Radiology departments as COVID-19 entry-door might improve healthcare efficacy and efficiency, and emergency department safety

In the Region of Murcia, Spain, general practitioners (GPs) see patients within a primary care network constituted by nine health areas with their own referral hospital. Each area includes basic health areas pivoting on primary care health centres. Given the previous experiences in Italy and other Spanish regions, with the arrival of the COVID-19 wave in February 2020, regional health authorities put forward a strategy based on primary care health centres GPs keeping patients with mild clinical presentation at home, similarly to the European recommendations [1]; at the Region of Murcia, they controlled more than 26,800 symptomatic cases suspected or diagnosed with COVID-19 and more than 34,600 asymptomatic contacts at home, only by close telephone follow-up [2]. But soon multiple conditioning factors came up and urged us to set up a secure primary care high-resolution radiology service (pcHRRS) in our health area, out of the emergency department (ED). These factors were the recommendations for chest X-rays even in the more resource-constrained scenario [3], the high volume of home-confined patients, the prevalence of dyspnoea in many patients, and the growing number of subjects asked by phone to go to the emergency or the radiology departments (RD) at the referral hospital without previous warning. The pcHRRS operated with a specific RD team 12-h daily, being at the same time the ED entry-door when appropriated. It was designed to offer GPs prompt objective respiratory clinical information from their home-based patients; to immediately transfer patients with pneumonia to the ED; to avoid overwhelming arrivals of respiratory patients to the ED; and, finally, to pilot and export the idea to the other health areas. Assuming the potential infection risk at overcrowded environments [4], we hypothesised that the pcHRRS would provide a much more efficient safer care. For that purpose, we aimed to analyse the pcHRRS efficiency in terms of reduced ED workload, waiting times, and admission triage through a simple radiology algorithm.

This cross-sectional study followed the SQUIRE guidelines and was approved by our hospital ethical committee (C.I. EST: 55/20). Patients’ informed consent was waived.

A top of 1494 confirmed and 5010 possible cases/week was reached in 23–29 March 2020, with a maximum of 119 confirmed cases/day on 25 March 2020. We considered that day to be the peak of the epidemic wave. The pcHRRS started on 26 March 2020 and has been active since then. From March 26th to 17 April 2020, 418 and 431 respiratory infection patients were seen through the pcHRRS and the ED, respectively. Ten scheduled patients did not attend the pcHRRS appointment. Those 418 patients accounted for 9.86% of the active confirmed or possible accumulated cases (233 and 4004 patients, respectively) followed up by telephone in that period of time, and 0.16% of our health area population (265.842 people).

The distribution of pcHRRS patients was: G1 325/418 (77.75%) and G2 93/418 (22.24%) patients (Additional file 1: Fig. 5). One patient with known fibrotic pulmonary lesions and other with a calcified granuloma where sent back home and included in G1. Among ED patients (G3), 224/431 (52%), 203/431 (47.10%) and 4/431 (0.93%) returned home, were admitted or refused admission, respectively. Eight pcHRRS patients encouraged to go home asked for ED assessment before being finally discharged. All of them went out from G1 (n = 317; 325 − 8) to be included in G3 (n = 439; 431 + 8) for the efficiency analysis; one G2 patient and four G3 patients refused medical attention at the ED and were excluded. Therefore, the final sample for our analysis was made up of 317 G1, 92 G2 and 435 G3 patients (Additional file 1: Fig. 5).

After starting the pcHRRS, the number of patients/day in the ED gradually decreased (Fig. 2). G1 patients stayed in hospital significantly less time than G2 and G3 subjects (0:41 ± 1:05 h; 5:25 ± 3:08 h; 5:36 ± 4:36 h, respectively; p < 0.001; Fig. 3), even when G2 and G3 patients returned home (0:41 ± 1:05 h; 3:36 ± 2:58 h; 3:50 ± 3:16 h, respectively; p < 0.001; Fig. 4). The time spent in the ED did not differ between G2 and G3 when they returned home (3:36 ± 2:58 h vs. 3:50 ± 3:16 h; p = 0.841), but was significantly shorter for G2 when patients were admitted (5:27 ± 3:08 h vs. 7:42 ± 5:02 h; p < 0.001; Fig. 4). Even considering the pcHRRS and ED times together in G2 patients, they waited less time than G3 patients, though the few 8/92 (8.7%) G2 patients returning home stayed a mean 1.08 + 3.36 h vs. 3:50 h (Fig. 4). The pcHRRS had a high yield for admission decisions, considering that G2 patients were admitted (84/93, 90.3%) more frequently than G3 (203/431, 47.1%; p < 0.001), with a rate per day always higher for G2 (mean rates: G2 0.92, range 0.67–1; G3 0.48, range 0.18–0.75), regardless of the epidemics time point (Fig. 5). Moreover, there were no significant differences in the ED radiology times from the electronic X-rays request to the X-ray examination (45.5 ± 57.7 min vs. 51 ± 41.3 min; p = 0.396), the X-ray examination to the report validation (46.3 ± 71.1 min vs. 41.8 ± 51.7 min; p = 0.569) and the X-rays request to the validation report (91.82 ± 89 min vs. 92.73 ± 67.8 min; p = 0.928) during the 5 days before and after the pcHRRS was started.

All 418 pcHRRS patients underwent posteroanterior and lateral chest X-ray views. Oximetry was implemented on March 31st, being applied systematically from patient 109. After getting empirically good results in four patients, 3DDT was systematically performed in addition to chest X-rays since April 1st, starting with patient 152. Regarding the 212 consecutive patients with all data available, their mean age was 46.75 ± 13.93 years, 87 (41%) men. Forty-eight (22.64%), 148 (69.81%) and 16 (7.54%) had abnormal, normal and questionable chest X-ray, respectively. All 54 patients with abnormal radiology examinations were referred to the ED. These 54 patients (G2) were older (50.46 ± 15.73 years) than G1 patients (45.41 ± 13.01 years; p = 0.019), men showing a trend to be more frequently referred to the ED than women (29/87, 33.33% vs. 27/115, 23.48%; p = 0.057) due to radiological abnormalities.

Among those first consecutive 212 patients, a follow-up chest X-ray was requested in 61 cases. That request was less frequent when initial chest X-ray was normal than when abnormal or questionable (18/148, 12.2% vs. 38/48, 75% vs. 5/16, 31.2%; p < 0.001). On the other hand, though follow-up radiographs worsen, respectively, in 15/48 (31.2%) and 2/16 (12.5%) patients with initial abnormal or questionable chest X-ray, no initial normal chest X-ray did it (0/148, 0%) (p < 0.001). No G1 patient needed hospital admission during the following weeks. Seven worsened clinically according to their GPs, but neither chest X-rays performed in the following 48 h worsened, nor did those carried out in four of them after the first 48 h. None of those patients required hospital admission. The 29 G2 patients who required hospitalisation for two days or more, and the four patients who required admission to the ICU, already had radiological findings suggestive of pneumonia in the first chest X-ray examination.

Oximetry was performed in 107/212 patients; in 6/212, data were missing. Mean blood oxygen saturation was lower in patients with abnormal (97.24 ± 1.52%, n = 17 vs. 98.49 ± 0.88%, n = 75; p < 0.001) or questionable chest X-ray (98.27 ± 0.80%, n = 15; p < 0.013).

During the first COVID-19 epidemic wave, respiratory patients were managed significantly faster through the pcHRRS, regardless of whether they returned home or were admitted. The pcHRRS contributed to halve the number of patients arriving to the ED for respiratory symptoms, triaging effectively the need for admission out of the ED.

The screening strategy recommended by the American College of Radiology and the Society of Thoracic Radiology for SARS-CoV-2 infection [6], based on RT-PCR and serology tests, has been applied in some countries [7]. But resources for laboratory tests in suspected SARS-CoV-2 infection were scarce and urged us to manage patients as potential infections in most occasions. Whatever the clinical setting, the RD was strategic when SARS-CoV-2 pneumonia had to be ruled out because hospital admission was then usually warranted [8]. Moreover, providing care for all suspected cases on a hospital basis may likely increase the number of infections in an already overwhelmed health system [7]. Therefore, according to our experience, a radiology entry-door may play a central role in ongoing waves of COVID-19 as our RD provided a faster and more efficient management, which was potentially safer [4].

The RD role to assess need for hospital admission in confirmed or possible COVID-19 patients has been a key innovative efficient strategy, but it was applied without a specific referral from the primary care network, and used CT as imaging technique [9]. Accordingly, the way home-confinement patients were managed and the straightforward imaging assessment are important differences in our case. Potential reduction in infection spread with our pcHRRS might have been achieved by keeping ppCOVID-19 outpatients with respiratory infection isolated in a specific route from the beginning [10, 11]. But, not less important, a personalised agenda contributes to clear facilities [12], guaranteeing a safe environment both for patients and for health workers [9, 11, 13‐15]. Risk contacts were significantly reduced through our pcHRRS by providing an extremely quick response in terms of specific appointments, safer X-ray technology and oximetry for decision-making, and expedited information transmission, always out of the ED. At the same time, it safely discriminated patients needing admission from those still ready to be managed at home. More than ¾ of ppCOVID-19 patients managed through the pcHRRS returned home and only 12% of those cases needed further X-rays assessment, who never showed radiological worsening. ppCOVID-19 patients seen through the pcHRRS and the ED routes were discharged in 77% and 52%, respectively, but the pcHRRS route was 5.6 times faster. Moreover, it had a remarkably high performance for discriminating ppCOVID-19 patients needing hospital admission, as over 90% of those referred to the ED were admitted. In a pandemic setting, chest CT has demonstrated high sensitivity but low specificity [16], except for expert radiologists [17]. However, the diagnosis of ppCOVID-19 in our environment has been and continues to be performed by chest X-ray examinations, according to the indications of reference scientific societies [18, 19]. CT scans have been exceptionally used for diagnosis, in severe symptomatic patients with X-ray findings not consistent with the clinical situation or equivocal. Indeed, some authors have agreed on the discriminating potential of chest X-ray as an independent factor for hospital admission [8]. In our case, that performance remained at the same level both at the epidemic peak and while it was declining.

Following our legal obligation [20], we have to keep a safe specific route for patients potentially infected by SARS-CoV-2, which necessarily needs X-ray assessment. We think that our RD workflow for ppCOVID-19 outpatients is this study’s main strength as we haven’t found reported a similar approach. Furthermore, the characteristics of age, gender and oximetry of our G2 patients were similar to those previously reported [8, 21], and the way both cohorts were finally built up within a strict home-confinement regional strategy reproduces a real setting whose results might be used as a model in ongoing pandemic waves. Furthermore, we believe that the higher pcHRRS discriminative performance makes beneficial to maintain and expand the RD entry-door role in the long term to be prepared for future COVID-19 waves or other similar pandemics, even more considering that the pcHRRS didn’t delay our usual radiology workflow of ED patients. But we also acknowledge some weaknesses. We have now complete clinical and radiological follow-up data in 212 patients, so many final outcomes are still under assessment. Although we didn’t consider for decisions any other clinical data beyond chest X-rays and oximetry during patients’ stay at our department, the clinical severity was established by the GPs the same day of the pcHRRS appointment and, when returning home, GPs closely followed them up. Furthermore, blood oxygen saturation levels were normal or almost normal in G1 and G2 groups, probable because our sample was mainly composed by mild forms of the disease. Therefore, we can assume that our decisions were safe. Furthermore, a retrospective assessment of the pcHRRS X-rays examinations by independent staff radiologists is still waiting. But the radiological reports were always validated by board-certified radiologists in patients without pneumonia findings, who never needed admission regardless of whether they were occasionally reassessed. Finally, we couldn’t compare the economic issues at this moment, but, in principle, only a resident without any additional income was required in the pcHRRS, while patients navigating through the ED would normally increase expenses due to other laboratory analysis and medical variability. In this manuscript, we present neither results on clinical–radiological results of our patients, including a Brixia Score analysis, nor data on 3DDT performance. That information was out of the scope of the current manuscript and will be presented in future submissions.

In summary, a RD positioned as a COVID-19 entry-door may be efficient in an epidemic setting to decrease respiratory patients at the ED, while potentially reducing infection hazards through safe and expedited decisions.