classified by daily urine output as anuric (50 mL/day), oliguric (500 mL/day), or nonoliguric.

With a move toward standardized definitions, recent studies have shown a rising incidence of AKI in hospitalized patients.[4, 5, 6] According to these series, AKI develops in up to 7% of hospitalized patients and in about 30% of those admitted to intensive care units. In one study of consecutive hospital admissions, patients classified by the RIFLE criteria had a sharp rise in the rate of in‐hospital mortality whether they had no change or improvement in creatinine (4.4%), or fell into a risk (15.1%), injury (29.2%), or failure (41.1%) class.[7] The in‐hospital mortality of critically ill patients with AKI is higher than 50%. AKI increases length of stay and hospital costs, and affects the clinical course after discharge.[8, 9] Small increases in serum creatinine during an intensive care unit stay predict increased 10‐year mortality above a critical illness alone.[10]

Risk factors for AKI include advanced age, male gender, African American ethnicity, and diabetes mellitus.[11] The most important risk factor, however, is preexisting chronic kidney disease (CKD).[12] AKI and CKD are tightly linked, each increasing the risk of the other.[13, 14, 15] Preexisting renal insufficiency is a key predictor of postoperative AKI and poor surgical outcomes.[16, 17]

AKI AND CLINICAL CONTEXT

The causes of AKI can be broadly divided into 3 categories: prerenal azotemia (a disorder characterized by renal hypoperfusion in which renal parenchymal tissue integrity is preserved), intrinsic kidney injury with parenchymal tissue injury, and postrenal AKI (dysfunction due to acute obstruction of the urinary tract). Table 2 lists several clinical scenarios sorted into these 3 categories.[18] The general epidemiology of AKI varies based on whether it was acquired in the community or in a hospital setting. Prerenal azotemia accounts for the bulk of community‐acquired AKI, followed in lesser frequency by postrenal and intrinsic etiologies. Prerenal azotemia continues to be the major cause of hospital‐acquired AKI, but intrinsic kidney injury becomes more common.[5, 19]

MEDICAL HISTORY

The initial goal of history taking is to establish whether the patient has AKI rather than the acute discovery of a more chronic process. A recent serum creatinine measurement can be valuable in this regard. In some cases the clinician must make a presumptive diagnosis of AKI while simultaneously reviewing past medical history and family history to assess for underlying CKD. A diagnosis of AKI is more readily established when it occurs during a hospitalization through review of urine output and serial laboratory values.

Symptoms of poor oral intake as well as salt and fluid losses from diarrhea or vomiting suggest a prerenal etiology. Subjective symptoms of lightheadedness, visual clouding, and near‐syncope with standing also suggest volume depletion. Patients should be asked about recent nonsteroidal anti‐inflammatory drug (NSAID) use, as these agents can exacerbate renal hypoperfusion through loss of prostaglandin‐mediated afferent arteriole dilatation. Angiotensin‐converting enzyme inhibitors and angiotensin receptor blockers, especially when combined with diuretics, can generate a hypoperfusion state. Heart failure and liver disease regularly result in an expanded extracellular fluid (ECF) compartment yet a reduced effective arterial volume and predispose to renal hypoperfusion.

A history of decreased urine output or anuria suggests postrenal AKI, but its absence does not rule out urinary tract obstruction. Voiding symptoms such as urinary frequency, hesitancy, or incontinence also raise the possibility of obstructive uropathy. Flank pain and hematuria often accompany obstruction from nephrolithiasis.

Symptoms of fever, skin rash, arthralgias, sinusitis, and/or hemoptysis raise the possibility of glomerulonephritis from infection, collagen vascular disease, or vasculitis. Risk factors for viral hepatitis and human immunodeficiency virus are important to clarify as are systemic symptoms of autoimmunity (eg, dry eyes, dry mouth, eye pain/emnflammation, or visual changes). The recent start of any new medication, including NSAIDs, antibiotics, or proton‐pump inhibitors, raises the possibility of a drug‐induced interstitial nephritis.[20] Statins are direct myotoxins, and the risk of rhabdomyolysis with renal injury increases with dose. Patients may not associate intravenous (IV) contrast or phosphate‐containing bowel preparations (eg, Fleet Phospho Soda), with the development of AKI, thus the clinician must carefully review for recent exposures that could result in intrinsic renal injury.[21]

PHYSICAL EXAMINATION

Estimation of the ECF volume and effective arterial volume are central to assessing the likelihood of renal hypoperfusion. Overt hypotension is the strongest indicator of hypoperfusion, and a careful review of initial blood pressure prior to worsening of renal function can provide significant information. Normal blood pressure does not exclude renal hypoperfusion, as acute tubular necrosis (ATN) may develop in chronically hypertensive patients whose blood pressures are acutely reduced.[22] Less‐severe volume depletion is suggested by an orthostatic pulse increase of more than 30 beats/minute, measured 1 minute after standing. Orthostatic hypotension, defined as a drop in systolic pressure of more than 20 mm Hg after standing, is less helpful, as it occurs in 10% of normovolemic subjects.[23] Dry axillae and mucous membranes with a furrowed tongue are useful signs of volume depletion. Poor skin turgor and slow capillary refill have not been shown to be reliable signs of hypovolemia in adults. The neck veins are usually flat when volume contraction exists, though engorged neck veins in the setting of elevated right‐sided pressures from heart failure or pulmonary hypertension may obscure this sign. Similarly, pulmonary rales, ascites, and peripheral edema may confound the exam in patients with underlying heart failure and/or cirrhosis.

Flank tenderness or a bladder palpable or percussable above the pelvic brim suggests possible urinary tract obstruction. Prostate exam should be performed on all men with AKI and a bimanual pelvic exam considered in women with changes in usual voiding pattern or with suspected gynecologic disease. Postvoid residual can be assessed at the bedside with either straight catheterization or bladder scan where available.

Signs of systemic disease associated with intrinsic AKI include fever, skin and joint findings of connective tissue disease, a new or changing heart murmur, purpura, and petechiae. Cholesterol emboli, disrupted by interarterial catheterization (eg, cardiac catheterization, angiography), cardiac or aortic surgeries, or, rarely, by systemic anticoagulation can shower throughout the vasculature, causing organ dysfunction and local inflammation. Kidney injury due to atheroemboli often has a stuttering course and may be separated in time from the vascular procedure by days to weeks. Physical exam findings of atheroembolic disease include livedo reticularis, blue toes, purpura, painful skin nodules, and gangrene. Retinal examination may reveal atheroembolic emboli (Hollenhorst plaques).[24, 25]

LABORATORY TESTING

Initial testing in AKI aims to assess the severity of injury as well as the likely mechanism of the injury. Estimation of glomerular filtration rate (GFR) gives an approximate measure of the number of functioning nephrons and hence an overall measure of renal function. Mathematical estimates of GFR, however, assume a steady state, and AKI, by definition, is not a steady state. This makes GFR estimates based on plasma creatinine unreliable. A rising serum creatinine concentration indicates that the renal injury is persistent or worsening, whereas a stable or falling creatinine concentration suggests recovery. Interventions that expand the ECF (eg, volume resuscitation with normal saline) will dilute the plasma creatinine concentration and must be considered when interpreting a falling creatinine concentration. A daily rise in the serum creatinine concentration of more than 1 mg/dL nearly always implies a GFR of 10 mL/min. Any change in serum creatinine must be interpreted with the nonlinear relationship of GFR and serum creatinine in mind (Figure 1).[26]

Figure 1

The fractional excretion of sodium (FENa) has been used to differentiate prerenal azotemia from intrinsic renal injury in patients with oligoanuria. Specifically, an FENa of 1% implies a prerenal cause for the oliguric AKI, whereas if it is >1%, then intrinsic renal injury is more likely. Unfortunately, there are significant limitations to this laboratory measure.[27] The FENa may be low (1%) in any intrinsic process that causes tissue ischemia, such as vasculitis, acute glomerulonephritis, atheroembolic disease, or from intense vasoconstriction such as after IV contrast administration. Patients with severe heart failure or portal hypertension often have avid sodium retention, and can have a FENa 1% even in the setting of ATN. Alternatively, the FENa may be elevated (>1%) in prerenal patients on diuretics, with osmotic diuresis, or in the setting of aldosterone deficiency.

Examination of the urinalysis and urine sediment provides valuable information about the etiology of the AKI. Prerenal and postrenal AKI typically present with a bland urine, without evidence of blood, protein, or leukocyte esterase on urinalysis and few cells or hyaline casts in the sediment. The urinalysis typically has a high specific gravity in prerenal AKI, reflecting intact tubules producing a concentrated urine. An active urinary sediment suggests intrinsic renal injury that is either the mechanism of the current AKI or indicative of underlying CKD. ATN, the most common cause of intrinsic renal injury, often produces a dirty urinalysis with many epithelial cells and muddy brown granular and epithelial cell casts. The urine is generally isosthenuric (ie, specific gravity of 1.010) due to loss of tubular function. A urinalysis positive for heme pigment but without red cells on microscopic analysis suggests the presence of either myoglobin from rhabdomyolysis or hemoglobin from hemolysis. Acute glomerulonephritis disrupts the usual glomerular barrier to large proteins and red cells and results in proteinuria and hematuria. Red cells that weather the journey from the glomerulus through the nephron often become dysmorphic with Mickey Mouse ear blebs in their membrane or are bound together by Tamm‐Horsfall protein into red cell casts. Acute interstitial nephritis results in pyuria, proteinuria, and white cell casts. Urinary eosinophils are neither sensitive nor specific for interstitial nephritis and have little utility in its diagnosis.[28, 29]

Given the limitations of serum creatinine as a marker of renal injury, a number of new urinary biomarkers have been recognized over the past decade.[30, 31, 32] These molecules are normal constituents of renal tubular cells that are upregulated and released into the urine in response to renal injury. Early measurement of these biomarkers might allow for detection of AKI within hours of the insult. The 2 biomarkers with the most promise include kidney injury molecule‐1 (KIM‐1) and neutrophil gelatinase‐associated lipocalin (NGAL). KIM‐1 is expressed by proximal tubular cells, and its production is sharply upregulated in response to ischemic injury. NGAL is a protein expressed primarily in immune cells, but also by renal tubular cells. Urinary NGAL levels rapidly rise in response to renal ischemia, and return to baseline following resolution of the injury. Although these urinary biomarkers are promising, they have a relatively low (70%75%) sensitivity and specificity, and have not yet been adopted into routine clinical practice.[33]

IMAGING

Renal ultrasound is useful both in the assessment of AKI as well as in the investigation for underlying CKD. Patients with long‐standing kidney disease frequently have small, echogenic kidneys consistent with fibrosis and nephron loss, or markedly distorted renal architecture in cystic diseases. Hydronephrosis and/or hydroureter suggest an acute or chronic urinary tract obstruction. However, this may not be present in the setting of early obstruction or ureteric encasement. Doppler ultrasonography of the renal vasculature can assess patency when vascular obstruction is suspected. The use of computerized tomography, magnetic resonance imaging, or angiography may be helpful in selected clinical circumstances, but their use is often limited due to the potential risk of contrast nephrotoxicity. Nuclear renal scans use less radiation than computerized tomography and are a preferred imaging modality for pediatric patients. When volume status is uncertain, echocardiography to assess both inferior vena cava volume and change in volume with respiration may be helpful.

MANAGEMENT

The general principles for management of AKI are to limit further injury and prevent systemic complications. Management of the patient with AKI greatly depends on which category of AKI is suspected, namely prerenal, intrinsic renal injury, or a postrenal (obstructive) cause. If a prerenal etiology due to true ECF volume depletion is suspected, volume resuscitation to replace baseline and ongoing losses is imperative. Careful attention to intake and output as well as serial volume assessment should dictate the strategy for resuscitation. Hyperchloremic acidosis is an expected consequence of normal saline resuscitation but is irrelevant to clinical outcomes.[34] NSAIDs, antihypertensives, especially those that affect the angiotensin/aldosterone system, and diuretics should be discontinued. Ongoing hypotension despite volume resuscitation suggests the possibility of blood loss, infection, or autonomic nervous system dysfunction. If this occurs, the patient may need to be transferred to an intensive care unit for pressor support to keep the mean arterial pressure >70 mm Hg. When prerenal AKI from reduced effective circulating volume is suspected, as in decompensated heart failure or cirrhosis, management must be tailored to the underlying pathophysiology.

If judicious volume resuscitation produces no improvement in renal function or if oliguria develops, repeat urinalysis and urine microscopy should be considered to assess for intrinsic renal injury. Aggressive volume resuscitation in the face of oliguria will not speed recovery from the intrinsic injury and may cause signs or symptoms of volume overload. This could also potentially necessitate renal replacement therapy earlier than anticipated.

In patients where an obstructive etiology for the AKI is identified, the obstruction must be relieved as soon and as safely as possible. In this regard, a timely urologic consultation may be helpful in assuring that urethral and/or ureteral conduits are placed rapidly. Interventional radiology can also assist in those patients who need percutaneous nephrostomies for the relief of the obstruction. In many patients with obstructive nephropathy, a timely intervention will avoid the need for renal replacement therapy.

The suspected mechanism of injury influences the management of intrinsic AKI. The management of ATN is primarily supportive, paying close attention to optimizing volume status, correcting electrolyte abnormalities, avoiding further nephrotoxic agents, and adjusting medication doses to the low GFR present. Over the last several decades, multiple studies have explored treatment strategies for established ATN using various drugs and biologic agents. All have been uniformly disappointing.

When the trajectory of AKI is uncertain and the creatinine continues to rise, all medication dosing should be adjusted for GFR 10 mL/min. Antibiotics routinely will require dose reduction, but all current medications should be reviewed for risk of accumulation in renal failure. Because the half‐life of oral hypoglycemic medications is unpredictable in AKI, these medications should be discontinued and replaced with insulin. Vigilance for hypoglycemia is necessary, as renal clearance of insulin is also reduced. Narcotics such as morphine and oxycodone, which are renally cleared, can produce unwanted sedation and respiratory depression if not discontinued. Fentanyl, methadone, and hydromorphone are safer choices for controlling pain in a patient with AKI.[35] Gabapentin is regularly used to treat symptoms of neuropathic pain, but can produce encephalopathy and myoclonus if not dose reduced in renal failure.[36] Clinicians should weigh the risk of overdose with underdose for each medication, namely antibiotics in critically ill patients.

TIMING OF NEPHROLOGY CONSULTATION

The optimal timing for nephrology consultation in hospital‐acquired AKI is uncertain, though several studies have suggested better outcomes, including shorter length of stay and reduced mortality, with early consultation.[37, 38, 39] A renal consult is indicated when intrinsic ATN does not reverse in a timely fashion. Renal replacement therapy should be instituted to limit the systemic complications of prolonged AKI and to allow time for the renal injury to improve or resolve over time. If acute glomerulonephritis or interstitial nephritis is suspected, an urgent consultation may be required for consideration of biopsy, immunosuppression, and guidance for further management. Early consultation may help limit drug toxicities and volume overload in the setting of decreased renal clearance. Guidance on vascular access (eg, peripherally inserted central catheter placement) may prevent future complications with hemodialysis access if the patient ultimately develops end‐stage renal disease (ESRD).[40]

PREVENTION OF AKI

Most studies of AKI prevention have focused on clinical scenarios where the likelihood of ATN was substantial such as in vascular or open heart surgery, or with the use of intravenous contrast agents.[41, 42] This topic remains controversial, though generally supported strategies include judicious volume expansion, avoidance of hypotension, and, when using contrast, limiting the volume of contrast and using iso‐osmolar formulations. As recent studies have shown uncertain benefit, the role for pretreatment with n‐acetylcysteine remains uncertain. Many clinicians, however, continue to use it as a preventive strategy as there are few side effects with this medication.

TAKE HOME POINTS

• AKI is common in hospitalized patients, with pre‐renal azotemia being the dominant etiology in both community‐acquired and hospital‐acquired AKI.
• CKD is an important risk factor for AKI. AKI increases the long‐term risk of developing CKD and ESRD.
• The diagnosis of AKI hinges on detailed medical history, careful physical exam, and key laboratory parameters including the urinalysis and urinary sediment.
• The management of AKI is tailored to the likely mechanism of injury. Reconsideration of the likely etiology is imperative if AKI fails to respond to initial attempts to reverse or limit injury.
• Early renal consultation for AKI is indicated when the etiology remains uncertain, AKI persists despite initial management, or acute glomerulonephritis or interstitial nephritis are suspected.

Acute kidney injury (AKI) is a clinical syndrome broadly defined as an abrupt decline in renal function occurring over a period of hours to days resulting in the retention of nitrogenous and metabolic waste products. Although the initial clinical manifestation of AKI may be oliguria, urine volume can remain normal or even increase. Patients may be asymptomatic, especially early in the course of AKI. The diagnosis is often made in hospitalized patients when biochemical screening reveals a recent increase in serum creatinine and/or blood urea nitrogen concentrations, or when there is a dramatic decrease in urine output.

Older studies looking at the incidence of AKI in hospitalized patients are difficult to interpret due to variable definitions of AKI. Those based on administrative databases were limited by lack of clinical context and/or variation in coding for AKI.[1]

There is no universally accepted operational definition of AKI, and more than 30 different criteria have been employed in various clinical studies. Difficulty in defining AKI lies in the lag time in the rise and fall of the serum creatinine concentration with injury and recovery, the variability of oliguria, and in the heterogeneity of patterns of renal injury. Two classification systems that attempt to capture the spectrum of AKI are the RIFLE (Risk, Injury, Failure, Loss, End Stage) criteria and the AKIN (Acute Kidney Injury Network) criteria.[2, 3] The AKIN criteria parallel the risk, injury, and failure stages of the RIFLE criteria and are the most applicable to characterizing AKI in the hospital (Table 1). AKI is commonly classified by daily urine output as anuric (50 mL/day), oliguric (500 mL/day), or nonoliguric.

With a move toward standardized definitions, recent studies have shown a rising incidence of AKI in hospitalized patients.[4, 5, 6] According to these series, AKI develops in up to 7% of hospitalized patients and in about 30% of those admitted to intensive care units. In one study of consecutive hospital admissions, patients classified by the RIFLE criteria had a sharp rise in the rate of in‐hospital mortality whether they had no change or improvement in creatinine (4.4%), or fell into a risk (15.1%), injury (29.2%), or failure (41.1%) class.[7] The in‐hospital mortality of critically ill patients with AKI is higher than 50%. AKI increases length of stay and hospital costs, and affects the clinical course after discharge.[8, 9] Small increases in serum creatinine during an intensive care unit stay predict increased 10‐year mortality above a critical illness alone.[10]

Risk factors for AKI include advanced age, male gender, African American ethnicity, and diabetes mellitus.[11] The most important risk factor, however, is preexisting chronic kidney disease (CKD).[12] AKI and CKD are tightly linked, each increasing the risk of the other.[13, 14, 15] Preexisting renal insufficiency is a key predictor of postoperative AKI and poor surgical outcomes.[16, 17]

AKI AND CLINICAL CONTEXT

The causes of AKI can be broadly divided into 3 categories: prerenal azotemia (a disorder characterized by renal hypoperfusion in which renal parenchymal tissue integrity is preserved), intrinsic kidney injury with parenchymal tissue injury, and postrenal AKI (dysfunction due to acute obstruction of the urinary tract). Table 2 lists several clinical scenarios sorted into these 3 categories.[18] The general epidemiology of AKI varies based on whether it was acquired in the community or in a hospital setting. Prerenal azotemia accounts for the bulk of community‐acquired AKI, followed in lesser frequency by postrenal and intrinsic etiologies. Prerenal azotemia continues to be the major cause of hospital‐acquired AKI, but intrinsic kidney injury becomes more common.[5, 19]

MEDICAL HISTORY

The initial goal of history taking is to establish whether the patient has AKI rather than the acute discovery of a more chronic process. A recent serum creatinine measurement can be valuable in this regard. In some cases the clinician must make a presumptive diagnosis of AKI while simultaneously reviewing past medical history and family history to assess for underlying CKD. A diagnosis of AKI is more readily established when it occurs during a hospitalization through review of urine output and serial laboratory values.

Symptoms of poor oral intake as well as salt and fluid losses from diarrhea or vomiting suggest a prerenal etiology. Subjective symptoms of lightheadedness, visual clouding, and near‐syncope with standing also suggest volume depletion. Patients should be asked about recent nonsteroidal anti‐inflammatory drug (NSAID) use, as these agents can exacerbate renal hypoperfusion through loss of prostaglandin‐mediated afferent arteriole dilatation. Angiotensin‐converting enzyme inhibitors and angiotensin receptor blockers, especially when combined with diuretics, can generate a hypoperfusion state. Heart failure and liver disease regularly result in an expanded extracellular fluid (ECF) compartment yet a reduced effective arterial volume and predispose to renal hypoperfusion.

A history of decreased urine output or anuria suggests postrenal AKI, but its absence does not rule out urinary tract obstruction. Voiding symptoms such as urinary frequency, hesitancy, or incontinence also raise the possibility of obstructive uropathy. Flank pain and hematuria often accompany obstruction from nephrolithiasis.

Symptoms of fever, skin rash, arthralgias, sinusitis, and/or hemoptysis raise the possibility of glomerulonephritis from infection, collagen vascular disease, or vasculitis. Risk factors for viral hepatitis and human immunodeficiency virus are important to clarify as are systemic symptoms of autoimmunity (eg, dry eyes, dry mouth, eye pain/emnflammation, or visual changes). The recent start of any new medication, including NSAIDs, antibiotics, or proton‐pump inhibitors, raises the possibility of a drug‐induced interstitial nephritis.[20] Statins are direct myotoxins, and the risk of rhabdomyolysis with renal injury increases with dose. Patients may not associate intravenous (IV) contrast or phosphate‐containing bowel preparations (eg, Fleet Phospho Soda), with the development of AKI, thus the clinician must carefully review for recent exposures that could result in intrinsic renal injury.[21]

PHYSICAL EXAMINATION

Estimation of the ECF volume and effective arterial volume are central to assessing the likelihood of renal hypoperfusion. Overt hypotension is the strongest indicator of hypoperfusion, and a careful review of initial blood pressure prior to worsening of renal function can provide significant information. Normal blood pressure does not exclude renal hypoperfusion, as acute tubular necrosis (ATN) may develop in chronically hypertensive patients whose blood pressures are acutely reduced.[22] Less‐severe volume depletion is suggested by an orthostatic pulse increase of more than 30 beats/minute, measured 1 minute after standing. Orthostatic hypotension, defined as a drop in systolic pressure of more than 20 mm Hg after standing, is less helpful, as it occurs in 10% of normovolemic subjects.[23] Dry axillae and mucous membranes with a furrowed tongue are useful signs of volume depletion. Poor skin turgor and slow capillary refill have not been shown to be reliable signs of hypovolemia in adults. The neck veins are usually flat when volume contraction exists, though engorged neck veins in the setting of elevated right‐sided pressures from heart failure or pulmonary hypertension may obscure this sign. Similarly, pulmonary rales, ascites, and peripheral edema may confound the exam in patients with underlying heart failure and/or cirrhosis.

Flank tenderness or a bladder palpable or percussable above the pelvic brim suggests possible urinary tract obstruction. Prostate exam should be performed on all men with AKI and a bimanual pelvic exam considered in women with changes in usual voiding pattern or with suspected gynecologic disease. Postvoid residual can be assessed at the bedside with either straight catheterization or bladder scan where available.

Signs of systemic disease associated with intrinsic AKI include fever, skin and joint findings of connective tissue disease, a new or changing heart murmur, purpura, and petechiae. Cholesterol emboli, disrupted by interarterial catheterization (eg, cardiac catheterization, angiography), cardiac or aortic surgeries, or, rarely, by systemic anticoagulation can shower throughout the vasculature, causing organ dysfunction and local inflammation. Kidney injury due to atheroemboli often has a stuttering course and may be separated in time from the vascular procedure by days to weeks. Physical exam findings of atheroembolic disease include livedo reticularis, blue toes, purpura, painful skin nodules, and gangrene. Retinal examination may reveal atheroembolic emboli (Hollenhorst plaques).[24, 25]

LABORATORY TESTING

Initial testing in AKI aims to assess the severity of injury as well as the likely mechanism of the injury. Estimation of glomerular filtration rate (GFR) gives an approximate measure of the number of functioning nephrons and hence an overall measure of renal function. Mathematical estimates of GFR, however, assume a steady state, and AKI, by definition, is not a steady state. This makes GFR estimates based on plasma creatinine unreliable. A rising serum creatinine concentration indicates that the renal injury is persistent or worsening, whereas a stable or falling creatinine concentration suggests recovery. Interventions that expand the ECF (eg, volume resuscitation with normal saline) will dilute the plasma creatinine concentration and must be considered when interpreting a falling creatinine concentration. A daily rise in the serum creatinine concentration of more than 1 mg/dL nearly always implies a GFR of 10 mL/min. Any change in serum creatinine must be interpreted with the nonlinear relationship of GFR and serum creatinine in mind (Figure 1).[26]

Figure 1

The fractional excretion of sodium (FENa) has been used to differentiate prerenal azotemia from intrinsic renal injury in patients with oligoanuria. Specifically, an FENa of 1% implies a prerenal cause for the oliguric AKI, whereas if it is >1%, then intrinsic renal injury is more likely. Unfortunately, there are significant limitations to this laboratory measure.[27] The FENa may be low (1%) in any intrinsic process that causes tissue ischemia, such as vasculitis, acute glomerulonephritis, atheroembolic disease, or from intense vasoconstriction such as after IV contrast administration. Patients with severe heart failure or portal hypertension often have avid sodium retention, and can have a FENa 1% even in the setting of ATN. Alternatively, the FENa may be elevated (>1%) in prerenal patients on diuretics, with osmotic diuresis, or in the setting of aldosterone deficiency.

Examination of the urinalysis and urine sediment provides valuable information about the etiology of the AKI. Prerenal and postrenal AKI typically present with a bland urine, without evidence of blood, protein, or leukocyte esterase on urinalysis and few cells or hyaline casts in the sediment. The urinalysis typically has a high specific gravity in prerenal AKI, reflecting intact tubules producing a concentrated urine. An active urinary sediment suggests intrinsic renal injury that is either the mechanism of the current AKI or indicative of underlying CKD. ATN, the most common cause of intrinsic renal injury, often produces a dirty urinalysis with many epithelial cells and muddy brown granular and epithelial cell casts. The urine is generally isosthenuric (ie, specific gravity of 1.010) due to loss of tubular function. A urinalysis positive for heme pigment but without red cells on microscopic analysis suggests the presence of either myoglobin from rhabdomyolysis or hemoglobin from hemolysis. Acute glomerulonephritis disrupts the usual glomerular barrier to large proteins and red cells and results in proteinuria and hematuria. Red cells that weather the journey from the glomerulus through the nephron often become dysmorphic with Mickey Mouse ear blebs in their membrane or are bound together by Tamm‐Horsfall protein into red cell casts. Acute interstitial nephritis results in pyuria, proteinuria, and white cell casts. Urinary eosinophils are neither sensitive nor specific for interstitial nephritis and have little utility in its diagnosis.[28, 29]

Given the limitations of serum creatinine as a marker of renal injury, a number of new urinary biomarkers have been recognized over the past decade.[30, 31, 32] These molecules are normal constituents of renal tubular cells that are upregulated and released into the urine in response to renal injury. Early measurement of these biomarkers might allow for detection of AKI within hours of the insult. The 2 biomarkers with the most promise include kidney injury molecule‐1 (KIM‐1) and neutrophil gelatinase‐associated lipocalin (NGAL). KIM‐1 is expressed by proximal tubular cells, and its production is sharply upregulated in response to ischemic injury. NGAL is a protein expressed primarily in immune cells, but also by renal tubular cells. Urinary NGAL levels rapidly rise in response to renal ischemia, and return to baseline following resolution of the injury. Although these urinary biomarkers are promising, they have a relatively low (70%75%) sensitivity and specificity, and have not yet been adopted into routine clinical practice.[33]

IMAGING

Renal ultrasound is useful both in the assessment of AKI as well as in the investigation for underlying CKD. Patients with long‐standing kidney disease frequently have small, echogenic kidneys consistent with fibrosis and nephron loss, or markedly distorted renal architecture in cystic diseases. Hydronephrosis and/or hydroureter suggest an acute or chronic urinary tract obstruction. However, this may not be present in the setting of early obstruction or ureteric encasement. Doppler ultrasonography of the renal vasculature can assess patency when vascular obstruction is suspected. The use of computerized tomography, magnetic resonance imaging, or angiography may be helpful in selected clinical circumstances, but their use is often limited due to the potential risk of contrast nephrotoxicity. Nuclear renal scans use less radiation than computerized tomography and are a preferred imaging modality for pediatric patients. When volume status is uncertain, echocardiography to assess both inferior vena cava volume and change in volume with respiration may be helpful.

MANAGEMENT

The general principles for management of AKI are to limit further injury and prevent systemic complications. Management of the patient with AKI greatly depends on which category of AKI is suspected, namely prerenal, intrinsic renal injury, or a postrenal (obstructive) cause. If a prerenal etiology due to true ECF volume depletion is suspected, volume resuscitation to replace baseline and ongoing losses is imperative. Careful attention to intake and output as well as serial volume assessment should dictate the strategy for resuscitation. Hyperchloremic acidosis is an expected consequence of normal saline resuscitation but is irrelevant to clinical outcomes.[34] NSAIDs, antihypertensives, especially those that affect the angiotensin/aldosterone system, and diuretics should be discontinued. Ongoing hypotension despite volume resuscitation suggests the possibility of blood loss, infection, or autonomic nervous system dysfunction. If this occurs, the patient may need to be transferred to an intensive care unit for pressor support to keep the mean arterial pressure >70 mm Hg. When prerenal AKI from reduced effective circulating volume is suspected, as in decompensated heart failure or cirrhosis, management must be tailored to the underlying pathophysiology.

If judicious volume resuscitation produces no improvement in renal function or if oliguria develops, repeat urinalysis and urine microscopy should be considered to assess for intrinsic renal injury. Aggressive volume resuscitation in the face of oliguria will not speed recovery from the intrinsic injury and may cause signs or symptoms of volume overload. This could also potentially necessitate renal replacement therapy earlier than anticipated.

In patients where an obstructive etiology for the AKI is identified, the obstruction must be relieved as soon and as safely as possible. In this regard, a timely urologic consultation may be helpful in assuring that urethral and/or ureteral conduits are placed rapidly. Interventional radiology can also assist in those patients who need percutaneous nephrostomies for the relief of the obstruction. In many patients with obstructive nephropathy, a timely intervention will avoid the need for renal replacement therapy.

The suspected mechanism of injury influences the management of intrinsic AKI. The management of ATN is primarily supportive, paying close attention to optimizing volume status, correcting electrolyte abnormalities, avoiding further nephrotoxic agents, and adjusting medication doses to the low GFR present. Over the last several decades, multiple studies have explored treatment strategies for established ATN using various drugs and biologic agents. All have been uniformly disappointing.

When the trajectory of AKI is uncertain and the creatinine continues to rise, all medication dosing should be adjusted for GFR 10 mL/min. Antibiotics routinely will require dose reduction, but all current medications should be reviewed for risk of accumulation in renal failure. Because the half‐life of oral hypoglycemic medications is unpredictable in AKI, these medications should be discontinued and replaced with insulin. Vigilance for hypoglycemia is necessary, as renal clearance of insulin is also reduced. Narcotics such as morphine and oxycodone, which are renally cleared, can produce unwanted sedation and respiratory depression if not discontinued. Fentanyl, methadone, and hydromorphone are safer choices for controlling pain in a patient with AKI.[35] Gabapentin is regularly used to treat symptoms of neuropathic pain, but can produce encephalopathy and myoclonus if not dose reduced in renal failure.[36] Clinicians should weigh the risk of overdose with underdose for each medication, namely antibiotics in critically ill patients.

TIMING OF NEPHROLOGY CONSULTATION

The optimal timing for nephrology consultation in hospital‐acquired AKI is uncertain, though several studies have suggested better outcomes, including shorter length of stay and reduced mortality, with early consultation.[37, 38, 39] A renal consult is indicated when intrinsic ATN does not reverse in a timely fashion. Renal replacement therapy should be instituted to limit the systemic complications of prolonged AKI and to allow time for the renal injury to improve or resolve over time. If acute glomerulonephritis or interstitial nephritis is suspected, an urgent consultation may be required for consideration of biopsy, immunosuppression, and guidance for further management. Early consultation may help limit drug toxicities and volume overload in the setting of decreased renal clearance. Guidance on vascular access (eg, peripherally inserted central catheter placement) may prevent future complications with hemodialysis access if the patient ultimately develops end‐stage renal disease (ESRD).[40]

PREVENTION OF AKI

Most studies of AKI prevention have focused on clinical scenarios where the likelihood of ATN was substantial such as in vascular or open heart surgery, or with the use of intravenous contrast agents.[41, 42] This topic remains controversial, though generally supported strategies include judicious volume expansion, avoidance of hypotension, and, when using contrast, limiting the volume of contrast and using iso‐osmolar formulations. As recent studies have shown uncertain benefit, the role for pretreatment with n‐acetylcysteine remains uncertain. Many clinicians, however, continue to use it as a preventive strategy as there are few side effects with this medication.

TAKE HOME POINTS

• AKI is common in hospitalized patients, with pre‐renal azotemia being the dominant etiology in both community‐acquired and hospital‐acquired AKI.
• CKD is an important risk factor for AKI. AKI increases the long‐term risk of developing CKD and ESRD.
• The diagnosis of AKI hinges on detailed medical history, careful physical exam, and key laboratory parameters including the urinalysis and urinary sediment.
• The management of AKI is tailored to the likely mechanism of injury. Reconsideration of the likely etiology is imperative if AKI fails to respond to initial attempts to reverse or limit injury.
• Early renal consultation for AKI is indicated when the etiology remains uncertain, AKI persists despite initial management, or acute glomerulonephritis or interstitial nephritis are suspected.