eAppendix A1. Top 5 Diagnosis Code Categories of Inpatient and Observation Patient Stays (N?=?43?853)
eAppendix A2. Top 5 Diagnosis Code Categories of Inpatient and Observation Patient Stays by Service (N?=?43?853)
eAppendix B1. Avoidable Days of Inpatient and Observation Patient Stays (N?=?43?853)
eAppendix B2. Avoidable Days of Inpatient and Observation Patient Stays by Service (N?=?43?853)
Customize your JAMA Network experience by selecting one or more topics from the list below.
Sheehy AM, Graf B, Gangireddy S, et al. Hospitalized but Not Admitted: Characteristics of Patients With “Observation Status” at an Academic Medical Center. JAMA Intern Med. 2013;173(21):1991–1998. doi:10.1001/jamainternmed.2013.8185
The Centers for Medicare & Medicaid Services (CMS) defines observation status for hospitalized patients as a “well-defined set of specific, clinically appropriate services,” usually lasting less than 24 hours, and that in “only rare and exceptional cases” should last more than 48 hours. Although an increasing proportion of observation care occurs on hospital wards, studies of patients with observation status have focused on the efficiency of dedicated units.
To describe inpatient and observation care.
Design and Setting
Descriptive study of all inpatient and observation stays between July 1, 2010, and December 31, 2011, at the University of Wisconsin Hospital and Clinics, a 566-bed tertiary academic medical center.
All patients with observation or inpatient stays during the study period.
Main Outcomes and Measures
Patient demographics, length of stay, difference between cost and reimbursement per stay, and percentage of patients discharged to skilled nursing facilities.
Of 43 853 stays, 4578 (10.4%) were for observation, with 1141 distinct diagnosis codes. Mean observation length of stay was 33.3 hours, with 44.4% of stay durations less than 24 hours and 16.5% more than 48 hours. Observation care had a negative margin per stay (−$331); the inpatient margin per stay was positive (+$2163). Adult general medicine patients accounted for 2404 (52.5%) of all observation stays; 25.4% of the 9453 adult general medicine stays were for observation. The mean length of stay for general medicine observation patients was 41.1 hours, with 32.6% of stay durations less than 24 hours and 26.4% more than 48 hours. Compared with observation patients on other clinical services, adult general medicine had the highest percentage of patients older than 65 years (40.9%), highest percentage female patients (57.9%), highest percentage of patients discharged to skilled nursing facilities (11.6%), and the most negative margin per stay (−$1378).
Conclusions and Relevance
In an academic medical center, observation status for hospitalized patients differed markedly from the CMS definition. Patients had a wide variety of diagnoses; lengths of stay were typically more than 24 hours and often more than 48 hours. The hospital lost money, primarily because reimbursement for general medicine patients was inadequate to cover the costs. It is uncertain what role, if any, observation status for hospitalized patients should have in the era of health care reform.
Hospitalizations with “observation status” are increasingly important in the United States. Quiz Ref IDUse of observation hospital services increased 26% for Medicare beneficiaries from 2006 through 2008; inpatient stays decreased by 4% in a similar period.1 Observation length of stay (LOS) rose from 26.2 to 28.2 hours from 2007 through 2009.2
Traditionally, observation care has been delivered in observation units for a limited number of short-stay patients with well-defined diagnoses. Current literature focuses almost exclusively on the efficiency of such dedicated units.3-5 However, over the past decade, federal legislation has reshaped observation status, shifting more observation care to hospital wards.6,7 This trend is important because observation status is not synonymous with inpatient admission.
The Centers for Medicare & Medicaid Services (CMS) defines observation status8-10 as follows8:
…a well-defined set of specific, clinically appropriate services, which include…treatment, assessment, and reassessment before a decision can be made regarding whether patients will require further treatment as hospital inpatients or if they are able to be discharged from the hospital…(and) in the majority of cases, the decision…can be made in less than 48 hours, usually in less than 24 hours. In only rare and exceptional cases do…outpatient observation services span more than 48 hours.
Anecdotal reports suggest that “hospitalization without admission” confuses both patients and health care providers and may create substantial financial issues.11-13 For Medicare patients, failing to meet admission criteria means that Medicare Part A will not cover care because observation care is “outpatient.” Although Medicare Part B and some private insurance will cover observation hospital services, Part B may have a higher deductible than Part A depending on the services delivered. Medicare Part B also has an additional 20% copay and does not cover inpatient pharmacy charges.9,14 In addition, observation days do not count toward the 3-day prequalifying stay that Medicare requires for patients needing skilled nursing facility care on discharge.9 In summary, when inpatient admission criteria are not satisfied, patients with observation status may face greater hospital and skilled nursing facility charges.
Because the actual impact of observation status on patient care has not been fully investigated,2,6 we conducted a retrospective descriptive study of all observation status and inpatient stays at the University of Wisconsin Hospital and Clinics from July 1, 2010, to December 31, 2011. We sought to describe observation clinical practice and to investigate the cost implications of inpatient and observation care.
The University of Wisconsin–Madison Health Sciences institutional review board approved the study. We retrospectively analyzed all observation and inpatient stays with encounter start dates during the 18-month study period.
The University of Wisconsin Hospital is a 566-bed tertiary care, academic medical center located in Madison, Wisconsin.15 More than 50% of patients have primary residence outside Dane County, where the hospital is located. The hospital does not have a dedicated observation unit; observation care occurs on general hospital wards.
Data were obtained from the institutional electronic medical record and ancillary clinical and administrative data systems, including the billing, case management documentation, and cost accounting systems. Staff of the business planning and analysis department deidentified patient data.
Observation status was assigned to patients by case managers using McKesson InterQual (McKesson Corporation), a manual also used by CMS to determine status.16,17 A minority (1174 of 43 853 [2.7%]) of encounters had a status change, with 93% of status changes occurring prior to discharge. These patients were classified based on their final status designation, with inpatient status changed to observation status (628 of 1174 [53.5%]) classified as observation, and observation status changed to inpatient status (546 of 1174 [46.5%]) classified as inpatient.
We first stratified patient encounters by observation or inpatient stay and then by the service responsible for the encounter. Services were categorized as adult general medicine (general internal medicine and family medicine), adult subspecialty nonsurgical (ie, cardiology, neurology), adult surgical (ie, general surgery), or pediatrics (all patients younger than 18 years).
We then obtained patient demographic information, characteristics of hospitalization, and financial data. Demographics included sex, age, and percentage older than 65 years. Characteristics of encounters included whether the encounter was for immediate care or unscheduled (acute/unscheduled [defined as a request for hospitalization occurring within 24 hours of encounter start]), day of the week stay began, length of stay, International Classification of Diseases, Ninth Revision (ICD-9) codes, and “avoidable days.” Avoidable days is an institutional term used to describe days that do not result from medical needs and are logged each day by case managers. Avoidable days are grouped in 3 categories: hospitalization not appropriate (ie, criteria for hospitalization not met); hospital cause for delay (ie, delay in test performed); or discharge delayed (ie, accepting facility requests delay). Repeat encounters for either an observation or inpatient stay during the study period were recorded, and discharges to skilled nursing facilities were documented. We also recorded primary insurance type, institutional billing department formulary costs per encounter, and reimbursement recovered.
A total of 43 853 hospitalizations occurred over the 18-month study. Of these, 39 275 (89.6%) were inpatient and 4578 (10.4%) were observation (Table 1). Of the 4578 observation encounters, 2404 (52.5%) were for adult general medicine service. Adult general medicine encounters also had the highest percentage of observation stays (2404 of 9453 [25.4%]) (Table 2).
Overall, 51.4% (2353 of 4578) of observation patients and 47.1% (18 514 of 39 275) of inpatients were women (Table 1). Adult general medicine observation patients and inpatients were more likely to be female, with all other categories having less than 50% female patients. Adult general medicine had the most observation patients and inpatients older than 65 years (Table 2).
Observation care was more frequently acute/unscheduled than inpatient care, and observation encounters more commonly started on weekends (Table 1). The majority of adult general medicine observation and inpatient encounters were acute/unscheduled. Observation encounters for adult subspecialty nonsurgical, adult surgical, and pediatrics all were more likely to be acute/unscheduled than their corresponding inpatient stays (Table 2).
Repeat stays accounted for 8.0% (366 of 4578) of observation encounters and 23.8% (9353 of 39 275) of inpatient stays (Table 1). Adult general medicine observation stays were most likely to be repeat encounters (Table 2).
Mean (SD) observation LOS was 33.3 (29.5) hours; 756 of 4578 observation encounters (16.5%) had LOS greater than 48 hours (Table 1). Adult general medicine observation encounters had the longest LOS. Less than a third (783 of 2404 [32.6%]) of encounters ended in discharges in under 24 hours, and 634 of 2404 (26.4%) had stays greater than 48 hours. Mean LOS was longer than the median in all groups, reflecting long LOS outliers for both observation and inpatient care (Table 2).
A total of 7.0% (320 of 4578) observation and 13.7% (5381 of 29 275) inpatient encounters concluded with patients being discharged to skilled nursing facilities (Table 1). Adult general medicine inpatient and observation stays had the highest percentage of discharges to skilled nursing facilities (Table 2).
Quiz Ref IDThe cost per encounter for observation care was less than for inpatient care. However, on average, reimbursement was insufficient to cover the cost of observation care. The net per-encounter loss for observation care was $331 (−8.7%) compared with a net gain for an inpatient stay of $2163 (+10.9%). There was a net loss of $9.94 per observation hour, compared with a net gain of $16.65 per inpatient hour (Table 1).
Of the 4 service groups, adult general medicine observation stays had the most negative margin. The net loss was $1378 per stay (−33.1%), or $33.53 per hour (Table 2). The net loss for pediatric observation stays was $66 (−2.3%) per encounter. In contrast, the net gain for adult subspecialty nonsurgical observation stays was $616 (18.6%) per encounter; for adult surgical observation stays, the net gain per stay was $1781 per encounter (46.4%).
A total of 51.8% (2372 of 4578) of observation and 48.8% (19 181 of 39 275) of inpatient encounters were covered by government insurance (Table 1). Adult general medicine observation encounters had more patients with government insurance and fewer patients with commercial insurance compared with the other service groups. Pediatrics observation encounters had the most commercial payers (Table 2).
Quiz Ref IDChest pain was the top observation diagnosis, which accounted for 12.1% of stays, followed by abdominal pain (3.7%) and syncope and collapse (3.0%). (Table 3; see also eAppendix A1 and A2 in the Supplement). There were 1141 distinct observation diagnosis codes. The top diagnosis for adult general medical observation stays was also chest pain (15.2%). There were 584 distinct adult general medical observation diagnosis codes. The other top diagnoses for observation stays were chest pain (24.4%) for the adult subspecialty nonsurgical service group, observation following other accident (3.0%) for the adult surgical service group, and abdominal pain (4.3%) for the pediatric service group (Table 4).
There were 3.8% of inpatient and 3.6% of observation encounters with avoidable days. For inpatients and observation patients, the top avoidable day category was discharge delay. The top individual cause of avoidable days was criteria for hospitalization not met (Table 3; see also eAppendix B1 and B2 in the Supplement). Of the 4 service groups, stays for adult general medicine observation and inpatient care had the most avoidable days (Table 4).
Quiz Ref IDIn a single-center study of patients with observation status at an academic medical center, more than half of all observation stays were on the adult general medicine service. Compared with observation patients on other services, the general medicine patients had longer LOS. They were also more likely to be discharged to a skilled nursing facility, to have more avoidable days, to have more acute/unscheduled admissions, and to have more repeat encounters. Adult general medicine observation patients were also most likely to have government insurance, and so they were potentially vulnerable to self-pay skilled facility costs and a greater out-of-pocket hospital bill than were inpatients for a given set of services. The result is financial hardship for many general medicine patients and substantial fiscal losses for hospitals and health care providers.
We found that many observation stays did not meet the CMS definition of observation, even when the CMS-endorsed InterQual criteria were used to determine status. First, CMS states that observation care should typically require less than 24 hours and only rarely last more than 48 hours, yet only 44.4% of patients with observation status were discharged in less than 24 hours, and 16.5% stayed more than 48 hours. In the adult general medicine service group, 26.4% of observation patients stayed more than 48 hours. Although our typical observation LOS may be longer than those at hospitals that do not provide tertiary care, “23-hour observation” is no longer the norm, and stays of more than 48 hours are no longer “rare and exceptional.” In addition, of 4578 observation stays, there were 1141 distinctly billed observation codes. Chest pain, the most common diagnosis, accounted for just 12.1% of observation stays. This wide variety of diagnoses, combined with complicated InterQual criteria, suggest that observation status is not well defined.
We had several unexpected financial findings. The cost for observation care was less than the cost for inpatient care, but reimbursement was markedly lower, resulting in operating losses and the transfer of some costs to patients. This finding is important in the context of savings typically reported with dedicated observation units. A recent review reported lower costs of observation unit care but did not report the actual reimbursement, an important metric when discussing hospital savings.5 In addition, in all but 2 studies included in the review, patients were not randomly assigned to the observation unit. Although efficiencies are possible from grouping patients in dedicated units, our data suggest that patient selection may explain many purported cost savings. Our findings also demonstrate that the entire spectrum of observation care should be addressed, not just care delivered in dedicated units.
Hospitals are obligated to follow stringent criteria for inpatient admissions in response to the threat of fines for “inappropriate” admissions. Under the 2003 Medicare Prescription Drug, Improvement and Modernization Act,7 recovery audit contractors were paid to find incorrect inpatient Medicare A and B claims, including inpatient stays that were actually observation stays.7 In a 6-state, 3-year pilot project, the auditors recovered $992.7 million for the government in inappropriate billings.18,19 Subsequently, the program was expanded to include all 50 states and all Medicare programs under the Tax Relief and Health Care Act (2006) and the Patient Protection and Affordable Care Act (2010).7
Although reducing erroneous Medicare billings is important, our findings suggest that many aspects of observation status and reimbursement should be reevaluated. Currently, the Center for Medicare Advocacy is litigating to end observation status,20 and proposed federal legislation would guarantee a skilled facility benefit to all hospitalized Medicare patients including those with observation status.21,22 In April 2013, CMS proposed rule changes that, if implemented, would confer inpatient status to some longer stay observation patients.23
Quiz Ref IDOur study has limitations. We studied 1 large academic medical center, and our findings may not be generalizable to other settings or hospitals. We only had access to variables captured in our electronic medical record and administrative databases. We could not identify repeat encounters that may have occurred at other institutions. Nonetheless, our findings demonstrate that observation care in clinical practice is very different than what CMS initially envisioned and creates insurance loopholes that adversely affect patients, health care providers, and hospitals. It is uncertain what role, if any, observation status for hospitalized patients should have in the era of health care reform.
Corresponding Author: Ann M. Sheehy, MD, MS, Division of Hospital Medicine, Department of Medicine, University of Wisconsin School of Medicine and Public Health, 1685 Highland Ave, MFCB 3126, Madison, WI 53705 (firstname.lastname@example.org).
Accepted for Publication: May 13, 2013.
Published Online: July 8, 2013. doi:10.1001/jamainternmed.2013.8185.
Author Contributions: Dr Sheehy had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Sheehy, Graf, Jacobs.
Acquisition of data: Sheehy, Heidke.
Analysis and interpretation of data: Sheehy, Graf, Gangireddy, Hoffman, Ehlenbach, Fields, Liegel, Jacobs.
Drafting of the manuscript: Sheehy, Graf.
Critical revision of the manuscript for important intellectual content: Gangireddy, Hoffman, Ehlenbach, Heidke, Fields, Liegel, Jacobs.
Obtained funding: Sheehy.
Administrative, technical, and material support: Gangireddy, Hoffman, Ehlenbach, Heidke, Fields, Liegel.
Study supervision: Sheehy, Graf.
Conflict of Interest Disclosures: Dr Graf reports receiving royalties from Smith & Nephew for patents he has assigned to them involving orthopedic implants used in arthroscopic surgery. The remaining authors have no relevant financial disclosures or conflicts to report.
Funding/Support: This study was supported by the University of Wisconsin Department of Medicine (Dr Sheehy). Dr Jacobs was supported by grant R01 AG033172-01 from the National Institute on Aging.
Role of the Sponsor: The sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.
Previous Presentation: Portions of data were presented at the Eighth Annual Midwestern Hospital Medicine Conference; October 11, 2012; Chicago, Illinois.
Additional Contributions: Daniel S. Dexter, BA, Project Manager at UWHC Business Planning and Analysis, assisted with the data set, and Andrew T. LaRocque, BBA, Analyst at the University of Wisconsin School of Medicine and Public Health, Division of Hospital Medicine, assisted with data presentation. Neither received additional compensation for their work on this study.