Thyroidectomy in Pediatric Patients with Graves’ Disease: A Systematic Review of Postoperative Morbidity

Literature Search

The literature search was performed using the electronic databases PubMed and Embase in April 2020 to identify studies evaluating short- and long-term postoperative morbidities in pediatric GD patients undergoing thyroidectomy. The following keywords were used: (Graves’ Disease) AND (Child OR Adolescent OR Juvenile) AND (Thyroidectomy OR Thyroid surgery OR Surgery). The exact search strategy used for both the PubMed and Embase databases is available as supplemental data. No time frame was used for the literature search. The search was extended by scanning the references of relevant articles and using the “related article” function of the databases.

Study Selection and Quality Assessment

Title and abstract screening was performed independently and blinded by A.S.Z. and C.F.M. using the Rayyan web app for systemic reviews [9]. Full text screening was performed by A.S.Z. and discussed with C.F.M., and in case of discussion, a third reviewer (A.S.P.T.) was consulted. Original studies on pediatric and/or adolescent GD patients who underwent a thyroidectomy reporting on postoperative morbidities in a minimum of 10 GD patients were included. When a study reported data on postoperative morbidities in pediatric patients who underwent a thyroidectomy because of a benign thyroid disorder, and the studied cohort consisted of ≥80% GD patients but data were not specified for GD patients only, the study was included as well. Full text availability was the last inclusion criterion. Exclusion criteria were conference abstracts, review articles, and articles published in another language than English, Dutch, German, French, Italian, or Spanish. Part of the full text screening was a critical appraisal to evaluate the quality per study following the Newcastle-Ottawa Scale (NOS), a quality assessment form for cohort studies. As (almost) all studies had a retrospective and observational character without comparison to another cohort, we did not score for comparability of the cohort(s). Studies were rated as low, fair, or good quality based on the number of stars scored in the selection and outcome domain: 3 stars for good quality, 2 stars for fair quality, and 0 or 1 star for low quality. Only studies that scored 3 stars in both domains were rated as overall good quality.

Data Extraction

Data extraction was performed by A.S.Z. and was double-checked by C.F.M. Data on hypocalcemia (transient and permanent), RLN injury (transient and permanent), infection, hemorrhage/hematoma, and keloid development were collected. Furthermore, the used definitions and diagnostic criteria for the reported morbidities were collected from each included study. The differences herein will be discussed in the Results and Discussion sections. Although the diagnostic criteria and methods differed between the studies, we calculated the cumulative reported frequency of each postoperative morbidity to give an indication of the overall frequency of postoperative morbidity.

Study Characteristics

The PubMed and Embase searches were performed on April 14, 2020, and yielded 1,728 unique articles. 1,623 articles were excluded after title and abstract screening, and an additional 83 articles were excluded after full text screening. Twenty-two mainly retrospective observational cohort studies were included in this review (Fig. 1). Based on the quality assessment, the quality of 4 of the included studies was rated as good [10-13], whereas the quality of the other 18 studies was rated as fair (see online suppl. Table 1; see online Supplementary Materials).

View Full Size

Flowchart illustrating the results of the literature search on studies evaluating postoperative morbidity in pediatric patients GD. GD, Graves’ disease.

Citation: European Thyroid Journal 10, 1; 10.1159/000511345

• Download Figure
• Download figure as PowerPoint slide

These 22 studies reported on the postoperative morbidity of 1,424 children and adolescents. The majority of the studied patients were female (Table 1). Thirteen studies focused mainly on total or near-total thyroidectomy (n = 705) [10, 11, 13-23]. In addition, Lobe and Wright [24] reported results in pediatric patients who underwent total thyroidectomy via a transaxillary endoscopic approach (n = 31). Subtotal thyroidectomy was performed in 5 studies (n = 550) [16, 25-28]. Four studies did not specify the surgical method used in the patients with GD (n = 110) [12, 29-31]. The mean duration of follow-up after surgery ranged widely among the studies (1–100 months). Cohen et al. [23] received additional follow-up data by calling patients who were not seen in the clinic over the previous 2 years. Table 1 also shows in which studies thyroidectomies were performed by a high-volume or a low-volume surgeon. Eight studies reported that the thyroidectomy was performed by a high-volume surgeon [10, 11, 13-15, 21, 29, 31]. In the study by Cohen et al. [23], surgery was performed by a low-volume surgeon. The remaining 13 studies did not report whether thyroidectomies were performed by a low- or a high-volume surgeon [12, 16-20, 22, 24-26, 28, 30]. In 3 studies, thyroidectomies were performed by a pediatric surgeon [14, 21, 22], in 3 studies by an endocrine surgeon [11, 26, 29], and in 4 by both pediatric and adult surgeons [12, 18, 19, 31]. In the other studies, the surgeon’s background was not specified.

Table 1.

Characteristics of included studies evaluating postoperative morbidities of thyroidectomy in pediatric patients with GD

View Table

In most studies, the majority of patients were initially treated with ATDs prior to thyroidectomy. Some studies specified the indication for thyroidectomy in their patients, including drug refractory hyperthyroidism, history of thyroid storm, noncompliance to ATD treatment, side effects of ATDs, compressive symptoms due to goiter, a coexisting suspicious thyroid nodule, concomitant germinoma, or the patient’s choice. In the study by Sherman et al. [16], 3 patients underwent thyroidectomy after failure of RAI treatment to achieve euthyroidism. Preoperative thyroid hormone levels were reported by Cohen et al. [23] and indicated hyperthyroidism (mean free T4 3.2 ng/dL [41.19 pmol/L] and mean total T3 338.7 ng/dL [5.20 nmol/L]). Patients in the study of Peroni et al. [11] received preoperative preparation with Lugol’s solution and were euthyroid or hypothyroid as assessed by TSH measurement. Sherman et al. and Breuer et al. [14, 16] did not report thyroid hormone concentrations before thyroidectomy, but their patients were preoperatively prepared with beta-blocker, Lugol’s solution, or saturated solution of potassium, probably achieving euthyroidism or hypothyroidism at the moment of thyroidectomy. Söreide et al. [27] reported that medical treatment before surgery consisted of iodine (93% of the patients), beta-blocker (73%), and ATD treatment (62%), but they did not specify thyroid hormone concentrations at the time of surgery. Sugino et al. [26] reported all patients to be euthyroid at the time of surgery, without reporting the used preoperative treatment protocol. The other included studies did not report on preoperative preparation or preoperative thyroid hormone levels in their patients [10, 13, 15, 17-22, 24, 25, 28-31]. Further characteristics of the included studies are shown in Table 1. The definitions and diagnostic methods used for the individual postoperative morbidities varied widely between the studies or were not specified in some of the studies (Table 2). None of the studies reported cases of postoperative mortality.

Table 2.

The definitions and diagnostic methods used for postoperative morbidities in included studies on thyroidectomy in children with GD

View Table

Hypocalcemia

Transient Hypocalcemia

The reported frequency of transient hypocalcemia ranged from 5.0 to 50.0%, with a cumulative reported frequency of 22.2% (269 out of 1,210 patients) (Table 3). Baumgarten et al. [21] reported transient hypocalcemia and transient hypoparathyroidism separately, with a reported frequency of transient hypoparathyroidism of 22% (n = 22 out of 123). De Jong et al. [20] reported both hypocalcemia at 24 h after thyroidectomy (48.1%) and hypoparathyroidism on discharge (46.2%; n = 24 out of 52). Zobel et al. [19] (n = 38) reported hypocalcemia immediately after surgery in 17 GD patients, with only 1 patient still being hypocalcemic after 6 months. This patient regained normal parathyroid function 1 year after surgery. This observation of regaining normal parathyroid function 1 year after surgery was also described in most children who underwent a total thyroidectomy because of another benign thyroid disorder or because of thyroid cancer [19]. Chen et al. [29] stated that GD was an independent risk factor to develop postoperative hypocalcemia in children undergoing a thyroidectomy (OR 3.99; p = 0.02).

Table 3.

The reported frequency of postoperative morbidities after thyroidectomy in pediatric patients with GD

View Table

RLN Injury

Infection

The frequencies of postoperative infections were reported in 7 studies [10-12, 16-18, 27]. Only 3 out of 344 evaluated patients were reported with a postoperative infection (cumulative reported frequency 0.9%) (Table 3). In 1 patient, the infection required reoperation [9]; in 1 patient, it required open drainage [16]; and 1 patient was treated with antibiotics [11].

Hemorrhage/Hematoma

Hemorrhage, postoperative bleeding, or a hematoma as a postoperative complication of thyroidectomy in pediatric GD patients was evaluated in 13 studies [10-12, 14-18, 21, 24-27]. The reported frequency of hemorrhage ranged from 0 to 4.8%, with a cumulative reported frequency of 1.7% (17 out of 980 patients) (Table 3). Despite the low reported frequency of hemorrhage after thyroidectomy, GD patients seem to have a higher risk of postoperative bleeding compared to patients who underwent a thyroidectomy for another cause (RR = 8.7 [95% CI: 1.06–71.85]; p = 0.02) [20]. A possible explanation for the higher risk of hemorrhage in GD patients may be the hypervascularity of the thyroid gland in these patients.

Keloid Development

Only 4 of the included studies evaluated the frequency of postoperative keloid development after thyroidectomy in pediatric GD patients [11, 18, 23, 25]. The reported frequency ranged from 0 to 7.4%, with a cumulative reported frequency of 3.7% (3 out of 82 patients) (Table 3).

Hypocalcemia

Hypocalcemia is the most common complication after thyroidectomy in pediatric GD. The primary cause of hypocalcemia is damage to or devascularization of the parathyroid glands during surgery. Therefore, in some studies, “hypoparathyroidism” and not “hypocalcemia” was reported as a postoperative morbidity. Transient hypocalcemia, which normally resolves within 6 months after thyroidectomy, can be distinguished from permanent hypocalcemia when calcium or active vitamin D supplementation (either oral or intravenous) is required for a longer time [32]. Although we calculated a cumulative frequency of hypocalcemia, it is important to realize that the definition of hypocalcemia was not uniform throughout the various included studies (Table 2). Cutoff serum levels for diagnosing hypocalcemia ranged from 1.87 to 2.15 mmol/L (equal to 7.5–8.7 mg/dL) in the included studies. In 10 studies, the diagnosis of hypocalcemia was solely based on serum calcium levels [10-12, 15, 18, 20-23, 29]. In contrast, in some studies, the presence of clinical symptoms (hand paresthesia, perioral numbness, muscle cramps, and Chvostek’s sign) was combined with serum calcium levels to diagnose postoperative hypocalcemia [14, 19, 26]. Elfenbein et al. [31] diagnosed hypocalcemia solely based on the presence of clinical symptoms (numbness or tingling of the hands, feet, or mouth after surgery) that improved after calcium supplementation. Overall, uniformity considering diagnostic criteria and cutoff values of hypocalcemia was missing. Furthermore, 10 of the included studies considered hypocalcemia to be permanent if recovery has not occurred within 6 months [10, 11, 13, 18, 20, 21, 23, 29-31]. Yu et al. [22] discussed this cutoff of 6 months because in their cohort, recovery of hypocalcemia occurred as late as 33 months. In addition, Cohen et al. [23] reported cases of “permanent” hypocalcemia that restored after 24 months and 5.7 years. Yu et al. [22] recommended to prolong the follow-up to at least 1 year for determining the permanence of hypocalcemia. This recommendation is supported by the finding of Zobel et al. [19] that most patients regain normal parathyroid function within 1 year after surgery. Based on the results of this systematic review, we suggest considering hypocalcemia as permanent if recovery has not occurred within 6 months after surgery. However, one should realize that parathyroid function can normalize even several years after thyroidectomy [21, 22]. We recommend to use a uniform cutoff value to diagnose hypocalcemia after thyroidectomy. It is our clinical experience that postoperatively patients often experience a transient drop in serum calcium levels that resolves quickly and does not lead to clinical signs of hypocalcemia. We suggest maintaining different cutoff levels for defining hypocalcemia post-thyroidectomy depending on the presence or absence of symptoms; patients with clinical symptoms of hypocalcemia (serum calcium level <2.0 mmol/L [8.0 mg/dL]) versus patients without clinical signs of hypocalcemia (serum calcium level <1.9 mmol/L [7.5 mg/dL]). This will prevent unnecessary calcium supplementation and shorten postoperative hospitalization.

Transient hypocalcemia is the most frequently reported morbidity after thyroidectomy performed in pediatric GD. In contrast, permanent hypocalcemia is relatively rare. This finding may be explained by the fact that the damaged parathyroid glands heal in the first months after surgery. The studies by Cohen et al. and de Jong et al. [20, 23] reported remarkably high complication rates of permanent hypocalcemia: 20.0 and 17.4%. Cohen et al. [23] hypothesized that the relatively high complication rate of permanent hypocalcemia in their cohort may be explained by the fact that poorly controlled GD patients were studied, which may increase the risk of postoperative complications. Furthermore, a low number of annually performed thyroidectomies by the surgeons in this study were raised as an additional explanation for the high complication rate. None of the surgeons in the study of Cohen et al. [23] performed >7 thyroidectomies in any given year. According to the American Thyroid Association (ATA) guideline, it is not recommended to perform thyroidectomies if there is lack of access to a high-volume surgeon (>30 thyroidectomies per year) [5]. A study in adult patients reported that GD was a significant predictor of postoperative complications in surgeries performed by low- and intermediate-volume surgeons [35]. If there is no access to high-volume surgeons, the ATA favors RAI therapy above thyroidectomy as definitive treatment for GD [5]. In addition to the guidelines, research has demonstrated that thyroidectomies performed by high-volume surgeons have better outcomes with less morbidity [36, 37]. The results of the other included studies illustrate that the chance of permanent hypocalcemia after thyroidectomy by high-volume surgeons in pediatric GD patients indeed is low. One should realize that the size of the underlying population has to be large for a pediatric surgeon to be able to perform >30 pediatric thyroidectomies per year. The study by de Jong et al. [20] also reported a relatively high incidence of permanent hypocalcemia (17.4%). In their study, the number of parathyroid glands remaining in situ (PGRIS) during surgery was monitored. Patients with GD had the highest incidence of a PGRIS score below 4 (52%) compared to patients with other indications for thyroidectomy. A 2-fold increase of the risk of long-term hypoparathyroidism was seen in patients with a PGRIS <4 (OR = 2.61 [95% CI: 1.01–6.73]; p = 0.04) [20]. As PGRIS scores were not reported in the other included studies, we cannot further evaluate the hypothesis that the number of PGRIS is an important predictor for the development of permanent hypocalcemia after thyroidectomy.

RLN Injury

The second most frequently reported complication after thyroidectomy in pediatric GD patients is transient RLN injury. The RLNs are located dorsal of the thyroid gland and are easily harmed during thyroid surgery through a variety of mechanisms (i.e., stretching, crushing, or transection of the RLN) [38]. Nerve regrowth may occur slowly (up to 12 months) but may not always be sufficient to overcome damage [38]. The RLN innervates both abductors and adductors of the vocal cord. Because the simultaneous contraction of these antagonist muscles is important, even light damage may cause inappropriate coordination and result in deterioration of the voice. Swallowing, coughing, and respiration may also be affected in RLN injury [38].

In the 18 included studies evaluating RLN injury, the methods to diagnose RLN injury varied or were not specified (Table 2) [10-12, 14-18, 21, 23-31]. Machens et al. [10] determined each patient’s vocal cord function with a laryngoscopy before and after surgery and used intraoperative nerve monitoring. Some studies performed direct laryngoscopy to visualize normal or abnormal vocal cord movement in patients who were experiencing a soft voice after surgery or if transection of the RLN was immediately recognized during surgery [14, 18, 26, 29]. Sinha et al. [12] reported that laryngoscopic confirmation was only performed in those patients with permanent hoarseness or a postoperative stridor and not in those with a temporary change of their voice.

The reported frequency of transient RLN injury ranged between 0 and 11.1%, with 1 study reporting a higher frequency of 20.0% (n = 2) [25]. The true frequency of RLN injury in the included studies is debatable because of heterogeneity in, or absence of information on, the used diagnostic methods. To identify RLN injury, it is recommended to perform a laryngoscopy to visualize the function of the vocal cords in combination with clinical symptoms [33, 38]. A laryngoscopy is recommended due to the fact that RLN injuries can be subclinical [38]. However, some clinical RLN injuries after thyroidectomy occur in the absence of vocal cord paralysis proven with laryngoscopy [39]. Only Machens et al. [10] and Chiapponi et al. [15] performed a pre- and postoperative laryngoscopy or a vocal cord function test in the pediatric patients who underwent thyroidectomy for GD. Machens et al. [10] reported frequencies of transient and permanent RLN injury of 3.4 and 0.0%, whereas Chiapponi et al. [15] reported frequencies of 4.8 and 0.0%, respectively. These results are likely to accurately reflect the complication rates of RLN injury after thyroidectomy performed by a high-volume surgeon in pediatric GD, due to their reliable diagnostic methods. We would recommend standardized methods to identify RLN injury in future studies to get a better impression on the true complication rates after thyroidectomy and to gain insight into the relationship between clinical signs of RLN neuropraxia and findings at laryngoscopy. On the other hand, in light of the very low frequency, one may question the necessity of diagnostic methods like postoperative laryngoscopy to identify possible RLN injury. Since transient hoarseness may also be caused by intubation, we suggest to consider laryngoscopy only if patients suffer from long-lasting deterioration (>6 months) of the voice or other clinical signs of RLN neuropraxia, not resolving after speech therapy. The diagnosis of transient RLN injury in children should then solely be made on the basis of clinical signs of RLN neuropraxia, avoiding invasive diagnostic procedures like laryngoscopy in pediatric patients.

Due to the variable diagnostic criteria and definitions of the studied morbidities, variance in cutoff values for hypocalcemia in the included studies, and original data were missing or not available, it was not possible to report the pooled complication rates of all 1,424 studied patients. To give an indication of the real frequency of postoperative morbidity, we have calculated cumulative frequency rates, based on the diagnostic criteria used in each study. This is the major limitation of our review. Also, the inclusion of 3 studies containing a few patients without GD (n = 14) may have influenced the results. Therefore, we recommend that future prospective studies on morbidities of thyroidectomy in pediatric GD patients should use widely accepted standardized diagnostic criteria, as proposed in this review. We have decided only to include studies with a minimum of 10 pediatric GD patients to reduce the risk of publication bias of seldom seen morbidities in very small cohorts or individual cases. However, by doing this, we may have missed some data on postoperative morbidities after thyroidectomy for pediatric GD. The same applies to data on pediatric GD patients hidden in mainly adult series that were not included in this review. The quality of the vast majority of the included studies was scored as fair mainly based on their retrospective and observational character. Future well-designed prospective studies are needed to improve the quality of evidence reporting on postoperative morbidities in pediatric GD patients. Nevertheless, to our knowledge, this systematic review is the most elaborate review on short- and long-term postoperative morbidities after thyroidectomy in children with GD and therefore gives a good overview of both short- and long-term postoperative complications in pediatric GD patients.

In conclusion, based on the results of this review, thyroidectomy in pediatric patients with GD has low postoperative morbidity. The minority of pediatric patients had transient forms of hypocalcemia and RLN injury. Permanent postoperative morbidities are relatively rare. Therefore, we conclude that thyroidectomy in pediatric GD patients is a safe treatment option, especially if performed by high-volume surgeons.