The number of interventional radiology procedures (IRPs) performed under fluoroscopic guidance is increasing globally.¹ This is mainly because their execution is less invasive than traditional open surgical techniques. Other advantages of IRPs include shorter recovery periods and reduced hospital stays compared to surgical procedures.² The IRPs are typically performed in an angiography suite, an interventional theatre fitted with a fixed X-ray fluoroscopy machine that guides diagnostic and interventional procedures.^3,4 The advantage of fluoroscopy over other imaging modalities is its ability to acquire real-time images with a high temporal resolution.⁵

The IRPs performed by radiologists in the angiography suite can be categorised as interventional neuroradiology (INR) and body interventional radiology (IR). The INR procedures include diagnostic cerebral angiography, cerebral aneurysm, arteriovenous fistula and arteriovenous malformation embolisation.⁶ The IR procedures can be further divided into vascular and non-vascular procedures. Vascular procedures are vessel angioplasty, stenting and solid organ embolisation (i.e., spleen, liver or kidney). Non-vascular body IR procedures encompass percutaneous transhepatic cholangiography, biliary stenting, percutaneous transhepatic biliary drainage (PTBD), ureteric stenting, nephrostomy and pigtail catheter drainage for various conditions.⁷

Fluoroscopy-guided IR procedures may take hours to complete, depending on the complexity of the case.⁶ Because of the need for cine acquisition and prolonged screening, fluoroscopic assessments may inadvertently expose patients to elevated radiation doses, thereby augmenting the risk of eliciting tissue reactions (formerly known as deterministic effects).^1,5 Examples of tissue reactions are skin injuries (i.e., erythema, ulcer or necrosis)⁸ and eye lens injuries (i.e., cataract) in patients who underwent head IRPs.⁹ Interventionalists also face a risk of developing cataracts in the eye.¹⁰ While tissue reactions occur above a certain radiation dose threshold, stochastic effects may be triggered by any amount of radiation exposure (no threshold).^1,11 They include leukaemia (except chronic lymphocytic leukaemia) and radiation-related malignancies (i.e., carcinoma of the thyroid, breast, lung, bowel, bone and skin).^12,13 Consequently, emphasis must be placed on IRPs to minimise radiation dose to the patients.¹⁴ Effective radiation management is achieved through optimisation, justification and dose limitation while maintaining the integrity and quality of the image.¹⁵

The utilisation of diagnostic reference levels (DRLs) for IRPs optimisation purposes was conceptualised by the International Commission on Radiological Protection (ICRP).¹⁵ Diagnostic reference levels are notional values specific to a particular investigation that can help identify exposures that are extraordinarily low or high.¹⁵ According to ICRP publication 135, the median of the distribution of DRL values for a specific procedure at a facility is termed the typical value. These values are used when a centre is too small to establish local diagnostic reference levels (LDRLs).¹⁵ Local diagnostic reference levels are the third quartile (75th percentile) of median values for X-ray rooms at a facility or a few facilities in a local area.¹⁵ The national diagnostic reference levels (NDRLs) are the third quartile of median values for a representative sample of X-ray facilities across an entire country, whereas the regional DRLs pertain to a representative sample of the NDRLs of several countries in a geographical region.¹⁵

The necessary radiation dose parameters for determining DRL are fluoroscopy time (FT), reference air kerma (K_a,r) and dose area product (DAP) or kerma area product (KAP).^5,16 These parameters are automatically recorded by the fluoroscopy unit after each procedure.^1,5 Local diagnostic reference level is indicated by the third quartile (75th percentile) of the dataset for the K_a,r, KAP and FT of every fluoroscopically guided IRP. The DRL values should not be exceeded in the context of prudent, routine practice and can be applied at the local, national or regional levels.¹

The Directorate of Radiation Control (DRC) instituted DRLs as a legal obligation in 2015 and required institutions to develop centre-wide LDRLs for specific procedures.¹⁴ Subsequently, X-ray machine licensees were mandated to appoint a physicist (medical) to develop and enforce a programme to optimise DRLs for all fluoroscopy-guided IRPs.^3,14 The DRC regulations state that LDRLs are also necessary to establish NDRLs. Despite these requirements, only a few centres have established and published LDRLs for IRPs.¹ The absence of DRLs may inadvertently lead to high radiation exposure for patients with no recourse.⁵

The available DRL data predominantly originate from high-income countries according to the World Bank criteria.¹⁶ However, these data do not adequately represent low-income and lower-middle-income countries, where over 80% of the global population resides. The World Bank classified South Africa as a lower-middle-income country.¹⁶ Published literature on DRLs in sub-Saharan Africa is markedly limited when compared with that in high-income countries. In 2015, fewer than 21% of the 135 lower-middle-income countries published DRLs for fluoroscopy-guided procedures.^1,16

The NDRLs are yet to be established in South Africa. Their establishment depends on the comprehensive development of LDRLs of individual centres throughout all nine provinces.^1,14 A review conducted in 2021 regarding South Africa’s DRL data revealed that only three (Gauteng, Free State, and Western Cape) out of the nine provinces published any DRL information.¹⁴ Thus, comprehensive data on LDRLs representing all nine provinces of South Africa are mandatory for successfully developing NDRLs. A few publications have established LDRLs for IRPs within South Africa.¹⁶ A study from Tygerberg Hospital in Stellenbosch (Western Cape province) by Malan et al. in 2020 determined the LDRL for prevalent IRPs in South Africa and compared it with internationally published data.¹⁷ They reported that LDRLs for all IR procedures did not exceed published levels.¹⁷

The most recent similar investigation from Soweto (Gauteng province), South Africa, was conducted at Chris Hani Baragwanath Academic Hospital (CHBAH) and published in 2023 by Slave et al.¹ It sought to develop LDRLs and compared them with other published DRLs. The results showed that LRDLs for uterine artery embolisation (1463.8 Gy/cm²), diagnostic cerebral angiograms (209.3 Gy/cm²) and interventional cerebral angiograms (275 Gy/cm²) at CHBAH exceeded the international and local DRLs.¹ The published DRL ranges were as follows: uterine artery embolisation (118.4 Gy/cm² – 214 Gy/cm²), diagnostic cerebral angiogram (55 Gy/cm² – 87.5 Gy/cm²), interventional cerebral angiogram (63 Gy/cm² – 233.4 Gy/cm²), bronchial artery embolisation (73 Gy/cm² – 131.4 Gy/cm²), PTBD (23 Gy/cm² – 145 Gy/cm²) and unilateral nephrostomy (10 Gy/cm² – 47 Gy/cm²).^1,17

There are no prior calculations or publications of LDRLs for IRPs in the IR angiography suite at Dr. George Mukhari Academic Hospital (DGMAH), rendering the hospital non-compliant. Hence, this study aimed to establish the DRLs for IRPs at DGMAH to comply with the statutory requirements and guidelines, contribute to the establishment of NDRL and reduce adverse radiation effects from overexposure.

A retrospective, quantitative, descriptive, cross-sectional study was conducted at DGMAH. This public academic hospital is located in Pretoria (South Africa) and is affiliated with the Sefako Makgatho Health Sciences University. The hospital has a total of 1652 beds. A convenience sampling technique was employed, in which consecutive IRPs performed between 01 January 2023 and 31 December 2023 were included. Achieved radiation doses of IRPs performed in the angiography suite on patients above the age of 18 years were included. The procedures encompassed both diagnostic and therapeutic interventions involving INR and body IR. Body IRPs included non-vascular and vascular fluoroscopy-guided interventions.

The IRPs performed in an angiography theatre without using fluoroscopic guidance were excluded from the study. Patients under 18 years of age were excluded. Register entries lacking KAP values were not considered. Equally, illegible dose entries were excluded. The IRPs with fewer than 15 cases were excluded. Although ICRP recommends at least 20 examinations for DRL estimation, it also encourages flexibility when fewer procedure data are available.¹⁵ Several recent local publications used a 15 IRP threshold for establishing DRLs; hence, the adoption of this threshold.^1,5,17 This also enables appropriate comparison of LDRL values among local facilities.

The established KAP LDRL at DGMAH were within the published international DRL ranges, except for permanent catheterisation and upper limb interventions. Permanent catheterisation KAP LDRL (10.7 Gy/cm²) was above the international range (0.9 Gy/cm² – 2.9 Gy/cm²). Upper limb interventions KAP LDRL (17.1 Gy/cm²) were also higher than the international range (10 Gy/cm² – 14.2 Gy/cm²). The two procedures were typically reserved for trainees who were inexperienced in performing IRPs. These findings are consistent with published DRL studies, which established that procedures performed by inexperienced operators were associated with higher radiation doses.^1,17

The highest international cerebral intervention LDRL (104 min) for FT was recorded in a study done in India.¹⁹ The discrepancies in DRL values across nations and regions may be attributed to differences in imaging equipment, local procedure protocols and operator techniques.¹⁵ Cerebral interventions DRLs were ranked high among all IRPs for dose (KAP) and radiation exposure duration (FT) in this research, consistent with several other publications.^{1,17,19,20,21,22,23,24,25} This is presumably associated with the complexities of cerebral interventions. Other studies that published DRLs did not include data for cerebral interventions, as shown in Table 2 and Table 3.^{2,26,27,28,29,30,31,32,33,34}

Table 2 and Table 3 demonstrate that the established LDRLs for KAP were comparable with the published DRLs from other institutions in South Africa, Africa, the Middle East, Asia, Europe and the United States of America. A similar trend was demonstrated for FT LDRLs in Table 4 and Table 5.

The high radiation dose (KAP) values were not always associated with long radiation exposure (FT). This observation aligns with both global^25,30 and domestic^1,17 research findings. It underscores that no specific dose parameters measured at the local level can be reliably extrapolated to infer DRLs at the national or regional level.^1,15 Thus, it is imperative to assess as many variables as feasible when ‘optimising radiation exposure’.¹⁵ These sentiments are reiterated in the guidelines proposed by the ICRP.¹⁵

Data on pulmonary intervention procedures lack comparable international and local benchmarks; as such, it is distinctive to this study. These procedures were mostly performed in patients with pulmonary artery thrombosis for thrombolysis using the EkoSonic Endovascular System (EKOS) from Boston Scientific. Development of a detailed institutional protocol is recommended to guide the process of investigating excessive radiation doses, beginning with an assessment of the equipment, followed by a review of procedural protocols and ultimately an evaluation of operator techniques.¹⁵

This investigation established LDRLs for IRPs at DGMAH, which were predominantly comparable to South Africa’s and international DRL ranges, except for permanent catheterisation and upper limb interventions, which exceeded these benchmarks. Cerebral interventions and ASVI exceeded the local DRL ranges but remained within international ranges. These procedures need to be optimised. The missing data limited further interpretation of results; this needs to be considered in future studies. This study contributed to the establishment of NDRLs, regional DRLs and assisted DGMAH in optimising radiation dose. All centres involved in radiation imaging are urged to establish LDRLs to enhance the estimation of NDRLs for further optimisation.