TITLE
Optimization of the diagnostic work-up in patients with suspected
obstructive lung disease.
Authors:
Frank J Visser1 Milena JMM van der Vegt1, Gert Jan van der Wilt2, Julius P
Janssen1
Canisius Wilhelmina Hospital, Nijmegen, The Netherlands
Departments of Pulmonary Medicine1 Canisius-Wilhelmina Hospital,
Nijmegen; The Netherlands.
Department of Epidemiology, Biostatistics and Health Technology
Assessment2, Radboud University Hospital, Nijmegen, the Netherlands.
Corresponding Author:
F.J. Visser, MD
Canisius-Wilhelmina Hospital
Weg door Jonkerbos 100
6525 SZ Nijmegen, The Netherlands
Tel: +31-24-3658755 Fax: +31-24-3658978 e-mail: f.visser@cwz.nl
Requests for reprints: FJ Visser
Funding source: none
No sources of support
Short title: diagnostic work-up in patients with obstructive lung disease
1
ABSTRACT
Background: Pulmonary function testing is a key procedure in the workup of patients suspected of asthma and COPD. Major contributions to
costs include visits and pulmonary function tests (PFT).
Objective:
To assess whether diagnostic test protocol contributed to optimising the
work up in these patients Methods: A prospective single blind,
randomised controlled study was performed. In the control group (CG), all
PFT as ordered by the lung specialist were carried out. In the experimental
group (EG), PFT were performed according to protocol. The primary end
point was the costs incurred to reach a final diagnosis .
Results: Total number of patients included was 179: 86 in the CG and 93
in EG. The mean number of tests to diagnosis was 3.8 in CG versus 2.9 in
EP (P<0.001) The mean number of redundant PFT before diagnosis was
1.2 in the CG versus 0.08 in the EG (P<0.001) .The number of patients
requiring an additional outpatient visit to complete diagnosis was higher
in the CG (P=0.02). The mean costs of workup per diagnosis were €227 in
the CG versus €181 in the EG (P<0.001).
Conclusions In this group of patients with suspected obstructive lung
disease, pulmonary function testing following a PFT protocol is more costeffective than pulmonary function testing at the discretion of lung
physicians.
1
Key words: COPD, asthma, cost-effectiveness, pulmonary function tests,
diagnostics., redundant tests
Introduction
Diagnosing asthma and COPD is an important part of the everyday
practice of pulmonary physicians. Pulmonary function tests (PFT) play a
key role in the workup of obstructive pulmonary diseases (1-3)
No exact figures exist for the annual costs associated with current
diagnostic processes, but they must be substantial.
Finding the optimal diagnostic work-up in patients with obstructive lung
disease is a challenge. The physician who largely orders PFT as a logical
next step in the work-up of astma and COPD runs the risk of unnecessary
testing; the physician who is more sparingly in test-ordering, on the other
hand, runs the risk of unnecessary outpatient visits. In view of the high
incidence of patients with obstructive lung diseases, it is important to find
the optimal diagnostic work-up in these patients. To this end, we have
developed a diagnostic protocol (Figure 1) , to be jointly used by
physicians and pulmonary function assistants. In our group some
physicians did already used this diagnostic protocol and some ordered
function tests liberally . Until the outset of this study there is no evidence
that protocol-driven lung function test ordering is more efficient. Our
hypothesis was that protocol driven test ordering will be more efficient
than liberally test ordering.
2
The aim of this study was to assess whether protocol-driven test ordering
reduced the number of redundant pulmonary function tests, decreased
outpatient visits and increased cost effectiveness. the work-up in these
patients, as compared to physician-driven test-ordering.
METHODS
A prospective randomised single blind trial was conducted at our
outpatient unit.
When we submitted the study for IRB approval, the chairman of the IRB
stated that this study was only a formal stratification of the current
practice, and would not lead to an extra
burden of examinations to patients, or expose patients to an increased
risk. Therefore, formal informed consent was not considered necessary.
And therefore the IRB approval was waived.
Study design
Our trial included during 10 months consecutive adult patients suspected
at the end of the first outpatient visit of having an obstructive pulmonary
disease; none of the patients had recently (in the preceding three years)
been diagnosed with asthma or COPD by a pulmonary physician. Patients
were referred mainly by general practitioners two patients referred
themselves to our outdoor department and some were referred by
cardiologists. We excluded patients not able to complete pulmonary
function tests adequately, patients who were referred to a pulmonary
physician because of an abnormal X-ray, pre-operative consultations, and
3
patients who have had an infection of the upper or lower airways, or a
possible exacerbation of obstructive airway disease in the past two
months.
In the first visit the physician takes a medical history, physical
examination, makes a differential diagnosis and ,orders laboratory testing,
chest X ray and so on. Only patients are included in the study for whom
the most likely diagnosis was “astma or COPD”.Physicians ordered
diagnostic tests as they deemed appropriate and added the reason for
pulmonary function testing e.g. suspected obstructive lung disease. At
the end of the first outpatient visit, nurses randomised the patients to the
control group (work-up at the discretion of the physician) or to the
experimental group (work-up in accordance with the protocol) by drawing
an opaque envelope.
PFT Lab assistants were notified of the outcome of randomisation, to
allow them to perform investigations as ordered or per protocol .
The pulmonary lungfunction protocol is visualised in Figure 1. Added to
this figure: when the FVC < lower limits of normal a TLC measurement
was done in order to exclude restrictive disease. When the patients
smoked > 10 pack years than a measurement of the diffusion capacity
was done.
A second physician independently examined the results for each patient
and classified the patients as follows:
4
Completely reversible (CR) : -1- Airway obstruction completely reversible
(FEV1 reversible ≥ 9% predicted value) to normal range after beta-2
agonists and/or anticholinergic, patients on reversibility testing were
tested for both bronchodilators, or -2- Airway obstruction completely
reversible after 14 days of 30 mg prednisone a day or -3- a normal PFT
but decreased PC20 histamine threshold.
This PFT group gives support to the diagnosis Asthma.
Non reversible obstructive(NRO):
Reduced FEV1 and FEV1/FVC .
Irreversible after beta-2 agonists and/or anti-cholinergics (FEV1 increase
of <9% of predicted value) and no return to normal range after 10 days of
30 mg of prednisone a day.
This PFT group gives support to the diagnosis COPD
Partly Reversible Obstructive (PRO): Reversibility present yet airflow
limitation persists. FEV1 increases ≥ 9% of predicted normal value but
does not return to normal range after bronchodilators.
This PFT group gives support to Asthma with persistent airflow limitation
or COPD with partial reversibility .
Normal PFT group No airflow limitation and normal PD20 histamine
threshold.
This group gives no support for an obstructive lung disease so other
diagnosis must be considered.
The second physician looked if the appropriate tests were done according
to the diagnostic flow and decided on the pulmonary pulmonary function
5
classification. He also looked to the decision about the final diagnosis of
the PFT referring physician
The second physician calibrated his findings with the findings of the PFT
ordering physician only if there were conflicting findings. In all these cases
we had a consensus on the final diagnosis.
A final diagnosis of COPD or asthma was made by the first physician on
the basis of medical history (smoking behaviour, allergy, asthma in family
, and or in childhood), PFT and clinical investigation like eosinophils count
, RAST test, etc..
In the protocol, criteria for obstruction were FEV1 < lower limit of normal
value and FEV1/FVC < lower limit of normal value according to Quanjer et
al (4). Airway hyper responsiveness was defined as PC20 histamine < 4
mg/ml. (5).Reversibility was defined as ≥9% improvement of the FEV1
from predicted normal value.(6-8) Steroid test consisted of 30 mg
prednisone for 10 days with reversal to normal range for the FEV1 as
advised by the Dutch Committee : diagnosis for Asthma and COPD (7)
Measurements of the Forced Vital Capacity (FVC) , Forced expiratory
volume in one second (FEV1) and Airway responsiveness (PC20
Histamine) were done according to ERS criteria.(1).
6
During the follow-up visit, results of all investigations, carried out at the
discretion of the physician or according to the protocol, were available to
the physician who then decided whether a final diagnosis could be made.
Follow-up visits and additional PFT were scheduled as deemed
appropriate.
Redundant PFT tests were defined as tests that were not absolutely
necessary to establish a final diagnosis, e.g., when a reversible
obstructive PFT was found, there was no need for a histamine provocation
test, but when a normal flow volume curve was found, a reversibility test
was considered redundant.
The economic analysis was conducted from a health-care perspective
including direct medical costs only. Where available, unit cost prices were
derived from a national guideline for economic analysis of health care
services (9). In other instances, real cost prices were calculated on the
basis of hospital administration data. (Table 1.)
Statistical analysis
The "Unpaired t test with Welch's correction" was used to test for
statistical significance in differences between the two groups. An alpha of
0.05 was used as significance level. For statistical calculations we used
GraphPad Prism5 for windows (www.graphpad.com).
Results
Of a total of 183 patients, 179 patients were included: 86 patients in the
control group and 93 in the experimental group. Four patients were
7
excluded; one had a malignancy, one failed to follow up, one patient died
within a week of the start of the study, and in one patient the protocol
was not followed. The second physician calibrated his findings with the
findings of the PFT ordering physician only if there were conflicting
findings. In all these cases we had a consensus on the final diagnosis.
Patient characteristics at baseline are presented in Table 2. Classification
on the basis of PFT group and final diagnoses are presented in Table 3. In
the control group, 35 patients were diagnosed with asthma versus 39
patients in the experimental group. For the Non Reversible Obstructive
group, these numbers were 8 and 9 patients, respectively. For the Partly
Reversible Obstructive group the numbers were 12 and 14, respectively.
Diagnosis in the normal PFT group consisted of sarcoidosis (1), Gastro
Oesophageal Reflux (GER) (5), rhinitis and or sinusitis (7) ,
hyperventilation syndrome (3) , persistent cough (21 of whom 15
smokers or ex smokers) or dyspnoea (8 of whom 3 smokers).
On the basis of clinical assessment, 52 patients were found to have COPD
of whom 7 in GOLD stage 1 , 24 in stage 2, 13 in stage 3 , and 8 in
stage 4,
The costs of procedures used to reach a final diagnosis in patients
suspected of obstructive lung diseases.
In the control group, mean total costs of testing and outpatient visits per
diagnosis were €227.17; in the experimental group these costs were
€180.89 (P<0.001) Figure 2. meaning a 20% reduction in costs
8
The number and costs of outpatient visits until diagnosis
In the control group, two outpatient visits were needed to reach a final
diagnosis in 71 patients; in 15 patients three outpatient visits were
needed (mean 2.17 median 2 ). In the experimental group, 90 patients
had two visits and 3 patients had three visits (total of 189 visits, mean
2.03, median 2 ) The difference in total visits between these groups was
statistically significant (P=0.02). Mean costs were 88.80 in the control
group and 83,00 in the experimental group (P=0.02) Figure 2.
The number and costs of PFT needed for diagnosis: In the control
group, a mean number of 3.81 PFT per patient were necessary to
diagnose asthma or COPD. In the experimental group, a mean number of
2.94 tests were necessary (P<0.001) Figure 2.
In the control group mean costs of PFT were €138,37 and in the
experimental group €97.89 (P<0.001).
The number and costs of redundant PFT used for diagnosis:
In the control group a mean number of 1.20 redundant PFT per patient
was performed. In the experimental group 0,08 unnecessary PFT were
performed (P<0.001). Mean costs of redundant tests for diagnosis in the
control group were €51.35 and in the experimental group €3.30 Figure 2.
Histamine provocation test and added costs of reversibility testing were
the two most important sources of redundant costs (Table 4) .
9
Time until final diagnosis
In the control group, a mean number of 33.02 days was necessary to
reach a final diagnosis, whereas in the experimental group, a mean
number of 35.94 days was needed. (P=0.51).
POST-HOC
We evaluated the added value in making a diagnosis of asthma with the
steroid test in our patients. Conforming to the protocol we needed in the
control group 11 steroid tests and in the experimental group 10 In none of
these tests the diagnoses changed to asthma because of a complete
reversibility after steroid use
DISCUSSION
In general, our study showed that the introduction of a problem-oriented
protocol for ordering lung function tests in patients with suspected
obstructive pulmonary disease can reduce redundant PFT and outpatient
visits. This resulted in a 20 % decrease of costs without an increase in
time to final diagnosis. Given the high incidence of PFT for obstructive
lung diseases, this results in a substantial saving of costs at a population
level. In our practice of 600 patients per year with suspected obstructive
lung disease, protocol-guided test-ordering can lead to an annual cost
reduction of €27.768 = €46.28 per patient. The most important part of
the saving is the reduced need for reversibility testing when there was a
normal flow volume curve obtained, and a reduced need for the time-
10
consuming unpleasant hyperresponsiveness (PC20 histamine) test when
obstruction with reversibility was obtained.
Our lung function protocol is based on the asthma and COPD guidelines
(7) of The Netherlands. It can also be used as a lung function protocol
within the ERS and ATS standards(3;4;6;10). Our patients we think are
not different from most other countries except that most patients were
referred to us from a family doctor, in some other countries a family
physician will be skipped more often. So a sort of selection bias is
possible. However the criteria how to diagnose COPD or asthma does not
depend on the physician the patient is going to. Of course the skills and
the tools are different between general practitioners and pulmonary
physicians but
for this study we included only the most basal lung
function tests and omitted test as e.g. diffusion capacity or exercise
testing. These test are not always needed to confirm a diagnosis of
asthma or COPD. We think indeed that this protocol (or workflow) will be
suitable for general practioners too except they need a lunfunction lab or
have to order it e.g. in a hospital based lungfunction lab as is possible at
out institution.
As we stated before some physicians did use the protocol driven lung
function protocol we made about 3 years before this study. These were
the author of this study and two other physicians followed mostly the
protocol driven testing. Two other physicians one more senior and one
11
younger did not implement this protocol driven ordering and stuck to their
old behaviour. They were not convinced of the time and lung function
tests sparing effect and wanted proof.
We all had a consensus on performing this study and behaviour did
change after finishing this study.
Trainees stays for four years in our hospital, we had that time five
trainees in all stages of their educational process. Now we asked our
trainees if possible to order according to a test-protocol rather than at
their own discretion. We learned them that this is more efficient.
We all (physicians) have to decide what additional research is needed for
of our patients and we cannot be interrupted, depending on the previous
step, too often to discuss the next step to do,. So if the pulmonary
function assistant can follow a diagnostic protocol without the need to ask
the opinion of the physician on how to proceed, the work will be more
efficient. The workflow is based on our national guideline which resembled
the international guidelines of the ERS.
Post hoc, we evaluated the added value in making a diagnosis of asthma
with the steroid test in our patients. In none of these tests the diagnoses
changed to asthma because of a complete reversibility after steroid use.
So we doubted about the need for such a test in a routine setting. After
this study we removed this test out of the pulmonary function protocol.
At the time of this study we had waiting lists for the outpatientdepartment and for lung function tests. These waiting lists motivated us to
12
search for a more efficient and less time consuming work-up. So we
expected that the waiting list in the CG would be longer, but overall only
half of the patients followed the protocol, and that was not enough to
shorten waiting The total time to diagnose from the first visit till the final
diagnosis, was not different in both groups but without waiting lists we
think we will have a difference between the CG and the EG in favour of the
EG.
A weakness in our study is a potential Hawthorne effect: knowing that
they were involved in a trial may have affected the physicians’ behaviour
(test ordering, making a final diagnosis). The only way to avoid this
problem is to conduct the study retrospectively, which would challenge its
internal validity. We therefore decided to conduct a random, parallel
design to ensure internal validity (equivalence between groups, minimal
likelihood of confounding). However, to the extent that our study suffered
from a Hawthorne effect, this will most likely have resulted in an
underestimate of the effect of protocol-guided test-ordering.
To minimize the Hawthorne effect we discussed after the inclusion period
the whole results within our group. Some doctors want to work efficiently,
and they ordered more than was the absolute minimum, so there were no
other visits to the physician necessary.
Conclusion,
13
Problem orientated pulmonary function testing significantly reduces
pulmonary function tests, costs and outpatient visits in the diagnosis of
asthma and COPD.
14
Tests
Costs
Flow volume curve
€ 15,00
reversibility (bronchus
€ 18,00
dilators) testing
TLCO (diffusion capacity)
€ 41,00
Hyper reactivity (histamine)
€ 84,00
TLC
€ 41,00
Outpatient visit
€ 41,00
Steroid test
€ 33,00
Table 1.: Costs of pulmonary function tests and visit
Control group
Experimental group
86
93
55.4
56.8
45
46
168.4
170.8
FEV1 % pred.
77.8
78.0
FVC %pred
91.3
91.0
Non smoker
41%
38%
Ex smoker
20%
23%
Current smoker
33%
29%
6%
10%
Number of
patients
Avg. age (years)
Sex (% male)
Height
Unknown smoker
Table 2: Patient characteristics
Classification on
Control group
Experimental group
41%
42%
17%
17%
14%
15%
27%
26%
41%
42%
5%
1%
COPD non reversible
17%
17%
COPD partial
10%
14%
27%
26%
the basis of PFT
CompletelyReversible
obstructive (CRO)
Non reversible
obstructive (NRO)
Partly Reversible
Obstructive (PRO).
Normal PFT (NO)
Final diagnosis
Asthma
Asthma partial
reversible
reversible
Other diagnosis
Table 3 : Classification on the basis of PFT and final diagnoses
Reduntant
Control group
Experimental group
PFTfunction
test
Number (%)
Costs in €
Number (%)
Costs in €
19 (22)
1596
1 (1)
84
30 (35)
540
1 (1)
18
TLC
30 (35)
1230
3 (3)
123
Diffusion
24 (28)
984
2 (2)
82
103 (120)
4323
7 (8)
307
Histamine
provocation
test
Reversibility
testing
Capacity for CO
Totals
Table 4. Redundant pulmonary function tests
Figure 1: Pulmonary function protocol for obstructive diseases. Criteria for
obstruction, airway responsiveness (PC20 histamine) ,reversibility, and
steroid test: see text.
Figure 2: Mean costs per patient per diagnosis. Redund.= Redundant
Reference List
(1) Miller MR, Hankinson J, Brusasco V, Burgos F, Casaburi R, Coates A et al.
Standardisation of spirometry Eur Respir J 2005; 26(2):319-338.
(2) Siafakas NM, Vermeire P, Pride NB, Paoletti P, Gibson J, Howard P et al.
Optimal assessment and management of chronic obstructive
pulmonary disease (COPD). The European Respiratory Society Task
Force. Eur Respir J 1995; 8(8):1398-1420.
(3) Global Initiative for Asthma (GINA). Global strategy for asthma
management and prevention
(updated 2005): . 2006.
Ref Type: Internet Communication
(4) Quanjer Ph.H., Tammeling G.J. Standardized lung function testing.
Official statement of the ERS. Eur Respir J 1993; 6(Suppl 16):5-40.
(5) Cockcroft DW. Bronchoprovocation methods: direct challenges. Clin Rev
Allergy Immunol 1994; 24(1):19-26.
(6) Pellegrino R, Viegi G, Brusasco V, Crapo RO, Burgos F, Casaburi R et al.
Interpretative strategies for lung function tests. Eur Respir J 2005;
26(5):948-968.
(7) Commissie diagnostiek astma bronchiale en COPD N. Standaard
diagnostiek van astma bronchiale en COPD. Pulmoscript 1994;
1994;(3):31-46.
(8) DalesRE STS, SuissaS.
Clinicalinterpretationofairwayresponsetoabronchodilator:epidemiologicc
onsiderations. Am Rev Respir Dis 1988; 138:317-320.
(9) Oostenbrink JB, Bouwmans CAM, Koopmanschap M.A., Rutten FFH.
Maual for costing studies. Methods and standard cost prices for
economic evaluations in health care. National Health Insurance Board
(in Dutch) 2004.
(10) Lung function testing: selection of reference values and interpretative
strategies:oficial statement ATS. Am Rev Respir Dis 1991; 144:12021216.