This study has shown that over a third 34.5% of the commonly performed CT procedures were inappropriate and could have been avoided or replaced by imaging modalities that use less or no ionizing radiation. The findings are almost similar to that found by Fransico et al among adults in Emergency Departments of a non-trauma tertiary care urban hospital, where they found that the rate of inappropriate orders was 36.3% for CT [19]. Additionally, a study done by Eng et al to assess for Utility, Appropriateness of Request and Report of Computed Tomography Scan for the Diagnosis of paranasal sinuses pPathologies in a Sub-Saharan Africa Urban Setting also found the level of appropriateness to be 37.4% [20].
On the other hand, there are as well contradictory findings. Some studies have found lower levels of inappropriateness than our findings that is to say, studies by Etsehiwot et al in Ethiopia found 52/443 (11.7%) and Tambe et al in Cameroon found 71/433 (16.4%) levels of inappropriate CTRFs [21, 22]. The discrepancies could arise from the differences in methodology, practice patterns and the form of care. Settings where radiologists are available and CTRs are vetted before the procedures are performed have lower levels of inappropriateness compared to settings where CTRFs aren’t vetted. Vetting of the CT referrals prior to examination by radiologists has been shown to enable a higher level of appropriate imaging Bianco et al in a study among hospitalized patients who receive CT and MRI scans in Southern Italy where more than 95% of CT and MRI referrals are vetted by a radiologist found lower levels of inappropriate CT procedures of 22.4% [23].
Furthermore, studies that excluded CT requisitions with incomplete clinical information had lower levels of inappropriate CT procedures compared to those that included all CT requisitions inspite of the adequacy of clinical information [21]. If we combine all CT requests that were categorized as inappropriate 321 plus the 40 requests that had missing clinical information, the 38.8% (361 of the total 931) rate of inappropriate CT requestions is consistent with an audit that evaluated the appropriateness of CT and MRI, in Luxembourg by Aurélien Bouëtté et al. which found 39% inappropriate CT requisitions [24]. In both clinical settings vetting of the referrals prior to examination is often not performed. Much as Vetting has been shown to enable a higher level of appropriate imaging, shortage of radiologists in Africa makes this intervention a big challenge. A review of this cohort 972 CT requests showed that only 1.9%, were vetted by a radiologist [25].
The findings of the present study also show that inappropriate rates varied significantly (p < 0.001) across the different hospitals with Nsambya Hospital having the highest levels of inappropriateness. The cause of these differences needs further evaluation. However, possible explanation might be that this facility contributed a higher number of CTRFs compared to other facilities. Additionally, different categories of prescribers from private or public, national, regional or teaching hospitals may have different categories of patients that might cause the differences observed for appropriateness.
The level of appropriateness for CT examinations was higher (74.2%) for children than for adults. The results are consistent with a previous study in Europe by Bouëtté et al who found that the level of appropriateness was higher (82% vs. 58%; p < 0.001) when the requests concerned pediatrics than adult patients. The reason for this may be that almost all non-traumatic pediatric CT scan requests are vetted by a radiologist for justification. This is a good practice which needs to be strengthened and scaled up for adult CT procedures.
Males were almost 2 times more likely to undergo an inappropriate CT scan compared to women 193(60%) vs 128(39.9%. These findings may be explained by the fact that the commonest indication for CT scan in this study was trauma and there is risk of injuries among young adult males than females, due to risky lifestyles i.e., motorcycle riding, alcohol abuse and other occupational hazards.
From the results of this study, out of every 5 CT exams performed, 4 of them were head and this anatomical region had the highest level (78.3%) of inappropriateness. This finding may be explained by the fact that most of the indications were due to head injury resulting from trauma. This can be further highlighted by the findings of a study at the Mulago National Referral and Teaching Hospital where the incidence of admissions related to Traumatic Brain Injury was 89 per 100,000 [26]. This finding is consistent with the study by J Becker et al that assessed appropriateness of CT and MRI in South Africa where they found that majority of CT scans requested in their study were of the brain [27]. Additionally, the high level of inappropriateness for head may be explained by the fact that there are no clinical imaging referral guidelines (CIGGs) to aid decisions in the management of Traumatic Head Injury. This is further confirmed by a study conducted in the emergency department of a national referral hospital in the same setting which showed that about one-third of head CT scans currently performed on adults with mild traumatic head injury could be avoidable by applying the CCHR without very low chances of missing any important injuries [28].
The study findings also indicated that non-trauma cases were more likely to have inappropriate Head CT examinations compared to trauma causes (54.3% vs 45.6%, P < 0.001). The common indications included chronic headache with no new features and a normal neurological examination, and pediatric seizures with neurological deficits. This finding may be explained by the fact that just like other low resource settings, Uganda is challenged with diagnostics related to non-communicable diseases and so imaging is always taken where applicable[29].
The reasons for underutilization of CT scans for other anatomical regions compared to the head (20% versus 80%) need further investigation. However possible explanation could be due to the availability and accessibility of the basic imaging modalities such as ultrasound and plain radiography compared to limited accessibility to the advanced imaging technologies such as CT /MRI scans due to availability given the few numbers and inhibiting high costs.
We also found that non-contrasted CT scans were almost two times more likely to be inappropriate compared to the contrasted ones 202 (65.4%) vs 105 (34.0%). This contradicts findings in a study done in Ethiopia by Demeke at al where they found that intravenous contrast agent had been used in 80.8% of the inappropriate CT exams[21]. However, our results agree with findings by Lehnert et.al which showed the highest percentage of inappropriate CT scans to be found for head CT without contrast (62%)[30].
The possible explanation could be that most contrasted scans tend to be elective, with a possibility of being vetted by a radiologist. This is a bit encouraging and should be strengthened given the radiation doses incurred from such procedures and also the possibility of significant long-term public health problems such as radiation induced cancers and contrast induced acute kidney injury.
The appropriateness of 40 out of the 931(4.3%) of CTRFs could not be assessed using the iGuides because of insufficient clinical information. This is consistent with findings of previous studies, which registered 0–4% [31]. The process of Justification in the field of radiology highly depends on clinical information especially for procedures with multiple or complex cross section imaging, those conditions which are not included in the available guidelines and procedures associated with high radiation exposures, such as Multi-Detector Computed Tomography[32]. Assessment of the level of completeness of clinical information and vetting of CT requisition forms in the same cohort showed all the CTRFs had incomplete clinical information and only 18 (1.9%) CRFs were vetted by a radiologist.
Study strength and limitation
The strengthen of this study is the large sample size and multi-center representation of CT services in the study setting. This increased the possibility for generalization of findings.
The retrospective data collection, especially the lack of appropriate data may have distorted the actual rate of appropriateness, since it is influenced by the adequacy of clinical information. However, the study aim was to conduct a situation analysis of the current practice as it is in order to develop targeted interventions for behavior change.
The actual rates of appropriateness of Ct requisition may differ from that one elicited using CIGs /ESR-iGuide, since appropriateness rate of requests was based on the available clinical information in the requests and CIGs as a gold standard without consulting other sources of information such as patients, patient’s files.
Converting free text clinical information regarding in regard to the requested procedure to fit the CIG and be recognized by computer was a challenge.
Some valid clinical reasons for ordering CT scan did not match the coding terminology possibly due to the limited software vocabulary. The CIG could not cover the whole spectrum of clinical situations given the background disease patterns in which it was developed. However, for a start, given the limited resources required to develop CIG de nevo, adoption and adaption rather than development of guidelines is more suitable to the African region.