Technical Note

Respiration 2012;83:344–352 Received: August 22, 2011

Accepted after revision: October 10, 2011
DOI: 10.1159/000334320
Published online: January 11, 2012

Accuracy and Precision of Desktop

Spirometers in General Practices
Tjard R.J. Schermer a, d Eddy H.A. Verweij a Riet Cretier a Annelies Pellegrino a, b
Alan J. Crockett d, e Patrick J.P. Poels a, c
a
COPD and Asthma Research and Development Unit, Department of Primary and Community Care, and
b
Department of Epidemiology, Biostatistics, and Health Technology Assessment, Radboud University Nijmegen
Medical Centre, Nijmegen, and c General Practice Bles en Poels, Huissen, The Netherlands; d Primary Care
Respiratory Research Unit, Discipline of General Practice, School of Population Health and Clinical Practice,
Faculty of Health Sciences, University of Adelaide, e School of Health Sciences, University of South Australia,
Adelaide, S.A., Australia

Key Words displacement spirometer. Mean age of the spirometers was

Accuracy ⴢ Diagnosis ⴢ Equipment evaluation ⴢ General 4.3 (SD 3.7) years. Average deviation from the waveform gen-
practice ⴢ Precision ⴢ Safety ⴢ Spirometry erator reference values (accuracy) was 25 ml (95% confi-
dence interval 12–39 ml) for FEV1 and 27 ml (10–45 ml) for
FVC, but some devices showed substantial deviations. FEV1
Abstract deviations were larger for pneumotachographs than for tur-
Background: Spirometry has become an essential tool for bine spirometers (p ! 0.0031), but FVC deviations did not dif-
general practices to diagnose and monitor chronic airways fer between the two types of spirometers. In the subset of
diseases, but very little is known about the performance of turbine spirometers, no association between age and device
the spirometry equipment that is being used in general performance was observed. Conclusions: On average, desk-
practice settings. The use of invalid spirometry equipment top spirometers in general practices slightly overestimated
may have consequences on disease diagnosis and manage- FEV1 and FVC values, but some devices showed substantial
ment of patients. Objectives: To establish the accuracy and deviations. General practices should pay more attention to
precision of desktop spirometers that are routinely used in the calibration of their spirometer.
general practices. Methods: We evaluated a random sample Copyright © 2012 S. Karger AG, Basel
of 50 spirometers from Dutch general practices by testing
them on a certified waveform generator using 8 standard
American Thoracic Society waveforms to determine accura- Introduction
cy and precision. Details about the brand and type of spi-
rometers, year of purchase, frequency of use, cleaning and Spirometry has become an essential tool for primary
calibration were inquired with a study-specific question- care doctors and nurses to diagnose and monitor chronic
naire. Results: 39 devices (80%) were turbine spirometers, 8 airways diseases [1–4]. In the Netherlands, spirometer
(16%) were pneumotachographs, and 1 (2%) was a volume ownership in general practices has increased substantial-

© 2012 S. Karger AG, Basel Tjard R.J. Schermer, MSc, PhD

0025–7931/12/0834–0344$38.00/0 Department of Primary and Community Care (117-ELG)
Fax +41 61 306 12 34 Radboud University Nijmegen Medical Centre, PO Box 9101
E-Mail [email protected] Accessible online at: NL–6500 HB Nijmegen (The Netherlands)
www.karger.com www.karger.com/res Tel. +31 24 361 4611, E-Mail T.Schermer @ elg.umcn.nl
ly in the past decade (from 25% in 1998 to 62% in 2007) 2007 [5]. Out of these 700 practices, we took random subsamples
[5]. In this same period of time, the rate of spirometry tests of 25 practices for each of the 12 provinces in our country. These
300 practices were contacted by telephone to inquire whether or
performed in Dutch primary care has tripled [5]. Spirom- not the practice owned one or more spirometers and if so, if they
etry tests performed in general practice have shown to be were willing to submit the device they most frequently used in
valid compared to tests performed in a pulmonary func- their regular patient care for performance testing. Details about
tion laboratory – provided that the tests are carried out by the brand and type of the spirometer, year of purchase, frequency
well-trained staff and that spirometers are regularly tested of use, cleaning and calibration routines, and history of previous
damage and repairs (if any) were inquired with a study-specific
for errors and deviations [6]. Nonetheless, substandard questionnaire. No formal sample size calculation was performed,
quality of spirometry tests routinely performed in general but we aimed to include 50 spirometers in the evaluation. Ethics
practices is a reason for concern [7–10]. approval or internal review board approval were not required for
General practices use various brands and types of this study.
desktop spirometers [5, 11, 12], which are generally less
Spirometer Testing and Outcomes
advanced but more affordable than the equipment typi- After shipment by express courier from the general practice
cally used in laboratories. A previous study has shown to the testing site (PT Medical BV, Leek, The Netherlands), spi-
that most brands of spirometers used in general practices rometers were tested by a trained operator on an ATS-certified
measure forced expiratory volume in 1 s (FEV1) and Standard wPWG S/N F02 waveform generator by applying 8
forced vital capacity (FVC) sufficiently accurate to diag- standard waveforms (Nos. 1, 4, 11, 15, 17, 19, 21, and 24) [28] dur-
ing a single session under ambient conditions which – in all cas-
nose chronic respiratory conditions [13]. However, the es – were in the operating temperature range of the devices [mean
spirometers in this particular study had been obtained temperature during testing: 22.9 ° C (SD 2.2), range 20.0–28.5 ° C;
       

directly from the manufacturers. Investigating such ‘off- air pressure 1,016 mb (SD 11.9), range 989–1,054 mb; humidity
the-shelf’ equipment may give a biased view on the per- 50.3% (SD 8.9), range 33–64%]. The waveform generator system
formance of devices that have routinely been used in was maintained in accordance with the manufacturers’ operat-
ing manual. The 8 waveforms were selected from the full set of
general practices for a period of time, especially while 24 ATS waveforms because together they represent a wide range
maintenance and calibration of spirometers in general of FEV1 and FVC values. A selection of the complete set of 24
practices is often well below laboratory standards [5, 11, waveforms has previously been used to evaluate lung function
12]. Lack of maintenance and calibration and wear of the equipment [29].
equipment over time are likely to negatively impact the Each waveform was applied 3 consecutive times to each spi-
rometer. Spirometer performance was expressed in terms of ac-
level of spirometer performance [14], which may have curacy and precision. To determine accuracy, we calculated the
consequences for the interpretations that are based on the mean of the 3 readings for the FEV1, FVC and forced mid-expi-
results of a spirometry test and the subsequent manage- ratory flow (FEF25–75) for each waveform, and compared these
ment of patients. with the gold standard reference values for these indices pro-
Several previous studies have looked at performance duced by the waveform generator. For FEV1 and FVC, we consid-
ered a spirometer to be inaccurate for a particular waveform in
markers of particular spirometry devices when used in case of a 6100 ml or 63.5% deviation (either positive or nega-
research or secondary care settings [15–27], but as far as tive) from the waveform generator reference value [28]. For
we are aware there are no published reports on the per- FEF25–75, we considered a 6250 ml/s or 65.5% deviation as inac-
formance of spirometers that are being used in general curate [28]. Precision of the spirometers was determined by cal-
practices as part of routine patient care. In this study, we culating the difference between the highest and the lowest of the
3 FEV1, FVC and FEF25–75 readings for each of the 8 waveforms.
evaluated a random sample of spirometers from Dutch In order to determine whether or not a spirometer was imprecise
general practices with regard to their accuracy and preci- for a particular waveform, the same cutoff points as for accuracy
sion. We applied recommendations from the American were used [28].
Thoracic Society (ATS) [28] to determine spirometer per-
formance. Analysis
We used SPSS (Statistical Package for Social Sciences), version
17.0, for all analyses (volumes and flows are expressed in ml and
ml/s, respectively). Mean values and 95% confidence intervals (CI)
were calculated for accuracy and precision estimates for the 8 sep-
Materials and Methods arate waveforms as well as for all waveforms combined. After cal-
culating accuracy and precision estimates for FEV1, FVC and
Study Design FEF25–75 for each of the spirometers tested, we plotted the differ-
We first took a random sample of 700 general practices from ences between the respective waveform generator reference values
all 4,533 practices registered in the national database of the Neth- and the spirometer readings against the mean of these values in
erlands Institute for Health Services Research (www.nivel.nl) in Bland-Altman plots [30]. The upper and lower limits of agreement

Performance of General Practice Respiration 2012;83:344–352 345

Spirometers
Table 1. Accuracy and precision for FEV1 and FVC per waveform and overall for the sample of 49 spirometers from general practices
[means (95% CI)]. Results for FEF25–75 are published as online supplementary material.

Wave- FEV1 FVC

form
No. generator accuracy accuracy precision precision % generator accuracy accuracy precision precision %
ml ml % ml ml ml % ml

1 4,395 42 0.9 20 0.5 6,000 7 0.1 29 0.5

(21, 62) (0.5, 1.4) (15, 26) (0.3, 0.6) (–26, 40) (–0.4, 0.7) (18, 40) (0.3, 0.7)
4 1,387 25 1.8 10 0.7 1,500 26 1.7 16 1.1
(17, 33) (1.2, 2.4) (6, 13) (0.5, 0.9) (15, 37) (1.0, 2.4) (12, 20) (0.8, 1.3)
11 1,784 20 1.1 10 0.6 2,701 35 1.3 29 1.1
(9, 32) (0.5, 1.8) (6, 14) (0.4, 0.8) (18, 51) (0.7, 1.9) (20, 38) (0.7, 1.4)
15 5,303 39 0.7 22 0.4 5,934 48 0.8 36 0.6
(15, 63) (0.3, 1.2) (17, 27) (0.3, 0.5) (22, 73) (0.4, 1.2) (22, 50) (0.4, 0.8)
17 2,597 18 0.7 14 0.5 5,829 49 0.8 52 0.9
(1, 35) (0.1, 1.4) (8, 19) (0.3, 0.7) (3, 95) (0.1, 1.6) (31, 73) (0.5, 1.2)
19 2,505 20 0.8 13 0.5 3,931 –9 –0.2 32 0.8
(7, 33) (0.3, 1.3) (10, 17) (0.4, 0.7) (–39, 22) (–1.0, 0.6) (21, 43) (0.5, 1.1)
21 3,519 24 0.7 16 0.4 4,446 34 0.8 38 0.8
(8, 39) (0.2, 1.1) (10, 21) (0.3, 0.6) (12, 57) (0.3, 1.3) (16, 60) (0.4, 1.3)
24 916 15 1.6 9 1.0 1,230 29 2.3 24 1.7
(7, 23) (0.8, 2.5) (7, 12) (0.7, 1.3) (11, 46) (0.9, 3.7) (5, 43) (0.5, 2.9)
Overall 25 1.0 14 0.6 27 1.2 35 1.0
(12, 39) (0.4, 1.6) (11, 17) (0.5, 0.7) (10, 45) (0.6, 1.9) (24, 47) (0.6, 1.4)

Accuracy = Mean of the 3 spirometer readings for the FEV1 and FVC with the gold standard reference values for these indices as
produced by the waveform generator; precision = difference between the highest and the lowest of the 3 FEV1 and FVC readings for
the applied waveform.

around the mean difference were set on 8100 ml for FEV1 and spirometry [5]. In total, 154 practices (62%) reported
FVC, and 8250 ml for FEF25–75, respectively [28]. Based on the age owning 61 spirometers. Of these 154 practices, 63 (41%)
distribution in the sample, we divided the subset of turbine spi- agreed to submit their spirometer for testing and to com-
rometers (the predominant type of sensor in our sample as well as
in Dutch general practices nationally [5]) into three categories: plete the questionnaire about its details and use. Eventu-
1–2; 3–5, and 66 years old. We used analysis of variance (ANOVA) ally 50 (32%) spirometers were submitted for testing on
to analyze associations between spirometer type and age and FEV1 the waveform generator. The main reasons for practices
and FVC accuracy and precision estimates. To correct for multiple not to submit their spirometer after all were that they
comparisons in these analyses, statistical significance was defined ‘could not miss their spirometer’ (35%) or ‘were too busy
as p ! 0.0031 (Bonferroni correction: ␣ = 0.05/2⫻8 waveforms).
to arrange shipment of the spirometer to the testing site’
(21%).
One spirometer was sent to the testing site but could
Results not be tested because the software necessary to operate
the spirometer was not available. Fourteen of the tested
Spirometers in the Study spirometers could not measure FEF25–75. Of the 49 spi-
Of the random sample of 300 general practices, 269 rometers tested, 38 (78%) had been manufactured by Mi-
(90%) practices could be contacted by telephone and 250 croMedical (now known as CareFusion), 6 (12%) by Vi-
(83%) were willing to provide information about their talograph, 3 (6%) by Clement Clarke International, and 1

346 Respiration 2012;83:344–352 Schermer /Verweij /Cretier /Pellegrino /

       

Crockett /Poels
   
Table 2. Summary of accuracy (a) and precision (b) performance oldest 18 years. Mean and median age for the subset of
criteria for FEV1 and FVC for the sample of 49 spirometers from turbine spirometers were 3.7 (SD 2.9) and 3.0 (interquar-
general practices
tile range: 4.0) years, respectively, with a minimum age of
a Accuracy 4 months and a maximum age of 12 years.
Online supplementary appendix 1 provides further
Number of waveforms Number of waveforms not meeting accuracy
not meeting accuracy1 criteria1 for FVC details for each of the tested spirometers in the sample
criteria for FEV1 (for all online supplementary material, see www.
0 1 2 3 4 5 6 7 8
karger.com/doi/10.1159/000334320).
0 21 10 4 1 1 1
1 1 1 Accuracy and Precision of the Spirometers
2 1 The total average deviation from the waveform gen-
3 erator reference values for the 8 applied waveforms (i.e.
4 1 accuracy) was 25 ml (95% CI 12–39 ml) for FEV1, 27 ml
5 1 (10–45 ml) for FVC, and –24 ml/s (–48 to 1 ml/s) for
6 1 FEF25–75, respectively. For precision, the respective mean
7 1 1 ranges were 14 (11–17) and 35 (24–47) ml and 45 (23–67)
8 1 1 1 ml/s. Table 1 shows the accuracy and precision results per
waveform (in ml as well as in %).
b Precision Figure 1 shows the difference against mean (Bland-
Altman) plots for FEV1 and FVC accuracy in milliliter by
Number of waveforms Number of waveforms not meeting precision
not meeting precision1 criteria1 for FVC
type of spirometer. The plot for FEV1 (fig. 1a) shows non-
criteria for FEV1 0 1 2 3 4 5 6 7 8
linearity of accuracy: the deviations between the genera-
tor reference values and the spirometer readings became
0 38 4 2 1 1 larger as the FEV1 value of the waveform increased. The
1 2 1 vast majority of the out-of-range FEV1 readings were at-
2 tributable to 5 of the 8 pneumotachographs in the sample
3 (for the deviations above the upper limit of agreement)
4 and to the single volume displacement spirometer (for the
5 deviations below the lower limit). Non-linearity was also
6 observed for the deviations in FVC readings (fig. 1b), with
7 deviations up to +590 and –490 ml. The deviations above
8 the upper limit of agreement were attributable to 19 tur-
bine spirometers, 7 pneumotachographs, and the volume
Numbers in the cells indicate numbers of spirometers.
1 ≥100 ml or ≥3.5% deviation (either positive or negative) from displacement spirometer. Online supplementary figure
the waveform generator reference value. 1c shows the Bland-Altman plot for FEF25–75 accuracy for
the subset of 35 spirometers that could measure this in-
dex.
With regard to accuracy, ANOVA testing showed that
FEV1 deviations were statistically significantly larger in
(2%) by Mijnhardt. Information about the brand and type the pneumotachographs than in the turbine spirometers
of the sensor was missing for 1 spirometer. Thirty-nine for all waveforms; FVC deviations did not differ between
spirometers (80% of all) had turbine flow sensors, 8 (16%) the two types of spirometers for any of the waveforms.
were pneumotachographs, and 1 (2%) was a volume dis- With regard to precision, the ranges in FEV1 readings
placement spirometer. Six spirometers (12%) had been re- were statistically significantly larger in the pneumo-
paired after having been purchased by the practice, and tachographs than in the turbine spirometers for 5 wave-
1 practice reported that the transducer of their spirome- forms (Nos. 1, 4, 11, 17, and 19). For the ranges in FVC
ter had once been replaced. readings, this was true for 3 waveforms (Nos. 1, 4, and 19).
Mean and median age for the total sample of spirom- Table 2 provides a summary of the combined accu-
eters were 4.3 (SD 3.7) and 3 (interquartile range: 4.5) racy and precision performance criteria for FEV1 and
years. The ‘youngest’ spirometer was 4 months old, the FVC for the total spirometer sample. With regard to ac-

Performance of General Practice Respiration 2012;83:344–352 347

Spirometers
 WF 24 WF 11 WF 17 WF 1
400 WF 4 WF 19 WF 21 WF 15

300

FEV1 spirometer – FEV1 WF generator (ml)

200

100 Limit of agreement

Mean deviation
0

–100 Limit of agreement

–200

–300
Turbine (n = 39)
Pneumotachograph (n = 8)
–400 Volume displacement (n = 1

0 1,000 2,000 3,000 4,000 5,000 6,000

a Mean of FEV1 spirometer and FEV1 WF generator (ml)

WF 24 WF 11 WF 21 WF 15
600 WF 17 WF 1
WF 4 WF 19
500

400
FVC spirometer – FVC WF generator (ml)

300

200

100

–100

–200

–300

–400
Fig. 1. Difference against mean plots for –500 Turbine (n = 39)
accuracy of FEV1 (a) and FVC (b) for all 49 Pneumotachograph (n = 8)
–600 Volume displacement (n = 1)
spirometers in the study sample. Each
symbol represents the result for 1 of the 8
1,000 2,000 3,000 4,000 5,000 6,000 7,000
applied waveforms (WF) (mean of 3 read-
b Mean of FVC spirometer and FVC WF generator (ml)
ings for WF 1, 4, 11, 15, 17, 19, 21, and 24
[28]) for 1 spirometer.

curacy, 21 (42%) devices fulfilled the FEV1 and FVC per- es fulfilled the FEV1 and FVC performance criteria for
formance criteria for all 8 waveforms, 11 (22%) devices all 8 waveforms or failed to meet the criteria for only 1
failed to meet the criteria for only 1 waveform, and the waveform. If the precision performance criteria were not
remaining 17 (36%) devices failed to meet the criteria for met, this was mostly due to FVC values being imprecise
61 waveform. With regard to precision, 42 (86%) devic- (table 2b).

348 Respiration 2012;83:344–352 Schermer /Verweij /Cretier /Pellegrino /

       

Crockett /Poels
   
 Influence of Age on Turbine Spirometer Performance
Figure 2 shows the difference against mean plots for 1–2 years old (n = 14)
400
FEV1 and FVC accuracy by age category for the subset of 3–5 years old (n = 15)
turbine spirometers. Only 4 devices (all aged 12 years) 300
≥6 years old (n = 8)

FEV1 spirometer – FEV1 WF generator (ml)

showed out-of-range deviations for 1 or 2 waveforms in
FEV1 readings (fig. 2a), whereas 17 devices (equally dis- 200
tributed over all 3 age categories) showed out-of-range
deviations for 61 waveforms in FVC readings (fig. 2b). 100
Apart from the fact that for FVC the most extreme de-
viations below the lower limit of agreement were attribut- 0

able to several devices from the oldest age category

–100
(fig. 2b), the plots show no clear associations between age
and performance in terms of accuracy. Online supple- –200
mentary figure 2c shows the Bland-Altman plot for
FEF25–75 accuracy for the subset of 27 turbine spirometers –300
that could measure this index. ANOVA testing did not
show any differences in FEV1 or FVC accuracy or preci- –400
sion estimates between the three age categories.
0 1,000 2,000 3,000 4,000 5,000 6,000
a Mean of FEV1 spirometer and FEV1 WF generator (ml)

Discussion

In this study, we tested a random sample of 49 spirom-

eters from Dutch general practices with regard to their 600 1–2 years old (n = 14)
3–5 years old (n = 15)
accuracy and precision using an ATS waveform generator 500 ≥6 years old (n = 8)
and a selection of standard ATS waveforms to determine 400
FVC spirometer – FVC WF generator (ml)

spirometer performance. Turbine spirometers dominat-

300
ed the sample, which is consistent with observations from
our larger survey in Dutch general practices [5]. With re- 200

gard to accuracy, for FEV1 as well as FVC the overall av- 100
erage deviation from the waveform generator reference 0
values was approximately +25 ml. With regard to preci- –100
sion, the average difference between the highest and the
–200
lowest of the three readings was 14 ml for FEV1 and 35 ml
for FVC. The single (18-year-old) volume displacement –300

spirometer in the sample should be considered an excep- –400

tion, as nowadays only very few practices will use such a –500
spirometer. Although the number of pneumotachographs
–600
in the sample was small (n = 8) and pneumotachographs
are more susceptible to environmental temperature and 1,000 2,000 3,000 4,000 5,000 6,000 7,000
humidity than turbine spirometers are [31], any compar- b Mean of FVC spirometer and FVC WF generator (ml)
ison between the different types of devices should be in-
terpreted with caution; our observations do suggest that
lack of accuracy of both FEV1 and FVC may be a problem Fig. 2. Difference against mean plots for accuracy of FEV1 (a) and
when general practices choose to use a pneumotacho- FVC (b) by age category for the subset of turbine spirometers (n =
37). Each symbol represents the result for 1 of the 8 applied wave-
graph for their spirometry. In general, accuracy seemed forms (WF) for 1 spirometer. The dotted lines represent the limits
to be more of a problem than precision in our spirometer of agreement (8100 ml).
sample. The subset of 38 turbine spirometers showed a
wide age range (4 months to 12 years) but despite the lim-
ited attention towards calibration and maintenance in the

Performance of General Practice Respiration 2012;83:344–352 349

Spirometers
practices who owned them, we did not observe an asso- pneumotachograph may be more prone to lose its accu-
ciation between the age of the devices and their actual racy. In the past decade, there have been technological
performance. advancements in the design and durability of turbine
sensors. A previous (1996) report about the performance
Strengths and Limitations of turbine sensors over time is therefore rather outdated
Initially, 63 practices agreed to submit their spirome- [21], but nonetheless it supports our observation that
ter for testing, but only 49 practices eventually did. The these sensors seem to retain their accuracy for a number
14 ‘missing’ spirometers may have caused selection bias of years.
because the main reason for not submitting was that the
practices could not miss them. This might indicate that Implications for Clinical Practice
these spirometers were more frequently used than the The majority of general practices in our study seemed
ones that we were able to test, and frequency of use may to pay no or only limited attention to calibration of their
cause faster wear and declining performance of the de- spirometers, which has previously been shown to be rath-
vice [32]. We do not know the actual motives of the 91 er common practice in the Netherlands [5]. For some
practices that did own a spirometer but were not able or practices, this lack of attention seemed not justified, as
willing to submit it for testing, but this may also have their spirometers showed substantial inaccuracy and/or
caused selection bias. We cannot exclude the possibility imprecision when tested on the waveform generator. On
that practices that owned a rather old spirometer were average, the spirometers slightly (⬃25 ml) overestimated
more likely to cooperate because they would like to see both FEV1 and FVC values, but some devices in the sam-
their spirometer tested and be reassured that it still per- ple – especially the pneumotachographs – showed sub-
formed sufficiently well. stantial deviations from the waveform generator values.
Budget restraints did not allow us to test each spirom- We do not have patient-related data from the general
eter according to the full procedure as recommended by practices to confirm this, but it is quite likely that devices
the ATS (i.e. 5 consecutive maneuvers for all 24 wave- that showed large measurement errors will have caused
forms) [28]. However, we did use the full procedure in a misinterpretation (especially overestimation) of FEV1
random subsample of 10 spirometers, and for these de- and FVC values in patients who have been assessed with
vices applying the full procedure did not change our in- these devices.
terpretation regarding their performance. One of the current issues in primary care is how spi-
Another limitation of our study is that there were no rometer performance can be checked in this setting, with
ultrasonic spirometers in our sample, because these were limited resources, expertise and access to laboratory fa-
only recently introduced in the Netherlands. Previous re- cilities. Clearly, regular use of a calibration syringe would
ports have shown that this type of spirometer seems rath- be desirable, even while this does not guarantee that the
er robust in the long term [20, 25], and may therefore be spirometer readings from patients’ forced maneuvers are
a relevant addition to the current assortment of spirom- indeed accurate [15, 33]. Regular biological calibration
eters Dutch general practices can choose from. In other (by a healthy test operator in the practice) [34, 35] is prob-
countries, ultrasonic spirometers already seem to be rath- ably the most feasible option. When purchasing a spirom-
er common in primary care practices [8, 11]. eter, general practices should consider the robustness of
the different types of sensors and the frequency of main-
Comparison with Previous Studies tenance and calibration they require, next to other im-
From their evaluation of 10 brand-new desktop spi- portant factors like hygiene, inclusion of the appropriate
rometers in a large sample of subjects who were also mea- reference equations, user friendliness, and price. Regard-
sured with standard diagnostic laboratory spirometers, ing accuracy and precision, the limited evidence current-
Liistro et al. [13] concluded that accuracy of FEV1 and ly available suggests that when attention to equipment
FVC was generally sufficient for turbine spirometers as maintenance and calibration is likely to be limited, a tur-
well as for pneumotachographs, but their study also bine or ultrasonic spirometer may be the better option for
showed that some brands of pneumotachographs seemed a general practice.
to perform not as good as turbine spirometers did in In conclusion, our evaluation of a sample of spirome-
terms of linearity and precision. Our observations seem ters from Dutch general practices showed that on average
to support this, and also suggest that once a spirometer these spirometers may slightly (⬃25 ml) overestimate
has been used in a general practice for a period of time, a FEV1 and FVC values, with larger deviations being more

350 Respiration 2012;83:344–352 Schermer /Verweij /Cretier /Pellegrino /

       

Crockett /Poels
   
likely at higher values for these indices. In general, accu- Acknowledgments
racy seemed to be more of an issue than precision was.
This study was sponsored by the Radboud University Nijme-
The vast majority of the sample consisted of turbine spi- gen Medical Centre. The authors appreciate the work put in the
rometers, but lack of accuracy was mainly attributable to project by Jorn de Vreede, who conscientiously tested the spirom-
the small number of pneumotachographs, especially with eters on the waveform generator. We also appreciate the time and
regard to FEV1. The subset of turbine spirometers showed effort it took the general practice staffs to submit their spirome-
a wide age range, from only a few months up to 12 years, ters for testing.
but we did not observe an association between the age of
the devices and their actual performance, despite the lim-
ited attention towards calibration and maintenance in Disclosure Statement
many of the practices that owned them. General prac-
None of authors has any conflict of interest to declare.
tices should pay more attention to the calibration of their
spirometer, either by using a calibration syringe or a bio-
logical control. When obtaining a spirometer, they should
take the robustness of the different types of sensors and
the frequency of maintenance and calibration they re-
quire into consideration.

References
1 Schirnhofer L, Lamprecht B, Firlei N, Kaiser 8 Walters JA, Hansen EC, Johns DP, Blizzard 16 Dompeling E, van Schayck CP, Folgering H,
B, Buist AS, Halbert RJ, Allison MJ, Stud- EL, Walters EH, Wood-Baker R: A mixed van den Hoogen HJ, van Weel C: Accuracy,
nicka M: Using targeted spirometry to re- methods study to compare models of spi- precision and linearity of the portable flow-
duce non-diagnosed chronic obstructive rometry delivery in primary care for pa- volume meter Microspiro HI-298. Eur Respir
pulmonary disease. Respiration 2011; 81: tients at risk of COPD. Thorax 2008;63:408– J 1991;4:612–615.
476–482. 414. 17 Rode A, Shephard RJ: Accuracy of an elec-
2 Levy ML, Fletcher M, Price DB, Hausen T, 9 Leuppi JD, Miedinger D, Chhajed PN, Buess tronic spirometer – a field trial. Respiration
Halbert RJ, Yawn BP: International Primary C, Schafroth S, Bucher HC, Tamm M: Qual- 1986;50:66–69.
Care Respiratory Group (IPCRG) Guidelines: ity of spirometry in primary care for case 18 Maree DM, Videler EA, Hallauer M, Pieper
diagnosis of respiratory diseases in primary finding of airway obstruction in smokers. CH, Bolliger CT: Comparison of a new
care. Prim Care Respir J 2006;15:20–34. Respiration 2010;79:469–474. desktop spirometer (Diagnosa) with a labo-
3 Rabe KF, Hurd S, Anzueto A, Barnes PJ, 10 Carr R, Telford V, Waters G: Impact of an ratory spirometer. Respiration 2001; 68:
Buist SA, Calverley P, et al: Global strategy educational intervention on the quality of 400–404.
for the diagnosis, management, and preven- spirometry performance in a general prac- 19 Rebuck DA, Hanania NA, D’Urzo AD, Chap-
tion of chronic obstructive pulmonary dis- tice: an audit. Prim Care Respir J 2011; 20: man KR: The accuracy of a handheld porta-
ease: GOLD executive summary. Am J 210–213. ble spirometer. Chest 1996;109:152–157.
Respir Crit Care Med 2007;176:532–555. 11 Johns DP, Burton D, Walters JA, Wood-Ba- 20 Walters JA, Wood-Baker R, Walls J, Johns
4 National Collaborating Centre for Chronic ker R: National survey of spirometer owner- DP: Stability of the EasyOne ultrasonic spi-
Conditions: Chronic obstructive pulmonary ship and usage in general practice in Austra- rometer for use in general practice. Respirol-
disease: national clinical guideline on man- lia. Respirology 2006;11:292–298. ogy 2006;11:306–310.
agement of chronic obstructive pulmonary 12 Dowson LJ, Yeung A, Allen MB: General 21 Dirksen A, Madsen F, Pedersen OF, Vedel
disease in adults in primary and secondary practice spirometry in North Staffordshire. AM, Kok-Jensen A: Long-term performance
care. Thorax 2004;59(suppl I):1–232. Monaldi Arch Chest Dis 1999;54:186–188. of a hand held spirometer. Thorax 1996; 51:
5 Schellekens D, Poels P, Pellegrino A, Cretier 13 Liistro G, Vanwelde C, Vincken W, Vande- 973–976.
R, Smeele I, Schermer T: Spirometry in voorde J, Verleden G, Buffels J: Technical and 22 Swart F, Schuurmans MM, Heydenreich
Dutch general practice. Results from a na- functional assessment of 10 office spirome- JC, Pieper CH, Bolliger CT: Comparison of
tional survey (in Dutch). Huisarts Wet 2008; ters: A multicenter comparative study. Chest a new desktop spirometer (Spirospec) with
51:434–439. 2006;130:657–665. a laboratory spirometer in a respiratory
6 Schermer TR, Jacobs JE, Chavannes NH, 14 Townsend MC, Hankinson JL, Lindesmith out-patient clinic. Respir Care 2003; 48:
Hartman J, Folgering HT, Bottema BJ, et al: LA, Slivka WA, Stiver G, Ayres GT: Is my 591–595.
Validity of spirometric testing in a general lung function really that good? Flow-type 23 Perez-Padilla R, Vázquez-García JC, Már-
practice population of patients with chronic spirometer problems that elevate test results. quez MN, Jardim JR, Pertuzé J, Lisboa C, et
obstructive pulmonary disease (COPD). Chest 2004;125:1902–1909. al: The long-term stability of portable spi-
Thorax 2003;58:861–866. 15 Nelson SB, Gardner RM, Crapo RO, Jensen rometers used in a multinational study of
7 Schermer T, Crockett A, Poels P, van Dijke J, RL: Performance evaluation of contempo- the prevalence of chronic obstructive pul-
Akkermans R, Vlek H, et al: Quality of rou- rary spirometers. Chest 1990;97:288–297. monary disease. Respir Care 2006; 51: 1167–
tine spirometry tests in Dutch general prac- 1171.
tices. Br J Gen Pract 2009;59:e376–e382.

Performance of General Practice Respiration 2012;83:344–352 351

Spirometers
24 Malmberg LP, Hedman J, Sovijärvi AR: Ac- 28 American Thoracic Society. Standardization 32 Johns DP, Side E, Kendrick AH, Williams TJ,
curacy and repeatability of a pocket turbine of spirometry, 1994 Update. Am J Respir Crit Walters EH: The effect of physiologic and
spirometer: comparison with a rolling seal Care Med 1995;152:1107–1136. mechanical aging on the performance of
flow-volume spirometer. Clin Physiol 1993; 29 Hankinson JL, Filios MS, Kinsley KB, Pet- peak flowmeters. Chest 1998;113:774–779.
13:89–98. sonk EL: Comparing MiniWright and spi- 33 van den Boom G, van der Star LM, Folgering
25 Skloot GS, Edwards NT, Enright PL: Four- rometer measurements of peak expiratory H, van Schayck CP, van Weel C: Volume cal-
year calibration stability of the EasyOne por- flow. Chest 1995;108:407–410. ibration alone may be misleading. Respir
table spirometer. Respir Care 2010; 55: 873– 30 Bland JM, Altman DG: Statistical methods Med 1999;93:643–647.
877. for assessing agreement between two meth- 34 Ninaber MK, Schot R, Fregonese L, Stolk J: A
26 Barr RG, Stemple KJ, Mesia-Vela S, Basner ods of clinical measurement. Lancet 1986; 1: syringe simulation of biological controls for
RC, Derk SJ, Henneberger PK, et al: Repro- 307–310. quality assessment of prospective lung vol-
ducibility and validity of a handheld spirom- 31 Miller MR, Hankinson J, Brusasco V, Burgos ume measurements. Respiration 2008; 76:
eter. Respir Care 2008;53:433–441. F, Casaburi R, Coates A, Crapo R, Enright P, 187–192.
27 Jensen RL, Teeter JG, England RD, White HJ, van der Grinten CP, Gustafsson P, Jensen R, 35 Poels PJ, Schermer TR: Value of recommend-
Pickering EH, Crapo RO: Instrument accu- Johnson DC, MacIntyre N, McKay R, Nava- ed spirometer accuracy checks on office spi-
racy and reproducibility in measurements of jas D, Pedersen OF, Pellegrino R, Viegi G, rometers in primary care unknown. Respi-
pulmonary function. Chest 2007; 132: 388– Wanger J: Standardisation of spirometry. rology 2007; 12:151.
395. Eur Respir J 2005;26:319–338.

352 Respiration 2012;83:344–352 Schermer /Verweij /Cretier /Pellegrino /

       

Crockett /Poels