Recent decades have seen a high increase in the medical use of ionising radiation, of multi-slice computed tomography (CT) in particular (1,2,3,4). According to the 2006 report of the National Council on Radiation Protection and Measurements (NCRP) (5), exposure of the US general population to ionising radiation from diagnostic procedures had increased seven times since the early 1980s. Overall medical exposure, including treatment, accounted for 48 % of total exposure (background and medical), while CT alone accounted for 24 %. Between the 1980s and 2006, the estimated cumulative individual dose from all sources almost doubled (from 3.6 mSv to 6.2 mSv). Save for environmental exposure estimates (6), corresponding medical exposure data for Croatia are missing.
General practitioners (GPs) often refer patients to plain radiography, CT scans, and other forms of imaging to assist them in diagnosis and treatment. However, estimates are that half of these procedures are unnecessary and mainly owed to patient’s wishes, defensive medicine, and media influence (7). Furthermore, previous systematic reviews have shown suboptimal radiation knowledge of referring physicians (1, 8), with many of them not knowing even the basic terms and principles regarding radiation safety, like the “as low as reasonably achievable” (ALARA) principle (9, 10). Our aim was therefore to see how Croatian GPs would fare in this respect by testing their knowledge of diagnostic radiological procedures and involved radiation doses.
This study is an extension of a study investigating patients’ knowledge of radiation (11) and was approved by the Ethics Committee of the Merkur University Hospital, Zagreb (approval No. 0311–1347 of 14 February 2018).
We distributed a questionnaire to GPs from all over Croatia, most of whom worked in Zagreb, the nation’s capital. Participants received the questionnaire either by email or as a handout at a national GP conference with an accompanying letter explaining the aim of the study. Participation was anonymous, voluntary, and implied consent of those who completed the questionnaire.
This questionnaire is based on similar, previously published surveys (12 13,14). The first part covers demographics, including gender, age, specialisation, years of medical practice, and whether the participants have ever taken a radiation protection course. The second part investigates general knowledge about radiation exposure in common radiological procedures and whether the participants inform their patients about medical imaging and associated risks. Eight items require yes or no answers and eight are multiple-choice. The last question investigates specific knowledge about the exposure doses of nine imaging methods. The participants are asked to assign to each method the equivalent number of units, assuming that 1 unit corresponds to the effective dose of standard chest X-ray (0.1 mSv) (15).
Replies were evaluated by an independent panel of radiology experts. Correct answers are available at the United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) website (15). Multi-choice answers to the last question were analysed based on the article published by Wong et al. (16). Each answer was scored between 1 and 6, depending on the range corresponding to standard chest X-ray effective dose described above. If the answer was 0, the score was 1; if 1–49 (units), the score was 2, and so on. If the question was left blank or answered as „do not know“, it was scored 0.
We then constructed a radiation knowledge index (RKI), a score based on respondents’ correct answers to yes/no and multiple-choice questions. Each correct answer was awarded one point, while incorrect and missing answers received zero points.
The respondents were then divided into three groups according to their specialisation [family medicine specialists, family medicine residents, and other (no specialisation or any other specialist working as GP)], and group averages for each item compared with the right answer using the one-sample t-test. If the group average did not significantly differ from the answer, it was considered “correct”. The difference between group average and correct answers, named right answer–mean answer difference (RMD), was applied to describe the extent of answer error. In total, there were nine RMD values for each group, and the mean RMD served to estimate how wrong the answers of the specific group were.
RMD means were compared between the three groups using a paired-sample t-test. All statistical analyses were run on STATISTICA for Windows (version 14.1.0, TIBCO, Palo Alto, CA, USA).
A total of 120 GPs completed the questionnaire. Among those who received it by email (N=84), the response rate was nearly 100 %. The response rate of the rest (N=36) was 50 %.
Table 1 shows the respondents’ demographic information. Most were women and family medicine specialists and most had never attended a radiation protection course.
Sociodemographic data of participating general practitioners from Croatia (N=120)
| Mean age (and range) (years) | 44.2 (26–65) | |
| Gender* (%) | Female | 83.1 |
| Male | 16.9 | |
| Specialisation (%) | Family medicine | 57.5 |
| Family medicine residents | 23.3 | |
| Others (no specialisation or any other specialisation working as GP) | 19.2 | |
| Mean years of service (and range) | 17.1 (1–40) | |
| Have you ever taken any radiation protection course? (%) | yes | 5.8 |
| no | 94.2 |
two respondents did not specify gender
Tables 2 and 3 show the prevalences of answers to the second set of questions regarding GPs’ practice and general knowledge about radiological procedures to which they refer their patients.
Replies of Croatian GPs (N=120) to the second set of yes/no questions establishing their referral practices and general knowledge of radiological procedures (preferable/correct answers are in boldface)
| Yes | No | |
|---|---|---|
| % | ||
| Have you ever been asked by a patient to provide additional information about the radiological procedure to which you refer them? | 69.2 | 30.8 |
| Do you inform your patients sufficiently about radiological procedures to which you refer them? | 88.3 | 11.7 |
| Is informing patients necessary? | 97.5 | 2.5 |
| Can some radiological procedures irradiate patient's body? | 100.0 | |
| Can radiation harm patients' health? | 91.6 | 8.4 |
| Can single exposure to ionising radiation during a radiological procedure increase the risk of cancer? | 47.5 | 43.3 |
| Can repeated exposure to ionising radiation during radiological procedures increase the risk of cancer? | 94.2 | 4.2 |
| Is your knowledge of ionising radiation involved in radiological procedures sufficient for your daily work at the clinic? | 27.5 | 72.5 |
Replies of Croatian GPs (N=120) to the third set of multiple choice questions establishing their knowledge of radiation exposure through diagnostic imaging procedures (correct answers are in boldface)
| Question | Reply (%) |
|---|---|
| Which of the selected imaging procedures do not expose patients to ionising radiation? | |
| Plain radiography | 0.8 |
| CT scan | 1.7 |
| MRI | 97.5 |
| Angiography | 0 |
| When selecting the appropriate radiologic procedure, what is the priority? | |
| To minimise radiation dose | 27.5 |
| To address the clinical issue | 69.2 |
| To reduce diagnostic expenses | 0 |
| To avoid radiation | 2.5 |
| Don’t know | 0.8 |
| What age is most at cancer risk because of radiation exposure? | |
| 10 years | 73.3 |
| 30 years | 5.8 |
| 50 years | 5.8 |
| 70 years | 3.3 |
| Don’t know | 11.7 |
| Which of the selected imaging procedures involves highest ionising radiation doses? | |
| Plain radiography | 89.2 |
| MRI | 0 |
| Mobile phone | 7.5 |
| Ultrasound | 0 |
| Don’t know | 3.3 |
| Repeated brain CT can cause: | |
| Headaches | 8.5 |
| Cataract | 59.3 |
| Nothing | 0.8 |
| Don’t know | 31.4 |
| The lowest lethal effective dose of radiation is: | |
| 0.5 Sv | 1.7 |
| 5 Sv | 15.8 |
| 50 Sv | 11.7 |
| 500 Sv | 12.5 |
| Don’t know | 58.3 |
| The most sensitive organ to ionising radiation is: | |
| Skin | 8.3 |
| Bone marrow | 69.2 |
| Thyroid gland | 15.8 |
| Don’t know | 6.7 |
| How important it is to know about ionising radiation in diagnostic imaging | |
| Very important | 59.2 |
| Important | 37.5 |
| Less important | 3.3 |
| Unimportant | 0 |
Table 4 shows the prevalences of answers to the question about effective radiation doses involved in nine specific radiological procedures. Many GPs admitted that they did not know the answers and many provided incorrect answers, save for brain MRI. A glaring example is that as many as 20 % did not know if kidney ultrasound involved ionising radiation. In addition, although most respondents were women, only 21.7 % answered correctly about the radiation dose involved in mammography.
Replies of Croatian GPs (N=120) to the question: "If the single chest X-ray is taken as a dose unit, how much radiation does a person receives during the following procedures?" (correct answers are in boldface)
| Chest X-ray unit* (score) | |||||||
|---|---|---|---|---|---|---|---|
| Procedure | 0 (1) | 1–49 (2) | 50–99 (3) | 100–199 (4) | 200–499 (5) | ≥500 (6) | Do not know/blank (0) |
| Abdominal CT | 1.7 | 15.8 | 17.5 | 10.8 | 14.2 | 21.7 | 18.3 |
| Intravenous urography | 6.7 | 20.8 | 15.8 | 15.8 | 15.0 | 3.3 | 22.5 |
| Barium meal | 11.7 | 20.8 | 14.2 | 13.3 | 12.5 | 2.5 | 25.0 |
| Lumbar spine CT | 2.5 | 15.0 | 18.3 | 14.2 | 22.5 | 3.3 | 24.2 |
| Brain MRI | 73.3 | 1.7 | 2.5 | 0.8 | 0.0 | 0.8 | 20.8 |
| Brain MRI with contrast | 58.3 | 10.8 | 2.5 | 1.7 | 3.3 | 0.0 | 23.3 |
| Lumbar spine X-ray | 47.5 | 20.0 | 6.7 | 3.3 | 0.8 | 0.0 | 21.7 |
| Mammography | 50.8 | 21.7 | 6.7 | 2.5 | 0.0 | 0.0 | 18.3 |
| Renal US | 80.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 20.0 |
1 unit = 0.1 mSv
The radiation knowledge index of our respondents ranges between 4 and 16 (mean ± SD = 10.06±2.511) and is significantly higher in male GPs (P<0.05) (Table 5). Between family medicine specialists, residents, and others, however, it does not differ significantly (Table 6).
Differences in the radiation knowledge index (RKI) between male and female respondents (N=120)
| Gender | N | Mean rank | Sum of ranks | P* |
|---|---|---|---|---|
| Male | 20 | 75.25 | 1505.00 | 665.000; 0.023 |
| Female | 98 | 56.29 | 5516.00 | |
| Total** | 118 |
Mann-Whitney U test;
two respondents did not specify their gender and were excluded from analysis
Differences in the radiation knowledge index between specialisations (N=120)
| Specialisation | N | Mean rank | P* |
|---|---|---|---|
| Family medicine specialists | 69 | 57.33 | 1.602; 0.449 |
| Family medicine residents | 28 | 62.68 | |
| Others | 23 | 67.35 | |
| Total | 120 |
Kruskal-Wallis (H) test; Others – physicians with no specialisation or other specialisation working as GPs
Table 7 shows that these groups showed poor knowledge about radiation doses involved in various diagnostic procedures as their mean scores mostly significantly differed from the right (correct answers). The exceptions are family medicine specialists answering correctly to questions about lumbar CT and kidney ultrasound, residents answering correctly about brain MRI and kidney ultrasound, and others answering correctly about lumbar CT, brain MRI, and kidney ultrasound.
Group mean scores in the answers to the last question by specialisation among Croatian GPs (N=120)
| Abdominal CT | Intravenous urography | Barium meal | Lumbar spine CT | Brain MRI | Brain MRI with contrast | Lumbar spine X-ray | Mammography | Renal ultrasound | |
|---|---|---|---|---|---|---|---|---|---|
| Correct answer | 6 | 4 | 5 | 4 | 1 | 1 | 3 | 1 | 1 |
| Family medicine specialists | 3.95* | 3.5* | 3.1* | 3.76 | 1.17* | 1.49* | 1.73* | 1.51* | 1 |
| Family medicine residents | 3.77* | 2.77* | 2.68* | 3.32* | 1.21 | 1.5* | 1.3* | 1.61* | 1 |
| Others | 4.57* | 3.26* | 3.21* | 3.75 | 1.11 | 1.26 | 1.6* | 1.45* | 1 |
Significant right answer – mean answer difference (RMD), i.e., significant difference between mean score and correct answer maximum score (P<0.05); Others – physicians with no specialisation or other specialisation working as GPs; CT – computed tomography; MRI – magnetic resonance imaging
The association between the RKI and years of service was not statistically significant (Spearman’s rho=−1.065, P=0.074) for the entire sample or for each group (Kruskal-Wallis H=6.539, P=0.478).
Considering that the distribution of RMD values did not deviate significantly from normal (Table 8), we compared the arithmetic means of RMDs between the groups using a paired-sample t-test, which reveals significant differences between family medicine specialists and residents as well as between residents and others (Table 9).
Test of distribution normality for RMD values for the three specialisations
| RMD | Kolmogorov-Smirnova | Shapiro-Wilk | ||||
|---|---|---|---|---|---|---|
| Statistic | df | Significance | Statistic | df | Significance | |
| Family medicine specialists | 0.312 | 9 | 0.012 | 0.844 | 9 | 0.065 |
| Family medicine residents | 0.224 | 9 | 0.200 | 0.910 | 9 | 0.317 |
| Others* | 0.211 | 9 | 0.200 | 0.878 | 9 | 0.151 |
physicians with no specialisation or other specialisation working as GPs;
Lilliefors significance correction; RMD – Right answer - mean answer difference
Pairwise comparison of RMD scores between specialisation groups
| Paired differences | t | df | Significance (two-tailed) | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Mean | SD | SE | 95 % Confidence interval of the difference | ||||||
| Lower | Upper | ||||||||
| Pair 1 | Family medicine specialists vs residents | −0.261 | 0.255 | 0.085 | −0.457 | −0.065 | −3.078 | 8 | 0.015 |
| Pair 2 | Family medicine specialists vs others | 0.078 | 0.244 | 0.081 | −0.109 | 0.265 | 0.956 | 8 | 0.367 |
| Pair 3 | Residents vs others | 0.339 | 0.249 | 0.083 | 0.148 | 0.530 | 4.085 | 8 | 0.004 |
Values in boldface denote significant difference between pairs (P<0.05); Others – physicians with no specialisation or other specialisation working as GPs; RMD – Right answer - mean answer difference; SD – standard deviation; SE – standard error of the mean
Our study highlights a disconcerting lack of knowledge among GPs regarding radiation exposure from diagnostic imaging, a finding consistent with previous research. Most GPs had not attended radiation protection courses, mirroring the study by Willoughby et al. (7), where only 28 % of GPs had prior training. Research has shown that such training significantly improves physicians’ ability to estimate radiation risks and doses (17, 18).
A key issue is the limited awareness of guidelines for radiological referrals. Many physicians are unfamiliar with the ALARA principle (3, 9, 10, 19 20, 21, 22), and studies indicate that adherence to referral guidelines can significantly reduce unnecessary imaging (21).
Even so, most of our GPs reported a higher rate of patient communication informing patients about radiographic procedures than other specialties reported by Lumbreras et al. (19), who also noted that prior education on radiation exposure increased the likelihood of discussing potential risks with patients. Lam et al. (1) reported that while patients preferred discussing risks with their referring physician, more than half of physicians felt this responsibility should fall on radiologists.
Considering the factors influencing referral decisions, most respondents reported the clinical issue as the key factor in selecting imaging modalities, with only a minority prioritising minimal exposure. This is in line with the report by Borgen et al. (3), who found that GPs and hospital physicians valued radiation dose less than the diagnostic benefit of imaging. In fact, 88.3 % of GPs in this study referred patients for imaging that was unlikely to alter treatment, largely due to patient expectations. In some settings, overuse of imaging was owed to considerations such as workload pressure and wish to “avoid a lawsuit” (20).
Although all participants acknowledged that some radiological procedures involve radiation exposure, 8.4 % disagreed that radiation could be harmful. Some even believed that mobile phones posed greater risks than plain (X-ray) radiography. Previous studies have reported a similar lack of awareness, with many physicians unable to tell the difference between deterministic and stochastic biological effects of radiation (3, 4).
Another critical gap was in recognising which imaging modalities use ionising radiation. While MRI was correctly identified as non-ionising by 97.5 % of respondents, some mistakenly believed CT (1.7 %) and X-ray (0.8 %) were non-ionising. Previous studies have found that 1–35 % of physicians misclassified MRI, and 0.5–24 % misidentified ultrasound as a radiation source (1, 2, 3, 7, 8, 10, 17, 23).
In response to the last question, our GPs struggled to correctly estimate radiation doses of various imaging procedures; many did not know the answer, and a great majority answered incorrectly, save for brain MRI, which most correctly recognised as non-irradiating procedure. In fact, an overwhelming majority underestimated the effective doses of the imaging procedures involving ionising radiation. This is probably the key finding of our study, and it confirms earlier findings that underestimation of effective doses is widespread among physicians regardless of specialisation (3, 7, 8, 17, 23,24 25,26).
The pairwise comparison of mean RMDs shows that residents performed worse in radiation dose estimation than family medicine specialists and other physicians (Table 9; P<0.015 and P<0.004, respectively), yet did not differ from these two groups significantly in the radiation knowledge index, most likely because RKI includes not only the last question on specific doses for each imaging procedure. We find this finding somewhat surprising, considering that residents had finished their medical school more recently than those with completed specialisations and should be better acquainted with updated information. However, this finding confirms previous research indicating that medical students have substantial knowledge deficits in this area ( 27,28,29). In contrast, greater clinical experience appears to contribute to improved understanding over time.
This study has limitations, particularly regarding the sampling methods. The initial snowball sampling may have introduced selection bias, and the 50 % response rate in the second phase could reflect nonresponse bias. Additionally, while a gender-based difference in radiation knowledge was observed, the study design was not suited to explore the underlying reasons. Future research should investigate factors contributing to these disparities and evaluate the effectiveness of targeted educational interventions for improving physicians’ radiation knowledge and referral practices.
This cross-sectional study clearly reveals major gaps in knowledge about radiological procedures among Croatian general practitioners, which seems to be consistent with other countries. One of the reasons is the lack of continuing professional development in this respect, that is, a lack of radiation courses that would fill these gaps and refresh their memories. We therefore urge for additional pre- and post-graduate radiological training and utilising tools such as referring guidelines to improve the current situation. Furthermore, more nuanced studies are needed to identify specific gaps that need addressing.