{"paper_id":"6f66a10d-4790-455b-876c-cad7335dd23e","body_text":"1 \n \nTitle:  1 \nImplementat ion of national guid el in es  o n ant enat al magne sium sulf at e for neonatal 2 \nneur opro tection in En g lan d, Sc ot lan d and Wale s : Ext e nde d evaluatio n of  th e effectiv ene ss 3 \nand cost-ef fec t iv en es s of t h e N at ion al PReCeP T P rogramme  4 \n 5 \nWord count :  3947 6 \n 7 \nAu thor s: 8 \nHannah B Edwards(1,2), Carlos Sillero-Rejon(1,2), Hugh McLeod(1,2), Elizabeth M Hill(1,2), 9 \nBrent C Opmeer(2,3), Colin Peters(4), David Odd(5,6), Frank de Vocht(1,2), Karen Luyt(7,8) 10 \nInstit utions: 11 \n1.  Population Health Sciences, Bristol Medical School, Univer sity of Brist ol, Beacon House 12 \nQueens Road Brist ol, BS8 1QU, UK 13 \n2.  The National Institute for Health and Care Resear ch Applied Resear ch Collaboration W est 14 \n(NIHR ARC West) at Univer sity Hospitals Bristol and Weston NHS Founda tion T rust, 9th 15 \nFloor , Whit efriars, Lewins Mead, Brist ol, BS1 2NT , UK 16 \n3.  Vilans, National Center of Expertise for Long T erm Car e, P . O. Box 8228, 3503 RE, Utrecht, 17 \nThe Netherlands 18 \n4.  Neonatology , Royal Hospit al for Childr en Glasgow , 1345 Govan Road, Govan, Glasgow , 19 \nG51 4TF , UK 20 \n5.  Population Medicine, Cardiff Univer sity , Heath Park, Cardiff , CF14 4YS, UK. 21 \n6.  Neonatology ,  Car diff and V ale University Health Board Univer sity Hospit al of W ales, 22 \nHeath Park, Cardiff , CF14 4XW , UK 23 \n7.  T ranslational Health Sciences, Bristol Medical School, University of Bristol, Beacon House 24 \nQueens Road Brist ol, BS8 1QU, UK 25 \n8.  Neonatology , St Michael's Hospital, Southwell Street, Bristol, BS2 8EG, UK 26 \n 27 \n 28 \nCorresponding a ut hor : Hannah B Edwards Hannah.edwards@bristol.ac.uk  29 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \nNOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.\n\n2 \n \nAB ST R ACT  30 \nBackgroun d : Since 2015, UK national guidelines have recommended antenatal magnesium 31 \nsulfate (MgSO 4 ) for mothers in preterm labour (<30 weeks’ gestation) to reduce the risk of 32 \ncerebral palsy in the preterm baby. However, implementation of this guideline in clinical 33 \np r a c t i c e  w a s  s l o w ,  a n d  Mg S O 4  use varied between maternity units. In 2018, the PReCePT 34 \nprogramme, an evidence-based Quality Improvement intervention to improve use of 35 \nMgSO 4 ,  was rolled-out across England. Earlier evaluation found this programme to be 36 \neffective and cost-effective over the first 12 months. We extended the original evaluation to 37 \ndetermine the programme’s longer-term impact over four years, its impact in later preterm 38 \nbirths, impact of the COVID-19 pandemic, and to compare MgSO 4  use in England, Scotland, 39 \nand Wales. 40 \nMetho ds: Quasi-experimental longitudinal study using data from the National Neonatal 41 \nResearch Database on babies born <30 weeks’ gestation and admitted to an NHS neonatal 42 \nunit. Primary outcome was the proportion of eligible mothers receiving MgSO 4 , aggregated 43 \nto the national level. Impact of PReCePT on MgSO 4 use was estimated using multivariable 44 \nlinear regression. The net monetary benefit (NMB) of the programme was estimated. 45 \nResults: MgSO 4  administration rose from  65.8% in 2017 to 85.5% in 2022 in En gland. 46 \nPReCePT was associated with 5.8 percentage points improvement in uptake (95%CI 2.69 to 47 \n8.86, p<0.001). Improvement was greater when including older preterm births (<34 weeks’ 48 \ngestation, 8.67 per centage points, 95%CI 6.38 to 10.96, p<0.001). Mo st gains occurred in the 49 \nfirst two years following implementation. PReCePT had a NMB of £597,000 with 89% 50 \nprobability of being cost-effective. Following implementation, English up take appeared to 51 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n3 \n \naccelerate compared to Scotland and Wales. There was some decline in use coinciding with 52 \nthe onset of the pandemic. 53 \nConclusions: The PReCePT Quality Improvement programme cost-effectively improved use 54 \nof antenatal MgSO 4 , with benefits to the babies who have been protected from cerebral 55 \npalsy.  56 \n 57 \nWhat is alr eady kno wn on t his  t opic  58 \n• Ant enatal magnesium sulphate (MgSO 4 ) reduces the risk of cer ebral palsy in babies born 59 \npret erm. 60 \n• The National PReCePT Quality Improvement Pr ogramme (NPP) eff ectively and cost-61 \neff ectively improved use of MgSO 4  in England in the first 12 months of implementation, 62 \nbut sustaining quality improvements over ti me is often challenging. 63 \nWhat th is  stu dy ad ds  64 \n• Using a quasi-experimental design and r outinely collected, longitudinal, patient-level 65 \ndata, this study found that the NPP had sustained effectiveness and cost-effectiveness 66 \nover four year s following implementation.  67 \n• Impr ovement may have been accelerated in England, compared to Scotland and W ales, 68 \nwhere the NPP was not formally implement ed. 69 \nHow t his s t udy might aff ec t  research, practice or  polic y  70 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n4 \n \n• This study demonstrates that dedicated national pr ogrammes can cost-effectively 71 \nachieve impr ovements in perinatal care. The PReCePT model could be used as an 72 \nimplementation blueprint for other quality impr ovement initiatives in perinatal car e. 73 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n5 \n \nINTR OD UCTI ON  74 \nSince 2015 the World Health Organisation (WHO)(1) and the UK National Institute for Health 75 \nand Care Excellence (NICE)(2) have recommended administration of magnesium sulfate 76 \n(MgSO 4 ) in preterm deliveries <30 weeks’ gestation as a core part of maternity care. This 77 \nfollows strong evidence that when given antenatally to women in preterm labour, MgSO 4  78 \nreduces the risk of cerebral palsy (CP) in preterm babies by around 30%(3). Historically, use 79 \nof this treatment has been inconsistent, with only 64% of eligible women in England being 80 \ntreated in 2017. High regional variation in uptake also indicates inequalities in perinatal 81 \ncare(4).  82 \nAs well as the significant impact of CP on affected individuals and their families(5), there are 83 \nlifetime societal costs of approximately £1m per affected indiv idual(6), and £1.8 billion 84 \nannually on NHS clinical negligence litigation (half of the total NHS litigation expenditure)(7). 85 \nIncidence of CP has been estimated at around 1.5 per 1000 livebirths in the UK(8), with 86 \npreterm birth as the leading risk factor(9-11). This highlights the importance of funding 87 \neffective, and cost-effective, strategies to reduce the risk of CP associated with preterm 88 \nbirth. It is estimated that one case o f CP can be prevented, below 30 weeks’ gestation, for 89 \nevery 37 mothers treated with MgSO 4 , and around 200 cases of CP per year could be 90 \navoided by con sistent administration of MgSO 4  during labour(3). 91 \nIn 2018, NHS En gland rolled-out the National PReCePT (Prevention of cerebral palsy in 92 \npreterm labour) Programme (NPP). This was a quality improvement (QI) programme for 93 \nmaternity units, providing clinical guidance, training, learning resources, midwife backfill 94 \nfunding, and QI support, to improve maternity staff awareness, and increase use of MgSO 4  95 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n6 \n \nfor mothers in preterm labour. The aim was to reach ≥85% uptake in eligible mothers across 96 \nall maternity units in England. The programme was delivered by regional Academic Health 97 \nScience Networks (AHSNs, now Health Innovation Networks). Evaluation of the first 12 98 \nmonths of the programme found it to be effective, improving MgSO 4  use by an estimated 99 \n6.3 percentage points, with an estimated net monetary benefit of £866 per preterm baby, 100 \nand >95% probability of being cost-effective(12). However it is unknown whether these 101 \nimprovements were sustained over time, and sustainability in large scale implementation  102 \nprogrammes is often a problem (and measurement of sustained effect often neglected)(13). 103 \nThe primary aim of this study, therefore, was to evaluate the NPP’s longer-term, sustained 104 \neffectiveness and cost-effectiveness over the first four years following implementation. A 105 \nkey secondary aim was to explore the impact of the NPP on all babies born up to 34 weeks’ 106 \ngestation; NICE guidelines recommend treatment for births up to 30 weeks’, and 107 \n‘consideration of treatment’ for older preterm births up to 34 weeks’. Other secondary aims 108 \nwere to explore the impact of the COVID-19 pandemic on MgSO 4  use, and to compare 109 \nMgSO 4  use in England with that in the devolved nations Scotland and Wales. This evaluation 110 \nis part of a larger programme of work, including qualitative interviews to explore how the 111 \ndevolved nations were responding to the NICE guidance, reported elsewhere(14). 112 \nMET H O DS  113 \nDe sign  114 \nThis was a quasi-experimental study for the evaluation of the NPP’s effectiveness and cost-115 \neffectiveness. The pre-registered Statistical Analysis Plan and Health Economic Analysis Plan 116 \nwere uploaded to the Open Science Framework prior to analyses: ht tps: //o sf.io/be76 s/ . 117 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n7 \n \nInt ervent io n  118 \nThe intervention being implemented was the National PReCePT Programme (NP P), as 119 \ndescribed above and fully detailed elsewhere(12), for the adoption of MgSO 4  a s  a  120 \nneuroprotectant in preterm births. 121 \nSett ing  122 \nNHS maternity units in England, Scotland and Wales. Within maternity units, analysis was 123 \nperformed on aggregated data on babies born preterm <30 weeks’ gestation, and admitted 124 \nto an NHS neonatal unit, between January 2014 to December 2022. All maternity units in 125 \nEngland, Scotland and Wales were included, excepting the five units in England that took 126 \npart in the original PReCePT pilot study(15) and were therefore not part of the NPP. 127 \nData so urces 128 \nData on eligible babies and their mothers were obtained from the National Neonatal 129 \nResearch Database (NNRD), which holds individual-level, pseudonymised, routinely 130 \ncollected patient data on babies admitted to an NHS neonatal unit. Costs associated with 131 \nthe NPP were estimated in the original evaluation(12). 132 \nEffe c t iv e ness e valuation 133 \nOutcome 134 \nThe main implementation outcome was MgSO 4  uptake over time. MgSO 4  uptake was 135 \ndefined as the proportion of eligible mothers recorded as receiving MgSO 4  a t a ma terni ty  136 \nunit, expressed as a percentage. For consistency with nationally reported data, mothers  137 \nw i t h  m i s s i n g  Mg S O 4  data were excluded from this calculation. This was computed per 138 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n8 \n \nmonth per unit, and reported as a percentage. For consistency with nationally reported data 139 \nonly data on singletons and the first born (i.e. one infant) from each multiple birth were 140 \nincluded in the calculation.  141 \nDescriptive analysis 142 \nMaternity unit, population characteristics, and MgSO 4  use were descriptively reported by 143 \nnation and time-period. 144 \nPrimary analysis  145 \nPrimary analysis was an interrupted time series using English data aggregated to the 146 \nnational level (mean national MgSO 4  uptake per month, across all English maternity units). A 147 \nmultivariable linear regression model was used to estimate the difference in mean MgSO 4  148 \nuptake from before (the 1-year period before) to after (the four-years follow-up) 149 \nimplementation of the NPP in England. The model adjusted for an underlying linear time 150 \ntrend, and mother and baby characteristics aggregated nationally per month (mean 151 \nmaternal age, Index of Multiple Deprivation (IMD) decile(16), baby’s birthweight adjusted 152 \nfor gestational age as a z-score, and proportion reported smokers, white British ethnicity, 153 \ntype of birth (c-section versus vaginal delivery), and multiple births. The model was further 154 \nadjusted for a non-linear temporal component to account for the ceiling effect and 155 \nreduction in rate of change at levels close to the ceiling. Data on paternal age and ethnicity 156 \nwere explored as potential confounding factors, but were excluded due to high levels of 157 \nmissing data, and expected collinearity with other variables (maternal and paternal age 158 \ntends to correlate, as does maternal and paternal ethnicity). Potential interaction was 159 \nexplored between mean MgSO 4  uptake and level of maternity unit (Neonatal Intensive Care 160 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n9 \n \nUnit (NICU), the highest-level unit versus Special Care Baby Unit (SCBU) or Local Neonatal 161 \nUnit (LNU), lower level units)). This was because it was anticipated that performance might 162 \ndiffer by type of unit: data from the original study indicated that smaller units tended to  163 \nhave lower starting uptake levels, so more room for improvement compared to larger 164 \nNICUs. 165 \nSensitivity and subgroup analyses 166 \nAs sensitivity analyses the above model was run on data aggregated to (a) the maternity-167 \nunit-level rather than the national level, and (b) the individual rather than national level. 168 \nThese two models additionally adjusted for type of maternity unit (NICU vs SCBU/LNU), 169 \nregional clustering by AHSN, and were weighted on the number of eligible births per unit 170 \nper month. Other sensitivity analyses included assessing (c) the impact of excluding a ‘fuzzy’ 171 \nimplementation start window of +/-2 months, to account for some units starting slightly 172 \nearlier or later than their officially recorded start date; (d) the impact of excluding the final 2 173 \nmonths of data, due to concerns about completeness of the most recent data for some 174 \nunits; (e) the impact of using a longer pre-NPP comparison period of four years; and (f) the 175 \nimpact of including more mature pre-term babies up to 34 weeks’ gestation in the analysis. 176 \nA sub-group analysis was performed on the 40 units that had participated in a connected 177 \nstudy, an RCT nested within the main NPP(17), as their performance could plausibly differ 178 \nfrom other maternity units.  179 \nEconomic e valuation  180 \nMgSO 4  treatment cost-effectiveness 181 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n10  \n \nEconomic analysis combines evidence of the treatment (MgSO 4 ) effect with evidence of the 182 \nimplementation (NPP) effect(18). For the former, estimates of the cost-effectiveness of 183 \nMgSO 4  treatment were adopted from Bickford and colleagues' results(6, 19, 20), with their 184 \nestimates converted to GBP currency and 2019 prices (Supplementary Table 1). This analysis 185 \naccounts for the health impacts and cost savings of MgSO 4  administered in births <30 and 186 \n<32 weeks’ gestation.  187 \nNPP implementation costs and effectiveness  188 \nThe mean implementation cost per unit of the NPP was estimated from data supplied by the 189 \nNPP team and PReCePT study team, and reported previously(12, 17). The total cost of the 190 \nNPP at a national level was estimated at £936,747(12) (which includes the funded backfill of 191 \nclinical time for NPP ‘champion’ midwives at each maternity unit, and regional support from 192 \nAHSNs). 193 \nFrom the multivariable linear regression model described above, we estimated the NPP 194 \neffectiveness as the difference between the predicted level of MgSO 4  u se o ver time 195 \ncompared to a counterfactual level of MgSO 4  use, representing what may have occurred in 196 \nthe absence of the NPP, assuming a continuation of the pre-NPP trend in MgSO 4  uptake. The 197 \nmain measure of NPP effectiveness was the area-between-the-curves. Primary analysis used 198 \na linear distribution to estimate the counterfactual based on the pre-NPP predicted trend. 199 \nSensitivity analysis used a beta distribution to estimate this counterfactual, to account for 200 \nMgSO 4 uptake as a proportion. 201 \nPolicy cost-effectiveness analysis 202 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n11  \n \nThe cost-effectiveness analysis was conducted from a societal lifetime perspective. Net 203 \nmonetary benefit (NMB) of the NPP was estimated over the four years since its launch, by 204 \ncombining analysis of the cost s and effectiveness of the NPP with the lifetime societal cost, 205 \nand health gains associated with MgSO 4  treatment. This analysis used a framework 206 \npreviously developed to conduct economic evaluations of implementation initiatives(21). A 207 \nwillingness-to-pay threshold of £20,000 per Quality Adjusted Life Year (Q ALY) gained was 208 \nused to determine NPP cost-effectiveness, following NICE guidelines(22). A positive NMB 209 \nindicated that the implementation initiative was cost-effective. The net increment of the 210 \nnumber of patients that received MgSO 4 , and the implementation cost-effectiveness per 211 \nadditional patient treated, was also estimated. The analysis used the area-between-the-212 \ncurves estimate of NPP effectiveness with a linear counterfactual, and a sensitivity analysis 213 \nusing a beta counterfactual.  214 \nProbabilistic analysis was conducted using a Monte Carlo simulation with 10,000 samples 215 \ndrawn from parameter distributions. Point estimates, probabilistic distribution assumptions, 216 \nand parameter source estimates are reported in Supplementary Table 2. Cost-effectiveness 217 \nplanes and cost-effectiveness acceptability curves were plotted for willingness-to-pay 218 \nthresholds from zero to £100,000 per QALY gained for the policy cost-effectiveness of the 219 \nNPP intervention. 220 \nSecondary economic analysis 221 \nAs evidence on the lifetime cost-effectiveness of antenatal MgSO 4  covers babies born up to 222 \n32 weeks’ gestation, cost-effectiveness analysis was performed only for babies <30 and <32 223 \nweeks’ gestation.  224 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n12  \n \nStatistical software Stata version 17 and R version 4.3.1 were used for all statistical analyses. 225 \nRESUL TS  226 \nBaseline character istics  227 \nIn 2017, the year before NPP roll-out, a total of 4091 babies born under 30 weeks’ 228 \ngestational age were admitted to neonatal units in England, 296 in Scotland, and 182 in 229 \nWales. The majority of births were in maternity units with a NICU (62.6% in England, 83.5% 230 \nin Scotland, 63.0% in Wales). Other than the number of babies admitted, study populations 231 \nwere largely comparable across the three nations with respect to other covariates (Table 1).  232 \nHis t orical trends   233 \nIn 2014, MgSO 4  uptake was around 20% in England, 40% in Scotland, and 10% in Wales. 234 \nUptake improved over time in all three nations, with the rate of change slowing down at 235 \nhigher levels of treatment (ceiling effect). Although national lev els appeared to converge in 236 \nthe latest 2022 data, there was visual suggestion that since the launch of the NPP, uptake 237 \nmay have been accelerated in England compared to the devolved nations. However, due to 238 \nrelatively smaller numbers, there was high variation in monthly uptake for Scotland and 239 \nWales, which  limited formal assessm ent of parallel trends (Figure 1). 240 \nPr e/ po s t - NP P c omparison  241 \nIn England, overall MgSO 4  uptake rose from 65.8% in 2017 to 85.5% in 2022 (62.3% to 81.4% 242 \nin Scotland, 61.6% to 86.6% in Wales). The amount of missing data fell from around 5% in 243 \n2017 to under 1% in 2022. Imminent delivery was the most commonly recorded reason for 244 \nnot giving MgSO 4 , accounting for around 15% of eligible babies in 2017, dropping to around 245 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n13  \n \n10% in England and Wales in 2022. The number recorded as not offered MgSO 4  f e l l  f r o m  246 \naround 7% to around 1% across all three nations (Supplementary Table 3).  247 \nEstimat e of N P P effectiv eness    248 \nThe adjusted model estimated an average 5.8 (95% CI 2.7 to 8.9, p<0.001) percentage point 249 \nincrease in MgSO 4  uptake in England across the four years post-NPP, compared to the one 250 \nyear pre-NPP. Much of the gains app eared to take place as a step-change in the first two 251 \nyears of the programme. There were additional gains in years three and four (at which point 252 \nthe improvement became statistically significant), although confidence intervals overlap 253 \nwith estimates from the first two years (Table 2).  254 \nEstimates were robust to sensitivity analyses (Table 2). There was indication of greater 255 \nimprovement in MgSO 4  use when including babies up to 34 weeks’ gestational age in the 256 \nanalysis (8.7 percentage point increase in MgSO 4  uptake, 95%CI 6.4 to 11.0, p<0.001), and in 257 \nthe 40 units in the PReCePT RCT (8.6 percentage point increase in MgSO 4  uptake, 95%CI 2.4 258 \nto 14.9, p=0.007). There was some evidence that lower-level maternity units improved more 259 \nthan higher-level units (SCBUs and LNUs: 9.1 percentage points change, 95%CI 3.9 to 14.3, 260 \np=0.001. NICUs: 4.1 percentage points change, 95%CI 0.6 to 7.6, p=0.022) , reflecting the fact 261 \nthat NICUs tended to have higher starting levels.  262 \nImpact o f the COVI D-19 pa nd emic  263 \nIn 2020 there was a slight declining trend in MgSO 4  u s e  c o i n c i d i n g  w i t h  t h e  s t a r t  o f  t h e  264 \nCOVID-19 pandemic, which continued to the end of the dataset at the end of 2022. The use 265 \nof antenatal steroids (another, more well-established protective treatment for preterm  266 \nbabies) had an almost identical decline over this same period (Supplementary Figure 1).  267 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n14  \n \nEconomic e valuation   268 \nThe impact of the NPP is illustrated in Figure 2, using a linear distribution (Figure 2a) and a 269 \nbeta distribution (Figure 2b) as counterfactuals. Probabilistic analysis estimated that the 270 \nadditional use of MgSO 4  attributed to the NPP was equivalent to 3.0 percentage point 271 \nimprovement on average over seven months, which equates to an additional 64 of the 2136 272 \npre-term (<30 weeks’ gestation) babies receiving treatment (Table 3). The lifetime and 273 \nsocietal NMB of the NPP was about £597000, or £279 per preterm baby. The probability of 274 \nthe NPP being cost-effective was 89% (Table 3, Supplementary Figure 2).  275 \nThe sensitivity analysis with the beta distribution counterfactual, estimated a longer period 276 \nof impact (an additional three months) over which there was additional use of MgSO 4  277 \nattributed to the NPP equivalent to a 2.9 percentage point improvement on average over 10 278 \nmonths, which equates to an additional 92 of 3129 pre-term babies receiving treatment 279 \n(Figure 2b). Accounting for the total cost of the NPP, and the lifetime health gains and cost 280 \nsavings of MgSO 4  treatment, the NMB was estimated to be £1.3m, or £400 per preterm 281 \nbaby. The probability of the NPP being cost-effective in this analysis was 100% (Table 3).  282 \nExpanding the analysis to include babies up to 32 weeks’ gestation estimated additional use 283 \nof MgSO 4  attributed to the NPP equivalent to a 4.4 percentage point improvement on 284 \naverage over nine months, which equates to an additional 215 pre-term babies treated. As 285 \nthe total cost of the NPP was fixed, and not sensitive to the number of babies treated, the 286 \nNMB of the NPP was about £4.2m, or  £853 per pre-term baby. The probability of NPP being 287 \ncost-effective was 100% at the willingness-to-pay threshold of £20,000 per QALY gained 288 \n(Supplementary Table 4 and Supplementary Figure 3). For the sensitivity analysis with the 289 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n15  \n \nbeta distribution the NMB of NPP was about £5.4m, or £700 per preterm baby. The 290 \nprobability of cost-effectiveness was 100% (Supplementary Table 4).  291 \nExpanding the analysis further again to include babies up to 34 weeks’ gestation shows that 292 \nthe additional use of MgSO 4  attributed to the NPP was equivalent to 7.3 percen tag e poin t  293 \nimprove ment  on average over seven months. This means an additional 961 babies receiving 294 \ntreatment. Using a beta distribution for the counterfactual showed a 6.5 percentage point 295 \nimprovement attributed to the NPP over 12 months (698 additional  patients). The NMB of 296 \nthe NPP was not calculated for this more mature group of preterm babies, as currently there 297 \nis no available estimate for the cost-effectiveness of MgSO 4  treatment for babies born 298 \nabove 31 +6  weeks’ gestation. 299 \nDI S CU S SI ON  300 \nThe original NPP evaluation found evidence of improved MgSO 4  use over the first 12 months 301 \nfollowing implementation. This extended evaluation found that the improvements have 302 \nlargely been sustained over the first four years following implementation, although there 303 \nwas suggestion of a slight decline in use coinciding with the pandemic. The benefits applied 304 \nboth to the target population of births <30 weeks’ gestation, but also to more mature 305 \npreterm babies up to 34 weeks’ gestation. The programme was associated with a net 306 \nmonetary benefit of about £0.6m for babies up to 29 weeks’ gestation, rising to about 307 \n£4.2m when babies up to 31 weeks’ gestation are included. Compared to the devolved 308 \nnations, uptake appeared to improve faster in England in the first two years following the 309 \nNPP launch. By the end of 2022 however, the three nations were broadly comparable with 310 \ndelivery of MgSO 4  to around 81-87% of eligible mothers. This is at the higher end of levels 311 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n16  \n \nreported internationally (69% to 87%(23-26)) following guidelines or interventions to 312 \nincrease MgSO 4 uptake.  313 \nComparisons between England, Scotland, and Wales should be interpreted with caution: 314 \nfirstly because there is high variability due to small numbers in the devolved nations data, 315 \nwhich limits formal statistical comparison of trends. Secondly because Scotland and Wales 316 \nwere implementing their own MgSO 4  initiatives (e.g. MCQIC(27), PERIPrem Cymru(28)), 317 \ncomplicating their position as a cont rol group. Thirdly because there was also accessing of 318 \nthe English PReCePT toolkit and implementation resources during this time period (66 319 \ndownloads from Wales, 32 from Scotland, 2018-2022, AHSN data, unpublished). This 320 \n‘contamination’ means that the boundaries of the target population are fuzzy, and again the 321 \ndevolved nations cannot be consider ed optimal controls. Finally, the three nations’ trends in 322 \nuptake prior to the NPP were also not parallel, due to variation in starting levels in 2014, 323 \nand this meant that formal statistical comparison of their improv ements (for example 324 \nincluding all three nations in an interrupted time series or difference in difference analysis) 325 \nwas not appropriate.  326 \nThe COVID-19 pandemic could plausibly have impacted on MgSO 4  use: the broad impact on 327 \nstaffing and quality of care affecting all parts of the NHS(29), and a specific impact on 328 \nexpecting mothers who may have presented at hospital later due to concerns about 329 \ninfection and giving birth alone, could both lead to missed opportunities to give MgSO 4 (30, 330 \n31). Analysis of future data will be important to explore what may be a temporary negative 331 \neffect of the pandemic, versus what might be a natural waning of the initially positive 332 \neffects of the NPP. From the observation that antenatal steroid use (historically well-333 \nestablished at high levels) declined almost identically to MgSO 4  u s e  o v e r  t h e  p a n d e m i c  334 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n17  \n \nperiod, the national decline may be associated with the pandemic. This finding is concerning 335 \nand may illustrate the fragility of complex healthcare systems. Further follow up is 336 \nimportant to monitor both metrics.  337 \nThe observation of greater improvements in uptake when including more mature preterm 338 \nbirths (all babies born up to 34 weeks’ gestation) is interesting. While the more mature 339 \npreterm infants represent a much larger proportion of preterm births, interpretation is 340 \ncomplex; their profile of underpinning antenatal disease, and hence their presentation to 341 \nhealthcare, may vary from the more extreme preterm presentation. This could be expected 342 \nto impact on the subsequent ease of delivering antenatal MgSO 4 . Evidence on the protective 343 \neffect of MgSO 4 in babies 30-34 weeks’ is less clear(3, 32, 33) and although long term 344 \nneurological impacts remain higher in these groups than term-born peers(34), due to their 345 \ngreater numbers, even small shifts in risks may have substantial population benefits(35). 346 \nIt is likely to be more difficult to improve from 85% to 90% uptake, compared to improving 347 \nfrom 65% to 70% uptake. Further overall increases in MgSO 4  use may be a challenge without 348 \nconcerted effort at the lower-performing units. However, as some units do report higher 349 \n(>90%) uptake (perhaps through better triaging and monitoring of symptoms), we argue 350 \nthat their performance should be used as the benchmark for quality of care, and investment 351 \nin supporting the lower-performing units is likely to be cost-effective. 352 \nThe creation of clinical guidelines alone is often not enough to ensure that evidence-based 353 \ninterventions become standard practice. A relevant example here is the case of antenatal 354 \nsteroids, which in the absence of a programme dedicated to getting this evidence into 355 \npractice, took several decades for their use to become standard care. In contrast, and in the 356 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n18  \n \nc o n t e x t  o f  t h e  N P P ,  t h e  s a m e  i m p r o v e m e n t s  i n  u s e  o f  Mg S O 4 were achieved within a few 357 \nyears. 358 \nStr engths a nd limitat ions  359 \nThis evaluation included effectiveness and cost-effectiveness analysis to capture the impact 360 \nof the NPP implementation. These types of analysis have been highlighted as key 361 \ncomponents of implementation science research(18, 36). Results from the main and 362 \nsensitivity analyses were consistent. Data covered a period of eight years, giving adequate 363 \ntime for analysis of trends. The study benefitted from high quality, routinely collected, 364 \nnational, longitudinal patient-lev el data. The key advantages of this comprehensive real-365 \nworld data is that it provides high generalisability (included all maternity units in England, 366 \nScotland, and Wales, reflecting the nationwide situation), shows effectiveness in real-world 367 \nconditions, and is less vulnerable to some biases such as recall, observer, and attrition bias. 368 \nA limitation is that this sort of data does not necessarily include all the covariates of interest, 369 \nand data quality and completeness is not always consistent across all sites. 370 \nA key limitation is that residual confounding cannot be excluded. We have tried to minimise 371 \nthe impact of confounding through robust analytic methods, and interpret findings with 372 \ncaution. In addition to the PReCePT programme, there are other factors that likely will have 373 \nimpacted on uptake, including the publication of definitive evidence on the protective  374 \neffective of MgSO 4 in 2009(37) MgSO 4 use being reliably recorded as a Neonatal Data 375 \nAnalysis Unit (NDAU) audit quality metric for maternity units in 2014-15, and its use 376 \nbecoming a formal recommendation in the NICE Guidance in 2015(2). Other factors related  377 \nto PReCePT include the original pilot study publishing positive results in 2017; and 378 \ndiscussions with unit leads about the proposed NPP in 2017. The National Neonatal Audit 379 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n19  \n \nProgramme (NNAP) concluded in their report on 2020 data that “This rapid improvement [in 380 \nMgSO 4  use], particularly seen in England, is likely to result from the targeted approach of 381 \nthe PReCePT quality improvement initiative.”(38)  382 \nAnother limitation is that the population investigated here included liveborn babies  383 \nadmitted to a neonatal unit, rather than the total population of mothers eligible for MgSO4. 384 \nThe rate of MgSO 4  uptake may be different between these two populations, although any 385 \nhealth benefit would only be realised in those investigated in this work. It would be an 386 \nadvantage if future research could explore, and compare, uptake in both populations, and 387 \nthe survival and neurodevelopmental outcomes for those preterm babies who did, and did 388 \nnot, receive MgSO 4 .  389 \nThe Health Foundation and Health Data Research UK recently listed PReCePT as a case study 390 \nmodel for a Learning Health System (i.e. a systematic approach to iterative, data-driven 391 \nquality improvement(39)) and we propose that the PReCePT model could be used as an 392 \nimplementation blueprint for other quality improvement initiatives.    393 \nCON C LU SION S  394 \nImplementation of the National PReCePT Programme has plau sibly helped accelerate 395 \nuptake of MgSO 4  in England, improving maternal and neonatal care, and positively 396 \nimpacting society in terms of direct patient benefit and future cost savings. Failure to deliver 397 \nMgSO 4  to eligible mothers should be considered inadequate care, and not financially 398 \nsustainable for the NHS. MgSO 4 as a quality metric should continue to be closely monitored, 399 \nand further intervention may be warranted to achieve optimal treatment lev els. Future 400 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n20  \n \nresearch should quantify the patient outcomes in this same population, specifically the 401 \ncases of cerebral palsy prevented, associated with the improvements in use of MgSO 4 .  402 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n21  \n \nAb breviation s  403 \nCP Cerebral Palsy \nMgSO 4  Magnesium Sulphate \nNDAU Neonatal Data Analysis Unit \nNMB Net Monetary Benefit \nNN RD  National Neonatal Research Database  \nNPP National PReCePT Programme \nPReCePT  Prevention of Cerebral Palsy in PreTerm Labour  \nQI  Quality Improvement  \n 404 \nREFERE NCE S  405 \n1. Or g aniza tion ) WWH. WHO re commen da tion s  on int erven tion s to im prov e pre ter m birth 406 \noutco me s . Swit zerl and; 2 015.  C on trac t N o.:  ISBN : 978 9 2 4 150 898 8. 407 \n2. (NI CE) NI fH a C E. Pr et erm labo u r  a nd bir th  (NG 25 ). 2015.  408 \n3. Crowth er C A, Mid dl eto n PF,  Voys ey M, A skie L, Dul e y L,  Pryde P G, e t  a l. As se s s i ng the 409 \nneurop rot ectiv e ben e f i ts fo r babi e s  o f an tena tal m agne sium sulpha t e :  An indiv idu al p art ic ipa nt da t a  410 \nmeta -an aly si s. PLOS Me dicin e . 2017;1 4 (1 0):e1002 398.  411 \n4. (NNAP ) N NA P . Annu al rep ort 201 8. UK: R CPH ; 2018 .  412 \n5. Odding E , Roebr oec k ME, Stam H J. The e pidemiol ogy o f  cer ebral p al sy: Inc idenc e,  413 \nimpa ir m ent s  a nd ri s k  fact or s. Di s a bility a nd Rehab ili tati on. 20 06;28 (4):183 - 9 1.  414 \n6. Bic kford CD ML, Mi tton C,  Kr u s e M , Synne s  A R, S awc huck  D, e t al.  M agne sium sul pha t e  for  415 \nfe tal n eu ropr o tec tion: a co s t - ef fec tivene s s anal y s i s . BM C H e a lth Se rvi ce s Re sea rc h. 2013;13(1 ):52 7.  416 \n7. Res oluti on N. NHS Re solu tion An nua l re p ort and a cco un ts 201 9/ 20 20 L ondon ; 2020  July 417 \n2020 .  Cont ract N o.: H C 49 9.  418 \n8. Kobezda T, Rehm A. Liv e birth s  w ith ce re b r a l pal sy a t  a t er t i ary mat ernity ho s pi t a l : inc idenc e 419 \nand ass oc iat ed ri sk fac t or s ove r  a  17 -ye a r  period . Jou rnal o f  obs t etric s an d gyna e colo gy : the  journal 420 \nof th e In sti tut e of Obs t etric s and  G y naec ology . 2022;4 2(7) :2771-8 .  421 \n9. O'Shea  TM, Allr ed E N , D a mmann O, Hi rtz D, Kuba n KCK, P ane th N, e t al.  The E LG A N  s tudy of 422 \nthe brai n  and rel a te d dis or d e rs in e xtrem ely low g es ta tion al ag e new born s. E arly Human 423 \nDeve lopm ent . 200 9;85(11 ):719 - 2 5 .  424 \n10. Vinc er M J ,  All en A C, Jo s eph K S, Stin son D A, Sc ott H,  W o od E. Inc re asi ng prev alenc e of 425 \nce r e bra l pal sy a mong v ery pret erm inf an ts : a po pula tion -ba s ed s tudy. P edia tric s. 200 6;118(6) :e16 21-426 \n6. 427 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n22  \n \n11. Himpens E, V a n de n Bro ec k C,  O o s tr a  A , Cald e rs P, Va nha e sebrouck  P. Prev al enc e , type,  428 \ndist r i buti on, and s ev e rity o f c er e b r a l pal s y i n relation to ge st a t io nal ag e: a me ta -a naly t i c revi ew. 429 \nDeve lopm ental Me dicine  &  C hild Neu rol ogy . 2008;5 0(5):334 -40.  430 \n12. Edw ards  H B , Re dani el  MT, Si lle ro -Rejon C, Marg elyte R, Pe ter s  TJ , Tilli ng K, e t  al. Natio nal  431 \nPRe Ce PT Progr amme : a be fo re -and -a fter  e valua t i on o f th e impl emen tati on of a  nationa l qu ali t y  432 \nimprove ment pr ogramme to increa s e th e u pt a ke o f m agne sium sulf at e i n pre term deliv er i e s . 433 \nArchiv e s  of D i s ea s e in Childh o od - F etal  a nd Neona t a l Editi on.  2023;10 8 (4) :342-7 .  434 \n13. Proct or E, Luke  D, Calhou n A, McMill e n C , Browns on R, Mc Crary S , e t al. Su sta ina bility  of 435 \nev idenc e -ba s e d h eal thc are : re se arch a ge nda , meth odolog ica l adva nce s , an d inf ra s t ructur e suppor t. 436 \nImple men t Sc i. 2015 ;10:88.  437 \n14. Edw ards  H B S R C, Pitha ra -Mc Keown C , De  Vocht F , McL eo d H, Redw ood S ,  H i ll L , O pm eer B, 438 \nOdd D, Lu yt K. P Re Ce PT Devol ved N ation s: Ev alua tion Rep ort  Univ er s ity  o f Bri s t ol;  2 024 5th F ebruary 439 \n2024 . 440 \n15. Burhou se A, L ea C , Ra y S , Baile y H, Davi e s  R,  H a r di ng H, et a l . P reven ting  ce r ebr a l palsy i n 441 \npret erm labo u r : a  multi organ i s a tion al qu a lity improv ement appro ac h to th e a dop tion and s pr ead o f 442 \nmag nesium sulpha te for n europ rot ectio n . BMJ O pe n Q ua l ity. 20 17;6 (2) .  443 \n16. GOV.UK. Nati ona l s t ati stic s: Eng li sh indic es  o f dep rivatio n 2019 . In: Mini s t ry  of Ho us in g CL G, 444 \nedito r. 20 19.  445 \n17. Edw ards  H B , Re dani el  MT, Si lle ro -Rejon C, Pi thar a- McKeow n C, M arge lyt e R, Sto ne T, et al . 446 \nQua lity improvemen t i nte r v enti on s t o  in crea se th e upt ake o f  magne sium sulp ha te  in pret erm 447 \ndeli verie s for t he pr even t i on of ce r ebral  palsy ( PRe Ce PT s tudy) : a clus te r r a ndomi sed c on t roll ed tri al. 448 \nBJOG: An I nte r na t i onal Journ al of O bste t r i cs &  Gyna ecol ogy. 2 023;n/a (n /a ).  449 \n18. Proct or E, Silme re H , Raghav an R, H ov ma nd P, Aa r o n s  G, B ung er A , et al .  O u tc ome s  fo r  450 \nimple menta tion r e sea rch: conc ep tual di stinc t i on s , m e a sur e m ent cha ll enge s, a nd re s e a rch ag enda . 451 \nAdm Pol icy  Ment He alt h. 201 1;38(2 ):65 - 76. 452 \n19. Shi h STF , Tonmuk aya kul U, Imms C , Redd ihough  D, Grah am HK, C ox  L, e t  al. Ec ono mic  453 \nev aluation and c o s t o f inte rve nti on s  f o r c erebral pa l s y : a sys tema tic r e view . Devel op mental 454 \nmedi cine and c hild ne u r ol ogy. 2 018;60(6 ):54 3-58 .  455 \n20. Cahi ll A G, Odibo AO,  Stou t MJ, Gr ob m a n WA, Maco n e s  GA , Ca ugh ey  AB. M agne si um sulf at e 456 \nther apy for  the  pre ventio n o f  ce rebr al p a lsy in p r e term i n f a n ts: a  deci s ion -a n alytic  and e conomi c 457 \nana lysi s. American J o ur n a l of Obs t etric s & G y neco l ogy. 2011;205 (6) :542.e1 -. e 7.  458 \n21. Si llero -Rejon C H W, Mc Leod H, Opme e r  B,  Luy t  K . Eco nomic evalu ation in i mplem enta tion  459 \nscien ce : A r ec oncil ed app roac h ev alu atin g  programmes to inc r e a s e  th e upt ake  of magn es i um 460 \nsulpha te i n p re -t e r m birt h s . Pr e -prin t ed.  Re sea r c hSqu a r e : 461 \nhttp s : / /www .r e s ea r c hsqu ar e.com /ar ticle / r s - 2 678140/v2 ;  2023.  462 \n22. NICE . Dev e loping  N I CE  guide lin es : th e manu al  (P MG20 ). L ondon : NI CE; 2 014.  463 \n23. Ow L L, Ke nnedy A, McCar thy E A ,  W a lke r S P. Fe a s ibi lity o f im plementi ng  magne s i u m s ulph a te 464 \nfor ne urop rot ection in a ter tiary o bs t et ri c uni t . Au s t ralia n a nd New Zeal an d Jou rn al of O bs t e t ric s  a nd 465 \nGynae colog y. 201 2;52 (4):356 -60 .  466 \n24. Keir AK, Sheph e r d E , McIn ty r e S, Rumbo l d A, Gr ov e s C, C rowthe r C, e t  al . Ant ena t al  467 \nmag nesium sul fat e to p reve nt  c erebral p a ls y . A rchiv e s  of di se a s e  in c hildhoo d Fet al a nd neona t a l 468 \neditio n . 2022;1 07(2 ):225 -7.  469 \n25. Choll a t C, B er tr a nd E, P eti t-L edo A, d e  V a nsay C ,  Voi s i n C, D abaj  I, e t al . Cer ebral P a ls y  in 470 \nVery P re term In fa n ts: A Nine -Ye a r  Pro s p ec tive S tudy in a Fr enc h Popul a tio n-B a se d Tertiary C ent er.  471 \nThe  Jour n a l of Ped ia tr i c s. 2021;2 37 :183- 9.e 6.  472 \n26. Doyle  LW, S pittle  AJ, O ls en JE , Kwong  A, Bolan d RA , Le e KJ, e t al . Tra ns lating an te natal 473 \nmag nesium sulpha te n e ur op rot ec t i on fo r infan t s  born <28 w e eks' g e s ta tion  in to practic e : A 474 \ngeo graphic al c oho rt  study. A u st N Z J Obstet Gyna ecol . 2021;6 1(4 ):513-8 .  475 \n27. Sc ott i s h P ati ent Sa fe ty  Progr amme (S PSP ) Mat erni ty a nd Chil dre n Q ua li ty  Imp r ov ement 476 \nColl abo rative (M C QI C) . In: S co tlan d HI , e ditor .: NH S  Sc o t l and ; 202 3.  477 \n28. PERI P r e m Cymru for p rof e ssiona l s. N HS Wal e s ; 2 023.  478 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n23  \n \n29. Jone s H, S ea bor ne M, Cow ley L , Odd D, P aranj ot hy  S, A k ba ri A, e t a l. Popul a tion bi rth 479 \noutco me s  i n 2020 and ex peri enc e s  o f  expec tan t mothe r s  dur i ng th e CO V ID -19 pa nde mic: A  'born in 480 \nWale s' mi xed me t ho d s  study u sing r ou ti ne  data . P LoS One. 2022;17 (5 ):e 0267176 . 481 \n30. Da sgupta T,  H o rga n G, Pe ter son  L, Mis try  HD ,  Ball s E, W il s on M , et al . Wome n’ s  ex perienc es 482 \nof ma terni ty care i n th e Unit ed King dom during t h e COVI D-19 p andemic : A follow -up system a tic 483 \nrevie w an d qualit ativ e ev id enc e s y nthe si s . W ome n and Bi rth . 20 24;37(3 ):101 588.  484 \n31. Aydi n E, Gla sgow  KA, Wei s s  S M, Khan Z, Aus t in  T, Joh ns on MH, e t al.  G i ving birt h i n a 485 \npand emic: women' s  bi rth ex pe rienc e s  in Engl and during  COVI D -19. BMC P r e gnanc y Chi ldbirth . 486 \n2022 ;22(1) :304.  487 \n32. Crowth er C A, A shwood  P, Middl e t o n PF, Mc Phe e A, T r a n T, Ha rding  JE, e t al . P ren atal 488 \nIntr aven ou s  Ma gne sium a t 30 - 3 4 W eek s’  Ge st a t i on and Neu rodev el opment al  O u t c omes in 489 \nOff spring: T he M AGE N T A  Randomiz ed Cl i nica l Trial.  JAMA . 2 023;330(7 ):60 3-14 .  490 \n33. Lo uis  J M , Randi s TM. Intr apar tum Mag ne s i um f or N e u r op rot ec t i on : Revi siting  Ge statio na l 491 \nAge Cr i teri a . J AM A . 2023;330 ( 7 ) :5 97 -8.  492 \n34. Odd D, Ev an s  D , Emond A. P ret erm Bir th,  Age  at Sc hool En try and L ong Te r m E duca t i onal 493 \nAchi eveme nt.  PLO S ONE. 2016 ; 11(5) :e01 551 57. 494 \n35. Odd D, Ev an s  D , Emond AM. Predi c tion o f sch oo l outcom e  af te r  pr e t e r m birth: a c oh or t 495 \nst u d y. Archiv e s o f dis ea se in c hildh ood . 2 019;10 4(4) :348-53 .  496 \n36. Hull L , Goulding  L, K hadjes ari Z,  D a vi s R, Heale y A, B a kolis I, e t a l. De signing  high- qua lity 497 \nimple menta tion r e sea rch: devel o pment , appl ication , f ea sibility and p reli minary e v alua t io n of t he 498 \nimple menta tion scie nc e re se arc h deve lo pment ( ImpRe s ) tool a n d guide . Impl em e ntati on Sc ienc e . 499 \n2019 ;14(1) :80.  500 \n37. Doyle  LW, C rowth er C A, Mi ddl eton P, M a r re t S,  Rou s e  D. M agne sium s ulpha t e  f o r  women a t 501 \nrisk o f pre term bi rth for neu ropr ot ection  o f  t he  fe tu s . The Coc hra ne da ta ba se  of sy s tem a t i c revi ew s. 502 \n2009 (1): Cd004661 .  503 \n38. (NNAP ) N NA P . Annu al rep ort 202 0. UK: R CPH ; 2020 .  504 \n39. Hardie T  H T , Thornt on-L ee  N, Ho me J, Pe reira P . Develo ping le arning he alth syste ms in t h e 505 \nUK: Prio ritie s for  ac tion.  2022 S e ptember  2022 . 506 \n  507 \n 508 \n 509 \n 510 \n 511 \n 512 \n 513 \n 514 \n 515 \n 516 \n 517 \n 518 \n 519 \n 520 \n 521 \n 522 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n24  \n \nTable 1: Baby, mother , and mater nity unit  c har ac t eristics  by  na t ion at ba seline 0  523 \n England Scotland Wales \nSocio-demographic characteristics of babies     \nNumber o f  ba bie s (N) 1  409 1 2 9 6 182  \nG e s t at i o na l a g e ( m e d i a n we ek s ,  I QR ) 2 7 . 7  \n( 2 6 . 0, 28 .9 )  \n27 . 9  \n(2 6 .6,28. 9 )  \n28 .0  \n( 2 6. 6, 2 8 .9 )  \n22 wee k s  14 ( 0 . 3 % ) 0 ( 0. 0 %)  3 ( 1.7%)  \n23 wee k s  22 5 (5.5 % ) 1 2  ( 4. 1 %)  6 ( 3.3%)  \n24 wee k s  37 4 (9.1 % ) 2 4  ( 8. 5 %)  9 ( 5.0%)  \n2 5  w e ek s  4 0 7 ( 10 .0 %)  1 9 (6 .4 % ) 1 2 (6 . 6 % ) \n26 wee k s  530  (1 3 . 0 % ) 34 ( 1 1 .5 %)  2 2  ( 1 2 . 1%)  \n27 wee k s  644  (1 5 . 7 % ) 58 ( 1 9 .6 %)  3 4  ( 1 8 . 7%)  \n28 wee k s  886  (2 1 . 7 % ) 74 ( 2 5 .0 %)  5 6  ( 3 0 . 8%)  \n29 wee k s  10 11 ( 2 4 . 7 % ) 74 ( 2 5 .0 %)  4 0  ( 2 2 . 0%)  \nB i r th w ei gh t ( m ed i an g r am s ,  I QR ) 9 61   \n(760, 1180)  \n101 0 \n( 78 0 , 1 21 8)  \n10 3 0  \n(7 8 2 , 124 0 )  \nMale sex  (N, %) 22 71 ( 5 5 . 5 % ) 1 5 6 (52. 7 %)  102  (56.0%)  \nM u l t i p l e  b ir t h s  ( N ,  % )  1 0 0 4  ( 2 4. 5% )  7 8  ( 26 . 4% )  40  ( 22 .0 % )  \nSocio-demographic characteristics of parents     \nNumber o f  mo th e rs (N) 2   357 3 2 5 4 162  \nMaterna l  ag e (ye ar s, me an,  S D)  3 0 .5 ( 6 . 1)  2 9 . 5  (5. 9 ) 28.7  ( 6.0) \nMo t hers r e p orti n g  w h ite B r itis h e thn icity (N ,  %)  \nNon -w hite  B ri tish \nMi ss i ng data  \n17 93 ( 5 0 . 2 % ) \n11 86 ( 3 3 . 2 % ) \n594  (1 6 . 6 % )  \n1 5 5 (61. 0 %)  \n3 4  (1 3. 4 % ) \n6 5  (2 5. 6 % ) \n116  (71.6%)  \n15  ( 9.3%)  \n31  ( 19 .1 % )  \nLe v el of  deprivati on (IMD quint il e ,  N,  %) 3      \n( Most d e p rive d ) 1 12 15 ( 3 4 . 0 % ) 88 ( 3 4 .7 %)  5 1  ( 3 1 . 5%)  \n2 804  (2 2 . 5 % ) 60 ( 2 3 .6 %)  3 0  ( 1 8 . 5%)  \n3 611  (1 7 . 1 % ) 33 ( 1 3 .0 %)  3 7  ( 2 2 . 8%)  \n4 487  (1 3 . 6 % ) 36 ( 1 4 .2 %)  2 0  ( 1 2 . 4%)  \n( L east d e p rive d ) 5 391  (1 0 . 9 % ) 28 ( 1 1 .0 %)  2 1  ( 1 3 . 0%)  \nMi ss i ng data  65 ( 1 . 8 % ) 9 ( 3. 5 %)  3 ( 1.9%)  \nAn y  re p orte d  smoki ng h ist ory (N,  %)  602  (1 6 . 9 % ) 38 ( 1 5 .0 %)  4 3  ( 2 6 . 5%)  \nClinical characteristics of mothers    \nHypert e n s i on i n pregna n cy  (N, %)  16 3 (4.6 % ) 1 6  ( 6. 3 %)  5 ( 3.1%)  \nPrema tu re  ru pture o f  me mbranes  (N, %)  603  (1 6 . 9 % ) 49 ( 1 9 .3 %)  30  ( 18 .5 % )  \nC a e s a re a n  s e c t i o n ( N , % )  1 8 8 8  ( 5 2. 8% )  15 2  ( 5 9 . 8% )  87  ( 53 .7 % )  \nAn te n a ta l  steroid s give n (N,  %)  32 68 ( 9 1 . 5 % ) 2 3 1 (90. 9 %)  149  (92.0%)  \nMaternity Unit characteristics     \nTotal  n um ber of  m ate rn ity u n its  \nWi t h  no  ne o nata l se r vice  \nWi t h  S pecial C a re Ba by Unit  / L o c al  Ne o na ta l  U ni t  \nW it h  Ne o nata l  I ntensive Care Uni t  \n15 0  \n3 \n10 6  \n41 \n18 \n3 \n6 \n9  \n 12  \n2 \n7 \n3  \nBi rt hs p e r l e v e l  of u nit  (N,  %)  \nWi t h  no  ne o nata l se r vice  \nWi t h  S pecial C a re Ba by Unit  / L o c al  Ne o na ta l  U ni t  \nW it h  Ne o nata l  I ntensive Care  Uni t  \n  \n6 ( 0. 2% )  \n13 29 ( 3 7 . 2 % ) \n22 38 ( 6 2 . 6 % )  \n \n6 ( 2. 4 %)  \n3 6  (1 4. 2 % ) \n2 1 2 (83. 5 %)  \n \n11  ( 6.8%)  \n49  ( 30 .3 % )  \n102  (63.0%)  \nAverag e  n um ber of  e l igi b l e  births  per h o sp ita l pe r \nmon th (me an, sd)  \n2 .8  (2 .1 )  2 . 3  (1 .3 ) 2 .1  (1 .6 ) \n0. Bas eli ne  p e rio d  is m eans/p ropo rtions a c r os s J a n-D e c 2 017, the y e ar b e fo r e th e  N P P wa s r oll ed-ou t in Eng l a n d .  524 \n1 .B a b ie s  u p  t o  3 0  we ek s ’  g es t a t i on a l ag e.  Des cr i p t ive  d a ta  i s  on  a l l b a bie s  i n clu di ng  mu lt i ple s .  A na ly s is  i s  res t r ic te d  t o  s ing l etons a nd  fir st-525 \nbo r n of multi p l e s .  526 \n2 . Un i q ue  m o t he r I Ds 527 \n3 . En g l ish , Sco t ti s h  a n d  Wels h  In d i ce s  o f Mu lt i ple  Dep r i vat i o n  are  c a l c u l a te d  di ffe ren tl y  an d  are  no t c o mp ar a b le be twee n  n a t io n s . 528 \n 529 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n25 \n \nTable 2: Change in Mg SO 4  uptak e in England fr om bef ore to  a f ter  t he NP P 0  530 \n Step-change in \nMgSO 4  uptake after \nintervention \n(percentage \npoints) 1  \n95 %  CI  p- v a l ue  Change in slope \npost- \nintervention \ncompared to pre-\nintervention \n95 %  CI  p - v a lu e  Overall slope \npost-intervention \n95 %  CI  p- v a l ue  \nUnadjusted 2  1 1. 8 1 0. 0  to  13 .6  < 0 . 0 01  -  -  -     \nFul l y  a d justed 3  5 . 8 2 . 7 to  8 . 9 <0.0 0 1 -0 .87  -1 .18  t o  - 0 . 57 < 0 .001  0 . 02 -0 .0 6 t o  0.1 0 0 . 6 2 2 \nCumulative increase per-year across the follow-up period 2     \nYear 1  \n \n3 . 1 -1 .3  t o  7.4 0 . 167  -0 .76  -1 .48  t o  - 0 . 05 0. 0 3 7 0 . 19 -0 .3 1 t o  0.6 9 0 . 4 5 0 \nYea r  1 - 2  \n \n3 . 5 -0 .2  t o  7.3 0 . 067  -0 .57  -0 .91  t o  - 0 . 23 < 0 .001  0 . 25 0 . 09 t o 0 . 4 1 0 . 0 0 2 \nYea r  1 - 3  \n \n5 . 6 2 . 1 to  9 . 1 0 . 002  -0 .86  -1 .25  t o  - 0 . 48 < 0 .001  0 . 05 -0 .0 5 t o  0.1 6 0 . 3 0 2 \nYe a r 1- 4  5 . 8 2 . 7 to  8 . 9 <0.0 0 1 -0 .87  -1 .17  t o  - 0 . 57 < 0 .001  0 . 02 -0 .0 6 t o  0.1 0 0 . 6 2 2 \nSecondary, sensitivity and sub-group analyses 2     \nAn a l y sis at  mat erni ty -un i t lev el 4  5 . 8 2 . 8 to  8 . 8 <0.0 0 1 -0 .83  -1 .47  t o  - 0 . 18 0. 0 1 2 0 . 003  -0 .0 5 t o  0.0 6 0 . 9 2 3 \nAn a l y sis at  in d iv id u a l-le v el 4  5 . 9 2 . 5 to  9 . 3 0 . 001  -0 .76  -1 .17  t o  - 0 . 34 < 0 .001  0 . 02 -0 .0 3 t o  0.0 7 0 . 4 4 9 \nEx clu di n g  im pleme ntati on start  \nw i nd o w  ( 2  m on t h s  e a c h s i de  of  s t a r t )  \n6 . 6 2 . 9 to  1 0 . 3  <0.0 0 1 -0 .86  -1 .17  t o  - 0 . 54 < 0 .001  -0 .0 1 -0 .0 9 t o  0.0 7 0 . 8 2 0 \nE x c l ud i ng  f i na l  2  m on t h s  ( ? da t a  \nqua lity) \n5 . 8 2 . 6 to  8 . 9 <0.0 0 1 -0 .82  -1 .14  t o  - 0 . 50 < 0 .001  0 . 04 -0 .0 4 t o  0.1 2 0 . 3 4 6 \nCo m p a ring 4  y ears  pre with  4 yea rs \npo st  \n6 . 0 3 . 7 to  8 . 4 <0.0 0 1 -0 .74  -0 .82  t o  - 0 . 65 < 0 .001  0 . 01 -0 .0 7 t o  0.0 8 0 . 8 9 5 \nPR e C e P T  c R CT  u nits o n ly (n = 40)  8 . 6 2 . 4 to  1 4 . 9  0 . 007  -0 .79   -1 .37  t o  - 0 . 20 < 0 .001  0 . 04 -0 .0 9 t o  0.1 6 0 . 5 6 3 \nI n c l ud i n g  b a b i es  up to 3 4 w ee ks  \ngestatio n \n8 . 7 6 . 4 to  1 1 . 0  <0.0 0 1 -0 .69  -0 .90  t o  - 0 . 48 < 0 .001  0 . 046  -0 .0 2 t o  0.1 1 0 . 1 5 7 \n0 A ll d at a on sin gl et o n s an d  first b o rn of m ulti pl es <30 w e e k s’ g e st ati o n an d ad mitt ed to an  NHS N e o na tal  un i t  531 \n1 P e r c ent a g e  poi nt diff er e n c e  i n  uptake betw e en m ean  a c ros s th e 12m pr e-N P P, a nd  mean ac ross the  four y ea r s  p ost-N P P 2  C ru d e  r eg r e s s i o n of  u pt a k e p o s t - i m pl e m e nt a t i on  c o m pa r e d w i t h  p r e - i m p l e m en t a t i o n .  532 \n3  F ul l y  a d j u s te d  m od e l:  in c l u de s  i n ter a c t io n  be twe e n  p re -p os t  per i od  a nd  s t ud y  mo n th  to  c a p tu re  c h a n ge  i n  s lop e a s  we ll a s  s te p -c h an ge . Adjust ed  for co v ari at es a s m o n thly ag g r egat es: m ean  ma t e rnal  a g e ,  533 \npropor tion  w h o i dentifi e d as of wh i t e B r iti sh et hn i city, mean I M D de ci l e ,  p ro p o r ti on  o f mu l ti pl e b irth s, prop o rti o n wi th pr egn a n cy hyp ert ensi on, r eport ed sm oker s, type of b i rth ( c-s e c t io n  ver sus v agi nal del i v e ry ),  534 \nb i r t h w e i g ht  a dj u s t e d f o r  g e st at i on a l  a g e  a s  a  z - s c o r e .  M o d e l  w ei g ht e d  o n un i t  s i z e .  535 \n4 A d diti o na lly adjust ed  for l ev e l  of bi rth u nit, and r e g ional c lust eri ng b y A HSN.   536 \n 537 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n26 \n \nTable 3. Pr obabilistic C os t -Effectiven e s s Re s u lts o f the N PP fr om Int erru pted Time Ser ies  A nalys is (<30 we eks ’  gestat ion)  538 \n Main analysis  \n(lin e ar dis tri buti o n \nc o u nt erfactu a l)  \nSensitivity analysis  \n(beta  d i s tri buti o n \nc o un te r f a c tu a l ) \nPe r i o d  o f  benefi t, m o nt h s  7  1 0 \nNu mber of pre-t erm b abies  (≤  3  w eeks),  N  213 6 31 2 9 \nCh a n ge i n pr o po rt i o n  o f  pre-ter m  babi es treate d \nwit h M g SO 4 , Δb i , %  3. 0% (1 . 5 %; 4.5% )  2.9%  (2. 3% ; 3. 6%)  \nNet I n c remen t of  pre- te rm  babies  t re at ed  w it h \nMgSO 4, Δpat  64  (32 ; 97 ) 92 (72; 11 2 ) \nNet c ost  of  im plement ati o n ,  ΔC i , £ 9 3 6 7 4 7 9 367 47  \nI m p l em en ta ti on  c os - e f fe ct iv en e s s ,  Δ Ci / ΔPat , £ \npe r  ad diti o na l  p at ie n t  trea ted  145 76 ( 29 28 4 ; 9 669 ) 10 2 19  (13 040 ;  838 6 ) \nL i f e t im e  he a lt h  e f fe c t  o f  Mg S O4 t r e a tmen t  p e r  \npatie nt , Δbt , Q ALY  0. 24 ( 0.1 6; 0 . 3 3 )  0.2 4 ( 0 . 1 6 ; 0. 33)  \nLifeti me c ost s  of M g SO 4  treatm ent p er p ati ent,  \nΔct , £ -1 9 0 6 4  (- 133 10;  -25 6 48)  -1 906 4 (- 1 331 0; -2 5 64 8)  \nNet Mon et a r y  B e n efit  of th e  P olic y ,  NMB P , £\n1\n 596 538  ( -2 21 748 ;  15 417 86 ) 12 5 15 11  ( 5581 15 ;  207 1 24 4 ) \nProba bil i t y of bei ng c o s t-e ff ec tive, %  8 9%  1 00%  \n1 A t a w ill i n gnes s -t o - pa y  t hres hold  of £ 20 , 00 0 p e r  Q AL Y gai ne d 539 \n 540 \n 541 \n 542 \n 543 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\n27 \n \n 544 \nFigure 1: Mg SO4 upt ak e in En gland, Sco tland and Wales, 2 01 4 t o 20 22  545 \nFigure 2: Pr edicted Mg S O 4 upt ake, C ounter factual and A rea-B etween-C u rves f rom  In terr upt ed Time Series an alys es  546 \n 547 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint \n\nFigure 1. MgSO 4 uptake in England, Scotland and Wales, 2014 to 2022 \n \n \n \n \n \nFigure 2. Predicted MgSO 4 uptake, Counterfactual and Area-Between-Curves from \nInterrupted Time Series analyses \n \n \n \n \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted July 16, 2024. ; https://doi.org/10.1101/2024.07.16.24310419doi: medRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}