Browse by Year
/ 1998
/ January
/ Friday, January 30, 1998
[Federal Register: January 30, 1998 (Volume 63, Number 20)]
[Rules and Regulations]
[Page 5105-5139]
From the Federal Register Online via GPO Access [wais.access.gpo.gov]
[DOCID:fr30ja98-22]
[[Page 5105]]
_______________________________________________________________________
Part IV
Department of Health and Human Services
_______________________________________________________________________
Health Care Financing Administration
_______________________________________________________________________
42 CFR Part 413
Medicare and Medicaid Programs; Salary Equivalency Guidelines for
Physical Therapy, Respiratory Therapy, Speech Language Pathology, and
Occupational Therapy Services; Final Rule
[[Page 5106]]
DEPARTMENT OF HEALTH AND HUMAN SERVICES
Health Care Financing Administration
42 CFR Part 413
[HCFA-1808-F]
RIN 0938-AG70
Medicare and Medicaid Programs; Salary Equivalency Guidelines for
Physical Therapy, Respiratory Therapy, Speech Language Pathology, and
Occupational Therapy Services
AGENCY: Health Care Financing Administration (HCFA), HHS.
ACTION: Final rule.
-----------------------------------------------------------------------
SUMMARY: This final rule sets forth revisions to the salary equivalency
guidelines for Medicare payment for the reasonable costs of physical
therapy and respiratory therapy services furnished under arrangements
by an outside contractor. This final rule also sets forth new salary
equivalency guidelines for Medicare payment for the reasonable costs of
speech language pathology and occupational therapy services furnished
under arrangements by an outside contractor. The guidelines do not
apply to inpatient hospital services and hospice services. The
guidelines will be used by Medicare fiscal intermediaries to determine
the maximum allowable cost of those services.
EFFECTIVE DATE: This rule is effective April 1, 1998. The rule is
applicable for services furnished on or after April 1, 1998. This rule
is a major rule as defined in Title 5, United States Code, section
804(2). Pursuant to 5 U.S.C. section 801(a)(1)(A), we have submitted a
report to Congress on this rule.
ADDRESSES: To order copies of the Federal Register containing this
document, send your request to: New Orders, Superintendent of
Documents, P.O. Box 371954, Pittsburgh, PA 15250-7954. Specify the date
of the issue requested and enclose a check or money order payable to
the Superintendent of Documents, or enclose your Visa or Master Card
number and expiration date. Credit card orders can also be placed by
calling the order desk at (202) 512-1530 or by faxing to (202) 512-
2250. The cost for each copy is $8.00. As an alternative, you can view
and photocopy the Federal Register document at most libraries
designated as Federal Depository Libraries and at many other public and
academic libraries throughout the country that receive the Federal
Register.
FOR FURTHER INFORMATION CONTACT: Jackie Gordon, (410) 786-4517.
SUPPLEMENTARY INFORMATION:
I. Background
Section 1861(v)(5) of the Social Security Act (the Act) requires
the Secretary to determine the reasonable cost of services furnished to
Medicare beneficiaries ``under an arrangement'' with a provider of
services, by therapists or other health-related personnel. The Health
Care Financing Administration (HCFA) pays the provider directly for
these services, rather than paying the therapist or supplying
organization. Under section 1861(w)(1) of the Act, this payment
discharges the beneficiary from liability to pay for the services.
Section 1861(v)(5) of the Act also specifies that the reasonable costs
for these services may not exceed an amount equal to the salary that
would reasonably have been paid for the services (together with any
additional costs that would have been incurred by the provider or other
organization) to the person performing them if they had been performed
in an employment relationship with a provider or other organization
(rather than under such arrangement), plus allowances for certain
expenses that may be incurred by the contracting therapy organization
in furnishing the services as the Secretary in regulations determines
to be appropriate.
These statutory requirements are implemented in existing
regulations at 42 CFR 413.106. The regulations apply to the services of
physical, occupational, speech language pathologists, and other
therapists and services of other health specialists (other than
physicians) furnished under arrangements with a provider of services, a
clinic, a rehabilitation agency, or a public health agency. The
regulations provide for:
<bullet> Hourly salary equivalency amounts comprised of:
--A prevailing hourly salary rate based on the 75th percentile of the
range of salaries paid to full-time employee therapists by providers in
the geographic area, by type of therapy.
--Fringe benefit and expense factors to take into account fringe
benefits generally received by an employee therapist, as well as
expenses (such as maintaining an office, insurance, etc.) that a
therapist or therapist organization might incur in furnishing services
under arrangements.
<bullet> A standard travel allowance to recognize time spent in
traveling to the provider's site or the patient's home.
<bullet> As provided for in existing regulations at Sec. 413.106(e)
and explained in section 1412 of the Provider Reimbursement Manual, the
following are additional allowances for costs incurred for services
furnished by an outside supplier. In addition to the guidelines
established for the adjusted hourly salary equivalency amount and the
travel allowance, the following costs incurred for services furnished
by an outside supplier are recognized, provided the services are
properly documented as having been received by the provider.
--Overtime, if an outside supplier utilizes the services of its
employees (including the services of aides and assistants) at an
individual provider in excess of the provider's standard workweek.
Several commenters stated that there should be no limits on overtime
compensation. The proposed rule did not specifically introduce new
limits on payment for overtime. The proposed rule provided that a
provider would receive payment for overtime but if the therapist worked
over 40 hours it would not receive the expense factor portion of the
hourly salary equivalency guideline amount.
--Administrative and supervisory duties, if an outside supplier
provides more than one therapist and at least one therapist spends more
than 20 percent of his or her time supervising other therapists and
performing administrative duties.
--Depreciable or leased equipment, including maintenance costs of
equipment remaining at the provider's site, that the outside supplier
uses in furnishing direct services to the provider's patients (may also
include equipment that is transported from one provider site to another
but excludes equipment owned by the provider).
--Supplies furnished by the supplier for direct patient care (e.g.,
gases and sprays for respiratory therapy), excluding items such as
envelopes, stamps, and typewriters that are reimbursed as overhead
expenses and included in the fringe benefit and expense factor.
--Travel expenses, based on 10 times the General Services
Administration mileage rate for each day an outside supplier travels to
a provider site.
--Aides, who are paid as an add-on. Several commenters requested that
we pay aides as a function of the hourly salary equivalency amount at
50 percent of these amounts.
--Assistants, who are paid as a function of the hourly salary
equivalency amount at 75 percent of these amounts. (All therapy types
use assistants except respiratory therapists.)
The provider must supply the intermediary with documentation that
[[Page 5107]]
supports these additional costs to the intermediary's satisfaction.
These are the only additional costs that will be recognized.
The regulations at 42 CFR 431.106 (b)(5) and (c) also provide for
an exemption for limited part-time or intermittent services if the
provider required the services of an outside supplier for a particular
type of therapy service and the total hours of services performed for
the provider, by type of service, average less than 15 hours per week
for those weeks in the cost reporting period during which services were
furnished by nonemployee therapists. (Travel time is not counted in the
computation, even if the actual time is used.) If a provider qualifies
for this exemption, the reasonable cost of such services is evaluated
on a reasonable rate per unit of service basis, except that payment for
these services in the aggregate, during the cost reporting period, may
not exceed the amount that would be allowable had the provider
purchased these services on a regular part-time basis for an average of
15 hours per week for the number of weeks in which services were
furnished. Where the contract provides for a method of payment other
than rate per unit of service (e.g., hourly rate or percentage of
charges), payment cannot exceed the guideline adjusted hourly amounts
plus other allowable costs, even though the services are performed on a
limited or intermittent part-time basis.
In addition, the existing regulations at Sec. 413.106(f)(1) have
provided for an exception because of binding contract. An exception was
granted to a provider that entered into a written binding contract with
a therapist or contracting organization prior to the date the initial
guidelines are published for a particular type of therapy. Before the
exception was granted, however, the provider was required to submit the
contract to its intermediary, subject to review and approval by the
HCFA regional office. This exception may be granted for the contract
period, but no longer than 1 year from the date that the guidelines for
the particular therapy are published. During the period in which a
binding contract exception was in effect, the cost of the services was
evaluated under the prudent buyer concept. (Section 1414.1 of the
Provider Reimbursement Manual contains instructions on this exception.)
This exception did not apply to providers who entered into a
contingency contract with a therapist or contracting organization or
another provider. In a contingency contract, the provider and
contractor agree that if Medicare does not reimburse the provider for
the rate at which the contract is set, the provider and contractor
agree that the contractor will make up the difference. We do not
consider a contingency contract a binding contract. (We are eliminating
this exception in this final rule. See Section II. On responses to
public comments on proposed rule for further discussion.)
Also, the existing regulations at Sec. 413.106(f)(2) provide for an
exception for unique circumstances or special labor market conditions.
An exception may be granted when a provider demonstrates that the costs
for therapy services established by the guidelines are inappropriate to
a particular provider because of some unique circumstances or special
labor market conditions in the area. As explained in section 1414.2 of
the Provider Reimbursement Manual, exceptions will be granted only in
extraordinary circumstances. Before the exception may be granted, the
provider must submit appropriate evidence to its intermediary to
substantiate its claim. The provider's request for an exception,
together with substantiating documentation, must be submitted to the
intermediary each year, no later than 150 days after the close of the
provider's cost reporting period. Because providers had been required
to submit cost reports to intermediaries no later than 90 days after
the close of their cost reporting periods, we had required that the
provider's request for an exception, together with substantiating
documentation, also be submitted to the intermediary no later than 90
days after the close of its cost reporting period. On June 27, 1995 (60
FR 33137), we changed the due date for submission of cost reports to
150 days after the close of the provider's cost reporting period.
Accordingly, as explained under Section II.F. of this preamble, we are
revising the time period for a provider's request for an exception,
together with substantiating documentation, to 150 days after the close
of its cost reporting period. If the circumstances giving rise to the
exception remain unchanged from a prior cost reporting period, however,
the provider need only submit evidence to the intermediary 150 days
after the close of its cost reporting period to establish that fact.
In order to establish an exception for unique circumstances, the
provider must submit evidence to establish that it has some unique
method of delivering therapy or other services, which affects its
costs, that is different from the other providers in the area. The
exception will be effective no earlier than the onset of the unique
circumstances.
In order to substantiate an exception for special labor market
conditions, the provider must submit evidence enabling the intermediary
to establish that the going rate in the area for a particular type of
service is higher than the guideline limit and that such services are
unavailable at the guideline amounts. It is the duty of the provider to
prove to the satisfaction of the intermediary that it has reasonably
exhausted all possible sources of this service without success.
The intermediary collects information on the rates that other
providers in the area generally pay therapists or other health care
specialists. Once this information is collected, the intermediary will
determine whether other providers in the area, in comparison to the
provider requesting the exception, generally pay therapists or other
health care specialists higher rates than the guideline amounts.
Under existing Sec. 413.106(b)(6), HCFA issues guidelines
establishing the hourly salary equivalency amounts in geographical
areas for therapy services furnished to Medicare beneficiaries under
arrangements. These guidelines apply only to the amount of payment the
Medicare program makes to a provider for therapy services obtained
under arrangements. The guidelines are not intended to dictate or
otherwise interfere in the terms of a contract that a provider may wish
to enter into with a therapist or therapist organization. The
guidelines do not apply to services furnished by employees of a
hospital or employees of other providers. There is also an exception to
the guidelines for inpatient hospital services provided by hospitals
paid under the prospective payment system or subject to rate-of-
increase limits (Sec. 413.106(f)(4)), in which case the services are
evaluated under the Medicare program's reasonable cost provisions as
described at Sec. 413.5). The salary equivalency guidelines also will
not be applied to skilled nursing facilities (SNFs) that are paid under
the prospective payment system for therapy services provided under
arrangements for cost reporting periods beginning on or after July 1,
1998. (This includes low volume SNFs currently electing prospective
payment under section 1888(d) of the Act.) In addition, the salary
equivalency guidelines will not be applied to HHAs who are paid under
the prospective payment system for therapy services provided under
arrangements for cost reporting periods beginning on or after October
1, 1999. The salary equivalency guidelines also will not apply for
outpatient therapy services provided by a SNF or an outpatient
rehabilitation
[[Page 5108]]
provider for services provided to SNF patients on or after July 1, 1998
when payment for those services is made on a fee schedule basis.
(Providers of Part B outpatient therapy services provided to Medicare
beneficiaries whose nursing home stays are not paid by Medicare will be
paid on a fee schedule basis for services furnished on or after July 1,
1998.) The guidelines also will not apply to an outpatient
rehabilitation provider, a comprehensive outpatient rehabilitation
facility (CORF), an HHA providing outpatient rehabilitation services to
patients who are not homebound, or the outpatient department of a
hospital when payment for those services is made on a fee schedule
basis beginning on January 1, 1999. Shown below is a chart outlining
the provisions of the Balanced Budget Act of 1997. The salary
equivalency guidelines will cease to apply to the enumerated provider
types once the Balanced Budget Act provisions become effective.
----------------------------------------------------------------------------------------------------------------
Provider type BBA provision Effective date
----------------------------------------------------------------------------------------------------------------
Hospital Outpatient Therapy Services....... Payment on a fee schedule basis.. Calendar year 1999.
SNF Inpatient Services (Includes therapy Payment on a Prospective Payment Cost reporting periods beginning
services and applies to free-standing and System basis. on or after July 1, 1998.
hospital-based providers).
SNF Outpatient Therapy Services............ Fee Schedule..................... For services beginning July 1,
1998.
CORFs (applies to free-standing and Fee schedule..................... Calendar year 1999.
hospital-based providers).
Outpatient Rehabilitation Providers........ Fee schedule..................... Calendar year 1999.
CMHCs...................................... Payment under the outpatient Calendar year 1999.
hospital Prospective System
Payment basis.
Outpatient Therapy Services Provided by HHA Fee Schedule..................... Calendar year 1999.
But Not Under HHA benefit.
HHA Services (Includes therapy services and Payment on a Prospective System Cost reporting periods beginning
applies to free-standing and hospital- Payment basis. on or after October 1, 1999.
based providers).
----------------------------------------------------------------------------------------------------------------
* A $1500 annual limitation on services provided to Medicare beneficiaries will be applied beginning January 1,
1999 where therapy services are provided by providers under the outpatient physical therapy benefit (which
includes speech language pathology services) and occupational therapy benefit.
However, we are establishing regulations that provide that the
salary equivalency guidelines will apply in situations where
compensation, at least in part, to a therapist employed by the provider
is based on a fee-for-service or on a percentage of income (or
commission). The entire compensation will be subject to the guidelines
in cases where the nature of the arrangements are most like an under
``arrangement'' situation, although technically the provider may treat
the therapists as employees. The guidelines will be applied in this
situation so that an employment relationship is not being used to
circumvent the guidelines.
The guidelines apply to SNFs providing therapy services under
arrangements that elect prospective payment under section 1888(d) of
the Act because that prospective payment system (PPS) only applies to
routine and capital services and does not apply to ancillary services
which include therapy services.
Section 413.106(d) provides that, prior to the beginning of a
period to which a guideline will be applied, HCFA will publish a notice
in the Federal Register establishing the guideline amounts to be
applied to each geographical area by type of therapy. We have issued
schedules of salary equivalency guidelines for the reasonable costs of
physical therapy services since 1975, and for respiratory therapy
services since 1978. On September 30, 1983, we published a final notice
(48 FR 44922) that revised the methodology used to establish the
schedules, as well as the guidelines themselves. The guidelines
continue to apply to physical therapy and respiratory therapy services
provided under arrangements, as set forth in Sec. 413.106, with
hospitals, home health agencies (HHAs), SNFs, hospital-based HHAs,
hospital-based SNFs, CORFs, and outpatient rehabilitation providers
(ORPs). (Since we are issuing guidelines for occupational therapists,
the guidelines also will apply to community mental health centers
(CMHCs) that provide occupational therapy services furnished under
arrangements. However, because CMHC therapy services will be paid under
the outpatient hospital prospective payment system beginning with
services furnished during calendar year 1999, at that time the
guidelines will no longer apply to those occupational therapy
services).
The September 30, 1983 final notice provided that, for providers
with cost reporting periods beginning after October 1, 1982, the
published guidelines would be revised upward by the projected 0.6
percent monthly inflation rate, not compounded. It also provided that,
if for any reason we did not publish a new schedule of guidelines to be
effective for cost reporting periods beginning on or after October 1,
1983 or did not announce other changes in the existing schedule, the
existing guidelines would remain in effect, increased by the projected
0.6 percent monthly inflation rate, not compounded, until a new
schedule of guidelines was issued. This monthly inflation rate was
based on a Data Resources Incorporated (DRI) forecast of the annual
rate of increase in each component of the salary equivalency amounts
(that is, salary, fringe benefits, rent, and other expenses), with each
component weighted to form a composite rate of increase for the 12-
month period ending March 31, 1984.
II. Provisions of the March 28, 1997 Proposed Rule
On March 28, 1997 we published in the Federal Register a notice of
proposed rulemaking (62 FR 14851) that proposed changes in the
methodology used to establish the salary equivalency guidelines. We
proposed to establish salary equivalency guidelines for occupational
therapy and speech-language pathology services that are contracted by
providers. We also proposed to revise the guidelines that were
currently in place for contracted physical therapy and respiratory
therapy services. In the proposed rule:
<bullet> The prevailing hourly salary rates were derived:
--From the 75th percentile of hourly therapist salaries of blended data
from several sources of hospital and SNF wage rate data (weighted by
relative employment levels in hospitals and nursing homes) to develop a
national ``best estimate'' of prevailing salary levels as a basis for
the guidelines.
[[Page 5109]]
--We calculated guideline levels for fourth quarter 1995 and trended
forward to April 1998.
<bullet> We computed fringe benefits as a percent of total
compensation using fiscal year 1994 Medicare cost reports for hospitals
under the prospective payment system.
<bullet> The expense component was based on an estimate of the
costs of maintaining a therapy services office.
<bullet> The standard travel allowance was set at 50 percent of the
hourly salary equivalency amount.
<bullet> The published amounts were to be adjusted to take into
account projected rates of inflation that occurred after the initial
effective date.
The proposal provided for a 60-day period for public comment. The
proposed rule also provided that the guidelines would not be effective
until at least 60 days after the date of publication of the final rule.
We received 409 pieces of correspondence on the proposed
guidelines. A significant number of comments focused on major aspects
of the proposed methodology that required us to perform an extensive
evaluation of the methodology before revised guidelines could be
issued. A summary of the public comments and our responses follow.
III. Summary of Public Comments and Departmental Responses
A. Data Sources for Salary Equivalency Guidelines
We proposed to use the latest available Bureau of Labor Statistics
(BLS) hospital occupational/industry wage survey data along with data
from several other sources of hospital and nursing home data to develop
the salary equivalency guidelines. This was the first time that we had
proposed using data sources in addition to the BLS data in issuing the
salary equivalency guidelines. We based this decision on the following:
First, BLS carried out its last hospital occupational/industry wage
surveys in 1989 and 1991 and for budgetary reasons has discontinued
conducting this survey. Accordingly, even if we had chosen to use BLS
survey data as our primary source for the proposed rule, we would have
needed to investigate other rehabilitation therapy survey data sources
for projecting the 1989 and 1991 data to a current base period such as
1995 and for use in future guidelines. In addition, although the 1989
and 1991 BLS survey data continue to meet the rigorous publication
standards of BLS and provide the only statistically reliable national/
regional data for wages by occupation of which we are aware, questions
have been raised as to whether the BLS data meet the Senate Committee
on Finance's recommendation on timeliness. We took this concern into
consideration explicitly in the proposed rule. Furthermore, the BLS
hospital occupational/industry wage surveys of 1989 and 1991 include
only hospital data. The last BLS nursing home occupational/industry
wage survey was conducted in 1985. We believed it was reasonable to use
combined hospital and SNF wages in the determination of the guidelines
as was done previously because therapist wage levels are primarily
determined in occupational labor markets, not in separate or isolated
industry labor markets. We also needed to review the SNF therapist data
so that we could determine the wage levels in SNFs holding all other
factors constant (including local labor market conditions, and working
conditions).
Comment: We received numerous comments regarding the strengths and
weaknesses of the various data sources that we proposed to use to
determine the guidelines.
Response: We intend to utilize five additional data sources for
hospital wages and two additional data sources for freestanding SNF
wages, each of which we discuss in detail below. We acknowledge the
commenters observations of strengths and weaknesses present in several
of the data sources. However, to delete any one data source would give
more weight to the remaining data sources, which have their own
strengths and weaknesses. To delete any data source with any weakness
relating to statistical reliability would leave only the BLS data which
are not as timely as we would have preferred. Although we received many
comments about the strengths and weaknesses of the various data sources
that we did use, we did not receive compelling evidence to either add
or delete any data source or change the equal weight given to each data
source.
A summary of the different data sources appears below the summaries
of the public comments we received and our responses to those comments.
1. BLS Data--General
BLS collected average hourly earnings (AHE) data for all four types
of therapists in 1989. However, the January 1991 BLS survey included
only the average hourly earnings for full-time physical and respiratory
therapists (BLS January 1991 average hourly earnings for full-time
physical and respiratory therapists were found in the BLS Occupational
Wage Survey: Hospitals, January 1991, pp. 36-119). The hospitals in
this survey employed 50 or more workers. We therefore needed to
estimate 1991 average hourly wages for speech language pathologists and
occupational therapists at the full labor market rate. To do so, we
started with the BLS 1989 survey of all four types of therapists as a
baseline (BLS Industry Wage Survey: Hospitals, March 1989 (the latest
previous survey), pp. 33-118). The hospitals in the 1989 survey
employed 100 or more workers. Our analysis of the University of Texas
survey data for U.S. hospitals indicated that the wages for speech
language pathologists and respiratory therapists increased at similar
rates between 1989 and 1993. Wages for occupational therapists also
increased at rates similar to that for physical therapists during that
period. Therefore, we determined that we could employ the 1991 to 1989
growth rates of respiratory therapist wages and of physical therapist
wages in order to estimate 1991 wage levels for speech language
pathologists and occupational therapists, respectively.
To update the data for the four therapist types from 1991 to later
periods, we derived rates of increase for the period from January 1991
through January 1994 (the period which predates the additional data
sources that HCFA used) and based 50 percent on American Hospital
Association Panel wage data and 50 percent on the average hourly
earnings for hospital workers published by the BLS Current Employment
Statistics Survey, SIC Code 806 (Hospitals). The additional industry
data sources, detailed below, that HCFA used were surveyed in 1994-
1995.
For the period from January 1994 through October 1995, we updated
the BLS occupational industry wage data for the four therapy types
using the BLS Current Employment Statistics Survey for hospital worker
hourly earnings. By incorporating the American Hospital Association
data, which had a higher rate of increase than the BLS data during the
January 1991-January 1994 period, HCFA captured the relatively faster
growth in therapist wages during the period, resulting in wage levels
that reflected current market conditions in January 1994. As mentioned
above, we used the BLS Current Employment Statistics Survey to trend
therapist wage increases from 1994 to 1995.
Comment: One commenter stated that most data sources that HCFA
used, especially BLS and Mutual of Omaha, were not statistically valid.
Specifically, the commenter argued that the BLS data were biased and
the extrapolation of the BLS survey to non-surveyed areas was
[[Page 5110]]
not a valid statistical procedure, especially since there was no known
relationship between surveyed areas and non-surveyed areas. Several
commenters noted that the National Association for the Support of Long-
Term Care (NASL) and the American Health Care Association (AHCA)
surveys provide timely and accurate data and should be the only data
sources used for the salary equivalency guidelines in SNFs. One
commenter concluded that the BLS survey had a ``high response rate''
and the data were reliable.
Response: We agree that no available data source is ideally suited
for all purposes. The data sources used may contain biases that we were
unable to remove using standard statistical editing routines. We
believe that the biases go in both directions and tend to offset each
other. Given that the mean hourly wages of therapists generally cluster
in rather small ranges, we believe that an average of the various
sources, including any inherent biases, fairly represents the national
wage rate for each of the four therapist types. We agree that the NASL
and AHCA databases are timely, but each has shortcomings regarding
representativeness. We address specific comments concerning the Mutual
of Omaha data and the issue of separate salary equivalency guidelines
for each setting later.
Comment: One commenter stated that Congress does not want HCFA to
use the BLS data because Congress discontinued funding for these
surveys in 1992.
Response: Congress discontinued funding for these surveys for
reasons unrelated to the salary equivalency guidelines. The BLS surveys
were replaced by the Occupational Compensation Survey (OCS). We could
not use the OCS because it did not contain the level of detail by
occupation required for use in establishing salary equivalency
guidelines.
2. National Association for the Support of Long-Term Care (NASL)
In March 1996, NASL, representing a portion of the rehabilitation
therapy industry, submitted an October 1995 sample survey of salaried
therapists in hospitals and nursing homes to HCFA, as allowed under our
regulations. This survey did not meet the requirements of the
regulations at Sec. 413.106(b)(6), since the survey design,
questionnaires, and instructions were not approved by HCFA prior to the
start of the survey. The survey did provide data that were current in
SNFs and hospitals, and some documentation was furnished. We,
therefore, conducted a special analysis of this NASL survey data,
including a limited audit of the survey records. Based on this analysis
and limited audit, we determined that the survey was not adequate as a
sole or primary source of data in determining the guidelines, but could
be useful in combination with other data sources. There were several
reasons for this determination:
<bullet> The data were not audited or certified by an independent
party. We were permitted to conduct an audit of the survey records only
under stringent restrictions designed to protect the confidentiality of
the survey respondents. Those restrictions made it impossible for us to
verify the survey results. For example, we were unable to compare
submitted survey data with data from other sources.
<bullet> The verification survey, conducted to determine the
reliability of data submitted by mail, did not appear to be adequate.
Only five providers were included in the verification survey.
Specifically, we were not satisfied that the verification sample was
either sufficiently large or adequately representative.
<bullet> The survey is not sufficiently representative. There were
variable response rates for hospitals and SNFs. The response rate for
hospitals was 10.8 percent and the response rate for SNFs was 29.9
percent. In addition, the sample seemed to include an
overrepresentation of large hospitals and chain-affiliated SNFs.
Because there is an underrepresentation of small hospitals and non-
chain SNFs in the NASL survey, we cannot be assured with this small
response rate that the large hospitals and chain-affiliated SNFs will
adequately represent the small hospitals and non-chain SNFs not
included in the survey. (The GAO stated in its report, ``Medicare Early
Resolution of Overcharges for Therapy in Nursing Homes is Unlikely'',
August 16, 1996, p. 7, regarding the NASL survey data, ``However, the
survey response rate was low (10 percent for hospitals and 30 percent
for SNFs), which raises questions about how representative the data
are.'' In a footnote on that page, GAO points out, ``Official
government surveys generate a much higher response rate. The BLS White
Collar Pay Survey (one component of which was the hospital salary data
survey on which the draft guidelines were based) has an overall
response rate of 82 percent. Typically, BLS response rates exceed 80
percent).''
<bullet> Despite requests for the raw unedited data file, the file
was not provided to us.
<bullet> We have questions about the validity of certain edits.
<bullet> We were also concerned that supervisory time and
compensation in lieu of benefits were not consistently reported.
Additionally, we were concerned that the supervisory time included in
the NASL survey was above a certain threshold that we use in developing
the guidelines.
Comment: Some commenters challenged HCFA's characterization of the
NASL data and felt that HCFA should give greater weight to the NASL
data for a variety of reasons.
Response: In general, the mean wages from the various data sources
we used were rather tightly clustered. None of the commenters offered
compelling evidence that NASL data should be weighted preferentially.
Therefore, we did not change the weighting of any of the data sources
used.
Comment: One commenter stated that the NASL data have response
rates comparable to those achieved in unspecified BLS studies, hospital
industry studies, and long-term care studies. The same commenter
pointed out that the NASL data consisted of responses from 711
institutions while the BLS data were from 628 institutions. Another
commenter stated that the NASL survey suffered from a low response
rate.
Response: The NASL surveyed hospitals, hospital-based SNFs, and
freestanding SNFs while the BLS surveyed hospitals only. The response
rate of the BLS survey was 84 percent, in contrast to the response rate
of the NASL survey, which was 20 percent in the aggregate (10 percent
for hospitals and 29 percent for SNFs). We agree with the comment that
the response rate for the NASL data was low with respect to statistical
sampling theory. While, the validity and reliability of a sample survey
depends primarily upon the representativeness of the sample, not on the
number of responses (assuming an adequate sample size), we have
concerns about the representativeness of the NASL survey. These
concerns, along with the low response rate to the survey, lead us to
believe that the NASL data should be given no greater weight than the
data from other sources.
Comment: One commenter asserted that the NASL survey followed a
rigorous statistical design in consultation with HCFA and that the NASL
data were as good as the data HCFA used.
Response: HCFA did comment and make suggestions on some aspects of
the
[[Page 5111]]
statistical design. NASL did not, however, implement all of the
suggestions that HCFA felt were necessary for a valid statistical
design. Nevertheless, we are using the NASL data in conjunction with
data from several other sources, giving it the same weight as all other
data sources.
Comment: One commenter defended the quality of the NASL data by
stating that HCFA performed an audit of the data, although limited by
conditions set by NASL.
Response: The restrictions set by NASL were such that essentially
all that HCFA was able to perform during its on-site visit to NASL was
a review. The data were not audited or certified by an independent
party. We were permitted to review the survey records only under
stringent restrictions designed to protect the confidentiality of the
survey respondents. Those restrictions made it impossible for us to
verify the survey results. For example, we were unable to compare
submitted survey data with data from other sources.
Comment: One commenter noted that the NASL survey benefitted from a
verification survey.
Response: We concur that verification surveys are beneficial, but
our review of the NASL survey disclosed that the number of provider
verifications actually conducted was extremely limited. As stated
earlier, there were only 5 verifications on 711 responses, a number too
small to give statistical significance to the result.
Comment: Several commenters recommended that HCFA use only the NASL
and/or AHCA data from SNFs to develop rates for SNFs.
Response: As stated above, HCFA has blended SNF and hospital data
in our previous notice and we see no valid reason not to do so again.
In addition, we found a number of shortcomings with the NASL data and
the AHCA data, which we found to be biased toward SNF chains and to
include some supervisory data. We edited the data as much as possible
to improve data quality, but did not use either data source alone to
develop rates for SNFs. We address the issue of separate salary
equivalency guidelines for each provider setting later in this final
rule.
Comment: Several commenters pointed out that the NASL data were the
most timely data available.
Response: We agree that the NASL data were the most timely data
available, but, as discussed earlier, timeliness alone does not
sufficiently meet the criterion for validity and reliability.
Comment: One commenter noted that the NASL data were skewed toward
larger hospitals.
Response: We concur that the sample responses were skewed toward
larger hospitals as well as larger SNF chains but, as stated earlier,
some of the other data sources are biased in other ways as well. The
extent of response bias within the reweighted data is not possible to
quantify without some additional survey work. Again, by combining data
sources with different biases, we believe that the biases tend to
offset each other as evidenced by the clustering of means.
3. Texas National Hospital Survey (1994 National Survey of Hospital and
Medical School Salaries, University of Texas Medical Branch, Galveston,
TX, 1994, pp. 15-19)
The University of Texas National Hospital Survey data are from
October 1994. This annual survey of hospitals is voluntary. The survey
has been conducted for many years for hospitals in various regions of
the country to use as a benchmark of regional wage levels for specific
health professional occupations. While there are data from all regions
of the United States, the survey was not designed to meet the rigorous
BLS standards for representativeness or statistical validity at the
regional level. It does, however, give reasonable levels at the
national level when compared to other data sources.
Comment: One commenter stated that it was inappropriate for HCFA to
use the University of Texas survey of hospitals in the United States
because the data ``includes medical schools with a low wage bias to
establish rates of pay for therapists that are working primarily in
SNFs.''
Response: The commenter's assertion is incorrect because the mean
wages from the University of Texas data clustered with the mean wages
from other data sources. Specifically, the University of Texas mean
hourly wage ranged from being $0.19 higher to $0.83 lower than the mean
hourly wage for the four therapy types using all the data sources--a
range well within reasonable boundaries associated with statistical
variation. For physical therapists, the University of Texas mean wage
was $20.29; the mean wage from all sources of hospital wage data was
$21.00, a difference of 3 percent. For occupational therapists, the
University of Texas mean wage was $19.28; the mean wage from all
sources of hospital wage data was $19.73, a difference of 2 percent.
For speech language pathologists, the University of Texas mean wage was
$18.58; the mean wage from all sources of hospital wage data was
$18.67, a difference of less than one percent. For respiratory
therapists, the University of Texas mean wage was $15.74; the mean wage
from all sources was $15.58, a difference of negative one percent.
4. American Health Care Association (AHCA) Data
The AHCA report includes data on both SNFs and hospitals. The SNF
data for January 1995 are both current and industry-specific. However,
the data are unevenly edited and appear to include some supervisors and
additional salary in lieu of benefits. The sample is heavily weighted
by large chains that are members of the Association. The SNF data,
unlike BLS data, appear as both employee-weighted and facility-weighted
averages and, therefore, do not permit computation of a median or 75th
percentile levels for individual workers.
Comment: One commenter objected to HCFA's observations concerning
the 1994 and 1995 AHCA survey data and indicated that HCFA's criticisms
were unreasonable, given the lack of alternative sources and the
constant enhancement of the AHCA database since 1987. In particular,
the commenter objected to HCFA's observations that the AHCA data were
``unevenly edited and appear to include supervisors and additional
salary in lieu of benefits,'' stating that HCFA fails to acknowledge
discussions addressing these issues. The same commenter suggested that
HCFA give the AHCA data greater weight because they were both timely
and accurate, noting that: (a) AHCA data are exhaustively and
consistently screened and cleaned with participants and the database is
certified by Buck Consultants as being representative; (b) Buck
Consultants has taken steps to insure that supervisory data are
excluded from the data; (c) there are no wages or salary in lieu of
benefits in the data; and (d) this is an annual study, given the same
scrutiny each year and, therefore, should increase the degree of
confidence that HCFA has in the data. Other commenters acknowledged the
bias in the AHCA data toward large chains and indicated that HCFA could
correct the AHCA survey for large company bias as well as individual
data point analysis and exclusion of supervisory rates.
Response: We acknowledge the steps taken to improve the quality of
the AHCA data over time, and agree that the quality of the data has
improved. Our analyses of the 1994 and 1995 AHCA survey indicate that
the survey is still not representative of Medicare-certified
facilities; it represents primarily large chains that are members of
AHCA. We
[[Page 5112]]
made the same observations as did some commenters regarding AHCA data
deficiencies and took steps to exclude supervisory data. HCFA did not
have the necessary information to correct for large company bias. We
believe that the biases tend to offset the data as evidenced by the
clustering of mean wages. Further, individual worker data are not
available to validate the reasonableness of the means for each
institution. For these reasons, it would not be appropriate for HCFA to
modify the weights given to the AHCA data, or to use these data as the
sole source in developing the salary equivalency guidelines.
Comment: Another commenter asserted that the NASL and AHCA data
probably contained more responses from therapists than were contained
in the BLS studies and that the occupational nature of therapists
should outweigh the industry focus created by counting numbers of
institutions.
Response: The 1989 and 1991 BLS samples had responses from 536 and
628 hospitals, respectively. The 1989 and 1991 BLS data that we used
contained responses from 12,672 certified therapists as follows: 3,668
in physical therapy (1991); 1,742 in occupational therapy (1989); 668
in speech language pathology (1989); and 6,594 in respiratory therapy
(1991). The post-edit NASL survey had responses from 191 hospitals, 50
hospital-based SNFs, and 351 freestanding SNFs. The post-edit NASL
survey contained responses from 5,741 registered/certified therapists
as follows: 1,720 in physical therapy; 1,204 in occupational therapy;
680 in speech language pathology; and 2,137 in respiratory therapy. The
AHCA data contained responses from 3,515 certified therapists: 1,806
physical therapists; 1,405 occupational therapists; and 304 speech
language pathologists. The commenter was apparently seeking to give
more weight to the NASL and AHCA data because ``the number of
therapists reported in the NASL and AHCA survey probably exceeds the
numbers reported in the BLS studies * * *'' implying that the two
industry data bases are more reliable for that reason. In fact, the BLS
studies (12,672 therapists) we used contained 37 percent more
therapists than the NASL and AHCA data combined (5,741 and 3,515,
respectively).
5. Maryland Health Services Cost Review Commission Data
The Maryland Health Services Cost Review Commission conducts an
annual census of occupational wage rates for all Maryland hospitals. We
analyzed data from the 1995 census. While this is a complete census
covering over 50 hospitals, it is for Maryland only. In addition,
speech-language pathologists are not included as a separate
occupational category.
Comment: One commenter noted that the Maryland Health Services Cost
Report Commission's database is not representative of the United States
because the data are from only one State. Further, the commenter noted
that speech language pathologists are not separately identified in the
data.
Response: Despite its shortcomings, the strengths of the Maryland
census are that it is timely, accurate, and contains data from
providers of various sizes in geographically diverse urban and rural
areas. It is a rich data source for variations in occupational wage
levels by degree of urbanization. In fact, the mean hourly wage for
physical therapists in the Maryland data was $20.78; the mean wage from
all sources of hospital wage data was $21.00, a difference of only 1
percent. The mean hourly wage for occupational therapists in the
Maryland data was $20.60; the mean wage from all sources of hospital
wage data was $19.73, a difference of 4 percent. The mean hourly wage
for respiratory therapists in the Maryland data was $16.20; the mean
wage from all sources of hospital wage data was $15.58, a difference of
four percent. We used the data because we concluded that its strengths
outweighed its weaknesses for our specific purpose.
6. 1995 American Rehabilitation Association (ARA) Salary Survey
The ARA collected July 1994 data from its members that are medical
and residential rehabilitation providers. Among ARA members are CORFs
that provide physical therapy, respiratory therapy, speech language
pathology, and occupational therapy services to Medicare and Medicaid
beneficiaries. The response rate was low and the Association indicated
in its report that these data cannot be presumed to represent the full
population of rehabilitation facilities. However, this survey appears
to give reasonable wage levels at the national level when compared to
other data sources. Information on SNFs was not reported due to an
inadequate sample size.
Comment: One commenter noted that the ARA survey had a low response
rate and that it could not be assumed to be representative. Another
commenter noted that despite the low response rate, the results
appeared to yield reasonable wage levels nationally.
Response: We agree with the observations of both commenters.
Although the data could not be assumed to be representative, they were
reasonable and fairly close to the other data sources we used. In fact,
the mean hourly wage for physical therapists in the ARA freestanding
hospital data was $20.82; the mean wage from all sources of hospital
wage data was $21.00, a difference of less than 1 percent. The mean
hourly wage for occupational therapists in the ARA freestanding
hospital data was $18.90; the mean wage from all sources of hospital
wage data was $19.73, a difference of only 4 percent. Similarly, the
mean hourly wage for physical therapists in the ARA rehabilitation unit
data was $21.12; the mean wage from all sources of hospital wage data
was $21.00, a difference of less than one percent. The mean hourly wage
for occupational therapists in the ARA rehabilitation unit data was
$19.82; the mean wage from all sources of hospital wage data was
$19.73, a difference of less than one percent. As is the case with the
other data sources, we used the ARA data because we concluded that its
strengths outweighed its weaknesses.
7. Mutual of Omaha Data
Mutual of Omaha, an HCFA intermediary, conducted a survey of about
2,000 Medicare SNF providers in 1995. Data were collected on contract
therapy prices and salary rates for occupational therapy and speech
language pathology.
Comment: Several commenters stated that the Mutual of Omaha survey
was not statistically valid because of inadequate sample design, no
analysis of respondents vs. nonrespondents, too small a sample size,
overrepresentation of hospital-based SNFs and contract therapists, no
physical therapist or respiratory therapist data, and data that were
limited to aggregate facility data as opposed to data points for each
employee. The weight of many comments is reflected in their assertions
that the average wage rates of occupational therapists and speech
language pathologists reflected in the Mutual of Omaha data are out of
line with other data sources.
[[Page 5113]]
Response: We agree that the Mutual of Omaha survey does not meet
the rigorous sample design requirements of the BLS survey data included
in our estimates. However, we did use it in combination with the other
described data sources. The Mutual of Omaha data are similar to other
data sources such as AHCA and the American Rehabilitation Association
(ARA) that reflect universes other than the national. The Mutual of
Omaha estimate of the mean hourly wage level of occupational therapists
in SNFs in October 1995 that we used in the salary computation was
$22.90, compared to the mean wage rate of all SNF data sources of
$20.33. The Mutual of Omaha mean wage rate for occupational therapy is
thus 13 percent above the mean wage rate of all data sources. The
Mutual of Omaha mean wage rate for speech language pathologists in SNFs
in October 1995 was $20.34 compared to the mean wage rate of $19.26.
The Mutual of Omaha mean wage rate for speech language pathologists is
thus 6 percent above the mean wage rate of all SNF data sources.
8. Unused Data Source--``A Study of Respiratory Care Human Resources in
Hospitals 1992''
This survey was conducted by the American Association of
Respiratory Care's (AARC) Task Force on Professional Direction in
conjunction with consultants from Arthur Andersen & Co. The AARC
surveyed 2,732 of 4,900 hospitals having respiratory care departments
and received 858 responses (31 percent response rate), comprising 17
percent of all hospitals with respiratory care departments.
Comment: One commenter inquired as to why HCFA did not use this
study by AARC in conjunction with consultants from Arthur Andersen &
Co.
Response: HCFA used data from academic (e.g., University of Texas),
government and industry-wide surveys for hospitals, SNFs, etc. that
included occupational specific data. HCFA did not use data sources
specific to one occupational category from its own professional
association, e.g., American Occupational Therapy Association data.
Using specific occupational data from a particular association may have
biased the results relative to the other occupational categories, given
the wide discretion used in defining wages, income, and statistical
design among the four occupational groups.
B. Methodology
In order to establish the proposed hourly salary equivalency
amounts, we determined the ``best estimate'' of wages for both
hospitals and SNFs. We first found mean wage rates for each of the data
sources listed above.
BLS surveyed average hourly earnings (AHE) for all four therapies
in 1989. However, their January 1991 survey included the average hourly
earnings only for full-time physical and respiratory therapists. (BLS
January 1991 average hourly earnings for full-time physical and
respiratory therapists were found in the BLS Occupational Wage Survey:
Hospitals, January 1991, pp. 36-119. The hospitals in this survey
employed 50 or more workers.) We, therefore, needed to estimate 1991
average hourly wages for speech language pathology and occupational
therapy. To do so, we started with the BLS 1989 survey of all four
therapies as a baseline (BLS Industry Wage Survey: Hospitals, March
1989 (the latest previous survey), pp 33-118). The hospitals in the
1989 survey employed 100 or more workers. Our analysis of the
University of Texas data for U.S. hospitals indicated that the wages
for speech language pathology and respiratory therapy increased at a
similar rate between 1989 and 1993. Wages for occupational therapy and
physical therapy also increased at a similar rate during that period.
Therefore, we determined that we could employ the 1989 ratios of speech
language pathology to respiratory therapy, and of occupational therapy
to physical therapy, in order to estimate 1991 wage levels for speech
language pathology and occupational therapy. Specifically, multiplying
the ratio of 1989 average hourly occupational therapy wages to 1989
average hourly physical therapy wages by 1991 physical therapy wages
yielded estimated 1991 occupational therapy wages. The following
formula summarizes the computation (all values are average hourly
wages):
[(March 1989 AHE, OT) / (March 1989 AHE, PT)] x (January 1991 AHE,
PT) = (estimated January 1991 AHE, OT).
Similarly, multiplying the ratio of 1989 average hourly speech
language pathology wages to 1989 average hourly respiratory therapy
wages by the 1991 average hourly respiratory therapy wages yielded
estimated 1991 average hourly speech language pathology wages. Again,
the following formula summarizes the computation (all values are
average hourly wages):
[(March 1989 AHE, speech language pathology) / (March 1989 AHE,
respiratory therapy)] x (January 1991 AHE, respiratory therapy) =
estimated January 1991 AHE, speech language pathology.
The American Health Care Association data provided facility-
weighted mean wage rates for SNFs. The Association has estimated that 5
percent of the SNF wage rates represented supervisors and additional
wages paid in lieu of fringe benefits. We used that estimate to reduce
the Association survey wage data to a nonsupervisory, no additional
salary in lieu of benefits basis.
We converted annual data in the American Rehabilitation Association
and University of Texas surveys to hourly wages using a divisor of 2080
hours, which represents a standard work year.
The Maryland Health Services Cost Review Commission census data
provided wage data, paid hours, and numbers of personnel for each
hospital. We eliminated data for employees who worked less than 35
hours or more than 40 hours a week to restrict the computation to full-
time employees only. We then determined the average hourly wage for
each hospital by dividing aggregate wages by the number of paid hours.
Finally, we computed the average hourly wages across all hospitals,
weighted by the number of employees in each hospital.
NASL data were first divided by 52 to arrive at weekly salary, then
divided by the number of hours worked per week which were also given in
the survey, to obtain hourly wage rates. As in the case of the Maryland
census data, we eliminated data for employees who worked less than 35
hours, or more than 40 hours a week to restrict the computation to
full-time employees only.
We trended all data to the 1995 fourth quarter as described in
detail in the March 1997 proposed rule. We then determined the salary
equivalency guideline amounts for 1998 in five steps. Those five steps
were: (1) Determine average wages by therapy type, separately for
hospitals and nursing homes; (2) blend the hospital and nursing home
average wages by therapy type, to yield average wages by therapy type
for the four occupational markets; (3) approximate the 75th percentile
of wages by therapy type; (4) calculate salary equivalency guideline
levels for fourth quarter 1995, by adding amounts for fringe benefits,
rent, etc.; and (5) update these guideline amounts to April 1, 1998,
the proposed effective date.
In the first step, we determined the mean wage levels, by therapy
type, for hospitals in each of the available data sources. (Data
sources used for hospitals were: BLS, Industry Wage Survey: Hospitals,
March 1989 and
[[Page 5114]]
Occupational Wage Survey: Hospitals, January 1991; University of Texas
1994 National Survey of Hospital and Medical School Salaries; American
Rehabilitation Association's surveys of freestanding hospitals and of
rehabilitation units, 1995 Salary Survey; Maryland Health Services Cost
Review Commission's census of hospitals; American Health Care
Association hospital report's data profile, 1994 AHCA Survey; and NASL
1995 survey of hospitals). We similarly determined the mean wage
levels, by therapy type, for nursing homes in each of the available
data sources. (Data sources used for SNFs were: 1995 NASL survey of
SNFs; American Health Care Association survey of SNFs, 1995 AHCA
Survey; and the 1996 survey of SNFs by Mutual of Omaha). We then
averaged the mean wage levels from the available data sources by
therapy type, separately for hospitals and nursing homes.
In the second step, we blended the hospital and nursing home
average wage levels, by therapy, to yield average wage levels by
therapist type across the four occupational markets. We employed a
blending process used in the previous salary equivalency guidelines
notice (48 FR 44922, September 30, 1983), to weight the occupational
averages by relative employment levels in hospitals and nursing homes,
respectively. To establish appropriate weights, we used employment of
therapists in nursing homes (Standard Industrial Classification (SIC)
Code 805) and in hospitals (SIC Code 806), as found in the BLS
Occupational Employment Statistics survey (OES). (The most recent
available survey of employment in nursing homes is for 1993, while the
most recent survey data of employment in hospitals is for 1995.) We
applied these weights to the mean hospital and SNF wage rates by the
four therapist types as determined in the first step. The BLS
Occupational Employment Statistics survey shows that the hospital
industry is a major employer of therapists of all types, while SNFs
employ fewer salaried therapists. The weights for hospitals and nursing
homes, respectively, are: for physical therapy, 85 percent and 15
percent; for occupational therapy, 85 percent and 15 percent; for
speech language pathology, 82 percent and 18 percent; and for
respiratory therapy, 99 percent and 1 percent.
In the third step we approximated the 75th percentile of the
blended wage rates for each therapy occupation. It was necessary to
approximate the 75th percentile because, unlike our previous
computations of the guidelines, in this proposal we could not determine
percentile values directly from each of the sources. We have observed
in the BLS data and a regression analysis we performed on NASL data
that the 75th percentile was approximately 110 percent of the mean. We,
therefore, proposed to increase each of the four blended wage averages
by 10 percent to approximate the 75th percentile of wages in each
discipline across the occupational market. (In response to comments on
the proposed rule, however, we have increased the factor to estimate
the 75th percentile from 110 percent of the mean to 112 percent of the
mean to reflect inherent variations that we were not able to quantify.)
The inherent variations are due to estimating national rates for each
of the four rehabilitation therapies, then using the GPCI to
approximate wage and fringe levels in all geographic areas of the
United States. Data does not exist to verify that, for each of the four
therapies, every local labor market in the United States is accurately
portrayed by the GPCI.
Salary equivalency guidelines are based on the therapists' time in
the facility. Adjustments to average hourly earnings data were
necessary to include a reasonable allowance for vacation, sick leave,
and administrative time. In order to convert the average hourly
earnings from an hours paid basis to an hours worked basis, we applied
a factor of total paid hours divided by hours worked (2,080 <divide>
1,808) to the average hourly earnings determined thus far, which is the
same methodology used in the previous notice. The 1,808 figure was
computed based on 2,080 hours (40 hours/week x 52 weeks; a standard
work year) less 15 vacation days, 10 sick leave days and 9 holidays
equal to 34 days, or 272 hours. Data on leave benefits come from the
BLS Employee Benefits Survey. (U.S. Department of Labor, Bureau of
Labor Statistics: Employee Benefits in Small Private Establishments,
1992, Bulletin 2441, U.S. Government Printing Office, May 1994, pp. 10-
20.)
In the fourth step, we added fringe benefit and expense factors to
the prevailing salary rates determined for each therapy type. The
fringe benefit and expense factors are intended to recognize fringe
benefits that are received by an employee therapist, as well as
overhead expenses that a therapist or therapist organization might
incur in furnishing services under arrangements. These factors are
expressed as percentages of the prevailing hourly rate and are applied
to every hour of service furnished at the provider site. Fringe
benefits may include vacation and sick pay, insurance premiums, pension
payments, allowance for job-related training, meals, severance pay,
bonuses, etc.
We computed fringe benefits as a percent of total compensation
using fiscal year 1994 Medicare cost reports for hospitals under the
prospective payment system. We believe these data are the best proxy
for therapist fringe benefit information, which is not available for
SNFs. We used the Medicare cost reports for prospective payment system
hospitals to obtain fringe benefit information because these data are
carefully scrutinized; they are used to adjust the labor portion of
hospital payments under the prospective payment system. Also, the BLS
Employment Cost Index (ECI) for March 1994 showed that fringe benefits
for professional and technical workers in hospitals and nursing homes
were similar. In the proposed rule, the fringe benefit component was
about 14 percent of the total salary equivalency guideline amount. In
the final rule, we have, instead, added the amount determined from the
adjustment to average hourly earnings for vacation, sick leave, and
administrative time to the fringe benefit amount excluding leave
determined from the hospital cost reports. By including paid leave in
fringe benefits rather than in salary, the final weight for fringe
benefits is about 20 percent of the guideline amount or about 28
percent of total compensation.
The expense component takes into account expenses a therapist or
therapist organization might have, such as maintaining an office,
purchasing insurance, etc. We based the expense component of the
guidelines on an estimate of the costs of maintaining a therapy
services office. The general methodology for computing the expense
component is similar to that used in the September 30, 1983 notice (48
FR 44922) but the factors have been revised. This component has rental
and non-rental portions.
To determine the rental portion of the expense component, we used
the 1995 rental rate data compiled by the Building Owners and Managers
Association International (BOMA) and published in the 1996 BOMA
Experience Exchange Report for Downtown and Suburban Office Buildings.
(Building Owners and Managers Association International: 1996 BOMA
Experience Exchange Report, Washington, DC, 1996, p. 17.) BOMA reported
a national rent average, excluding utility cost, of $18.37 per square
foot per year. We applied an occupancy factor of .887 to take into
account the space used for rental building hallways, elevators, etc.,
that
[[Page 5115]]
are included in the BOMA rent figure, but are not part of the area
rented for an office. We then added the BOMA utilities cost of $1.82
per square foot. We determined total rental cost, assuming a rental
area of 250 square feet, the same rental area used in prior schedules
of guidelines. The total 1995 rental cost was divided by 1,808 (the
hours factor applied to average hourly earnings) to compute rental cost
per hour worked in 1995.
The expense component includes costs of maintaining an office, such
as wages and salaries of administrative and clerical help, insurance,
telephones, etc. Medicare pays for services at their reasonable cost.
It has been reported to HCFA that an effective and efficient
rehabilitation therapy firm incurs overhead expenses of about 25
percent. We estimate this component, including rent, to be within a
reasonable cost range of 28.2 percent of total expenses in 1995. The
1995 rent per square foot amount and the other expenses amount were
constant across the four therapy types, implying that the share of
these costs vary by therapy type (the share for rent is lowest for
physical therapy since the physical therapy wage rates are the
highest).
As described in detail in the proposed rule, we added the fourth
quarter 1995 dollar values of the ``blended'' wages, fringe benefits,
rent, and the remainder of the other expenses factors to obtain salary
equivalency guideline amounts for fourth quarter 1995. We updated the
resultant fourth quarter 1995 salary equivalency guideline amounts to
April 1998, using a Standard & Poor's DRI 1997:4 forecast.
1. Occupational Labor Market
In calculating the salary equivalency guidelines proposed on March
28, 1997, HCFA used a blend of hospital and SNF therapist wages. We
also used a blend of hospital and SNF therapist wages in the
establishment of salary equivalency guidelines for physical and
respiratory therapy in the September 30, 1983 notice. The use of a
blended wage reflects the influence of occupational labor market
conditions on rehabilitation therapist wages, given the substantial
degree of mobility between the settings. In the proposed rule, the
labor market for therapists was characterized as an integrated
occupational market in which therapists working in hospitals and SNFs
have the potential to migrate between the two settings with relatively
little difficulty resulting from differences in job requirements. We
noted, however, that wage levels across settings for the same
occupation may differ due to reasonable compensating wage differentials
associated with working conditions, risk of injury, and geographic
location. Wage differentials may also be associated with differences in
worker characteristics, such as experience and skill. When these
factors are taken into account, the ability to move across settings
should ensure that the wage levels between these settings bear a
reasonable relationship over time.
Comment (general): Many of the comments on the proposed rule have
focused on the issue of compensating wage differentials, asserting that
HCFA should not blend wages of hospital and SNF therapists in the
establishment of salary equivalency guidelines. These comments maintain
that wage differentials that exist between the two settings can be
fully explained by a combination of higher skill requirements and a
less agreeable work setting in SNFs. For this reason, commenters claim
that the full difference in wages should be recognized by HCFA.
Response: The assertion that differences in skills and work
environment fully explain current differences in wage rates rests on
the assumption that compensating wage differentials between hospitals
and SNFs are equivalent to the actual wage differentials observed at a
point in time. However, there are a number of factors which may cause
actual wage differentials to vary from those associated solely with
differences in skills or environment. These factors include adjustments
to short-term shifts in demand, entrance barriers to the therapy
professions which have slowed adjustment to these shocks, and
distortions to the operation of markets for therapy services and labor
within the SNF sector caused by the inflation of prices and wages by
Rehabilitation Therapy Firms (RTFs) to quickly gain market share as
well as the lack of sufficient efforts to minimize costs by SNFs.
HCFA contracted with Standard & Poor's DRI to study this issue.
Their data indicate that HCFA's proposed salary equivalency rates,
incorporating the adjustment to the 75th percentile of the wage
distribution, are more than sufficient to cover legitimate compensating
wage differentials for skills and work environment in SNFs, as well as
the wage differential which would result from increases in demand for
therapy services in SNFs given cost-minimizing behavior by SNFs and
RTFs.
Comment: Some commenters stated that there is no legal foundation
for using a blended wage rate for hospitals and skilled nursing
facilities to set salary equivalency rates.
Response: We do not believe that the statute prohibits use of a
blended wage rate. We used a methodology based on blending wages from
therapists in hospitals and SNFs in the September 30, 1983 notice which
revised salary equivalency guidelines for physical and respiratory
therapists. In that notice, HCFA established the prevailing salary
component based on a blended hourly wage for hospitals and nursing home
hourly wage at the 75th percentile of the wage distribution. As
discussed in more detail in the Statutory Issues section below, we
believe that this approach comports with Congressional intent as
expressed in the relevant legislative history.
Comment: One commenter stated that the growing wage differential
between therapists in the SNF setting and those in the hospital setting
implies that the labor markets for therapists in these two settings are
separate and distinct. According to the commenter, this indicates that
the concept of an occupational labor market cannot be used as the basis
for establishing salary equivalency rates based on a blend of hospital
and SNF wages.
Response: The key factor involved in determining the extent to
which an occupational labor market is integrated is the
substitutability of professional skills across settings. This
determines the potential for mobility between the two settings. If
workers can flow relatively freely across industry settings, and
markets are functioning competitively, this means that wage rates in
different settings will be influenced by the supply and demand
conditions for that occupation in all settings. This does not mean that
wages will be equivalent. Compensating wage differentials for differing
skills and environments will result in a reasonable relationship of
wages across all settings.
The term ``occupational labor market'' implies some range of
shared, and, therefore, substitutable skills. The question then becomes
whether the extent to which skills required in the two settings are
overlapping, whether the educational requirements are similar, and
whether substantial retraining is required in order for therapists to
move from one setting to another. An examination of these issues for
therapists in hospitals versus SNFs indicates that the required
educational qualifications and skills are extremely similar. All
therapists complete the same accredited education programs and
substantive retraining for individuals moving between these two
settings is not standard. For therapists
[[Page 5116]]
employed in hospitals and SNFs, it is clear that, while not identical,
the skills needed to perform their jobs are highly substitutable. This
is evidenced in commenters' observations by the shift in employment of
roughly a quarter of physical therapists and speech language
pathologists formerly employed in hospitals who have been moved to SNFs
or HHAs without substantive retraining. In addition, both hospitals and
SNFs routinely hire occupational therapists, physical therapists, and
speech language pathologists directly out of college, indicating that
the body of required skills is covered by the general educational
programs completed by all accredited therapists.
The existence of actual wage differentials between two settings
does not indicate that an integrated occupational labor market does not
exist. These differences are not solely those associated with different
skill requirements or working conditions. Short-term differentials may
reflect disequilibrium in response to rapid shifts in employment in the
presence of transaction costs, costs of information, and lags in the
adjustment of the occupational labor supply. These are reasonable wage
differentials that are consistent with cost minimizing behavior.
Differentials may also reflect differences in incentives to minimize
costs between the two settings.
The fact that SNFs may have little economic incentive to minimize
costs, beyond the point where they are subject to risk of audit, will
likely result in higher relative prices for therapy and higher
therapist wages in the SNF sector. Rehabilitation therapy firms may
take advantage of this incentive structure to push prices and wages
beyond prudent buyer rates. Because SNFs have little incentive to
switch suppliers unless the price is far above their current rates,
higher prices will not cause the rehabilitation therapy firm to lose
market share. There is no market pressure to push prices down and less
pressure for rehabilitation therapy firms to minimize costs than would
be the case in competitive cost-minimizing markets.
Comment: Some commenters were concerned that therapists attracted
to SNFs and HHA settings are different from those attracted to
hospitals and, therefore, do not compete in the same labor market. The
wage differential between hospitals and SNFs is a reasonable
compensating wage differential associated with these differences in
skills and work environment. These differences make it harder for SNFs
to recruit qualified therapists. Specific differences between hospitals
and SNFs cited by commenters were:
(a) Therapists in SNFs must work independently with less
supervision. For this reason, SNFs require a more experienced
workforce.
(b) The work environment in SNFs is less appealing than that in
hospitals.
--There is less variety in the case mix,
--Smaller therapy departments in SNFs mean less collegiality, less
potential for advancement, and fewer opportunities for training; and
--Patients in SNFs are more difficult to work with.
Response: We agree that compensating wage differentials potentially
exist between different work settings for therapists. In using a
blended hospital/SNF wage rate as the basis for salary equivalency
rates, the question is the magnitude of these differentials and whether
they are covered by the use of the blended wage rate at the 75th
percentile of the wage distribution.
In response to industry requests for additional statistical
research on this issue, we contracted with Standard & Poor's DRI to
estimate the magnitude of justifiable wage differentials for physical
and occupational therapists, and speech-language pathologists. The
resulting data presents estimates of compensating wage differentials
associated with skills and work environment across industry settings
for 1979-89, wage differentials associated with short-term labor market
disequilibrium under conditions of cost-minimization, and wage premiums
resulting from the failure of many SNFs to behave as cost-minimizers.
The DRI data estimated a net compensating wage differential
associated with education, experience, and work environment which is
very small in comparison to the actual disparity in wages between the
two sectors in 1995. The conclusion of this study was that our proposed
salary equivalency rates, based on the blended wage approach, were more
than sufficient to cover reasonable compensating wage differentials
between hospitals and SNFs and an additional positive short-term
differential for SNF therapist wages associated with the estimated
increase in the relative demand for therapists in SNFs under the
condition that SNFs behave as cost-minimizers.
The commenters maintained that the current disparity of wages is
solely reflective of compensating wage differentials associated with
the underlying fundamentals of skills and work environment. However,
the actual wage differential will be equal to the compensating wage
differential only in cases where product and labor markets are
competitive (i.e. suppliers, providers and consumers are cost
minimizers) and in equilibrium. Neither of these assumptions are met in
the case of the labor market for therapists in rehabilitation therapy
firms and SNFs.
The nursing home reform requirements of the Omnibus Budget
Reconciliation Act of 1987 (OBRA '87), which took effect in 1990,
caused a rapid increase in the relative demand for therapist labor in
SNFs. The continued shift of employment from hospitals, educational
institutions, and other settings towards SNFs, which is associated with
the current observed wage differential, means that the occupational
labor market has not yet reached equilibrium. This indicates that some
part of the wage disparity between hospitals and SNFs is reflective of
continued efforts to sharply increase the share of the pool of
therapists who are employed in SNFs over a fairly short period of time.
In the presence of costs associated with changing jobs and costs of
information about available positions and associated wage rates, these
efforts can be expected to result in a temporary wage differential even
with cost minimizing behavior. This effect is heightened by constraints
on the number of new graduates from accredited therapy programs, since
job mobility is less costly for new graduates than established
therapists. Long queues for entry into accredited therapy programs
indicate that current occupational wages are well above the level
needed to attract new entrants to the professions. This suggests that
difficulties in expanding the capacity of educational programs is
contributing to the cross-sectoral adjustment process. These effects on
therapist wages in SNFs are beyond the control of the SNFs in
minimizing costs, and should therefore be covered by salary equivalency
guidelines.
DRI also has shown that the market for therapy services in SNFs
does not function in the normal parameters of a cost minimization
framework. This analysis relied on a model framework originally
developed by Joseph Newhouse (1978) <SUP>1</SUP> for the analysis of
the behavior of medical cost increases under conditions where cost-
sharing requirements for consumers vary. This model has the implication
that the
[[Page 5117]]
supply of medical services will exhibit increasing inefficiency as the
coverage of costs approaches 100 percent, resulting in higher volumes,
prices, and wages than would otherwise be the case.
---------------------------------------------------------------------------
\1\ Newhouse, Joseph P. The Erosion of the Medical Marketplace,
R-2141-1-HEW, The Rand Corporation, Santa Monica, California,
December 1978.
---------------------------------------------------------------------------
Therapy services provided under arrangement in SNFs represent a
service that closely approximates 100 percent coverage. Medicare Part
A, which accounts for 58 percent of such services, requires zero cost-
sharing for the first 20 days. The daily coinsurance rate that
beneficiaries must pay for days 21 through 100 for 1998 is $95.50.
Medicare pays for all costs over this coinsurance rate. However,
because the daily rate in a SNF is usually higher than the coinsurance
rate (in 1995, the latest year for which data is available, the average
daily rate in a nursing home was $127.16), beneficiaries will pay the
full coinsurance amount whether they receive therapy services or not.
The extra cost of therapy services is usually paid for by Medicare.
Medicare Part B coverage, which begins after 100 days and accounts for
32 percent of therapy services provided in SNFs, requires a 20 percent
copayment which is primarily covered by Medigap policies.<SUP>2</SUP>
An additional 5 percent of services are covered by Medicaid, again with
zero cost-sharing. While 5 percent of contract therapy services are
covered by private insurance (and therefore may be subject to some
cost-sharing), this fraction of the market is too small to introduce
any significant sensitivity to price into this market.
---------------------------------------------------------------------------
\2\ Weiner and Zeid, Comparing Current Cost with Salary
Equivalency Reimbursement for Physical Therapy, Occupational
Therapy, and Speech-Language Pathology, Washington DC, April 1995.
---------------------------------------------------------------------------
In many cases, higher contract therapy costs result in higher
relative reimbursement from Medicare for allocated overhead as well as
for the direct costs of contract therapy services. This reimbursement
methodology and market structure has implications for the behavior of
firms that supply contract therapy services to SNFs. Cost-effective
therapy firms will have little advantage in this market and will not
tend to gain any substantial market share compared to the case of
freely functioning market. In this situation, Newhouse argues that a
competitive cost minimizing supply curve for the industry does not
exist. This invalidates claims that prices, input and output
quantities, and wages in this setting are at cost-minimizing rates that
reflect freely functioning markets.
The Standard & Poor's DRI data produced estimates of hospital/SNF
wage differentials associated with reasonable compensating wage
differentials based on worker and job characteristics as well as
reasonable differentials based on short-term disequilibrium associated
with increases in demand, given cost-minimization by SNFs. The
estimated differentials associated with characteristics of the
workforce and work environment were produced by the estimation of wage
equations relating the hourly wages of individual employees throughout
the U.S. economy with human capital variables such as education and
experience, as well as the systematic differences across occupations
and industry groups that are associated with work environment. This
estimation was based on regression analysis using pooled cross-
sectional data from the 5 percent Public Use Microdata Samples from the
1980 and 1990 decennial census. The census sample incorporates
information for therapists in all settings with information on salary,
hours worked, educational attainment, demographic characteristics, and
location. These data were carefully screened for potential inaccuracies
associated with self-reporting and reviewed for consistency with
licensure requirements and consistency with other available data
sources. The estimation period ended before the implementation of OBRA
'87 in 1990. This indicates that wage differentials associated with the
resulting unanticipated increase in demand, and those associated with
failure to minimize costs in an environment with little restraint on
volume and prices, will not bias the estimated compensating wage
differentials.
DRI data show that in 1989 SNFs were actually able to hire
similarly qualified therapists for a slightly lower wage than could
hospitals, holding skills and environment constant. While it is not
possible to obtain comparable multivariate estimates based on other
data sources because of the lack of available information on skill
variables and other occupational groups, we note that several other
sources from 1989 and surrounding years confirm that actual wage
differentials for SNFs relative to hospitals were small positives or
negatives. The American Speech-Language Hearing Association (ASHA)
reported a negative differential for SNFs relative to hospitals in
1989, while the American Occupational Therapy Association (AOTA)
reports a small positive differential for 1990 (1989 is not available).
The differentials reported by ASHA and AOTA are close to those seen in
the Census wage data without adjustment for skills and work
environment.
Data from the 1990 decennial census indicates that, on average,
physical and occupational therapists working in SNFs do have more
experience than physical and occupational therapists in hospitals,
possibly because some therapists in SNFs need to work more
independently or with less supervision. The estimated wage differential
associated with the greater degree of experience, however, was small
(in both cases less than 5 percent.) The reason for the small
differential appears to be that the greater degree of experience is
usually past the point where additional experience results in a
substantial increase in wages. Rapid increases in wages associated with
experience occur during the first decade of practice with wage
increases for additional years of experience adding little in the terms
of wage gains. Thus, additional years of experience past this point add
relatively little to the wages these therapists can demand. On the
other hand, speech language pathologists in the census sample reported
slightly less experience on average than those employed in hospitals.
However, since the compensating differential resulting from the
Census wage equations applies to the year 1989 (1990 Census), it is
important to analyze how these conditions might have changed between
1989 and 1995. The principle reason why more experienced therapists
might be required in the SNF environment, according to comments, was
the relative lack of supervision for these therapists, when compared to
hospitals, which have larger, more established therapy departments. The
key issue becomes the determination of the direction of changes in the
level of supervision since the year 1989 on which our estimates are
based.
Given the rapid expansion in the volume of therapy services
provided in SNFs, and the larger number of therapists practicing within
a given SNF, it follows that the opportunity to consult supervisory
personnel has actually grown over the past 6 years. This suggests that,
while a gap in the average years of work experience may exist, the size
of the gap is likely to be smaller than was the case in 1989.
Comments on the unappealing work environment in SNFs focused on two
areas: (1) The nature of the work, that is there is less variety in the
case mix, and (2) the lack of collegiality, potential for career
advancement, and training opportunities associated with smaller therapy
departments in SNFs relative to hospitals.
[[Page 5118]]
To apply these estimates to the later period, we must analyze how
these conditions that contribute to the less appealing work environment
would have changed between 1989 and 1995. It would appear that there
would actually be more variety in case mix in 1995 than in 1989 due to
the expansion of therapy services in the SNF setting and the trend
towards discharging hospital patients to SNFs earlier. With the
significant increases in therapy programs in SNFs, it is likely that
career advancement, training opportunities, and the opportunity to work
with other therapists would have grown similar to that of hospitals.
Given the changes in the SNF and hospital environments over the
past 6 years, these environments are likely to have grown more similar,
on average, than otherwise. It, therefore, appears unlikely that the
relative appeal of the two settings would be far different than in
1989. DRI's estimates of compensating wage differentials can therefore
be applied to the later period.
Comment: Several commenters stated that, by combining hospital and
SNF wages, HCFA was not recognizing the full compensating wage
differential for therapists in SNFs.
Response: Our salary equivalency rates cover the full compensating
wage differential for therapists in SNFs (which, as explained above, is
very small), and a reasonable wage differential for the estimated costs
of the increased demand that would have occurred after OBRA '87, under
the condition that all SNFs behaved as cost-minimizers.
The observed relative wages for therapists in SNF settings have
become distorted by the lack of cost minimization efforts in the
provision of therapy supply services in SNFs. Blending hospital and SNF
wage rates, as we have done in the past, provides a methodology that
covers compensating differentials associated with skills and work
environment, while avoiding the validation of increases associated with
the absence of sufficient cost minimization efforts. As hospitals and
educational institutions adjusted wages upwards at a rate slower than
rehabilitation therapy firms and SNFs to retain staff, access to
therapy services in these settings would decrease, while the volume of
services available in SNFs continued to increase beyond the point where
the benefits conveyed to patients justified the costs incurred.
2. Trending Old Data To Reflect Current Conditions
Comment: HCFA should use the percent increases in physical therapy
wages to update speech language pathology wages from 1989 BLS speech
language pathology data to 1991 rather than using the percent changes
in respiratory therapy wages.
Response: The University of Texas data source was the only source
that had all four therapy types over time with relatively consistent
definitions and methodology. The University of Texas data indicated
that the speech language pathology wage growth from 1989 to 1991
correlated better with respiratory therapy wage growth than with
physical therapy wage growth. Therefore, we used the same percentage
growth for speech language pathology wages as existed for respiratory
therapy wages from 1989 to 1991.
Comment: One commenter suggested that the use of 6 -and 8- year-old
BLS hospital data is inappropriate and does not satisfy the Senate
Finance Committee's recommendation for timely and accurate data. This
commenter also suggested that trending forward does not mitigate
distortions of using old data and does not capture the significant
changes in the marketplace over the past several years. Other
commenters stated that because the BLS data were relatively old, HCFA
should give a lesser weight to the BLS data or not use these data at
all.
Response: We believe that our methodology for aging the baseline
BLS data is consistent with Congressional intent. We used the most
recent BLS data available on therapists employed in hospitals and
trended it forward using the best data sources of which we were aware.
If the commenter's assertion that trending forward does not capture the
significant changes in the therapy marketplace were correct, then the
BLS hospital data for therapists trended forward to 1995, would be
substantially different from the best industry data sources for 1995.
In fact, the trended BLS data tend to be at the center of the clustered
industry data sources for 1995. For physical therapists, the trended
BLS mean wage was $20.90; the mean wage from all sources was $21.00, a
difference of less than one percent. For occupational therapists, the
trended BLS mean wage was $19.67; the mean wage from all sources was
$19.73, a difference of less than one percent. For speech language
pathologists, the trended BLS mean wage was $19.30; the mean wage from
all sources was $18.67, a difference of 3 percent. For respiratory
therapists, the trended BLS mean wage was $15.48; the mean wage from
all other sources was $15.58, a difference of less than one percent.
Therefore, we do not think it would be appropriate to give the BLS data
a smaller weight or remove BLS from the mix of data sources.
3. Blending Hospital and SNF Data for Occupational Labor Market Wage
Comment: Many commenters believed that use of hospital data in the
blend was (1) inappropriate because only SNF data should have been
used; (2) irrelevant because the rule applies exclusively to
nonhospital settings; (3) flawed because there are problems with the
sources of hospital data that HCFA used; and (4) incorrect because of
the large difference in wage levels between hospitals.
Response: Hospital wage levels by therapy type were used, in part,
because hospital therapists constitute a large part of the therapist
labor market. In addition, hospitals are a major source of therapists
hired by SNFs and rehabilitation therapy firms that contract with SNFs
to furnish therapy services. Also, the salary equivalency guidelines do
apply to contracted therapy services provided in the outpatient
departments of hospitals. Following traditional labor market theory for
professional services, we believe that there is an occupational labor
market for therapists, with compensating differentials for workers for
worker characteristics and job requirements. Had we used only SNF wage
data, the result would have reflected the relatively higher rates that
the rehabilitation therapy firm can afford to pay to bid therapists
away from other sectors that operate in a more financially constrained
contract environment. With respect to commenters' assertion that
hospital data are irrelevant in determining wages for therapists that
work primarily in nonhospital settings, our analyses supports our
position that an occupational labor market exists. We discussed these
issues in more detail in Section III.B.1 the Occupational Labor Market
of this final rule.
We believe that our ``best estimate'' approach incorporated the BLS
occupational/industry data in a reasonable way with other data from
less statistically reliable but more current sources. Each set of data
has equal weight in developing the ``best estimate'' for therapist
wages in hospitals and in SNFs. Each of the data sources we used is
discussed more fully in section III.A, Data Sources for Salary
Equivalency Guidelines of this final rule.
[[Page 5119]]
Comment: Many commenters challenged our blending of hospital and
SNF wage levels, but offered various blending recommendations in the
event that we use blending in the final rule. Many commenters agreed
that the employment weights as proposed for respiratory therapy wages
(99 percent hospital and 1 percent SNF) were correct. These same
commenters, however, offered a wide range of alternative employment
weights for use in blending hospital and SNF wages for physical and
occupational therapists and speech language pathologists. Some of these
commenters offered alternative employment weights which included HHAs
as well as SNFs and hospitals.
Response: We believe that an occupational labor market, with
compensating differentials, exists and that blending is required to
achieve equitable wage levels across settings as indicated in the
discussion above.
We also believe that the proposed blending method is reasonable.
When we blended the wages for each type of therapy, we used hospital
therapist employment and SNF therapist employment to develop the
relative shares. For SNFs, we used BLS' 1993 Occupational Employment
Statistics survey data, the latest and most complete employment data
for SNFs available from a government source. For hospitals, we used
BLS' 1995 Occupational Employment Statistics survey data, also the
latest and most complete employment data for hospitals available from a
government source. Some commenters preferred that employment data for
both settings be for the same year. We agree. Data on a same-year
basis, however, are not yet available. Based on industry discussions,
we believe that, as contract therapy services to SNFs have grown, there
has been a corresponding drop in the relative share of employed
therapists in SNFs. Without new data to substantiate this hypothesis,
we felt that the most appropriate option was to use the 1993
Occupational Employment Statistics survey for employment of therapists
in SNFs, by therapy type, as we did in the proposed rule (Occupational
Employment Statistics SNF data were collected for 1990 and 1993). This
approach may overstate SNF employment relative to hospitals in 1995,
and therefore the blended wage may also be slightly overstated.
One reason for the discrepancy between BLS Occupational Employment
Statistics and commenters' suggested shares of employment in SNFs and
hospitals is that commenters have included contract therapists in their
employment count for SNFs and hospitals. Using employment setting
(contract and employed) rather than employer to determine the share in
SNFs and hospitals increases the SNF share and decreases the hospital
share to use in blending. As rehabilitation therapy firms have hired
therapists away from SNFs and hospitals to work as contract therapists,
it is likely that the percentage of employed therapists in SNFs
relative to employed therapists in hospitals has decreased. As
indicated earlier, the SNF employment weights that we use may be too
high.
Comment: One commenter believed the blend is invalid because SIC
codes (806 for hospitals, 805 for SNFs) do not differentiate between
registered therapists, therapy assistants, and therapy aides. In
addition, a few commenters noted that audiologists are included in the
OES survey figures for speech language pathologists.
Response: The BLS Occupational Employment Statistics survey has a
separate occupational category for registered therapists. Therapy aides
and assistants are in a separate category and are excluded from
therapist employment numbers. Regarding the audiologist data included
with speech language pathology data in the Occupational Employment
Statistics survey, we believe that including audiologist data will not
significantly skew the employment shares in hospitals and SNFs for
speech language pathologists.
Comment: Another commenter proposed using total wages (wage bill
share of costs) rather than employment shares in blending hospital and
SNF wages.
Response: Using wage bill shares of costs for weights would double
the weight given to wages. The wage bill share is the number of hours
of service times the hourly wage for each therapy type in each setting.
The wage bill approach would, in effect, use SNF and hospital wage
levels twice in the calculation rather than once as is appropriate.
Comment: Several commenters were concerned that in the blending
methods we used we have not shown the difference in wage levels between
hospital and SNF, and that we have not proven the statistical validity
of combining these two wage values (hospital and SNF) to arrive at the
wage portion of the salary equivalency guidelines.
Response: In this final rule, we followed the same procedure that
we used in the 1983 rebasing of the salary equivalency guidelines for
physical and respiratory therapy, as published in the September 30,
1983 Federal Register (48 FR 44924). The industry requested a
statistical analysis of our blending therapist wages by relative
employment. We commissioned a complete statistical analysis of
therapist wage differentials under contract with Standard & Poor's DRI.
A full discussion of the results of this study appears in Section
III.B.1., the Occupational Labor Market of this final rule.
4. 75th Percentile
Comment: Commenters indicated that HCFA has significantly
underestimated the 75th percentile differential. Some want therapy-
specific differentials applied to each therapy type and many offered
alternatives to the 10 percent that HCFA used to approximate the 75th
percentile differential.
Response: HCFA estimated the 75th percentile differential using
several different data sources. We did not use the data sources that
did not have the 75th percentile available. The 75th percentile
differential in hospitals varied from 6.9 percent to 14.7 percent
depending on the data source and therapy type. The 75th percentile
differential in SNFs varied from 9.7 percent to 26.9 percent depending
on the data source and therapy type. None of the SNF sample surveys met
the sample design criterion of the Federal Government, resulting in
wider variation than would otherwise be the case.
When we ran regressions on the NASL data for hospitals and SNFs
with adjustments for region, ownership, and chain or individual
establishment, the 75th percentile for hospitals ranged from 9 to 11
percent, depending upon therapy type, while the 75th percentile for
SNFs ranged from 12 to 14 percent, again depending upon therapy type.
The 75th percentile differential varies so widely because if two
samples with the same means are compared, one meeting the BLS sample
design standard and the other below the BLS standard, the 75th
percentile differential will tend to be smaller for the BLS-type sample
than for the other sample. The alternatives offered by commenters come
from samples that do not meet BLS sample design standards.
We proposed a 10-percent differential to approximate the 75th
percentile for all therapy types because we believed that we had
selected a reasonable estimate for the range of average 75th percentile
differentials for the various therapy types. We have increased the
differential from 10 percent to 12 percent to allow for factors that we
may not have quantified previously. In choosing 12 percent for all
therapy types, we believe that we have selected
[[Page 5120]]
a reasonable estimate of the 75th percentile differential.
5. Calculations
Comment: Some commenters suggested that HCFA use BLS' 1991 Employee
Benefits for Medium and Large Private Establishments rather than its
1992 Small Establishments survey. As an adjunct to this comment, some
commenters indicated that the number of productive hours HCFA used was
too high, and that HCFA should make adjustments for breaks and lunches,
family leave, jury duty, funeral leave, and military leave.
Response: The average number of employees per establishment in SIC
Code 805, Nursing and Personal Care Facilities, calculated from the BLS
ES-202 survey in 1995, was fewer than 100. The average number of
employees per facility in the AHCA survey's sample data for 1995 was
fewer than 100, despite the fact that this data source is skewed toward
larger SNF chains. These figures support our decision that employee
benefits for small firms should be used in determining the number of
productive hours with which to adjust the hourly wage from hours paid
to hours worked.
The 1994 Small Private Firms survey reports even fewer paid leave
days (vacation, sick leave, and holidays) than did the 1992 survey. For
5-year employees, subtracting paid leave and 2 days for continuing
education from the standard work year (2,080 hours), still brings the
number of productive hours very near to our 1,808 productive hours
figure.
When data from the BLS Employment Cost Index Employer Costs for
Employee Compensation for March 1995 or March 1996 are used, only State
and Local Government Health Services and one of its subcategories,
State and Local Hospitals, have employees who work fewer productive
hours than the 1,808 hours used in HCFA's hours adjustment. All other
white collar, professional and technical occupations, as well as the
Health Services and Service Producing industries, work more productive
hours than we used in this calculation. (The calculation to reach
productive hours is 2,080 hours--272 hours of paid leave = 1,808
productive hours. Paid leave days numbered 33.7 (rounded to 34 days)
and multiplied by 8 hours per day to equal 272 hours of paid leave.
This adjustment equals approximately 15 percent of therapist hourly
wages.)
Data from both the recent BLS Employee Benefits Survey and the BLS
Employer Costs for Employee Compensation Survey support our choice of
number of productive hours worked per year. It is our policy to limit
paid leave to vacation, sick leave, and holidays.
Comment: Many commenters believed that HCFA's estimate of the
fringe benefit share of compensation is too low. Most commenters
mentioned about 30-31 percent of salary for a fringe benefits share,
while another mentioned 27 percent of salary for the standard fringe
benefit factor. Some commenters indicated that our proposed fringe
benefit share of 14 percent was too low; others asked that the ECI for
fringe benefits be used to determine fringe benefit share.
Response: For our fringe benefits calculation, HCFA used Health
Care Provider Cost Report Information System (HCRIS) prospective
payment system hospital cost reports to determine the share of
compensation other than leave that fringe benefits constitute. The
amount determined was 19.5 percent of total compensation excluding
leave, or about 24.2 percent of salary. Adding the fringe benefit of
paid leave (15 percent of salary associated with the productive hours
adjustment) to the fringe benefits determined from the cost reports
results in an overall fringe benefit rate of 39.2 percent of salary.
Thus, fringe benefits, including paid leave, constitute 28.2 percent of
total compensation, which is similar to the shares recommended by
commenters.
Comment: A few commenters stated that the allowed rental space of
250 square feet for office space was not sufficient to meet direct and
indirect space requirements. These commenters suggested that greater
allowance should be made for human resources management, program
support, compliance, and general business support.
Response: Contract therapists work in office space outside of the
rehabilitation therapy firm for which HCFA pays as part of Medicare
payments to providers. We believe that the 250 square feet for each
contract therapist allowed for rehabilitation therapy firm office space
is more than adequate to allow for human resources management, program
support, and general business support as well as space for individual
therapists.
Comment: Commenters offered what they termed ``technically
justifiable corrections'' that would have added between $5.20 and
$13.38 to the proposed guidelines, depending on the therapy type.
Others suggested salary equivalency guidelines somewhat closer to
HCFA's proposed guideline amounts. Several commenters stated that
speech language pathology guidelines should be as high or higher than
those for physical therapy. The commenters pointed out that speech
language pathologists have greater education requirements than do the
other types of therapists for whom salary equivalency guidelines were
proposed. In addition, commenters indicated that the services speech
language pathologists perform merit higher guidelines than we proposed.
Other commenters indicated that if salary equivalency guidelines do
not reflect accurately contract rates in the RTF industry, SNFs and
other providers will be unable to obtain medically necessary services
for Medicare beneficiaries. Some commenters believed that, in addition
to having difficulty in procuring therapy services, SNFs may find their
profit margins depressed to the point where some may close.
Commenters reported that some rehabilitation therapy firms pay
therapists sign-on bonuses and offer cruises as special incentives.
Response: We carefully analyzed all the industry ``technically
justifiable correct'' alternative levels for salary equivalency
guidelines and made modifications where we believed them to be
appropriate. We carefully reviewed recommended changes to employment
weights, fringe benefit shares, rental space, overhead shares, and 75th
percentile differentials. Modifications were made where they were
justified by the data, as stated in other sections of this rule. We
recognize that sign-on bonuses and other incentives such as cruises,
noted by some commenters, increase the operating costs of
rehabilitation therapy firms and result in higher wages than cost
conscious purchasers can afford to pay. This regulation requires HCFA
to set the salary equivalency guidelines at levels reasonably close to
costs that providers would incur for their own employees. These bonuses
and other incentives have contributed to distorting the therapist
market. We do not believe that Medicare should recognize these
extraordinary costs, which may not be considered to be related to
patient care, as part of its salary equivalency guidelines. Finally, we
believe that the therapist market, including rehabilitation therapy
firms, will adjust to these new guidelines without disrupting access to
care. Indeed, because of provisions in the Balanced Budget Act of 1997,
the new guidelines will not be the only change controlling provider
behavior.
[[Page 5121]]
C. One Schedule for Respiratory Therapists
We proposed to use one schedule of guidelines for respiratory
therapists, in contrast to the three schedules that we issued in the
September 30, 1983 notice. This decision was based on the fact that
HCFA does not differentiate in covering respiratory therapists by
different levels. Therefore, to make coverage conform with payment for
respiratory therapy services, we proposed one schedule for respiratory
therapists. Information from fiscal intermediaries and the American
Association for Respiratory Care indicates that industry practice is to
use only one schedule. For respiratory therapists in 1991, BLS showed
two wage classes and a summary wage level. The summary level was the
consistent category present for all metropolitan statistical areas
(MSAs) and encompassing all nonsupervisory levels of responsibility.
This final rule includes one schedule of guidelines for all therapy
types.
Comment: One commenter was concerned that the American Hospital
Association (AHA) and BLS data do not distinguish between wages for a
Certified Respiratory Therapy Technician and a Registered Respiratory
Therapist and fail to take the salary differentials of the two levels
into consideration. The commenter felt this was important for two
reasons. First, SNFs usually require the more experienced registered
respiratory therapists. Second, with a single rate, HCFA would
introduce an incentive for SNFs ``to contract for the less costly, yet
less experienced and less trained CRTT, rather than the more advanced
registered respiratory therapist to provide respiratory care services
to the medically acute SNF patient,'' raising questions about the
delivery of appropriate quality patient care.
Response: We used the AHA and BLS data for trending purposes only
and assumed that certified respiratory therapy technician and
registered respiratory therapist wages rose at the same rate. This rule
implements one schedule of guidelines for respiratory therapists,
regardless of whether services are rendered by a certified respiratory
therapy technician or a registered respiratory therapist. We developed
a single respiratory therapy wage rate that includes wages for
certified respiratory therapy technicians and registered respiratory
therapists, weighted for the various levels in respiratory therapy. The
single wage rate for the BLS data aged to 1995 was $15.48 per hour,
compared to $15.58 per hour for all data sources, a difference of less
than one percent.
Regarding the commenter's concern of introducing an incentive to
SNFs to use ``less experienced and less trained'' certified respiratory
therapy technicians rather than registered respiratory therapists, we
believe that as long as the therapist is qualified to provide
respiratory therapy services, then the provider will furnish quality
care. Therefore, both a certified respiratory therapy technician or a
registered respiratory therapist should be qualified to provide
respiratory therapy services.
D. Geographic Adjustment Factors
1. Use of Urban Portions of the Prospective Payment System Hospital
Area Wage Index for Geographic Adjustment
We proposed using the urban portion of the prospective payment
system hospital area wage index to adjust the guideline amounts for
local labor-related cost variations. We chose the urban portions of the
prospective payment system hospital area wage index because we felt
that SNFs compete in the same labor markets as hospitals, HHAs, and
other health care providers. There was also precedent for using the
hospital area wage index since two other long-term care Medicare
benefit programs, SNF and HHA care, use it to adjust for local labor
cost variation.
Comment: Several commenters stated that using the hospital area
wage index to adjust the salary equivalency guidelines for geographic
variation exaggerates the market variances both within and across
States. The commenters suggested that the geographic wage variation in
rehabilitation therapist labor markets is less than the geographic
variation in hospital industry labor markets. Therefore, they concluded
that using the hospital area wage index creates variations among
localities that are much too large. The commenters offered suggestions
that they believed would more adequately reflect the actual geographic
variations in therapist wages. One of these suggestions was the
Geographic Practice Cost Index (GPCI) used under the Resource-Based
Relative Value Scale (RBRVS) of the Physician Fee Schedule. Other
suggestions included aggregating data into State or regional rates
similar to those under the existing guidelines, or creating State
guideline amounts close to the national average with exceptions for
markets that have extreme variations. Commenters also suggested using
the hospital area wage index, but applying it to a smaller portion of
the labor-related costs to reduce distortions within and across states.
Another commenter suggested using the reclassified prospective payment
system hospital area wage index, instead of the pre-reclassified area
wage index. That would give SNFs and therapy suppliers the same
advantages that prospective payment system hospitals receive since SNFs
compete in the same labor markets as hospitals.
Response: As recommended by commenters, we are using the GPCI
contained in the Physician Fee Schedule (62 FR 59052, October 31, 1997)
instead of the hospital area wage index. We will use the Work, Practice
Expense, and Malpractice GPCIs, and apply them to therapist
compensation and overhead shares. Therapist compensation and overhead
shares come from the therapy-specific input price indexes as developed
by HCFA. There was no direct source of data on therapist malpractice
cost shares. To estimate a malpractice share, we analyzed the
malpractice shares from relevant rehabilitation therapy Current
Procedural Terminology (CPT) codes under the Physician Fee Schedule. We
determined that, on average, malpractice represents roughly 3.0 percent
of total expenses for these therapy CPT codes. We used 3.0 percent for
the malpractice share of the GPCI and subtracted 3.0 percentage points
from the overhead share of the GPCI to avoid accounting for malpractice
twice. The shares for these therapy-specific input price indexes are
presented in the table below.
----------------------------------------------------------------------------------------------------------------
Cost shares from therapy-specific input price
indexes
----------------------------------------------------
Therapist cost category GPCI Speech
Physical Occupational language Respiratory
therapist therapist pathologist therapist
----------------------------------------------------------------------------------------------------------------
Therapist Compensation............. Work.................. 0.74 0.72 0.71 0.67
Therapist Practice Expense......... Practice Expense...... 0.23 0.25 0.26 0.30
[[Page 5122]]
Therapist Malpractice.............. Malpractice........... 0.03 0.03 0.03 0.03
----------------------------------------------------
Total.......................... ...................... 1.00 1.00 1.00 1.00
----------------------------------------------------------------------------------------------------------------
The guideline amounts are calculated by the following equation:
Locality SEG amount
=
National SEG amount
x
[(Work GPCI x Therapy-Specific Compensation share)
+
(Practice Expense GPCI x Therapy-specific Overhead share)
+
(Malpractice GPCI x Therapy-specific Malpractice share)]
The GPCIs and guideline amounts for each therapy type for each GPCI
locality are in Table I under section V of this final rule.
We decided to use the GPCI for several reasons. The Balanced Budget
Act of 1997 mandates that many therapy services that are now reimbursed
based on the salary equivalency guidelines will be shifted to the
physician fee schedule, and, thus, therapist wages will be indexed by
the GPCI as early as July 1, 1998. We, therefore, saw that using the
GPCI was the direction for future therapy wage adjustments. We assessed
the appropriateness of using the GPCI and found that, of the available
indexes, the GPCI most accurately reflects the local labor costs of
therapists. Using the GPCI produces a less widespread geographic
distribution of guideline amounts. Also, many commenters asked us to
provide statewide rates as opposed to MSA rates provided in prior
salary equivalency guideline notices.
We decided to apply the GPCI to the therapist compensation share as
determined by the therapy-specific input price index. We then used the
practice expense GPCI to approximate the relative cost differences by
geographic area of practice expenses (clerical and managerial
compensation, office costs, and other costs) used to provide therapy
services. In addition, we use the malpractice expense GPCI to
approximate the relative cost differences by geographic area for
malpractice expenses incurred in providing therapy services. The
application of these GPCIs is analogous to the methods used under the
Physician Fee Schedule.
As mandated by section 4541 of the Balanced Budget Act of 1997,
many services presently covered under the therapy guidelines will be
paid under the physician fee schedule beginning in January 1999. Since
the physician fee schedule is adjusted for geographic variation by the
GPCI, both the current and future payment systems will reflect similar
geographic wage adjustments providing a smoother transition from the
salary equivalency guidelines to the physician fee schedule.
Comment: Commenters suggested that the hospital area wage index in
the proposed rule did not reflect the known geographic differences in
therapist wages in different settings, specifically, hospital-employed
therapists as compared to SNF-employed therapists. Many commenters
suggested that until HCFA can demonstrate that the geographic variation
in the wages in other settings are comparable, the use of the PPS
hospital area wage index should be abandoned.
Response: We responded to the variation in wage levels among
different settings in our responses to comments on the occupation labor
market for therapists under section III.B. of this final rule. We have
no data that indicate that the geographic adjustment needs to be done
by setting if the national baseline amounts by setting are
appropriately handled.
Comment: A commenter recommended that HCFA use nursing home
employed therapist wage data, and that the recent revision to the SNF
Medicare cost report would be useful in this regard. The commenter
suggested this as a long-term option and was willing to accept
modification of the hospital area wage index as a short-term solution
for reducing the influence of the geographic adjuster.
Response: We have decided to use the GPCI from the physician fee
schedule as the geographic adjuster. Currently, however, therapist wage
data are not available on the Medicare SNF cost reports. Also, the
Balanced Budget Act of 1997 provides for payment for outpatient
rehabilitation services on a fee schedule basis which uses the GPCI as
the wage index. Since we will only have salary equivalency guidelines
for a short period of time, we have not developed a separate wage index
for therapy services using nursing home employed therapist wage data.
2. Methodology for Determining Rural Rates Under Salary Equivalency
We proposed to calculate the guidelines in rural (non-urban) areas
in a given State as the weighted average of the prospective payment
system hospital wage index for MSAs within a State's boundaries. We
proposed this method because our analyses indicated that the therapy
market for rural areas tends to reflect the prevailing compensation
conditions of the surrounding urban areas in the region. By weighing
the urban areas in a state by the amount of hours associated with the
delivery of PPS hospital care, the rural rate would reflect the larger
weight given to MSAs with the most hospital hours. These urban areas
with most of the hospital hours also tend to have higher wage index
values.
Comment: Several commenters were concerned that rates in some rural
areas may be set too low. Some commenters indicated that this would
impede access to quality health care by Medicare beneficiaries because
it would be difficult to recruit and retain therapists in rural areas.
These commenters offered no recommendation on how to mitigate this
potential problem.
Response: Since we have decided, based on industry comments and
HCFA analyses, to use the work, practice expense, and malpractice GPCIs
from the physician fee schedule, we analyzed rural rates using the
GPCI. Unlike the hospital area wage index, the GPCI prov |
