Lung ultrasound as a screening tool for SARS‐CoV‐2 infection in surgical patients

Abstract

Purpose

To evaluate the diagnostic performance of lung ultrasound (LUS) in screening for SARS‐CoV‐2 infection in patients requiring surgery.

Methods

Patients underwent a LUS protocol that included a scoring system for screening COVID‐19 pneumonia as well as RT‐PCR test for SARS‐CoV‐2. The receiver operator characteristic (ROC) curve was determined for the relationship between LUS score and PCR test results for COVID‐19. The optimal threshold for the best discrimination between non‐COVID‐19 patients and COVID‐19 patients was calculated.

Results

Among 203 patients enrolled (mean age 48 years; 82 males), 8.3% were COVID‐19‐positive; 4.9% were diagnosed via the initial RT‐PCR test. Of the patients diagnosed with SARS‐CoV‐2, 64.7% required in‐hospital management and 17.6% died. The most common ultrasound findings were B lines (19.7%) and a thickened pleura (19.2%). The AUC of the ROC curve of the relationship of LUS score with a cutoff value >8 versus RT‐PCR test for the assessment of SARS‐CoV‐2 pneumonia was 0.75 (95% CI 0.61–0.89; sensitivity 52.9%; specificity 91%; LR (+) 6.15, LR (−) 0.51).

Conclusion

The LUS score in surgical patients is not a useful tool for screening patients with potential COVID‐19 infection. LUS score shows a high specificity with a cut‐off value of 8.

In a high prevalence setting of SARS‐CoV‐2 infection, lung ultrasound severity score (LUSS) > 8 showed a high specificity. However, it cannot be recommended as a screening tool in patients requiring surgery due to its low sensitivity. B lines and subpleural consolidations had the best performance at identifying patients with pneumonia.

1. INTRODUCTION

By the end of 2019, an outbreak of SARS‐CoV‐2 infection rapidly spread from China to the rest of the world. ¹ The clinical manifestations of SARS‐CoV‐2 ranged from asymptomatic to severe pneumonia accompanied by organ function damage. ² , ³ , ⁴ The available data to date suggest that at least one‐third of SARS‐CoV‐2 infections are asymptomatic. ⁵ Nevertheless, asymptomatic patients can be a source of transmission. Li et al. ⁶ reported a nosocomial transmission from one asymptomatic thoracic surgery patient; 12 individuals, including health care staff and postoperative patients, had a positive exposure, and 11 individuals tested positive within 3–10 days of exposure.

Identifying asymptomatic patients with SARS‐CoV‐2 infection is important for taking appropriate measures to protect health care workers and other patients against nosocomial infections. Furthermore, surgical patients with asymptomatic or symptomatic SARS‐CoV‐2 infection have an increased risk of perioperative morbidity and mortality. ⁷ , ⁸ , ⁹ These findings are particularly relevant for patients requiring urgent surgery, because such procedures cannot be delayed.

Given the considerable perioperative morbidity and mortality associated with operating on COVID‐19 patients and the risk of nosocomial transmission, it is recommended that surgical patients undergo appropriate screening prior to surgery, without unnecessarily delays. ¹⁰ Preoperative COVID‐19‐positive testing rates range between 0.74% and 0.86%. ⁷ , ¹¹

A diagnostic work‐up that included clinical evaluation, history of exposure to SARS‐CoV‐2, and testing with reverse transcription‐PCR (RT‐PCR) is recommended. Computed tomography (CT) has been proposed for characterizing pulmonary involvement in COVID‐19 patients, due to its ability to detect lung changes related to SARS‐CoV‐2 infection. ¹⁰ , ¹² However, CT has the disadvantage of additional costs of screening, the burden of ionizing radiation, and low cost‐effectiveness in asymptomatic patients. Indeed, studies have shown limited added value of chest CT in preoperative screening. ¹³ , ¹⁴ , ¹⁵

Interest has emerged in the use of lung ultrasound (LUS) as an alternative first‐line imaging modality for screening patients. Soldati et al. ¹³ proposed a protocol for standardization of the use of LUS in COVID‐19 patients, using landmarks on chest anatomic lines and a scoring system that allows clinicians to record the highest score obtained in each area. The potential role of LUS in characterizing lung involvement in COVID19 is still debated. While in some studies LUS has been a useful tool for the early detection of SARS‐CoV‐2 infection ¹⁴ , ¹⁵ , ¹⁶ some recent studies found it is not a reliable imaging tool in ruling out covid19 pneumonia in patient presenting to the emergency department. ¹⁷

The purpose of this study was (1) to evaluate the diagnostic performance of LUS in screening for SARS‐CoV‐2 infection in patients requiring urgent surgery; and (2) to identify the cutoff value of the LUS score for COVID‐19 pneumonia that discriminates patients with SARS‐CoV‐2 infection.

2. MATERIALS AND METHODS

2.1. Study design and human subjects

This prospective study was designed for reporting the diagnostic accuracy of LUS for diagnosing SARS‐CoV‐2 infection during the screening process of patients requiring emergency surgery, following the STARD guideline. ¹⁸ The study was carried out in an academic hospital in Cali, Colombia, equivalent to a Level I trauma center. The study was conducted in accordance with the Declaration of Helsinki and was approved by the Research Ethics Committee of our institution.

2.2. Data source

Between May 15, 2020, and November 17, 2020, all adult patients were enrolled who were <72 hours from admission and who required urgent surgery for any cause. We excluded patients <18 years old, prisoners, patients with chronic lung disease, heart or renal failure, patients who were intubated on admission, and patients who declined to participate. Demographics, clinical parameters, laboratory values, PCR‐testing for SARS‐CoV‐2, imaging features of LUS, and outcome variables were collected. Data were collected using a database that allows real‐time data entry. The RT‐PCR testing results and outcomes were later registered by the researchers. A 15‐day follow‐up was completed in patients with negative RT‐PCR results.

2.3. Data collection

In our hospital during a SARS‐CoV‐2 outbreak, collection of LUS score and nasopharyngeal swabs for RT‐PCR assays were included in the screening process of the routine work‐up for patients who required any surgical procedure. All attending physicians (emergency medicine and surgeons) had experience in point‐of‐care ultrasound as a standard diagnostic tool; they were also trained to perform LUS.

2.4. PCR testing for SARS‐CoV‐2

Clinical specimens for COVID‐19 diagnostic testing were obtained from nasopharyngeal swabs and processed by RT‐PCR assay. In cases of positive results, patients undergo their scheduled surgery and are thereafter admitted to the COVID‐19 Care Unit. In cases of negative SARS‐CoV‐2 test results, but that are highly suspected to be infected due to the presence of symptoms (fever, cough, dyspnea, diarrhea, dry cough, ageusia or anosmia) or previous contact with positive cases within the last 14 days, patients were re‐tested for SARS‐CoV‐2 at day 5 post admission using RT‐PCR. If the second RT‐PCR result was negative and symptoms persisted, yet another RT‐PCR assay was performed on day 10. If SARS‐CoV‐2 infection remained strongly suspected, a final RT‐PCR nasopharyngeal swab was performed on day 15. We defined a confirmed COVID‐19 case in patients with a positive RT‐PCR test result.

2.5. Ultrasound data collection

Lung ultrasound was performed by certified operators using a portable ultrasound device (Sonosite Edge and Sonosite Edge II®). Linear and phased array probes were used. Sonographers were unaware of the patients' clinical manifestations and of the nasopharyngeal swab results. Patients were examined in a sitting position (when possible) with a continuous scan of intercostal spaces covering 14 areas (3 posterior, 2 lateral, and 2 anterior) using the protocol by Soldati et al. ¹³ For each of the 14 regions, a score ranging from 0 to 3 was assigned, according to the following findings: the pleural line is continuous and regular (0 pts); the pleural line is indented (1 pt); the pleural line is broken (2 pts); the presence of dense and largely extended white lung with or without larger consolidations (3 pts). At the end of the procedure, the sonographer recorded the highest score obtained for each area. The total LUS score was calculated by summing the scores of the 14 zones (range of possible scores: 0–36).

2.6. Statistical analysis

Statistical analyses were performed using Stata 15.1® (College Station, TX). Categorical variables were presented as frequencies and percentages. The normality of continuous variables was examined by the Shapiro–Wilk test. Afterward, they were presented as mean and standard deviation or median and IQR, according to the normality of the data. Operative characteristics were also calculated. The ultrasounds were interpreted as negative when no suspicious signs were identified or when the sum of the scores was zero. Conversely, the ultrasounds were interpreted as positive when any suspicious sign was found or when the score was ≥1. Regarding SARS‐CoV‐2 infection, each patient was classified as positive if any RT‐PCR test results were positive and classified as negative if the results were negative (Figure 1). After the classification, operative characteristics with their respective 95% CI were calculated for every ultrasonographic sign. The severity of the ultrasonographic findings was computed by summing the severity of the identified sign in each of the 14 prespecified points, thereby generating the Lung Ultrasound Severity Score (LUS). An ROC curve was constructed to evaluate the discriminative ability and the best cutoff for the obtained score.

FIGURE 1.

Ultrasound in COVID‐19 screening in urgent surgical patients. Flow diagram of the participants and their results

2.7. Sample size

The operative characteristics of ultrasound for diagnosing SARS‐CoV‐2 infection were unknown when the study was planned. However, sensitivity and specificity were reported by Testa and coworkers for the early diagnosis of H1N1 pneumonia (94% and 85%, respectively). ¹⁹ Buderer's method was used to calculate the sample size. ²⁰ With a 95% CI and an error margin of 10%, samples of 433 and 217 are appropriate for a prevalence of COVID‐19 of 5% and 10%, respectively. ²¹ For every 100 included patients, the prevalence was evaluated to adjust the sample size.

2.8. Ethical considerations

This investigation was approved by the investigation committee and the ethics committee on May 12 and 14 of 2020 respectively (record #041, act #160‐2020). Given that the pulmonary ultrasound and the RT‐PCR assay were incorporated into usual care, the requirement to obtain informed consent was waived.

3. RESULTS

3.1. Patient features

A total of 292 patient candidates for urgent surgery were evaluated with LUS; of these patients, 89 were excluded, the majority of whom because they did not undergone surgery (Figure 1). The remaining 203 patients were enrolled, whose median age was 48 years and 121 (59.6%) of whom were women and 82 (40.4%) of whom were men. The clinical characteristics of the patients are summarized in Table 1. The most frequent surgical group was trauma and emergency surgery (39.9%), followed by orthopedic surgery (14.8%) and gynecology (3.4%; Table 1).

TABLE 1.

Ultrasound for COVID‐19 screening in urgent surgical patients

Variable	Result
Sex
Female, n (%)	121 (59.6)
Male, n (%)	82 (40.4)
Age, median (IQR)	48 (35–66)
Group of surgery
Emergency surgery, n (%)	45 (22.2)
Trauma, n (%)	36 (17.7)
Orthopedics, n (%)	30 (14.8)
Gynecology, n (%)	7 (3.4)
Oncologic surgery, n (%)	1 (0.5)
Other, n (%)	84 (41.4)
Respiratory symptoms
In the first evaluation, n	14 (6.9)
In posterior evaluations, n	4 (2.0)
SARS‐Co‐V‐2 infection
Present, n (%)	17 (8.4)
Absent, n (%)	186 (91.6)
Infection severity, n = 17
Ambulatory, n (%)	6 (35.3)
Hospital, without oxygen, n (%)	2 (11.8)
Hospital with oxygen, HFNC or NIMV, n (%)	3 (17.6)
Hospital with mechanical ventilation, n (%)	3 (17.6)
Dead, n (%)	3 (17.6)

Note: General characteristic.

3.2. RT‐PCR testing and symptoms

Fourteen patients (6.9%) reported respiratory symptoms during the first evaluation. Of these patients, four (2.0%) reported symptoms in posterior evaluations. In 12 of them (70.6%) a SARS‐CoV‐2 infection was confirmed.

In total, 17 patients (8.4%) had a confirmed SARS‐CoV‐2 infection. RT‐PCR test was positive in the first evaluation in 10 patients (58.8%) and in additional posterior evaluations in seven patients (41.2%). Of the confirmed COVID‐19‐positive cases, 11 required in‐hospital management, three received invasive mechanical ventilation, and three died.

3.3. Lung ultrasound findings

Lung ultrasound was negative for COVID‐19 in 128 patients (63.1%; Table 2). The most frequent ultrasound findings were B lines (n = 40; 19.7%), followed by pleural abnormalities, such as thickened pleura (n = 39; 19.2%) and pleural discontinuity (n = 27; 13.3%). Areas of lung consolidations were less frequently seen: subpleural consolidation was documented in 19 cases (9.4%) and lobar consolidations in three cases (1.5%; Table 2).

TABLE 2.

Ultrasound for COVID‐19 screening in urgent surgical patients

Finding	n (%)
Positive	75 (37.9)
Ultrasound finding
Thickened pleura	39 (19.2)
Pleural discontinuity	27 (13.3)
B lines	40 (19.7)
Subpleural consolidation	19 (9.4)
Lobar consolidation	3 (1.5)
Lung ultrasound score
0	128 (63.1)
1–7	50 (24.6)
8–20	16 (7.9)
>20	9 (4.4)

Note: Ultrasound findings.

The LUS score constructed from the sum of the findings in the 14 examined points (LUS) revealed 1–7 pts in 50 patients (24.6%), 8–20 pts in 16 (7.9%), and a >20 pts in 9 (4.4%) patients (Tables 2 and 3).

TABLE 3.

Ultrasound for COVID‐19 screening in urgent surgical patients

Finding	Covid‐19 (+)		Covid‐19 (−)
	US (+)	US (−)	US (+)	US (−)
Ultrasound finding
Any positive ultrasound finding	12	5	63	123
Thickened pleura	2	15	37	149
Pleural discontinuity	2	15	15	161
B lines	7	10	30	156
Subpleural consolidation	6	11	13	173
Lobar consolidation	0	17	3	183
Lung ultrasound score cut‐point
>0	12	5	63	123
>7	9	8	16	170
>20	6	11	3	183

Note: Contingency table of the ultrasound findings.

3.4. Operative characteristics of the ultrasonographic findings

The presence of one or more abnormal LUS finding had a sensitivity of 70.6%, a specificity of 66.1%, an L.R. (+) of 2.08, an L.R. (−) of 0.44, and an AUROC of 0.68 (95% CI 0.57–0.80). Table 4 shows the operative characteristics of the ultrasonographic findings.

TABLE 4.

Ultrasound for COVID‐19 screening in urgent surgical patients

Finding	Sensitivity	Specificity	L.R. (+)	L.R. (−)	AUROC (95% CI)
Ultrasound positive	70.6	66.1	2.08	0.44	0.68 (0.57–0.80)
Ultrasonographic finding
Thickened pleura	11.8	80.1	0.59	1.1	0.46 (0.38–0.54)
Pleural discontinuity	11.8	85.6	0.88	1.0	0.49 (0.41–0.57)
B lines	58.8	83.9	3.65	0.49	0.71 (0.59–0.49)
Subpleural consolidation	35.3	93.0	5.05	0.69	0.64 (0.52–0.76)
Lobar consolidation	0	94.4	0	1.0	0.49 (0.48–0.50)
Lung ultrasound score
>0	70.6	66.1	2.08	0.44
>7	52.9	91.9	6.15	0.51	0.75 (0.61–0.89)
>20	35.3	98.4	21.88	0.66

Note: Operative characteristics of the ultrasonographic findings.

Abbreviations: AUROC, area under the receiver operating characteristic curve; L.R., Likelihood Ratio.

Of the individual findings, B lines and subpleural consolidation had the best performance, with a sensitivity of 58.8% and 35.3%; a specificity of 85.6 and 83.9; an L.R. (+) of 3.65 and 5.05; an L.R; (−) of 0.49 and 0.69; and an area under the corresponding ROC curve (AUROC) of 0.71 and 0.64, respectively (Table 4), which did not differ statistically from the LUS AUROC (p = 0.209). Despite the lack of statistically significant differences in the AUROC, the graphic comparison of the curves shows a greater specificity of B lines in the range above 3 (Figure 2).

FIGURE 2.

Graphic comparison of different ROC curves

The AUROC of the LUS for SARS‐CoV‐2 pneumonia identification was 0.76 (0.62–0.90). The best‐observed sensitivity was 70.6%, with a cut‐off of 1 pts, which decreased to 35.3% with a cut‐off of 20. On the other hand, the specificity increased to 91.9 with a cut‐off of 8 pts and to 98.4% with a cut‐off of 20 pts. The corresponding L.R. (+) were 6.15 and 21.88, respectively (Table 4).

4. DISCUSSION

During the study period, Colombia faced its second COVID‐19 wave. The Valle del Cauca was one of the most affected regions in Colombia, with an incidence rate of 77 per 100 000 individuals. ²² Hospitals in our city were at maximum capacity due to SARS‐CoV‐2 cases, with 8.3% of infected cases requiring in‐hospital management. ²²

Patients in our study were treated by different surgical subspecialists, with the majority being treated by specialists in trauma surgery and orthopedic surgery, as reported by other authors. ²³ , ²⁴ In a study by Lei et al. ⁹ the median age of positive COVID‐19 patients who underwent elective surgery was 55 years (IQR 43–63) and, as in our study, the majority of patients were women.

In our study, 8.4% of surgical patients tested positive for SARS‐CoV‐2, 17% of whom did not report any respiratory symptoms. The asymptomatic proportion of patients is estimated to range between 18% and 57% in other cohorts. ⁴ , ²⁵ , ²⁶ In our study, 17.5% of positive cases died due to respiratory complications of COVID‐19. Di Martino et al. ²⁷ reported a similar percentage of positive cases (7%) in ambulatory surgery patients, but only a 1.4% mortality rate attributed to disease progression. This discrepancy may have arisen from the fact that their patients were not emergency surgery cases. The mortality rate in the study by Li ⁶ in thoracic surgery patients with SARS‐CoV‐2 was 30.8%, which is significantly higher than previously reported rates.

During the pandemic, LUS became a frequent tool to identify disease severity and to facilitate screening of potentially infected patients, as described by other authors. ²⁸ , ²⁹ LUS was also used in 2009 for rapid point‐of‐care triage and management of patients during the H1N1 influenza virus outbreak. ³⁰ For the detection of pneumonia, LUS can achieve a specificity of 75%–94% and a sensitivity of 85%–95%, ¹⁶ , ³¹ , ³² , ³³ and it is superior to X‐ray and comparable to thoracic CT. ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ LUS appears to be a promising technique for the early identification of pneumonia in patients with suspected COVID‐19 infection in the context of an active pandemic ¹⁴ ; however, few studies to date have evaluated LUS in surgical patients during a SARS‐CoV‐2 outbreak.

COVID‐19 typically induces an interstitial diffuse bilateral pneumonia. The sonographic signs of interest include B lines in various forms (e.g., focal, multifocal, or confluent B lines) as well as light beams, irregular or fragmented aspect of the pleural line, and small peripheral or subpleural consolidations. ¹⁶ , ⁴⁰ , ⁴¹

The most frequent LUS findings in our study were B lines, followed by pleural abnormalities and, less frequently, subpleural consolidations; these pathological features have been previously described. ¹⁶ , ⁴² , ⁴³ , ⁴⁴ Our results differ from those reported by Tung‐Chen et al. ⁴⁵ who found subpleural consolidations on the posterior lower lobes to be the most common finding. They also found LUS compatible with moderate affection in most of their patients, which differs from our findings of mild affection. This difference may be because the cohort in Chen et al. comprised highly symptomatic patients.

We identify that a cutoff LUS of >8 identified patients with SARS‐CoV‐2 pneumonia with a sensitivity 52.9% and specificity of 91.9%. The discriminative ability of the LUS, determined by an AUROC of 0.76, was acceptable, according to Hosmer and Lemeshow. ⁴⁶ Nevertheless, its ability as a screening tool is completely limited by the high rate of false negatives observed. On the other hand, specificity increased, as LUS did. LUS scores above 7 would confirm the diagnosis if the clinical characteristics are of intermediate or high probability of a Covid‐19 infection.

The comparison of the LUS score with the sum of the points for the individual components showed a similar behavior of B lines and a slightly poorer behavior of subpleural consolidation. Specificity of B lines with scores higher or equal than 3 was higher than 90%. Our findings are similar to those reported by Fonsi et al. ⁴⁷ in their study on patients admitted to the emergency department in an Italian Level I trauma center. Lung ultrasound has a sensitivity, specificity, positive predictive value, and negative predictive value of 68%, 79%, 88%, and 52%, respectively. The AUC was 0.745 (95% CI 0.606–0.884). ⁴⁷ Zanforlin et al. ¹⁴ study performed a retrospective analysis and reported a higher AUC of 0.83 applying a cutoff LUS > 3, with a higher sensitivity of 81% and lower specificity 80%, compared with our findings.

Quarato et al. ¹⁷ estimate sensitivity of admission LUS for the detection of SARS‐CoV‐2 lung involvement using Chest‐CT as reference standard to assess LUS usefulness in ruling out COVID‐19 pneumonia in the Emergency Department. They found a lower rate of detection for COVID19 lung findings using LUS than Chest‐CT. LUS showed a global sensitivity of 52.44% in detecting pulmonary lesions, a very similar sensitivity to the one obtained in our study.

Our results, which were obtained in an emergency setting with the purpose of rapidly identifying COVID‐19 pneumonia in surgical patients, differ from those reported before the onset of this pandemic. ³⁴ , ³⁶ , ⁴⁷ Our findings do not dismiss LUS in clinical settings; however, we do not recommend LUS as a screening tool in acute surgical patient for identifying SARS‐CoV‐2 pneumonia, due to its low sensitivity.

There are some important considerations regarding LUS in SARS‐CoV‐2 surgical patients. For instance, LUS performance remains poor at detecting deep alveolar lesions, ⁴⁸ which may explain some of the false‐negatives results of the LUS. Furthermore, a considerable number of patients are asymptomatic for pneumonia or have different degrees of lung involvement, which may confound the diagnosis. ⁴⁹ Raiteri et al. ⁵⁰ study showed that approximately 30% of expectedly healthy COVID‐19 negative subjects presented some lung ultrasound abnormalities, limiting LUS as a screening tool for lung involvement in COVID‐19 suspected or confirmed patients. Lastly, in a time of low prevalence disease, the ultrasound findings may overlap with other medical conditions, which could hinder the accurate diagnosis of COVID‐19. ⁴⁹

Some limitations need to be accounted for in this study. This was a single‐center experience, which may lack inter‐rater reliability. Multicenter studies are needed for a more precise sensitivity and specificity of LUS in surgery patients. Finally, the total sample size of the study as well as the SARS‐ CoV‐2 positive subset of patients were small.

LUS is an important tool in emergency settings—it is radiation free, time‐saving, and has a broad availability. In a high prevalence setting of SARS‐CoV‐2 infection, a LUSS > 8 showed a high specificity, and B lines and subpleural consolidations had the best performance at identifying patients with pneumonia. However, LUSS cannot be recommended as a screening tool in surgical emergency patients, due to the low sensitivity of LUS.

CONFLICT OF INTEREST

The authors declare that they have no competing interests.

ETHICS STATEMENT

This investigation was approved by the investigation committee and the ethics committee (record #041, act #160‐2020). Given that the pulmonary ultrasound and the RT‐PCR assay were incorporated into usual care, the requirement to obtain informed consent was waived.

García AF, Ángel‐Isaza AM, Chica J, et al. Lung ultrasound as a screening tool for SARS‐CoV‐2 infection in surgical patients. J Clin Ultrasound. 2022;50(9):1271‐1278. doi: 10.1002/jcu.23358

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.