How platform and SaaS choices affect care delivery

Arguments about software sprawl usually run on invented statistics. This note collects what the peer-reviewed literature actually reports, with the study designs attached, so the Zero Neo hypothesis is argued against evidence rather than against vendor material.

Every figure below comes from a published paper, cited at the end of its section. Where the literature is thin or mixed, the note says so rather than rounding toward the argument.

In a cross-sectional survey of 870 US physicians drawn from the American Medical Association Masterfile, the mean System Usability Scale score for their electronic health record was 45.9, which places it in the bottom 9 percent of scores across more than 1,300 prior studies of other technologies and in the grade F band. Adjusting for age, sex, specialty, setting, hours, and call, each one-point improvement in that score was associated with 3 percent lower odds of burnout.

A 2025 cross-sectional study of 1,424 health professionals at a Norwegian university hospital, six months after a sector-wide record replaced the previous one, found a median usability score of 25 out of 100, ranging from 15 among doctors to 40 among laboratory technicians, and examined its association with burnout, insomnia, and turnover intention.

A 2023 systematic review of electronic health record stress and burnout in hospital settings and a 2025 scoping review of usability and documentation burden both reach the same qualitative conclusion from wider evidence bases: the association is consistent across studies, and the mechanisms are usability, documentation load, and interruption.

The cost is not only fatigue. A 2024 Health Affairs study reports that documentation burden displaces other high-value use of the record among primary care physicians, specifically the exchange of health information with other organizations, which is the exact activity a fragmented estate most needs.

That finding matters more than a fatigue statistic for this program. It says the burden is not a wellbeing side effect; it consumes the clinician attention that coordination depends on.

A workflow and information-flow analysis of cross-institutional referrals identifies where loops fail to close: information arrives incomplete, arrives in a form nobody can act on, or arrives with no owner on the receiving side. The failures are at the boundaries between systems and organizations, not inside any one application.

A qualitative study and literature review of care coordination in the United States attributes the same class of gap to lack of interoperability, and a 2025 mixed-methods cohort study of integrating record systems between primary and specialist care examines the other direction, quantifying effects on diagnosis tracking, care management, and coordination.

This is the strongest evidence in the note for the Zero Neo framing, because it locates the defect between applications rather than in any of them. Buying a better application on either side does not, by itself, close a loop that crosses both.

Most of this evidence is cross-sectional and self-reported, so it establishes association rather than causation, and burnout in particular has causes far beyond software. Usability scores are sensitive to how recently a system was replaced, which is why a six-month post-go-live score is not a steady-state score. Single-site studies do not generalize on their own.

Documentation burden is also measured inconsistently across the literature, in minutes, in note length, in after-hours time, and in perception, which makes cross-study comparison unreliable. A 2024 review sets out how varied those measures are.

Anyone quoting a single dramatic number from this field, including us, should expect to be asked which study, which population, and which measure.

No study cited here shows that consolidating onto platforms an organization already owns fixes any of it. The literature establishes a cost. It does not endorse our remedy, and treating it as endorsement would be the same error the note is written against.

What Zero Neo would have to contribute is the missing half: a before-and-after measurement on a real workflow, with usability, time on task, loop closure rate, and total operating cost recorded on both sides. Until that exists, this note is context for the hypothesis, not evidence for it.