统计问题第132问：非随机干预性试验

Question

In March 2008, a regulation was introduced in New York City that required restaurants to provide calorie information on all food purchases. The regulation applied only to restaurants with 15 or more stores nationwide. Researchers assessed the effects of the regulation on individual purchases. The primary outcome was the energy content of individual purchases, based on customers’ register receipts and the information provided on menus within the restaurant.

A random sample of 168 locations of the most popular fast food chains in New York City was obtained. For nine weeks in the spring of 2007 and then again in the spring of 2009, all adult lunchtime customers at the sampled locations were interviewed. In total, 7309 adult customers were interviewed in 2007 and 8489 in 2009. The mean number of calories purchased after the regulation was slightly higher than that before, although the difference was not statistically significant (846v 828; P=0.22).

Which one of the following best describes the above study design?

·a) Non-randomised controlled trial.

·b) Cohort study.

·c) Before and after study.

·d) Cross sectional study.

提示：正确答案只有一个。

Answer

The above study is best described as a before and after study (answer c).

A non-randomised study evaluates the effectiveness of an intervention, comparing outcomes for those who received the intervention with those who did not. As the name suggests, people are not randomly allocated to comparison groups. People may be allocated using a non-random method—for example, alternate allocation—or be within a group as a result of choice by the study participants themselves. Many types of non-randomised study exist. The study design may be experimental, with researchers deliberately intervening and recording the effects of an intervention. An example of an intervention in an experimental study design is the introduction of a therapeutic regimen by the researchers. Non-randomised studies may also be observational, with researchers observing the effects of peoples’ behaviour on their health. Interventions in observational studies include lifestyle choices such as smoking. Answersa, b, c, and d are all examples of non-randomised study designs.

In a before and after study (answer c), the effects of an intervention are evaluated by comparing the outcomes of study participants investigated before the intervention with those measured afterwards. The participants investigated before and after the intervention may be the same people or different samples. The aim of the above study was to assess the effects of the regulation on the energy content of individual purchases. The distinguishing feature of the study was the time line of events—a sample of residents of New York City were interviewed before the intervention, the regulation was then introduced, and another sample was interviewed afterwards. Because the intervention was deliberate, the above study has an experimental design.

A non-randomised controlled trial (answer a) has an experimental design with the non-random allocation of participants to comparison groups. Typically the two groups would be followed prospectively and investigated concurrently. A non-randomised controlled study design could have been used to study the effects of the regulation described above. A sample of cities would be allocated in a non-random manner to the intervention (implementation of regulation) or control group. A random sample of locations of the most popular fast food chains in each city would then be taken, with all adult lunchtime customers being interviewed over the same period.

A cohort study (answer b) is observational in design. A defined group of people—the cohort—is followed prospectively over a study period and the subsequent medical history of these people recorded. Cohort studies investigate whether cohort members exposed to a risk factor have an increased or decreased risk of disease compared with those members not exposed to the risk factor. For example, a cohort study might investigate whether dietary fibre intake (high or low) was associated with the development of diverticular disease over a 20 year period. Although dietary fibre intake is a lifestyle choice, it is regarded as an intervention.

A cross sectional study (answer d) is carried out at a single point in time, with the purpose of describing peoples’ behaviour, attitudes, or beliefs regarding a particular matter. Two cross sectional studies were undertaken in the above study, one before and another after the regulation. However, because these two cross sectional studies were separated by the intervention, answer c best describes the study design.

Care is needed when interpreting the results of the above study. Any changes in food purchases may not have resulted from the regulation but from other differences between the two samples or periods of observation. For example, there may have been changes in the menus offered and their energy content, or differences in weather may have influenced energy intake. It is also possible that people interviewed before and after the regulation had different dietary preferences. More generally, in non-randomised studies participants are not randomly allocated to groups, so the groups may not be comparable in their demography or characteristics, and this may affect the outcome. Therefore, any differences between the groups in outcome may not result from the intervention but from other differences between the groups.

所以答案是选择  c 

每天学习一点，你会更强大！