The Institute of Medicine reports To Err is Human and Crossing the Quality Chasm have drawn great attention to quality improvement and patient safety in the hospital setting.13 With the growth of the hospitalist field over the past several years,4 there has been increasing discussion about the importance of assuring quality of care, and some have argued that improving health care quality and reducing avoidable errors may be among the hospitalist's most important functions.5 Most discussions about the quality of hospital care have concerned the inpatient stay itself. However, the growth of hospital medicine, with its inherent discontinuity between inpatient and outpatient physicians, has intensified interest in the transition period from hospital discharge until first outpatient appointment.

At discharge, physicians may prescribe medications, order home health services, and arrange follow‐up appointments. It is often assumed a patient will remain stable after discharge and will follow up at the outpatient physician's office. Previous research has shown there may be problems with these assumptions. A patient may not understand the postdischarge treatment plan as well as the physician thinks.6 A recent study found that adverse events after discharge were common and often preventable.7 A follow‐up study confirmed that approximately 25% of patients had an adverse event after hospital discharge and that most adverse events caused symptoms but did not result in an emergency department visit, hospitalization, or death.8 Another study also found the prevalence of medical errors following hospitalization was high because of the discontinuity of care from the inpatient to the outpatient setting.9 These errors resulted in an increased rate of rehospitalization.

Telephone follow‐up may be a useful method of bridging the gap in care between discharge and the first outpatient appointment.10, 11 In most previous studies it was 2 or 3 weeks after discharge before patients were contacted or their records studied. By this point, patients who had done poorly may already have been readmitted or sought care at alternative locations. In one small study, pharmacists found that 12 of 79 patients (15%) contacted by telephone within 2 days of discharge12 had symptoms there were new or had worsened since discharge. The purpose of the present investigation was to extend these previous findings through a large multicenter study of how frequently patients had new or worsened symptoms within several days of discharge.

METHODS

RESULTS

During the study period, 48,236 patients were discharged to their homes from an acute care hospital. The IPC call center successfully contacted 16,135 patients after discharge, of whom 368 patients (2.3%) were excluded because of incomplete answers, leaving 15,767 as the valid study population (effective response rate = 32.4%). Of these, 98.9% were contacted within the first 5 days. The primary reasons for nonresponse or noninclusion in the present analysis were no answer after 2 attempts (52%) and missing or incorrect phone numbers (16%). If there was an answering machine, a message was left for the patient to call back. Those who called back accounted for fewer than 1% of all the patients.

A comparison of participants versus nonparticipants is shown in Table 1. The mean age of surveyed patients was 60.1 years, and 57% were female. The most common categories of insurance coverage were Medicare and HMO. The inpatient severity of illness of most patients was minor to moderate. Self‐reported health status was normally distributed, with the greatest percentage of patients rating their health as fair or good. On average, nonparticipants were younger than participants, more likely to be male, had a different pattern of health insurance, and a slightly lower severity of illness. The top 10 DRGs were the same for the respondents and nonrespondents, and the order of these 10 diagnoses was very similar.

Of the 15,767 patients contacted, 11.9% (N = 1876) reported symptoms that were new or had worsened since leaving the hospital. Sixty‐four percent of these patients had new symptoms, and 36% had worsening symptoms. These two groups were combined for analysis in this study because for both groups, identification and action are important. Of the patients with new or worse symptoms, 37% required no assistance from the nurse because they had already notified a doctor and/or were doing something about it. The other 63% either had not notified their doctor or had concerns about their signs and symptoms. The most common action taken by the nurse was patient education regarding the symptoms. Of those requiring nurse intervention in addition to education, the most common intervention was to contact the patient's primary care provider or specialist about the patient's symptoms, followed by contacting the hospitalist. In 72% of nurse interventions, the patient's primary care physician or a specialist was contacted about the new or worsened symptoms. Other interventions included contacting the physician's office to obtain a prescription for a medication for the symptom, to get an appointment for the patient, or to reschedule an appointment to be earlier. A referral to an emergency room or urgent care center was given to 4% of patients.

Mean age of the patients with new or worsened symptoms was 60.5 years. The age distribution of symptomatic and asymptomatic patients was not significantly different, whether comparing by mean, median, or decades. Table 2 illustrates factors associated with the increased rate of new or worsened symptoms. Women were more likely than men to report symptoms (13.0% vs. 10.3%, P < .0001). As health status worsened, the percentage of patients with new or worsened symptoms increased (P < .0001), as it did with increased inpatient SOI (P < .0001). There was no correlation between self‐rated health status and SOI score based on DRG score, suggesting they measured different parameters. Table 3 lists the percentages of patients reporting new or worsened symptoms for the most common DRGs. The only significant distinction was that patients discharged with a DRG of chest pain were less likely to report symptoms than were all patients.

The symptoms the patients reported are categorized in Table 4 without distinction as to whether they are primary or secondary. Gastrointestinal symptoms were the most common category of symptoms, followed by general symptoms, cardiovascular symptoms, and pain. The most common symptoms reported were fatigue/weakness, nausea/vomiting, and edema.

The call center assessed whether the patient had difficulty making a follow‐up appointment and whether an appointment was scheduled within 2 weeks of discharge. Patients with new or worsening symptoms were only minimally more likely to have scheduled follow‐up (61.0% vs. 58.4% for patients not reporting new or worsening symptoms, P < .05). Symptomatic patients had a higher prevalence of medication issues, defined as not picking up their prescriptions or not understanding how to take their medication (22.2% vs. 6.8% for asymptomatic patients; P < .0001). Likewise, the prevalence of symptomatic patients having problems receiving scheduled home health care services was 5.8%, compared with a prevalence of 3.6% for asymptomatic patients (P < .0001).

DISCUSSION

Enhancing the quality of care provided by hospitalists means not only improving care during hospitalization but also assuring patient stability between discharge and outpatient follow‐up. As part of efforts to improve transition management, the call center at IPC attempted to contact all patients discharged home within several days of discharge. Of 15,767 patients surveyed between May 1, 2003, and October 31, 2003, 11.9% (N = 1876) had new or worsening symptoms since leaving the hospital only 2 or 3 days earlier. We had hypothesized that older patients might be more symptomatic than younger ones, but found no difference in the prevalence of new or worsening symptoms based on age. Women were more likely to be symptomatic than men.

We defined appropriate postdischarge follow‐up as having an appointment with an outpatient physician within 2 weeks. A previous study of psychiatric patients documented that keeping a follow‐up appointment significantly reduces the risk of rehospitalization.17 Similar data do not exist for medical patients. Our data demonstrated that symptomatic patients were only minimally more likely to have made a follow‐up appointment with their outpatient physician within the first 2 weeks than were those patients who were not symptomatic. As part of patient education at discharge, clinicians routinely counsel patients to call their outpatient physician should they experience new or worsening symptoms once at home. Inpatient physicians may assume this recommendation provides a safety net for the patients should they develop problems after discharge. However, our finding that almost 40% of patients with new or worsening symptoms within 2‐3 days of discharge had not made a follow‐up appointment with their physician suggests many patients fall through this safety net. Although there was a slight statistically significant difference between the groups, this difference was not clinically significant. One potential limitation of our data is that we did not examine whether there was a correlation between the day of the week that a patient was discharged and inability to make a follow‐up appointment.

As part of the survey script (see Appendix), we inquired whether patients were able to pick up their prescriptions and whether they understood how to take their medication. A high percentage of patients in our study reported having one of these medication issues in the first several days following hospital discharge, providing an opportunity for early intervention and prevention of medical error. Forster and others have demonstrated that adverse events and medical errors are common in the postdischarge period, affecting 23%‐49% of patients.79 Errors in the transition from inpatient to outpatient care increased the 3‐month rate of rehospitalization.9 New or worsening symptoms represented the most common adverse event.8 Noting that many of these postdischarge complications could be preventable if detected early, Forster suggested system changes such as earlier follow‐up with the outpatient physician or a postdischarge telephone call to check on the patient's status.7, 18 Future studies are planned to further analyze our data on medication issues and to determine if these problems are more prevalent for certain medications or diagnoses.

Comprehensive discharge planning remains an essential step in the discharge process. This may involve prescribing medications, arranging home health care services, and arranging outpatient follow‐up. The traditional hospital discharge process does not adequately ensure that patients understand their discharge plan and are able to comply with it. Calkins et al. compared physicians' perceptions of patients' understanding of medication side effects and activity restrictions with patients' actual understanding.6 They found that, compared with what was reported by patients, physicians overestimated the time spent discussing discharge plans and how well patients understood medication side effects and activity restrictions.

An important method for reducing patient problems is to contact patients by telephone after discharge in order to identify any health care issues. Previous research has confirmed that follow‐up telephone calls improve health outcomes and decrease resource utilization of patients, mainly those discharged from the emergency department.10, 11, 1923 A study of telephone follow‐up after ambulatory care visits did not find significant benefits of this procedure.24 In one of the few studies of telephone calls after hospitalization, pharmacists contacted patients 2 days after discharge and were able to detect and resolve medication‐related problems in 19% of patients and learned of new or worsening symptoms in 15%. Patient satisfaction was improved, and the intervention resulted in a lower rate of repeat visits to the emergency room within 30 days of discharge.12 Another study of telephone follow‐up following hospital discharge compared proactively calling all patients with providing a phone number that patients can call if they have questions. The study demonstrated that very few patients called the number provided, but of those patients called by the nursing service, more than 90% had questions about self‐care and recovery.25 These findings demonstrate the value of proactively contacting patients in the first several days after discharge, when problems can be detected and interventions initiated earlier.

One potential concern with this study was the low response rate. This was a retrospective analysis of an existing discharge management call‐center system, not a prospective study. We were not able to reach 52% of the patients discharged after 2 attempts by telephone. To have our call center make additional attempts to reach each patient by telephone would require a significant increase in the size of the call center, because at the time of the study, the staff was handling more than 370 patients discharged home a day. The telephone number of 16% of patients was missing or incorrect. We have since developed internal quality improvement mechanisms to decrease this percentage. After subtracting the patients we were unable to reach and those whose phone number was missing or incorrect, we were able to contact 32.4% of all the patients discharged home.

Several reasons explain the response rate found by many prospective research studies. In most studies of telephone follow‐up, patients must be able to consent to participate in order to be considered eligible inclusion. This raises the response rate because patients who do not consent to participate, have language barriers, or have no telephone are excluded from the study. In our study none of these types of patients were excluded. There are 2 additional differences between our study and many published studies that involve telephone surveys. Ours was not a prospective research study, and we contacted many more patients than did other studies. For example, a study by Forster et al. involved only 581 patients, and the research staff was diligent in its efforts to reach the patients,7 making up to 20 attempts for each patient. They reported a response rate of 69%. If they had included patients who were non‐English speakers or had no phone in their study, the response rate would have been 59%. Shesser et al. were able to reach 144 of 297 patients in their study of emergency room follow‐up, for a 48.4% response rate.25 The response rate in a study of telephone consultation with asthma patients was quite similar to ours. They enrolled 932 eligible patients, of whom they were able to reach 278, for a response rate of 30%.

It is possible that the rate of symptoms and the other variables we measured relative to this would have been different if we had been able to reach 100% of patients. There were some demographic differences between the patients we were able to reach and those we were not (Table 1). The nonresponders were slightly younger and slightly more likely to be female. Nonresponders were more likely to have Medicaid or commercial insurance or be self‐pay and were less likely to have Medicare. In addition, nonresponders had less severe illness. Although this scenario is highly unlikely, if none of the nonresponders had new or worsening symptoms, the rate of symptoms would only have been 3.86%. Conversely, it is possible but also very unlikely that a greater percentage of the nonresponders had new or worsening symptoms. Given the demographics of our study participants, we would expect a potentially slightly lower rate of signs and symptoms.

The present study had several other limitations. First, all patients surveyed were cared for by IPC‐employed physicians. It is possible that reported rates of symptoms and other postdischarge issues are not generalizable to other hospitalist practices. However, the present data were collected at more than 100 health care facilities in 10 health care markets, and the patients were cared for by more than 200 physicians. Therefore, it is unlikely these results would have been significantly influenced by a particular physician's or institution's practice patterns.

Second, because of the large number of facilities involved and that we could only track readmissions to facilities where our own hospitalists practice, we were not able to report 30‐ or 90‐day readmission rates or emergency room visit rates. In a prospective study, these would be important variables to track in order to assess the clinical relevance of the symptoms. We could track this data for some institutions, but for most of them, the quality of data was not sufficient to be meaningful or to make conclusions.

An additional limitation is that the call center did not differentiate between clinically minor and major symptoms. The inclusion of symptoms perhaps considered minor might have elevated the reported symptom frequency. However, the definitions of minor and major symptoms are very subjective, and a clinician's definitions might differ from those of a patient who is at home and uncomfortable. For example, nausea or loss of appetite related to new medications may be considered minor clinically but could be devastating to the patient experiencing them, leading the patient to stop taking the medication. Conversely, symptoms that may be considered nonsignificant by the patient may be interpreted as indicating clinically significant disease by a physician. Therefore, we would argue that, regardless of the severity of the symptom, follow‐up with a clinician is important.

Another limitation is based on our definition of an adequate follow‐up appointment as one scheduled within 2 weeks of discharge. It might be argued that if a patient's new symptoms were considered minor clinically, then a follow‐up interval greater than 2 weeks might be considered adequate. However, as already noted, a patient's criteria for considering a symptom minor and not requiring follow‐up may differ from a clinician's criteria. Also, the standardized discharge process requires that the hospitalist identify a physician for outpatient follow‐up and specify the period when the patient is to see the physician. Because of the inherent variability in having a many hospitalists practicing in many hospitals, not all patients had a scheduled appointment at discharge. We were not able to determine whether patients had an appointment date and time for follow‐up before discharge or had only received instructions to call the office for an appointment.

The Institute of Medicine, in its report Crossing the Quality Chasm, identified the coordination of care across services and sites of care as one of the health care system's redesign imperatives.2 Hospitalists are in a unique position to address transition care issues. Managing the transition from inpatient to outpatient care is vitally important, and hospitalists should play an essential role in designing a transition management system for discharged patients. Although individual efforts by hospitalists are essential to assuring postdischarge contact with patients, there is increasing agreement that system solutions are needed to improve the quality of care in the transition period following hospitalization. Improving a health care process involves more than working harder; it involves working differently.3 It is therefore imperative that hospitalist programs develop effective systems to manage the transition period until safe arrival by the patient in the outpatient physician's office.

In summary, 11.9% of patients contacted by a telephone call center within several days of discharge had new or worsening symptoms since discharge. There was no difference by age in the prevalence of symptoms. Patients who rated their health status as fair to poor were more likely to be symptomatic. Symptomatic patients were also more likely to have difficulty obtaining or understanding how to take their medications and receiving home health services. Patients who felt poorly were only minimally more likely to have made an appointment for follow‐up with their outpatient physician. It is hoped that by identifying patients who are doing poorly after discharge and intervening as necessary, we can improve the health outcome of our patients, as well as reduce the number of emergency room visits and readmission rates. Although actions by individual physicians are important, a system to manage the postdischarge transition period is essential for improving posthospitalization outcomes.

The Institute of Medicine reports To Err is Human and Crossing the Quality Chasm have drawn great attention to quality improvement and patient safety in the hospital setting.13 With the growth of the hospitalist field over the past several years,4 there has been increasing discussion about the importance of assuring quality of care, and some have argued that improving health care quality and reducing avoidable errors may be among the hospitalist's most important functions.5 Most discussions about the quality of hospital care have concerned the inpatient stay itself. However, the growth of hospital medicine, with its inherent discontinuity between inpatient and outpatient physicians, has intensified interest in the transition period from hospital discharge until first outpatient appointment.

At discharge, physicians may prescribe medications, order home health services, and arrange follow‐up appointments. It is often assumed a patient will remain stable after discharge and will follow up at the outpatient physician's office. Previous research has shown there may be problems with these assumptions. A patient may not understand the postdischarge treatment plan as well as the physician thinks.6 A recent study found that adverse events after discharge were common and often preventable.7 A follow‐up study confirmed that approximately 25% of patients had an adverse event after hospital discharge and that most adverse events caused symptoms but did not result in an emergency department visit, hospitalization, or death.8 Another study also found the prevalence of medical errors following hospitalization was high because of the discontinuity of care from the inpatient to the outpatient setting.9 These errors resulted in an increased rate of rehospitalization.

Telephone follow‐up may be a useful method of bridging the gap in care between discharge and the first outpatient appointment.10, 11 In most previous studies it was 2 or 3 weeks after discharge before patients were contacted or their records studied. By this point, patients who had done poorly may already have been readmitted or sought care at alternative locations. In one small study, pharmacists found that 12 of 79 patients (15%) contacted by telephone within 2 days of discharge12 had symptoms there were new or had worsened since discharge. The purpose of the present investigation was to extend these previous findings through a large multicenter study of how frequently patients had new or worsened symptoms within several days of discharge.

METHODS

RESULTS

During the study period, 48,236 patients were discharged to their homes from an acute care hospital. The IPC call center successfully contacted 16,135 patients after discharge, of whom 368 patients (2.3%) were excluded because of incomplete answers, leaving 15,767 as the valid study population (effective response rate = 32.4%). Of these, 98.9% were contacted within the first 5 days. The primary reasons for nonresponse or noninclusion in the present analysis were no answer after 2 attempts (52%) and missing or incorrect phone numbers (16%). If there was an answering machine, a message was left for the patient to call back. Those who called back accounted for fewer than 1% of all the patients.

A comparison of participants versus nonparticipants is shown in Table 1. The mean age of surveyed patients was 60.1 years, and 57% were female. The most common categories of insurance coverage were Medicare and HMO. The inpatient severity of illness of most patients was minor to moderate. Self‐reported health status was normally distributed, with the greatest percentage of patients rating their health as fair or good. On average, nonparticipants were younger than participants, more likely to be male, had a different pattern of health insurance, and a slightly lower severity of illness. The top 10 DRGs were the same for the respondents and nonrespondents, and the order of these 10 diagnoses was very similar.

Of the 15,767 patients contacted, 11.9% (N = 1876) reported symptoms that were new or had worsened since leaving the hospital. Sixty‐four percent of these patients had new symptoms, and 36% had worsening symptoms. These two groups were combined for analysis in this study because for both groups, identification and action are important. Of the patients with new or worse symptoms, 37% required no assistance from the nurse because they had already notified a doctor and/or were doing something about it. The other 63% either had not notified their doctor or had concerns about their signs and symptoms. The most common action taken by the nurse was patient education regarding the symptoms. Of those requiring nurse intervention in addition to education, the most common intervention was to contact the patient's primary care provider or specialist about the patient's symptoms, followed by contacting the hospitalist. In 72% of nurse interventions, the patient's primary care physician or a specialist was contacted about the new or worsened symptoms. Other interventions included contacting the physician's office to obtain a prescription for a medication for the symptom, to get an appointment for the patient, or to reschedule an appointment to be earlier. A referral to an emergency room or urgent care center was given to 4% of patients.

Mean age of the patients with new or worsened symptoms was 60.5 years. The age distribution of symptomatic and asymptomatic patients was not significantly different, whether comparing by mean, median, or decades. Table 2 illustrates factors associated with the increased rate of new or worsened symptoms. Women were more likely than men to report symptoms (13.0% vs. 10.3%, P < .0001). As health status worsened, the percentage of patients with new or worsened symptoms increased (P < .0001), as it did with increased inpatient SOI (P < .0001). There was no correlation between self‐rated health status and SOI score based on DRG score, suggesting they measured different parameters. Table 3 lists the percentages of patients reporting new or worsened symptoms for the most common DRGs. The only significant distinction was that patients discharged with a DRG of chest pain were less likely to report symptoms than were all patients.

The symptoms the patients reported are categorized in Table 4 without distinction as to whether they are primary or secondary. Gastrointestinal symptoms were the most common category of symptoms, followed by general symptoms, cardiovascular symptoms, and pain. The most common symptoms reported were fatigue/weakness, nausea/vomiting, and edema.

The call center assessed whether the patient had difficulty making a follow‐up appointment and whether an appointment was scheduled within 2 weeks of discharge. Patients with new or worsening symptoms were only minimally more likely to have scheduled follow‐up (61.0% vs. 58.4% for patients not reporting new or worsening symptoms, P < .05). Symptomatic patients had a higher prevalence of medication issues, defined as not picking up their prescriptions or not understanding how to take their medication (22.2% vs. 6.8% for asymptomatic patients; P < .0001). Likewise, the prevalence of symptomatic patients having problems receiving scheduled home health care services was 5.8%, compared with a prevalence of 3.6% for asymptomatic patients (P < .0001).

DISCUSSION

Enhancing the quality of care provided by hospitalists means not only improving care during hospitalization but also assuring patient stability between discharge and outpatient follow‐up. As part of efforts to improve transition management, the call center at IPC attempted to contact all patients discharged home within several days of discharge. Of 15,767 patients surveyed between May 1, 2003, and October 31, 2003, 11.9% (N = 1876) had new or worsening symptoms since leaving the hospital only 2 or 3 days earlier. We had hypothesized that older patients might be more symptomatic than younger ones, but found no difference in the prevalence of new or worsening symptoms based on age. Women were more likely to be symptomatic than men.

We defined appropriate postdischarge follow‐up as having an appointment with an outpatient physician within 2 weeks. A previous study of psychiatric patients documented that keeping a follow‐up appointment significantly reduces the risk of rehospitalization.17 Similar data do not exist for medical patients. Our data demonstrated that symptomatic patients were only minimally more likely to have made a follow‐up appointment with their outpatient physician within the first 2 weeks than were those patients who were not symptomatic. As part of patient education at discharge, clinicians routinely counsel patients to call their outpatient physician should they experience new or worsening symptoms once at home. Inpatient physicians may assume this recommendation provides a safety net for the patients should they develop problems after discharge. However, our finding that almost 40% of patients with new or worsening symptoms within 2‐3 days of discharge had not made a follow‐up appointment with their physician suggests many patients fall through this safety net. Although there was a slight statistically significant difference between the groups, this difference was not clinically significant. One potential limitation of our data is that we did not examine whether there was a correlation between the day of the week that a patient was discharged and inability to make a follow‐up appointment.

As part of the survey script (see Appendix), we inquired whether patients were able to pick up their prescriptions and whether they understood how to take their medication. A high percentage of patients in our study reported having one of these medication issues in the first several days following hospital discharge, providing an opportunity for early intervention and prevention of medical error. Forster and others have demonstrated that adverse events and medical errors are common in the postdischarge period, affecting 23%‐49% of patients.79 Errors in the transition from inpatient to outpatient care increased the 3‐month rate of rehospitalization.9 New or worsening symptoms represented the most common adverse event.8 Noting that many of these postdischarge complications could be preventable if detected early, Forster suggested system changes such as earlier follow‐up with the outpatient physician or a postdischarge telephone call to check on the patient's status.7, 18 Future studies are planned to further analyze our data on medication issues and to determine if these problems are more prevalent for certain medications or diagnoses.

Comprehensive discharge planning remains an essential step in the discharge process. This may involve prescribing medications, arranging home health care services, and arranging outpatient follow‐up. The traditional hospital discharge process does not adequately ensure that patients understand their discharge plan and are able to comply with it. Calkins et al. compared physicians' perceptions of patients' understanding of medication side effects and activity restrictions with patients' actual understanding.6 They found that, compared with what was reported by patients, physicians overestimated the time spent discussing discharge plans and how well patients understood medication side effects and activity restrictions.

An important method for reducing patient problems is to contact patients by telephone after discharge in order to identify any health care issues. Previous research has confirmed that follow‐up telephone calls improve health outcomes and decrease resource utilization of patients, mainly those discharged from the emergency department.10, 11, 1923 A study of telephone follow‐up after ambulatory care visits did not find significant benefits of this procedure.24 In one of the few studies of telephone calls after hospitalization, pharmacists contacted patients 2 days after discharge and were able to detect and resolve medication‐related problems in 19% of patients and learned of new or worsening symptoms in 15%. Patient satisfaction was improved, and the intervention resulted in a lower rate of repeat visits to the emergency room within 30 days of discharge.12 Another study of telephone follow‐up following hospital discharge compared proactively calling all patients with providing a phone number that patients can call if they have questions. The study demonstrated that very few patients called the number provided, but of those patients called by the nursing service, more than 90% had questions about self‐care and recovery.25 These findings demonstrate the value of proactively contacting patients in the first several days after discharge, when problems can be detected and interventions initiated earlier.

One potential concern with this study was the low response rate. This was a retrospective analysis of an existing discharge management call‐center system, not a prospective study. We were not able to reach 52% of the patients discharged after 2 attempts by telephone. To have our call center make additional attempts to reach each patient by telephone would require a significant increase in the size of the call center, because at the time of the study, the staff was handling more than 370 patients discharged home a day. The telephone number of 16% of patients was missing or incorrect. We have since developed internal quality improvement mechanisms to decrease this percentage. After subtracting the patients we were unable to reach and those whose phone number was missing or incorrect, we were able to contact 32.4% of all the patients discharged home.

Several reasons explain the response rate found by many prospective research studies. In most studies of telephone follow‐up, patients must be able to consent to participate in order to be considered eligible inclusion. This raises the response rate because patients who do not consent to participate, have language barriers, or have no telephone are excluded from the study. In our study none of these types of patients were excluded. There are 2 additional differences between our study and many published studies that involve telephone surveys. Ours was not a prospective research study, and we contacted many more patients than did other studies. For example, a study by Forster et al. involved only 581 patients, and the research staff was diligent in its efforts to reach the patients,7 making up to 20 attempts for each patient. They reported a response rate of 69%. If they had included patients who were non‐English speakers or had no phone in their study, the response rate would have been 59%. Shesser et al. were able to reach 144 of 297 patients in their study of emergency room follow‐up, for a 48.4% response rate.25 The response rate in a study of telephone consultation with asthma patients was quite similar to ours. They enrolled 932 eligible patients, of whom they were able to reach 278, for a response rate of 30%.

It is possible that the rate of symptoms and the other variables we measured relative to this would have been different if we had been able to reach 100% of patients. There were some demographic differences between the patients we were able to reach and those we were not (Table 1). The nonresponders were slightly younger and slightly more likely to be female. Nonresponders were more likely to have Medicaid or commercial insurance or be self‐pay and were less likely to have Medicare. In addition, nonresponders had less severe illness. Although this scenario is highly unlikely, if none of the nonresponders had new or worsening symptoms, the rate of symptoms would only have been 3.86%. Conversely, it is possible but also very unlikely that a greater percentage of the nonresponders had new or worsening symptoms. Given the demographics of our study participants, we would expect a potentially slightly lower rate of signs and symptoms.

The present study had several other limitations. First, all patients surveyed were cared for by IPC‐employed physicians. It is possible that reported rates of symptoms and other postdischarge issues are not generalizable to other hospitalist practices. However, the present data were collected at more than 100 health care facilities in 10 health care markets, and the patients were cared for by more than 200 physicians. Therefore, it is unlikely these results would have been significantly influenced by a particular physician's or institution's practice patterns.

Second, because of the large number of facilities involved and that we could only track readmissions to facilities where our own hospitalists practice, we were not able to report 30‐ or 90‐day readmission rates or emergency room visit rates. In a prospective study, these would be important variables to track in order to assess the clinical relevance of the symptoms. We could track this data for some institutions, but for most of them, the quality of data was not sufficient to be meaningful or to make conclusions.

An additional limitation is that the call center did not differentiate between clinically minor and major symptoms. The inclusion of symptoms perhaps considered minor might have elevated the reported symptom frequency. However, the definitions of minor and major symptoms are very subjective, and a clinician's definitions might differ from those of a patient who is at home and uncomfortable. For example, nausea or loss of appetite related to new medications may be considered minor clinically but could be devastating to the patient experiencing them, leading the patient to stop taking the medication. Conversely, symptoms that may be considered nonsignificant by the patient may be interpreted as indicating clinically significant disease by a physician. Therefore, we would argue that, regardless of the severity of the symptom, follow‐up with a clinician is important.

Another limitation is based on our definition of an adequate follow‐up appointment as one scheduled within 2 weeks of discharge. It might be argued that if a patient's new symptoms were considered minor clinically, then a follow‐up interval greater than 2 weeks might be considered adequate. However, as already noted, a patient's criteria for considering a symptom minor and not requiring follow‐up may differ from a clinician's criteria. Also, the standardized discharge process requires that the hospitalist identify a physician for outpatient follow‐up and specify the period when the patient is to see the physician. Because of the inherent variability in having a many hospitalists practicing in many hospitals, not all patients had a scheduled appointment at discharge. We were not able to determine whether patients had an appointment date and time for follow‐up before discharge or had only received instructions to call the office for an appointment.

The Institute of Medicine, in its report Crossing the Quality Chasm, identified the coordination of care across services and sites of care as one of the health care system's redesign imperatives.2 Hospitalists are in a unique position to address transition care issues. Managing the transition from inpatient to outpatient care is vitally important, and hospitalists should play an essential role in designing a transition management system for discharged patients. Although individual efforts by hospitalists are essential to assuring postdischarge contact with patients, there is increasing agreement that system solutions are needed to improve the quality of care in the transition period following hospitalization. Improving a health care process involves more than working harder; it involves working differently.3 It is therefore imperative that hospitalist programs develop effective systems to manage the transition period until safe arrival by the patient in the outpatient physician's office.

In summary, 11.9% of patients contacted by a telephone call center within several days of discharge had new or worsening symptoms since discharge. There was no difference by age in the prevalence of symptoms. Patients who rated their health status as fair to poor were more likely to be symptomatic. Symptomatic patients were also more likely to have difficulty obtaining or understanding how to take their medications and receiving home health services. Patients who felt poorly were only minimally more likely to have made an appointment for follow‐up with their outpatient physician. It is hoped that by identifying patients who are doing poorly after discharge and intervening as necessary, we can improve the health outcome of our patients, as well as reduce the number of emergency room visits and readmission rates. Although actions by individual physicians are important, a system to manage the postdischarge transition period is essential for improving posthospitalization outcomes.

The Institute of Medicine reports To Err is Human and Crossing the Quality Chasm have drawn great attention to quality improvement and patient safety in the hospital setting.13 With the growth of the hospitalist field over the past several years,4 there has been increasing discussion about the importance of assuring quality of care, and some have argued that improving health care quality and reducing avoidable errors may be among the hospitalist's most important functions.5 Most discussions about the quality of hospital care have concerned the inpatient stay itself. However, the growth of hospital medicine, with its inherent discontinuity between inpatient and outpatient physicians, has intensified interest in the transition period from hospital discharge until first outpatient appointment.

At discharge, physicians may prescribe medications, order home health services, and arrange follow‐up appointments. It is often assumed a patient will remain stable after discharge and will follow up at the outpatient physician's office. Previous research has shown there may be problems with these assumptions. A patient may not understand the postdischarge treatment plan as well as the physician thinks.6 A recent study found that adverse events after discharge were common and often preventable.7 A follow‐up study confirmed that approximately 25% of patients had an adverse event after hospital discharge and that most adverse events caused symptoms but did not result in an emergency department visit, hospitalization, or death.8 Another study also found the prevalence of medical errors following hospitalization was high because of the discontinuity of care from the inpatient to the outpatient setting.9 These errors resulted in an increased rate of rehospitalization.

Telephone follow‐up may be a useful method of bridging the gap in care between discharge and the first outpatient appointment.10, 11 In most previous studies it was 2 or 3 weeks after discharge before patients were contacted or their records studied. By this point, patients who had done poorly may already have been readmitted or sought care at alternative locations. In one small study, pharmacists found that 12 of 79 patients (15%) contacted by telephone within 2 days of discharge12 had symptoms there were new or had worsened since discharge. The purpose of the present investigation was to extend these previous findings through a large multicenter study of how frequently patients had new or worsened symptoms within several days of discharge.

METHODS

RESULTS

During the study period, 48,236 patients were discharged to their homes from an acute care hospital. The IPC call center successfully contacted 16,135 patients after discharge, of whom 368 patients (2.3%) were excluded because of incomplete answers, leaving 15,767 as the valid study population (effective response rate = 32.4%). Of these, 98.9% were contacted within the first 5 days. The primary reasons for nonresponse or noninclusion in the present analysis were no answer after 2 attempts (52%) and missing or incorrect phone numbers (16%). If there was an answering machine, a message was left for the patient to call back. Those who called back accounted for fewer than 1% of all the patients.

A comparison of participants versus nonparticipants is shown in Table 1. The mean age of surveyed patients was 60.1 years, and 57% were female. The most common categories of insurance coverage were Medicare and HMO. The inpatient severity of illness of most patients was minor to moderate. Self‐reported health status was normally distributed, with the greatest percentage of patients rating their health as fair or good. On average, nonparticipants were younger than participants, more likely to be male, had a different pattern of health insurance, and a slightly lower severity of illness. The top 10 DRGs were the same for the respondents and nonrespondents, and the order of these 10 diagnoses was very similar.

Of the 15,767 patients contacted, 11.9% (N = 1876) reported symptoms that were new or had worsened since leaving the hospital. Sixty‐four percent of these patients had new symptoms, and 36% had worsening symptoms. These two groups were combined for analysis in this study because for both groups, identification and action are important. Of the patients with new or worse symptoms, 37% required no assistance from the nurse because they had already notified a doctor and/or were doing something about it. The other 63% either had not notified their doctor or had concerns about their signs and symptoms. The most common action taken by the nurse was patient education regarding the symptoms. Of those requiring nurse intervention in addition to education, the most common intervention was to contact the patient's primary care provider or specialist about the patient's symptoms, followed by contacting the hospitalist. In 72% of nurse interventions, the patient's primary care physician or a specialist was contacted about the new or worsened symptoms. Other interventions included contacting the physician's office to obtain a prescription for a medication for the symptom, to get an appointment for the patient, or to reschedule an appointment to be earlier. A referral to an emergency room or urgent care center was given to 4% of patients.

Mean age of the patients with new or worsened symptoms was 60.5 years. The age distribution of symptomatic and asymptomatic patients was not significantly different, whether comparing by mean, median, or decades. Table 2 illustrates factors associated with the increased rate of new or worsened symptoms. Women were more likely than men to report symptoms (13.0% vs. 10.3%, P < .0001). As health status worsened, the percentage of patients with new or worsened symptoms increased (P < .0001), as it did with increased inpatient SOI (P < .0001). There was no correlation between self‐rated health status and SOI score based on DRG score, suggesting they measured different parameters. Table 3 lists the percentages of patients reporting new or worsened symptoms for the most common DRGs. The only significant distinction was that patients discharged with a DRG of chest pain were less likely to report symptoms than were all patients.

The symptoms the patients reported are categorized in Table 4 without distinction as to whether they are primary or secondary. Gastrointestinal symptoms were the most common category of symptoms, followed by general symptoms, cardiovascular symptoms, and pain. The most common symptoms reported were fatigue/weakness, nausea/vomiting, and edema.

The call center assessed whether the patient had difficulty making a follow‐up appointment and whether an appointment was scheduled within 2 weeks of discharge. Patients with new or worsening symptoms were only minimally more likely to have scheduled follow‐up (61.0% vs. 58.4% for patients not reporting new or worsening symptoms, P < .05). Symptomatic patients had a higher prevalence of medication issues, defined as not picking up their prescriptions or not understanding how to take their medication (22.2% vs. 6.8% for asymptomatic patients; P < .0001). Likewise, the prevalence of symptomatic patients having problems receiving scheduled home health care services was 5.8%, compared with a prevalence of 3.6% for asymptomatic patients (P < .0001).

DISCUSSION

Enhancing the quality of care provided by hospitalists means not only improving care during hospitalization but also assuring patient stability between discharge and outpatient follow‐up. As part of efforts to improve transition management, the call center at IPC attempted to contact all patients discharged home within several days of discharge. Of 15,767 patients surveyed between May 1, 2003, and October 31, 2003, 11.9% (N = 1876) had new or worsening symptoms since leaving the hospital only 2 or 3 days earlier. We had hypothesized that older patients might be more symptomatic than younger ones, but found no difference in the prevalence of new or worsening symptoms based on age. Women were more likely to be symptomatic than men.

We defined appropriate postdischarge follow‐up as having an appointment with an outpatient physician within 2 weeks. A previous study of psychiatric patients documented that keeping a follow‐up appointment significantly reduces the risk of rehospitalization.17 Similar data do not exist for medical patients. Our data demonstrated that symptomatic patients were only minimally more likely to have made a follow‐up appointment with their outpatient physician within the first 2 weeks than were those patients who were not symptomatic. As part of patient education at discharge, clinicians routinely counsel patients to call their outpatient physician should they experience new or worsening symptoms once at home. Inpatient physicians may assume this recommendation provides a safety net for the patients should they develop problems after discharge. However, our finding that almost 40% of patients with new or worsening symptoms within 2‐3 days of discharge had not made a follow‐up appointment with their physician suggests many patients fall through this safety net. Although there was a slight statistically significant difference between the groups, this difference was not clinically significant. One potential limitation of our data is that we did not examine whether there was a correlation between the day of the week that a patient was discharged and inability to make a follow‐up appointment.

As part of the survey script (see Appendix), we inquired whether patients were able to pick up their prescriptions and whether they understood how to take their medication. A high percentage of patients in our study reported having one of these medication issues in the first several days following hospital discharge, providing an opportunity for early intervention and prevention of medical error. Forster and others have demonstrated that adverse events and medical errors are common in the postdischarge period, affecting 23%‐49% of patients.79 Errors in the transition from inpatient to outpatient care increased the 3‐month rate of rehospitalization.9 New or worsening symptoms represented the most common adverse event.8 Noting that many of these postdischarge complications could be preventable if detected early, Forster suggested system changes such as earlier follow‐up with the outpatient physician or a postdischarge telephone call to check on the patient's status.7, 18 Future studies are planned to further analyze our data on medication issues and to determine if these problems are more prevalent for certain medications or diagnoses.

Comprehensive discharge planning remains an essential step in the discharge process. This may involve prescribing medications, arranging home health care services, and arranging outpatient follow‐up. The traditional hospital discharge process does not adequately ensure that patients understand their discharge plan and are able to comply with it. Calkins et al. compared physicians' perceptions of patients' understanding of medication side effects and activity restrictions with patients' actual understanding.6 They found that, compared with what was reported by patients, physicians overestimated the time spent discussing discharge plans and how well patients understood medication side effects and activity restrictions.

An important method for reducing patient problems is to contact patients by telephone after discharge in order to identify any health care issues. Previous research has confirmed that follow‐up telephone calls improve health outcomes and decrease resource utilization of patients, mainly those discharged from the emergency department.10, 11, 1923 A study of telephone follow‐up after ambulatory care visits did not find significant benefits of this procedure.24 In one of the few studies of telephone calls after hospitalization, pharmacists contacted patients 2 days after discharge and were able to detect and resolve medication‐related problems in 19% of patients and learned of new or worsening symptoms in 15%. Patient satisfaction was improved, and the intervention resulted in a lower rate of repeat visits to the emergency room within 30 days of discharge.12 Another study of telephone follow‐up following hospital discharge compared proactively calling all patients with providing a phone number that patients can call if they have questions. The study demonstrated that very few patients called the number provided, but of those patients called by the nursing service, more than 90% had questions about self‐care and recovery.25 These findings demonstrate the value of proactively contacting patients in the first several days after discharge, when problems can be detected and interventions initiated earlier.

One potential concern with this study was the low response rate. This was a retrospective analysis of an existing discharge management call‐center system, not a prospective study. We were not able to reach 52% of the patients discharged after 2 attempts by telephone. To have our call center make additional attempts to reach each patient by telephone would require a significant increase in the size of the call center, because at the time of the study, the staff was handling more than 370 patients discharged home a day. The telephone number of 16% of patients was missing or incorrect. We have since developed internal quality improvement mechanisms to decrease this percentage. After subtracting the patients we were unable to reach and those whose phone number was missing or incorrect, we were able to contact 32.4% of all the patients discharged home.

Several reasons explain the response rate found by many prospective research studies. In most studies of telephone follow‐up, patients must be able to consent to participate in order to be considered eligible inclusion. This raises the response rate because patients who do not consent to participate, have language barriers, or have no telephone are excluded from the study. In our study none of these types of patients were excluded. There are 2 additional differences between our study and many published studies that involve telephone surveys. Ours was not a prospective research study, and we contacted many more patients than did other studies. For example, a study by Forster et al. involved only 581 patients, and the research staff was diligent in its efforts to reach the patients,7 making up to 20 attempts for each patient. They reported a response rate of 69%. If they had included patients who were non‐English speakers or had no phone in their study, the response rate would have been 59%. Shesser et al. were able to reach 144 of 297 patients in their study of emergency room follow‐up, for a 48.4% response rate.25 The response rate in a study of telephone consultation with asthma patients was quite similar to ours. They enrolled 932 eligible patients, of whom they were able to reach 278, for a response rate of 30%.

It is possible that the rate of symptoms and the other variables we measured relative to this would have been different if we had been able to reach 100% of patients. There were some demographic differences between the patients we were able to reach and those we were not (Table 1). The nonresponders were slightly younger and slightly more likely to be female. Nonresponders were more likely to have Medicaid or commercial insurance or be self‐pay and were less likely to have Medicare. In addition, nonresponders had less severe illness. Although this scenario is highly unlikely, if none of the nonresponders had new or worsening symptoms, the rate of symptoms would only have been 3.86%. Conversely, it is possible but also very unlikely that a greater percentage of the nonresponders had new or worsening symptoms. Given the demographics of our study participants, we would expect a potentially slightly lower rate of signs and symptoms.

The present study had several other limitations. First, all patients surveyed were cared for by IPC‐employed physicians. It is possible that reported rates of symptoms and other postdischarge issues are not generalizable to other hospitalist practices. However, the present data were collected at more than 100 health care facilities in 10 health care markets, and the patients were cared for by more than 200 physicians. Therefore, it is unlikely these results would have been significantly influenced by a particular physician's or institution's practice patterns.

Second, because of the large number of facilities involved and that we could only track readmissions to facilities where our own hospitalists practice, we were not able to report 30‐ or 90‐day readmission rates or emergency room visit rates. In a prospective study, these would be important variables to track in order to assess the clinical relevance of the symptoms. We could track this data for some institutions, but for most of them, the quality of data was not sufficient to be meaningful or to make conclusions.

An additional limitation is that the call center did not differentiate between clinically minor and major symptoms. The inclusion of symptoms perhaps considered minor might have elevated the reported symptom frequency. However, the definitions of minor and major symptoms are very subjective, and a clinician's definitions might differ from those of a patient who is at home and uncomfortable. For example, nausea or loss of appetite related to new medications may be considered minor clinically but could be devastating to the patient experiencing them, leading the patient to stop taking the medication. Conversely, symptoms that may be considered nonsignificant by the patient may be interpreted as indicating clinically significant disease by a physician. Therefore, we would argue that, regardless of the severity of the symptom, follow‐up with a clinician is important.

Another limitation is based on our definition of an adequate follow‐up appointment as one scheduled within 2 weeks of discharge. It might be argued that if a patient's new symptoms were considered minor clinically, then a follow‐up interval greater than 2 weeks might be considered adequate. However, as already noted, a patient's criteria for considering a symptom minor and not requiring follow‐up may differ from a clinician's criteria. Also, the standardized discharge process requires that the hospitalist identify a physician for outpatient follow‐up and specify the period when the patient is to see the physician. Because of the inherent variability in having a many hospitalists practicing in many hospitals, not all patients had a scheduled appointment at discharge. We were not able to determine whether patients had an appointment date and time for follow‐up before discharge or had only received instructions to call the office for an appointment.

The Institute of Medicine, in its report Crossing the Quality Chasm, identified the coordination of care across services and sites of care as one of the health care system's redesign imperatives.2 Hospitalists are in a unique position to address transition care issues. Managing the transition from inpatient to outpatient care is vitally important, and hospitalists should play an essential role in designing a transition management system for discharged patients. Although individual efforts by hospitalists are essential to assuring postdischarge contact with patients, there is increasing agreement that system solutions are needed to improve the quality of care in the transition period following hospitalization. Improving a health care process involves more than working harder; it involves working differently.3 It is therefore imperative that hospitalist programs develop effective systems to manage the transition period until safe arrival by the patient in the outpatient physician's office.

In summary, 11.9% of patients contacted by a telephone call center within several days of discharge had new or worsening symptoms since discharge. There was no difference by age in the prevalence of symptoms. Patients who rated their health status as fair to poor were more likely to be symptomatic. Symptomatic patients were also more likely to have difficulty obtaining or understanding how to take their medications and receiving home health services. Patients who felt poorly were only minimally more likely to have made an appointment for follow‐up with their outpatient physician. It is hoped that by identifying patients who are doing poorly after discharge and intervening as necessary, we can improve the health outcome of our patients, as well as reduce the number of emergency room visits and readmission rates. Although actions by individual physicians are important, a system to manage the postdischarge transition period is essential for improving posthospitalization outcomes.

Soon after they form, most new medical fields begin agitating for a special certification, something that says, We're here, and we're different. As I've noted previously in the Journal of Hospital Medicine, the field of hospital medicine resisted this impulse in its early years, fearing that any special designation or certification would actually harm the field's growth and status.1 The concern was that managed‐care organizationsconvinced by the evidence that hospitalists improve efficiency and might improve quality2would react to any new hospitalist sheepskin by mandating that anyone providing hospital care to its covered patients have one. The backlash from primary care physicians locked out of the hospital by such a mandate would have been swift and ultimately damaging to hospitalists. In addition to these political considerations, the early field of hospital medicine lacked the academic credibility and scientific underpinning needed for specialty designation.3

Times have changed. There are now more than 15,000 hospitalists in the United States, and nearly half of American hospitals have hospitalists on their medical staffs. In many markets, including my own, hospitalists care for most internal medicine inpatients, as well as significant numbers of pediatric and surgical patients. The field has achieved academic legitimacy, with this journal, several textbooks, large and flourishing groups in every academic medical center, and several residency tracks and fellowship programs.4, 5 The Society of Hospital Medicine (SHM) has grown to more than 6000 members, become a widely respected and dynamic member of the community of professional societies, and published its core competencies.6

With this as a background, in 2004 SHM asked the American Board of Internal Medicine (ABIM) to consider a program of certification for hospitalists. As a past SHM president and now a member of the ABIM Board of Directors, I am privileged to have a bird's‐eye view of the process. In this article, I reflect on some of the key issues it raises.

THE NUTS AND BOLTS OF BOARD CERTIFICATION

Since the first board (ophthalmology) was formed in 1917, 24 specialty boards have emerged, all under the umbrella of the American Board of Medical Specialties (ABMS).7 Because no one type of physician can do it all, certifying boards have had to struggle not only with how to assess competency in existing disciplines, but with the dynamic and often controversial questions raised when new fields emerge. In the past few decades, certifying boards have grappled with specialties formed around new procedures (such as cardiac electrophysiology), discrete populations (geriatrics, palliative care), complex diseases (HIV medicine), and sites of care (intensive care medicine, emergency medicine). It is this latter category that now includes hospital medicine.

In the past, it was relatively simple for a physician to obtain board certification. Residency or fellowship training was believed to confer on its graduates the presumption of competence and professionalismthe program director's attestation served as the graduate's Good Housekeeping seal of approval. Passing the board exam was the final step, ensuring that newly minted graduates had the requisite knowledge and judgment to practice in their fields.

Remarkably, for the first half century of the specialty boards, all certifications lasted for a physician's professional lifetime. Beginning with the 1969 decision of the American Board of Family Practice to limit the validity of its certificates to 7 years, all ABMS member boards now time limit their certifications, usually to 7‐10 years.7 Of course, in an environment of rapidly changing medical knowledge and new procedures, periodiceven continuousdemonstration of competence is increasingly expected by the public.

For ABIM, the mechanism to promote lifelong learning and demonstrate ongoing competence in the face of a rapidly changing environment is known as maintenance of certification (MOC).8 Through MOC, board‐certified internists demonstrate their ongoing clinical expertise and judgment, their involvement in lifelong learning and quality improvement activities, and their professionalism. Because MOC involves no new training requirements and includes an assessment of a physician's actual practice, it provides a potential mechanism, heretofore untapped, of demonstrating a unique professional focus that emerges after the completion of formal training.

HOSPITALIST CERTIFICATION AND THE MOC PROCESS

As ABIM considered a separate certification pathway for hospital medicine, it faced a conundrum. The vast majority of hospitalists are general internists (most of the rest are generalists in family medicine or pediatrics) who entered hospital medicine at the completion of their internal medicine training or after a period of primary care practice. Job opportunities for hospitalists are plentiful, andexcept for additional training in quality improvement, systems leadership, care transitions, palliative care, and communication9there is little clinical rationale to prolong internal medicine training for hospitalists (some individuals may opt for fellowships to enhance their leadership skills or to launch a research career,5 but few would argue for mandatory additional clinical training in hospital medicine at this time).

So, in the absence of formal training, how could the ABIM (or other boards) recognize the focused practice of hospitalists? This question must be framed within a broader challenge: Is it possible and appropriate for certifying boards to recognize expertise and focus that is accrued not through formal training, but through actual practice experience and accompanying self‐directed learning?

In 2006, the ABIM took up this question, producing a report (New and Emerging Disciplines in Internal Medicine II [NEDIM II]) that delineated several criteria to guide whether a new field merited focused recognition through MOC (Table 1). Judging by these criteria, hospital medicine appears to be a suitable first candidate for recognition of focused practice through MOC.

PRELIMINARY THOUGHTS ON FOCUSED RECOGNITION IN HOSPITAL MEDICINE

The ABIM has endorsed the concept of recognition of focused practice in hospital medicine and charged a subcommittee (that I chair) with working out the details. It would be premature to describe the committee's deliberations in detail (particularly because the final plan needs to be approved by both the ABIM and the ABMS), but the following are some key issues being discussed.

First, demonstration of focused practice requires some minimum volume of hospitalized patients. In the absence of hard data defining a threshold number of cases for hospitalists, we are likely to endorse a number that has face validity and that reliably separates self‐identified hospitalists from nonhospitalist generalists. As with all volume requirements, we will struggle over how to handle academic physicians, physician‐administrators, and physician‐researchers who limit their overall clinical practice but who spend most of their clinical time in hospital medicine and the bulk of their nonclinical time trying to improve hospital care.

The requirements to demonstrate performance in practice and lifelong learning may be more straightforward. As with all such MOC requirements, the ABIM is increasingly looking to use real practice data, trying to harmonize its data requirements with those of other organizations such as insurers, Medicare, the Joint Commission, or for pay‐for‐performance initiatives. Despite the operational challenges, this effort is vital: for MOC (including focused recognition) to be highly valued by patients, purchasers, and diplomates, it will increasingly need to measure not only what physicians know, but also what they do.

Finally, there is the test. It is likely that a secure exam for MOC with Recognition of Focused Practice in Hospital Medicine will involve core content in internal medicine (information that every internist should know), augmented by substantial and challenging content in hospital medicine. Because it will be vital that a competent hospitalist understand key elements of outpatient practice, the exam will not be stripped of ambulatory content but will likely have fewer questions on topics that hospitalists are unlikely to confront (osteoporosis, cancer screening).

ONGOING ISSUES

As hospital medicine continues its explosive growth, it is important to develop ways to make board certification relevant to hospitalists. The ABIM believes that modifying the MOC process to recognize physicians who have focused their practice and achieved special expertise in hospital medicine is a good way to launch this effort. Ultimately, this process is likely to evolve, particularly if separate training pathways for hospital medicine emerge. For now, the development of Recognition of Focused Practice in Hospital Medicine will further legitimize the new field, provide ABIM with insights into how to recognize physicians who have advanced through practice‐based learning rather than through training, and help to guide other certifying boards (particularly family medicine and pediatrics) considering hospitalist certification. In the end, the process will need to be user‐friendly for and satisfying to diplomates, flexible enough to allow for career transitions (both toward and away from hospital medicine), and sufficiently rigorous to be credible to all stakeholders, particularly patients.

Soon after they form, most new medical fields begin agitating for a special certification, something that says, We're here, and we're different. As I've noted previously in the Journal of Hospital Medicine, the field of hospital medicine resisted this impulse in its early years, fearing that any special designation or certification would actually harm the field's growth and status.1 The concern was that managed‐care organizationsconvinced by the evidence that hospitalists improve efficiency and might improve quality2would react to any new hospitalist sheepskin by mandating that anyone providing hospital care to its covered patients have one. The backlash from primary care physicians locked out of the hospital by such a mandate would have been swift and ultimately damaging to hospitalists. In addition to these political considerations, the early field of hospital medicine lacked the academic credibility and scientific underpinning needed for specialty designation.3

Times have changed. There are now more than 15,000 hospitalists in the United States, and nearly half of American hospitals have hospitalists on their medical staffs. In many markets, including my own, hospitalists care for most internal medicine inpatients, as well as significant numbers of pediatric and surgical patients. The field has achieved academic legitimacy, with this journal, several textbooks, large and flourishing groups in every academic medical center, and several residency tracks and fellowship programs.4, 5 The Society of Hospital Medicine (SHM) has grown to more than 6000 members, become a widely respected and dynamic member of the community of professional societies, and published its core competencies.6

With this as a background, in 2004 SHM asked the American Board of Internal Medicine (ABIM) to consider a program of certification for hospitalists. As a past SHM president and now a member of the ABIM Board of Directors, I am privileged to have a bird's‐eye view of the process. In this article, I reflect on some of the key issues it raises.

THE NUTS AND BOLTS OF BOARD CERTIFICATION

Since the first board (ophthalmology) was formed in 1917, 24 specialty boards have emerged, all under the umbrella of the American Board of Medical Specialties (ABMS).7 Because no one type of physician can do it all, certifying boards have had to struggle not only with how to assess competency in existing disciplines, but with the dynamic and often controversial questions raised when new fields emerge. In the past few decades, certifying boards have grappled with specialties formed around new procedures (such as cardiac electrophysiology), discrete populations (geriatrics, palliative care), complex diseases (HIV medicine), and sites of care (intensive care medicine, emergency medicine). It is this latter category that now includes hospital medicine.

In the past, it was relatively simple for a physician to obtain board certification. Residency or fellowship training was believed to confer on its graduates the presumption of competence and professionalismthe program director's attestation served as the graduate's Good Housekeeping seal of approval. Passing the board exam was the final step, ensuring that newly minted graduates had the requisite knowledge and judgment to practice in their fields.

Remarkably, for the first half century of the specialty boards, all certifications lasted for a physician's professional lifetime. Beginning with the 1969 decision of the American Board of Family Practice to limit the validity of its certificates to 7 years, all ABMS member boards now time limit their certifications, usually to 7‐10 years.7 Of course, in an environment of rapidly changing medical knowledge and new procedures, periodiceven continuousdemonstration of competence is increasingly expected by the public.

For ABIM, the mechanism to promote lifelong learning and demonstrate ongoing competence in the face of a rapidly changing environment is known as maintenance of certification (MOC).8 Through MOC, board‐certified internists demonstrate their ongoing clinical expertise and judgment, their involvement in lifelong learning and quality improvement activities, and their professionalism. Because MOC involves no new training requirements and includes an assessment of a physician's actual practice, it provides a potential mechanism, heretofore untapped, of demonstrating a unique professional focus that emerges after the completion of formal training.

HOSPITALIST CERTIFICATION AND THE MOC PROCESS

As ABIM considered a separate certification pathway for hospital medicine, it faced a conundrum. The vast majority of hospitalists are general internists (most of the rest are generalists in family medicine or pediatrics) who entered hospital medicine at the completion of their internal medicine training or after a period of primary care practice. Job opportunities for hospitalists are plentiful, andexcept for additional training in quality improvement, systems leadership, care transitions, palliative care, and communication9there is little clinical rationale to prolong internal medicine training for hospitalists (some individuals may opt for fellowships to enhance their leadership skills or to launch a research career,5 but few would argue for mandatory additional clinical training in hospital medicine at this time).

So, in the absence of formal training, how could the ABIM (or other boards) recognize the focused practice of hospitalists? This question must be framed within a broader challenge: Is it possible and appropriate for certifying boards to recognize expertise and focus that is accrued not through formal training, but through actual practice experience and accompanying self‐directed learning?

In 2006, the ABIM took up this question, producing a report (New and Emerging Disciplines in Internal Medicine II [NEDIM II]) that delineated several criteria to guide whether a new field merited focused recognition through MOC (Table 1). Judging by these criteria, hospital medicine appears to be a suitable first candidate for recognition of focused practice through MOC.

PRELIMINARY THOUGHTS ON FOCUSED RECOGNITION IN HOSPITAL MEDICINE

The ABIM has endorsed the concept of recognition of focused practice in hospital medicine and charged a subcommittee (that I chair) with working out the details. It would be premature to describe the committee's deliberations in detail (particularly because the final plan needs to be approved by both the ABIM and the ABMS), but the following are some key issues being discussed.

First, demonstration of focused practice requires some minimum volume of hospitalized patients. In the absence of hard data defining a threshold number of cases for hospitalists, we are likely to endorse a number that has face validity and that reliably separates self‐identified hospitalists from nonhospitalist generalists. As with all volume requirements, we will struggle over how to handle academic physicians, physician‐administrators, and physician‐researchers who limit their overall clinical practice but who spend most of their clinical time in hospital medicine and the bulk of their nonclinical time trying to improve hospital care.

The requirements to demonstrate performance in practice and lifelong learning may be more straightforward. As with all such MOC requirements, the ABIM is increasingly looking to use real practice data, trying to harmonize its data requirements with those of other organizations such as insurers, Medicare, the Joint Commission, or for pay‐for‐performance initiatives. Despite the operational challenges, this effort is vital: for MOC (including focused recognition) to be highly valued by patients, purchasers, and diplomates, it will increasingly need to measure not only what physicians know, but also what they do.

Finally, there is the test. It is likely that a secure exam for MOC with Recognition of Focused Practice in Hospital Medicine will involve core content in internal medicine (information that every internist should know), augmented by substantial and challenging content in hospital medicine. Because it will be vital that a competent hospitalist understand key elements of outpatient practice, the exam will not be stripped of ambulatory content but will likely have fewer questions on topics that hospitalists are unlikely to confront (osteoporosis, cancer screening).

ONGOING ISSUES

As hospital medicine continues its explosive growth, it is important to develop ways to make board certification relevant to hospitalists. The ABIM believes that modifying the MOC process to recognize physicians who have focused their practice and achieved special expertise in hospital medicine is a good way to launch this effort. Ultimately, this process is likely to evolve, particularly if separate training pathways for hospital medicine emerge. For now, the development of Recognition of Focused Practice in Hospital Medicine will further legitimize the new field, provide ABIM with insights into how to recognize physicians who have advanced through practice‐based learning rather than through training, and help to guide other certifying boards (particularly family medicine and pediatrics) considering hospitalist certification. In the end, the process will need to be user‐friendly for and satisfying to diplomates, flexible enough to allow for career transitions (both toward and away from hospital medicine), and sufficiently rigorous to be credible to all stakeholders, particularly patients.

Soon after they form, most new medical fields begin agitating for a special certification, something that says, We're here, and we're different. As I've noted previously in the Journal of Hospital Medicine, the field of hospital medicine resisted this impulse in its early years, fearing that any special designation or certification would actually harm the field's growth and status.1 The concern was that managed‐care organizationsconvinced by the evidence that hospitalists improve efficiency and might improve quality2would react to any new hospitalist sheepskin by mandating that anyone providing hospital care to its covered patients have one. The backlash from primary care physicians locked out of the hospital by such a mandate would have been swift and ultimately damaging to hospitalists. In addition to these political considerations, the early field of hospital medicine lacked the academic credibility and scientific underpinning needed for specialty designation.3

Times have changed. There are now more than 15,000 hospitalists in the United States, and nearly half of American hospitals have hospitalists on their medical staffs. In many markets, including my own, hospitalists care for most internal medicine inpatients, as well as significant numbers of pediatric and surgical patients. The field has achieved academic legitimacy, with this journal, several textbooks, large and flourishing groups in every academic medical center, and several residency tracks and fellowship programs.4, 5 The Society of Hospital Medicine (SHM) has grown to more than 6000 members, become a widely respected and dynamic member of the community of professional societies, and published its core competencies.6

With this as a background, in 2004 SHM asked the American Board of Internal Medicine (ABIM) to consider a program of certification for hospitalists. As a past SHM president and now a member of the ABIM Board of Directors, I am privileged to have a bird's‐eye view of the process. In this article, I reflect on some of the key issues it raises.

THE NUTS AND BOLTS OF BOARD CERTIFICATION

Since the first board (ophthalmology) was formed in 1917, 24 specialty boards have emerged, all under the umbrella of the American Board of Medical Specialties (ABMS).7 Because no one type of physician can do it all, certifying boards have had to struggle not only with how to assess competency in existing disciplines, but with the dynamic and often controversial questions raised when new fields emerge. In the past few decades, certifying boards have grappled with specialties formed around new procedures (such as cardiac electrophysiology), discrete populations (geriatrics, palliative care), complex diseases (HIV medicine), and sites of care (intensive care medicine, emergency medicine). It is this latter category that now includes hospital medicine.

In the past, it was relatively simple for a physician to obtain board certification. Residency or fellowship training was believed to confer on its graduates the presumption of competence and professionalismthe program director's attestation served as the graduate's Good Housekeeping seal of approval. Passing the board exam was the final step, ensuring that newly minted graduates had the requisite knowledge and judgment to practice in their fields.

Remarkably, for the first half century of the specialty boards, all certifications lasted for a physician's professional lifetime. Beginning with the 1969 decision of the American Board of Family Practice to limit the validity of its certificates to 7 years, all ABMS member boards now time limit their certifications, usually to 7‐10 years.7 Of course, in an environment of rapidly changing medical knowledge and new procedures, periodiceven continuousdemonstration of competence is increasingly expected by the public.

For ABIM, the mechanism to promote lifelong learning and demonstrate ongoing competence in the face of a rapidly changing environment is known as maintenance of certification (MOC).8 Through MOC, board‐certified internists demonstrate their ongoing clinical expertise and judgment, their involvement in lifelong learning and quality improvement activities, and their professionalism. Because MOC involves no new training requirements and includes an assessment of a physician's actual practice, it provides a potential mechanism, heretofore untapped, of demonstrating a unique professional focus that emerges after the completion of formal training.

HOSPITALIST CERTIFICATION AND THE MOC PROCESS

As ABIM considered a separate certification pathway for hospital medicine, it faced a conundrum. The vast majority of hospitalists are general internists (most of the rest are generalists in family medicine or pediatrics) who entered hospital medicine at the completion of their internal medicine training or after a period of primary care practice. Job opportunities for hospitalists are plentiful, andexcept for additional training in quality improvement, systems leadership, care transitions, palliative care, and communication9there is little clinical rationale to prolong internal medicine training for hospitalists (some individuals may opt for fellowships to enhance their leadership skills or to launch a research career,5 but few would argue for mandatory additional clinical training in hospital medicine at this time).

So, in the absence of formal training, how could the ABIM (or other boards) recognize the focused practice of hospitalists? This question must be framed within a broader challenge: Is it possible and appropriate for certifying boards to recognize expertise and focus that is accrued not through formal training, but through actual practice experience and accompanying self‐directed learning?

In 2006, the ABIM took up this question, producing a report (New and Emerging Disciplines in Internal Medicine II [NEDIM II]) that delineated several criteria to guide whether a new field merited focused recognition through MOC (Table 1). Judging by these criteria, hospital medicine appears to be a suitable first candidate for recognition of focused practice through MOC.

PRELIMINARY THOUGHTS ON FOCUSED RECOGNITION IN HOSPITAL MEDICINE

The ABIM has endorsed the concept of recognition of focused practice in hospital medicine and charged a subcommittee (that I chair) with working out the details. It would be premature to describe the committee's deliberations in detail (particularly because the final plan needs to be approved by both the ABIM and the ABMS), but the following are some key issues being discussed.

First, demonstration of focused practice requires some minimum volume of hospitalized patients. In the absence of hard data defining a threshold number of cases for hospitalists, we are likely to endorse a number that has face validity and that reliably separates self‐identified hospitalists from nonhospitalist generalists. As with all volume requirements, we will struggle over how to handle academic physicians, physician‐administrators, and physician‐researchers who limit their overall clinical practice but who spend most of their clinical time in hospital medicine and the bulk of their nonclinical time trying to improve hospital care.

The requirements to demonstrate performance in practice and lifelong learning may be more straightforward. As with all such MOC requirements, the ABIM is increasingly looking to use real practice data, trying to harmonize its data requirements with those of other organizations such as insurers, Medicare, the Joint Commission, or for pay‐for‐performance initiatives. Despite the operational challenges, this effort is vital: for MOC (including focused recognition) to be highly valued by patients, purchasers, and diplomates, it will increasingly need to measure not only what physicians know, but also what they do.

Finally, there is the test. It is likely that a secure exam for MOC with Recognition of Focused Practice in Hospital Medicine will involve core content in internal medicine (information that every internist should know), augmented by substantial and challenging content in hospital medicine. Because it will be vital that a competent hospitalist understand key elements of outpatient practice, the exam will not be stripped of ambulatory content but will likely have fewer questions on topics that hospitalists are unlikely to confront (osteoporosis, cancer screening).

ONGOING ISSUES

As hospital medicine continues its explosive growth, it is important to develop ways to make board certification relevant to hospitalists. The ABIM believes that modifying the MOC process to recognize physicians who have focused their practice and achieved special expertise in hospital medicine is a good way to launch this effort. Ultimately, this process is likely to evolve, particularly if separate training pathways for hospital medicine emerge. For now, the development of Recognition of Focused Practice in Hospital Medicine will further legitimize the new field, provide ABIM with insights into how to recognize physicians who have advanced through practice‐based learning rather than through training, and help to guide other certifying boards (particularly family medicine and pediatrics) considering hospitalist certification. In the end, the process will need to be user‐friendly for and satisfying to diplomates, flexible enough to allow for career transitions (both toward and away from hospital medicine), and sufficiently rigorous to be credible to all stakeholders, particularly patients.