Abstract
E
Although the importance of patient-reported outcomes may seem obvious, to date only a few studies have incorporated them in study design. In fact, a 2015 review could identify only 4 out of 103 ongoing CLS trials that assessed psychosocial outcomes, including treatment satisfaction, acceptance, and satisfaction with CLS technology, fear of hypoglycemia, and general QoL. 10 A review of published CLS trials between 2015 and 2016 demonstrates some improvement in this area. Using the search terms “Artificial Pancreas” or “Bionic Pancreas” or “Closed Loop” not “Review” and “Diabetes” on PubMED, we identified five CLS articles reporting psychosocial outcomes as primary endpoints 11 –15 and two with these outcomes as secondary endpoints. 5,8 The psychosocial measures assessed in trials varied, with most using previously validated questionnaires, 11 –13,15 several presenting new (or study-specific) questionnaires, 8,13,15 and others reporting qualitative data on participants' experience and perception of CLS technology. 5,11,12,14 Although these studies reflect increasing recognition of the need for psychosocial evaluation in CLS trials, most were narrow in the scope of examined outcome variables.
The study reported by Sharifi et al. 9 is exemplary in its assessment of a broad range of psychological and behavioral outcome variables that reflect different aspects of QoL, ranging from emotional status to cognitive function to sleep quantity and quality. For emotional status and well-being, two diabetes-specific constructs were measured at baseline and after treatment conditions—diabetes-related distress and fear of hypoglycemia. Treatment satisfaction associated with each device was also assessed, which is critical because user satisfaction is influenced not only by objectively measured clinical outcomes but also by the individual's subjective perceptions concerning ease of use, functional usefulness, burden, and benefit. For investigation of sleep quality and quantity, Sharifi et al. 9 utilized both subjective and objective measures. While undergoing trials with SAP-LGS and CLSs, participants wore actigraphy devices, which were used to derive a number of measures of sleep, including latency, total time, number of awakenings, and sleep efficiency. After each condition, participants also completed subjective ratings of their sleep quality. This is the first study to systematically investigate the possible impact of glucose control systems on sleep quality and quantity. Reduced sleep disturbance may be an unrecognized but important outcome from the patient perspective, especially for those individuals whose sleep is disrupted by the occurrence of or anxiety related to nocturnal hypoglycemia (including parents of children with T1DM). Relatedly, it is important to assess whether diabetes technology substantially inhibits sleep through device alarms and the bulk of additional equipment, a qualitative outcome reported by some patients in another study. 12
Sharifi et al. 9 also obtained objective measures of cognitive function two times each day (am and pm) during both 4-day SAP-LGS and CLS trials, using a computerized complex choice reaction time task. In addition to task performance parameters (speed and accuracy), participants gave subjective ratings of their perceived mental alertness at each cognitive test. No previous studies have introduced the possibility that nighttime use of SAP-LGS or CLS devices could have an impact on daytime alertness and cognitive functioning, secondary to improvements in either sleep or glycemic control (e.g., reduced hypoglycemia). Ideally, future longer term CLS studies will follow this example and assess a broad range of psychosocial and behavioral outcomes to gain a deeper understanding of the impact of diabetes technology on the daily lives and experiences of individuals using devices.
The Sharifi et al. 9 study also highlights the inherent complexity involved in conducting a thorough assessment of patient-centered outcomes in diabetes technology research. It may involve far more than administering well-validated pre-post psychosocial questionnaires (although these are certainly essential). In addition to quantitative data, qualitative data may be important to obtain a more comprehensive and deeper understanding of the person's experiences of technology use. Other methodological approaches to operationalizing relevant outcomes may be needed, such as ecological momentary assessment, which involves taking measures of mood, symptoms, function, and behavior at the time they are occurring in real-world circumstances. 16,17 Sharifi et al.'s 9 method of simultaneously obtaining subjective and objective measures of cognitive function during the day exemplifies this approach. Their employment of actigraphy to assess sleep variables demonstrates the use of sophisticated nonglycemic technology to measure a behavior. This multimodal methodological approach to measuring patient outcomes can be challenging and it is worth noting that the research team who designed and conducted this study included behavioral scientists, who are well trained in the assessment of complex psychological and behavioral variables.
Despite its rigorous methods and inclusion of relevant psychosocial and behavioral variables, it should also be noted that the Sharifi et al. 9 study found no changes in patient-reported outcome measures, with the one exception of increased treatment satisfaction for adults in the CLS condition. Most likely the short duration of the study conditions (four nights) was a major contributor to these results. However, characteristics of the participant sample may also have played a role. As the authors noted, baseline scores on diabetes-related distress and fear of hypoglycemia were quite low in their participants, indicating that diabetes was having a minimal negative effect on their emotional status. Participants with HbA1c levels more than 8.4% (69 mmol/mol) were excluded, and all had a history of or current use of diabetes technology (e.g., insulin pumps, CGM, or SAP).
Past studies have almost exclusively tested CLS in select, highly sophisticated patient samples who tend to have greater levels of diabetes knowledge and are highly engaged in intensive diabetes management, including the use of insulin pumps. There is a compelling need to study individuals who are not as technologically “savvy” or inclined. These patients are likely to have more difficulty adapting to the technology, but could potentially benefit from these tools the most. There is also a need to study individuals who are experiencing significant levels of diabetes-related emotional distress and to investigate ways of using CLS and other diabetes technologies to help patients struggling with suboptimal diabetes management and control. However, it is important to remember that we are still in the early stages of scientific investigation into the potential of artificial pancreas systems to improve the lives of people with diabetes. With the improvement of device safety and reliability, expansion of trial sample sizes and inclusion, and increases in the length and number of home use trials, the CLS research community will be better able to address these important questions. The study by Sharifi et al. 9 and recent others offer encouraging signs that research is moving in the direction of recognizing the essential role human factors and QoL play in achieving this potential.