Are You REAL-ly Paying Attention? The Importance of Attention Controls

Written By

Reviewed By


Pyatak EA, Carandang K, Vigen CLP, et al. Occupational therapy intervention improves glycemic control and quality of life among young adults with diabetes: the resilient, empowered, active living with diabetes (REAL Diabetes) randomized controlled trial. Diabetes Care. 2018;41(4):696-704.

Critically-evaluating the literature is essential to engage in evidence-based practice.  A key component of assessing studies involves determining whether the comparator groups are appropriate.  Most pharmacists are familiar with the use of placebos for evaluating drug treatments, but how many of us have considered the comparator groups in behavioral interventions?  For these situations, employing attention placebo controls (APC) is important in order to assure the comparator groups are similarly treated and managed.


The term APC has appeared in clinical trials involving behavioral interventions with psychological and biological endpoints since the mid-1960's.1 APCs are used to address nonspecific effects that may be imparted when clinicians or study investigators pay attention to people. These effects are due to a type of psychological reactivity known as the Hawthorne effect, defined as the alteration of study subjects’ behavior due to their awareness that they are being observed.Notably, the Hawthorne effect may increase or diminish the magnitude of the effect depending on whether patients perceive scrutiny as a motivating or discouraging factor for a behavior.2


Hence, APCs are implemented to ensure interventions, not attention alone, are impacting the therapeutic outcome(s). Conditions set forth by the APC group should consider factors that may affect both intervention and control groups, including, but not limited to, the frequency, manner, and duration of contact with patients during the study.3 These variables are typically unaccounted for when researchers use certain control groups (Table 1), such as wait list, “usual” care, and no control groups.4  Without a well-matched attention placebo control, it is not possible to determine whether the observed benefits are due to the intervention (e.g. pharmacist’s care) or merely increased attention (e.g. three additional healthcare visits).


Table 1. Examples of Behavioral Control Groups Utilized in Clinical Trials4



Control Group Description

Attention Control

Mimics the amount of time and attention received by the intervention group.

No Control

No treatment at any point in the study. Behavioral or biological endpoints are still monitored.

Standard of Care

Appropriate, accepted, and widely used treatment per expert agreement promoted by researchers.

Usual Care

Behavioral intervention provided without direction from the researchers.

Wait List

Treatment withheld from the waiting list group until the intervention group treatment is completed, then the waiting list group is provided with the intervention. Behavioral and biological endpoints are monitored in both groups for the entire study.


Let’s take a look at a concrete example where the investigators employed an APC.  The study is entitled Resilient, Empowered, Active Living with Diabetes (REAL Diabetes) trial.5  REAL Diabetes was a randomized, controlled trial that assessed the impact of an intervention delivered by occupational therapists compared to APC on the glycemic control and psychosocial well-being of young adults diagnosed with type 1 or type 2 diabetes mellitus. Table 2 summarizes the trial’s intervention and APC groups. Individuals in the intervention group received in-person educational sessions tailored to the patients' individual goals according to their willingness to change, current medication regimen, and personal preferences. Individuals in the APC group received an initial home visit where they were provided with standard educational materials. The APC group was then contacted by the researchers biweekly using scripted phone calls to maintain the component of contact. Both groups received care (either the tailored care or APC care) for twelve weeks. The primary endpoint was the difference between A1c at baseline and six months. Secondary endpoints included psychological outcomes such as diabetes-related quality of life (QOL), diabetes distress, and depressive symptoms. The results are presented in Table 2.


Table 2. The REAL Diabetes Trial Results: Intervention vs Attention Placebo Control (APC)


REAL Study Component

Intervention Group

APC Group

Attention Factor

Type and timing of attention

10 – 16 hours of treatment distributed at the patients' convenience over 6 months.

Initial 15-minute home visit, then 11 biweekly phone calls for a total of 12 weeks.

Care Received

Individualized assessments and goal setting for living with diabetes, access and advocacy, activity and health, social support, emotions and well-being, and long-term health.

Standardized education materials from the National Diabetes Education Program and

Primary Outcome

(Change in A1c)



Secondary Outcomes

Diabetes-related QOL

(-9 [max negative impact] to +9 [max positive impact])



Glucose monitoring (days/week)

(Range 0-7 days)



Medication adherence (days/week)

(Range 0-7 days)



Diabetes distress

(PAID-SF; range 0-20)



Life satisfaction

(SWLS; range 5-25)



Depressive symptoms

(PHQ-8; range 0-27)



Abbreviations = QOL: quality of life; PAID-SF: Problem Areas in Diabetes- short form; SWLS: Satisfaction with Life Scale; PHQ-8: Patient Health Questionnaire 8


The results of the REAL Diabetes study demonstrated improvements in glycemic control and changes patients behaviors and quality of life similar to “pharmacist care” studies.  But was the APC in REAL Diabetes properly implemented? The Hawthorne effect may positively or negatively impact behavior, making APCs difficult to design.2 There are two major concepts that should be considered when constructing APCs to avoid confounding. First, the attention received by the control group should be reasonably equivalent to that of the intervention group.6 Second, the elements of care provided to the APC group should be perceived as credible but should not influence responses from the participants.7,8


To address the first requirement, investigators typically match the total time spent with the placebo group to that received by the active treatment group.6 In the REAL Diabetes intervention group, occupational therapists provided a minimum of ten hours of treatment using the prescribed content modules and were given the liberty to extend the intervention to up to sixteen hours for individuals with complex care needs. The frequency of visits with patients in the intervention group was delivered on "an individual basis." Therefore, the duration and frequency of contact were flexible in the intervention group. In contrast, calls delivered to the APC group were at fixed intervals (e.g. every two weeks) and did not include a prescribed range of contact hours. Thus, there was potentially a difference not only in total time spent with patients in each of the two groups, but also in the quality of the time. The impersonal nature of phone follow-up (vs. in-person visits) may lack a necessary component(s) to build a working therapeutic relationship between the patient and provider.9,10 If the APC does not involve the same quality or degree of attention then it becomes questionable if it is truly meets the first criteria.


To address the second criteria for constructing an appropriate APC, investigators must not only adjust for the attention factor, but also ensure the attention control is credible and will not influence the behavior of the study participants.7,8 Many researchers have argued that the construction of a behavioral placebo is impossible, while others believe it a worthwhile endeavor.11 The difficulty is creating a control which has the façade of a credible treatment but does not include any of the “active” components delivered to the intervention group. This is challenging, as motivated patients tend to volunteer for studies, and may confound the passive APC component by seeking the active treatment components. Additionally, passive components could be unethical in certain situations, such as withholding medical advice.3 In REAL Diabetes, the active components of motivational interviewing and personalized diabetes self-management education were replaced with self-guided readings and biweekly phone calls in the APC group. It is possible that the APC participants felt that the treatment provided was of poor quality because it required them to learn the material “on their own.”


To assess whether the APC was credible and did not influence behavior, behaviorial studies sometimes employ three treatment arms: an intervention group, an APC group, and a “usual care” group, where care is provided in a manner that’s considered “routine” in the specific practice.4 In this type of design, the three arms are compared to assess the equipoise of the active (intervention) and passive (APC) components.  The REAL Diabetes study did not include three comparator groups.  However, to their credit, the investigators did compare the baseline demographics and number of treatment sessions in the invention and APC groups – all of which were similar.


Unlike “usual care” and other comparators commonly employed which do not control for the amount of attention participants received, APCs are a more rigorous approach to creating control groups in clinical trials.  However, they are not without their own set of limitations and can be confounded by differences in patients’ perceptions of attention. The credibility of the APC must be carefully considered. Additionally, post-hoc analyses should be performed to assess for other unanticipated confounders. When evaluating (or designing) research studies regarding ambulatory care pharmacist services, the appropriateness of the control group should be considered. Does the lack of APCs in most “pharmacist care” studies diminish the findings?  Should we routinely employ APCs in future health services research? Let us know what you REAL-ly think!