From data to trajectory.

Chapter 1

The historical trend: a century of progress

The long-run trend in maternity outcomes is unmistakably one of major improvement. In pre-industrial Britain, maternal mortality has been estimated at around 1,200 deaths per 100,000 births, and through much of the 19th century rates were still around 400 to 500 per 100,000 [12,13]. Antisepsis, antibiotics, blood transfusion, safer anaesthesia, modern obstetrics, improved nutrition and organised hospital care drove a profound reduction over the 20th century [12].

Infant and neonatal outcomes followed the same pattern. In England and Wales, infant mortality fell from 6.3 per 1,000 live births in 1993 to 3.9 in 2013, and neonatal mortality fell from 4.2 to 2.7 per 1,000 over the same period [1]. Stillbirth rates also fell, though more gradually, reflecting gains in antenatal surveillance, foetal medicine, neonatal care and public health [1,14].

These gains matter because they show the service can improve substantially when science, organisation, workforce capability and data align. They also carry a caution. A system can look historically successful while becoming complacent about present stagnation [1,3].

Chapter 2

The recent trend, and why it has stalled

Since the mid-2010s, progress has slowed or stalled in stillbirth, neonatal mortality and preterm birth, and maternal mortality has moved the wrong way. Nuffield Trust's 2026 review concludes that England has not met the national maternity safety ambitions set for 2025 [3,15].

Maternal mortality tells the starkest version of the story, and it tells it in exactly the way this briefing is about. Read as a single number, it looks like a rise. Read as a trajectory, it is worse than that. From 8.79 per 100,000 maternities in 2017 to 2019, the rate climbed to a peak of 13.41 in 2020 to 2022, lifted by COVID, then eased to 12.80 in the latest figures for 2022 to 2024 [2]. Easing sounds like recovery. It is not. The rate never came back to where it started. It has come to rest about a fifth above the 2010 baseline the safety ambition was measured from, and the three most recent readings sit, statistically, on top of one another. The system did not bend the curve. It found a higher plateau and stopped there. And beneath the average sits a harder fact. A Black woman in England remains markedly more likely to die in pregnancy or soon after than a white woman, a gap that has persisted across every recent report. Behind that average are the same women, in the same communities, year after year. An intelligence model that only disaggregates the gap, without forcing the board to act on it, is a more precise way of watching it endure [2,15].

Neonatal and stillbirth trends are not improving fast enough either. Neonatal mortality remained 1.4 per 1,000 live births in 2023, above the ambition of 1.0. Stillbirth fell from 4.4 per 1,000 in 2015 to 3.9 in 2023, still well above the 2.6 target. Preterm birth rose to 7.9% in 2022, almost back to the level that first triggered the ambition [3].

Why the recent trend has stalled

• More complex case mix. Maternal age has risen, more births are to women with comorbidity, obesity remains a significant risk factor, and neonatal services care for more complex babies [3,14].
• Intense workforce pressure. Only 19% of midwives in the 2024 NHS Staff Survey felt their organisation was staffed well enough. One in five NICU shifts were reportedly inadequately staffed in a 2024 audit [3].
• Entrenched inequalities. Compounded risks linked to ethnicity and deprivation persist [3,15].
• Culture and operating conditions. Reviews continue to describe services where women are not listened to, escalation is poor, and learning is inconsistent [3,16].
• Measuring activity, not change. The service often measures compliance more effectively than it can tell whether a change has truly altered the underlying system [4,17].

Source: Nuffield Trust, maternity safety ambitions against latest data, February 2026.

Chapter 3

The wrong comparison: why we cannot copy Scandinavia

Drawn from the author's Friday Fish and Chip Paper, written on returning from Sweden.

First, the thing that must not be lost in the argument that follows; where care fails a mother or a baby who could have been saved, no account of social structure excuses it. The Ockenden review of Nottingham concluded that 94 stillbirths, 62 neonatal deaths and 6 maternal deaths between 2012 and 2025 were avoidable, 162 deaths in all, with a further 520 mothers and babies harmed [21]. Those deaths were failures of care that better intelligence and governance could have caught. Social structure did not cause them. Hold that fixed. Then look at the comparison everyone reaches for, because the lesson usually drawn from it is wrong.

The NHS is routinely held against Scandinavian health systems and found wanting. Norway retains the title of the nation with the lowest maternal death rate, and senior political voices argue that England should simply be more like Norway [2,3]. The instinct is understandable. The conclusion drawn from it is not safe.

Benchmarking the NHS against Norway or Sweden measures a difference. It tells us what might be possible. It does not tell us how to attain it here, because it misses almost everything that produces the difference in the first place. Comparing the NHS to Scandinavia is like comparing a comprehensive school in Burnley to a fee-paying school in Stockholm: the comparison is real, but the gap is mostly in what each child arrives with and what sits around the school, not in the teaching. The point is the resources and the starting conditions, not the people inside the gates.

The differences are real, and they sit outside the hospital

The Nordic nations that beat the NHS on maternal mortality, cancer survival and child health do not have better hospitals. They have less poverty, less inequality, higher social trust, newer homes, better schools, and governments their citizens broadly believe in. Denmark's income inequality (a Gini coefficient around 0.28) is lower than the UK's (around 0.33). Norway saved its North Sea oil revenue into a sovereign wealth fund now worth close to two trillion dollars; the UK spent its own. Sweden invests around 3.4% of its GDP in research and development, against roughly 2.6% in the UK. The UK carries the oldest housing stock in Europe, much of it built before 1980 and a meaningful share failing basic quality standards. Around three quarters of Norwegians say they trust their government; UK trust sits well below that and is falling.

Our health outcomes are in large part determined by the social inequality carried by a majority of our population. In Norway, a mother's outcome is not meaningfully shaped by her ethnicity or how poor she is. In England, it is. None of that appears in any healthcare benchmark.

The histories and the populations are not comparable

Britain's structure of advantage and disadvantage has deep historical roots, and Sweden's is nothing like it. The populations differ just as sharply. Sweden has 10.6 million people at 23.6 per square kilometre. England has around 57 million at 434 per square kilometre, in one of the most ethnically diverse societies on earth, carrying national debt of about 100% of GDP. The four Nordic nations together hold some 28 million people; England's maternity services serve roughly twice that. None of that is in any healthcare benchmark either.

The cultural gap is bigger than the data gap

The Nordic principle of Janteloven, that no individual stands above the collective, is a civilising habit reproduced across generations, not a policy that can be adopted. Where Nordic social mobility takes around two generations, in the UK it takes at least five. When people feel their society is broadly fair, they behave differently. They present earlier, they trust clinicians, they trust the data, and they trust the system enough to let it make hard decisions about resources. Norway taxed its oil and saved it for the generations to follow; the UK had more oil for longer and spent it. That was a decision about what a state is for, and the same question sits behind every health funding decision the NHS has ever faced.

So what should benchmarking actually do

We should want fewer preventable deaths of mothers and babies, and we should be precise about where they come from. The Nottingham deaths were failures of care, not of social structure. That is the line the benchmark blurs. The point is not complacency. The point is precision about what is fixable inside the health system and what is not.

The benchmarking comparison is not wrong; it is incomplete. What we conclude from it is wrong, because it frames a civilisational choice as a managerial problem. We can never simply be Norway, and the NHS cannot fix what it did not cause. It can get very good at its one job, and for that it needs an intelligence model honest enough to tell us, given the society we serve, where the genuinely preventable harm sits and where effort will change it.

Chapter 4

The scrutiny system

Maternity services are now one of the most scrutinised parts of the NHS. What began as a series of high-profile local scandals is increasingly recognised as a wider systemic problem. Concerns once seen as isolated to outlier hospitals are now described by the CQC's national maternity review and HSSIB as systemic and widespread [3].

Baroness Valerie Amos was appointed to lead the independent national maternity and neonatal investigation, examining services across England with an initial focus on 12 trusts [18,19]. Her interim findings identified six major sources of system pressure: capacity constraints, culture and leadership, racism and discrimination, poor responses to harm, deteriorating estates and workforce shortfalls [16]. Donna Ockenden's work on Shrewsbury and Telford, and on Nottingham, exposed repeated failures in governance, listening, escalation and culture; the East Kent investigation, Reading the Signals, added weight to the failure to detect valid warning signals in time [20,21,22,23].

Main current scrutiny mechanisms

• Independent national maternity and neonatal investigation chaired by Baroness Amos [18,19].
• Ockenden reviews, including Shrewsbury and Telford, and Nottingham [20,21].
• East Kent investigation, Reading the Signals, and oversight processes [22,23].
• CQC maternity inspections and well-led assessments [3,24].
• NHS Resolution Maternity Incentive Scheme and Safety Action framework [10,25].
• MOSS, the Maternity Outcomes Signal System, for national and regional signal detection [6].
• MBRRACE-UK maternal and perinatal mortality reviews [2,26].
• PMRT, HSIB and HSSIB investigations, NMPA, NNAP, GIRFT and other national audit mechanisms [3,4].

This volume of scrutiny would be understandable if it had translated into sustained traction. The evidence suggests it has not done so reliably [3,4].

Chapter 5

Why the review cycle has lost traction

There is a particular way that well-meaning systems fail: they mistake activity for progress. Faced with tragedy, an organisation does the thing it knows how to do. It commissions, it reviews, it recommends. The decade's response to maternity harm has produced 748 recommendations [3]. No board can hold 748 of anything in its head. A system that answers grief with another report is not being careless; it is being busy in the place of being effective, and from the inside the two are hard to tell apart. The real test is not whether the system has responded. It is whether anything has changed on the labour ward at three in the morning.

“Everyone is drowning. The system is completely overloaded.”

Staff voice cited in HQIP's So What of Maternity Data report

Nuffield Trust's 2026 assessment is explicit that recommendations are not being consistently acted on [3]. The HQIP report found repeated frustration at duplication, ambiguity and overload, including calls for recommendations to be pulled into one set. Staff described the system as completely overloaded [4].

Why traction is lost

• Recommendations are too numerous, insufficiently prioritised, and weakly linked to methods of implementation [3,4].
• Different bodies produce overlapping requirements with different definitions, timelines and audiences, fragmenting local effort [4].
• Many reviews are retrospective and operationally weak. They identify what went wrong but do not give boards a real-time way to see deterioration earlier [3,22].
• Incentives can distort behaviour, rewarding evidence of compliance over the disciplined improvement that actually changes care [11,17].
• Culture is emphasised but treated as a generic explanation rather than a set of observable team, leadership and escalation behaviours [3,27].
• Staff are demoralised and exhausted, so each new requirement lands on people with less left to give [4].

The cumulative effect is a service repeatedly told what good looks like, but less often given a coherent intelligence architecture for moving safely from where it is to where it needs to be [4,17].

Chapter 6

Data and trajectory: you cannot govern a verdict

There is a simple distinction at the heart of maternity governance that boards too often miss. An outcome is a verdict. A stillbirth, a maternal death, a serious neonatal injury: these are the final words on a process that has already run its course. By the time the outcome is recorded, the decisions, the staffing, the escalation and the culture that produced it are already in the past.

You cannot govern a verdict. You can only mourn it, investigate it, and answer for it.

Yet most maternity assurance is built around exactly these end-point measures, reported late, in aggregate, after the fact. It tells a board what happened. It cannot tell a board what is about to happen, and it cannot be acted on to change the result, because the result is already in. Good governance therefore has to move upstream, to the things that are still in the board's gift to change. These are the conditions that produce the outcome, and they are visible long before the outcome itself.

What boards can actually govern

• The data, read as a trajectory rather than a snapshot, so drift is visible while it can still be corrected.
• The skills present in the service, and the depth of senior and specialist cover across the week and the night.
• The skill mix on any given shift, including reliance on temporary staff and the loss of experienced midwives.
• The resources allocated to the service, including training, governance and the analytical capacity to interpret its own data.
• The rotas, the escalation routes and the on-call resilience that determine whether help arrives in time.
• The behaviours of teams and leaders, including whether women are listened to and whether concern is escalated or suppressed.
• The discipline that tests whether a change actually altered the process of care, rather than simply recording that activity took place [4,35].

Each of these is upstream of the verdict. Each can be measured now, governed now, and changed now. This is what it means to move from data to trajectory. Not to predict the unpredictable, but to govern the conditions that load the dice, while there is still time to unload them.

Chapter 7

Not short of data: MSDS is indispensable but underused, and MOSS is still late

The Maternity Services Data Set is indispensable because it is the main patient-level national dataset covering NHS-funded maternity care, from booking through to postnatal discharge [5,28]. It is one of the few national assets capable of linking activity, outcomes, demographics and care pathways at scale, and it is the credible foundation for each unit to benchmark against itself, and within itself, with statistically meaningful trends, as well as for national and local-to-national comparison [4,29].

Yet MSDS remains underused. HQIP found that 71% of respondents had difficulty accessing local data from nationally collated datasets. Around half cited problems with data quality, completeness and timeliness. 68% said resource constraints for quality improvement were often or always problematic, and 66% said the same of time for analysis [4]. Some did not know how to access the National Maternity Dashboard, and some distrusted the data because it did not match what staff saw clinically [4].

MSDS has the value. The difficulty is that many services receive it as raw or semi-processed information rather than decision-ready intelligence, and must spend scarce local effort cleaning and interpreting it. The dataset is strategically essential but operationally under-activated [4,29].

MOSS: a real advance, but still a late signal

MOSS brings together near-real-time maternity outcome data at trust, ICB, regional and national level, and generates coded alerts when unusual patterns emerge. Developed in response to East Kent, it is explicitly intended to identify emerging safety concerns earlier, a real improvement on waiting for annual reports [6,22].

But MOSS is still late in an important sense. It detects signals in outcome patterns after they are already visible in routine data. It tells a service that something may now be going wrong. It does not reveal the earlier deterioration in workforce fragility, pathway instability, quality drift or service stress that made the outcome signal more likely. It is a warning system focused on major harms, and those harms are still rare events per unit that take a long time to reach statistical significance. They are the consequence of what the MSDS trends reveal before the catastrophe. MOSS is one instrument, not the whole intelligence system [6,9].

Chapter 8

The patient voice: an essential measure made unusable

If the mother's story matters, then the systematic measurement of her experience should be one of the most powerful instruments a board has. At present it is one of the weakest, not because the data is absent, but because it arrives in a form that cannot be governed.

The CQC maternity survey is the national measure of women's experience. It is rigorous, but it reaches boards around six months after the women were surveyed. It carries many questions, and its results show the scale of the challenge without showing precisely where and how change is possible within a given unit. Each maternity service is its own complex adaptive system, and an annual national average, half a year out of date, cannot tell a board where to push in its own service this month [4,24].

A faster, sharper instrument

Boards would benefit from a monthly release of maternal feedback, built around a single, well-understood question. Question Two of the survey asks, in effect, how likely a woman is to recommend the unit to her friends and family. Used as a net promoter score and reported with no more than a month's lag, that one measure becomes a live, trend-based signal of how a service feels to the women using it, unit by unit, month by month.

This is not a replacement for the full survey. It is the difference between a verdict and a trajectory, applied to experience. A monthly recommendation score is an upstream signal. It moves before outcomes move, it is intelligible to every board member, and it can be acted on while the women whose experience it reflects are still in the service.

If the Nottingham board had held a monthly recommendation score for its own maternity unit, tracked as a trend with no more than a month's lag, would the service have been allowed to deteriorate as far as it did? A falling line, visible every month, is much harder to look past than a national average that lands half a year too late.

Chapter 9

Why the SMI is needed: one truth, trend focus, action and intuitive AI

The maternity system needs a more integrated, trend-sensitive and explanatory view than MSDS dashboards or MOSS alone can provide. Before reaching for any product, it is worth being clear about what that takes, because most of it costs nothing and belongs to the board already.

Start with the method, not the product

Reading data as a trajectory rather than a snapshot. Statistical process control in place of RAG ratings. The monthly recommendation score. A defined set of leading indicators. Every key metric disaggregated by ethnicity and deprivation. None of that needs a vendor, and the evidence that disciplined, measured intervention changes maternity outcomes is independent of any one supplier. The Saving Babies' Lives Care Bundle, independently evaluated, is judged likely to have contributed to a fall in stillbirths, and NHS England's own Making Data Count work shows that trend charts catch what RAG ratings miss [35,37]. The method is the point. A board that takes only the method from this paper, and buys nothing, has already begun to move from governing verdicts to governing trajectory.

Where a board wants to go faster, an integrated index is an accelerant, not a precondition. The Strasys Maternity Index is one such index, a trust-level benchmark using more than 100 measures across current performance, trend and volume-adjusted risk [7,8]. It does five things a board would otherwise have to assemble by hand:

• One version of the truth. The index draws clinical, safety, workforce and equity indicators into a single comparative frame, reducing the risk that different parts of the organisation are looking at different stories [8,9].
• Trends, not snapshots. It incorporates 12 month trend analysis, central because safety failure is rarely a single bad month but a gradual drift, letting boards intervene before a major incident or a MOSS escalation [7,8].
• Contextualised narrative. It generates clear narratives from the data trends, so the reader understands what is happening and why, and therefore where improvement effort should go. Boards see where they stand against peers, ask sharper questions, and see how resources are being used to hold the trajectory.
• Actionability. Its measures map to Ockenden recommendations, CQC well-led criteria and NHS Resolution safety requirements, so intelligence translates into specific actions, owners and governance routes [8,10].
• Intuitive AI. MIA, the associated maternity intelligence agent, supports interpretation by surfacing emerging risks and linking patterns across sources, making advanced analysis accessible to leaders without time to interrogate dozens of datasets. The value is the translation of complexity into insight humans can challenge, test and act on [9,30].

A necessary honesty. No single index has yet been shown, in a published evaluation, to prevent a maternity tragedy, and this briefing does not claim one has. The claim is narrower and more defensible. Each component is proven on its own, benchmarking, trend analysis, equity disaggregation and statistical process control, and the gap they leave between them, the missing middle between a raw dataset and a late alarm, is real and currently unfilled. An integrated index such as the SMI is the proposal for filling it, to be judged on the quality of the decisions it changes, not on its novelty.

How this fails

The intelligence pack becomes one more monthly return, produced for the board and read by no one on the ward. The board watches a line fall for a year and still does not act, because seeing was never the hard part. Equity becomes a tab that is opened and closed. The index is bought and the behaviour it was meant to change is not. Each of these has happened to good services. No software guards against these. What does is a board that treats a moving line as a reason to act, and a director of midwifery given the room and the resource to act on it.

Chapter 10

A rounded board view: outcomes, workforce, finance and viability

If boards are to make sound decisions, they need more than an outcomes dashboard. Outcomes are essential but lagging. A rounded board view should integrate at least four interacting domains: outcomes and safety signals; workforce capacity and skill mix; finance and operational resilience; and longer-term service viability [10,11].

• Workforce is a leading indicator. Nuffield Trust highlights widespread staffing strain, supervision gaps, rota gaps and recruitment freezes. The Amos interim findings and the professional bodies stress that meeting a headline ratio is not the same as having the right senior cover and resilience in escalation [3,16,31].
• Skill mix matters. Safety depends not only on how many staff are present, but which staff, in what combinations, with what experience, and at what hour. Heavy reliance on temporary staff, or the loss of experienced midwives, can change risk before outcome metrics worsen enough to trigger an alert [3,9,32].
• Finance matters. Services under sustained financial pressure freeze recruitment, defer training, and underinvest in governance and business intelligence. The cut in maternity Service Development Funding from £95 million in 2024/25 to £2 million in 2025/26 weakens the local improvement infrastructure needed to turn recommendations into delivery [3,33].
• Service viability matters. Boards need sight of birth volumes, complexity, estate quality, transfer risk, neonatal dependency, network support and on-call resilience, to judge not only whether outcomes are acceptable today, but whether the service model is safe and sustainable over the next one to three years [16,34].

The role of NHS Resolution, and the case for one shared truth

There is a striking imbalance at the centre of the system. NHS Resolution exists, in practice, largely to settle litigation costs and court judgements after harm has occurred. Maternity claims dominate its liabilities. The system pays out of the order of £1.3 billion a year settling maternity harm, and left £2 million in the maternity Service Development Funding to help services prevent it, once that fund was cut from £95 million [10,25,33]. It spends vast sums on the consequences of harm and a fraction on its causes.

This is the logic of governing the verdict rather than the trajectory, expressed in pounds. Money flows to the outcome, after the fact, rather than to the upstream conditions that determine it. It is the most expensive possible way to run a safety system, because litigation pays for the harm twice, once in damage to families and once in cost to the service.

It raises an obvious question. Should NHS Resolution and the regulators not be part of the one version of the truth that an instrument like the SMI provides? If the body that pays for harm, the bodies that inspect for it, and the boards responsible for preventing it all worked from the same trend-based intelligence, the system could begin to move effort and money upstream, towards the conditions that prevent claims rather than the settlements that follow them [8,10,11].

Two hard truths sit beneath that question. The first is financial. When prevention pays back years later, in NHS Resolution's accounts or a neighbouring organisation's budget, no single board has the incentive to fund it now. Redirecting that money upstream is a national decision about who holds the risk and the reward across the system, not something one trust can do alone. The second is political. Where viability points to changing or concentrating a service, every option meets a campaign, a local headline and the threat of judicial review. The paediatric version of this argument was honest that the analysis is the easy part and the politics is the blocker, and the same honesty is owed here. Intelligence can show a board that a service model will not be safe over a three-year horizon. It cannot, by itself, carry the public through what follows. It can, though, surface the case in time to have that conversation calmly, rather than in the aftermath of a tragedy.

Chapter 11

What boards must now do

Boards need to move from passive assurance to active inquiry. Reviews suggest boards have too often been comfort-seeking rather than truth-seeking, presented with large volumes of data and no clear interpretive frame. The answer is better-structured oversight: a disciplined intelligence model that boards actually use [10,11].

Changing the pack is the easy half. The harder half is how the board behaves around it. A truth-seeking board asks for the falling lines first, not the reassuring ones. It rewards the executive who brings bad news early. It treats a flat "nothing to see here" as itself a signal. It reads the trajectory before the narrative, and it returns to the same risk month after month until it is closed, rather than moving on once it has been discussed. None of that is bought with an index. It is set by the chair.

This now carries a personal edge that boards did not feel a decade ago. Nottinghamshire Police is conducting a corporate manslaughter investigation into Nottingham University Hospitals, one of the largest of its kind in the history of the NHS, and the trust has been fined £1.6 million for exposing women and babies to avoidable harm [38]. Maternity board members are right to ask whether they are personally and corporately exposed. The strongest protection is also the right thing to do: to see deterioration early, record that you saw it, and act on it. A board that can show it governed the trajectory, rather than waited for the verdict, is a board that did its job. That is the accountability case for everything in this paper.

The actions below put that into practice.

• Adopt a single maternity intelligence pack. Integrate outcome trends, MOSS alerts, MSDS-derived indicators, equity breakdowns, workforce and skill mix, user experience and viability into one monthly board view [4,6,10].
• Insist on trend charts, not static RAG ratings. Require SPC or equivalent charts to distinguish special-cause variation from ordinary fluctuation [4,35].
• Make equity visible in every discussion. Routinely disaggregate every key metric by ethnicity, deprivation and other relevant characteristics [3,15].
• Track leading indicators, not just serious harms. Monitor staffing fragility, skill mix, vacancy and agency use, CTG training, escalation compliance, perineal trauma, postpartum haemorrhage, critical care admission, complaints and speaking-up indicators alongside deaths and stillbirths [3,9,10].
• Use MOSS as a trigger, not the full picture. Any signal should prompt a structured review of wider factors, workforce, finance and viability, not a narrow incident-style response [6,10].
• Commission an integrated explanatory layer. Through the SMI or a comparable approach, link multiple measures, identify deteriorating trajectory, and suggest lines of enquiry and action [8,9].
• Clarify ownership and escalation. For every major risk domain, name an executive owner, a clinical lead, a time-bound action plan and a board review date [10,11].
• Support local analytical capability. Invest in dedicated maternity informatics, audit and improvement support rather than assuming frontline teams can absorb this work [4].
• Test sustainability, not just compliance. Ask whether improvements would survive staff turnover, funding loss and reduced external scrutiny [17,27].
• Create psychological safety for uncomfortable truths. Ensure board discussions reward escalation, candour and challenge over defensiveness and reputation management [11,36].

Questions boards should ask every month

Picture a unit where, over four months, the recommendation score drifts down, agency use climbs, and CTG-training compliance slips, while the serious-incident count stays flat. Read as three separate returns, each looks survivable. Read as one trajectory, they are the early shape of a service in trouble, visible a year before it would surface in a death. The value of the intelligence lies less in the chart than in the conversation it forces in the boardroom, and the consultant midwife given the room and the resource to act on the ward, while there is still time.

Chapter 12

A roadmap to rebuild the governance architecture

The takeaway for a hospital board and its ICB is a practical, sequenced programme, not another set of recommendations to file. The six phases below move a system from fragmented assurance to a single, trend-based intelligence architecture that links outcomes to workforce, finance and viability, across both the trust and the place it serves [8,10,11].

Doing this with the funding you have

A board reading this in 2026 has less money, not more. National maternity Service Development Funding has been cut from £95 million to £2 million, and midwifery vacancies are real. So the first test of this programme is whether it can be afforded, and it can, because most of it is not a purchase. It is a change in what the board already does with data it already holds.

Trajectory charts replace RAG ratings using data already collected. The monthly recommendation score is drawn from a CQC survey question already asked. Leading indicators come from workforce and rota data already gathered. Equity disaggregation is a way of cutting numbers the service already holds. None of that needs a contract. It needs a decision.

A single intelligence pack replaces the parallel dashboards and ad hoc papers a board currently drowns in. Pulling the local response to those 748 recommendations into one prioritised set removes work rather than adding it. An integrated index is an accelerant for a board that wants to move faster, not a condition of starting.

How to read this roadmap

Phases 1 to 3 are foundational and can begin immediately, on existing budgets. Phases 4 to 6 embed the architecture so it survives turnover, funding pressure and reduced external scrutiny. Each phase names the lead level (trust board, ICB, or joint) and a clear exit test before moving on.

The architecture at a glance

These phases compound. One version of the truth makes early warning possible. Early warning makes accountability meaningful. Accountability at place makes the network governable. A governed network can learn from itself. In that order, a board moves from governing verdicts to governing trajectory.

Chapter 13

Conclusion

The central issue is no longer whether the service cares enough, whether there are enough recommendations, or whether enough data exist. It is that intelligence has not yet been organised to give boards one coherent, trusted, timely and actionable view of risk and trajectory. Historical gains show that sustained improvement is possible. The recent period, stalled progress, maternal mortality stepped up and stuck above target, and persistent inequality, shows the current improvement architecture is not sufficient.

MSDS remains the indispensable base dataset, but it is underused. MOSS is a real step forward in signal detection, but it becomes active only after adverse patterns appear. The SMI addresses the missing middle: one version of the truth, longitudinal trends, comparative risk, intuitive AI, and an operational route from insight to action.

For boards, the implication is clear. Good maternity governance now requires a rounded intelligence model, combining outcomes, workforce, skill mix, finance and viability, that treats trend, inequality and sustainability as core governance questions. And it requires boards to remember why this work matters more than almost any other they do. A mother in labour is doing the hardest, most frightening work of her life, trusting a team to keep her and her baby safe. The boards that best support good and outstanding services will be those that move beyond static assurance and build a disciplined capacity to see deterioration early, ask better questions, and act before scrutiny becomes scandal.

A practical next step

The Strasys Maternity Index brings outcomes, workforce, finance and viability into one benchmarked, trend-based view. To see where a service stands, explore the index at smi.strasys.uk or speak with the Strasys team.

Briefing paper

References

1. Nuffield Trust, "Stillbirths and neonatal and infant mortality," 2024. nuffieldtrust.org.uk
2. MBRRACE-UK / NPEU, "Saving Lives, Improving Mothers' Care: Maternal Mortality 2020 to 22" and "Maternal Mortality 2022 to 24," 2024 and 2026. npeu.ox.ac.uk
3. Nuffield Trust, "Is England making progress to improve maternity care? An explainer," March 2026. nuffieldtrust.org.uk
4. HQIP, "The 'So What' of Maternity Data," 2022. hqip.org.uk
5. NHS England / NHS Digital, "Maternity Services Data Set (MSDS)," DCB1513. standards.nhs.uk
6. NHS England, "Maternity Outcomes Signal System (MOSS): Standard Operating Procedures" and national rollout, November 2025. england.nhs.uk
7. Strasys, "Strasys Maternity Index (SMI)," 2025. strasys.uk
8. Strasys, "Strasys Maternity Index (SMI): measures and benchmarking," 2025. strasys.uk
9. Strasys, "Maternity Safety and Risk / MIA (Maternity Intelligence Agent)," 2025. strasys.uk
10. NHS Resolution, "Maternity Incentive Scheme," 2025. resolution.nhs.uk
11. Sands / Tommy's Joint Policy Unit, "Better board oversight needed to save babies' lives" (board oversight recommendations), 2022. tommys.org
12. Cambridge Group for the History of Population, "How dangerous was childbirth in the past?," 2024. campop.geog.cam.ac.uk
13. Irvine Loudon, "Maternal mortality in the past and its relevance to developing countries today," Am. J. Clin. Nutr., 2000.
14. ONS, "Childhood, infant and perinatal mortality in England and Wales," previous releases. ons.gov.uk
15. MBRRACE-UK / NPEU, "Saving Lives, Improving Mothers' Care 2025" (2021 to 23 data, including deprivation and ethnicity inequalities), 2025. npeu.ox.ac.uk
16. National Maternity and Neonatal Investigation, "Interim Report," 26 February 2026. matneoinv.org.uk
17. Institute for Healthcare Improvement (IHI), "Sustaining Improvement," White Paper. ihi.org
18. GOV.UK, "National maternity investigation launched to drive improvements," June 2025. gov.uk
19. GOV.UK, "Independent maternity and neonatal investigation: terms of reference," 2025. gov.uk
20. GOV.UK / Donna Ockenden, "Final report of the Ockenden Review (Shrewsbury and Telford)," March 2022. gov.uk
21. GOV.UK / Donna Ockenden, "Ockenden review (Nottingham University Hospitals): final report," June 2026. gov.uk
22. GOV.UK / Dr Bill Kirkup, "Maternity and neonatal services in East Kent, Reading the Signals," October 2022. gov.uk
23. Department of Health and Social Care, "Government response to Reading the Signals (East Kent)," July 2023. gov.uk
24. Care Quality Commission (CQC), "State of Care 2024/25, Focus on maternity," October 2025. cqc.org.uk
25. NHS Resolution, "Annual report and accounts: maternity claims," 2024/25.
26. MBRRACE-UK / NPEU, "Perinatal mortality surveillance reports," 2025. npeu.ox.ac.uk
27. NHS England, "NHS Sustainability Guide," 2010 (updated). england.nhs.uk
28. NHS England / NHS Digital, "Maternity Services Data Set (MSDS): scope and coverage," DCB1513.
29. NHS England, "National Maternity Dashboard and Data Viewer: access and use," 2024. england.nhs.uk
30. Strasys, "Maternity Safety and Risk / MIA: interpretation and human oversight," 2025. strasys.uk
31. Royal College of Midwives (RCM), "Workforce," 2025. rcm.org.uk
32. Royal College of Obstetricians and Gynaecologists (RCOG), "Safe staffing in maternity," 2025. rcog.org.uk
33. Royal College of Midwives (RCM), "Shocking budget cuts will compromise the delivery of safe maternity care," April 2025. rcm.org.uk
34. NHS England, "Planning, assuring and delivering service change for patients," 2018. england.nhs.uk
35. NHS England, "Making data count," including SPC resources, 2019. england.nhs.uk
36. National Guardian's Office and NHS England, "Freedom to Speak Up," 2025. nationalguardian.org.uk england.nhs.uk
37. NHS England, "Saving Babies' Lives Care Bundle (version 3)," 2023; and Widdows K et al., "Saving Babies' Lives Project Impact and Results Evaluation (SPiRE)," independent evaluation, 2018. england.nhs.uk
38. Nottinghamshire Police, "Operation Perth: corporate manslaughter investigation into Nottingham University Hospitals NHS Trust," announced May 2025; and Care Quality Commission enforcement action (trust fined £1.6 million), 2023 to 24.