Full text

Investigators designing intervention studies make many decisions that influence the credibility and interpretation of research results. One key decision is the selection of an appropriate control group, or comparison group. For an intervention study to demonstrate that a treatment is effective or superior to standard practice, the intervention group should be compared with the correct control group (Lindquist, Wyman, Talley, Findorff, & Gross, 2007; Malay & Chung, 2012). Western Journal of Nursing Research (WJNR) reviewers often comment on comparison groups chosen for intervention research reports. This editor’s special article highlights some considerations for selecting comparison groups.

Comparison with the ideal control group decreases the influence of outside factors and biases on outcome variables and increases internal validity. Researchers should consider various conditions as they select a control or comparison group. These include study goals, the presence and quality of existing interventions, potential influence of attention on outcome variables, available study subjects and sites, and fiscal resources (Barkauskas, Lusk, & Eakin, 2005).

True control groups, otherwise referred to as no-treatment control groups, receive no intervention treatment (Barkauskas et al., 2005). However, denying the treatment intervention would be unethical in some cases. Also, true control participants may lose interest in the study or seek out treatment. True control groups are a viable choice for intervention studies when withholding treatment presents no ethical concerns, when the intervention arena lacks an accepted “usual care,” and when no other interventions with documented effectiveness are available.

Attention control groups receive attention similar to the treatment group but do not receive the active intervention content. Ideally, attention control groups receive an identical dose of attention as the treatment group. Such groups may help minimize the influence of non-specific factors, such as relationships between interventionists and subjects or visits to a clinic, that could affect outcomes (Donovan, Kwekkeboom, Rosenzweig, & Ward, 2009). This type of control group should be selected when attention or other non-specific factors may influence outcome variables.

Participants in placebo control groups do not receive the experimental intervention, but instead receive what appears to be an intervention. Sham interventions are one type of placebo comparison group used in device research (La Vaque & Rossiter, 2001). Participants receive an apparent treatment similar in form to the intervention group; however, the placebo lacks the active ingredient and so is expected to have no effect on study outcomes (Barkauskas et al., 2005). Placebo control groups are a rigorous and efficient design that tends to minimize subject bias and investigator bias (European Medicines Agency, 2001). Placebo subjects may be less likely to be lost to attrition than true control subjects because they may experience or believe they experience some benefit from the placebo. Researchers could select placebo control groups when a placebo effect is suspected, when withholding treatment presents no ethical or legal concerns, or it is required for the intervention approval, such as in some drug/device studies (Barkauskas et al., 2005).

Participants in usual care comparison groups, sometimes referred to as existing-practice or standard-care comparison groups, receive the usual treatment designed for their circumstances; meanwhile, intervention group participants receive a different treatment intervention (Barkauskas et al., 2005; Freedland, Mohr, Davidson, & Schwartz, 2011; Lindquist et al., 2007). Issues associated with usual care groups include the possibility of great variability in usual care and thus usual care outcomes; ceiling effects if usual care is very effective; and achieving a sample size large enough to detect a difference between groups due to the effectiveness of usual treatment (Barkauskas et al., 2005). Comparing a new intervention with usual care establishes whether the new intervention is better than existing treatments. This is more important knowledge than whether the new intervention is better than no treatment. Select this type of comparison group when withholding usual care is unethical and when the effectiveness of the intervention treatment can be compared with usual care.

Minimal-intervention comparison groups receive minor enhancements to usual care conditions (Freedland et al., 2011), typically to address methodological or ethical issues that accompany ordinary usual care. Therefore, this type of comparison group is appropriate when methodological or ethical concerns exist related to implementing usual care alone. A common concern with this type of comparison group is enhancing usual care so much that it resembles an intervention, making it difficult to detect outcome differences between comparison and intervention groups.

Active comparison/control groups receive an existing treatment that has been shown to be effective, while the intervention group receives a new type of treatment that is suspected to have the same or better effects (European Medicines Agency, 2001). Active comparison participants usually receive the same amount of attention from research staff as intervention participants (Lindquist et al., 2007). This type of comparison group should be selected when the objective of the study is to compare the efficacy of two treatments (European Medicines Agency, 2001).

Participants in component comparison groups, or dismantling comparison groups, receive part of an existing efficacious intervention (e.g., education on diet alone), while the intervention group receives all of the existing intervention (e.g., exercise plus the education on diet, Lindquist et al., 2007; Street & Luoma, 2002). Select this type of comparison group when the goal is to develop a more cost-effective intervention that maintains effectiveness. Alternatively, participants in additive comparison groups receive an additional and new treatment component to an existing intervention (e.g., adding an educational component to a supervised physical activity intervention), while the intervention group receives the existing intervention alone. This type of comparison group serves well when the goal is to increase the effectiveness of an existing intervention.

Dose-level comparison groups receive the same treatment as the intervention group, but at a different dose (Barkauskas et al., 2005). For example, the comparison group may receive two contacts with study staff, while the treatment intervention group receives six contacts. Attrition is common issue associated with this approach when participants not receiving an effective dose-level drop out of the study. This type of comparison group works best when dosages can be controlled and their effects measured.

Wait-list control groups begin the study as true control, attention control, or usual care groups while the intervention group receives treatment. Then, after the intervention group’s outcome measures have been collected, the wait-list group receives the same treatment as the intervention group (Barkauskas et al., 2005; Lindquist et al., 2007). Some studies collect outcome data from the wait-list control subjects after they receive the intervention to compare pre-intervention with post-intervention variables. Wait-list control designs may reduce attrition if control subjects believe the treatment will be effective. However, these designs may increase the cost by implementing the treatment to both groups and increasing the length of the study.

External control groups use participants from a different population than the intervention group (European Medicines Agency, 2001). External control participants could be a group from which data were collected at an earlier time (i.e., historical controls) or concurrently from a different setting (European Medicines Agency, 2001; Malay & Chung, 2012). External control groups generally are true control groups that receive either no intervention or usual care conditions, but they sometimes fit the criteria for other comparison groups. A common concern about external control studies is their lack of rigor. Because random assignment to groups is not possible, baseline characteristics of treatment and control groups may differ significantly, and other population-based or external factors may influence the intervention effects. Researchers may select external control groups when an internal control group (i.e., concurrent participants from the same population as the intervention group) cannot be obtained, but researchers should attempt to match characteristics of treatment and control subjects.

A study using any type of control or comparison group may demonstrate efficacy of the treatment intervention by showing that it is superior to the comparison group (European Medicines Agency, 2001). However, depending on the study goals and other factors, certain comparison or control groups may be more applicable for a particular study than others because they reduce the influence of outside factors and biases on outcome variables (Lindquist et al., 2007; Malay & Chung, 2012). Authors of intervention research should fully describe the control or comparison condition, including the rationale for the comparison group they select. WJNR welcomes intervention research reports because they are essential for developing nursing knowledge.

Vicki S. Conn, PhD, RN, FAAN University of Missouri, Columbia, MO, USA Tamara Coon Sells, PhD University of Missouri, Columbia, MO, USA

END

来源 | 健康界

版权归原作者所有，若有违规、侵权请联系我们