Zhong Qiuyue, Gelaye Bizu, Fann Jesse R, Sanchez Sixto E, Williams Michelle A
Department of Epidemiology, Harvard School of Public Health, 677 Huntington Ave, K505 Boston, MA, USA.
Department of Epidemiology, Harvard School of Public Health, 677 Huntington Ave, K505 Boston, MA, USA.
J Affect Disord. 2014 Apr;158:148-53. doi: 10.1016/j.jad.2014.02.012. Epub 2014 Feb 10.
We sought to evaluate the validity of the Spanish language version of the patient health questionnaire-9 (PHQ-9) depression scale in a large sample of pregnant Peruvian women using Rasch item response theory (IRT) approaches. We further sought to examine the appropriateness of the response formats, reliability and potential differential item functioning (DIF) by maternal age, educational attainment and employment status.
This cross-sectional study was conducted among 1520 pregnant women in Lima, Peru. A structured interview was used to collect information on demographic characteristics and PHQ-9 items. Data from the PHQ-9 were fitted to the Rasch IRT model and tested for appropriate category ordering, the assumptions of unidimensionality and local independence, item fit, reliability and presence of DIF.
The Spanish language version of PHQ-9 demonstrated unidimensionality, local independence, and acceptable fit for the Rasch IRT model. However, we detected disordered response categories for the original four response categories. After collapsing "more than half the days" and "nearly every day", the response categories ordered properly and the PHQ-9 fit the Rasch IRT model. The PHQ-9 had moderate internal consistency (person separation index, PSI=0.72). Additionally, the items of PHQ-9 were free of DIF with regard to age, educational attainment, and employment status.
The Spanish language version of the PHQ-9 was shown to have item properties of an effective screening instrument. Collapsing rating scale categories and reconstructing three-point Likert scale for all items improved the fit of the instrument. Future studies are warranted to establish new cutoff scores and criterion validity of the three-point Likert scale response options for the Spanish language version of the PHQ-9.
我们试图使用拉施克项目反应理论（IRT）方法，在大量秘鲁孕妇样本中评估患者健康问卷-9（PHQ-9）抑郁量表西班牙语版本的有效性。我们还进一步研究了反应格式的适当性、可靠性以及按母亲年龄、教育程度和就业状况划分的潜在项目功能差异（DIF）。
这项横断面研究在秘鲁利马的1520名孕妇中进行。采用结构化访谈收集人口统计学特征和PHQ-9项目的信息。将PHQ-9的数据拟合到拉施克IRT模型，并测试类别排序是否合适、单维性和局部独立性假设、项目拟合、可靠性以及DIF的存在情况。
PHQ-9西班牙语版本表现出单维性、局部独立性，并且对拉施克IRT模型具有可接受的拟合度。然而，我们检测到原始四个反应类别存在反应类别无序的情况。将“超过一半天数”和“几乎每天”合并后，反应类别排序正确，且PHQ-9符合拉施克IRT模型。PHQ-9具有中等内部一致性（个人分离指数，PSI = 0.72）。此外，PHQ-9的项目在年龄、教育程度和就业状况方面不存在DIF。
PHQ-9西班牙语版本显示出有效筛查工具的项目特性。合并评分量表类别并为所有项目重建三点李克特量表提高了该工具的拟合度。有必要开展进一步研究，以确定PHQ-9西班牙语版本三点李克特量表反应选项的新临界分数和效标效度。
