Data Items and Collection Process
Information obtained from each article pertaining to the primary outcomes of interest included authors, year of publication, number of study participants, mean age, BMI, pre- and post-op AHI and LSAT (Table 1). BMI was documented in terms of either kg/m2, percentile, or Z-score across all studies; for consistency, we converted BMI units to either “normal,” “overweight,” or “obese.” Normal BMI is defined as BMI < 85 percentile for age or a Z-score < 1, overweight is defined as BMI between 85 and 95 percentile for age or a Z-score between 1 and 2, and obese is defined as BMI > 95 percentile for age or a Z-score > 2.9  For studies which included a variety of single or multi-level surgical sites (e.g., lingual tonsillectomy alone, or revision adenoidectomy plus turbinoplasty), only outcome measures that involved tongue surgery were extracted for review. These measures were primarily from PSG data including mean AHI change or LSAT change after surgery (primary outcome), and the percentage of subjects who were successfully treated by surgery (responders, secondary outcome). The definition of a responder in this study depends on preoperative AHI. If the preoperative AHI > 5 events/hr then response was defined as post-operative AHI < 5 and/or a 50% reduction in AHI. Patients with a preoperative AHI of < 5 were considered responders if postoperative AHI was <1 which is considered surgical cure. This definition was adapted from the definition used by He et al.15