AM Speets, Y van der Graaf, AW Hoes, et al

Chest radiography in general practice:

indications, diagnostic yield and consequences
for patient management
Anouk M Speets, Yolanda van der Graaf, Arno W Hoes, Sandra Kalmijn, Alfred PE Sachs, Matthieu JCM Rutten,
Jan Willem C Gratama, Alexander D Montauban van Swijndregt and Willem PThM Mali

INTRODUCTION
ABSTRACT Chest radiography (CXR) is an important diagnostic
Background method for evaluation of the airways, pulmonary
Chest radiography (CXR) is frequently performed in parenchyma and vessels, mediastinum, heart, pleura
Western societies. There is insufficient knowledge of its and chest wall.1 It is one of the most widely used
diagnostic value in terms of changes in patient
diagnostic imaging techniques in Western societies;
management decisions in primary care.
on average 236 CXRs per 1000 patients per year are
Aim
To assess the influence of CXR on patient management performed and this technique accounts for 25% of
in general practice. the annual total numbers of diagnostic imaging
Design of study procedures.2 In the Netherlands, annually
Prospective cohort study. approximately 500 000 CXRs are requested by GPs.3
Setting The frequency with which even relatively
Seventy-eight GPs and three general hospitals in the inexpensive and non-invasive diagnostic tests are
Netherlands. performed leads to high costs in health care.
Method Unnecessary diagnostic investigations may lead to
Patients (n = 792) aged ≥18 years referred by their GPs
incidental findings, or to additional unnecessary
for CXR were included. The main outcome was change
in patient management assessed by means of
diagnostic procedures or even over treatment.
questionnaires filled in by GPs before and after CXR. Current guidelines for CXR are aimed mainly at
Results diseases instead of at the complaints with which
Mean age of the patients was 57.3±16.2 years and patients present themselves, and even lacking in the
53% were male. Clinically relevant abnormalities were Netherlands.1,4–10 We are aware of only few studies on
found in 24% of the CXRs. Patient management CXR in patients referred by GPs. Geitung et al11
changed in 60% of the patients following CXR. Main
concluded that the clinical utility of CXR was high
changes included: fewer referrals to a medical
specialist (from 26 to 12%); reduction in initiation or enough to justify its costs, and Lim et al12 showed
change in therapy (from 24 to 15%); and more frequent that GPs do act on results of abnormal CXRs. The
reassurance (from 25 to 46%). However, this
reassurance was not perceived as such in a quarter of
these patients. A change in patient management AM Speets, PhD, MSc; WPThM Mali, MD, PhD, Department
occurred significantly more frequently in patients with of Radiology; Y van der Graaf, MD, PhD; AW Hoes, MD,
complaints of cough (67%), those who exhibited PhD; S Kalmijn, MD, PhD; APE Sachs, MD, PhD, Julius
abnormalities during physical examination (69%), or Centre for Health Sciences and Primary Care, University
those with a suspected diagnosis of pneumonia (68%). Medical Centre Utrecht, Utrecht, The Netherlands. MJCM
Rutten, MD, Department of Radiology, Jeroen Bosch Hospital,
Conclusion
‘s-Hertogenbosch, The Netherlands. JWC Gratama, MD, PhD,
Patient management by the GP changed in 60% of
Department of Radiology, Gelre Hospitals, Apeldoorn, The
patients following CXR. CXR substantially reduced the
Netherlands. AD Montauban van Swijndregt, MD, PhD,
number of referrals and initiation or change in therapy,
Department of Radiology, ‘Onze Lieve Vrouwe Gasthuis’,
and more patients were reassured by their GP. Thus,
Amsterdam, The Netherlands.
CXR is an important diagnostic tool for GPs and seems
a cost-effective diagnostic test.
Address for correspondence
Keywords Anouk Speets, Department of Radiology (E01.335),
chest radiography; general practice; patient care University Medical Centre Utrecht, PO Box 85500, 3508 GA,
management. Utrecht, The Netherlands. E-mail: aspeets@umcutrecht.nl

Submitted: 4 November 2005; Editor’s response:

24 April 2006; final acceptance: 8 June 2006.
©British Journal of General Practice 2006; 56: 574–578.

574 British Journal of General Practice, August 2006

 Original Papers

studies of Guyer et al13 and Keogan et al14 reported

clinically relevant abnormalities in 21 and 23% of
patients referred for CXR by GPs, respectively. How this fits in
Clearly, the full value of CXR cannot be assessed in To our knowledge, this prospective cohort study is the first study that has
terms of positive findings alone. The relevance of assessed the influence of both positive and negative findings of chest
detected abnormalities must be assessed with radiography (CXR) on the change in patient management in general practice
and evaluated the consequences of the CXR according to the patient. CXR led
respect to clinical practice, because positive
to changes in patient management in 60% of the patients referred by GPs,
findings may be incidental and without any
which is one of the prerequisites for successfully influencing clinically relevant
consequences. Negative examinations can also patient outcomes. CXR resulted in fewer referrals to a medical specialist, a
have potential value when they result in changes of reduction in the number of patients with initiation or change in therapy, and
patient management and can be very helpful in more frequent reassurance of the patient. CXR is an important diagnostic tool
reassuring the patient. Neither of these studies cited for GPs and seems a cost-effective diagnostic test.
both positive and negative findings in detail, nor
assessed the value of CXR in terms of changes in
patient management. day. In general all CXRs are reported by a
The objective of this study was to assess the radiologist within 24 hours. Any significant
influence of both positive and negative findings of abnormalities will be verbally reported to the GP,
CXR on the change in patient management in before the official radiologic report is sent by mail.
general practice and to evaluate the consequences Therefore, significant abnormalities will normally be
of the CXR according to the patient. received by GPs within 1 day, and they directly can
adjust their patient management plan. When no
METHOD significant pathology is detected with CXR, it can
Participants take up to 4 days before the GP receives the official
This prospective cohort study was conducted from radiologic report. After the GP received the report,
April 2003 to December 2004. In total, 78 GPs in the he or she filled in a second questionnaire; again
catchment area of one of three participating general including the suspected diagnosis and anticipated
hospitals located in three main cities in the patient management plan.
Netherlands (Jeroen Bosch Hospital in The reports of CXR were collected in the three
‘s-Hertogenbosch; Gelre Hospitals in Apeldoorn; hospitals to determine the findings of CXR. These
‘Onze Lieve Vrouwe Gasthuis’ in Amsterdam) were findings were categorised into six groups (the first
involved. Twenty-eight GPs (36%) worked in a solo four groups were considered clinically relevant
practice; 58 (74%) were male, and 40 GPs (51%) abnormalities):
graduated between 1968–1980, 19 (24%) between
1980–1990 and 19 (24%) between 1990 and 1997. • Malignancy;
All patients aged 18 years and older who were • Pneumonia;
referred for CXR (standard posteroanterior and lateral • COPD/asthma/chronic bronchitis;
view) by their GP to one of these hospitals were • Other clinically relevant abnormalities (heart failure
included in the study. The patients received an and unclear abnormalities that required further
exclusion form from their GP, which they could return investigation according to the radiologist);
to the study coordinator if we were not allowed to • The follow-up of abnormalities detected previously
use their data for this study. on CXR;
• No abnormality.
Measurements
All GPs were asked to fill in a standardised form Six months after the CXR a short questionnaire
before requesting a CXR, including information on was sent to all patients, in order to assess the
history, physical examination, indication, suspected consequences of CXR according to the patient
diagnosis, and proposed patient management. The (response rate = 79%). They could choose one of the
anticipated patient management was filled in as if following options: definite diagnosis; better
no CXR would be performed. The management treatment; reassurance; nothing; or other. With this
options included: referral to a medical specialist; information we could check whether reassurance of
initiation or change in therapy; reassurance of the the patient as reported by the GP was really
patient; and follow-up by the GP (watchful waiting perceived as reassurance by the patient.
or additional diagnostic testing). The GP could
choose only one of these management options. Statistical analysis
After the GP requests a CXR a patient can be The primary outcome measure for our study was the
referred for CXR to the general hospital at the same proportion of patients in whom there was a change in

British Journal of General Practice, August 2006 575

 AM Speets, Y van der Graaf, AW Hoes, et al

suspected diagnosis was pneumonia (24%) and

Table 1. Patient characteristics (n = 792). malignancy (18%) (Table 1).
The radiology reports of CXR showed no
n (%)
abnormality in 416 patients (53%) and follow-up of an
Mean ±SD in years 57.3±16.2
abnormality detected previously on CXR in 179
Sex
patients (23%). Clinically relevant abnormalities were
Male 423 (53)
Female 369 (47)
found in 197 CXRs (25%), these included: malignancy
(n = 11; 1%); pneumonia (n = 44; 5%);
Prior diagnoses
Malignancy (various locations n = 29; lung n = 9) 38 (5) COPD/asthma/chronic bronchitis (n = 99; 13%); and
Cardiovascular 95 (12) other clinically relevant abnormalities that required
Pneumonia 76 (10) further investigation according to the radiologist
COPD/asthma/chronic bronchitis 143 (18) (n = 43; 5%). As expected, all patients with a
History taking malignancy were referred to medical specialists after
Smoking 142 (18)
CXR, with the exception of one patient, in this case
Pain 172 (22)
Respiratory complaints
the GP wanted to wait for the results of the additional
Haemoptysis 53 (7) CT-scan before further action. Patients with
Cough 394 (50) pneumonia were mainly treated by the GP with a
Dyspnoea 199 (25) prescription of antibiotics. Noticably, 29 patients (4%)
Other symptoms of respiratory infectiona 117 (15)
with no abnormalities detected on CXR were referred
General complaints
Weight loss 31 (4)
to a medical specialist. Fifteen patients had unclear
Fever 53 (7) complaints that needed further examination, in nine
General malaise 101 (13) patients lung pathology was excluded and these
Abnormalities during physical examinationb 317 (40) patients were referred to another medical specialist,
Suspected diagnosis pneumonia according to GP 193 (24) such as a cardiologist. In four patients a clinically
Suspected diagnosis malignancy according to GP 142 (18) relevant abnormality was found with another
a examination (such as abdominal ultrasound), and
Abnormal sputum, nasal congestion, throat symptoms, and complaints of a cold.bA
physical examination was considered abnormal when abnormalities were detected with CXR was used as a screening tool in one patient.
auscultation (such as wheeze), percussion (such as dullness), or palpation (such as pain). The proportion of patients in whom CXR resulted
SD = standard deviation.
in a change in patient management was 60% (95%
CI = 57 to 64%). Main changes in patient
patient management by the GP following CXR. This management plans after CXR included: a reduction
proportion and the corresponding 95% confidence in anticipated referrals to a medical specialist from
interval (CI) were calculated using the statistical 203 (26%) to 97 (12%); a reduction in the number of
program Confidence Interval Analysis.15 Additionally, patients with initiation or change in therapy from 187
subgroup analyses were performed to assess (24%) to 119 (15%), which was demonstrated mainly
whether the patient and GP characteristics by a reduction in the anticipated prescription of
influenced the proportion of change in patient drugs such as antibiotics; and more frequent
management. Associations were tested with χ2 tests reassurance of the patient, from 195 (25%) to 363
and regarded as significant when the P-value was (46%) patients (Table 2).
≤0.05. Data were analysed using SPSS for Windows Subgroup analyses revealed that the proportion
version 11.0. of patients in whom patient management changed
after CXR was significantly higher among patients
RESULTS who complained of cough (67%), who exhibited
In total, 870 patients aged 18 years or older were abnormalities during physical examination (69%) or
referred for CXR. Patient management plans for 78 had a suspected diagnosis of pneumonia (68%)
patients (9%) were not filled in by the GP before (Table 3). The characteristics of the GPs (solo or
and/or after CXR. These patients were excluded from group practice, sex and year of graduation) had
the study, resulting in a study population of 792 little influence on the proportion of change in
patients. Their patient characteristics were management of 60%.
comparable with the included patients. Almost one-fifth of the patients who returned the
The mean age of the patients at time of CXR was questionnaire reported that CXR had no
57.3 years (standard deviation = 16.2) and 53% consequences, and approximately 50% of the
were male. Fifty per cent of the patients had a patients were reassured after CXR. It was noted
history of cough and 25% of dyspnoea. that a quarter of the 363 patients who were
Abnormalities with physical examination were found reportedly reassured by their GP after CXR failed to
in 40% of the patients. The most common perceive the result of the CXR as reassurance.

576 British Journal of General Practice, August 2006

 Original Papers

Table 2. Patient management plans for GPs before and after chest radiography n (%).
Before (n)
After n (%; 95% CI) Referral medical specialist Therapeutic management Reassurance Follow-up by GPa Total
Referral medical specialist 48 22 8 19 97b
(24; 18 to 30) (12; 8 to 17) (4; 2 to 8) (9; 6 to 14) (12; 10 to 15)
Therapeutic management 29 45 15 30 119b
(14; 10 to 20) (24; 18 to 31) (8; 5 to 12) (14; 10 to 20) (15; 13 to 18)
Reassurance 89 57 141 76 363b
(44; 37 to 51) (30; 24 to 37) (72; 66 to 78) (37; 30 to 43) (46; 42 to 49)
Follow-up by GPa 37 63 31 82 213
(18; 14 to 24) (34; 27 to 41) (16; 11 to 22) (40; 33 to 46) (27; 24 to 30)
Total 203 187 195 207 792
(26; 23 to 29) (24; 21 to 27) (25; 22 to 28) (26; 23 to 29)
a
Follow-up by GP: predominantly watchful waiting or additional diagnostic testing, such as spirometry or laboratory investigation. bThe differences in proportions
of patient management after chest radiography were significant with a P-value ≤0.05 (95% confidence intervals did not overlap).

DISCUSSION performed. This could result in an overestimation of

Summary of main findings intended referrals to medical specialists. This study
The proportion of patients for whom patient does not prove that the patient actually benefits from
management changed following CXR was 60%. the diagnostic procedure, such as in terms of
Main changes included: fewer referrals to a medical morbidity, mortality or quality of life. However, the
specialist (from 26 to 12%); a reduction in the study is the first to show that the procedure often leads
number of patients with initiation or change in to changes in patient management, which is one of the
therapy (from 24 to 15%), especially fewer prerequisites for successfully influencing clinically
prescriptions of drugs such as antibiotics; and more relevant patient outcomes. Finally, the interval of
frequent reassurance of the patient (from 25 to 46%). 6 months between the short questionnaire after CXR
Subgroup analyses revealed that the proportion of
patients in whom patient management changed after Table 3. Proportion of changes in patient management after
chest radiography in relevant subgroups.
CXR was significantly higher among patients with
complaints of cough (67%), those who exhibited Change in management
abnormalities during physical examination (69%) or n % P-value
those with a suspected diagnosis of pneumonia (68%). All patients 476 60
Age
Strengths and the limitations of the study <60 years 249 59 0.56
≥60 years 227 61
Primary care patients referred for CXR have a broad
range of complaints, the indications for the tests vary Sex
Male 260 61 0.40
widely, and many different diseases can be detected
Female 216 59
with CXR. Both a cross-sectional design or
Prior diagnoses
randomised controlled trial were not feasible, for Malignancy
reasons such as absence of a valid reference test, (various locations n = 20; lung n = 8) 28 74 0.080
complex logistics, the need for large sample sizes to Pneumonia 32 50 0.10
show statistically significant differences, and ethical COPD/asthma/chronic bronchitis 90 63 0.44
considerations. A prospective cohort study provided a History taking
pragmatic and valid way to assess the effectiveness of Smoking 93 66 0.15
Haemoptysis 32 60 0.97
CXR, however some limitations must be mentioned.
Cough 264 67 <0.001
An important limitation of this type of study is the fact Dyspnoea 129 65 0.12
that the GPs assess the value of the diagnostic test by Fever 36 68 0.23
filling in the patient management before and after the Abnormalities during
test. In this way GPs may influence the estimated value physical examinationa 219 69 <0.001
of a diagnostic test, because most GPs would be keen Suspected diagnosis 132 68 0.007
to present their requested test as having some value. pneumonia according to GP
Besides, it was impossible to verify whether or not the Suspected diagnosis 91 64 0.29
GP really would have conducted the anticipated malignancy according to GP
a
patient management in accordance with the plan A physical examination was considered abnormal when abnormalities were detected with
made on the standardised form before CXR was auscultation (such as wheeze), percussion (such as dullness), or palpation (such as pain).

British Journal of General Practice, August 2006 577

 AM Speets, Y van der Graaf, AW Hoes, et al

could have influenced the accurateness of recalling Implications for clinical practice
and reporting by the patients. In conclusion, the GP’s patient management strategy
was changed for 60% of patients following CXR.
Comparison with existing literature CXR substantially reduced the number of referrals to
To our knowledge this is the first study that has a medical specialist and initiation or change in
investigated the influence of CXR on patient therapy, and more patients were reassured by their
management in general practice. Our results are in line GP. Thus, CXR is an important diagnostic tool for
with the conclusion of Geitung et al11 that the clinical GPs and seems a cost-effective diagnostic test.
utility of CXR was high enough to justify its costs after
performing a study in 55 patients in general practice. Ethics committee
The studies of Guyer et al13 and Keogan et al14 reported This study was approved by the Medical Ethics Review
Board of the University Medical Centre Utrecht (02/121)
clinically relevant abnormalities in 21% of 1.163 and
Competing interests
23% of 2.017 patients referred for CXR by GPs, None
respectively. The 24% clinically relevant abnormalities Acknowledgements
found in our study is comparable. In addition, our study It would not have been possible to conduct this study without
the participation of all GPs from the catchment areas of the
showed that the full value of CXR cannot be assessed three hospitals. We wish to thank the three trial nurses, Han
in terms of positive findings alone. Negative findings de Koning working in the Jeroen Bosch Hospital in ‘s-
are important for exclusion of diseases and, therefore, Hertogenbosch, Ireen Brussee from the Gelre Hospitals in
Apeldoorn and Cecil Kressenhof from the ‘Onze Lieve Vrouwe
for reassurance of the patient. However, such findings Gasthuis’ in Amsterdam, for their help with all the logistics in
can also result either in referral of patients to a medical the three hospitals. We thank Cees Haaring from the
University Medical Centre Utrecht for making the database
specialist for further evaluation of their complaints and his assistance with the data management. Finally, we
when a CXR fails to show any abnormalities, or in the wish to thank Christina Hooper for the revision of this article.
referral of patients to another medical specialist, such
as a cardiologist, when lung pathology is excluded. REFERENCES
1. American College of Radiology. ACR standard for the performance of
The changes in GPs’ patient management plans pediatric and adult chest radiography. American College of Radiology:
after CXR in patients with a higher proportion of United States, 2001.
change in patient management (that is, cough, 2. United Nations Scientific Committee on the Effects of Atomic
Radiation. Medical radiation exposures. United Nations Scientific
exhibited abnormalities during physical examination or Committee on the Effects of Atomic Radiation UNSCEAR 2000 Report
a suspected diagnosis of pneumonia) were fewer to the General Assembly, with Scientific Annexes. Sources and effects of
ionizing radiation, Volume I: Sources. New York: United Nations, 2000.
anticipated referrals to a medical specialist, a reduction
3. Speets AM, Kalmijn S, Hoes AW, et al. Frequency of chest radiography
in the number of patients with initiation or change in and abdominal ultrasound in the Netherlands: 1999–2003. Eur J
therapy and more frequent reassurance of the patient. Epidemiol 2005; 20: 1031–1036.
4. European Commission. Referral guidelines for imaging. Radiation
It is widely known that thorough history taking and protection 118. Luxembourg: Office for Official Publications of the
physical examination before commencement of a more European Communities, 2000.
advanced workup, such as a radiological examination, 5. Coblentz CL, Matzinger F, Samson LM, et al. CAR standards for chest
radiography. Canada: Canadian Association of Radiologists, 2000.
is very important. This study showed that even after a
6. Health Services Utilization and Research Commission. Chest
history and physical examination of the patient the radiography: a summary of the evidence supporting selective clinical
influence of CXR on patient management was practice guidelines and recommendations for implementation. Saskatoon:
Health Services Utilization and Research Commission, 1997.
substantial. We expected that the ability of GPs to
7. Stolberg HO, Buckley N, Coblentz CL, et al. Guidelines for chest X-rays
establish a more specific patient management plan in asymptomatic populations. Ontario: The College of Physicians and
Surgeons of Ontario, 1999.
after gaining detailed information of the patient with
8. World Health Organization Scientific Group. Chest and cardiovascular
physical examination would result in a smaller system. Effective choices for diagnostic imaging in clinical practice.
proportion change in management after CXR. However Geneva: World Health Organization, 1990.
the proportion of change in patient management 9. Bearcroft PW, Small JH, Flower CD. Chest radiography guidelines for
general practitioners: a practical approach. Clin Radiol 1994; 49: 56–58.
increased to almost 70% in patients with abnormalities
10. Burbridge B, Douglas D, Kriegler S. Chest X-ray ordering related to
detected during physical examination. varied clinical scenarios: a survey of Saskatchewan physicians. Can
Almost 80% of the questionnaires were returned Assoc Radiol J 2005; 56: 219–224.
11. Geitung JT, Skjaerstad LM, Gothlin JH. Clinical utility of chest
by the patients, which increased the validity of these roentgenograms. Eur Radiol 1999; 9: 721–723.
results. Approximately 50% of patients were 12. Lim WS, Macfarlane JT, Deegan PC, et al. How do general
reassured by their GP after CXR. Our study showed practitioners respond to reports of abnormal chest X-rays? J R Soc
Med 1999; 92: 446–449.
that in almost one-quarter of the patients who were
13. Guyer PB, Chalmers AG. Chest radiography for general practitioners-
reassured by their GP after CXR, the patient did not a low yield investigation. J R Coll Gen Pract 1983; 33: 477–479.
perceive this as reassurance. Therefore, CXR did not 14. Keogan MT, Padhani AR, Flower CD. Chest radiography for general
practitioners: scope for change? Clin Radiol 1992; 46: 51–54.
have much value for these patients, because no
15. Altman DG, Machin D, Bryant TN, Gardner MJ. Statistics with
referral or treatment followed after the radiological confidence: confidence intervals and statistical guidelines. London: BMJ
investigation and reassurance was not achieved. Books, 2000.

578 British Journal of General Practice, August 2006