Children's Hospital Boston, Division of Endocrinology, 300 Longwood Ave, Boston, MA 02115, USA.
Pediatrics. 2010 Nov;126(5):938-44. doi: 10.1542/peds.2009-3649. Epub 2010 Oct 25.
Recombinant human growth hormone (GH) is approved for treatment of children with idiopathic short stature, and endocrinologists often depend on algorithms to predict adult height. Because algorithm performance often is included in treatment decisions, we sought to evaluate agreement among height prediction formulas.
We identified 3 commonly used algorithms for height prediction, the Bayley-Pinneau, Roche-Wainer-Thissen, and Khamis-Roche methods. We constructed simulated samples of children with typical distributions of ages, heights, weights, bone ages, and parental heights seen in patients with idiopathic short stature, and we applied the algorithms to the simulated sample to determine whether predicted adult height was <160 cm for boys or <150 cm for girls (<1.2nd height percentiles for adults).
We found substantial disagreement among algorithms in the proportions of simulated cases with predicted adult heights of <1.2nd percentile, a cutoff value that may influence GH treatment decisions. With the Bayley-Pinneau formula, 43% of boys and 81% of girls had predicted adult heights below this threshold; with the Khamis-Roche method, only 3% of boys and 0.2% of girls had predicted heights of <1.2nd percentile. Roche-Wainer-Thissen predictions were between those values. Overall agreement of the methods was poor (κ = 0.21) for boys and negative for girls.
Wide variation exists among formulas used to predict adult heights. Because these algorithms may be used in decisions regarding whether to initiate GH treatment and assessment of the efficacy of GH in research trials, it is important for parents, pediatricians, and investigators to recognize the considerable variation involved in height predictions.
重组人生长激素（GH）获批用于治疗特发性身材矮小的儿童，内分泌学家通常依赖算法来预测成年身高。由于算法性能通常包含在治疗决策中，我们试图评估身高预测公式之间的一致性。
我们确定了 3 种常用的身高预测算法，即 Bayley-Pinneau、Roche-Wainer-Thissen 和 Khamis-Roche 方法。我们构建了具有特发性身材矮小患者中常见的年龄、身高、体重、骨龄和父母身高分布的模拟样本，并将这些算法应用于模拟样本，以确定男孩的预测成年身高是否<160cm 或女孩的预测成年身高是否<150cm（<成人身高第 1.2 百分位数）。
我们发现，在预测成年身高<第 1.2 百分位数的模拟病例比例方面，算法之间存在很大差异，这一截断值可能会影响 GH 治疗决策。使用 Bayley-Pinneau 公式，43%的男孩和 81%的女孩的预测成年身高低于这一阈值；而使用 Khamis-Roche 方法，只有 3%的男孩和 0.2%的女孩的预测身高<第 1.2 百分位数。Roche-Wainer-Thissen 的预测值介于两者之间。这些方法的总体一致性对于男孩较差（κ=0.21），对于女孩为负。
用于预测成年身高的公式存在很大差异。由于这些算法可能用于决定是否开始 GH 治疗以及评估 GH 在研究试验中的疗效，因此父母、儿科医生和研究人员认识到身高预测中存在的大量差异非常重要。
