NHS App login sessions grew faster than monthly users in England in August 2026 because both the number of users and average sessions per user increased. Compared with August 2025, sessions rose 43.8% and monthly unique users 26.7%. These administrative counts do not establish improved care or reduced workload. August release; England-level source data
The app recorded 78.2 million sessions and 15.5 million distinct users logging in that month. The release was published on 25 September and was the latest located in the original source check on 8 October. The analysis separates accumulated accounts, monthly activity and actions within individual services.
August 2026: counts with different units
The figures below come from the England-level CSV, checked against Table 1 of the summary workbook. Changes are recalculated against August 2025 within that same release and rounded to one decimal place.
| Measure | August 2026 | Change from August 2025 | What is counted |
|---|---|---|---|
| New account registrations | 270,851 | −5.5% | Registration events during the month |
| All-time account registrations | 42,441,120 | +9.2% | Cumulative registrations, including accounts no longer retained |
| Registered users | 39,061,451 | +7.6% | Retained active accounts at month end |
| Monthly unique user logins | 15,522,703 | +26.7% | Distinct user identifiers logging in that month |
| Login sessions | 78,153,850 | +43.8% | Separate recorded sessions |
| Repeat prescription orders | 7,620,985 | +27.4% | Requests sent to a GP through the app |
| Unique users ordering prescriptions | 6,121,973 | +26.8% | Distinct users triggering a prescription request |
| Messages delivered | 36,499,843 | +61.1% | In-app messages, excluding rejected messages |
“Registered users” does not mean everyone used the app that month. The metadata defines a retained-account measure that excludes known deceased users, deleted accounts and accounts removed for inactivity. Monthly unique logins use distinct identifiers observed during the month; login sessions count separate sessions. All-time registrations retain historical registration events. Definitions and methods
These measures should not be added together or treated as stages of a single patient funnel.
Usage grew through both more users and more sessions per user
Dividing sessions by monthly unique users gives 5.03 sessions per user in August 2026, compared with 4.44 in August 2025. That ratio increased by 13.5%, alongside the 26.7% rise in monthly users. The formula is total monthly sessions ÷ monthly unique logins. Source data
This helps explain why session growth exceeded user growth: both the count of users and average recorded sessions per user were higher. It does not reveal whether repeated use reflected useful engagement, administrative effort or difficulty completing a task. The average also says nothing about the median user or how activity was distributed.
In England’s August 2026 figures, the monthly-user count was 39.7% of the month-end registered-user count: 15,522,703 ÷ 39,061,451. This is a comparison of aggregates with different time and exclusion boundaries, rather than a matched cohort’s activation rate. The remainder cannot be labelled digitally excluded or assumed never to use the app.
New registrations fell while monthly use rose. That combination is a reason to report registration and usage separately; it does not identify the cause of either change.
Feature activity needs its own denominator
August’s prescription figures imply 1.24 request events per requesting user: 7,620,985 ÷ 6,121,973. The metadata records an event when a repeat prescription request is sent to a GP. One request can include several medications. The figures do not count medicines dispensed or show whether a request was approved or collected. Prescription definitions
Messaging has another unit. Recalculating from the published counts gives 41.7% of messages read within 24 hours and 45.7% within 72 hours. These are message-level measures, not percentages of patients. The 72-hour window includes the first 24 hours, so the percentages must not be added. Late-month messages read after month end may only enter a subsequent release. Messaging data, read-window methodology
A recorded opening is not evidence that the reader understood the message or completed the requested action.
Notification percentages describe different groups
The release’s opt-in counts produce 66.1% for registered users and 88.7% for users who logged in during August. The smaller, recently active group has a higher recorded opt-in rate; it should not stand in for all registered accounts. Because of a 31 August outage, these notification figures use a 30 August snapshot. Release and outage note
There is also an exclusion difference: the active opt-in numerator excludes deleted accounts, while the monthly-login denominator includes all recorded unique logins. Neither percentage guarantees that a particular message reached a device. Notification methodology
What teams can reasonably take from this release
The national series shows wider and more intensive recorded use compared with the same month a year earlier. It gives teams a way to distinguish account reach, login activity and service demand without treating them as interchangeable. It does not establish reduced workload, fewer missed appointments or improved health outcomes.
NHS England recommends comparing equivalent periods across years because usage is seasonal. Its data-quality guidance also warns that collection methods and service availability change, and that a link out to a third-party service need not mean the task was completed. Historical ICB figures were remapped in the April 2026 release and should not be casually compared with older releases. Supporting information and comparability
For a local service question, the next evidence is therefore the relevant feature’s availability, definition and completion data. The NHS App series and our health-technology coverage provide context; national login totals alone cannot answer that local question.