The Tool That Worked Once: A Goal-Setting Trial in Complex Pediatric Care
A stepped-wedge trial gave 67 families a web tool to set and track care goals for medically complex children. Nearly all used it in the room; most never touched it after. The signal that survived came from the conversation, not the software.
Dr. Sven Jungmann
CEO
Thirty-four of the thirty-five families used the tool in the clinic. Then nineteen of them never used it again. That gap — almost universal uptake in the room, near-total abandonment after the visit — is the most instructive thing this trial produced, and it deserves more attention than the headline result that the authors were chasing.
The clinical territory is unforgiving. Children with medical complexity carry several chronic conditions, see several specialists, and depend on coordination that no one owns. The system forgets between appointments; parents re-narrate the same history each visit, and the shared sense of what the family is working toward quietly comes apart. The study, in JMIR Human Factors, tests whether a more structured way of setting goals can keep that thread from breaking.
The design, stated plainly
This was a stepped-wedge cluster randomized controlled trial: eleven providers at a single academic centre — across two outpatient clinics, pediatric neurology and complex primary care — switched from usual care to the tool at staggered, randomly assigned times rather than all at once. Sixty-seven parent-child dyads enrolled (32 control, 35 intervention); 27 in each arm remained at follow-up. Enrolment ran from April 2019 to December 2020, with follow-up through March 2021. The intervention itself, a web platform called GoalKeeper, has two halves: one used face-to-face in the visit, where the clinician helps a parent turn worries and hopes into specific, measurable, time-bound goals; and one built for tracking those goals between visits. The primary outcome was the quality of goal-setting one month later, scored on the goal-setting domain of the Patient Assessment of Care for Chronic Illness Care (PACIC), a validated parent-reported questionnaire.
The half that never ran
Start with the part that did not happen, because it frames everything else. After leaving the clinic, parents logged a median of zero entries in the tracking module (interquartile range 0–3). Nineteen of the thirty-five entered nothing; thirteen entered data between one and ten times; three used it more than ten. So whatever benefit the trial detected cannot be credited to between-visit follow-up — that machinery scarcely turned over. The continuity feature, as deployed, did not drive the effect, because it barely operated.
It is tempting to conclude that the single in-clinic conversation therefore did all the work. That reading is plausible, and the data lean toward it, but a stepped-wedge trial of this size was not built to pit one component against the other and cannot prove which ingredient mattered. The honest claim is the narrower one: the part designed to extend care across time went largely unused, and the part bound to a human encounter is what was left.
A split verdict on what helped
On the primary outcome the result is a near miss. Goal-setting quality held roughly steady in the intervention group and fell in the control group over the month — a change of +0.03 versus −0.43 on the PACIC scale — but the difference reached only a trend (F49=3.52, P=.06), short of the conventional threshold. The secondary composite measure of overall care quality pointed the same way and did clear the bar: +0.01 versus −0.48 (F49=4.28, P=.04). The defensible reading is modest. Families who received the structured conversation did not slide the way families under usual care did.
That decline in the control arm is the quiet finding underneath. In one month, with nothing added, parents of these children reported measurably worse goal-setting and care quality. The tool did not so much lift families as arrest a drift that ordinary care left unchecked — a smaller claim, but a more honest one.
“Highly burdened parents did not skip the tracking because they lacked interest. They skipped it because they lacked time — and that is a finding about design, not about parents.”
What it cannot carry
The constraints are real and the authors name them. This is a small, single-system trial in English-speaking families with home internet and, for most, some college education; attrition was higher in the intervention arm; the study window was short; and the primary endpoint only trended. A parent-reported questionnaire at one month sits a long way from the outcomes that ultimately matter — fewer emergency visits, fewer missed diagnoses, less caregiver exhaustion. None of those were measured. This is a carefully reported signal, not a foundation to build a service on.
The lesson reaches past pediatrics and past this particular tool. A large share of digital health rests on the premise that patients and families will feed it a steady stream of data between encounters. Here the people with the most at stake — parents of the sickest children — supplied almost none, not for want of motivation but for want of time and capacity. The value, such as it was, came from a structured human moment that the software helped organise, not from its between-visit apparatus. For anyone building tools for overwhelmed users, that ordering is worth keeping: strengthen the conversation that already happens before you hand someone homework that competes with caring for a sick child.
Source: Lin J, Huber B, Amir O, Assis-Hassid S, Gehrmann S, Gajos K, Grosz B, Sanders L. Novel Web-Based Technology to Promote Goal-Setting in Complex Chronic Illness: Randomized Controlled Trial. JMIR Human Factors 2026;13:e70402. A small single-centre stepped-wedge randomized trial (67 dyads) whose primary outcome reached only a trend (P=.06); the patient-relevant claim rests on a one-month, parent-reported questionnaire, not on clinical outcomes.
