What is an observation stay? Evaluating the use of hospital observation stays in Medicare

Abstract

Background/Objectives:

Observation stays are increasingly common for older adults, yet little is known about the extent to which it is being used as CMS originally intended for unscheduled or acute problems, and whether different types of services are reflected in current billing practices.

Design:

Observational cohort study.

Setting/Participants:

867,165 qualifying observation stays identified from 451,408 patients using Medicare fee-for-service claims data from a nationally representative 20% beneficiary sample between January 1 2014 and November 30 2014.

Measurements:

Using descriptive and multivariable logistic model analytic approaches, we evaluated the patient, stay, and hospital characteristics associated with the most common billing practice for observation stays (charge revenue center 0761 exclusively) versus all other practices.

Results:

Sixty-three percent of observation stays were billed exclusively under the 0761 revenue center and were more likely to be for pre-planned chronic conditions consisting of short-term treatments (e.g., chemotherapy, radiation therapy, wound care, paracentesis, epidural spinal injection). These stays appeared to be used for recurrent single-day visits, given its strong association with prior visits and a high rate of re-observation (41.4%), with frequent return stays appearing in a seven-day pattern.

Conclusion:

Nearly two-thirds of observation stays are billed using only the 0761 revenue code and appear to be for prescheduled, repeat treatments—differing substantially from CMS’s explicitly stated purpose as a form of care used while a health care provider determines whether a patient presenting for unscheduled or acute conditions requires inpatient hospital admission or can be safely discharged. Guidance is needed from CMS to clarify the appropriate role of observation stays, with discussion as to whether episodic single-day, planned treatment for chronic conditions not originating in the ED should be billed as observation stays or placed under another mechanism. Subsequent research is needed to understand how the current use of observation stays impact patient out-of-pocket costs.

INTRODUCTION

Outpatient (observation) hospitalization stays are increasingly common, yet precisely how current observation stay designations are used by the health care system community is unknown. Because observation stays are billed differently than inpatient admissions, repeated use of observation stays may have unintended financial consequences for patients. Inpatient admissions are covered under Medicare Part A: patients pay a $1,408 deductible with no coinsurance for the first 60 days. Post-acute skilled nursing facility care is covered when the stay is over three consecutive inpatient midnights. Whereas, observation stays are billed under Medicare Part B: most beneficiaries pay a monthly premium (typically $144.60) with a $198 annual deductible after which they pay 20% of the Medicare-approved amount for covered services with no out-of-pocket limit. With an average out-of-pocket cost of $528 per observation stay¹, a single observation hospitalization may cost a beneficiary less than an inpatient stay, but repeat observation stays and subsequent skilled nursing facility costs can add up in the absence of an out-of-pocket cap on these services, ultimately leading to high financial burden.

Medicare defines observation stays as services for “short term treatment, assessment, and reassessment that are furnished while a decision is being made regarding whether patients will require further treatment as hospital inpatients or if they are able to be discharged from the hospital.”¹ Often, although not always, observation care is ordered for patients who present for Emergency Department (ED) evaluation; as described by Medicare, these are intended to be unplanned visits.² Given the explicitly stated purpose of observation services defined by Medicare, it raises the question of whether overuse or inappropriate use of the observation designation for services exists.

Like inpatient services, hospitals bill for observation services using revenue center codes. According to Medicare’s claims processing manual, observation services should be coded under 0760 (treatment or observation room-general classification) and 0762 (treatment or observation room-observation room).² In practice, however, observation stays are also classified under two additional codes: 0761 (treatment or observation room-treatment room) and 0769 (treatment or observation room-other). Billing 0760 or 0762 for observation service instead of 0761 or 0769 has been in place since at least 2003, with little additional guidance in subsequent years. Previous work by Sheehy et al. described the heterogeneity existing in the use of these four codes to report observation services.³ They found almost half (46%) of all 30-day readmission observation stays were billed under 0761-only, even though 0761 is not one of the Medicare-designated codes. This raises the possibility that different codes are being used to represent different types of services underneath the umbrella of “observation.”³ Within research, the codes used to identify observation services vary between studies, raising the question: what are the differences between how Medicare has recommended observation stays be used versus how clinicians and health care systems are actually using them?

Given the large volume of observation stays billed solely under the 0761 revenue code, the aim of this brief is to better understand the utilization and predictors of the 0761 revenue code within a national sample of Medicare beneficiaries.

METHODS

Study sample

We employed the University of Wisconsin Method by Sheehy et al.³ to identify all observation stays between January 1, 2014 and November 30, 2014 in the Medicare 20% nationally representative fee-for-service beneficiary sample. The definition of eligible observation stays was aligned with CMS’s existing 30-day inpatient hospitalization measure denominator criteria.⁴ We compared observation stays billed exclusively under the 0761 revenue code (no other codes billed) to observation stays billed under all other combinations of observation revenue codes.

Covariates

Our multivariable prediction model consisted of patient-, stay-, and hospital-level characteristics. Predictors were drawn from theoretical models of health service utilization.^5,6 Characteristics included patient comorbidities from the baseline year prior (including the Elixhauser Comorbidity Index, Hierarchical Condition Category (HCC) score, and disability status), factors occurring during the index stay (originating from the ED, length of stay, discharge to skilled nursing facility, principal diagnoses according to multi-level Clinical Classifications Software categories), patient sociodemographics (age, gender, rural urban commuter area classification, race/ethnicity, Medicaid status, neighborhood disadvantage via the Area Deprivation Index⁷), and hospital-level characteristics (medical school affiliation, for-profit/nonprofit status, inpatient discharge volume).

Analysis

Since repeat stays are central to the study and have important patient and hospital implications for both costs and care provision, we provide an analysis at the observation stay-level. We used logistic regression with generalized linear modeling techniques to predict which stays would be billed under revenue center code 0761-only, with clustered robust standard errors to account for patient-level clustering. Finally, we calculated 30-day re-observation rates and plotted the frequency of 30-day re-observation from each index stay.

RESULTS

Study sample

There were 867,165 qualifying observation stays (age M=72; SD=13) identified from 451,408 beneficiaries using the University of Wisconsin method; 547,185 (63%) were billed solely under the 0761 code (see Supplementary Material Figure 1 for sample derivation).

Predictors of being billed under 0761-only revenue code

Results of the multivariable logistic regression suggest that stays billed for observation services under 0761-only were more likely to be for recurrent, chronic problems (Supplemental Table S1). Stay-level predictors had the strongest associations with 0761-only stays. Notably, the lowest odds of being billed exclusively under 0761 were found with stays originating in the ED (OR=0.01; 95% CI=0.01–0.01), a length of stay longer than one day (two days OR=0.04; 95% CI=0.03–0.04; three days OR=0.03; 95% CI=0.03–0.04; four or more days (OR=0.20; 95% CI=0.18–0.24), and a stay resulting in discharge to a SNF (OR=0.11; 95% CI=0.09–0.12). Patients billed under 0761-only were less likely to have principal diagnoses involving diseases of the digestive (OR=0.35; 95% CI=0.33–0.38), circulatory (OR=0.42; 95% CI=0.39–0.44), or genitourinary systems (OR=0.67; 95% CI=0.62–0.72). Conversely, beneficiaries were over four times more likely to be billed under 0761-only if the primary diagnosis code for their stay was related to diseases of the skin and subcutaneous tissue (OR=4.23; 95% CI=3.87–4.61), disease of the musculoskeletal system and connective tissue (OR=1.86; 95% CI=1.75–1.98), and neoplasms (OR=1.72; 95% CI=1.61–1.84).

Patient sociodemographic, hospital and other characteristics during the year prior were also predictive. Specifically, the odds of being billed solely under 0761 increased 31% for each observation stay in the prior year (OR=1.31; 95% CI=1.26–1.36). Being billed under 0761-only was also associated with higher patient complexity and illness burden in the prior year as measured by the HCC score (OR=1.06; 95% CI=1.05–1.07) as well as a diagnosis of complicated diabetes (OR=1.25; 95% CI=1.18–1.32), but less likely for drug abuse (OR=0.70; 95% CI=0.63–0.78). Certain hospital characteristics were also more likely to be associated with a 0761-only visit, including a major medical school affiliation (OR=1.57; 95% CI=1.52–1.62) and a high annual discharge volume (OR=1.49; 95% CI=1.39–1.60). In addition, African-Americans (OR=0.72; 95% CI=0.68–0.75) and those classified under other race/ethnicity categories relative to Whites (OR=0.72; 95% CI=0.68–0.76), as well as living in the most disadvantaged neighborhoods (OR=0.88; 95% CI=0.84–0.92) were less likely to experience 0761-only stays.

30-day re-observation

The 30-day re-observation rate for 0761-only stays was unusually high by any measure (41.4%) and especially when compared to non-0761-only stays (6.8%; Table 1). The average number of subsequent re-observations in the 30 day time frame was also higher in the 0761-only group (M=1.082; 95% CI=1.076–1.089) compared to other stays (M=0.096; 95% CI=0.094–0.098). The same is true for those with at least one re-observation in the 30 days (0761-only M=2.613 days, 95% CI=2.600–2.626 vs. other stays M=1.405 days, 95% CI=1.386–1.424).

Table 1.

30-day re-observation by observation stay type

30-day re-observationa	Overall (n=867,165)	Non 0761-only (n=319,980)	0761-only (n=547,185)
N	248,499	21,837	226,662
Rate (95% CI)	28.66% (28.56 – 28.75)	6.82% (6.74 – 6.91)	41.42% (41.29 – 41.55)
# of re-observations, mean (95% CI)	0.718 (0.714 – 0.723)	0.096 (0.094 – 0.098)	1.082 (1.076 – 1.089)

CI, Confidence Intervals

^aFor any related observation stay revenue center code

A plot of time to the first re-observation in the 30-day period post-discharge suggests a high density of return stays in a seven-day pattern for 0761-only stays (Figure 1). Those billed under other observation revenue center combinations also had increased frequency of repeat stays at 7, 14, 21, and 28 days, though the magnitude was much less pronounced, with stays at these 7-day increments representing 51% of first re-observations in the 0761-only group compared to 17% in other stay combinations.

Figure 1.

Day to first re-observation in 30-days post index observation stay discharge

DISCUSSION

In this study, we evaluated how observation stays are being used and billed in the Medicare system nationwide. This is the first step in a larger effort to determine the impact of CMS’s observation stay policy on health outcomes, health care system costs, and individual financial burden.^8–14 By first examining the existing patterns of use, this study informs the current understanding of what the health care system community collectively considers to be an observation stay, in clinical practice.

Our findings suggest nearly two-thirds of stays are billed solely under the 0761 revenue code (n=547,185) and most often for reasons that fall outside Medicare’s specific definition of an ‘observation stay’. Stays billed entirely under the 0761 revenue code are markedly different than those billed under other codes combinations. These 0761-only visits frequently occur for conditions generally considered to be chronic or episodic, rather than acute, differing greatly from CMS’ explicitly stated purpose of observation services as a form of care used while a health care provider determines whether a patient presenting for unscheduled or acute problems requires inpatient admission or can be safely discharged.² Conditions that often require procedural intervention are more likely to be coded as 0761-only, including decubitus ulcers, chronic lower extremity ulcers, ascites, and thoracic or lumbosacral neuritis or radiculitis chemotherapy (see Supplemental Table S2 for differences in principal diagnoses). In addition, 0761 is often used for the treatment of malignancy, specifically maintenance radiotherapy and maintenance chemotherapy, as well as other conditions requiring parenteral therapies, such as ulcerative colitis. Conditions traditionally thought of as requiring acute therapies, in contrast, such as sepsis, acute chest pain, and urinary tract infections, are less likely to be billed under the 0761 revenue center code. In addition, 97.5% of the 0761-only observation stays were for a single day and did not originate from the ED (62% of all observation stays identified in the sample), compared to 17.8% of stays billed using another revenue center combinations.

We hypothesize that the 0761 observation code is regularly used for pre-planned visits for short-term treatments such as chemotherapy, radiation therapy, wound care, paracentesis, and epidural spinal injection. While further studies are needed to confirm if procedure codes are associated with these 0761 hospitalizations, our work thus far suggests that many 0761 stays are atypical when it comes to CMS’ intentions for and providers’ understanding of observation policy. In addition, the cyclic seven-day nature of re-observations associated with these 0761 stays (a typical interval for planned care), combined with the finding that most are not originating in Emergency Departments, illustrates that these visits are not what providers or CMS would consider traditional observation stays.

These results have broad implications, both for health care policy and for patients. At the patient level, the use of observation stays in the 0761 pattern may have unintended cost consequences given the lack of a cap on out-of-pocket expenses for services bulled under Part B. In our data of 248,499 beneficiaries with at least one re-observation in the subsequent 30 days, the rate of re-observation for 0761-only stays was 86% larger than non 0761-only stays, indicating that the average patient in this subgroup experiences two or more observation hospitalizations in a 30-day period. At this rate, beneficiaries can rapidly accumulate unmanageable medical debt that, after three observation stays, would exceed the inpatient Part A deductible.

On a macro level, researchers and policymakers should be aware of the different uses of observation stays. Moving forward, an understanding that different types of care are captured under one large “observation” umbrella can result in better methods to isolate particular types of observation stays, which in turn can lead to better designed studies that result in more precise and accurate evidence needed to inform policy decisions.

Given these results, we believe that hospitals and health care systems need more guidance from CMS regarding the appropriate use of observation stay services. CMS should clarify whether observations should be used repeatedly and for planned reasons, or whether these 0761 services should be billed under another mechanism. Our findings on the existing patterns of use warrant further investigation, in particular, the recurring seven-day nature of visits under 0761 which may have considerable impact on patient out of pocket costs.