PubMed and MEDLINE® on Web of Science were searched for published manuscripts using search terms (thyroid cancer [MeSH Major Topic]) AND (incidental OR asymptomatic OR 'chance finding' OR accidental OR serendipitous OR coincidental). All English language articles on human studies of thyroid cancer risk and/or prognosis in relation to the mode of detection published from the inception of the databases until 24 March 2020 were considered for inclusion. An updated search was carried out on 5 April 2021 on PubMed using search terms ('thyroid cancer' or 'thyroid nodule' [MeSH Major Topic]) AND (incidental OR asymptomatic OR 'chance finding' OR accidental OR serendipitous OR coincidental). The bibliographies of manuscripts included in the review were also screened.
Screening and Assessing the Eligibility of Articles
The process of screening, inclusion and exclusion of manuscripts is shown in Figure 1. After removing duplicate manuscripts, titles and abstracts of remaining manuscripts were assessed for eligibility. Only primary studies comparing incidental and nonincidental detection of thyroid nodules and its impact on either the risk and/or prognosis of thyroid cancer were included. Studies on select cohorts (such as radiation-exposed population, multiple endocrine neoplasia type 2 patients, pregnant patients, microcarcinoma), nonmalignant thyroid nodules, secondary research papers (reviews), registry-based studies, single-arm studies, correspondence, editorial papers, animal studies, non-English language literature and studies lacking relevant data were excluded. All excluded manuscripts at every stage of the screening process were checked by a second researcher and any discrepancy was resolved through consensus or discussion with the senior author.
Flow chart of the literature search strategy for the systematic review (adapted from Preferred Reporting Items for Systematic reviews and Meta-analysis flow diagram17)
Definitions for Exposures and Outcomes
Incidental detection was defined as the detection of thyroid nodule(s)/cancer(s) on imaging for reasons unrelated to the thyroid. Studies reporting on other types of incidental detection (such as nodule/cancer detected on thyroid cancer screening or autopsy study) were excluded. Nonincidental detection was defined as thyroid nodule(s)/cancer(s) detected following investigation of symptoms (such as palpable thyroid nodules detected by the patient or a third party, local compressive symptoms such as difficulty swallowing or hoarse voice) or discovered during the evaluation of the thyroid gland.
The primary outcome for the risk studies was the occurrence of thyroid cancer, as defined in the manuscripts and included both histological and cytological assessments. For the prognosis studies, all described prognostic outcomes were included and were classed as direct prognostic markers that were time-dependent (such as disease-free survival, overall survival, residual/recurrence disease); and indirect prognostic markers that were not time-dependent (such as the size of the nodule, histological subtypes, lymph node and distant metastasis, lymphovascular/capsular invasion, extrathyroidal extension, bilaterality, multifocality/multicentricity, cancer staging, lymphocytic infiltration, distant Metastasis, patient Age, Completeness of resection, local Invasion, tumour Size [MACIS] and age, metastases, extent, size [AMES] scores).
Data Extraction and Analysis
A word-based data extraction form was developed and piloted in five studies and subsequently modified and used by two observers to extract the data from all included manuscripts. Discrepancies were resolved by consensus or addressed by the senior author. The final data set (available upon request) was transferred to an excel spreadsheet. All included studies were assessed for 'risk of bias' using the modified Newcastle–Ottawa Scale (NOS) for quality assessment, taking into account the different observational study designs (case-control studies, cohort studies and cross-sectional studies).
Descriptive reporting was done using Excel. Odds ratios and relative risks not reported in individual studies were calculated by the researcher; but testing for statistical significance was not performed. Two meta-analyses were performed using Review Manager v5.4 (https://tinyurl.com/y7kepe2e) on risk to produce a summated odds ratio. The first meta-analysis included case-control studies with comparable study populations and definitions of incidental detection (n = 4) whereas the second reported on data from good quality case-control studies (NOS score of >50%; n = 3). The three retrospective cohort studies were not subjected to a meta-analysis due to heterogeneity in populations and unit of the study presented (nodules/patients). Two of these studies had NOS scores of less than 50%. Meta-analysis was not carried out for the prognosis studies due to the heterogeneity in the definitions of incidental detection, subtypes of thyroid cancer and the outcomes assessed.
The Preferred Reporting Items for Systematic reviews and Meta-analysis (PRISMA) statement guided the planning and conduct of this systematic review and the PRISMA checklist (shown in Tables S1A,B) and flow diagram were used for reporting. This systematic review was registered with the PROSPERO database (Registration number CRD42020172291).
Clin Endocrinol. 2022;96(2):246-254. © 2022 Blackwell Publishing