Impact of AAO-HNS Guideline on Obtaining Polysomnography Prior to Tonsillectomy for Pediatric Sleep-Disordered Breathing

Abstract

Objective

To assess the impact of the 2011 American Academy of Otolaryngology-Head and Neck Surgery (AAO-HNS) guideline on practice patterns in obtaining pre-adenotonsillectomy (AT) polysomnography (PSG) for pediatric sleep-disordered breathing (SDB).

Study Design

Retrospective cohort study.

Setting

Tertiary children’s hospital.

Subject and Methods

The study population included all patients referred to our pediatric otolaryngology clinic for consideration of AT for SDB during two 12-month time periods: before (2010–2011) and after (2015–2016) publication of the 2011 AAO-HNS guideline. Demographic, insurance, comorbidity, and Pediatric Sleep Questionnaire (PSQ) variables were assessed for association with pre-AT PSG using bivariate and multivariate logistic regression analysis.

Results

A greater percentage of patients underwent pre-AT PSG in 2015–2016 vs. 2010–2011 (30% vs. 22%, p=0.001). On multivariate analysis, presence of neuromuscular disorder was the only predictor associated with pre-AT PSG in 2010–2011 (OR 3.00, 95% CI [1.10, 8.06], p=0.03). Presence of neuromuscular disorder (OR 2.54, 95% CI [1.51, 4.29], p<0.0001), craniofacial anomaly (OR 2.32, 95% CI [1.20, 4.50], p=0.013), or Down syndrome (OR 3.45, 95% CI [1.54, 7.72], p=0.003) was associated with pre-AT PSG in 2015–2016. Positive PSQ screen was significantly associated with decreased odds of pre-AT PSG in both time periods.

Conclusion

After publication of the 2011 AAO-HNS guideline, there was a significant but modest increase in pre-AT PSG utilization in children with SDB and high-risk comorbidities at our institution, consistent with AAO-HNS guideline recommendations. However, overall guideline adherence remains low and may reflect limitations in PSG testing capacity.

INTRODUCTION

Over 750,000 adenotonsillectomies (AT) are performed annually in the United States, for which pediatric sleep-disordered breathing (SDB) is the indication in up to 77% of cases.^1–3 Pediatric SDB is a condition affecting up to 12% of children with presentations ranging from mild snoring to obstructive sleep apnea (OSA).⁴ Polysomnography (PSG) is considered the gold standard for the diagnosis of SDB and assessment of SDB severity.

Multiple organizations have published guidelines on the use of PSG prior to AT, including the American Academy of Pediatrics (AAP), American Academy of Sleep Medicine (AASM), and American Academy of Otolaryngology-Head and Neck Surgery (AAO-HNS). The 2002 AAP and 2010 AASM guidelines both recommended PSG to confirm a diagnosis of OSA in any child being considered for AT.^5,6 In contrast, the 2011 AAO-HNS Clinical Practice Guideline: Polysomnography for Sleep-Disordered Breathing Prior to Tonsillectomy recommended pre-AT PSG in a more selective group of children considered at high risk for perioperative morbidity or where there is uncertainty regarding the diagnosis or severity of SDB due to discordance between reported symptoms and physical examination findings. Risk factors for perioperative morbidity included obesity, neuromuscular disorders, Down syndrome, sickle cell anemia, and craniofacial anomalies. Furthermore, severe OSA on PSG was recommended as an indication for overnight monitoring.⁷ The recent 2019 AAO-HNS Clinical Practice Guideline: Tonsillectomy in Children included the additional recommendation for pre-AT PSG in children aged less than 2 years.⁸

Despite various guideline recommendations, the majority of children who undergo AT for SDB do not have a preoperative PSG.^9,10 A recent study at our institution demonstrated that in children undergoing AT for SDB in the 12 months immediately following publication of the 2011 AAO-HNS guideline, age less than 3 years was the only independent predictor of obtaining pre-AT PSG.¹¹ However important limitations of this study were assessment of a single time period immediately after 2011 AAO-HNS guideline publication and potential selection bias in including only patients who underwent AT. The purpose of the current study is therefore to assess the impact of the 2011 AAO-HNS guideline on practice patterns by comparing the predictors for obtaining preoperative PSG in all pediatric patients being considered for AT for SDB before and after publication of the guideline.

METHODS

This was a retrospective cohort study of all children who were referred to the multi-provider pediatric otolaryngology clinic at a single tertiary children’s hospital (Doernbecher Children’s Hospital at Oregon Health and Sciences University (OHSU)) for consideration for AT for SDB. The study was approved by the OHSU Institutional Review Board.

Patients were identified by screening a billing database for International Classification of Diseases, 9th Revision (ICD-9) and 10th Revision (ICD-10) codes corresponding to snoring, OSA, and tonsillar hypertrophy (ICD-9 codes 786.09, 474.10; ICD-10 codes R06.83, J35.3, G47.30, G47.33, G43.39, G47.36, R06.81). Inclusion criteria were: (1) age less than 18 years at the time of encounter and (2) primary indications for AT of either SDB or OSA. Exclusion criteria were: (1) non-SDB primary indication for AT including tonsillitis or recurrent streptococcal pharyngitis and (2) presence of other known airway disorders such as laryngomalacia, subglottic stenosis, or tracheobronchomalacia. Patients were divided into two groups: patients seen from September 1, 2010 to September 30, 2011 (prior to publication of 2011 AAO-HNS guideline) and patients seen from October 1, 2015 to October 18, 2016 (after publication of 2011 AAO-HNS guideline).

The electronic medical record (Epic Systems, Verona, Wisconsin, United States) was reviewed for demographic, insurance, comorbidity, Pediatric Sleep Questionnaire (PSQ), and PSG data. Demographic variables included age, which was grouped into age less than 3 and age greater than or equal to 3, and gender. Insurance types were grouped into Oregon Medicaid, Washington Medicaid or other, commercial, or none. Comorbidities assessed included neuromuscular disorder, craniofacial anomaly, Down syndrome, asthma, and congenital heart disease. Obesity was not assessed in this study due to a large proportion of patients with incomplete height and weight data in the electronic medical record. There were no patients meeting study criteria with mucopolysaccharidosis or sickle cell disease. Positive screen on PSQ was defined as a score greater than 0.33. The electronic medical record was reviewed for PSG tests ordered before and after each patient was seen in our pediatric otolaryngology clinic.

Bivariate analysis using Student’s t-test for continuous variables and Pearson’s chi-square test for categorical variables was used to test for significant differences between patients seen from 2010–2011 and those seen from 2015–2016. Univariate logistic regression was performed on each variable to assess for significant associations with after-visit PSG orders. Variables with associations approaching significance (P < 0.2) were included in a multivariate logistic regression model to assess for independent significant associations with after-visit PSG orders. All analyses were performed with Stata 12.1 (College Station, Texas, United States).

RESULTS

A total of 1,852 patients were referred to our pediatric otolaryngology clinic for consideration of AT in 2010–2011 and 2015–2016, of which 1,669 patients met study criteria. Of these, 475 patients were seen for consideration of AT for SDB in 2010–2011 vs. 1,194 patients in 2015–2016.

The overall mean age of the study population was 6.8 ± 3.8 years (Table 1). The mean age was slightly higher in the 2015–2016 group compared to the 2010–2011 group (7.1 ± 3.9 vs. 6.2 ± 3.3 years, P < 0.0001) (Table 1). The overall proportion of males in the study population was 55%. The majority of patients presented with Medicaid or commercial insurance, though the proportion of uninsured patients increased significantly between 2010–2011 and 2015–2016 (11% vs. 26%, P < 0.0001) (Table 1).

Table 1.

Cohort Characteristics by Time Period

Characteristic	Overall N = 1,669	Year 2010–2011 N = 475	Year 2015–2016 N = 1,194	p-value
Age (mean ± SD) Age <3 years	6.8 ± 3.8 215 (13)	6.2 ± 3.3 72 (15)	7.1 ± 3.9 143 (12)	<0.0001 0.08
Gender  Male  Female	922 (55) 747 (45)	250 (53) 225 (47)	672 (56) 522 (44)	0.18
Insurance  Oregon Medicaid  Commercial  WA Medicaid/Other  None	646 (39) 646 (39) 12 (1) 365 (22)	225 (47) 198 (42) 2 (0) 50 (11)	421 (35) 448 (38) 10 (1) 315 (26)	<0.0001
Comorbidities  Neuromuscular disorder  Craniofacial anomaly  Down syndrome  Asthma  Congenital heart disease	149 (9) 90 (5) 87 (5) 193 (12) 102 (6)	39 (8) 13 (3) 18 (4) 58 (12) 30 (6)	110 (9) 77 (6) 69 (6) 135 (11) 72 (6)	0.51 0.002 0.10 0.61 0.83

The proportion of patients with craniofacial anomalies in the study population increased significantly from 3% in 2010–2011 to 6% in 2015–2016 (P < 0.002) (Table 1). The proportion of patients with all other included comorbidities (neuromuscular disorder, Down syndrome, asthma, congenital heart disease) remained stable between the two time periods. Asthma (12%) and neuromuscular disorders (9%) were the most common comorbidities in the study population overall.

Approximately 16% of patients presented to our clinic with a history of previous PSG testing, which remained stable over time (Table 2). However, the overall proportion of patients in which an after-visit, pre-AT PSG was ordered increased from 22% in 2010–2011 to 30% in 2015–2016 (P < 0.001) (Table 2). The proportion of patients in which a PSG was ordered increased across most comorbidities including neuromuscular disorder (from 38% to 53%, P = 0.13), craniofacial anomaly (38% to 51%, P = 0.42), and congenital heart disease (50% to 53%, P = 0.80), however the only comorbidity for which this was statistically significant was asthma (17% to 39%, P = 0.004) (Table 2). Interestingly, the rate of testing among children with Down syndrome did not substantially change between the two time periods (56% to 57%, P = 0.94) (Table 2).

Table 2.

Frequency of Ordering PSG

PSG Ordered	Overall N = 1,669	Year 2010–2011 N = 475	Year 2015–2016 N = 1,194	p-value
Prior to clinic presentation	260 (16)	80 (17)	180 (15)	0.37
Prior to AT surgery	457 (27)	103 (22)	354 (30)	0.001
Comorbidities  Neuromuscular disorder  Craniofacial anomaly  Down syndrome  Asthma  Congenital heart disease	73 (49) 44 (49) 49 (56) 62 (32) 53 (52)	15 (38) 5 (38) 10 (56) 10 (17) 15 (50)	58 (53) 39 (51) 39 (57) 52 (39) 38 (53)	0.13 0.42 0.94 0.004 0.80

In multivariable regression analysis, the only significant positive predictor of pre-AT PSG in 2010–2011 was presence of a neuromuscular disorder (OR 3.00, 95% CI [1.10, 8.06], P = 0.03) (Table 3). In contrast, in 2015–2016 presence of a neuromuscular disorder (OR 2.54, 95% CI [1.51, 4.29], P < 0.0001), craniofacial anomaly (OR 2.32, 95% CI [1.20, 4.50], P = 0.013), or Down syndrome (OR 3.45, 95% CI [1.54, 7.72], P = 0.003) were all significant positive predictors of pre-AT PSG (Table 3).

Table 3.

Adjusted Associations with Ordering PSG

Characteristic	Year 2010–2011 OR [95%CI]	p-value	Year 2015–2016 OR [95%CI]	p-value
Positive screen on Pediatric Sleep Questionnaire	0.34 [0.18, 0.66]	0.001	0.54 [0.38, 0.77]	0.001
Age <3 years	0.79 [0.21, 3.01]	0.73	0.67 [0.41, 1.06]	0.12
Female gender	0.87 [0.49, 1.53]	0.63	0.93 [0.68, 1.26]	0.63
Insurance  Oregon Medicaid  Commercial  WA Medicaid/Other  None	Ref 1.49 [0.74, 3.00] Infinite 1.54 [0.64, 3.70]	-- 0.26 -- 0.33	Ref 1.19 [0.84, 1.68] 1.71 [0.35, 8.43] 0.71 [0.48, 1.06]	0.33 0.51 0.09
Comorbidities  Neuromuscular disorder  Craniofacial anomaly  Down syndrome  Asthma  Congenital heart disease	3.00 [1.10, 8.06] 1.08 [0.22, 5.30] 3.55 [0.52, 24.08] 0.79 [0.29, 2.16] 1.87 [0.52, 6.69]	0.03 0.93 0.19 0.64 0.34	2.54 [1.51, 4.29] 2.32 [1.20, 4.50] 3.45 [1.54, 7.72] 1.43 [0.90, 2.26] 1.55 [0.70, 3.46]	<0.001 0.013 0.003 0.13 0.28

^*Final multivariable model included provider, insurance type, age <3 years, gender, PSQ screening, and comorbidities listed above

Positive screen on PSQ was a consistently negative predictor of pre-AT PSG in both 2010–2011 (OR 0.34, 95% CI [0.18, 0.66], P = 0.001) and 2015–2016 (OR 0.54, 95% CI [0.38, 0.77], P = 0.001) (Table 3).

The remaining covariates included in our model (age less than 3 years, gender, and insurance type) did not have significant associations with obtaining pre-AT PSG (Table 3).

DISCUSSION

Previous survey-based studies have demonstrated significant variability among otolaryngologists in adherence to AAO-HNS guidelines on pre-AT PSG. Friedman et al. compared data from surveys of American Society of Pediatric Otolaryngology (ASPO) members in 2004 and 2011 and noted overall increased utilization of PSG for SDB from 2004 to 2011, but inconsistent adherence to the specific recommendations of the 2011 AAO-HNS guideline.⁹ A more recent survey of ASPO members in 2016 found that only 21% of members had altered their utilization of PSG for SDB in response to the 2011 guideline.¹²

Unlike previous studies that have relied on survey data, the current study provides an objective assessment of predictors of ordering PSG and adherence to the 2011 AAO-HNS guideline at our institution. A prior study from our institution demonstrated an overall low rate of adherence to the 2011 AAO-HNS guideline in the 12 months immediately after its publication, with age less than 3 years being the only positive predictor for pre-AT PSG.¹¹ Important limitations of this prior study were evaluation of a single time period immediately after guideline publication and inclusion of only patients who underwent AT. The current study assesses all patients seen in our clinic for evaluation of tonsillar hypertrophy or SDB, not only those who underwent AT, and directly compares two time points before and after guideline publication.

With this larger and less selected sample, we found significantly improved adherence to the guideline 4 years after publication of the 2011 AAO-HNS guideline compared to immediately prior to publication. Our results demonstrate a significant shift in utilization of PSG in patients presenting for consideration of AT for SDB. Prior to the guideline, the only positive predictor identified for pre-AT PSG was presence of neuromuscular disorder. After its publication, presence of neuromuscular disorder, craniofacial anomaly, or Down syndrome were all significantly associated with pre-AT PSG. Age less than 3 years was not an independent predictor of pre-AT PSG in the current study. Of note, age less than 3 years is not a recommended criterion for pre-AT PSG in the 2011 guideline but rather for hospital admission after AT.⁷ An important consideration is that this improvement in adherence may not be entirely driven by the 2011 guideline recommendations but by other factors such as increased access to PSG and a growing body of literature demonstrating associations of severe OSA and comorbidities such as neuromuscular disorders or obesity with increased risk of perioperative complications.¹²⁻¹⁸

It is worth noting that while our data demonstrates that adherence to the guideline has modestly improved, overall adherence remains low as evidenced by 51–57% rates of ordering PSG in patients with high-risk comorbidities after guideline publication. It is possible that the relatively small sample sizes of patients with neuromuscular disorders or craniofacial anomalies limited the statistical power to detect a difference between the two time periods in simple bivariate analysis. A smaller sample size in the earlier time period may similarly have limited the power to detect significant associations in the multivariable analysis. Examination of the point estimates of association among the comorbidities suggests that while Down syndrome was not significantly associated with PSG testing in the earlier time period, the odds ratio point estimate was very similar to the odds ratio in the later time period, which is consistent with the very similar proportion of Down syndrome patients tested in the two time periods. Taken together, these results suggest that while there was a net increase in the proportion referred for PSG among patients with these comorbidities, the overall increase was modest, and it remains unclear whether this can be attributed to the guideline recommendations or to other unmeasured temporal trends. As with other institutions across the country, a likely significant contributor to low rates of PSG ordering is limited sleep lab capacity with long wait times of up to several months for PSG.^10,11 Other potential explanations for overall limited adherence to the guideline include controversy regarding the role of PSG in the diagnosis and management of SDB,^3,19 high cost of PSG testing,³ and use of alternative pre-AT diagnostic tools such as questionnaires or pulse oximetry.^16,20

Interestingly, a positive PSQ screen was a strongly negative predictor for obtaining pre-AT PSG during both time periods. Patient history and physical examination have been shown to poorly correlate with PSG results,²¹ however more recent screening questionnaires have shown significantly greater potential accuracy in diagnosing SDB. The PSQ questionnaire has been increasingly used in the research setting as a way to systematically screen for symptoms of SDB.²² Various studies have demonstrated rates of sensitivity for the PSQ ranging from 78% to 85% and specificity ranging from 72% to 87%.^22–26 This level of sensitivity and specificity is not generally considered adequate as a clinical substitute for the gold standard overnight PSG in diagnosing OSA. However, the PSG itself has been shown to be an imperfect gold standard, particularly with regard to assessing the effects of SDB or OSA on cognitive, behavioral, and quality of life outcomes.^17,27,28 A recent study by Rosen et al. demonstrated that PSQ but not PSG data was predictive of improvements in executive functioning, behavior, sleepiness, and quality of life after AT.²⁹ Potential explanations for our finding of decreased likelihood of obtaining a pre-AT PSG in patients with a positive PSQ include reliance on strong clinical suspicion of OSA in patients who are clearly symptomatic with a positive PSQ screen, and parental desire to avoid delaying treatment in the setting of potentially long wait times for PSG.

It is important to recognize that the retrospective and single institutional nature of the current study does potentially limit the generalizability of its findings. Additionally, the small number of patients with certain comorbidities including craniofacial anomaly and Down syndrome in the pre-guideline group may have been inadequate to demonstrate significance in aspects of the statistical analysis. Another limitation of the current study was the inability to perform a complete multivariate analysis of all recommendations from the 2011 AAO-HNS guideline due to incomplete patient height and weight data and low prevalence of sickle cell disease and mucopolysaccharidosis in our patient population. Additional guideline recommendations for pre-AT PSG in patients with discordant clinical examination findings and reported severity of SDB were also not evaluated in the current study. Therefore, future study with prospective, multi-institutional data would be valuable in understanding the true rate of adherence to AAO-HNS guidelines.

CONCLUSIONS

After publication of the 2011 AAO-HNS guideline, there was a significant but modest increase in pre-AT PSG utilization in children with SDB and high-risk comorbidities at our institution. Consistent with guideline recommendations, neuromuscular disorders, craniofacial anomalies, and Down syndrome were all associated with increased likelihood of obtaining a pre-AT PSG after guideline publication. However, overall guideline adherence remains low and may reflect limitations in PSG testing capacity.