Patrick Mauldin, PhD

 Tertiary referral centers have created inpatient units to meet the needs of specific patient populations but sometimes are forced to place patients on other units that, although having the basic necessary skillsets for treating the patient, are not focused on that diagnosis area. The objective of this study was to look at outcomes of patients admitted to these different inpatient settings.

 Retrospective review of patient data from a single tertiary academic medical center from August 1, 2014, to June 30, 2015, comparing patients admitted to primary versus secondary inpatient services. Patients admitted to the inpatient children’s hospital, psychiatric hospital, labor and delivery unit, or subacute transitional care unit were excluded.

 Demographics of patients in the primary versus secondary units were compared to look for systematic differences between the 2 patient populations. To control for confounding variables, a gamma regression analysis was conducted for length of stay (LOS) and total cost, whereas a logistic regression was conducted for mortality.

 Admitting to the primary unit resulted in 5.5% lower observed LOS, controlling for other patient variables, but it came at a 17.8% higher total cost of care provided compared with secondary units. Mortality was also found to be lower on primary units (odds ratio, 0.864) but did not cross the threshold of statistical significance (

 Patients admitted to the primary unit had a lower LOS with higher costs of care. There was a trend toward improved mortality, although it was not statistically significant.

This is a retrospective analysis of outcomes for patients admitted to 1 tertiary academic medical center to either the primary or secondary specialized unit of their admitting diagnosis.

After controlling for patient acuity, admitting to the primary specialized unit resulted in a 5.5% lower observed length of stay in the hospital compared with other units.

Costs were found to be 17.8% higher on primary units compared with secondary units.

 = .101), mortality was lower on primary units (odds ratio, 0.864).

This study suggests that admitting patients in a referral center to an inpatient unit designed to focus on their specific disease condition might have a positive impact on their outcomes.

The decision to admit to a specific unit must be balanced with the potential for boarding of patients in noninpatient settings when inpatient capacity rates are high.

As healthcare becomes increasingly complex, patients admitted to the hospital require specialized services from both a physician and nursing perspective. Many hospitals have developed specific inpatient units to meet the needs of unique patient populations. However, with inpatient occupancy rates up in many tertiary referral centers,

 hospitals may not be able to place patients on the primary specialty unit to meet the specific needs of their medical condition. This must be balanced with the fact that increasing inpatient occupancy rates have been shown to lead to adverse outcomes.

 Therefore, patients may need to be placed on secondary specialty units to optimize capacity demands to compensate for resource constraints of the inpatient setting.

When no inpatient beds are available, patients may be delayed in the emergency department (ED), leading to the boarding of patients, which occurs when a patient has been accepted to an inpatient service but remains located in a noninpatient setting, usually in the ED. Boarding has been found to increase the length of stay (LOS) for both patients awaiting admission and those eventually discharged from the ED.

 Increased ED occupancy rates have also been found to lead to increased morbidity and mortality rates. One study found that the highest quartile of ED occupancy ratios—the total number of patient beds divided by the number of licensed ED beds&mdash;compared with the lowest quartile was associated with significantly increased odds of 1-day (odds ratio [OR], 1.42), 2-day (OR, 1.31), and 3-day (OR, 1.27) mortality.

 Another study's findings demonstrated a 3% increase in mortality with a 10% increase in ED bed occupancy, along with a 3% increase in hospital admission during a return visit to the ED.

Another patient population affected by high inpatient occupancy rates is those awaiting transfers from lower-acuity medical/surgical units to the intensive care unit (ICU) or from outside facilities. Delays in transfer to an ICU setting when clinically indicated due to a lack of inpatient capacity have been found to increase mortality.

 Conversely, a lack of medical/surgical beds may result in patients boarding in the ICU setting.

To date, no study has looked at the impact of admitting a patient based on the primary disease focus of a particular unit. There is existing evidence that trauma patients do best when cared for on a trauma unit.

 One study from a French hospital found longer LOS and higher readmission rates for patients not admitted to the primary specialty unit.

 However, no existing study has explored LOS and readmissions across all specialties in an American inpatient population. This study looked at the impact on patient outcomes based on the initial placement of a patient admitted to an inpatient setting and that placement’s impact on LOS, mortality, and total cost of care delivered.

This study is a retrospective review of all inpatient admissions to the Medical University of South Carolina Medical Center, a single tertiary academic medical center in Charleston, South Carolina, from August 1, 2014, to June 30, 2015. Institutional review board approval was obtained for this study. Patient information and all variables were extracted from the electronic health record (Epic Systems; Verona, Wisconsin). Expected LOS for patients was provided based on benchmarking data by Vizient, Inc (Irving, Texas). Patients admitted to the inpatient children’s hospital, psychiatric hospital, labor and delivery unit, or subacute transitional care unit were excluded from the analysis because no alternative inpatient units were available to them at the time of admission. Patients were considered to be admitted to a primary specialty unit versus a secondary unit based on an algorithm used by a centralized bed management team. This algorithm was agreed on by nursing and hospital leadership to ensure that the minimum core competencies required to take care of patients were also available on secondary units, even if not their primary disease focus. Although all units to which the patients were admitted had the appropriate basic nursing skillsets and technological support to meet the needs of the patients, primary specialty units had greater focus on these particular types of patients and handled a higher volume of them on a regular basis. In our institution, there are units focused on neurologic, musculoskeletal, oncology, gastrointestinal, transplant, trauma, cardiovascular, and general medicine admitting diagnoses. Any unit can be a primary or secondary unit, depending on the patient’s primary admitting diagnosis and service. Providers preferred to have their patients on these primary specialty units to help cohort their inpatient service in close proximity  and maximize their efficiencies to deliver care, but the teams of providers taking care of these patients did not alter the care delivered whether the patient was on the primary or secondary unit.

SAS version 9.4 (SAS Institute; Cary, North Carolina) was used to conduct the data analysis. Characteristics of patients admitted to the primary versus secondary units were compared to look for systematic differences between the 2 patient populations. 

 tests were used to compare differences in age; χ

 tests were used to compare categorical variables, including gender, African American race, insurance status, readmission, and mortality. Gamma regression was used for LOS, total cost, and expected LOS (

). Observed inpatient LOS, total cost, and mortality were the outcome variables of focus. Costs were based on standardized charges for services rendered and supplies used, regardless of reimbursement rates of individual patients and insurers. Confounding variables that could affect patient morbidity and acuity were factored into the analysis. These included age, gender, race, readmissions, and expected LOS. The expected LOS is a national benchmark calculated based on the discharged diagnoses of the patient as the expected time that patients with a similar presentation would stay in the hospital. To control for confounding variables of patient acuity or demographics that may impact the outcome variables, gamma regression analyses were conducted for LOS and total cost, given the propensity of positive continuous right-skewed outcomes, whereas a logistic regression was conducted for mortality (binary). Given the high correlation of age with insurance, insurance was not included in the multivariate analysis to avoid multicollinearity.

A total of 13,294 patients with 17,625 encounters were included in the study. Table 1 presents the characteristics of patients placed on their primary unit compared with patients placed on a secondary unit. Overall, patients placed on the primary unit were older (by 0.9 years), less likely to be female (49.7% vs 54.1%, respectively), and less likely to be African American (35.0% vs 42.0%). Patients placed on their primary unit also were less likely to have Medicaid and more likely to have private insurance. Readmission rates were higher for patients originally admitted to the secondary unit (15.5% vs 12.7%). The LOS, both mean and median, were found to be higher on the secondary units than the primary units (6.98 vs 6.67 and 4.25 vs 4.04 days, respectively). In addition, the mean and median costs of care were found to be higher on the primary units than on the secondary units ($23,417 vs $19,876 and $13,448 vs $11,031, respectively). There was no statistically significant difference between mortality and the expected LOS for primary and secondary unit placements.

Controlling for all other factors, the placement of a patient on the primary service was associated with a 5.5% lower observed LOS (

) compared with on the secondary service. For every 1-year increase in age, observed LOS increased by 0.2%. Each additional day in expected LOS was associated with an 11.1% higher observed LOS. Patients who had a 30-day readmission were found to have a 15.1% higher observed LOS. Female patients were found to have a 2.7% lower LOS than their comparable male counterparts. African American patients had a 11.5% higher LOS compared with non—African Americans. Increasing age (OR, 1.029; 95% CI, 1.024-1.034), expected LOS (OR, 1.073; 95% CI, 1.064-1.083), and African American race (OR, 1.253; 95% CI, 1.066-1.472) were associated with higher mortality, and female gender was associated with lower mortality (OR, 0.772; 95% CI, 0.660-0.903). The placement of a patient on a primary service was found to be associated with a lower mortality rate, although the difference did not approach statistical significance (

= .101), and a 17.8% higher total cost compared with the secondary unit, after controlling for all other confounding variables. Every year of increasing age led to a 0.2% higher cost, and those with 1 day higher expected LOS were found to have 10.3% higher costs. Female patients were found to have 7.0% lower total costs compared with males. African Americans were found to have 4.2% lower total costs than non—African Americans.

The findings of this study reveal that the placement of patients on their primary units leads to a lower LOS, after controlling for age, gender, race, expected LOS, and readmissions. Given that the same group of physicians and support services takes care of patients whether they are on the primary or secondary unit, it is possible that the specialized skillset of nurses on primary units, along with an infrastructure to support specific patient populations, may have a positive impact on improving patient care. Another potential variable not captured in the analysis is that primary units may have better familiarity and relationships with resources available for patients when they are ready to transition to the postacute care setting outside of the hospital. This may expedite the transitions of care and result in a lower observed LOS.

In addition, a noticeable difference between primary and secondary unit placement was the increased costs seen on the primary unit. Whether or not this increased amount of resources spent on patients led to improved outcomes remains to be seen. There may have been specific therapies provided on the primary units that were not available on the secondary units, leading to the difference. One other variable not factored into the total costs is the higher readmission rate on the secondary units, as reimbursement for readmissions will be reduced and eventually eliminated by Medicare and other insurers over time.

Also notable are the demographic differences, regarding age, gender, race, and insurance type, seen between the primary and secondary unit placements. It is unclear why patients with Medicaid and younger, female, and African American patients were more likely to be placed on secondary units. Although the placement of patients is not based on these factors, there may be underlying systemic influences leading to this discrepancy, such as differences in elective versus unscheduled admissions. Although it was controlled for in statistical modelling, the assignment of patients to specific units did not seem to be completely randomized.

Given the negative effects associated with placing a patient on a secondary unit, the decision on where to admit a patient when dealing with high inpatient occupancy rates must be balanced with the decision of whether to leave a patient boarding in the ED or post—anesthesia care unit or at an outside hospital. Future studies should explore whether or not allowing boarding while waiting for availability on the primary unit is better for the patient versus admitting to the secondary unit.

As this was a study conducted in a single academic medical center, the results may not be generalizable to other sites, especially those that do not specialize their medical/surgical inpatient units. Patients in this patient population may be sicker than those typically seen in community hospitals, so results in a community setting may vary. In addition, the placement of patients was based on their admitting diagnoses, which may or may not reflect the true condition or acuity of the patient. One risk of a too-low LOS is an increased rate of readmissions, whether to the same or different hospitals, and there was no way to determine if patients were readmitted to outside hospitals. The number of minority races beyond African American captured in the sample size was extremely small, making an analysis beyond the binary one presented here underpowered. This study lacked sufficient power to explore when a patient was first admitted to the secondary unit, then later transferred to the primary unit, and what effect this had on patient outcomes. Finally, patients were more likely to be placed on secondary units when the inpatient census was higher, and higher occupancy rates may confound outcomes, including higher LOS and mortality. This study was unable to attribute the inpatient occupancy rate to specific patients at the time of their admission.

Admitting patients to an inpatient hospital unit that has specialized services to focus on each patient’s disease process has been shown to lower the observed LOS with a trend toward improved mortality. The trade-off in these improved outcomes is a higher total cost for the care provided. In the future, studies of patient placement should focus on a comprehensive model to meet the goals of the triple aim of improving the patient’s experience and improving the health of the population, all while reducing the cost per capita.&ensp;

The authors would like to acknowledge Amy Wilson, PhD, and Anna Ford for the initial collection and analysis of the data.

Indiana University Health, South Central Region (DAH), Bloomington, IN, Medical University of South Carolina (ZS, NH, PM), Charleston, SC.

 The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

 Concept and design (DAH, ZS, NH, PM); acquisition of data (DAH); analysis and interpretation of data (DAH, ZS, PM); drafting of the manuscript (DAH, ZS, PM); critical revision of the manuscript for important intellectual content (DAH, NH); statistical analysis (DAH, ZS, PM); and administrative, technical, or logistic support (NH).

 Daniel A. Handel, MD, MBA, MPH, Indiana University Health, South Central Region, 601 W Second St, PO Box 1149, Bloomington, IN 47402. Email: dhandel1@iuhealth.org.

1. Halpern NA, Pastores SM. Critical care medicine in the United States 2000-2005: an analysis of bed numbers, occupancy rates, payer mix, and costs. 

2010;38(1):65-71. doi: 10.1097/CCM.0b013e3181b090d0.

2. Boyle J, Zeitz K, Hoffman R, Khanna S, Beltrame J. Probability of severe adverse events as a function of hospital occupancy. 

2014;18(1):15-20. doi: 10.1109/JBHI.2013.2262053.

3. Chrusch CA, Olafson KP, McMillan PM, Roberts DE, Gray PR. High occupancy increases the risk of early death or readmission after transfer from intensive care. 

2009;37(10):2753-2758. doi: 10.1097/CCM.0b013e3181a57b0c.

4. Schilling PL, Campbell DA Jr, Englesbe MJ, Davis MM. A comparison of in-hospital mortality risk conferred by high hospital occupancy, differences in nurse staffing levels, weekend admission, and seasonal influenza. 

2010;48(3):224-232. doi: 10.1097/MLR.0b013e3181c162c0.

5. Wise J. Cutting medical bed occupancy is linked to reduced patient mortality in UK hospital, study shows. 

6. Madsen F, Ladelund S, Linneberg A. High levels of bed occupancy associated with increased inpatient and thirty&#8209;day hospital mortality in Denmark. 

2014;33(7):1236-1244. doi: 10.1377/hlthaff.2013.1303.

7. McCarthy ML, Zeger SL, Ding R, et al. Crowding delays treatment and lengthens emergency department length of stay, even among high-acuity patients. 

2009;54(4):492-503.e4. doi: 10.1016/j.annemergmed.2009.03.006.

8. Jo S, Jin YH, Lee JB, Jeong T, Yoon J, Park B. Emergency department occupancy ratio is associated with increased early mortality. 

2014;46(2):241-249. doi: 10.1016/j.jemermed.2013.05.026.

9. McCusker J, Vadeboncoeur A, Lévesque JF, Ciampi A, Belzile E. Increases in emergency department occupancy are associated with adverse 30-day outcomes. 

2014;21(10):1092-1100. doi: 10.1111/acem.12480.

10. Robert R, Reignier J, Tournoux-Facon C, et al; Association des Réanimateurs du Centre Ouest Group. Refusal of intensive care unit admission due to a full unit: impact on mortality. 

2012;185(10):1081-1087. doi: 10.1164/rccm.201104-0729OC.

11. Lloyd JM, Elsayed S, Majeed A, et al. The practice of out-lying patients is dangerous: a multicentre comparison study of nursing care provided for trauma patients. 

2005;36(6):710-713. doi: 10.1016/j.injury.2004.11.006.

12. Stowell A, Claret PG, Sebbane M, et al. Hospital out-lying through lack of beds and its impact on care and patient outcome. 

2013;21:17. doi: 10.1186/1757-7241-21-17.

Articles

Placement of patients in an inpatient hospital setting is associated with lower length of stay and mortality at the expense of higher costs.

Inpatient Placement: Associations With Mortality, Cost, and Length of Stay

Nursing home (NH) patients are frequently transferred to emergency departments (EDs) and/or hospitalized in situations in which transfer might have been avoided. This study describes the frequency of NH transfers for ambulatory care—sensitive conditions (ACSCs) and estimates associated expenditures.

Retrospective cohort study of 62,379 NH patients with Medicare coverage receiving care in South Carolina between 2007 and 2009.

Subjects were analyzed to determine the frequency of acute ED or hospital care for conditions. Comparison is made to similar patients transferred for acute treatment of non-ACSCs. Generalized linear models were used to estimate the costs attributable to treating ACSCs.

Over 3 years, 20,867 NH subjects were transferred from NHs to acute care facilities, and 85.3% of subjects had at least 1 episode of care for an ACSC. An average of 13,317 subjects per year were transferred for an average of 17,060 episodes of ED or hospital care per year between 2007 and 2009. More ACSC patients transferred to EDs were subsequently admitted to the hospital (50.4% vs 25%; P <.0001). In adjusted analyses, mean ED costs per episode of care ($401 vs $294; P <.0001) were higher, but mean hospitalization costs per episode of care were lower ($8356 vs $10,226; P <.0001) for ACSC patients compared with non-ACSC patients.

A significant proportion of Medicare NH patients are treated acutely for ACSCs, which are associated with higher healthcare utilization and costs. Better access to onsite evaluation might enable significant cost savings and reduce morbidity in this population.

Nursing home patients are frequently transferred to emergency departments (EDs) or admitted to acute care facilities for ambulatory care—sensitive conditions (ACSCs) in situations in which more timely access to onsite primary care services might have diminished the need for transfer.

		85.3% of transferred patients had at least 1 ACSC diagnosis.

		50.4% of ACSC patients seen in EDs were subsequently admitted to the hospital (vs 25% for patients without ACSCs, P <.0001).

		Mean ED costs per episode of care ($401 vs $294; P <.0001) were higher for ACSC patients, but mean hospitalization costs per episode of care were lower ($8356 vs $10,226; P <.0001).

Millions of Americans receive important, but costly, long-term and rehabilitative care in US nursing homes (NHs) each year. Approximately 1.8 million Americans live in NHs or other skilled nursing facilities, and this number is expected to double in the next 40 years.

In South Carolina alone, approximately 16,000 residents live in more than 170 NH facilities.

In addition, up to 4.9 million Medicare recipients receive rehabilitative care each year in US NHs following acute hospitalization.

 Medicare, the largest payer for rehabilitation care, spent an estimated $288 million for post acute rehabilitation and hospice care in 2007.

NH patients are frequently transferred to acute hospital emergency departments (EDs) and subsequently hospitalized.

 Hospitalization often results in iatrogenic complications and declining health trajectories among elderly patients.

In a recent study, nearly one-fourth of long-term NH residents experienced at least 1 ED visit in 6 months, and more than 60% of ED visits resulted in hospitalization.

Overall, reported annual hospital admission rates for long-term care NH patients range from 9% to 59%.

Similarly, 23.5% of Medicare patients discharged to skilled nursing facilities are readmitted to acute hospitals within 30 days, resulting in an estimated $4.3 billion in excess costs.

An extensive and varied literature exists describing factors associated with NH to ED transfers and hospitalizations.

 Identified predictors include sociodemographic factors, such as male gender and increasing age; medical conditions including congestive heart failure, chronic respiratory disease, and dementia; provider-specific factors related to the availability of nurses, physicians, and nurse practitioners; and regulatory factors such as reimbursement rates and bed-hold policies.

When surveyed, providers have indicated several factors that contribute to overhospitalization.

Some of these include patient and family preferences; lack of knowledge on the part of patients and families regarding end-of-life care options such as hospice care, lack of familiarity with patients by covering providers, and lack of timely (<4 hours) access to physician or nurse practitioner (NP) evaluation. Acknowledgment of concern for medico-legal liability is largely absent from the literature, but this may still be a factor in decisions to hospitalize.

Many patients, providers, and policy makers feel that a significant proportion of NH-to-hospital transfers are potentially avoidable. Indeed, the preventability of such episodes of care has been the subject of much study and debate, but there is no gold standard for determining preventability.

 In 1 systematic review, low-quality inpatient care was associated with a 55% increase in risk for early hospital readmission.

However, estimates of hospitalization preventability range from 9% to 48%.

Within this context, it is appealing to analyze a set of diseases for which improvements in care might prevent hospitalization.

Ambulatory care—sensitive conditions (ACSCs) are those for which hospital admission could potentially be prevented by timely primary, or ambulatory, care intervention. According to the Agency for Healthcare Research and Quality (AHRQ): “Hospitalization for an ambulatory care-sensitive condition is considered to be a measure of access to appropriate primary healthcare. While not all admissions for these conditions are avoidable, it is assumed that appropriate ambulatory care could prevent the onset of this type of illness or condition, control an acute episodic illness or condition, or manage a chronic disease or condition.”

For example, with early evaluation, a patient with cellulitis might receive oral antibiotics that could avoid a course of intravenous antibiotics for which hospitalization is necessary. Similarly, a patient with chronic heart failure receiving early dietary modification and diuretic therapy might avoid hospitalization for volume overload and pulmonary edema.

Thus, ACSC analysis represents a useful method for identifying potentially avoidable acute care utilization based on the assumption that facilities with disproportionately high rates have problems with timely access to appropriate primary care.

 Several reports indicate that avoidable hospitalizations for ACSCs are common and costly.

 Carter and colleagues examined an elderly Massachusetts cohort and found that 20.3% of ED visits among NH residents were for ACSCs and that NH residents had a relative risk ratio for hospital admission of 2.23 (95% CI, 1.744-2.855) compared with community-dwelling patients. In addition, acute care costs for ACSCs in NH patients are quite high, estimated, for example, at 23% of the $971 million spent on NH residents in New York state.

The current literature on ACSC in NH patients has several limitations. Most studies have analyzed ED visit rates or hospitalizations, but not both.

 Also, relatively few studies have analyzed costs of care for ACSCs among nursing home patients.

Finally, a greater number of robust studies are needed to more accurately analyze the actual costs attributable to ACSCs in NH patients. Such figures can be used to estimate cost savings based on the projected or actual effect sizes of different interventions designed to reduce NH transfers. The purpose of this project was to determine the frequency of ED visits and hospitalizations for ACSCs among NH patients and to determine Medicare expenditures for these patients.

We constructed a statewide cohort of Medicare-insured patients receiving care in South Carolina NHs during calendar years 2007 to 2009. Data sources included Medicare Provider Analysis and Review (MedPar), Medicare Outpatient Carrier, and Medicare Annual Beneficiary Summary files from CMS. Data sets were linked using a unique patient identifier code. Selection criteria included: 1) age ≥18 years; 2) enrolled in Medicare; and 3) billed for an acute hospitalization, ED visit, long-term NH stay, or inpatient rehabilitation in a South Carolina NH. Subjects were included in the present cohort if they had a NH admission and discharge date within the study time frame. We categorized subjects as having had acute ED visits or hospitalizations if they had encounters that fell during the period of a NH stay. Episodes of acute care were further categorized as ED-only visits, ED-hospitalization visits, or hospitalization-only visits. This study was reviewed and approved by our local institutional review board. All analyses were conducted using SAS version 9.3 (SAS Institute, Cary, North Carolina).

ACSCs have been used in numerous prior health services research studies examining access to primary care services, including studies of NH and skilled care populations.

For this study, we used a modification of the coding scheme for ACSCs recommended by the AHRQ.

 Subjects were categorized as having been treated for ACSCs diagnoses or not. Ambulatory care-sensitive conditions and relevant 

International Classification of Diseases,

. Given the age range of our population, we excluded congenital syphilis (

 090.x), hemophilus meningitis age 1-5 only (

 780.3), and failure to thrive age <1 year (

 783.4). We further categorized ACSCs as “acute and/or preventable” versus “chronic.” Finally, given the nature of our research question and in order to better characterize total hospital costs attributable to ACSCs, we examined subjects with ACSCs as either a primary or secondary diagnosis.

The primary outcomes for this study were measures of acute healthcare utilization and associated hospital costs for patients treated for ACSCs compared with those treated acutely for non-ACSCs. We determined both the number of subjects treated and episodes of care per year for each ACSC. We further determined the mean number of ED visits and hospitalizations per subject and the overall number of visits. The primary cost outcome for adjusted analyses was mean Medicare expenditures for ED visits and/or hospitalization.

In addition to treatment for ACSCs, we also examined several other covariates. Age and length of NH stay were analyzed as continuous variables. Gender (male/female) and race (white/other) were analyzed as categorical variables. Baseline comorbidity was analyzed using chronic condition flags present in the Medicare Annual Beneficiary Summary file, provided the date of first diagnosis occurred prior to the index NH stay. These comorbidities were: acute myocardial infarction/ischemic heart disease, Alzheimer’s disease/dementia, atrial fibrillation, cataract, chronic kidney disease, chronic obstructive pulmonary disease (COPD), congestive heart failure, diabetes mellitus (DM), depression, osteoporosis, stroke, and cancer.

Descriptive statistics (means, medians, and proportions) of demographic variables, length of hospital stay, chronic conditions present in beneficiary summary file prior to the index episode, and acute care utilization (eg, hospital readmission, intensive care unit [ICU] admission, etc) were calculated by ACSC status (with vs without an ACSC). We also calculated the frequency of episodes of care per year stratified by chronic and acute preventable conditions. In order to determine costs attributable to the care of ACSC adjusted for relevant covariates described above, we fitted a generalized linear model estimated via generalized estimating equations and a gamma distribution with a log link. This approach enabled us to estimate costs with robust standard error estimates for making inferences regarding expenditures. Adjusted mean costs are reported in US dollars calculated from the model-predicted costs, with all dollar values adjusted to 2009 dollars using the Consumer Price Index.

This cohort represented a 100% sample of Medicare recipients treated in South Carolina NHs during calendar years 2007 to 2009. 

 describes characteristics of NH subjects treated in EDs or hospitalized for ACSCs compared with those treated for non-ACSCs. Overall, 17,794 of 20,867 transferred patients (85.3%) were treated for at least 1 ACSC condition as a primary or secondary diagnosis during an episode of acute care. Mean age, the proportion of female subjects, and the proportion of nonwhite subjects were similar between groups. NH length of stay and the proportion of long-stay (>30 days) NH patients were higher among subjects in the ACSC group. Both groups had significant numbers of subjects with preexisting chronic diseases, but chronic ACSCs including congestive heart failure (69.2% vs 54.7%), COPD (47.7% vs 37.6%), and DM (56.5% vs 39.9%) were higher in the ACSC group compared with the non-ACSC group.

Table 1 describes the types of treated ACSCs grouped according to acute and preventable conditions or chronic conditions. Over the 3-year study period, 4680 patients were treated in the ED and/or hospital for a total of 5433 episodes of acute and preventable conditions. Dehydration/volume depletion and kidney/urinary tract infections were the most frequent acute ACSCs treated. More patients were seen for chronic ACSCs than acute/preventable, with 8637 subjects treated for a total of 11,627 episodes of care per year. Hypertension, DM, and congestive heart failure were the most frequent chronic conditions treated.

 displays several domains of acute ED and hospital utilization in ACSC subjects compared with non-ACSC subjects. During the 3-year study period, the 20,867 NH patients experienced 27,382 episodes of care for ACSC compared with only 7200 episodes of care without ACSC. A significantly higher proportion of NH patients seen in the ED with ACSC diagnoses were subsequently admitted to the hospital (50.4% vs 24%; 

 <.0001). NH-to-hospital transfers representing hospital readmissions were not significantly different between the ACSC and non-ACSC condition groups (15.1% vs 16%; 

 = .26). ICU admission was almost twice as common in the ACSC condition group (10.2% vs 5.6%; 

<.0001). Patients treated for ACSC diagnoses had higher mean numbers of ED visits per subject (1.5 vs 1.2; 

<.0001) and hospitalizations per subject (1.3 vs 1.1; 

<.0001) than did non-ACSC subjects, with the total number of ED visits and hospital visits occurring in similar proportions.

 highlights estimated costs for ED care and hospitalization among NH subjects transferred for ED care and hospitalization for ACSCs compared with non-ACSCs after adjustment for demographic factors and comorbid chronic diseases. Adjusted mean ED costs were approximately $100 more ($401 vs $294; 

 <.0001) for episodes of ED care for ACSC patients compared with non-ACSC patients. However, adjusted mean hospitalization costs were approximately $1900 less ($8356 vs $10,226; 

 <.0001) for ACSC patients compared with non-ACSC patients. Given the markedly higher numbers of ACSC subjects, total annual expenditures for ACSCs were significantly higher in each category. Model coefficients are displayed in 

This report represents the first study to analyze ED and acute hospital utilization in a Medicare cohort at the state level while using robust methods for estimating attributable healthcare costs. Our results confirm that a majority of patients transferred from NHs to hospitals for ED care and/or hospitalization are treated for ACSCs as a primary or secondary diagnosis. ACSC patients appear more likely to be admitted to the hospital from the ED and more likely to be admitted to the ICU, based on bivariate analyses. In fully adjusted regression models, overall utilization and Medicare expenditures are higher for patients with ACSCs transferred from NHs to EDs than for patients without ACSCs. Total annual costs were significantly higher in ACSC patients compared with non-ACSC patients. This is important because prior studies have suggested that early access to primary and preventive care may obviate or diminish the need for acute care in patients with these conditions. Clearly, most acute care transfers are likely appropriate, especially in instances of critical illness. However, our results suggest that additional measures to improve onsite management of ACSCs might prevent avoidable NH-to-acute care transfers among stable patients and/or decrease the severity of illness among patients destined to require acute care transfers. Both of these phenomena could have favorable effects on healthcare utilization and expenditures.

Previously successful strategies to reduce NH-to-ED transfers and hospitalizations have focused on improving access to care through innovative payment policies, improving care quality for specific diseases, and increasing referrals for hospice care.

 For example, Casarett and colleagues (2005) introduced a simple communication intervention to identify NH patients and families whose care preferences were congruent with palliative approaches to care.

Patients randomized to this intervention had higher hospice enrollment rates, half as many acute hospitalizations, and fewer days spent in acute care hospitals. Kane and colleagues (2004) described the effects of an innovative Medicare payment program that increased the availability of NPs to NH residents.

 Intervention patients had half as many hospitalizations (2.43 per 100 patients per month vs 4.67 per 100 patients per month; 

 <.001) compared with control patients at other facilities, as well as 9% lower hazard for mortality compared with non-intervention patients at the same facilities. The program was estimated to save $103,000 per year per NP.

Regarding clinical management interventions, Loeb and colleagues (2006) demonstrated in a cluster-randomized trial that early initiation of a clinical pathway for onsite treatment of pneumonia resulted in a 12% absolute reduction in hospitalizations (95% CI, 5%-18%; 

<.001) and a trend toward decreased mortality. This intervention was also evaluated to be cost-effective, estimated to save $1016 per resident (95% CI, $207-$1824).

Unfortunately, the innovations described here have not been widely disseminated, and many NH professionals still regard access to timely, well-informed evaluations for acutely ill residents as a significant limitation in most nursing homes.

Many have called for new and effective strategies to reduce NH transfers and hospitalizations given that prior efforts at NH regulation and market-based incentives have had only modest effects on quality metrics.

This report should be evaluated in light of its limitations. First, our analysis involves a single state, South Carolina, and the observed frequencies and costs of acute care for ACSCs may not be generalizable nationally. However, based on 2006-2007 data from the Commonwealth Fund, South Carolina ranked 28th in the nation in Medicare hospital admissions per 100,000 beneficiaries, near the median national rate.

Second, our data set did not include information on costs related to ambulance transportation. Grunier and colleagues reported that in 1 cohort of long-term care patients, more than 90% of patients transferred to the ED-required ambulance transport.

Thus, our results may underestimate total costs associated with ED and hospital transfers, especially for Medicare patients. We analyzed costs in aggregate without breaking them down into relevant cost centers. Thus, we are unable to comment on, for example, the relative contribution of radiographic tests versus laboratory testing or pharmacy costs. Finally, there were baseline differences in the prevalence of several chronic conditions between patients in the ACSC condition group and the non-ACSC condition group. It is possible that differences in severity of illness influenced patients’ clinical conditions and/or provider decision making during episodes of acute illness. Nevertheless, non-ACSC subjects had high levels of chronic disease, including heart disease, dementia, chronic kidney disease, depression, stroke, and cancer.

This study has relative strengths as well. We were able to examine a large cohort and to outline frequency and attributable costs for the acute care of ACSCs in Medicare. Our cost estimates include both ED and inpatient costs and represent actual payments received by Medicare, preferable to examining hospital charges as has been done in previous reports.

Our findings have several implications for healthcare policy makers and intervention developers. CMS, in provisions related to the Affordable Care Act (ACA), has recently encouraged the development of community-based care transitions programs, as well as pilot initiatives, specifically focused on reducing avoidable NH transfers.

 Section 3025 of the ACA established penalties for hospitals with higher-than-expected risk-adjusted hospital readmission rates for congestive heart failure, an ACSC, and COPD was added to this list in 2014.

Finally, several bundled care demonstration projects are under way for posthospital care that aim to improve care transitions for Medicare enrollees.

All of these endeavors can benefit from reliable estimates of cost attributable to ACSCs among NH residents.

In considering potential interventions to reduce acute care transfers, it is plausible to hypothesize that real-time video teleconferencing might overcome barriers that restrict access to onsite physician and NP providers in NH settings. This redesign of the delivery system could be easily coupled with decision support similar to successful clinical management pathways for ACSC conditions as in prior clinical trials.

Wade and colleagues (2010) reviewed 36 trials utilizing real-time video telehealth interventions in a variety of settings, and the majority of these interventions offered health outcomes superior or equivalent to those handled in person, and they were deemed cost-saving or cost-neutral.

Real-time video telehealth has been used successfully in rural settings to increase subspecialty physician access, as well as in the home setting to improve chronic disease management.

Daly and colleagues (2005) demonstrated the feasibility of establishing real-time videoconferencing in NHs, but no trial to date has assessed the effectiveness of this type of intervention.

 One barrier to implementation of telehealth and other interventions relates to expected cost outlays and estimating potential cost savings. Our analysis should prove useful to intervention developers in this respect.

In conclusion, we observed that a majority of Medicare patients transferred from NH to acute-care settings had at least 1 ACSC, and associated costs were substantial.

Ralph H. Johnson VA Medical Center, Health Equity and Rural Outreach Innovations Center (RNA), Charleston, NC; Division of General Internal Medicine and Geriatrics (RNA, JZ, PM, WPM) and Department of Public Health Services (MG), the Medical University of South Carolina, Charleston, SC; Clemson University School of Nursing, Clemson, SC, and Health Sciences South Carolina (JC), Columbia, SC.

This project was supported by the South Carolina Clinical & Translational Research Institute, with an academic home at the Medical University of South Carolina, through NIH grant numbers UL1 RR029882 and UL1 TR000062.

The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Concept and design (RNA, JC, MG, PM); acquisition of data (RNA, JC); analysis and interpretation of data (RNA, JC, MG, WPM, JZ); drafting of the manuscript (RNA, MG, JZ); critical revision of the manuscript for important intellectual content (RNA, MG, WPM, PM); statistical analysis (JC, MG, JZ); provision of study materials or patients (JC); obtaining funding (RNA, JC, WPM, PM); and supervision (RNA).

 R. Neal Axon, MD, MSCR, Ralph H. Johnson VAMC, 109 Bee St, Charleston, SC 29401. E-mail: 

1. Kaye HS, Harrington C, LaPlante MP. Long-term care: who gets it, who provides it, who pays, and how much? 

This article outlines the frequency of transfers of patients with ambulatory care-sensitive conditions from nursing homes to emergency departments or hospitals, and provides reliable estimates of associated costs.

Patrick Mauldin, PhD

Articles

Inpatient Placement: Associations With Mortality, Cost, and Length of Stay

Frequency and Costs of Hospital Transfers for Ambulatory Care-Sensitive Conditions