Locker David, Allen P Finbarr
Faculty of Dentistry, University of Toronto, 124 Edward Street, Toronto, Ontario M5G 1G6, Canada.
J Public Health Dent. 2002 Winter;62(1):13-20. doi: 10.1111/j.1752-7325.2002.tb03415.x.
Using the item-impact method, we developed an alternative short-form Oral Health Impact Profile (OHIP) that has good psychometric properties and minimal floor effects.
OHIP data were collected from a sample of older Canadians at two points in time. Data from the first administration were used to develop a 14-item short-form measure; data from the second compare the latter's psychometric properties with those of the original short form developed by Slade (1997), who used a controlled regression procedure.
The short form based on the item-impact method had only two items in common with the short form derived from the regression approach and contained more high-prevalence items. The regression short form was subject to marked floor effects, while the impact short form had floor effects comparable to those of the full 49-item OHIP. The former discriminated between dentate and edentulous subjects, while the latter did not. Both discriminated between dentate subjects who did and did not wear dentures, those with and without dry mouth, and those with and without chewing problems. Both were also significantly associated with self-ratings of oral health, satisfaction with oral health, and self-perceived need for dental treatment. The strength of the associations was somewhat stronger with the regression short form, indicating that it performed better as a discriminatory instrument. However, because of its floor effects, it was markedly less sensitive to change than the impact short form. There was an indication that item-impact methods of shortening oral health-related quality of life measures produced more stable results across samples than the statistical approach.
Because the content validity of short-form measures is always compromised, different short forms are required for different purposes and different patient populations. The regression short form developed by Slade (1997) is likely to be better when the aim is to discriminate, while the impact short form developed here may be preferable when the aim is to describe the oral health-related quality of life of populations or to detect change.
我们采用条目-影响法开发了一种替代性的简化版口腔健康影响程度量表（OHIP），该量表具有良好的心理测量特性且地板效应最小。
从加拿大老年人样本中在两个时间点收集OHIP数据。第一次施测的数据用于开发一个14条目的简化版量表；第二次施测的数据将后者的心理测量特性与斯莱德（1997年）开发的原始简化版量表进行比较，斯莱德采用的是一种对照回归程序。
基于条目-影响法的简化版量表与回归法得出的简化版量表仅有两个条目相同，且包含更多高患病率条目。回归简化版量表存在明显的地板效应，而影响简化版量表的地板效应与完整的49条目OHIP相当。前者能区分有牙颌和无牙颌受试者，而后者不能。两者都能区分佩戴和未佩戴假牙的有牙颌受试者、有和无口干的受试者以及有和无咀嚼问题的受试者。两者也都与口腔健康自评、对口腔健康的满意度以及自我感觉的牙科治疗需求显著相关。回归简化版量表的关联强度稍强，表明其作为一种区分工具表现更好。然而，由于其地板效应，它对变化的敏感度明显低于影响简化版量表。有迹象表明，缩短与口腔健康相关的生活质量测量量表的条目-影响法在不同样本中产生的结果比统计方法更稳定。
由于简化版量表的内容效度总会受到影响，不同目的和不同患者群体需要不同的简化版量表。当目的是进行区分时，斯莱德（1997年）开发的回归简化版量表可能更好，而当目的是描述人群的口腔健康相关生活质量或检测变化时，此处开发的影响简化版量表可能更可取。
