Sure Start: What Was Promised, What Was Measured, and When It Was Known

The question

Sure Start was a national programme of area-based, joined-up support (professionals from different services working from the same local centre) for families with children under five. At its peak in 2009–10 it accounted for £2.7 billion of public spending (2023–24 prices). Robust evidence on what it achieved arrived about a quarter of a century after launch.

Parliament was examining the value for money of Sure Start in 2007, before the first cohorts had left primary school. This essay sets out what was promised, what was measured and in what order, when each measurement could first have been made, and what the results do and do not show. The question running through it is one of timing: how long an investment in early childhood needs before a fair judgement is possible, and what that asks of political commitment.

What was promised and built

Launch. Sure Start funding was announced in 1998 (£450 million over three years, following the Comprehensive Spending Review), the first trailblazer areas were announced in January 1999, and the programme doubled from 250 to 500 local programmes in 2000 (Naomi Eisenstadt, the first Director of Sure Start, in the appendix of key dates to Providing a Sure Start). It aimed at highly disadvantaged neighbourhoods. Services were brought together in local centres: health services, parenting support, early learning and childcare, and help with parental employment.

Joined-up working. The idea was that a family with a young child could reach the professionals it might need in one local place: midwives and health visitors, early education and childcare staff, parenting support and employment advice. Evaluators of the later children’s centres described this as integrated, multi-agency services at a single local point of access (Sammons and colleagues, 2022). It also let the professionals supporting one family share what they knew, consult one another and agree a shared plan, instead of each holding part of the picture. That bears on safeguarding. Lord Laming’s inquiry (2003) into the death of Victoria Climbié, an eight-year-old girl who died in 2000 after months of abuse, is reported to have listed failure of communication between staff and agencies among the main failures. The government’s Every Child Matters green paper (2003), published with its response, named poor co-ordination and a failure to share information among the problems behind such failures. Later national analyses of serious case reviews, the local reviews held after a child’s death or serious injury, kept returning to the point. The 2003–05 analysis (Brandon and colleagues) was reported to have found inter-agency communication problems, as in all earlier studies, and the 2011–14 analysis found that in many reviews important information was not passed on. Co-location does not by itself secure sharing: the national evaluation of children’s centres reported tensions over communication and data-sharing between partners, and staff rated multi-agency working as needing improvement (Sylva and colleagues, 2015, p. 5). The IFS findings do not separate the effect of joined-up working from the rest of the Sure Start offer, because delivery data are lacking.

Expansion. A decade-long childcare strategy in 2004 called for a children’s centre in every community. The programme became universal in intent and grew to over 3,000 centres by 2010.

Funding at the peak. In 2009–10 Sure Start accounted for about 30% of public spending on programmes for under-fives (Institute for Fiscal Studies, 2025).

What was measured, and when

Early findings. The National Evaluation of Sure Start was led by Professor Edward Melhuish, then at Birkbeck, University of London. The first impact report (National Evaluation of Sure Start Research Team, November 2005) used data collected after programmes had run for about three years, which it said was typically how long a programme took to offer its full range of services. It described its findings as “at best, early indications of whether SSLPs might be affecting the well-being of children and families”. It reported “extremely few overall main effects”, small benefits for children of non-teenage mothers, and apparent adverse effects for children from more disadvantaged families, meaning those with teenage mothers, lone parents or no one in work. Read on its own, the phrase “early indications” sounds more positive than the report intended. The age-three findings were published in The Lancet in November 2008 (5,883 children in 93 Sure Start areas, compared with 1,879 in 72 deprived areas without a programme): five of 14 outcomes were beneficial, with better social behaviour in children, less negative parenting and a better home-learning environment. An earlier paper (Melhuish and colleagues, Journal of Child Psychology and Psychiatry, 2007) reported variation between programmes. Eisenstadt’s 2022 review lists the national evaluation’s four impact studies as published in 2005, 2007, 2010 and 2012. She says the 2005 study found children of teenage mothers in Sure Start areas doing less well than those in comparison areas (the report itself names teenage mothers among several disadvantaged groups), and that the subgroup differences had disappeared by the follow-up. The report that revisits the subgroups is dated March 2008: it followed up the nine-month-olds from the 2005 study at age three, and compared them with 1,879 families from the Millennium Cohort Study rather than with families in areas due to get a programme. It found “almost no evidence of adverse effects”, and effects that “appeared generalisable across population sub-groups (e.g., workless households, teen mothers)” (NESS Research Team, 2008, p. v). Its authors cautioned that it was “impossible to completely discount the possibility that the virtual absence of adverse effects in the current report is due to methodological differences”, and that the change may reflect better services, greater attention to the hard to reach, and longer exposure; they called the positive results “modest” (pp. v–vi). In a Health Foundation podcast (28 May 2024) she described that first finding as a shock and said the programme responded with assertive outreach to teenage mothers. A conference slide deck by Jacqueline Barnes, a member of the evaluation team, describes the same pattern at age three.

Age five. The national evaluation’s age-five report (DFE-RR067, published after June 2010; the report states no date) compared children with a Millennium Cohort Study group. Of 21 outcomes, eight differed significantly: six positive, two negative. Children in Sure Start areas were less likely to be overweight and had better physical health; mothers reported a more stimulating home learning environment, a less chaotic home and less harsh discipline. Mothers also reported more depressive symptoms, and parents were less likely to attend school meetings. Teacher-rated measures showed no difference. The authors call the effects “putative”, because the two samples were surveyed two years apart by different teams, and suggest the age-three advantage in self-regulation faded because the comparison group caught up once free early education reached nearly all children.

The 2007 audit finding. The House of Commons Committee of Public Accounts reported in 2007 that the national evaluation compared whole areas, not users with non-users, so it was “difficult to tell” whether centres gave value for money. The sponsoring Department could not state at the hearing how much had been spent since 1998. It later confirmed £4.8 billion on early years initiatives between April 1997 and March 2006, of which £2.1 billion went to Sure Start local programmes and children’s centres. The Committee also found that most local authorities did not understand what services should cost, and more than half were not monitoring performance.

Age seven. The national evaluation team’s age-seven report found that four of 15 outcomes showed positive effects of Sure Start local programmes. Mothers reported less harsh discipline and a more stimulating home learning environment. There were no consistent differences in children’s educational, social, behavioural or health outcomes, and no negative effects. The team suggested children in other areas may have caught up, since free part-time pre-school had become available to nearly all three-year-olds from 2004. The comparison group also came from a different cohort study, which limits causal conclusions.

Children’s centres. The Oxford-led Evaluation of Children’s Centres in England (Pam Sammons and colleagues, with Professor Kathy Sylva, December 2015) found consistent but small positive effects, most below an effect size of 0.30. Effects were strongest for mothers and families: better maternal mental health, less chaotic homes and a better home learning environment. Effects on children were fewer. There was no significant impact on household employment or child health. Centres offering more named programmes and working across agencies were associated with better outcomes. The authors noted that centres were being cut and restructured while the study ran.

The long-run evaluation. In May 2025 the Institute for Fiscal Studies, with Nuffield Foundation funding, published a capstone report bringing together its work on educational attainment, hospitalisations, youth offending, special educational needs and social care. It defined access as living within 2.5 kilometres of a centre. For hospitalisations it used the number of centres per 1,000 children under five in a local authority.

Hits

All figures are from the IFS May 2025 report.

Education. Children with early access did significantly better in assessments at ages 7, 11 and 16. The proportion reaching five GCSEs at grades A* to C rose by 0.9 percentage points, about 2%.

Hospital admissions. Admissions rose by 10% of baseline at age one and fell by 8.5% of baseline by age 15. At the peak that meant about 6,700 extra admissions of one-year-olds a year and more than 13,150 fewer admissions of 11- to 15-year-olds. IFS read part of this as earlier detection and prevention, with more contact early and less need later.

Teenage mental health. Young people with early access reported half as many internalising difficulties, such as depression and anxiety, as siblings with less access. Mental-health-related admissions at ages 12 to 14 fell by around 50% of baseline.

Attendance and support needs. Unauthorised absence at age 15 fell 15%. Special educational needs support rose at age five and fell at 7, 11 and 16.

Benefits and costs. IFS’s summary table gives the up-front cost per cohort as £2.7 billion, the same figure the report gives for total Sure Start spending in the 2009–10 peak year, in 2023–24 prices. The cost-benefit chapter explains that IFS uses the 2010 annual cost (over £2 billion running costs plus about £0.5 billion capital) “as a rough estimate of what it would have cost to provide services with the same funding for all five years of eligibility” for one cohort, so the per-cohort cost is a stand-in, not a measured figure (Carneiro et al., 2025, p. 81, footnote 30). Future amounts are discounted at 3.5% a year (p. 79). Government benefits were £2.4 billion: £0.6 billion from lower later spending and £1.9 billion from projected tax. Individual earnings benefits were £3.1 billion. Total benefits of £5.5 billion are about twice the cost. The net cost to government, once savings and projected tax are counted, is about £270 million: £2,690 million up-front cost against £2,420 million in long-run savings, of which £1,860 million is extra tax revenue (Table 11.1, p. 86). That is about 10% of the up-front cost, so the authors say Sure Start “did not entirely ‘pay for itself’” for government. These are central estimates in 2023–24 prices. These are projections of lifetime earnings, not observed repayments, and the authors describe them as “subject to considerable uncertainty” (pp. 83–84). The ratio of about £11 of private benefit per £1 of net cost falls to £6.09 if earnings and taxes are 10% lower than estimated (p. 86). Effects measured for children within 2.5km of a centre are scaled up by 2.9 to estimate effects on users (p. 80).

Misses and unknowns

Crime. Children with early access were more likely to have a criminal record by 16, driven by an 11% rise in cautions. A 6% fall in more serious criminal convictions was not statistically significant. IFS says it cannot rule out no real impact on convictions and custodial sentences.

Care. There was no significant effect on the number of children looked after between ages 7 and 15. Time in care fell about 11% at ages 7 to 11. Social care contact among secondary-age pupils was often higher, though not significantly. There was no impact on Education, Health and Care plans.

Behaviour. No significant effect on self-reported externalising behaviour such as conduct problems and aggression. Effects on school exclusions were not significant.

Employment and reach. The 2015 children’s centre evaluation found no significant impact on household employment. The national evaluation’s own reports on reach changed over time. In the first in-depth implementation study, about three-quarters of programme managers said they were not yet reaching the whole community but were optimistic that they might, given time; they were “less confident about reaching more vulnerable groups, despite their best efforts” (Tunstill et al., 2005, p. 8). That is managers’ own view, not measured reach. The following year the national evaluation reported that “in many areas SSLPs found that services were taken up most enthusiastically by the least disadvantaged families, and this could put off the families who were in more need of them” (NESS, 2006, p. 8). The same report described outreach and home visiting as a way of targeting those in most need and persuading them to attend. By the age-seven report the evaluators wrote that the effects for lone-parent and workless households “can be regarded as evidence that SSLPs were being successful in affecting ‘hard to reach’ groups within SSLP areas”, and that programmes “now appear to be engaging the most vulnerable groups in the most deprived areas” (National Evaluation of Sure Start, age-seven report, DFE-RR220). A 2013 report by the WAVE Trust and a Department for Education practitioner group repeats that later view, adding that engagement often took considerable time. The extracts read do not say what changed between the early and later reports.

What IFS itself concludes. Sure Start was “not a silver bullet”. Light-touch universal early years services cannot replace later specialist ones.

How a single evidence base moved

An October 2024 IFS report reported about 13% fewer convictions at 16 and a fifth fewer custodial sentences. The May 2025 report supersedes it. IFS changed how it weights areas: it now uses a consistent measure of area deprivation from 2000 and no longer weights on pre-Sure-Start trends in the outcome. The report says this produced particularly large changes for crime and behaviour in school. The earlier crime headline no longer stands.

When to test

The principle. A fair test of any slow process needs three things: a measure suited to the stage the process has reached, enough elapsed time for the process to mature, and a baseline and comparison fixed before the results arrive. Remove any one and a judgement can be made, but it will not be a test.

Two illustrations follow. They show the logic of timing and are not evidence about Sure Start.

The soufflé. Taken from the oven after ten minutes, a soufflé is liquid in the middle. That says nothing about the recipe. A cook knows when to look: the rise, the colour, the set of the top, at a time the recipe gives.

The orchard. Fruit picked in July from a tree that crops in October is hard and sour. Nobody concludes the tree is barren. A grower reads colour, firmness and sugar, and knows roughly when each will show.

Where the comparison breaks. The cook and the grower know the timing in advance, from many repeats. For a new early years programme nobody did. The time to maturity was itself unknown, so the only way to learn it was to measure at several ages and keep measuring. A defence of patience with no stated test points is unfalsifiable: “wait longer” can excuse any programme. The strength of the argument depends on naming in advance what would be measured at each stage and what would count as failure.

Science already accepts long follow-up. The Cambridge Study in Delinquent Development followed 411 inner-London boys from childhood. One analysis of its data found that separation from a parent who went to prison predicted antisocial outcomes up to age 32, after controlling for other risks (Murray and Farrington, 2005). Studies of that length are standard when the outcome is a life course.

Long follow-up changes the answer. A RAND review of 20 rigorously evaluated early childhood programmes included benefit-cost analysis for a subset: seven individual programmes plus meta-analyses. Of the seven, one was not effective and another had outcomes that could not be given a dollar value; for the rest, returns ran from $1.26 to $17.07 per dollar invested. The largest ratios were associated with programmes with longer-term follow-up, because they allowed measurement at older ages of outcomes such as educational attainment, delinquency and crime, and earnings. The largest, just over $17 per dollar, was for the Perry Preschool programme, which had the longest follow-up. The authors judged their estimates for effective programmes “likely to be conservative”, because not all benefits could be given a dollar value (Karoly, Kilburn and Cannon, 2005, pp. xvii, xxv, 114). Long follow-up does not guarantee a good result: one costed programme was not effective.

An alternative to waiting. Cohen, Piquero and Jennings (2010) estimated the lifetime cost to society of poor outcomes for at-risk youth, valued at birth: $2.1–3.7 million for a career criminal, and $250,000–285,000 for child abuse and neglect, for example (from the published abstract). WAVE Trust (2013, Appendix 4) is reported to have presented this work as supporting a break-even approach; the paper’s abstract does not describe one. A National Academies report on economic evidence describes break-even analysis as an option when evidence of impact is not available, to show how large impacts would need to be for a programme to be worthwhile (National Academies of Sciences, Engineering, and Medicine, 2016). Such a test sets a threshold in advance, instead of waiting decades for outcome data. It is a published alternative to the argument for patience, and it depends on assumptions about costs and effect sizes that the later Sure Start evidence shows can move.

A staged test for an early years programme. Different stages call for different measures:
- Infancy and early childhood: parent and home measures, early contact with services, school readiness at five.
- Primary school: attainment, special educational needs support, attendance.
- Adolescence: GCSE results, hospital admissions, mental health, offending.
- Adulthood: earnings, health, and the next generation’s outcomes.

Sure Start’s evaluations followed roughly this order. The sources read do not show whether a commitment to follow the children through every stage was made in advance; Eisenstadt wrote that the programme was not built on rigorously evaluated evidence. Funding decisions were taken before the later stages could be observed.

A second example: the Elizabeth Line

What was said at the time. The National Audit Office reported in May 2019 that the Department for Transport had raised Crossrail’s funding to £17.6 billion in December 2018, about £2.8 billion above the 2010 level, and that the central section could not open as planned in December 2018. It attributed the cost rise to a compressed schedule, the contractual model and an unrealistic plan, and said some project choices “clearly damaged public value”. The same report said overall value for money could not be concluded until the services opened and final costs were known. A later NAO report (July 2021) gave the 2010 budget as £14.8 billion and Crossrail Ltd’s forecast cost to complete as £18.9 billion, a 28% nominal increase, against an agreed funding package of £18.8 billion. Both figures exclude about £1.1 billion for new trains and the depot, which Transport for London funded separately. The YFL essay Public Financing Models and Cost Outcomes: Crossrail, HS2 and the Shadow-Toll Roads records staged opening between 2022 and 2023, three and a half years late. Transport for London’s published schedules put the planned opening of the central section at December 2018 and the full line at December 2019. The central section opened on 24 May 2022, through running from Reading, Heathrow and Shenfield began on 6 November 2022, and the full timetable started on 21 May 2023: about three and a half years late at each stage. The outer branches had already been running as TfL Rail since 2015 and 2018 No audited final cost for the whole project appears to have been published. Transport for London’s post-opening evaluation (May 2025) still calls £18.9 billion the “estimated final cost”, and its investment programme report for the last quarter of 2022/23 put Crossrail Ltd’s own share at £15.9 billion, excluding on-network works by Network Rail, with the project closing on 26 May 2023 (Transport for London, 2023; 2025).

What was measured after opening. Transport for London reported in December 2024 more than 350 million journeys since opening, with journeys rising from 138 million in 2022/23 to 210 million in 2023/24. Its business case gave a benefits-to-cost ratio of 1.9 to 1. Most monetised benefits come from time savings and congestion relief.

What the criticism got right. The overspend and the delay were real and remain on the record. What changed after opening is that the judgement of the whole investment could include how the line was used.

What is still unfinished. Transport for London calls its findings early and interim, says long-term value for money needs longer monitoring, and says about 38% of Elizabeth Line demand either did not exist before or cannot be traced to a source. Its figures are its own and the document does not say they were independently audited. The 1.9 to 1 ratio is close in size to the roughly twofold ratio IFS reports for Sure Start, and like IFS’s it rests on projection.

The difference. A railway’s use can be counted within weeks of opening. An early years programme’s main outcomes appear over a childhood. The Elizabeth Line shows the short end of the same lesson: a project judged on cost before it opened, judged again on use after it opened, and still being evaluated for value. Sure Start’s equivalent wait is measured in decades. As the YFL essay Public Financing Models and Cost Outcomes: Crossrail, HS2 and the Shadow-Toll Roads notes, Transport for London’s estimate of the line’s economic contribution is the sponsor’s own projection, not an independently audited outcome.

When it could be known, and when it was judged

The arithmetic of age. Sure Start was introduced in 1999. A child born that year reached school age in 2004 or 2005 and sat GCSEs in 2015 or 2016. Outcomes at 16 could not be observed in any cohort before the middle of the 2010s. Earnings, the largest part of IFS’s benefit estimate, can only be projected from those results.

The sequence of judgement. The 2007 committee examined cost-effectiveness while the national evaluation was still running. The Department told the committee that its five-year evaluation would finish in 2008–09, and that medium- and long-term effects on academic performance and health would take longer. The first impact report (2005) called its own findings “at best, early indications” and said stronger grounds for conclusions would exist once children were followed up at 36 months. Eisenstadt wrote in 2011 that programmes took roughly three years to become operational, mostly commissioning construction, while outcomes were expected within months. The age-three results were published in 2008, the age-five report after June 2010, and the age-seven report, with no consistent child outcomes, in 2012 (the Department for Education research brief is dated June 2012). The ring-fence on children’s centre funding was removed in 2011, and early intervention funding then fell by more than 64% by 2017/18. The IFS reports on teenage outcomes came after that, in 2019 and 2025.

The three conditions applied. Measured against the conditions set out in “When to test”, the sources read show the following.

• A measure suited to the stage. Eisenstadt wrote that programmes took roughly three years to become operational while outcomes were expected within months, so early measures often described a programme that was still being built. The outcomes IFS later reported, such as GCSE results and teenage hospital admissions, could not be measured until children reached those ages.
• Enough elapsed time. Parliament examined value for money in 2007. The first children to attend sat their GCSEs in 2015 or 2016, so the judgement came years before the outcomes IFS later reported.
• A baseline and comparison fixed in advance. Here the sources point the other way. The national evaluation compared whole areas, and its age-five authors called their effects “putative” because the two samples were surveyed two years apart by different teams. Eisenstadt gave reasons a randomised trial was not run. IFS’s method was developed after the programme had run, and changing how areas were weighted removed an earlier crime result.

The sources read do not show test points and failure criteria set before the results arrived. That is a reading of those sources, not a finding that no plan existed.

What the long-run results add. IFS reports that for educational attainment and behaviour in school, effects were larger in adolescence than at some earlier points in childhood. Hospital admissions rose at age one and fell by 15. Mental-health effects appeared in the teenage years. Evidence from ages 3, 5 and 7 therefore gave an incomplete and in places different picture from evidence at 16.

Other early-years programmes. Donna Molloy, then Deputy Chief Executive of Foundations, the What Works Centre for Children and Families, said in the Health Foundation podcast with Eisenstadt (28 May 2024) that “we do have a tendency just to look at short-term effects from many of our government interventions and flagship programme and we just don’t stay the course”. She cited the Family Nurse Partnership: early evaluations “were disappointing” and funding was withdrawn, but “6, 7 years on” the data showed gains in school readiness for disadvantaged children. “We didn’t stay the course”, she said.

A hypothesis, flagged. The timeline fits a claim that the cost of Sure Start was judged before outcomes could be measured, and that seeing results required a commitment lasting a generation. Two limits apply. First, the long-run results were not uniformly positive: crime, numbers in care and employment did not show clear gains even with time. A longer wait would not necessarily have produced a better verdict, only a better-informed one. Second, IFS studied the children who attended. Effects on their own children, a further generation on, have not been measured.

Why robust evidence was late

The record. The 2007 committee found weak cost data and monitoring. A National Audit Office review in 2021 found that in 2019 only 8% of spend on major government projects (£35 billion of £432 billion) had robust evaluation plans, and 77 of 108 projects, about 64% of spend, had none. Seven of 16 chief analysts cited a lack of pressure and demand from senior policy colleagues, and a 2019 review by the Prime Minister’s Implementation Unit found few incentives to evaluate and few consequences for not doing so.

An advocate’s account, 2013. A report by the WAVE Trust and a Department for Education practitioner group, with forewords by two ministers, said there is “much less data from the UK” than the United States on early years programmes, and gave as a reason “a historical reluctance to fund early years’ research because of its assumed long-term nature”. It said UK evaluations were often on low budgets, limited in scope and sometimes methodologically flawed, and that the UK studies “lack the breadth and rigour of the American evaluations” (WAVE Trust, 2013, chapter 3 and Appendix 4). WAVE is an advocacy charity, and the report states that its views do not summarise government policy. It is a published statement of the problem by a body that also argued for more early years investment. The YFL essay Examining the Evaluators: Why Are They Listened To? describes the WAVE Trust’s status, funding and governance.

Why no randomised trial. Eisenstadt said in the Health Foundation podcast (28 May 2024) that “one of the big arguments was in the design that we had of letting local determination on what was on offer”, which meant a randomised controlled trial could not be run. In a 2022 review she gave further reasons: the diversity of local programmes; ministers’ view that it was unfair to consult local parents on what they wanted and then randomly refuse some of them the new services; and, as Sure Start expanded, researchers’ difficulty finding poor areas without a programme to compare against (Therapeutic Care Journal, 1 June 2022). Penny Kane wrote a commentary, “Sure Start Local Programmes in England”, published alongside the age-three paper in The Lancet (8 November 2008). Lewis (2011, p. 80, with Kane’s specific criticisms in her footnote 14) says Kane argued that it was too early, and too difficult, to assess Sure Start’s impact, pointing to the variation between local programmes and to comparison data collected two years before the Sure Start data. A press report of 15 November 2008 (Medical News Today) has her saying that the rejection of a randomised trial “has made interim evaluations difficult”. The commentary itself has not been read, so that wording is unchecked. In a 2011 LSE blog post Eisenstadt wrote that Sure Start was not built on rigorously evaluated evidence, that programmes took roughly three years to become operational, mostly commissioning construction, while outcomes were expected within months, and that “some of the disappointing results of the first outcome study were because very little was actually happening”. She also wrote that “the disappointments were about the failure to deliver any real cognitive gains for children”, while parents saw social and economic gains. That choice shaped what could later be measured.

Design. In the same podcast Eisenstadt said the evaluation was designed so that progress could be reviewed about every two years and the programme redesigned in response, giving the teenage mothers example above. Her book Providing a Sure Start (2011) is described in its publisher’s chapter summary as saying many changes to the service model responded to the evaluation results; the chapter itself has not been read. The published reports came out in 2005, 2008, 2010 and 2012 (Eisenstadt’s review gives 2007 for the second), so the gaps were not always two years. Molloy said that long-term follow-up and evaluation is “just something we so rarely see in this country”, citing the IFS work as an unusual example. On Sure Start she described “the time lag in the evaluation results” set against the interest in making policy decisions, and said that “some decisions would have been made when we just didn’t have the information about what was being achieved by the programme”.

Data. The IFS evaluation used administrative data held across departments and a method developed after the programme had run. IFS notes a lack of data on how services were delivered and used, which limits what can be said about mechanisms. Its advice for Family Hubs is to build evaluation in from the start.

Who supplied the evidence. Much of the evidence in this essay came from bodies outside government: the Institute for Fiscal Studies, with funding from the Nuffield Foundation; the Sutton Trust, a charity founded in 1997 by Sir Peter Lampl; the WAVE Trust; and University of East Anglia teams commissioned by government to analyse serious case reviews. The YFL essay Examining the Evaluators: Why Are They Listened To? describes what each of these bodies is, how it is paid for and who checks it. It notes that the safeguards the bodies describe are their own statements, so the standing of this evidence rests partly on the bodies’ own accounts of themselves.

A current case. The YFL essay After Munro. Who is listening now? describes the reverse timing problem in children’s social care: a national reform rolled out after trials in ten areas, with a further seven trialled for under a year (figures as reported by Community Care). Sure Start’s cost was judged before outcomes could be measured; that reform was scaled before testing had run its course. The two cases are different eras and designs, and this is an observation, not a comparison.

Hypothesis, not finding. These sources are consistent with weak system-wide incentives to evaluate, not the preferences of one party. The YFL essay Bureaucratic Incentives and the Limits of New Public Management sets out theories of why government bodies lack the discipline that markets impose; it does not itself address evaluation. No comparison with other countries, or between parties, has been read, so whether the UK is unusual is unknown.

What was cut

Funding. The coalition government removed the ring-fence on children’s centre funding in 2011, merging it into the Early Intervention Grant. The Sutton Trust, a charity founded in 1997 by Sir Peter Lampl, reported that the early intervention allocation to local authorities, of which children’s centres were only one part, fell by more than 64%, from £3.28 billion in 2010/11 to £1.17 billion in 2017/18 (Smith et al., 2018, p. 10). The figure is taken from research by Action for Children, the National Children’s Bureau and The Children’s Society (2017), and ignores councils’ other income. Wider spending on children and young people’s services fell by a smaller 16% in real terms between 2010/11 and 2015/16 (p. 10). In the Sutton Trust’s own survey, 69% of responding authorities said their children’s centre budget had decreased over the previous two financial years (p. 25), and 84% said financial pressures were the main driver of change (p. 5).

Thinner funding. Eisenstadt said in the same podcast that the most intensive funding was kept in the original centres in poor areas under Labour, with the others funded more thinly. When responsibility moved to the Coalition government, she said, “they spread it thinly across and that made a massive difference to impact”. She described the results as coming “only from the first 500 that were very well funded”. That is her reading of the IFS findings, not evidence of her own, and it was recorded in May 2024, a year before the IFS report of May 2025. That report agrees that the original local programmes kept much higher budgets per centre: spending per centre in 2010 was less than a third of the 2003 level, though former local programmes “maintained much higher budgets than newer centres” (Carneiro et al., 2025, p. 18). Using an updated method, it finds that the effects of the original local programmes and the later children’s centres “seem to be similar” on test scores at ages 5, 7 and 11, and that both types of centre were effective on special educational needs support (pp. 75–76, 98–99). At age 7, children near a local programme improved about twice as much. The exception is GCSE results: they are available only for children who grew up near programmes opened by 2003, so the lifetime earnings projections rest on those programmes (p. 76, footnote 22). Eisenstadt’s remark is therefore best read as her reading of findings available before the 2025 report, which qualifies it. She also defended the expansion: if Sure Start “had just stayed with those 500 programmes in poor areas, it would’ve been cut immediately”.

Closures. The Sutton Trust reported 3,632 centres at the peak in August 2009 (Smith et al., 2018, p. 4). The national database recorded a 14% fall by October 2017, and the report’s own survey of local authorities found 16%. Counting only registered centres, the fall was more than 30%, which suggests more than 1,000 centres may have closed (p. 4). The report says the true figure lies somewhere between about 16% and 30% or more depending on definition, because “children’s centre” had no clear definition (p. 12). Many remaining centres were part-time, merged into clusters, or targeted at referred families.

Guidance. National guidance in 2013 shifted focus toward families in greatest need and dropped the requirement for full daycare and qualified teachers in the most disadvantaged areas.

Later programmes. The Family Hubs and Start for Life programme began in 2022–23. IFS noted that its spending was under £600 million in a recent year, against £2.7 billion in Sure Start’s peak year, 2009–10, in 2023–24 prices.

Critics of the model, supporters of the purpose

Several of those who criticised Sure Start’s design backed its aim. Speaking at a Conservative Party conference fringe event in 2013, Eisenstadt, who started the programme, said: “We really have missed a trick, and I say that as someone who started the programme. One of the things we got wrong is that we tried to get children’s centres to do too many things.” The trick she described was language development from birth to two. At the same event she argued for concentrating centres in the poorest areas, with schools and health centres covering the rest (Children & Young People Now, 1 October 2013). Frank Field, a Labour MP, reviewed poverty and life chances for the Prime Minister in 2010. He wrote: “I do not believe that we can make the Foundation Years the success they must become without Sure Start. But the concept of the Sure Start Children’s Centre needs radical reform.” He meant a return to the original aim of “providing greatest help to the most disadvantaged”, while avoiding Sure Start “becoming simply a service for poorer families” (Field, 2010, p. 22).

Sure Start in party politics. The sources read show Sure Start’s future being argued in election campaigns. A Nursery World report of 2 April 2015, a month before that year’s general election, quotes Tristram Hunt, then Labour’s shadow education secretary: “The Tories pose a real threat to Sure Start”. The report says Labour warned that over 1,000 centres could close if the Conservatives won; that was a Labour projection made during the campaign, not an official count. The same report says David Cameron had said before the 2010 election that he would protect Sure Start. Eisenstadt wrote in 2011 that Sure Start established a government role in regulating and partly funding services for young children. These are two sources. They do not show how far Sure Start was treated as a political target, or by whom.

This is a pattern across a small number of sources, not a finding about critics generally. It does not show that criticism caused the cuts. The documented mechanism in the sources read is the change in how funding was allocated and ring-fenced.

What the evidence does and does not show

Shows: gains in education and teenage health, a pattern of earlier contact followed by lower later admissions, and benefits that grew with age. Effects were small and uneven in places, and absent in others.

Does not show: that Sure Start reduced offending or the number of children in care. It does not show that the sums repaid the cost to government within a short period. The roughly £2 of benefit per £1 spent is a projection that counts private earnings and future tax.

Does not settle: why robust UK evidence on early years programmes was scarce. Published statements, from the 2007 committee, the National Audit Office, the evaluation’s own commentators and an advocacy charity, describe the gap. They do not establish its cause.

Does not settle: which services produced which effects, including whether joined-up working made the difference, since delivery data are lacking. Nor does it settle whether the early critics or the later supporters were more right, because they were describing different programmes at different stages. It does show that the programme’s cost was being judged years before the first cohorts could be assessed at 16, and that the later evidence differed from the earlier picture.

---

References

Carneiro, P., Cattan, S., Conti, G., Crawford, C., Farquharson, C. and Ridpath, N. (May 2025). The short- and medium-term effects of Sure Start on children’s outcomes. Institute for Fiscal Studies.

Carneiro, P., Cattan, S., Conti, G., Crawford, C., Drayton, E., Farquharson, C. and Ridpath, N. (October 2024). The effect of Sure Start on youth misbehaviour, crime and contacts with children’s social care. Institute for Fiscal Studies. (Superseded by the May 2025 report.)

House of Commons Committee of Public Accounts (2007). Sure Start Children’s Centres, thirty-eighth report, session 2006–07. House of Commons.

National Evaluation of Sure Start team. The impact of Sure Start local programmes on seven year olds and their families. Department for Education, Research Report 220.

Melhuish, E., Belsky, J., Leyland, A.H., Barnes, J. and the National Evaluation of Sure Start Research Team (2008). Effects of fully-established Sure Start Local Programmes on 3-year-old children and their families living in England: a quasi-experimental observational study. The Lancet, 372(9650), 1641-1647.

National Evaluation of Sure Start team. The Impact of Sure Start Local Programmes on Five Year Olds and Their Families. DFE-RR067. Department for Education.

Kane, P. (2008). Sure Start local programmes in England. The Lancet, 372(9650), 1610-1612.

Lewis, J. (2011). From Sure Start to children’s centres: an analysis of policy change in English early years programmes. Journal of Social Policy, 40(1), 71-88.

National Evaluation of Sure Start Research Team (November 2005). Early Impacts of Sure Start Local Programmes on Children and Families: Report of the Cross-sectional Study of 9- and 36-Month Old Children and their Families. Research Report NESS/2005/FR/013. Department for Education and Skills.

Medical News Today (15 November 2008). Report of the Lancet Sure Start local programmes study and commentary.

NESS Research Team (March 2008). The Impact of Sure Start Local Programmes on Three Year Olds and Their Families. Research Report NESS/2008/FR/027. Department for Children, Schools and Families. ISBN 978 1 84775 103 4.

Melhuish, E., Belsky, J., Anning, A., Ball, M., Barnes, J., Romaniuk, H., Leyland, A. and the NESS Research Team (2007). Variation in community intervention programmes and consequences for children and families: the example of Sure Start Local Programmes. Journal of Child Psychology and Psychiatry, 48(6), 543-551.

Sammons, P., Hall, J., Smees, R., Goff, J., with Sylva, K., Smith, T., Evangelou, M., Eisenstadt, N. and Smith, G. (December 2015). Evaluation of children’s centres in England: the impact of children’s centres. Department for Education.

Smith, G., Sylva, K., Smith, T., Sammons, P. and Omonigho, A. (4 April 2018). Stop Start: Survival, decline or closure? Children’s centres in England, 2018. Sutton Trust.

Department of Health and Home Office (January 2003). The Victoria Climbié Inquiry: report of an inquiry by Lord Laming. Cm 5730.

Chief Secretary to the Treasury (September 2003). Every Child Matters. Cm 5860. Presented to Parliament by command.

Brandon, M., Belderson, P., Warren, C., Howe, D., Gardner, R., Dodsworth, J. and Black, J. (2008). Analysing child deaths and serious injury through abuse and neglect: what can we learn? A biennial analysis of serious case reviews 2003–05. DCSF-RR023. Department for Children, Schools and Families.

Sidebotham, P., Brandon, M., Bailey, S., Belderson, P., Dodsworth, J., Garstang, J., Harrison, E., Retzer, A. and Sorensen, P. (May 2016). Pathways to harm, pathways to protection: a triennial analysis of serious case reviews 2011 to 2014: final report. Department for Education.

Sylva, K., Goff, J., Eisenstadt, N., Smith, T., Hall, J., Evangelou, M., Smith, G. and Sammons, P. (June 2015). Organisation, services and reach of children’s centres: Evaluation of children’s centres in England (ECCE, Strand 3), research brief. DFE-RB433. University of Oxford.

Sammons, P., Sylva, K., Hall, J., Evangelou, M. and Smees, R. (2022). Challenges facing interventions to promote equity in the early years: exploring the ‘impact’, legacy and lessons learned from a national evaluation of Children’s Centres in England. Oxford Review of Education.

National Audit Office (December 2021). Evaluating government spending. National Audit Office.

Tunstill, J., Meadows, P., Allnock, D., Akhurst, S., Garbers, C. and the NESS Implementation Team (January 2005). Implementing Sure Start local programmes: An in-depth study. Department for Education and Skills. ISBN 1 84478 408 8.

NESS Implementation Team (November 2006). Outreach and Home Visiting Services in Sure Start Local Programmes. National evaluation report 017, NESS/2006/FR/017. Department for Education and Skills. ISBN 978 1 84478 848 4.

National Audit Office (May 2019). Completing Crossrail. National Audit Office.

Transport for London (2023). Investment Programme Report, Quarter 4 2022/23. Transport for London.

Transport for London (30 May 2025). Elizabeth line post-opening evaluation: full report. Transport for London.

National Audit Office (July 2021). Crossrail: a progress update. HC 299, session 2021-22. National Audit Office.

Transport for London (December 2024). Evidencing the value of the Elizabeth line. Transport for London.

Murray, J. and Farrington, D.P. (2005). Parental imprisonment: effects on boys’ antisocial behaviour and delinquency through the life-course. Journal of Child Psychology and Psychiatry, 46(12), 1269-1278.

WAVE Trust (June 2013, second edition). Conception to age 2 – the age of opportunity. WAVE Trust. Addendum to the Department for Education’s Supporting Families in the Foundation Years. Chapter 3 and Appendix 4 read; other chapters searched by keyword only.

Karoly, L.A., Kilburn, M.R. and Cannon, J.S. (2005). Early Childhood Interventions: Proven Results, Future Promise. MG-341. Santa Monica, CA: RAND Corporation. (18 October 2005.)

Cohen, M.A., Piquero, A.R. and Jennings, W.G. (2010). Estimating the costs of bad outcomes for at-risk youth and the benefits of early childhood interventions to reduce them. Criminal Justice Policy Review, 21(4), 391-434.

National Academies of Sciences, Engineering, and Medicine (2016). Advancing the Power of Economic Evidence to Inform Investments in Children, Youth, and Families. National Academies Press.

Nursery World (2 April 2015). Conservatives “pose a real threat” to Sure Start, warns Labour. Nursery World.

Eisenstadt, N. and Molloy, D. (28 May 2024). Sure Start: a model for long-term policymaking? The Health Foundation podcast, episode 44, hosted by Jennifer Dixon. Quoted from the published transcript, which contains labelling errors.

Eisenstadt, N. (2011). Providing a Sure Start: How government discovered early childhood. Policy Press.

Jozwiak, G. (1 October 2013). Conservative conference: children’s centres fail to deliver, says Sure Start group chair. Children & Young People Now.

Field, F. (December 2010). The Foundation Years: preventing poor children becoming poor adults. The report of the Independent Review on Poverty and Life Chances. Cabinet Office.

Eisenstadt, N. (1 June 2022). Sure Start review. Therapeutic Care Journal.

Eisenstadt, N. (28 September 2011). Sure Start blog post, LSE British Politics and Policy blog.

Nursery World (19 October 2011). Report of Eisenstadt: Sure Start a policy success but has “lost the babies”.