Reach Every Patient

The problem

Most GP practices reach most patients most of the time. A digital invitation goes out via NHS App, SMS booking link, or email. Some patients book. The practice moves on.

Standard patient recall is built around the average patient: registered on the NHS App, with a working mobile number, digitally literate enough to act on an automated invitation, and with the practical capacity to book and attend. For everyone who doesn't fit that profile, the system doesn't fail out loud; it just quietly fails for that person.

The result is a predictable gap in who does and who does not take up proactive care. This can be seen in plateauing screening rates. Annual reviews that go uncompleted. The quiet reflection of practice staff realising that the patients who most need proactive care are least likely to receive it.

Five million women in England are currently not up to date with cervical screening. MMR vaccine coverage has declined for five consecutive years and now sits below the 95% herd immunity threshold. NHS Health Check attendance has been flat at around 48% of those invited for years. Physical health checks for people with severe mental illness are received by only 55% of eligible patients nationally, against an NHS target of 75%.

These are not new problems. They have persisted despite the programmes existing for years, despite financial incentives through QOF, and despite investment in recall infrastructure. The gap is structural, not incidental. It reflects the fact that standard recall was designed around the patients who are easiest to reach, not the patients who are hardest.

Appt Health was built around a different question: not "how do we reach most patients?" but "what does it take to reach every patient?"

Who gets left behind: the evidence

NHS England's Core20PLUS5 framework identifies the groups most likely to experience health inequalities in primary care. It is also, in practice, a map of who standard recall fails most predictably.

Patients in the most deprived 20%

The deprivation gradient in preventive care uptake is one of the most consistent findings in the primary care literature. For bowel cancer screening, uptake in the most deprived quintile is 35% compared to 61% in the least deprived. Colorectal screening uptake is nearly 14 percentage points lower in the most deprived areas and almost 16 percentage points lower in the most ethnically diverse areas compared to the least deprived and least diverse. The digital-first recall channel compounds this: NHS App registrations are 34% lower in the most deprived practices and appointment bookings via the App are 39.7% lower compared to the least deprived practices.

The reasons are structural. Patients in more deprived areas have lower rates of NHS App registration. The kind of job roles often done by individuals in deprived areas often makes responding to a daytime digital invitation harder. Housing instability means contact details go out of date. Distrust of healthcare institutions, sometimes well-founded, reduces the likelihood of acting on an automated message.

Patients from ethnic minority communities

Uptake inequality by ethnicity is large and consistent across programmes. South Asian bowel screening uptake is 32.8% versus 61.3% for non-Asian patients, with Muslim patients at 26.1%. For cervical screening, BAME women are more likely to be non-attenders than white British women (44–71% non-attendance versus 12%), and Black women are most likely to be diagnosed through emergency presentation for cervical and breast cancer. Language barriers, cultural and religious factors, medical mistrust, and the impersonal nature of automated digital invitations are all documented barriers. Crucially, non-English-speaking patients convert lower on every digital channel, which points to message design and personalisation as the main lever to pull, rather than channel choice alone.

Patients with a learning disability

Only around half of eligible adults with a learning disability receive an NHS Annual Health Check in a given year, despite the check being specifically designed for this group. People with a learning disability are twice as likely to die from preventable causes and four times as likely to die from treatable causes compared to the general population. Women with learning disabilities are specifically less likely to attend cervical screening. The barrier is not primarily unwillingness: it is that standard digital recall assumes a level of navigation and independent scheduling that many patients in this group cannot manage without support.

Patients with serious mental illness

Serious mental illness is associated with a 15 to 20 year reduction in life expectancy, with two thirds of those premature deaths attributable to preventable physical conditions, primarily cardiovascular disease. Approximately 551,000 people in England have a severe mental illness. Physical health checks for this group are incentivised through QOF, yet national achievement sits at around 55%, well below the 75% NHS England target, and as low as 24% in some parts of the country. Removing QOF incentives causes BMI check uptake to fall by 14.3 percentage points and alcohol check uptake by 11.9 percentage points, demonstrating how dependent this group's engagement with preventive care is on active, persistent practice-level effort.

Inclusion health groups

For patients experiencing homelessness, those with no fixed abode, and Gypsy, Roma and Traveller communities, standard digital recall is largely ineffective. Contact details are unreliable. NHS App registration is near zero. Trust in institutional healthcare is low. These patients cycle through failed invitation attempts without generating any visible signal of persistent non-response, making them invisible in standard practice reporting.

Older and digitally excluded patients

As of 2024, approximately 8.5 million adults in England lack foundation-level digital skills for everyday online tasks. Around 30% of non-internet users report difficulty interacting with the NHS. Nine in ten people with very low digital skills are over 50. Patients over 75 represent one of the groups with the highest preventive care need and the lowest likelihood of being registered on NHS App or acting on an email invitation.

Why standard recall fails: the channel problem

Reaching every patient requires understanding why single-channel recall cannot do it. The data from Appt Recall's production analytics, across more than 150 GP practices, means we have a precise understanding of these dynamics.

The four primary digital recall channels do fundamentally different jobs. They reach different proportions of patients, and the patients they reach convert at very different rates.

NHS App has the highest conversion of any channel: 25.5% of patients who receive an App notification book directly. But it reaches only 15.6% of patients, because most are not registered or do not have the app installed or notifications enabled. Accessible SMS reaches nearly three quarters of patients but converts at a more modest 17.4%. Email and SMS booking link reach broadly but convert at around 12%.

Read across these rows and the trade-off is clear: no single channel is best on both dimensions. A programme that defaults to NHS App posts high-looking conversion numbers while quietly missing most of the population. A programme that defaults to accessible SMS reaches almost everyone but leaves conversion on the table for patients who respond better elsewhere.

This is not just an efficiency argument. It is an equity argument. The patients least likely to be registered on NHS App are the patients in the most deprived areas, the oldest patients, and those with the highest health burden. Defaulting to NHS App would structurally favour the most engaged, most digitally included patients: precisely the patients who are least likely to need the invitation to act.

The same pattern holds across demographic subgroups. NHS App conversion peaks for patients aged 60–74, at 43.6%, and then falls away sharply for over-75s, who barely appear in App data at all. Non-English-speaking patients convert lower on every channel: NHS App 21.9% versus 25.2% for English-speaking patients; Accessible SMS 12.9% versus 17.0%; NHS Email 9.3% versus 12.8%. The channel mix does not close this gap on its own.

How Appt reaches the patients standard recall misses

The evidence on who gets left behind points to three distinct barriers. Some patients are unreachable through the default channel. Some are reachable but contacted at the wrong time. Some receive an invitation that doesn't connect because the framing, reading level, or language doesn't match who they are. Standard recall treats all three as the same problem and applies the same solution. Appt treats them as separate problems requiring different responses.

Channel: a portfolio, not a default

Rather than defaulting to a single channel, Appt Recall operates a coordinated portfolio in sequence. An NHS App notification goes first to patients who are registered and likely to respond to it. Patients who do not respond, or who cannot be reached by App, receive an SMS. Persistent non-responders across automated channels escalate to coordinator-led contact, where human outreach takes over from automated invitation.

At every stage, the next intervention fires only for patients who haven't yet booked. Nobody receives a second invitation because they were easy to reach. They receive it because the first one didn't work. The 6.8% of SMS patients who actively decline an invitation are not messaged further for that care invitation; we treat opt-outs as a signal to respect.

This architecture means the patients who benefit most from the portfolio approach are the patients who are hardest to reach: those who cannot be contacted by App, those who don't respond to the first attempt, those who require a different channel entirely to engage. The portfolio approach means that we actually reach under-served patients while picking up easy-to-reach ones on the way.

Timing: personalised, not uniform

Send time is a meaningful lever for some patients and irrelevant for others. SMS booking rates peak at 9am and decline steadily through the day, falling below 7% by evening. NHS App shows the same pattern. The two link-based channels, SMS booking link and NHS Email, are nearly flat across the day, consistent with their mechanism: the patient clicks through to live capacity whenever they choose, so the send moment matters less.

But the effect is not uniform across groups, and the differences are operationally significant.

Patients aged 60–74 show the sharpest early-morning advantage, booking at over 34% when contacted by SMS at 8am, with a steep decline through the day. For patients aged 30–44, the time-of-day curve is nearly flat, and other factors matter more than timing. For the most deprived practices, the strong early-morning peak seen in mid-deprivation areas disappears entirely: the time-of-day curve is flatter, and optimising for timing alone will not close the booking rate gap.

The language interaction is the most consequential for equity. English-speaking patients on SMS peak at 8am (26–31% booking rate). Non-English-speaking patients show a different pattern, with their peak shifting toward mid-afternoon. A programme that sends every patient at 8am is, in effect, choosing the optimal time for English-speaking patients and a suboptimal one for non-English-speaking patients. That is not a neutral design choice: it is a built-in disadvantage for the group already converting at a lower rate on every channel.

Appt's model translates these patterns into per-patient send time recommendations across five variables: channel, age band, language, gender, and deprivation quintile. For a 65-year-old English-speaking patient receiving an SMS, it recommends early morning. For a non-English-speaking patient on the same channel, it shifts toward mid-afternoon. For link-based channels, where timing barely moves the needle, the model allocates capacity accordingly.

Messaging: adapted, not generic

Channel and timing determine whether an invitation reaches a patient at the right moment. The message determines whether they act on it.

The language gap in booking rates does not close with channel or timing changes alone. It persists across every channel and every send time. The 6.8% of patients who actively decline an SMS invitation are also providing a direct signal that the message didn't land. Both groups point to framing, reading level, and language as the primary lever.

The programme-level data makes this concrete. NHS Health Checks, on the same platform and through the same channels, run in the opposite direction: non-English-speaking patients book at 64.6% versus 54.1% for English-speaking patients, and booking rates are highest in the most deprived quintile (55.1%) rather than the least. The difference is a result of changing our approach to better fit each patient, with clearer perceived relevance and stronger alignment with the health concerns of the communities it is designed to serve. When a message connects with what a patient already believes matters for them, the structural barriers documented in the literature shrink considerably.

For the non-English-speaking cohort and for patients who actively decline SMS invitations, a structured programme of message testing, comparing the framing, language, and personalisation, is the next stage of evidence development.

The evidence: programme by programme

We are big believers that we should not be marking our own homework. We want to be transparent about how our technology performs so we can build a better understanding of how some of the health inequalities that we see in the data are constructed and to gather ideas of how we can counter them. The table below shows Appt Recall's direct booking rates across its primary programmes, alongside the comparable invitation response rates available in the published literature. Two points of methodology matter for reading it correctly.

First, the booking rate shown is the direct booking rate: the proportion of contacted patients who booked through the Appt Recall invitation itself. A proportion of patients recorded as non-responders go on to book through other means following the invitation. The figures below are therefore floor estimates of Appt Recall's contribution to programme uptake, not the ceiling.

Second, the cervical screening cohort is specifically non-responders: women who had not responded to their standard recall invitation before being contacted by Appt Recall. This is a materially harder population to convert than the eligible population as a whole. All other programmes reflect the broader eligible cohort.

‍

The cervical screening and NHS Health Check comparisons are the most direct. Standard re-invitation of cervical non-attenders converts around one in twenty using standard postal approaches. Appt Recall converts more than one in three from the same cohort. For NHS Health Checks, the best digital invitation approaches in published RCTs achieve around 30% attendance. Appt Recall achieves 54.1%, with the inversion on deprivation (the most deprived patients booking at the highest rate) not replicated in any published trial.

For lung cancer screening, Appt Recall's rate sits significantly above the national TLHC programme average and is comparable to the best-performing sites in the programme, which deploy community-based mobile scanner units to remove logistical barriers. Appt Recall achieves a comparable result through channel, timing, and message optimisation rather than physical infrastructure investment. The highest performing Appt Recall lung cancer screening sites are achieving over 65% uptake rates, among the best in the country.

The flu figure sits slightly below the national at-risk group average, but that average includes spontaneous attenders and patients who come in without any invitation. Appt Recall's 35.2% is drawn from the subset who had not yet come in: by definition, a harder population.

The outcome: deprivation does not predict whether patients book

Across all programmes and all demographic groups, the single most important finding in Appt Recall's data is this: once a patient is inside an Appt Recall programme, their deprivation decile does not predict whether they book.

Measured at patient level across all channels and all programmes in the last 12 months, booking rates across the deprivation spectrum are:

The external literature predicts the opposite. Bowel screening uptake is 26 percentage points lower in the most deprived quintile than the least. NHS App appointment bookings are 39.7% lower in the most deprived practices. The expected deprivation penalty on booking does not appear in our data.

The resolution is not that deprivation doesn't matter. It is that the deprivation gap in NHS screening is primarily a contact and channel problem, not an unwillingness problem. Standard recall produces a booking gap because it produces a contact gap. Appt Recall's contact rate is actually slightly higher for the most deprived patients (58.0%) than for the least deprived (50.9%): the system reaching harder rather than defaulting to those easiest to reach. With contact solved, the booking rate equalises.

This finding is qualified appropriately. It describes patients inside an Appt Recall programme. It does not speak to patients who never enter a programme. The honest claim is: among patients reached by Appt Recall, deprivation does not predict whether they book. 35% of Appt Recall's practice base falls in the most deprived IMD quintile, so our programmes are disproportionately serving the populations where the gap is largest.

What this means in practice

A practice using Appt Health does not simply send more invitations. It runs a fundamentally different kind of recall programme: one that starts by understanding which patient-programme combinations are underperforming, builds an engagement strategy differentiated by channel availability, age, language, and deprivation, contacts patients through the right channel at the right time, and gives care coordinators the capacity and information to follow up where automated outreach is not enough.

Our NHS Health Check results are the clearest illustration of what becomes possible. Booking rates in the most deprived quintile are 55.1%, above the least deprived quintile's 33.3%. Non-English-speaking patients book at 64.6%. These are not the outcomes of a programme that treats deprived or non-English-speaking patients as harder to reach. They are the outcomes of a programme where the channel, timing, and message were matched to the patient.

The point about deprivation and engagement is important enough to state plainly. The patients labelled as hard to reach in one programme are actively engaged in another. The barrier to booking is not primarily about who the patient is. It is about the intersection of who the patient is, what they are being asked to attend, and how the invitation reaches them. When all three are right, the gap closes.

That is what reaching every patient means in practice. Not a higher average across the population that is already engaged but a different distribution: one where a patient's likelihood of receiving and acting on a preventive care invitation is no longer determined by how well they fit the design assumptions of the recall system that contacted them.

Evidence summary

Appt Recall supports NHS practices in closing screening and preventive care uptake gaps through multi-channel, personalised outreach. Appt Coordinate supports proactive management of patients with long-term conditions. Both products are designed around the patients standard systems leave behind.

Booking rate figures: Appt Recall production analytics (analytics.appt-health.co.uk), RoundEvents and Strategies tables, last 12 months. Denominator: Booked + NoResponse + Declined (delivery failures excluded). IMD deciles from English Indices of Deprivation 2019, mapped on PatientLsoa. QMUL, consent, and Appt Coordinate strategies excluded.