Results

Parts of this chapter have been reproduced from the published papers by Cavallaro et al.1 and Cavallaro et al.2 These are open-access articles distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) licence, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given and an indication of whether changes were made. See: http://creativecommons.org/licenses/by/4.0/.

Chapter outline

This chapter begins with a description of the study cohort, providing a comparison of the characteristics of mothers who were and were not enrolled in the FNP. We then present the findings from Objective 1, which aimed to determine which groups of adolescent mothers receive FNP across LAs in England. For Objective 2, we describe child and maternal outcomes for the study cohort, providing an unadjusted comparison of outcomes for those who were or were not enrolled in the FNP (and according to maternal risk factors). We then present findings from the propensity score analysis (with sensitivity analyses) and from our multivariable logistic regression analysis. We show findings for subgroups according to maternal risk factors. Lastly, we present results from Objective 3, which aimed to evaluate the contextual factors associated with benefiting from the FNP.

Description of study cohort

Description of mothers in study cohort

Mothers who were enrolled in FNP were strikingly different from those who were never enrolled (Table 5). FNP mothers were younger, more likely to be admitted to hospital for adversity-related diagnoses or to attend A&E in the 2 years prior to 20 weeks of pregnancy and more likely to have their booking appointment after 20 weeks of pregnancy. FNP mothers were also more likely to have been in care or have a CPP, more likely to be recorded as having SEN provision, FSM and be in the most deprived quintile according to Income Deprivation Affecting Children Index (IDACI), more likely to have been excluded or be persistently absent and less likely to achieve 5 A*–Cs at General Certificate of Secondary Education (GCSE) level (but more likely to have achieved expected levels at Key Stage 1). Further information on the FNP cohort for mothers aged 13–19 based on data from the FNP IS is provided in Appendix 5, Table 33. Similar patterns were seen for mothers aged 20–24 (see Appendix 5, Tables 34 and 35).

Objective 1 – which groups of adolescent mothers receive Family Nurse Partnership across local authorities?

Parts of this section have been reproduced from the published paper on FNP enrolment by Cavallaro et al.2 This is an open-access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) licence, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given and an indication of whether changes were made is given. See: http://creativecommons.org/licenses/by/4.0/.

Key messages

• Only 23.2% (95% CI 23.0% to 23.5%) were enrolled in the FNP (25,680 of 110,520 eligible mothers).
• Enrolment rates varied substantially across 122 sites (range: 11–68%), and areas with greater numbers of first-time adolescent mothers achieved lower enrolment rates.
• Mothers aged 13–15 were most likely to be enrolled (52%; adjusted OR 2.65, 95% CI 2.39 to 2.94 compared with 18- to 19-year-olds) but accounted for only 2% of all eligible mothers.
• Only 26% of adolescent mothers with markers of vulnerability (including living in the most deprived areas and previous mental health-related hospital admissions) were enrolled.

Enrolment among mothers 13–19

We restricted the study cohort to eligible mothers aged 13–19 giving birth between April 2010 and 2017. Among all 110,520 eligible mothers, 25,680 (23.2%; 95% CI 23.0% to 23.5%) were enrolled in the FNP. This percentage ranged across 122 FNP sites, from 11% in Cumbria to 68% in Wandsworth (Figure 6). Appendix 2, Table 26, describes the 136 LAs, enrolment dates and FNP enrolment rates by FNP site.

FIGURE 6 Percentage enrolment in the FNP among eligible mothers aged 13–19, living in a LA with an active FNP site at the time of first antenatal appointment, by LA – England, births between 1 April 2010 and 31 March 2017. Note: the 122 FNP sites active during the study period covered 136 LAs (numbers and geographic boundaries of sites and LAs changed over the study period). Different sites were active for different periods within the 2010–7 cohort; the FNP was never commissioned in 15 LAs (see Figure 1). (Figure not reproduced here; see the original article.)

Overall, 12.5% (95% CI 12.4% to –12.7%) of all (25,680–204,690) first-time mothers aged 13–19 giving birth in England between March 2010 and April 2017 were enrolled in the FNP.

Risk factors for enrolment among mothers aged 13–19

Most eligible first-time mothers aged 13–19 in LAs with active FNP sites were white (85%), aged 18–19 at the time of birth (55%) and living in the most deprived quintile (49%) (Table 6). Five per cent of first-time adolescent mothers had ever been Looked After before 20 weeks of pregnancy, 32% had ever been persistently absent and 63% had attempted but not achieved 5 A*–C GCSEs. Characteristics of eligible mothers were similar between LAs that commissioned or had never commissioned FNP (see Appendix 5, Table 36). Overall, 66% of adolescent mothers in the eligible population had at least one vulnerability marker.

The percentage of eligible mothers enrolled in the FNP was highest (52%) among those aged 13–15 years than 20 years old at childbirth (12%), although 13- to 15-year-olds accounted for only 2% of eligible mothers. The percentage of eligible adolescent mothers enrolled increased slightly from 21% in the least deprived quintile to 24% in the most deprived. Forty per cent of adolescent mothers with a history of mental health- or adversity-related admissions were enrolled, as well as 44% of mothers ever Looked After. Overall, 26% of adolescent mothers with any vulnerability marker were enrolled.

Results from the adjusted model (see Table 6) showed that younger mothers were prioritised for enrolment [the OR decreased from 2.65 (95% CI 2.39 to 2.94) in 13- to 15-year-olds to 0.56 (95% CI 0.53 to 0.59) in mothers aged 20, compared to 18- to 19-year-olds]. Other risk factors included ever been a CLA (OR 1.92, 95% CI 1.81 to 2.04), ever had a CPP (OR 1.62, 95% CI 1.46 to 1.80) and ever identified as having SEN provision (OR 1.22, 95% CI 1.18 to 1.27).

Stratified analyses according to site enrolment level, English region and year of delivery among mothers 13–19

Low-enrolment FNP sites included 51% of all eligible mothers in their catchment areas but enrolled ≤ 21% of mothers in their catchment area, while high-enrolment sites included 9% of all eligible mothers and enrolled > 36% of mothers in their area (see Appendix 5, Table 37).

The effect of age and ethnicity on enrolment was more pronounced in low-enrolment sites (see Appendix 5, Table 38). The association between enrolment and age, ethnicity and deprivation varied across regions (see Appendix 5, Table 39). The age gradient appeared in all regions but was particularly pronounced in the South-West, East Midlands and South-East. In five of nine regions, mothers living in the most deprived areas were more likely to be enrolled than those in the middle quintile of deprivation. Conversely, London was the only region in which enrolment was higher in the least deprived areas (OR 1.68, 95% CI 1.08 to 2.63, compared with the middle quintile).

Risk factors for enrolment also varied over time (see Appendix 5, Table 40), in part due to changes in regional distribution of active sites.

Funnel plots of variation in enrolment rates

A substantial proportion of FNP sites’ enrolment rates fell outside the funnel plot limits, indicating that much of the variation in enrolment rates across sites for younger mothers (aged 13–17 at childbirth) was unexplained by chance (Figure 7). There was even more unexplained variation among mothers aged 18–20 at childbirth, as indicated by the majority of FNP sites falling outside the funnel plot limits. Among both age groups, adjusted enrolment rates were lower than expected in sites with larger numbers of eligible adolescent mothers (see Appendix 5, Figure 27).

FIGURE 7 Unadjusted funnel plots of variation in FNP enrolment rates among eligible first-time mothers aged 13–19 across FNP sites, by maternal age, births between April 2010 and March 2017. (Figure not reproduced here; see the original article.)

Objective 2 – what was the effect of Family Nurse Partnership enrolment on maternal and child outcomes?

Key messages

• We found no evidence of an association between FNP and indicators of child maltreatment, except for an increased rate of unplanned admissions for maltreatment/injury-related diagnoses up to age 2 for children born to FNP mothers.
• There was weak evidence that children born to FNP mothers were more likely to achieve a Good Level of Development (school readiness) at age 5 than those born to mothers who were not enrolled.
• Mothers enrolled in FNP were less likely to have a subsequent delivery within 18 months of the index birth than those not enrolled.

Descriptive analysis of outcomes of interest according to maternal risk factors

Birth, child and maternal outcomes varied substantially according to maternal risk factors (see Appendix 6). For example, children were more likely to be admitted to hospital for maltreatment or injury if they were born to younger mothers, those living in more deprived areas, those with a history of hospital admissions for adversity or mental health-related admissions, those with contact with social care services as a child and those with lower levels of education.

Descriptive (unadjusted) analysis of outcomes of interest according to enrolment in Family Nurse Partnership

Indicators of child maltreatment (unadjusted)

Over a fifth (22%) of children in our cohort were ever classified as a CiN by age 7, 4.5% had a CPP and 2.0% were Looked After at some point during this period. Children born to mothers in the FNP were more likely to have been discharged from hospital to social care services at birth and more likely to have ≥ 1 admission to hospital for maltreatment or injury-related diagnoses within 2 and 7 years of birth than those born to mothers who were not enrolled (Table 7). Children of mothers enrolled in the FNP were more likely to have a referral to social care services in the 7 years following birth than those of mothers who were not enrolled.

Child health, developmental and educational outcomes (unadjusted)

Low-birthweight babies and preterm births were more prevalent in mothers who were enrolled in FNP compared to those who were not (Table 8). Children born to mothers in the FNP were more likely to have ≥ 1 unplanned admission for any diagnosis and were more likely to attend A&E in the 2 and 7 years following birth. The mean number of admissions was similar across groups.

Children born to FNP mothers were less likely to achieve an expected level of development at school entry and less likely to achieve expected levels of development in Reading, Writing and Maths at Key Stage 1.

Maternal outcomes (unadjusted)

Mothers who were enrolled in the FNP were more likely than those who were not to have an unplanned admission to hospital for an adversity-related diagnosis or any diagnosis and more likely to attend A&E within 2 and 7 years of delivery (Table 9). Amongst mothers who gave birth before they started Year 11, mothers enrolled in the FNP were less likely than those who were not to achieve 5 A*–Cs and GCSEs.

Propensity score matching

The best-fitting model for the propensity score was a multilevel probit model, with mothers clustered in FNP sites. We included as predictors all available maternal risk factors up to 28 weeks gestation (at which point the vast majority of mothers had been enrolled) listed in Table 3. We also included interactions between maternal age and year of delivery and maternal age and gestational age at booking. We explicitly modelled the missing data categories for unknown ethnicity and mothers who did not link to NPD, as this provided a better balance between groups than using the missingness pattern information approach. There was a good overlap of propensity scores between FNP and non-FNP mothers, meaning that there were many mothers who were not enrolled in the FNP who had similar risk factors to those who were enrolled (see Appendix 7, Figure 28).

The most balanced groups were achieved by using one-to-one matching without replacement with a calliper width of 0.01. Using this approach, the mean difference in propensity scores between groups was < 0.001. The standardised differences of key maternal risk factors were small (effect sizes of 0.2, 0.5 and 0.8 are considered to be small, medium and large effect sizes, respectively; see Appendix 7, Table 48).95 However, since FNP mothers tended to have higher propensity scores (median = 0.39) than non-FNP mothers (median = 0.31), there were a small number of FNP mothers for whom we were unable to find a match (see Appendix 7, Figure 29). When matching the entire cohort of mothers aged 13–19 giving birth between 2010 and 2019, we were able to include 94.9% of mothers in the matched analysis. For the cohort with 2 years of follow-up (births between April 2010 and March 2017), we matched 95.7% of FNP mothers, and for the cohort with 7 years of follow-up (births between April 2010 and March 2011), we matched 99.9% of FNP mothers. The comparatively lower match rates for the later years were due to a smaller number of untreated mothers for these years. Using matching with replacement may have increased the proportion of matched mothers further but would have led to increased imbalance between groups.

Outcomes within matched cohort

Indicators of child maltreatment

There was an increased risk of unplanned admissions for maltreatment/injury-related diagnoses in the 2 years following birth amongst children of mothers who were enrolled in the FNP compared with those who were not (RR 1.15, 95% CI 1.07 to 1.24). There was weak evidence that FNP was associated with an increased risk of a hospital record indicating discharge to social services at birth (RR 1.23, 95% CI 1.00 to 1.51; p-value 0.046) and a decreased risk of a CPP up to 7 years after birth (RR 0.84, 95% CI 0.71 to 1.00; p-value 0.049). There were no other differences between groups in any of the indicators of child maltreatment that were evaluated (Table 10, Figure 8). Due to small numbers, we did not calculate RRs for the percentage of CiN referrals from health visitors.

FIGURE 8 Indicators of child maltreatment: adjusted RRs and 95% CIs comparing outcomes for mothers enrolled in the FNP vs. mothers who were not enrolled, for mothers aged 13–19 and giving birth in an area in which FNP was offered at the time of pregnancy. (Figure not reproduced here; see the original article.)

FIGURE 9 Child health, developmental and educational outcomes: adjusted RRs and 95% CIs comparing mothers enrolled in the FNP vs. mothers who were not enrolled, for mothers aged 13–19 and giving birth in an area in which FNP was offered at the time of pregnancy. KS1, Key Stage 1. (Figure not reproduced here; see the original article.)

FIGURE 10 Maternal outcomes: adjusted RRs and 95% CIs comparing outcomes for mothers enrolled in the FNP vs. mothers who were not enrolled, for mothers aged 13–19 and giving birth in an area in which FNP was offered at the time of pregnancy. a, Among mothers who were aged < 16 at the start of the academic year in which they reached 20 weeks gestation. b, Among mothers who were aged < 15 at the start of the academic year in which they reached 20 weeks gestation. (Figure not reproduced here; see the original article.)

Child health, developmental and educational outcomes

There was an increased risk of low birthweight amongst mothers who were enrolled in the FNP versus those who were not (RR 1.07, 95% CI 1.02 to 1.13). There was also an increased risk of unplanned admission for any diagnoses (in the 2 years following birth) and for A&E attendances (in the 2 and 7 years following birth).

There was weak evidence that children born to FNP mothers were more likely to achieve a Good Level of Development (school readiness) at age 5 (RR 1.05, 95% CI 1.00 to 1.09) than those born to mothers who were not enrolled. Children in the FNP arm were also more likely to be recorded as having FSM (see Table 11, Figure 9).

Maternal outcomes

Mothers who enrolled in the FNP were more likely to have unplanned admissions for adversity-related diagnoses, mental health conditions or any diagnoses and A&E attendances in the 2 and 7 years following birth, compared to those who were not enrolled (see Table 12, Figure 10). However, these mothers were less likely to have a repeat pregnancy within 18 months of the index birth (RR 0.92, 95% CI 0.88 to 0.97).

Subgroup analyses

Although there was variation in the association between FNP and outcomes across subgroups, there were no statistically significant interactions between enrolment in FNP and maternal age, ethnicity, area-level deprivation, maternal contact with social care services, child sex or year of delivery (see Figures 11–14, Appendix 7, Appendix 8, Figures 31–34). However, we did observe a significant interaction between region and subsequent births within 18 months (p = 0.0018). The reduction in subsequent births associated with FNP was seen most strongly in the South-West, South-East and London (see Appendix 8, Figure 33).

FIGURE 11 Child unplanned admissions for maltreatment/injury in the 2 years following birth: subgroup analysis presenting adjusted RRs and 95% CIs comparing mothers enrolled in the FNP vs. mothers who were not enrolled, for mothers aged 13–19 and giving birth in an area in which FNP was offered at the time of pregnancy. (Figure not reproduced here; see the original article.)

FIGURE 12 Good level of development: subgroup analysis presenting adjusted RRs and 95% CIs comparing mothers enrolled in the FNP vs. mothers who were not enrolled, for mothers aged 13–19 and giving birth in an area in which FNP was offered at the time of pregnancy. (Figure not reproduced here; see the original article.)

FIGURE 13 Subsequent delivery within 18 months: subgroup analysis presenting adjusted RRs and 95% CIs comparing mothers enrolled in the FNP vs. mothers who were not enrolled, for mothers aged 13–19 and giving birth in an area in which FNP was offered at the time of pregnancy. (Figure not reproduced here; see the original article.)

FIGURE 14 Maternal unplanned admissions for any diagnosis in the 2 years following delivery: subgroup analysis presenting adjusted RRs and 95% CIs comparing mothers enrolled in the FNP vs. mothers who were not enrolled, for mothers aged 13–19 and giving birth in an area in which FNP was offered at the time of pregnancy. (Figure not reproduced here; see the original article.)

Sensitivity analyses

We performed two sensitivity analyses where the approach to propensity score matching was changed to allow matching (1) within the same LA but in different time periods (comparing FNP mothers with similar mothers before FNP was offered in that LA) and (2) within the same time period but in different LAs (comparing FNP to similar mothers living in different LAs). It was more difficult to find closely matched groups within these sensitivity analyses since the characteristics of mothers differed over time and between LAs that did or did not offer the FNP. We therefore also present results from a matching with replacement strategy, aiming to retain a higher coverage of FNP mothers (see Appendix 2, Table 31). Results from these sensitivity analyses did not differ substantially from those of the main analysis.

Multivariable regression (adjusted) analyses

Effect estimates from the sensitivity analyses using multivariable regression (see Appendix 9, Tables 55–57) were broadly similar to those from the main propensity score analysis, with some differences between groups being attenuated.

Objective 3 – which programme and contextual factors influence the effect of Family Nurse Partnership?

Key messages

• Fifty-eight per cent of mothers enrolled in FNP completed the programme, and 42% left early.
• Mothers enrolled in FNP received an average of 38 visits and 42 hours of contact time with a family nurse.
• Fifty-nine per cent of mothers in pregnancy, 65% in infancy and 61% in toddlerhood met targets for the expected number of visits.
• Younger mothers and those with a history of unplanned hospital admissions for mental health, adversity or chronic health conditions, received a greater number of visits; other contextual factors, including nurse characteristics and area-level characteristics, had little effect on the number of visits received.
• Meeting the target for the number of expected visits in pregnancy was associated with a reduction in subsequent births within 18 months; meeting the target in infancy and toddlerhood was associated with an increase in the number of children with unplanned hospital admissions for maltreatment/injury up to age 2.

Description of attrition, fidelity targets and dosage in the Family Nurse Partnership

Of the 31,260 mothers aged 13–19 at LMP enrolled in FNP and giving birth between 2010 and 2019, with linkage to HES (see Figure 2), 28,155 gave birth before January 2018 and were included in this analysis. Of these, 28,120 had at least one visit recorded, 16,305 (58%) completed the programme and 42% left early (slightly higher than the target attrition rate of 40%). A small number of clients had some interrupted periods of enrolment where they left the programme and returned later on (see Appendix 4).

On average, mothers received a total of 38 visits during their enrolment in the FNP (Table 13), corresponding to an average 42 hours of contact time with a family nurse.

At each stage of the programme, there is a target number of visits that mothers should receive, and fidelity to these targets is measured (see Table 4). Mothers are expected to receive at least 80% of the maximum number of visits during pregnancy, 65% during infancy and 60% during toddlerhood. FNP also has targets for attrition: no more than 10% during pregnancy, 20% during infancy and 10% during toddlerhood.

Pregnancy stage

The majority of FNP mothers (25,925; 92%) remained in the programme for the full pregnancy (see Table 13). Eight per cent of mothers dropped out before delivery, and a small number left during pregnancy but returned later (see Appendix 4). For mothers giving birth before January 2018, the FNP fidelity target of < 10% attrition during pregnancy was met.

The maximum possible number of visits during pregnancy is 14 (for those enrolled at 16 weeks of pregnancy), and the average number of visits in our cohort was 10. A total of 16,695 (59%) mothers met the target of 80% of visits completed during pregnancy. The distribution of visits is summarised in Appendix 9.

Infancy stage

Of the 25,975 mothers who were still in the programme at the start of the infancy stage, 19,860 (77%) completed the infancy period in full; 70 mothers were in the infancy period for < 2 weeks, and a small number restarted the programme during the infancy period (i.e. left during pregnancy and returned during infancy). Overall, attrition during infancy was 23% (5855), slightly exceeding the 20% target.

For the infancy period, the maximum possible number of visits is twice that of pregnancy (28). In our cohort, the average number of visits in this stage received remained similar to those received in pregnancy (10 visits). A total of 16,945 (65%) met the target of 65% of visits completed for this stage. The distribution of visits is summarised in Appendix 9.

Toddlerhood stage

Of the 20,015 participants remaining in the programme at the toddlerhood stage, 16,175 (81%) completed the toddlerhood stage in full; 25 mothers were in the toddlerhood stage for < 2 weeks, and a small number of mothers left and returned during the toddlerhood stage. Overall attrition was 19% (3760), higher than the 10% target.

For the toddlerhood stage, the maximum possible number of visits is 22. The average number in our cohort was 14 visits, and 12,115 mothers (61%) met the 60% fidelity target (see Table 13). The distribution of visits is summarised in Appendix 9.

Participant, programme and nurse characteristics associated with dosage in the Family Nurse Partnership

There was substantial variation in whether sites met delivery targets for each stage of the programme, illustrated by the lack of overlap in CIs in Figure 15.

FIGURE 15 Percentage of mothers meeting fidelity targets for the number of expected visits in pregnancy, infancy and toddlerhood, according to FNP site, for births between April 2010 and January 2018 to mothers aged 13–19 at LMP. The solid line indicates the fidelity target. (Figure not reproduced here; see the original article.)

Younger mothers and those with a history of unplanned hospital admission for mental health, adversity or chronic health conditions received a greater number of visits over their time in the programme (see Appendix 10, Table 58). There was little variation in the average number of visits or proportion of mothers meeting fidelity targets according to nurse characteristics (see Appendix 10, Table 59). However, we did see some patterns according to the number of visits with a partner present: mothers whose partner was present for 10–50% of visits tended to receive more visits than those whose partner was never present or those whose partner was present for > 50% of visits. A similar pattern was seen with the presence of parents.

When adjusting for all these maternal factors (Table 14), the greatest drivers of meeting fidelity targets in the pregnancy stage were maternal age (RR 1.13, 95% CI 1.08 to 1.19 comparing mothers aged 13–15 with those aged 18–19), having a history of adversity (RR 1.06, 95% CI 1.02 to 1.09 compared to mothers with no history of adversity), being a CiN at enrolment (RR 1.18, 95% CI 1.14 to 1.21 compared to mothers not CiN at enrolment) and having a CPP at enrolment (RR 1.20, 95% CI 1.14 to 1.26). Mothers who had ever been recorded with persistent absence or who had been excluded were less likely to meet targets, and those who had achieved 5 A*–C GCSEs prior to enrolment were more likely to meet targets (see Table 14).

The greatest driver of meeting fidelity targets during the infancy and toddlerhood stages was having met the targets in the previous stages of the programme (Tables 15 and 16). The area-level factors we explored (% of mothers in the area who were adolescents, ethnic distribution, % of preterm births, % of mothers smoking at delivery and rates of referrals to children’s social care or CPPs) did not have a significant association with meeting fidelity targets, except for LA-level deprivation, where mothers in the most deprived LAs were more likely to have the expected number of visits.

Effect of programme and contextual factors and dosage on outcomes

Within the group of mothers enrolled in FNP, none of the additional participant, programme or nurse characteristics (see Appendix 10, Table 59) were associated with child unplanned admissions for maltreatment or injury up to age 2, a Good Level of Development at age 5 (school readiness), maternal unplanned admissions for any diagnosis in the 2 years following birth or subsequent births within 18 months.

However, we observed that meeting the target for the number of expected visits in pregnancy was associated with a reduction in subsequent births within 18 months (RR 0.87, 95% CI 0.78 to 0.97 compared with mothers who did not meet the target in pregnancy, Table 17). Meeting the target in infancy and toddlerhood was associated with an increase in the number of children with unplanned hospital admissions for maltreatment/injury up to age 2 (Table 17).