- Email: [email protected]
Survey on the use of pulmonary scintigraphy, spiral CT and conventional pulmonary angiography for suspected pulmonary embolism in the British isles
Survey on the Use of Pulmonary Scintigraphy, Spiral CT and Conventional Pulmonary Angiography for Suspected Pulmonary Embolism in the British Isles G.
J.
Received:
C.
BURKILL,
30 March
J.
1999
R.
Revised:
G.
BELL,
I I May
S.
1999
I’.
G.
Accepted:
PADLEY
26 May
1999
AIM: To determine current clinical practice in the radiological diagnosis of acute pulmonary embolism and assess the use of spiral volumetric computed tomography. METHOD: A survey of 327 acute hospitals including cardiothoracic and orthopaedic tertiary referral centres was undertaken to assess current utilization of lung scintigraphy, spiral computed tomography and pulmonary angiography in the investigation of suspected pulmonary embolism. Responses were received from 215/327 (66%) centres. RESULTS: Lung scintigraphy was provided by 208 hospitals (144 on-site and 64 off-site). Spiral CT services were provided by 111 (52%) hospitals (on- or off-site), 142 (66%) units had access to angiographic facilities. Sixty-three centres out of 215 (29%) offered both on-site lung scintigraphy and spiral CT while only 41/215 (19%) hospitals were able to undertake all three tests on-site. On average, 501 perfusion (Q) or ventilation-perfusion (V/Q) scintigrams were performed per hospital per year with 26 spiral CT studies and just 4.6 pulmonary angiograms. CONCLUSION: These data suggest that lung scintigraphy is frequently the only imaging test in patients other than chest radiography, despite the large number of indeterminate results reported in most series. Burkill. G. J. C.. Bell. J. R. G.. Padley. S. P. G. (1999). Clmicrtl Rtrdinlog~ 54, 8070 1999 The Royal College of Radiologists 810. Key words: helical, lung.
pulmonary radionuclide
embolism, studies
thrombosis.
The ability of computed tomography (CT) to detect pulmonary embolus was lirst described in 1980 [I]. However. it was not until the advent of spiral CT that sufficiently rapid data acquisition with satisfactory pulmonary artery opacilication became feasible [2] and raised hopes of a significant role for spiral CT in the diagnosis of pulmonary embolism (PE). Interest in the technique was stimulated by a combination of impressive sensitivity (100%) and specificity (96%) in early studies comparing CT and conventional pulmonary angiography [3], and the acknowledged shortcomings of lung scintigraphy [4] coupled with concerns over the safety [5] and availability [6] of the posited ‘gold standard’ of pulmonary angiography. A potential advantage of CT is that it may provide an alternative explanation for the patient’s symptoms: indeed the wholesale replacement of lung scintigraphy by spiral CT has been suggested and adopted by some centres in the United States 1781. The response to this call for change has been more cautious in the United Kingdom [9-I I]. This is in part. due to the lower availability of spiral CT, but also may reflect the utility of lung ventilation-perfusion (V/Q) scintigraphy in Author for correspondence and guanntor of study: Dr S. Padlry. Dep:utmerit of Radiology. Chelsea and Westminister Hospital. 369 Fulhnm Road. London. SW IO 9NH. UK. 0009-9260/99/120807+04
$12.00/O
pulmonary
angiopraphy.
computed
tomography
(CT),
patients with a low pre-test probability and normal chest radiograph [S]. The present survey was undertaken to assess the availability and current practice in the utilization of lung V/Q, spiral CT and conventional pulmonary angiography for the diagnosis of pulmonary thromboembolism.
METHODS A questionnaire and explanatory letter were sent to superintendent radiographers at 327 hospitals in the United Kingdom and Southern Ireland. All hospitals designated as a ‘Major Unit’ in the British Association for Accident and Emergency Medicine Directory 1996 were included. Specialized orthopaedie and cardiothoracic centres as detailed in the 1997 Medical Directory were also included. A second round of questionnaires were sent to the Head of department in units not responding to the first round of questionnaires. The questionnaire was divided into two parts. Part I was designed to establish the service provision of each department. Part I1 requested information concerning the total numbers of imaging investigations related to the diagnosis of pulmonary embolus. 0
1999 The Royal College
of Radiologists
808 Table
CLINICAL
1 - Investigation
availability
and hospital On-site
Q V/Q SCT ANGIO
dasignation Off-site
facilities
facilities
Total o,,- nd
DGH n= 179
Teaching II=26
Specialist ,I= IO
Total u=2lS
DGH
Teaching
Specialist
Toral
I12 109 72 74
25 25 I7 21
7 7 6 4
I44 141 95 99
58 60 I-l 37
I I I 2
3 3 I 4
62 64 I6 43
Q = perfusion.
(63%) (61%) (40%) (41%)
(965”) (967~) (65%) (81%)
V/Q = ventilation
perfusion.
(70%) (70%) (60%) (40%)
(677r) (66%) (44%) (46%)
SCT = spiral computed
tomography.
Respondents were asked to detail the number of patients undergoing sequential radiological tests to establish a diagnosis of PE. Identification of the current first line investigation was requested, and if this had recently changed. participants were asked to record when that change had been implemented.
RESULTS
Completed questionnaires were returned by 215 (66%) hospitals. One hundred and seventy-nine district general hospitals, 26 teaching hospitals and IO specialist centres (seven orthopaedic, three cardiothoracic). Two hundred and four of the 215 hospitals provided A&E medicine, 209 orthopaedic surgery and 36 cardiothoracic surgery. The availability of each investigation, by hospital designation is shown in Table I. One hundred and forty-four out of 215 (67%) hospitals provided an on-site lung scintigraphy (Q or V/Q) service: all but three of these had access to ventilation as well as perfusion studies. A further 64 hospitals (30%) had an off-site agreement to provide lung scintigraphy. The mean annual number of V/Qs undertaken in the I56 hospitals providing tigures was 274 per institution (Table 2). Similarly, in the I34 units undertaking perfusion studies the mean number per hospital per year was 227. Amongst the 23 teaching hospitals providing numerical data a higher number of lung scintigrams were ventilation and perfusion (353 per hospital per year) as opposed to perfusion only scintigrams (163 per hospital per year). Ninety-five out of 215 (44%) hospitals had on-site spiral CT and a further I6 (7%) provided this service off-site. Ninety-nine out of 215 (46%) units were able to offer on-site conventional pulmonary angiography and an additional 43 (20%) offered angiography off-site. Table 2 - Average per year
number
of investigations
undertaken
per hospital
Investigation
DGH
Teaching
SpecialisL
All units
Q V/Q
254 (103) 271 (IZ5) I5 (69) 3.1 (77)
I63 353 53 8
72 97 100 15
227 (13-l) 274 (156) 26 (87) 4.6 (103J
CT ANGIO
RADIOLOG
Figures in parenlheses indicate Key is the same as Table I
(23) (23) (16) (22)
number
(8) (Is) (2) fJJ
of hospitals
providing
full dak
(32%) (34%) (WC) (21%) ANGIO
(4%) (4%) (4%) (8%)
= pulmonary
(30%) (30%) (IO%) (‘Kwr)
angiopraphy.
(296) (30%) (7%) (‘Olr)
DGH = District
ofl-.~Sf-sik~
206 (96%) 20s (95%) I I I (52%) 141(66%) Gener;d
Hospital.
Of the I I I hospitals which had access to spiral CT either onsite (II = 95) or off-site (II = 16). 87 (78%) provided numerical data. These hospitals undertook a mean of 26 CT examinations per hospital per year specihcally for the diagnosis of PE: teaching hospitals performed on average 53 such CT examinations per hospital per year. Numerical data on pulmonary angiography was provided by a total of I03 of the 142 centres that indicated there was access to this investigation. On average. 4.6 angiograms were performed per hospital per year for the diagnosis of acute PE. with Il.4 per hospital over the past 3 years, giving an annual mean of 3.8 per hospital. Of the 23 teaching hospitals offering pulmonary angiography, 22 provided numerical data. At these institutions a mean of 8.0 pulmonary angiograms were undertaken per hospital in the previous year with 25.3 in the previous 3 years. giving an annual mean of X.4 per hospital. Of 59 hospitals providing numerical data concerning both lung scintigraphy and spiral CT. the average number of lung scintigrams per annum was 473 and CT examinations 31, giving a ratio of I5 : I, Of the 35 hospitals providing numerical data concerning all the imaging modalities. the average number of tests performed was 570 Q or V/Q. 40 spiral CT and 6.7 pulmonary angiograms. Hence when all these tests were available. for every pulmonary angiogram, six CT and 85 lung scintigrams were performed. The numerical information required to determine the usual sequence of investigations in suspected pulmonary embolus was insufficient to provide any meaningful data, with several respondents highlighting difficulties in acquiring this data. DISCUSSION
The present survey suggests that lung scintigraphy continues to play a major role in the diagnosis of pulmonary embolism. being performed approximately I9 times more commonly than spiral CT and over 100 times more often than pulmonary angiography. We suggest that this difference is not explained by a lack of availability of the alternatives to V/Q scintigraphy since, where all three tests were available, scintigraphy was still the most common investigation. The relatively low uptake of spiral CT and conventional pulmonary angiography is worrying. given the documented shortcomings of lung scintigraphy together with the magnitude of the diagnostic problem; it has been estimated that PE accounts for I% of UK Hospital admissions [ 121.
SllKVEY
ON THE USE OF PULMONARY
SCINTIGRAPHY.
The relatively high cost of spiral CT and conventional angiography may contribute to their under utilization. However. in one comprehensive cost-effectiveness analysis of the live most commonly used methods the best diagnostic and costeffective outcomes all included spiral CT in the imaging strategy, provided the specificity remained above 32%, (which past studies support ]3.13-161) and the sensitivity remained above 85% (a more contentious value) [ 171. There have been a number of studies reporting improved specificity following the disappointing results of V/Q in the PIOPED study: there was an overall specificity of IO% with only a minority of patients falling into the high probability ( 13%) or normal/near-normal scintigram (14%)) groups 141. Moditications to the PIOPED criteria have recently been proposed and have resulted in an improvement in the discrimnation between the intermediate and low probability scan categories [ 18-201. Our data indicate that spiral CT has so far failed to persuade clinicians to significantly alter practice. Although there is much data regarding the value of spiral CT in the diagnosis of PE [ I3- IS]. a recent review only brieHy highlighted the potential of spiral CT angiography 191. Instead. the author emphasized the value of clinical evaluation or serial venous doppler ultrasound in patients with indeterminate V/Q results when pulmonary angiography is not an option 191. The current guidelines of the British Thoracic Society state that ‘further evaluation of the technique (CT) is required before confident statements can be made about its place in the diagnosis of PE’. Spiral CT was positioned alongside pulmonary angiography following non-diagnostic echocardiography in patients with present or imminent collapse [IO]. A cautious approach to the investigation of patients with suspected PE may be justifed. Early reports of the sensitivity and speciticity of spiral CT of ( 100% and 96% respectively) in the diagnosis of PE were impressive [?I]. However, larger studies saw a decline in these tigures with sensitivities between X-958 and speciticities between 86-979 [ l3- IS]. Furthermore. when Goodman PI (I/. included subsegmental emboli in the analysis. sensitivity fell from 86% to 63% although the patient numbers were small 1211. Of greater concern are the results of a recent study. where experienced readers achieved a sensitivity of only 53% and specifcity of 97% for spiral CT with all emboli at segmental level or above [ 161. However. images were not viewed on :I work station and we do not know how well lung scintigraphy would have performed in these patients. The data in the present study highlight the continued underutilization of pulmonary angiography. Two thirds of hospitals in our survey had access to conventional pulmonary anpiogruphy: however less than I% of investigations for pulmonary embolus were pulmonary angiograms. a similar ratio to the hpure published in a 1991 survey [6] before the implications of the disappointing PIOPED results could be fully assessed [4]. Indeed in one unit. over half of patients with indeterminate V/Q findings received no further investigations with over one third of these being anticoagulated [22]. The reluctance to use pulmonary angiography may relate to its perceived dangers even though advances in catheter technology and the use of low osmolar non-ionic contrast media have resulted in a fall in the mortality and major complication rates 1331. However, these
SPIRAL Cl- AND CONVENTIONAL
PULMONARY
ANGIOGRAPHY
809
tigures are from a US centre with considerable experience in this technique. a relative rarity in the UK. Although the British Thoracic Society promotes the use of pulmonary angiography by stating a need for, ‘wider availability of pulmonary angiography, with appropriate teaching of training grades in interventional radiology’ [IO], it is impossible to provide that training if the test is rarely performed. It would appear that diagnostic perfection in pulmonary embolic disease is likely to remain an elusive goal. However. a more workable strategy to lower the number of patients left in diagnostic limbo does seem achievable. This may include a role for D-Dimer testing at the outset. Recent results of a cheap and rapid testing kit suggested a high negative predictive value, which not only substantially reduced the need for further tests, but also aided in the probability assessment of an abnormal V/Q [24,25]. Our large survey of UK NHS based practice reveals that V/Q scintipraphy remains the front-line imaging investigation in pulmonary embolus. A few centres have adopted spiral CT as the screening test in suspected PE. although the majority reserve CT for evaluation of the more problematic cases, particularly those with an abnormal chest radiograph. Pulmonary angiography continues to be a rarely performed test despite diagnostic difficulties with indeterminate V/Q results and calls from the British Thoracic Society for increased utilization. The evidence to date is not sufficiently compelling to support replacement of lung scintigruphy by spiral CT as the first line investigation for pulmonary embolism in the majority of patients. Likewise, neither the results of recent studies nor the prevailing attitudes of clinicians are likely to see CT replace pulmonary angiography as the gold standard. The publication of the large multicentre study recently conducted by the European Society of Thoracic Imaging is eagerly awaited. Of equal interest will be the response of UK radiologists and clinicians to the results of that study. REFERENCES I Godwm
JD. Webb WR. G;unw G. Ovenfors CO. Computed lomog’aphy of pulmonury rmboli~m. Am J Ro~wrg~vw/ 1980: I X:69 I-69.5. 2 Kalender WA. Seissler W. Klolz E. Vock P. Spiral volumelric CT with single-breath-hold technique, continuous rmnspor~. and continuous swarmer rotation. Ror/ido,q~ I9YO: 176: I8 I - 183. 3 Remy-Jardin M. Rrmy J. Wattinne L. Giraud F. Central pulmonary rhrombocmboli~m: diagnosis with spiral volumerric CT with the singlebrearh-hold technique - comparison with pulmonary angiography. Rrrdidog~ 1992: I X5:381 -387. 4 Value of the venlilalion/perfusion scan in ucule pulmonary embolism. Results ol’ the prospeck investigation of pulmonary embolism diagnosis (PIOPED). The PIOPED Invesligaors. JAMA lYYO:263:
‘753-2759. S Stein PD. Athunasoulis C. Alavi A YI d. Complications and validily of pulmonary angiography in sxle pulmonary embolism. Circdorior~
1992:85:462-468. 6 Coouer TJ. Hnvward MW, Hartox M. Survey on the use of pulmonary sci&mphy rind nngiography for suspected pulmonary thromboembolism in the UK. C/if1 Rdiol 1991:43:243-245. 7 Goodman LR. Lipchik RJ. Diagnosis of acuk pulmonary embolism: time for a new approach. Kndio/og,v 1996: 19925-27. 8 H;msell DM. Spiral computed romography and pulmonary embolism: current stale. C/in Kodiol 1997:S2:S7S-S81. 9 Fennerry T. The diagnosis oC pulmonary embolism. BMJ 1997314:
425-429. IO BTS Working Pany. Suspected approach. 77w,rr.v lY9752(Suppl
acute pulmonary 4):s l-74.
embolism:
a pracrical
810
CLINICAL
I I Robinson PJ. Venrilalion-perfusion lung scnnnin~ and spiral compurcd tomography of the lungs: compcling or complcmcntary mod;lliries’? 00 J NW/ Mm/ 199623: 1547-1.553. I2 Anderson FA Jr. Wheeler HB. Goldberg RJ CI nl. A population-based perspective of the hospiral incidence and case-fatality rates of deep vein thrombosis and pulmonary embolism. The Worcester DVT Study. Arch hrem Med 199l:lSl:933-938. I3 Remy-Jxdin M, Remy J. Deschildre F cr (I/. Diagnosis of pulmonary embolism with spiral CT: comparison with pulmonary angiography and scintigmphy. Rc1diolo.q 1996:200:699-706. I4 van Rossum AB. Treumiec FE. Kicft GJ. Smith SJ. Schepers Bok R. Role of spiral volumetric computed komopmphic scanning in the nssessmem of patients with clinical suspicion of pulmonary embolism and an ahnormal verltilation/perfusion lung scan. Tlwrrrr 19965 I :2328. I5 van Rossum AB, Pa~tynama PM. Ton ER PI [I/. Pulmonary embolism: vnlidntion of spiral CT angiography in I49 patients. Rot/i&~,q~ 1996:
RADIOLOGY
I8
I9
20
2I
22 23
20 I :467-470. I6
Drucker EA. Rivitz MS. Shepard JO cr ul. Acute Pulmonnry embolism: assessment of helical CT for diagnosis. Rodihgy I998:209:235241. I7 van Erkel AR. van Rossum AB. Blocm JL. Kievit J, Psltynama PM. Spiral CT anpiography for suspected pulmonnry embolism: n costeffecliveness analysis. Rcrdirhyp 199620 I :29-X.
1J
2.5
Gouschalk A. Sosm~nn HD. Coleman RE (‘I ul. Veniil;rlion-perfusi~lti scintipmphy in the PIOPED study. Pm II. Evaluation of the scintigraphic critcrin and interpretations. J Nucl Mcv/ 19933-I: I I I% I IX. Gray HW. Bcsscnt RG. Pulmomq embolism exclusion: ;I practical approach to low probability using the PIOPED data. Prospectiw Invrslip;~~ion of Pulmonary Embolism Diagnosis. Ew J NW/ Mod 1998:25:271-276. Freitas JE. Sarosi MG. N@c CC. Ycom;ma ME. Frcita AE. Juni JE. Modilied PIOPED crilrrin used in clinical prxticc. J NW/ Mcvl 1995:?6:1572-1578. Goodman LR, Curtin JJ. Mewissen MW PI (I/. Detection of pulmonary embolism in patients with unresolved clinical and scintigraphic diapnosis: helical CT wrsus angiopmphy. /\nt J R~~I~~II,~P~Io/ 1995: 164: I .x9- 1374. Murchison JT. Gown DR. Reid JH. Clinicnl utilization of the now diagnostic lung scinrigram. C/in Radio/ 1997:52:295-29X. Hudson ER, Smith TP, McDcrmort VG ~2,ctl. Pulmonary an~iogrnphy performed with iopamidol: complications in I .43-l palirnts. Rrrdio/r~~q\ 1996:198:6l-6.5. Epermayer P. Town Cl. Tumcr JG. Hcnton DC. Mcc AL. Beard MEJ. rare me;wrcUsefulness of the D-dimer, blood sas. and rcspiramry merits for excluding pulmonary emholism. 77uwr1.r 199X:53:830-834. Miller A. A fresh look al D-dimer in aspeclrd pulmonary embolism. 77l,w~ls 199853:X2 I -x22.
J.
Received:
C.
BURKILL,
30 March
J.
1999
R.
Revised:
G.
BELL,
I I May
S.
1999
I’.
G.
Accepted:
PADLEY
26 May
1999
AIM: To determine current clinical practice in the radiological diagnosis of acute pulmonary embolism and assess the use of spiral volumetric computed tomography. METHOD: A survey of 327 acute hospitals including cardiothoracic and orthopaedic tertiary referral centres was undertaken to assess current utilization of lung scintigraphy, spiral computed tomography and pulmonary angiography in the investigation of suspected pulmonary embolism. Responses were received from 215/327 (66%) centres. RESULTS: Lung scintigraphy was provided by 208 hospitals (144 on-site and 64 off-site). Spiral CT services were provided by 111 (52%) hospitals (on- or off-site), 142 (66%) units had access to angiographic facilities. Sixty-three centres out of 215 (29%) offered both on-site lung scintigraphy and spiral CT while only 41/215 (19%) hospitals were able to undertake all three tests on-site. On average, 501 perfusion (Q) or ventilation-perfusion (V/Q) scintigrams were performed per hospital per year with 26 spiral CT studies and just 4.6 pulmonary angiograms. CONCLUSION: These data suggest that lung scintigraphy is frequently the only imaging test in patients other than chest radiography, despite the large number of indeterminate results reported in most series. Burkill. G. J. C.. Bell. J. R. G.. Padley. S. P. G. (1999). Clmicrtl Rtrdinlog~ 54, 8070 1999 The Royal College of Radiologists 810. Key words: helical, lung.
pulmonary radionuclide
embolism, studies
thrombosis.
The ability of computed tomography (CT) to detect pulmonary embolus was lirst described in 1980 [I]. However. it was not until the advent of spiral CT that sufficiently rapid data acquisition with satisfactory pulmonary artery opacilication became feasible [2] and raised hopes of a significant role for spiral CT in the diagnosis of pulmonary embolism (PE). Interest in the technique was stimulated by a combination of impressive sensitivity (100%) and specificity (96%) in early studies comparing CT and conventional pulmonary angiography [3], and the acknowledged shortcomings of lung scintigraphy [4] coupled with concerns over the safety [5] and availability [6] of the posited ‘gold standard’ of pulmonary angiography. A potential advantage of CT is that it may provide an alternative explanation for the patient’s symptoms: indeed the wholesale replacement of lung scintigraphy by spiral CT has been suggested and adopted by some centres in the United States 1781. The response to this call for change has been more cautious in the United Kingdom [9-I I]. This is in part. due to the lower availability of spiral CT, but also may reflect the utility of lung ventilation-perfusion (V/Q) scintigraphy in Author for correspondence and guanntor of study: Dr S. Padlry. Dep:utmerit of Radiology. Chelsea and Westminister Hospital. 369 Fulhnm Road. London. SW IO 9NH. UK. 0009-9260/99/120807+04
$12.00/O
pulmonary
angiopraphy.
computed
tomography
(CT),
patients with a low pre-test probability and normal chest radiograph [S]. The present survey was undertaken to assess the availability and current practice in the utilization of lung V/Q, spiral CT and conventional pulmonary angiography for the diagnosis of pulmonary thromboembolism.
METHODS A questionnaire and explanatory letter were sent to superintendent radiographers at 327 hospitals in the United Kingdom and Southern Ireland. All hospitals designated as a ‘Major Unit’ in the British Association for Accident and Emergency Medicine Directory 1996 were included. Specialized orthopaedie and cardiothoracic centres as detailed in the 1997 Medical Directory were also included. A second round of questionnaires were sent to the Head of department in units not responding to the first round of questionnaires. The questionnaire was divided into two parts. Part I was designed to establish the service provision of each department. Part I1 requested information concerning the total numbers of imaging investigations related to the diagnosis of pulmonary embolus. 0
1999 The Royal College
of Radiologists
808 Table
CLINICAL
1 - Investigation
availability
and hospital On-site
Q V/Q SCT ANGIO
dasignation Off-site
facilities
facilities
Total o,,- nd
DGH n= 179
Teaching II=26
Specialist ,I= IO
Total u=2lS
DGH
Teaching
Specialist
Toral
I12 109 72 74
25 25 I7 21
7 7 6 4
I44 141 95 99
58 60 I-l 37
I I I 2
3 3 I 4
62 64 I6 43
Q = perfusion.
(63%) (61%) (40%) (41%)
(965”) (967~) (65%) (81%)
V/Q = ventilation
perfusion.
(70%) (70%) (60%) (40%)
(677r) (66%) (44%) (46%)
SCT = spiral computed
tomography.
Respondents were asked to detail the number of patients undergoing sequential radiological tests to establish a diagnosis of PE. Identification of the current first line investigation was requested, and if this had recently changed. participants were asked to record when that change had been implemented.
RESULTS
Completed questionnaires were returned by 215 (66%) hospitals. One hundred and seventy-nine district general hospitals, 26 teaching hospitals and IO specialist centres (seven orthopaedic, three cardiothoracic). Two hundred and four of the 215 hospitals provided A&E medicine, 209 orthopaedic surgery and 36 cardiothoracic surgery. The availability of each investigation, by hospital designation is shown in Table I. One hundred and forty-four out of 215 (67%) hospitals provided an on-site lung scintigraphy (Q or V/Q) service: all but three of these had access to ventilation as well as perfusion studies. A further 64 hospitals (30%) had an off-site agreement to provide lung scintigraphy. The mean annual number of V/Qs undertaken in the I56 hospitals providing tigures was 274 per institution (Table 2). Similarly, in the I34 units undertaking perfusion studies the mean number per hospital per year was 227. Amongst the 23 teaching hospitals providing numerical data a higher number of lung scintigrams were ventilation and perfusion (353 per hospital per year) as opposed to perfusion only scintigrams (163 per hospital per year). Ninety-five out of 215 (44%) hospitals had on-site spiral CT and a further I6 (7%) provided this service off-site. Ninety-nine out of 215 (46%) units were able to offer on-site conventional pulmonary angiography and an additional 43 (20%) offered angiography off-site. Table 2 - Average per year
number
of investigations
undertaken
per hospital
Investigation
DGH
Teaching
SpecialisL
All units
Q V/Q
254 (103) 271 (IZ5) I5 (69) 3.1 (77)
I63 353 53 8
72 97 100 15
227 (13-l) 274 (156) 26 (87) 4.6 (103J
CT ANGIO
RADIOLOG
Figures in parenlheses indicate Key is the same as Table I
(23) (23) (16) (22)
number
(8) (Is) (2) fJJ
of hospitals
providing
full dak
(32%) (34%) (WC) (21%) ANGIO
(4%) (4%) (4%) (8%)
= pulmonary
(30%) (30%) (IO%) (‘Kwr)
angiopraphy.
(296) (30%) (7%) (‘Olr)
DGH = District
ofl-.~Sf-sik~
206 (96%) 20s (95%) I I I (52%) 141(66%) Gener;d
Hospital.
Of the I I I hospitals which had access to spiral CT either onsite (II = 95) or off-site (II = 16). 87 (78%) provided numerical data. These hospitals undertook a mean of 26 CT examinations per hospital per year specihcally for the diagnosis of PE: teaching hospitals performed on average 53 such CT examinations per hospital per year. Numerical data on pulmonary angiography was provided by a total of I03 of the 142 centres that indicated there was access to this investigation. On average. 4.6 angiograms were performed per hospital per year for the diagnosis of acute PE. with Il.4 per hospital over the past 3 years, giving an annual mean of 3.8 per hospital. Of the 23 teaching hospitals offering pulmonary angiography, 22 provided numerical data. At these institutions a mean of 8.0 pulmonary angiograms were undertaken per hospital in the previous year with 25.3 in the previous 3 years. giving an annual mean of X.4 per hospital. Of 59 hospitals providing numerical data concerning both lung scintigraphy and spiral CT. the average number of lung scintigrams per annum was 473 and CT examinations 31, giving a ratio of I5 : I, Of the 35 hospitals providing numerical data concerning all the imaging modalities. the average number of tests performed was 570 Q or V/Q. 40 spiral CT and 6.7 pulmonary angiograms. Hence when all these tests were available. for every pulmonary angiogram, six CT and 85 lung scintigrams were performed. The numerical information required to determine the usual sequence of investigations in suspected pulmonary embolus was insufficient to provide any meaningful data, with several respondents highlighting difficulties in acquiring this data. DISCUSSION
The present survey suggests that lung scintigraphy continues to play a major role in the diagnosis of pulmonary embolism. being performed approximately I9 times more commonly than spiral CT and over 100 times more often than pulmonary angiography. We suggest that this difference is not explained by a lack of availability of the alternatives to V/Q scintigraphy since, where all three tests were available, scintigraphy was still the most common investigation. The relatively low uptake of spiral CT and conventional pulmonary angiography is worrying. given the documented shortcomings of lung scintigraphy together with the magnitude of the diagnostic problem; it has been estimated that PE accounts for I% of UK Hospital admissions [ 121.
SllKVEY
ON THE USE OF PULMONARY
SCINTIGRAPHY.
The relatively high cost of spiral CT and conventional angiography may contribute to their under utilization. However. in one comprehensive cost-effectiveness analysis of the live most commonly used methods the best diagnostic and costeffective outcomes all included spiral CT in the imaging strategy, provided the specificity remained above 32%, (which past studies support ]3.13-161) and the sensitivity remained above 85% (a more contentious value) [ 171. There have been a number of studies reporting improved specificity following the disappointing results of V/Q in the PIOPED study: there was an overall specificity of IO% with only a minority of patients falling into the high probability ( 13%) or normal/near-normal scintigram (14%)) groups 141. Moditications to the PIOPED criteria have recently been proposed and have resulted in an improvement in the discrimnation between the intermediate and low probability scan categories [ 18-201. Our data indicate that spiral CT has so far failed to persuade clinicians to significantly alter practice. Although there is much data regarding the value of spiral CT in the diagnosis of PE [ I3- IS]. a recent review only brieHy highlighted the potential of spiral CT angiography 191. Instead. the author emphasized the value of clinical evaluation or serial venous doppler ultrasound in patients with indeterminate V/Q results when pulmonary angiography is not an option 191. The current guidelines of the British Thoracic Society state that ‘further evaluation of the technique (CT) is required before confident statements can be made about its place in the diagnosis of PE’. Spiral CT was positioned alongside pulmonary angiography following non-diagnostic echocardiography in patients with present or imminent collapse [IO]. A cautious approach to the investigation of patients with suspected PE may be justifed. Early reports of the sensitivity and speciticity of spiral CT of ( 100% and 96% respectively) in the diagnosis of PE were impressive [?I]. However, larger studies saw a decline in these tigures with sensitivities between X-958 and speciticities between 86-979 [ l3- IS]. Furthermore. when Goodman PI (I/. included subsegmental emboli in the analysis. sensitivity fell from 86% to 63% although the patient numbers were small 1211. Of greater concern are the results of a recent study. where experienced readers achieved a sensitivity of only 53% and specifcity of 97% for spiral CT with all emboli at segmental level or above [ 161. However. images were not viewed on :I work station and we do not know how well lung scintigraphy would have performed in these patients. The data in the present study highlight the continued underutilization of pulmonary angiography. Two thirds of hospitals in our survey had access to conventional pulmonary anpiogruphy: however less than I% of investigations for pulmonary embolus were pulmonary angiograms. a similar ratio to the hpure published in a 1991 survey [6] before the implications of the disappointing PIOPED results could be fully assessed [4]. Indeed in one unit. over half of patients with indeterminate V/Q findings received no further investigations with over one third of these being anticoagulated [22]. The reluctance to use pulmonary angiography may relate to its perceived dangers even though advances in catheter technology and the use of low osmolar non-ionic contrast media have resulted in a fall in the mortality and major complication rates 1331. However, these
SPIRAL Cl- AND CONVENTIONAL
PULMONARY
ANGIOGRAPHY
809
tigures are from a US centre with considerable experience in this technique. a relative rarity in the UK. Although the British Thoracic Society promotes the use of pulmonary angiography by stating a need for, ‘wider availability of pulmonary angiography, with appropriate teaching of training grades in interventional radiology’ [IO], it is impossible to provide that training if the test is rarely performed. It would appear that diagnostic perfection in pulmonary embolic disease is likely to remain an elusive goal. However. a more workable strategy to lower the number of patients left in diagnostic limbo does seem achievable. This may include a role for D-Dimer testing at the outset. Recent results of a cheap and rapid testing kit suggested a high negative predictive value, which not only substantially reduced the need for further tests, but also aided in the probability assessment of an abnormal V/Q [24,25]. Our large survey of UK NHS based practice reveals that V/Q scintipraphy remains the front-line imaging investigation in pulmonary embolus. A few centres have adopted spiral CT as the screening test in suspected PE. although the majority reserve CT for evaluation of the more problematic cases, particularly those with an abnormal chest radiograph. Pulmonary angiography continues to be a rarely performed test despite diagnostic difficulties with indeterminate V/Q results and calls from the British Thoracic Society for increased utilization. The evidence to date is not sufficiently compelling to support replacement of lung scintigruphy by spiral CT as the first line investigation for pulmonary embolism in the majority of patients. Likewise, neither the results of recent studies nor the prevailing attitudes of clinicians are likely to see CT replace pulmonary angiography as the gold standard. The publication of the large multicentre study recently conducted by the European Society of Thoracic Imaging is eagerly awaited. Of equal interest will be the response of UK radiologists and clinicians to the results of that study. REFERENCES I Godwm
JD. Webb WR. G;unw G. Ovenfors CO. Computed lomog’aphy of pulmonury rmboli~m. Am J Ro~wrg~vw/ 1980: I X:69 I-69.5. 2 Kalender WA. Seissler W. Klolz E. Vock P. Spiral volumelric CT with single-breath-hold technique, continuous rmnspor~. and continuous swarmer rotation. Ror/ido,q~ I9YO: 176: I8 I - 183. 3 Remy-Jardin M. Rrmy J. Wattinne L. Giraud F. Central pulmonary rhrombocmboli~m: diagnosis with spiral volumerric CT with the singlebrearh-hold technique - comparison with pulmonary angiography. Rrrdidog~ 1992: I X5:381 -387. 4 Value of the venlilalion/perfusion scan in ucule pulmonary embolism. Results ol’ the prospeck investigation of pulmonary embolism diagnosis (PIOPED). The PIOPED Invesligaors. JAMA lYYO:263:
‘753-2759. S Stein PD. Athunasoulis C. Alavi A YI d. Complications and validily of pulmonary angiography in sxle pulmonary embolism. Circdorior~
1992:85:462-468. 6 Coouer TJ. Hnvward MW, Hartox M. Survey on the use of pulmonary sci&mphy rind nngiography for suspected pulmonary thromboembolism in the UK. C/if1 Rdiol 1991:43:243-245. 7 Goodman LR. Lipchik RJ. Diagnosis of acuk pulmonary embolism: time for a new approach. Kndio/og,v 1996: 19925-27. 8 H;msell DM. Spiral computed romography and pulmonary embolism: current stale. C/in Kodiol 1997:S2:S7S-S81. 9 Fennerry T. The diagnosis oC pulmonary embolism. BMJ 1997314:
425-429. IO BTS Working Pany. Suspected approach. 77w,rr.v lY9752(Suppl
acute pulmonary 4):s l-74.
embolism:
a pracrical
810
CLINICAL
I I Robinson PJ. Venrilalion-perfusion lung scnnnin~ and spiral compurcd tomography of the lungs: compcling or complcmcntary mod;lliries’? 00 J NW/ Mm/ 199623: 1547-1.553. I2 Anderson FA Jr. Wheeler HB. Goldberg RJ CI nl. A population-based perspective of the hospiral incidence and case-fatality rates of deep vein thrombosis and pulmonary embolism. The Worcester DVT Study. Arch hrem Med 199l:lSl:933-938. I3 Remy-Jxdin M, Remy J. Deschildre F cr (I/. Diagnosis of pulmonary embolism with spiral CT: comparison with pulmonary angiography and scintigmphy. Rc1diolo.q 1996:200:699-706. I4 van Rossum AB. Treumiec FE. Kicft GJ. Smith SJ. Schepers Bok R. Role of spiral volumetric computed komopmphic scanning in the nssessmem of patients with clinical suspicion of pulmonary embolism and an ahnormal verltilation/perfusion lung scan. Tlwrrrr 19965 I :2328. I5 van Rossum AB, Pa~tynama PM. Ton ER PI [I/. Pulmonary embolism: vnlidntion of spiral CT angiography in I49 patients. Rot/i&~,q~ 1996:
RADIOLOGY
I8
I9
20
2I
22 23
20 I :467-470. I6
Drucker EA. Rivitz MS. Shepard JO cr ul. Acute Pulmonnry embolism: assessment of helical CT for diagnosis. Rodihgy I998:209:235241. I7 van Erkel AR. van Rossum AB. Blocm JL. Kievit J, Psltynama PM. Spiral CT anpiography for suspected pulmonnry embolism: n costeffecliveness analysis. Rcrdirhyp 199620 I :29-X.
1J
2.5
Gouschalk A. Sosm~nn HD. Coleman RE (‘I ul. Veniil;rlion-perfusi~lti scintipmphy in the PIOPED study. Pm II. Evaluation of the scintigraphic critcrin and interpretations. J Nucl Mcv/ 19933-I: I I I% I IX. Gray HW. Bcsscnt RG. Pulmomq embolism exclusion: ;I practical approach to low probability using the PIOPED data. Prospectiw Invrslip;~~ion of Pulmonary Embolism Diagnosis. Ew J NW/ Mod 1998:25:271-276. Freitas JE. Sarosi MG. N@c CC. Ycom;ma ME. Frcita AE. Juni JE. Modilied PIOPED crilrrin used in clinical prxticc. J NW/ Mcvl 1995:?6:1572-1578. Goodman LR, Curtin JJ. Mewissen MW PI (I/. Detection of pulmonary embolism in patients with unresolved clinical and scintigraphic diapnosis: helical CT wrsus angiopmphy. /\nt J R~~I~~II,~P~Io/ 1995: 164: I .x9- 1374. Murchison JT. Gown DR. Reid JH. Clinicnl utilization of the now diagnostic lung scinrigram. C/in Radio/ 1997:52:295-29X. Hudson ER, Smith TP, McDcrmort VG ~2,ctl. Pulmonary an~iogrnphy performed with iopamidol: complications in I .43-l palirnts. Rrrdio/r~~q\ 1996:198:6l-6.5. Epermayer P. Town Cl. Tumcr JG. Hcnton DC. Mcc AL. Beard MEJ. rare me;wrcUsefulness of the D-dimer, blood sas. and rcspiramry merits for excluding pulmonary emholism. 77uwr1.r 199X:53:830-834. Miller A. A fresh look al D-dimer in aspeclrd pulmonary embolism. 77l,w~ls 199853:X2 I -x22.