08/11/09

Hospital “Staffed Beds”: A concept in need of clarification

Rexford Anson-Dwamena, MPH

Overview:
Customarily health planning professionals assess the size of a hospital by using measures such as
licensed beds, admissions, patient days, average daily census, and staffed beds. Most measures,
with the exception of staffed beds, are clearly defined by statute or by specific administrative,
regulatory or reimbursement procedures. “Staffed beds” as a concept, on the contrary, has become
a subjective variable that is often left to the hospital’s own judgment to access without further
guidance. Because hospital size for funding purposes can be measured by staffed beds, and
hospitals often have not contemplated the ramifications of inaccurate assessment of this variable, it
is important that certain health planning guidelines be established to provide minimum guidance to
hospitals.

Historically, health planning endeavors have relied on a probabilistic model comparing the pattern
of patient admissions with the normal or Poisson distribution. The object is to estimate at a high
level of probability (let us say 98% confidence level) that given an average daily census (ADC) a
hospital will ever be required to staff above a certain level. Of course this methodology does not
take into account the requirements of major disasters, but neither would the hospital’s internal
decision process regarding the number of staffed beds. We want to be able to say with high levels
of confidence, that a hospital with an ADC of 30 will seldom, if ever, have to staff for more than 43
patients throughout a given year. Given the nature of the probabilistic model, and the assumptions
of normal distribution that underlie the model, we would feel more comfortable if we were dealing
with a large hospital with an ADC of thousands, but even with small hospitals the estimates can
provide guidance into what staffing levels would be required in a “normal” year and a baseline for
comparing the actual levels at the end of the year.

Staffed Bed Calculations*:

Because small hospitals often only have general medical/surgical beds and do not have specialty
beds (Pediatrics, Obstetrics, Psychiatric, ICU/CCU) the facility can compute the required staffed
beds from their “Patient Days” data.

A small rural hospital, for instance with 6,000 patient days would first need to compute its ADC:

ADC = Patient Days/365 = 6,000/365 = 16.4

Under the assumptions of the normal or Poisson distribution the standard deviation equals the
square root of the mean.

SD = √ ADC = √16.4 = 4.05

Page 1 of 4
 08/11/09
To determine the Probability Factor (PF) we need only remember that if we go out 1.96 standard
deviations from the mean on a normal or Poisson distribution we can capture 95 percent of all
measured events. Therefore by multiplying the SD by 1.96 this would provide 95% certainty that
you would have enough staff for all but 9 days of the year [(365 - (.95 x 365))/2 = 9. [Note: We
must divide by 2 because we are only interested in the positive tail of the normal distribution]. To
calculate the staffed beds at the 95% confidence level for medical/surgical beds, therefore:

ADC + (PF x SD) =16.4 + (1.96 x 4.05) = 16.4 + 7.9 = 24 staffed beds

Likewise, by multiplying SD by 2.33, we would provide 98% certainty that enough staff would be
provided for all but 4 days of the year [(365 – (.98 x 365))/2 = 4]. To calculate the staffed beds at
the 98% confidence level for medical surgical beds, therefore:

ADC + (2.33 x SD) =16.4 + (2.33 x 4.05) = 16.4 + 9.4 = 26 staffed beds

Therefore, in general to estimate staffed given that you know the annual patient days and have
calculated the ADC of the hospital:

Staffed Beds = ADC + (PF x √ADC)

Policy implications:
Four caveats must be stated.

(1) the above statistic does not specify how many staff are required, but rather it provides an
estimate given the number of hospital patient days how many patients, with a high degree of
certainty, will be in the hospital on any given day
(2) the hospital is staffing for the number of patients, not to the maximum number of patients
that a given staffing level would allow. Because of the incremental nature of hospital
staffing, once a threshold of patients is reached, the hospital may actually increase the
staffing pattern such that many more patients than the estimated number could be taken care
of by the hospital. It is the number of patients that the hospital is staffing for, not the
number of maximum patients given a staffing pattern.
(3) staffed beds should reflect normal operations, not the maximum that could be obtained
under extreme emergency.
(4) staffed beds should reflect a baseline for administrative guidance, and if experience suggests
that the statistical estimates are grossly under or over the number of actually staffed beds
then the estimates can be adjusted. Because staffing is a “bottom line” issue hospital
administration should be aware if they are in reality overstaffing for the number of patients
within their facility.

The computational example given above represents an actual rural hospital in Virginia that had
listed its staffed beds as 48 but after the above analysis and several discussions with the CEO and
CFO it was determined that they had no idea how to estimate or calculate the number of staffed
beds. Of the total of 12 hospitals contacted that had staffed bed estimates considerably higher than
that predicted by the model, none of the administrative personnel could point to any institutional or
regulatory rule as to a methodology for calculating staffed beds.
Page 2 of 4
 08/11/09

For Virginia, the data on staffed beds for all hospitals can be found on the Virginia Health
Information website (see: http://www.vhi.org/hospitals.asp). Using the above calculations it can be
seen that many hospitals grossly overestimate their actual staffed beds. There is clearly no
systematic determination of what constitutes a staffed bed. The analysts at the VHI proffer that it is
a very tenuous and ill defined measure and probably should not be used for analytical purposes. It
is suggest here that to estimate the actual utilized capacity in Virginia the above formula be
consitently used for health planning purposes, as it is already used within the COPN regulations
with projected pediatric bed need computation. Namely, the applicant is instructed to:

Calculate the number of beds needed to assure that adequate bed capacity will exist with a 98%
probability for an unscheduled pediatric admission using the following formula: Number of
pediatric beds allowable = PADC + 2.33√PADC (12VAC5-240-30. Availability, p. 9)

It would appear that because Virginia is a Certificate of Public Need State (COPN) that many
hospitals are reluctant to appear to be drastically underutilizing their current bed capacity. But by
inflating staffed beds a false sense of availability of beds may be projected that unrealistically
suggests bed utilization.

In addition some hospital administrators may interpret staffed beds to be what they are reporting on
their Medicare cost Report** (See Appendix A: Form CMS-2552-96, line #12, column #1). But
directions for lines 6 through 11 are simply “Enter the appropriate statistic applicable to each
discipline for all programs.” “Appropriate” is nowhere defined! “Staffed” is nowhere used in the
instructions. Because some federal programs us the cost report data [e.g., for the Small Rural
Hospital Improvement Grant Program (CFDA 93.301) small is defined as 49 available beds or less,
as reported on the hospital’s most recently filed Medicare Cost Report,
http://ruralhealth.hrsa.gov/ship.htm] it is important for rural hospitals in this case to know what this
addition can entail. Fortunately the SHIP program has made a determination to use staffed beds as
the criteria but the methodology for determining staffed beds is nowhere specified.
Conclusion:

Hospital administrators should with do diligence estimate the number of staffed beds that they
actually have within their hospital. For health planning and for funding purposes the “staffed bed”
criteria remains one of the best indicators of actual usage of hospital beds.

Health planners should probably use the staff bed computations as specified above rather than those
numbers proffered by the individual hospitals. For reporting and analytical purposes the estimates
provided by state databases such as VHI are particularly unreliable. It may be advantageous for
health planning purposes to consistently calculate the “staffed bed” data rather than view such data
as a reporting requirement.

___________________________________
*For a standard treatment of the subject see, Strategic Health Planning: Methods and Techniques
Applied to Marketing/Management, Ablex Publishing, Norwood,.NJ, (January 1, 1991, 1998) p. 74.

**FORM CMS-2552-963605.1 Part I - Hospital and Hospital Health Care Complex Statistical
Data.--This part collects statistical data regarding beds, days, FTEs, and discharges.

Page 3 of 4
 08/11/09

APPENDIX A

Page 4 of 4