Prevalence of Thyroid Incidentalomas from 1995 to 2016: A Single-Center, Retrospective Cohort Study

Abstract

Background

Incidental detection of thyroid nodules on nonthyroid imaging may contribute to increased diagnosis of thyroid cancer. We investigated the prevalence of thyroid incidentalomas across imaging modalities among a predominately male veteran population.

Methods

Thyroid nodules were identified on nonthyroid-directed radiology reports using natural language processing. All reports from 1995 to 2016 for chest computed tomography (CT), carotid ultrasound (US), and neck magnetic resonance imaging (MRI) were reviewed. Individuals with multiple studies were included at their initial study and duplicates removed.

Results

A total of 25 763 carotid US, 23 526 chest CTs with contrast, 39 262 noncontrast chest CTs, and 9503 MRIs were reviewed. With duplicates removed, 14 642 carotid US, 12 923 chest CTs with contrast, 17 416 noncontrast chest CTs, and 6926 MRIs were included. Mean age was 66.2 years and 1834 were female (3.53%). Thyroid nodules were reported on 0.84% carotid US, 3.45% MRIs, 5.84% chest CTs with contrast, and 5.14% noncontrast chest CTs. Women had a higher rate of thyroid nodules on MRI (6.46% vs 3.20%, P = .003), chest CT with contrast (9.80% vs 5.72%, P = .007), and noncontrast chest CT (8.77% vs 5.02%, P = .002), but not on carotid US (1.99% vs 0.81%, P = .12). Incidentaloma prevalence increased with age on MRI, chest CT with or without contrast, but not on carotid US, and were more commonly reported from 2007 to 2016 compared to before 2007 across all modalities.

Conclusions

Thyroid incidentalomas are commonly reported, are more common among women, and increase with age. The rate of reported incidental thyroid nodules is increasing, likely contributing to the increase in thyroid cancer.

Incidental detection of thyroid nodules (thyroid incidentalomas) on nonthyroid-directed imaging may be contributing to increasing thyroid cancer diagnoses. In the United States, the yearly thyroid cancer incidence increased from 4.9 per 100 000 in 1975 to 14.3 per 100 000 in 2009, predominantly due to increases in papillary thyroid cancer [1]. The importance of incidentaloma detection to increased thyroid cancer diagnosis is unclear because the current literature is incomplete and conflicting. Two large database studies found that less than 1% of imaging studies report a thyroid incidentaloma [2, 3] suggesting that incidentalomas are not contributing to rising thyroid cancer rates. Smaller studies have found significantly higher incidentaloma rates. When neck computed tomographies (CTs) were evaluated, 16.8% were found to have a thyroid incidentaloma, with 10% of these incidentalomas being malignant, indicating that thyroid incidentalomas are a significant factor in thyroid cancer diagnosis [4]. In a population study, 68% of randomly selected patients with a mean age of 56.7 years had a thyroid nodule on ultrasound (US) [5]. In addition to these disparate findings, gaps in the literature regarding the frequency of incidental nodules among men and older people restrain conclusions about the importance of thyroid incidentalomas.

Increased thyroid cancer incidence sharply contrasts with the stable mortality of thyroid cancer. Thyroid incidentalomas and thyroid nodules are challenging because 7% to 15% of thyroid nodules represent thyroid cancer [6]. Thyroid cancer is estimated to be the 11th most common type of cancer diagnosed in 2018, with 53 990 new cases, an incidence of 14.5 per 100 000 men and women per year. Despite high prevalence, death rates for thyroid cancer are 0.5 per 100 000 men and women per year, and is estimated to have caused 2060 deaths in 2018, placing it as the 29th leading cause of cancer death [7]. Because of the cancer threat, the American Thyroid Association recommends thyroid incidentalomas be evaluated with a formal thyroid US, followed by further evaluation based on size and US characteristics similar to nodules detected in other ways [6].

Detection of thyroid incidentalomas can also be seen as inadvertent screening for thyroid nodules. The United States Preventive Services Task Force recommends against screening for thyroid cancer in asymptomatic adults [8] because the screening benefits were found to be small, but screening harms to be at least moderate, thus determining negative net benefit. Screening for thyroid cancer with US practically occurs in South Korea with opportunistic screening, which began in 1999. This led to a rapid increase in the incidence of thyroid cancer, and in 2011 the incidence was almost 70 per 100 000. Currently, thyroid cancer is the most common type of cancer diagnosed in South Korea. Despite this increase in diagnosis, thyroid-cancer mortality there is unchanged [9].

Given the current thyroid cancer epidemic, incomplete and conflicting studies on thyroid incidentalomas, and increasing use of imaging studies for screening purposes (such as chest CT for lung cancer screening), it is important to study the rates of reported incidental thyroid findings to help understand the growing thyroid cancer epidemic and to develop future research and treatment algorithms for thyroid incidentalomas. We chose to address this question by investigating the true prevalence of thyroid incidentalomas on chest CT, head/neck magnetic resonance (MRI), and carotid US within our predominately male patient population.

1. Participants and Methods

A. Participants

All radiology reports for chest CT with or without contrast (March 1995 to September 2016), carotid US (March 1995 to September 2016), and MRI cervical spine and MRI neck/orbit/face (June 2001 to September 2016) from the Minneapolis VA Health Care System were included in our study. These imaging studies were selected given their commonality and likelihood of including the thyroid gland in the image. Other imaging modalities such as neck CT were examined in preliminary data and not included because of their low frequency. Dates were chosen to include all available electronic data. Individuals who underwent multiple imaging studies of the same modality were included at the time of their initial study and subsequent imaging studies were removed. This project was conducted as a quality improvement initiative. It was reviewed by the Minneapolis VA Institutional Review Board and determined that the board’s approval was not required.

B. Incidentaloma Identification Method

We developed a natural language processing technique to identify thyroid incidentalomas reported in radiology text reports. This technique was developed by a trial and error method followed by validation. First, reports (body of the report and impression) were screened for the word “thyroid” and excluded if the word “thyroid” did not appear in the report text. Reports were then excluded if the word “thyroid” appeared near the word “normal” or “unremarkable.” To ensure these were incidental findings, reports were also excluded if the word “thyroid” appeared in the clinical history/reason for examination section of the report. The remaining reports were then manually reviewed by author T.D. to verify the presence of a thyroid nodule, defined as mention of 1) a nodule, mass, or lesion within the thyroid gland, 2) focal attenuation abnormality within the thyroid gland, 3) calcifications within the thyroid gland, or 4) nodularity of the thyroid gland. Reports were not included if they mentioned the following: previously seen thyroid nodule that is not identified on current imaging study, enlargement of 1 or both lobes of the thyroid without mention of nodularity of lesions, goiter without mention of nodularity or lesions, diffuse attenuation abnormality, or imaging performed for a palpable neck abnormality.

C. Validation of Method

A random representative sample of 500 imaging reports that were automatically excluded (either by not including the word “thyroid,” by having the word “normal” or “unremarkable” near the word “thyroid,” or by having the word “thyroid” in the clinical history/reason for examination) were manually reviewed. Of these 500, a total of 498 reports were properly excluded, giving this method a negative predictive value of 99.6%.

D. Statistical Analysis

Rates of incidental thyroid nodules were compared within each modality (MRI, CT no contrast, CT with contrast, and carotid US). To examine whether rates were different between men and women, age categories, and before 2007 vs 2007 or later, we compared proportions biviariately using logistic regression. This year was chosen to create roughly equal time interval cohorts for different imaging modalities. Additionally, it was hypothesized that there would be an age trend, so this was tested using the Cochran-Armitage Trend test. All statistical analyses were conducted in SAS 9.2.

2. Results

Between 1995 and 2016, a total of 25 763 carotid US, 23 526 chest CTs with contrast, and 39 262 noncontrast chest CTs were performed within the facility and reviewed, and between 2001 and 2016, a total of 9503 MRIs (cervical spine or face/orbit) were performed and reviewed, for a total of 98 054 imaging studies reviewed. After duplicate imaging studies were removed, 14 642 carotid US, 6926 MRIs, 12 923 chest CTs with contrast, and 17 416 chest CTs without contrast were included in our analysis. Of these 51 907 included imaging studies, the mean age was 66.2 years and 1834 were female (3.53%). See Fig. 1 for the number of each study performed per year, after duplicates were removed.

Figure 1.

Number of unique subject imaging studies performed annually by imaging modality.

Overall, 0.84% of carotid US (123 of 14 642), 3.45% of MRIs (239 of 6926), 5.84% of chest CTs with contrast (755 of 12 923), and 5.14% of chest CTs without contrast (895 of 17 416) reported a thyroid incidentaloma. Women had a higher rate of reported incidentalomas compared to men on MRI (6.46% vs 3.20%, P = .003), chest CT with contrast (9.80% vs 5.72%, P = .007), and chest CT without contrast (8.77% vs 5.02%, P = .002), but not on carotid US (1.99% vs 0.81%, P = .12). Reported incidentaloma prevalence significantly increased with age on MRI, chest CT without contrast, and chest CT with contrast, but not on carotid US; see Table 1 for details.

Table 1.

Rate of reported incidentalomas based on imaging study and age

Age, y	No. of Imaging Studies	% Reported Incidentaloma	P
Carotid US
≤ 50	444	0.68%	.23
51-60	2131	0.99%
61-70	5026	0.94%
71-80	5091	0.79%
> 80	1950	0.62%
Total	14642	0.84%
MRI
≤ 50	1946	2.57%	<.001
51-60	1982	2.93%
61-70	1850	4.27%
71-80	837	3.58%
> 80	311	7.07%
Total	6926	3.45%
Chest CT with contrast
≤ 50	1163	3.01%	<.001
51-60	2558	5.00%
61-70	4137	6.12%
71-80	3618	6.30%
> 80	1447	7.67%
Total	12 923	5.84%
Chest CT without contrast
≤ 50	1553	1.93%	<.001
51-60	3045	4.07%
61-70	6140	4.74%
71-80	4350	6.44%
> 80	2328	7.30%
Total	17 416	5.14%

Abbreviations: CT, computed tomography; MRI, magnetic resonance imaging; US, ultrasound.

Thyroid incidentalomas were more commonly reported from 2007 to 2016 compared to 1995 to 2006 on carotid US (1.15% vs 0.64%, P = .002), on chest CT with contrast (9.15% vs 4.20%, P < .001) and on chest CT without contrast (5.91% vs 4.04%, P < .001), and also on MRI from 2007 to 2016 compared to 2001 to 2006 (3.93% vs 2.75%, P = .006). See Fig. 2 for the annual rate of reported thyroid incidentalomas by imaging modality.

Figure 2.

Annual rate of reported thyroid incidentalomas by imaging modality.

3. Discussion

Within our institution for the entire interval from 1995 to 2016, a total of 5.44% of chest CTs, 3.45% of head/neck MRIs, and 0.84% of carotid US reported a thyroid incidentaloma. Reported incidentalomas increased with age and appear to have increased in the past decade compared to the prior decade. Incidentalomas were nearly twice as likely in women, although even among men the rate of incidental detection was substantially higher than the less than 1% seen in some previous studies [2, 3]. Thyroid nodules have previously been shown to be more common in women and with increasing age [5, 10] and these findings are consistent with this sex and age difference. Thyroid incidentalomas were more commonly reported within the past decade (2007-2016) compared to the prior decade (1995-2006). Although it is known that thyroid nodules are very common, there are currently limited data focused specifically on thyroid incidentalomas. Our study represents a robust validated cohort that more accurately defines the scope of thyroid incidentalomas.

There is a discrepancy in the current literature on the rate of reported thyroid incidentalomas. Two small studies have reported a roughly 16% rate of verifiable (present on dedicated review of images including the thyroid gland) thyroid incidentalomas on CT and MRI. The first included 231 imaging studies with personal review of each case and showed a rate of reported incidentalomas of 6% on CT and MRI, with an additional 10% of verifiable incidentalomas present but not reported [11]. Another study of 734 CT scans that were manually reviewed showed a 16.8% rate of verifiable incidentalomas, though this study did not specify the rate at which these nodules were reported in official readings [4]. Contrasting with these findings, 2 recent large database studies have shown a much lower rate of reported thyroid incidentalomas. A single-center, retrospective review of 97 908 imaging studies (CT, MRI, neck US, and positron emission tomography) found that only 0.4% of these imaging studies reported a thyroid incidentaloma [2]. In another single-center, retrospective study, 109 220 imaging studies (CT, MRI, nonthyroid-directed US, and chest radiograph) were reviewed and only 125 (0.1%) were identified as having a thyroid nodule [3]. Our large database study including 98 054 imaging studies finds a higher rate of incidentalomas, more consistent with the previous smaller study finding a 6% rate of reported incidentalomas when all images were manually reviewed [11].

There is no clear definition of how a thyroid incidentaloma is reported on CT or MRI studies. Our methods included natural language-processing techniques to exclude negative studies, was validated with a 99.6% negative predictive value, and all positive reports were manually reviewed for verification of the inclusion/exclusion criteria. The study by Davies et al of 109 220 imaging reports included “thyroid mass,” “thyroid nodule,” and “thyroid cyst,” whereas Uppal and colleagues’ study of 97 908 imaging reports searched for the word “thyroid,” with both studies then excluding previously known thyroid disease [3, 2]. Our inclusion criteria were developed to account for differences in how thyroid abnormalities are reported by different radiologists, and all incidentaloma cases were confirmed with manual review of the radiology report. Our inclusion criteria were 1) a nodule, mass, or lesion within the thyroid gland, 2) focal attenuation abnormality within the thyroid gland, 3) calcifications within the thyroid gland, or 4) nodularity of the thyroid gland. Calcifications within the thyroid gland were included because these have been shown to correlate with microcalcifications on US [12] and with malignancy [4]. The goal was to be specific to thyroid nodules, and not to any general thyroid structural abnormality. Our criteria allow for standardization in the reporting of thyroid incidentalomas, something previous studies have been lacking.

Although it is known that thyroid nodules are more common among female patients, our study represents the rare thyroid cohort of predominately male patients in a Veteran Affairs hospital. The previously mentioned study by Uppal et al of 97 908 imaging studies did not report the female to male ratio but did report that 80% of the patients with a thyroid incidentaloma were female, whereas the study by Yoon et al of 734 CT scans included 48% women [2, 4]. Despite the female predominance, thyroid nodules are still common among men and studies such as ours are needed to gain further information among this population.

It is unclear why thyroid incidentalomas were more commonly reported within the past decade compared to the prior decade. It is unlikely to be a true rise in prevalence of thyroid nodules based on prevalence data published before the year 2000 [10], and therefore is more likely due to a rise in the reporting rate of incidental thyroid nodules. This rise could be due to many factors, including advances in imaging quality and techniques, changes in reporting habits, or a true increase in thyroid nodules, and evaluation of these factors is beyond the scope of this study. It has previously been shown that there is high variability among radiologists when reporting thyroid incidentalomas [13].

There was a low rate of reported thyroid incidentalomas on carotid US within our facility at 0.84%. In comparison, a thyroid incidentaloma rate of 9.4% was found among 2004 patients who underwent carotid US between January 2000 through January 2002, and those with previously known thyroid disease were excluded [14]. Compared to our study, that study had a similar mean age (67.8 years compared to 66.2 years) but a much higher rate of female participants (59% compared to 3.5%). US is different compared to CT or MRI in that US is performed in real time and only select images are saved, compared to CT or MRI, which generates many images at predefined intervals for review. For example, a carotid US performed more cephalic would be less likely to visualize the thyroid gland and one performed more caudal would be more likely to visualize the thyroid gland, whereas a neck CT will always include the thyroid gland. One hypothesis is our low rate of reported thyroid incidentalomas on carotid US could be due to differences in US technique, and not representative of the larger population. This is a weakness of a single-center study and would need to be validated in a multicenter study.

A strength of our study is that it reviews only thyroid incidentalomas reported by radiology, either in the body of the report text or in the impression section of the report. This provides a clinically applicable perspective because clinicians are made aware of a thyroid incidentaloma only when it is included in the report. Prior studies have shown that not all verifiable thyroid incidentalomas are reported. In one study, imaging was reviewed on 500 randomly selected CT scans and 50 (10%) were found to have a thyroid nodule ranging in size from 0.5 to 2.2 cm, yet none of these nodules were reported in the radiology reports [2]. In another study, 16% of CT and MRI scans showed a verifiable thyroid nodule, yet only 6% of all imaging studies reported the nodule [11]. It has also been shown that there is high variability in radiologists’ reporting practices for thyroid incidentalomas when reported, including reporting in the body of the report only, reporting in the impression, and whether to give further recommendations such as dedicated thyroid US or biopsy [13]. Within our institution CT, MRI, and carotid US do not have standardized reporting templates and we therefore expect to see variability in reporting practices.

From 2000 to 2016 the number of veterans served by our health-care system increased steadily from 53 606 to 103 689. Data on the number of unique patients are unavailable before the year 2000. With this nearly 2-fold increase in patients, the number of included MRIs, carotid US, and contrast-enhanced chest CTs stayed relatively constant per year, whereas the number of noncontrast chest CTs per year increased 3-fold over this period (443 in 2000 and 1375 in 2016); see Fig. 1. This highlights shifting imaging patterns related to changes in practice patterns and implementation of lung cancer screening.

The American Thyroid Association recommends that thyroid US be performed in all patients with a suspected thyroid nodule, including thyroid incidentalomas, to answer the following questions: Is there truly a thyroid nodule? How large is the nodule? What are the sonographic characteristics of the nodule [6]? In contrast, The American College of Radiology does not recommend thyroid US in all patients with an incidental thyroid nodule on CT or MRI, and uses age and nodule size to determine whether a thyroid US is warranted [15]. This raises 2 key questions: First, should all verifiable thyroid incidentalomas be reported by radiology, and second, do all verifiable/reported thyroid incidentalomas warrant a formal thyroid US? One difficulty in making clinical decisions based on nondedicated thyroid US imaging is that CT has been shown to be an inaccurate modality for imaging thyroid nodules. One study examined the accuracy of characterizing thyroid abnormalities on CT compared with US and pathology, and found that no CT feature reliably distinguishes benign from malignant lesions, and that CT underestimates the number of nodules relative to US, though it did show that punctate calcifications on CT correlated with microcalcifications on US [12]. Our study better defines the extent to which thyroid incidentalomas are reported, a necessary first step toward looking at what is the appropriate evaluation of a thyroid incidentaloma. Future studies are needed to examine how thyroid incidentalomas are currently evaluated and what proper evaluation should include.

4. Conclusion

Our study confirms that roughly 4% to 5% of imaging studies of the chest and neck report a thyroid incidentaloma, consistent with 2 earlier smaller studies and in contrast to 2 previous large database studies. Reported thyroid incidentalomas are nearly twice as common in females, increase with age, and have become more common over the last 2 decades. Previous studies have found that thyroid incidentalomas have a significant rate of malignancy when further evaluated, yet reporting on incidental findings is variable and there are conflicting recommendations on the proper evaluation of these nodules because of limited research.

Glossary

Abbreviations

CT

computed tomography

MRI

magnetic resonance imaging

US

ultrasound

Additional Information

Disclosure Summary: The authors report no conflicts of interest for this work.