Background

Early school dropout is an economic, social, and individual problem. School dropout is a result of cumulative processes that occur over many childhood years. Despite the influence of level of education on health outcomes, primary prevention of dropout outside of the school setting is rare. In the Netherlands, the Youth Health Care (YHC) service may play a role in primary prevention of school dropout.

Objective

We hypothesized that data collected by YHC on family background and Strength and Difficulties Questionnaire (SDQ) scores at ages 10 and 14 is predictive of school dropout.

Methods

We analyzed Dutch YHC data from 24,988 children born in 1996 − 200. Early school dropout was defined as having left school without diploma by the age of 17. Two multilevel logistic regression models were built with predictors measured at the ages of 10 and 14. The model performance was assessed using ROC curve.

Results

A child’s SDQ was a strong predictor of early school dropout, in addition to gender and parents’ socio-economic status at age 10 and age 14. Models showed moderate prediction performance (ROC value 0.70/0.69, respectively).

Conclusions

The proposed prediction models are based on only few routinely collected socio-demographic factors and SDQ scores. We found these models can contribute to risk stratification by YHC as early as age of ten. This provides a window of opportunity for interventions that aim to strengthen school engagement. Further research and practical efforts to expand the set of predictors available to YHC (e.g., school performance) are expected to improve the quality of this prediction.

中文翻译：

---

校外辍学预测：荷兰青年保健服务机构进行早期风险分层的可能性。回顾性队列研究的结果

背景

早期辍学是一个经济、社会和个人问题。辍学是许多童年时期累积过程的结果。尽管教育水平对健康结果有影响，但在学校环境之外对辍学进行初级预防的情况很少见。在荷兰，青年保健（YHC）服务可能在辍学的初级预防中发挥作用。

客观的

我们假设 YHC 收集的关于 10 岁和 14 岁家庭背景和优势与困难问卷 (SDQ) 分数的数据可以预测辍学。

方法

我们分析了来自 1996 年至 200 年间出生的 24,988 名儿童的荷兰 YHC 数据。早期辍学被定义为在 17 岁之前没有文凭就离开了学校。建立了两个多级逻辑回归模型，其中预测变量在 10 岁和 14 岁时测量。使用 ROC 曲线评估模型性能。

结果

除了性别和父母在 10 岁和 14 岁时的社会经济地位之外，儿童的 SDQ 是早期辍学的有力预测因素。模型显示出适度的预测性能（ROC 值分别为 0.70/0.69）。

结论

所提出的预测模型仅基于少数常规收集的社会人口因素和 SDQ 分数。我们发现这些模型早在 10 岁时就可以帮助 YHC 进行风险分层。这为旨在加强学校参与的干预措施提供了机会之窗。为扩大 YHC 可用的预测因素集（例如，学校表现）而进行的进一步研究和实际努力有望提高该预测的质量。