目的

建立广东地区妇女乳腺癌预测模型,筛选高危人群,为乳腺癌的早期干预提供线索。

方法

在门诊患者和正常体检的妇女中,采用简单随机的方法选取35～60岁乳腺癌病例284例,以1∶2成组配对原则在相应年龄段内选取非乳腺癌妇女673例,进行回顾性病例对照研究。分析数据时,首先将数据按绝经前后分层;以是否为乳腺癌患者为结果变量,其余变量为自变量,分别通过l ogistic单因素、多因素回归分析筛选出绝经前、后乳腺癌的危险因素,并以此建立绝经前、后妇女乳腺癌预测模型;用判别分析法分别对绝经前、后预测模型的预测能力进行判别。

结果

绝经前妇女患乳腺癌的主要危险因素包括服用避孕药史、亲属中有乳腺癌患者、有不良情绪、有不良事件、劳动强度大、无乳腺组织活检史、很少参加运动等。绝经前妇女患乳腺癌的预测模型为:1/{1+exp[-(-4.314+1.333 服用避孕药+1.167亲属中有乳腺癌患者+0.404有不良情绪+0.613有不良事件+0.754劳动强度大-0.716有乳腺组织活检史-0.839经常运动)].。此模型有统计学意义(χ²=165.83,P=0.000),并且其预测不发生乳腺癌的正确率为74.2%,预测发生乳腺癌的正确率为58.4%。绝经后妇女患乳腺癌的主要危险因素包括亲属中有乳腺癌患者、文化程度低、情绪调节能力差等。绝经后妇女患乳腺癌的预测模型为:1/{1+exp[-(1.211+2.269 亲属中有乳腺癌患者-1.617文化程度高-0.635情绪调节能力强)].。此模型有统计学意义(χ²=116.32,P=0.000),并且其预测发生乳腺癌的正确率为78.9%,不发生乳腺癌的正确率为55.2%。

结论

绝经前、后妇女患乳腺癌的影响因素不完全相同。绝经后妇女乳腺癌预测模型对预测发生乳腺癌有一定的参考价值;绝经前妇女乳腺癌预测模型对预测不发生乳腺癌有一定的参考价值。有效的乳腺癌预测模型还需要整合更多的临床和生物学指标。

Objective

To establish a predicting model for breast cancer in order to screen high risk wo men and pr ovide cl ue f or early inter vention of breast cancer in Guangdong.

Methods

A retrospective case-control study was used to identif y risk f actors of breast cancer.Fr o m outpatients and wo men of physical exa mination,284 breast cancer patients aged fro m 35 to 60 years were rando mly selected and 673 nonbreast cancer patients at the sa me age fro m t he sa me population were selected accor ding to 1∶2 paired gr ouping principle.Breast cancer was dependent variable,and all other variables wereindependent variables.For data anal ysis,all cases were firstly divided into a pre menopausal group and a post menopausal gr oup.Si mple logistic regression and multivariate l ogistic regression models were applied to filter t he risk factors of breast cancer respectively.Then t hese i mportant f actors were used to establish t he premenopausal and post menopausal risk f orecast models respectively.The t wo models were evaluated by discri minant analysis.

Results

In the premenopausal wo men gr oup,the main risk factors of breast cancer incl uded medicine-contraception,breast cancer-family history,bad emotion,undesirable life,excessive wor king,absence of breast biopsy,less exercise,etc.The equation of logistic regression was:1/{1+exp[-(-4.314+1.333 medicine-contraception +1.167 breast cancer-fa mily histor y+0.404 bad emotion+0.613 undesirable life+0.754 excessive wor king-0.716 breast biopsy history-0.839 regular exercise)]..This model was significant in predicting breast cancer(χ²=165.83,P=0.000＜0.010).The correct rate of predicting non-breast cancer was 74.2%,and the correct rate of predicting breast cancer was 58.4%.In the post menopausal group,the main risk factors included breast cancerf a mily histor y,low education,and t he absence of e motion-adjusting ability.The equations of logistic regression was:1/{1+ exp[-(1.211+2.269 family history-1.617 high education-0.635 high ability of emotion adj ust ment)]..This model was significant in predicting breast cancer(χ²=116.32,P=0.000).The correct rate of predicting breast cancer was 78.9%,and t he correct rate of predicting non-breast cancer was 55.2%.

Conclusions

There were different infl uencing factors f or breast cancer bet ween pre menopausal and post menopausal wo men.The predictive model f or post menopausal wo men has val ue to predict risks of breast cancer;t he predictive model f or premenopausal wo men has val ue to predict non-breast cancer;however a efficacious predicting model of breast cancer still needs integrate biology and clinic index.