Beebe Timothy J, Talley Nicholas J, Camilleri Michael, Jenkins Sarah M, Anderson Kari J, Locke G Richard
Survey Research Center, Department of Health Sciences Research, Mayo Clinic College of Medicine, Rochester, Minnesota 55905, USA.
Med Care. 2007 Oct;45(10):959-65. doi: 10.1097/MLR.0b013e31805468b0.
There has been speculation that the Health Insurance Portability and Accountability Act (HIPAA) has muted participation in research but little direct evidence to substantiate those claims exists.
To evaluate the effects of including a HIPAA authorization form (HAF) on multiple measures of survey performance. RESEARCH DESIGN, SUBJECTS: A community survey of Olmsted County, Minnesota residents conducted between September 2005 and April 2006. A total of 6939 cases were randomly assigned to 2 experimental conditions where half of the subjects received a 1-page HAF (n = 3469) and the other half did not (n = 3470).
Response rate, nonresponse bias, and data quality.
At the end of data collection, a response rate of 39.8% was observed in the HAF condition and 55.0% in the No HAF condition (P < 0.0001). There was a negligible but statistically significant (P < or = 0.001) over-representation of males in the No HAF condition but no difference in the amount of missing data between the 2 groups. However, reports of general health and the percentage of respondents indicating that they were nonsmokers were both significantly (P < or = 0.01) lower in the No HAF condition than in the HAF condition.
Inclusion of a minimally burdensome version of the HAF reduced survey response rates by up to 15 percentage points. This could have implications for a study's statistical power. There was little evidence that the form affected nonresponse bias or data quality.
有人猜测《健康保险流通与责任法案》（HIPAA）抑制了研究参与度，但几乎没有直接证据证实这些说法。
评估纳入一份HIPAA授权表（HAF）对多项调查绩效指标的影响。
研究设计、对象：2005年9月至2006年4月对明尼苏达州奥尔姆斯特德县居民进行的一项社区调查。总共6939个案例被随机分配到2种实验条件下，其中一半受试者收到一份1页的HAF（n = 3469），另一半没有收到（n = 3470）。
应答率、无应答偏倚和数据质量。
在数据收集结束时，HAF组的应答率为39.8%，无HAF组为55.0%（P < 0.0001）。无HAF组男性的过度代表性可忽略不计但具有统计学意义（P ≤ 0.001），但两组之间缺失数据量没有差异。然而，无HAF组的总体健康报告以及表示自己不吸烟的受访者百分比均显著低于HAF组（P ≤ 0.01）。
纳入一份负担最小化版本的HAF使调查应答率降低了多达15个百分点。这可能会对研究的统计效力产生影响。几乎没有证据表明该表格会影响无应答偏倚或数据质量。
