You are on page 1of 6

A Systematic Review of the Cost-

Effectiveness of Noncardiac Transitional


Care Units*
Sean P. Keenan, MD; David Massel, MD; Kevin J. Inman, MSc; and
William J. Sibbald, MD, FCCP

Objective: To critically appraise and summarize the studies examining the cost-effectiveness of
noncardiac transitional care units (TCUs).
Data sources: We conducted a computerized literature search using MEDLINE, and Current
Contents from January 1, 1986 to December 31, 1995 and HealthSTAR from January 1, 1989 to
December 31, 1995 with the key words intermediate care unit, respiratory care unit, and
step-down unit. Bibliographies of all selected articles and review articles were examined. Personal
files were also reviewed.
Study selection: (1) Population: patients in a noncardiac TCU of an acute-care institution; (2)
intervention: addition of a noncardiac TCU to the institution; and (3) outcomes: patient
outcome.survival and associated costs.
Data extraction: The necessary data were abstracted and study validity was evaluated by two
independent reviewers using a modification of previously published criteria.
Data synthesis: The studies were summarized qualitatively; upon inspection, they were too
heterogeneous to allow quantitative analysis. While the studies all claimed that their TCUs were
cost-effective, the economic evaluation designs were flawed to such an extent that the validity of
the conclusions is suspect.
Conclusions: To date, the evidence in the literature is insufficient to determine under which
circumstances, if any, TCUs are a cost-effective alternative technology to the traditional
institution with only ICU and general ward beds. (CHEST 1998; 113:172-77)
Key words: cost-effective; intensive care; intermediate care unit; step-down unit; critical care
Abbreviation: TCU=transitional care unit

HP he cost associated with caring for patients in the future, necessitating an emphasis on efficient triag-
¦*¦critical care units of the United States has been ing of patients to areas of care appropriate to patient
estimated to be 1 to 1.5% of the gross national need.
product.1"3 Canada has seen a similar, rapid growth In the past, a significant proportion of patients
in the cost of critical care.4 Resources dedicated to were cared for in ICUs for monitoring purposes
health care will not increase appreciably in the alone15"7 or for single-system failure (prolonged
ventilation in hemodynamically stable patients). The
proportion of these patients admitted to ICUs varied
among hospitals and countries depending on the
*From the Richard Ivey Critical Care Trauma Center (Drs. resources available and the philosophies of the
Keenan and Sibbald and Mr. Inman) and the Department of
Cardiology (Dr. Massel), London Health Sciences Centre, Vic¬ health-care workers involved.89 As a result of these
toria Campus, University of Western Ontario, London, Ontario, observations, units were developed to provide vary¬
Canada.
This study was supported by the Departments of Critical Care, ing levels of noninvasive monitoring with or without
Cardiology, General Surgery, and Nursing, London Health Sci¬ the capability to ventilate patients. For this review,
ences Centre, Victoria Campus, University of Western Ontario, we have chosen to call these units transitional care
London, Ontario, Canada. Dr. Keenan is supported by a Cana¬
dian Lung Association/Medical Research Council of Canada
units (TCUs) (Table 1). TCUs are a technology in
Fellowship. received November themselves and, like all technologies, should be
Manuscript 26, 1996; revision accepted June critically evaluated prior to widespread adoption.
10, 1997. Patients in TCUs require a lower nurse: patient ratio
Reprint requests: Sean P. Keenan, MD, Department of Critical and may require fewer investigations when com¬
Care Medicine, London Health Sciences Centre, Victoria Cam¬
pus, 375 South St, London, ON, Canada, N6A 4G5 pared to patients in ICUs. While a number of
172 Clinical Investigations in Critical Care
Table 1.Potential Levels of Care Available
Level of Care Nurse: Patient Ratio Mechanical Ventilation Noninvasive Monitoring Invasive Monitoring
General ward Low(<l:10) Not available Intermittent oximetry Not available
TCU Intermediate (1:3 to 1:6) Available Not available*
ICU High (1:1, 1:2) Available Available Available
*Some TCUs may provide limited invasive monitoring.

authors claim that TCUs are, de facto, cost-effec¬ ties) is imperative. Finally, the conclusions drawn in the discus¬
sion must reflect the results presented. Conclusions based on
tive,10-15 it may be presumptuous to accept these invalid study design cannot themselves be respected as valid. A
claims without a rigorous review of the evidence. number of important articles have been written that allow greater
The objective of this systematic review was to explanation of these concepts than are possible here.1621 The
determine whether noncardiac TCUs are cost-effec¬ maximum score possible was 18. In addition, an overall rating of
tive. the study quality was independently made by the two reviewers
using a five-point Likert scale (very poor, fair, good, very
poor,
good) after completing the validity criteria.
Materials and Methods Analysis
Interobserver variability in assessment of study validity and
Search Strategy and Study Selection Criteria overall quality was described qualitatively, and differences in
We performed a computerized literature search using MED¬ opinion were decided by consensus. No attempt was made to
LINE and Current Contents from January 1, 1986 to December pool the results of the studies quantitatively due to the hetero¬
31,1995 and HealthSTAR from January 1, 1989 to December 31, geneity of study design and TCU populations.
1995 with the following key words: intermediate care unit,
respiratory care unit, and step-down unit. Our search was Results
restricted from 1986 onwards as noncardiac TCUs were only
introduced in the early to middle 1980s and the literature lagged
about 5 years behind this. We also reviewed the citations from all Summary of Selected Studies
articles retrieved to identify other articles of interest. Our Only three articles fulfilled the selection crite¬
selection criteria included the following: (1) population: patients ria.1315 Other studies on implementation of a TCU
in a noncardiac TCU of an acute-care institution; (2) intervention:
addition of noncardiac TCU to the institution; and (3)
a
patient outcome.survival and associated costs.
out¬ Table 2.Modified Checklist for Assessing Economic
comes:
Evaluations*
Assessment of Study Validity Checklist
The validity of studies that fulfilled our selection criteria was 1. Was a well-defined question posed in an answerable form?
independently appraised (S.P.K., K.J.I.) using modified criteria (No 0; yes 2)
for economic evaluation (Drummond and colleagues16.see Ta¬ 2. Was a comprehensive description of competing alternatives
ble 2). A properly posed question should identify the different given?
alternatives being compared and the relevant viewpoint of the (Neither alternative 0; one alternative 1; both alternatives 2)
economic evaluation. 3. Was there evidence that the program's effectiveness had been
The viewpoint selected can be that of the hospital, insurance established?
company, society, or others. A complete description of both
(No 0; yes 2)
alternatives should be given (a hospital with or without a TCU). 4. Were all the important and relevant costs and consequences for
The effectiveness of the two alternatives must be established each alternative identified?
either from previously conducted, valid studies or within the (Neither costs nor consequences 0; one of two or most of both
economic evaluation itself. All relevant costs and consequences 1; both 2)
must be included (usually dictated by the viewpoint of the 5. Were costs and consequences measured accurately in
economic evaluation). These costs and consequences must be appropriate physical units?
ascribed in physical units initially to allow generalizability to other (No 0; partially 1; all 2)
institutions and then valued in a credible and clear fashion. In 6. Were costs and consequences valued credibly?
most cases, adjustment for differential timing is required (costs (Neither 0; one of two 1; both 2)
7. Was an incremental analysis of costs and consequences of
may be incurred at different times for alternative therapies); alternatives performed?
however, in economic evaluations assessing short-term outcomes, (No 0; yes 2)
this is not necessary (we therefore did not include these criteria
8. Wasa sensitivity analysis performed?
in our review). Analysis should be conducted in an incremental
fashion, describing the increase in effectiveness relative to the (None 0; some but insufficient 1; sufficient number 2)
increase in cost (for example, the increased cost required to save 9. Did the presentation and discussion of study results include all
one further life). As a number of assumptions or estimates are
issues of concern to users?
often required, the use of sensitivity analysis (reanalyzing the data (None 0; partial 1; most 2)
after varying the assumptions to the extremes of their possibili¬ * Modified from Drummond et al.16

CHEST / 113 / 1 / JANUARY, 1998 173


were found but were excluded because they de¬ scored slightly higher on the validity criteria. While
scribed TCUs in specialized nonacute-care hospi¬ greater taken to substantiate the equiva¬
care was
tals2225 or they did not perform an economic lence of effectiveness of the alternative technolo¬
evaluation.10"12'22"2426"31 All three studies involved gies,343740 the patient population focused on were
respiratory TCUs. Two were conducted by the same those with respiratory failure who were undergoing
group of authors and all three originated in the weaning in the TCU, representing only 33 of their
United States (Table 3). 109 patients. They also reported more extensive cost
data than previously; however, they did not break
Validity Assessment of TCU Economic Evaluations down how they arrived at the cost for beds in the
Two of three studies1314 scored poorly using the ICU, TCU, and ward in a way that would allow
modified criteria for validity assessment of economic others to generalize their results. While sensitivity
evaluation (Table 4). Neither study began with a analysis was used, no reason was given for why they
well-defined question or established the relative used 2 SDs as their interval. The baseline assump¬
tions were listed but the acceptable ranges chosen
effectiveness of the two alternative technologies for sensitivity analyses should be based on known
(rather they referred to historical controls).32-34 The variability of each assumption. As a result of these
relevant viewpoint was never specified but appeared variations from the proposed validity criteria, the
to be somewhere between that of the third-party
conclusions reached by the authors (that their TCU
payer and the hospital. While their respective TCUs was cost-effective) must be questioned.
were well described, alternative setting (an insti¬
the The two reviewers agreed completely on six of the
tution without a TCU) was not. Consequences were
nine criteria. In criteria 4, 5, and 6, the reviewers
properly identified but only for the TCU alternative. disagreed on one of the three studies. The cause of
Costs were derived from cost-to-eharge ratios in the the disagreement was the use of cost-to-charge ratios
study by Elpern and colleagues,13 a method that is in the study in question. One reviewer believed that
difficult to interpret.35-36 In the first study by Krieger this should negate any points in these three criteria,
and associates,14 there appeared to be a better while the other reviewer was more liberal in his
approach to identifying and measuring some of the assessment. Together they agreed to give one point
costs; however, the cost analysis was restricted to for each criterion. The two reviewers were in total
ventilated Medicare patients, a minority of the total
patients admitted to the hospital.14 Neither study agreement on overall quality, rating all studies as
poor.
performed analoneincremental analysis (presented as cost
with associated outcome) or sen¬
comparison
sitivity analysis. As a result of the above flaws, the Discussion
conclusions that TCUs are cost-effective are of ques¬
tionable validity. This systematic review of the literature suggests
The second study by Krieger and coworkers15 that there is currently insufficient evidence to sup-

Table 3.Summary of Studies of Noninvasive TCUs*


Type of
Intermediate Nurse: Patient Estimated
Article, yr Study Design Setting Care Unit Beds, No. Ratio Patients, No. Time Frame Savings
(1) Elpern Observational Tertiary care Respiratory 11 (8 ventilators) 1:3 to 1:5 136 Fiscal yr, 1988 $20,000
et al,13 1991 cohort study hospital per patient
(2) Krieger Observational Tertiary care Respiratory 4 (all ventilators) 1:4 to 1:6 54 Oct 1986 to $l,975±$l,061f
et al,14 1988 cohort studv hospital 12 vent March 1987 per patient
15 airway
27 monitor
11 patients
analyzed
(3) Krieger Observational Tertiary care Respiratory 4 (all ventilators) 1:4 to 1:6 109 May 1987 to $1,551
et al,15 1990 cohort study hospital 33 vent May 1988 per patient
11 trach
65 monitor
94 analyzed
*vent=ventilated patients; airway=patients requiring airway protection alone; monitor=patients who required monitoring only;
trach=tracheostomy patients.
f±SD.

174 Clinical Investigations in Critical Care


Table 4.Validity Assessment of Economic Evaluations
Validity Criteria* Elpern et al,13 1991 Krieger et al,14 1988 Krieger et al,15 1990
1. Well-defined question 0 0 0
2. Description of competing alternatives 1 1 1
3. Program effectiveness established 0 0 0
4. All costs/consequences identified 1 1 1
5. Costs/consequences measured in appropriate physical units 1 1 1
6. Costs/consequences valued credibly 1 1 2
7. Incremental analysis performed 0 0 0
8. Sensitivity analysis performed 0 0 1
9. Complete discussion 1 1 1
Total 5/18 5/18 7/18
Overall quality assessment Poor Poor Poor
* Possible score of
0, 1, or 2.

port the viewpoint that the addition of a TCU to an accepted asdetermine


valid. The studies reviewed herein at¬
institution with ICU and general ward beds is cost- tempted toTCUs the cost-effectiveness of their
short of doing so because of
effective. We do not claim
that the addition of a TCU respective but fell
is not a cost-effective measure. Rather, we have a number of design flaws.
highlighted the design flaws of the economic evalu¬ The criteria used in this review to critically ap¬
ations available that compromise the validity of these praise cost-effective analyses were derived from
studies and any conclusions drawn. We believe that those described by Drummond and colleagues.16
while a TCU may be cost-effective, the design and Similar criteria were used by Udvarhelyi and cowork-
implementation of a TCU must be carefully con¬ ers17 in a review of published cost-effectiveness and
ducted and continuously evaluated to ensure that cost-benefit analyses. The Task Force on Principles
resources are utilized as planned. for Economic Analysis of Health Care Technology
The strengths of a systematic review include a released a report recently that presented similar
comprehensive search strategy, objective study se¬ criteria for critically appraising economic evaluations
lection criteria, and validity assessment of the pri¬ but provide additional comment on how to conduct
mary studies. We summarized these studies qualita¬ economic evaluations, including recommendations
tively as their study designs did not lend themselves on how sponsors and researchers should interact.18
to quantitative pooling. In this systematic review, we In 1996, the Panel on Cost-Effectiveness in Health
conducted an extensive search using computerized and Medicine published a series of three articles19-21
databases, bibliographies of selected and review designed to guide the conduct of economic evalua¬
tions. The authors emphasize the importance of the
articles, and personal files. We limited our search to
the time period of relevance as TCUs were not results of economic evaluations being able to inform
evaluated prior to this era. We restricted ourselves to resource allocation decisions. They suggest the use
evaluation of the English-language literature as we of a standard "reference case" format, using the
were most interested in the economic impact of perspective of society and valuing outcomes in terms
TCUs in North America. Unpublished studies were of health-related quality of life, for all economic
not included in our systematic review. While we evaluations. They suggest that additional specific
agree that it is important to consider unpublished baseline cases, using other perspectives and means
literature in any systematic review,4142 after witness¬ of expressing outcome, may be added to the eco¬
ing the poor rating on study validity of the published nomic evaluation to answer specific study questions,
studies, we do not believe that unpublished studies at the investigator's discretion.
would have changed our conclusions. It is intuitively appealing to assume that the
Economic evaluations are becoming increasingly introduction of a TCU into a hospital would enhance
common in the medical literature as physicians and overall efficiency of patient care by providing a more
health administrators struggle to balance patient cost-effective approach for patients requiring only
benefit with limited resources. It is imperative that intensive monitoring with or without mechanical
we appropriately evaluate each new technology, not ventilation. The studies presented in this review
only for its effectiveness, but also its associated cost appear to begin with the assumption that TCUs are,
prior to accepting its adoption. Economic evalua¬ by definition, cost-effective. This may explain why
tions are as open to bias as clinical trials and must many investigators did not attempt any form of
adhere to specific criteria for their results to be economic evaluation. 10-12,22-24,26-31 Tne few studies

CHEST / 113 / 1 / JANUARY, 1998 175


that did report an economic evaluation were of poor related to inappropriate admissions. The interpreta¬
quality. Despite this objection, how could TCUs not
be cost-effective and hence enhance efficiency?
tion of potential cost savings for the institution must
also be closely scrutinized. The claim that costs per
We will address the issue of effectiveness first. patient will be less with introduction of a TCU must
There are no studies in the literature to date (to our be coupled with the acknowledgment that overall
knowledge) that have demonstrated a clear differ¬
ence in effectiveness between a hospital setting with
costs to the institution will increase unless ICU beds
are closed at the same time. Therefore, from the
an ICU and ward beds alone compared to a similar perspective of the hospital, the total budget could
hospital setting that also has a TCU. While in a study possibly increase rather than decrease. Krieger and
by Franklin and coworkers26 an improved survival coworkers15 suggested closing ICU beds in their
was noted with introduction of a medical TCU, the
before-after design and lack of control for severity of
study, but it is not clear whether ICU beds were
illness makes it impossible to draw direct inferences.
actually closed.
In addition to careful planning of a TCU prior to
Indeed, the studies reviewed herein tried to empha¬ its opening, continuous monitoring of the appropri¬
size that survival was not negatively influenced by ateness of patient admissions and discharge must be
the introduction of a TCU. We must conclude, conducted. A trend may develop toward admitting
therefore, that currently there is no evidence that patients to a TCU rather than a general ward bed
allows one to determine whether the introduction of who are less sick than proposed in initial planning.
a TCU will improve, worsen, or not affect patient admission creep. Alternatively patients may be dis¬
outcomes.
If we make the assumption that TCUs do not
charged from the ICU to TCUs who are well enough
for a regular ward bed.ICU discharge creep. These
affect outcome, then to prove that a TCU is cost- drifts in admissions policies in addition to another
effective we must demonstrate a reduction in cost. In
order for TCUs to be cost-effective, they must be
possible trend.keeping patients in the TCU longer
than necessary.would all lead to unanticipated
well planned prior to implementation. The approach additional costs. We have summarized in Table 5
used by Hilton and coworkers,31 performing an some of the important factors to consider when
initial pilot study to determine the size of TCU assessing or attempting to control costs of a TCU.
required, should be adopted by all those planning to In conclusion, we were unable to find evidence
open a TCU. Strict guidelines for admissions to and from our systematic review of the literature that
discharge from a TCU also need to be developed and
adhered to. If units are not properly planned prior to
supports TCUs being cost-effective. We believe that
the cost-effectiveness of any new technology should
opening, they may be too small to provide the service be rigorously assessed prior to adoption. This was not
that is required, or too large, creating the opportu¬ performed for TCUs. There remains a need to
nity for admission of inappropriately low-risk pa¬ conduct economic evaluations using the appropriate
tients. In the latter case, the costs per patient care for viewpoint to determine the conditions under which
these low-risk patients would potentially be in¬ TCUs may be cost-effective. Ideally, these studies
creased by treating them in a TCU rather than a would be conducted as randomized, controlled trials
general ward. For example, in the studies by Elpern comparing the costs and outcomes of patients
and associates13 and the first study by Krieger and treated in various types of TCUs to similar patients
coworkers,14 the focus ofthe cost analysis was exclu¬ treated in the appropriate alternative setting, insti¬
sively on ventilated patients (and in the latter study14 tutions with ICU and general ward beds alone.
only Medicare patients were considered). However,
a large proportion of patients admitted to their TCUs
However, alternative study designs, such as a before-
were not represented by their economic evaluation
(61% and 80%, respectively). In the second study by Table 5.Factors Influencing Cost of TCU Care
Krieger and colleagues,15 private patients, compris¬
ing 14% of admissions, were excluded from the Factors
analysis. Of the remaining patients, the cost savings
were considerably less for the nonventilated patients
1. Nurse: patient ratio
2. Degree of support available (ventilation, etc)
when compared to those patients who were venti¬ 3. Use of protocols in patient care (weaning protocol may lead to
lated. One would suspect that as the proportion of early discharge)
nonventilated patients in a TCU of this type in¬
4. No. of beds relative to need (increased bed availability may lead
to lower acuity admissions)
creases, the overall cost savings would decrease or 5. Flexibility of admission policy
disappear. Careful planning prior to opening a TCU 6. Flexibility of discharge policy
Periodic review of admission/discharge process
would allow better estimation of the institution's 7.
needs and hopefully avoid future unnecessary costs 8. Periodic review of process of care.reassess/develop protocols

176 Clinical Investigations in Critical Care


after design or cluster randomization, may be more of the panel on cost-effectiveness in health and medicine.
pragmatic. The results of these economic evaluations JAMA 1996; 276:1253-58
21 Siegel JE, Weinstein MC, Russell LB, al. Recommenda¬
would allow us to answer the question: are noncar¬ et

diac TCUs truly cost-effective? tions for reporting cost-effectiveness analyses. JAMA 1996;
276:1339-41
22 Indihar FJ, Forsberg DP. Experience with a prolonged
respiratory care unit. Chest 1982; 81:189-92
References 23 Indihar FJ, Walker NE. Experience with a prolonged respi¬
1 Henning RJ, McClish D, Daly B, et al. Clinical characteristics ratory care unit.revisited. Chest 1984; 86:616-20
and resource utilization of ICU patients: implications for 24 Indihar FJ. A 10-year report of patients in a prolonged
organization intensive care. Crit Care Med 1987; 15:264-69
of respiratory care unit. Minn Med 1991; 74:23-27
2 Knaus WA, Thibault GE. Intensive care units today. In: 25 Scheinhorn DJ, Artinian BM, Catlin JL. Weaning from
McNeil BJ, Cravalho EG, eds. Critical issues in medical prolonged mechanical ventilation: the experience at a re¬
technology. Boston: Auburn House, 1982; 193-215 gional weaning center. Chest 1994; 105:534-39
3 Critical care in the United States. Anaheim, Calif: Society of 26 Franklin CM, Rackow EC, Mamdani B, et al. Decreases in
Critical Care Medicine, 1992 mortality on a urban medical service by facilitating
large
4 Jacobs P, Noseworthy TW. National estimates of intensive access to critical care: an alternative to rationing. Arch Intern
care utilization and costs: Canada and the United States. Crit Med 1988; 148:1403-05
Care Med 1990; 18:1282-86 27 Vitacca M, Clini E, Scalvini S, et al. Cardiopulmonary
5 Thibault GE, Mulley Ag, Barnett GO, et al. Medical intensive intermediate intensive unit: time course of 2 years activity.
care: indications, interventions, and outcomes. N Engl J Med Monaldi Arch Chest Dis 1993; 48:296-300
1980; 302:938-42 28 Byrick RJ, Power JD, Yeas JO, et al. Impact of an interme¬
6 Knaus WA, Wagner DP, Draper EA, et al. The range of diate care area on ICU utilization after cardiac surgery. Crit
intensive care services today. JAMA 1981; 246:2711-16 Care Med 1986; 14:869-72
7 Wagner DP, Knaus WA, Draper EA. Identification of low- 29 Byrick RJ, Mazer CD, Caskennette GM. Closure of an
risk monitor admissions to medical-surgical ICUs. Chest intermediate care unit: impact on critical care utilization.
1987; 92:423-28 Chest 1993; 104:876-81
8 Knaus WA, Le Gall JR, Wagner DP, et al. A comparison of
intensive care in the USA and France. Lancet 1982; 2:642-46 30 Maclntyre ML, Savoy-Bird S. Creating a step-down unit
9 Zimmerman JE, Knaus WA, Judson JA, et al. Patient selection opens a door to the future. Dimens Health Serv 1989;
for intensive care: a comparison of New Zealand and United 66:25-27
States hospitals. Crit Care Med 1988; 16:318-26 31 Hilton G, Madayag M, Shagoury C. Development of a
10 Bone RC, Balk RA. Noninvasive respiratory care unit: a surgical/trauma intermediate care unit. Clin Nurse Specialist
cost-effective solution for the future. Chest 1988; 93:390-94 1993; 7:274-79
11 Silver MR, Elpern EH, Balk RA, et al. Noninvasive respira¬ 32 United States Congress Office of Technology Assessment.
tory care units as an innovative alternative to ICUs: how Life sustaining technologies and the elderly. OTA-BA-306.
hospitals cut costs without compromising quality of care. J Washington, DC: US Government Printing Office, July 1987;
Crit Illness 1990; 5:1279-87 225
12 Patterson PA, Elpern EH, Silver MR. Advances in patient 33 Wagner DP. Economics of mechanical ventilation. Am Rev
care management: the NRCU. Crit Care Nurse 1991; 11: Respir Dis 1989; 140:S14-18
42-45 34 Douglass PS, Rosen RL, Butler PW, et al. DRG payment for
13 Elpern EH, Silver MR, Rosen RL, et al. The noninvasive long-term ventilator patients: implications and recommenda¬
respiratory care unit: patterns of use and financial implica¬ tions. Chest 1987; 91:413-17
tions. Chest 1991; 99:205-08 35 Finkler SA. The distinction between cost and charges. Ann
14 Krieger BP, Ershowsky P, Spivack D, et al. Initial experience Intern Med 1982; 96:102-09
with a central respiratory monitoring unit as a cost-saving 36 Gyldmark M. A review of cost studies of intensive care units:
alternative to the intensive care unit for Medicare patients
who require long-term ventilator support. Chest 1988; 93:
problems with the cost concept. Crit Care Med 1995; 23:
964-72
395-97 37 Spicher JE, White DP. Outcome and function following
15 Krieger BP, Ershowsky P, Spivack D. One year's experience prolonged mechanical ventilation. Arch Intern Med 1987;
with a noninvasively monitored intermediate care unit for 147:421-25
pulmonary patients. JAMA 1990; 264:1143-46 38 Morganroth ML, Morganroth JL, Nett LM, et al. Criteria for
16 Drummond MF, Stoddart GL, Torrance GW. Methods for weaning from prolonged mechanical ventilation. Arch Intern
the economic evaluation of health care programmes. Oxford, Med 1984; 144:1012-16
UK: Oxford University Press, 1987; 18-38 39 Davis H, Lefrak SS, Miller D, et al. Prolonged mechanically
17 Udvarhelyi IS, Colditz GA, Rai A, et al. Cost-effectiveness assisted ventilation: an analysis of outcome and charges.
and cost-benefit analyses in the medical literature: are the JAMA 1980; 243:43-45
methods being used correctly? Ann Intern Med 1992; 116: 40 Gracey DR, Gillespie D, Nobrega F, et al. Financial impli¬
238-44 cations of prolonged ventilator care of Medicare patients
18 Task Force on Principles for Economic Analysis of Health under the prospective payment system: a multicenter study.
Care Technology. Economic analysis of health care technol¬ Chest 1987; 91:424-27
ogy: a report on principles. Ann Intern Med 1995; 123:61-70 41 Dickersin K. The existence of publication bias and risk factors
19 Russell LB, Gold MR, Siegel JE, et al. The role of cost- for its occurrence. JAMA 1990; 263.1385-89
effectiveness analysis in health and medicine. JAMA 1996; 42 Cook DJ, Guyatt GH, Ryan G, et al. Should unpublished data
276:1172-77 be included in meta-analyses? Current convictions and con¬
20 Weinstein MC, Siegel JE, Gold MR, et al. Recommendations troversies. JAMA 1993; 269:2749-53

CHEST / 113 / 1 / JANUARY, 1998 177

You might also like