Article Text

Download PDFPDF

Targeting health visitor care: lessons from Starting Well
  1. C M Wright1,
  2. S K Jeffrey1,
  3. M K Ross2,
  4. L Wallis2,
  5. R Wood3
  1. 1
    PEACH Unit, Faculty of Medicine, University of Glasgow, Glasgow, UK
  2. 2
    Starting Well Health Demonstration Project, Greater Glasgow NHS Board, Glasgow, UK
  3. 3
    Public Health Sciences, University of Edinburgh, Edinburgh, UK
  1. Professor Charlotte M Wright, Community Child Health, PEACH Unit, QMH Tower, Yorkhill Hospitals, Glasgow G3 8SJ, UK; charlotte.wright{at}


Background: UK child health promotion guidelines expect health visitors to assess family needs before new babies are aged 4 months and offer targeted care on that basis thereafter. Data from an intensive family support programme were used to assess how accurately family needs can be predicted at this stage.

Design: A population based cohort of 1202 families with new babies receiving an intensive health visiting programme. Analysis of routinely recorded data.

Setting: Starting Well project, Glasgow, UK.

Predictors: Health visitor rating of family needs.

Main outcome measures: Families receiving high visiting rates or referred to social work services.

Results: Of 302 families rated high need, only 143 (47%) were identified by age 4 months. Visiting rates in the first year for those initially rated high need were nearly double those for the remainder, but around two thirds of those with high contact rates/referred to social work were not initially rated high need. Six family characteristics (no income, baby born preterm, multiple pregnancy, South Asian, prior social work/criminal justice involvement, either parent in care as a child) were identified as the commonest/strongest predictors of contact rates; 1003 (83%) families had one such characteristics and/or lived in a highly deprived area, including 228 (93%) of those with high contact rates and 157 (96%) of those referred to social work.

Conclusions: Most families at risk will not be identified on an individual basis in the early weeks. Most families in deprived areas need continued input if the most vulnerable families are to be reliably identified.

Statistics from

Over recent years policy has increasingly focused on a positive start in life as the key to health improvement and reduction of inequalities.1 2 The key NHS mechanism delivering early years support is the child health promotion programme, which is delivered predominantly by the health visiting service. The programme offers universal health promotion and parenting advice, screening, and immunisation and growth surveillance, as well as support for vulnerable families.

What is already known on this topic

  • Current child health promotion policy recommends a targeted service model, in which the majority of families receive a core programme of support.

  • Families with needs identified in the first 2–4 months of a child’s life receive more intensive tailored input.

  • Little information is available on the ability of health visitors to accurately predict family needs in the early weeks of a child’s life.

What this study adds

  • During an intensive family support programme, health visitors targeted more care to those families identified as having high needs.

  • Those ultimately requiring the most input were generally not identified as high risk in the first 4 months of their child’s life.

  • A majority of families in deprived areas will need continued input if the most vulnerable families are to be reliably identified.

The fourth Health for All Children report (HFAC4)3 recommended moving from relatively uniform child health promotion provision (with all families receiving a similar service) to a more explicitly targeted model (with all families receiving a reduced core programme of support, and families with specific needs receiving additional targeted input as required). Scotland has recently implemented HFAC4, and health visitors in Scotland are being asked to allocate all children to one of three levels of ongoing support when they reach 6–8 weeks of age.4 Children allocated to receive the reduced core programme receive no further routine health visitor contacts except at immunisation visits. Children allocated to additional support receive more frequent contacts as required, and children allocated to intensive support receive coordinated multi-agency input. In England, current guidance states that health visitors may take until a child’s first birthday to agree the ongoing level of input required, but similarly recommends no routine contacts in the first year after the 4-month immunisation.5

This approach thus assumes that health visitors can accurately predict in the early weeks which families require additional or intensive support. There is little support for this assumption in the published literature,68 and concern has been expressed about whether making assessments at such an early stage will be robust.9

In this paper we used data from the Starting Well programme to explore health visitors’ ability to assess family needs in early infancy and how this related to markers of vulnerability and subsequent service use and family outcomes. Starting Well was one of four major health demonstration projects designed to foster best practice in health improvement and reduction of health inequalities across Scotland.10 Starting Well operated from 2001 to 2005 (ie, before the implementation of HFAC4 in Scotland) in three urban areas of Glasgow with high socio-economic deprivation and a combined population of almost 65 000. Detailed descriptions of the project and its impacts are available elsewhere.11 12 All families with new babies born during the project were offered an intensive home visiting service starting in the antenatal period and extending up to the child’s third birthday instead of the usual child health promotion programme. Home visits were coordinated and mostly carried out by health visitors. A visiting schedule set out the recommended frequency of visits for families with “average” needs, but health visitors were free to agree more or less intensive schedules with individual families.


The subjects for this analysis were all families living in the Starting Well project area with a single birth (of one child, twins or more) born at least 1 year before the project end (late 2000 to 2004) and actually followed for at least a year. Families with more than one birth were excluded, as it was not possible to distinguish to which child in the family any one contact applied.

The Starting Well project database held all information collected on families and babies and their contacts with the service, with forms completed by field staff entered into an ACCESS database by administrative staff. The main sources of data for this analysis were information collected about the family around the time of the birth and in the early weeks and entered into the family health plan, Edinburgh Postnatal Depression Scale scores recorded by health visitors in the first 6–12 weeks, and forms completed after every family contact. Family postcodes were matched to Scottish Index of Multiple Deprivation (SIMD) decile13 and then stripped out before being passed to the PEACH Unit for analysis. There the data were cleaned and the completeness and prevalence of potentially useful variables explored. In some cases similar variables were combined to generate new variables to maximise their completeness and discrimination (eg, using entered maternal age and maternal date of birth to arrive at a more complete variable for maternal age).

By the time the baby was 8 weeks old, health visitors were asked to document families’ circumstances and health needs in a family health plan, and allocate a family needs score (FNS). The FNS reflected the health visitor’s subjective global assessment of a family’s level of need and requirement for ongoing visiting and support. An FNS of 2 indicated predicted need for visiting in line with the recommended visiting schedule, while a score of 1 or 3 predicted a need for lower or higher visiting, respectively. The precise nature of the additional support offered to families allocated an FNS of 3 was determined by the health visitors’ professional judgement rather than rigid protocols, reflecting the complex and variable nature of different families’ needs. Health visitors were expected to review and update the FNS as required.


The first FNS allocated to a family was identified along with whether a family was ever rated high need (FNS 3). The total number of contacts in the first year was identified for each child. These showed a skew distribution so that log contact rates were used as an outcome. Families with high contact rates in the first year were defined as those in top 20th percentile, a proportion chosen to supply reasonable numbers in the extreme group. Likely or relevant predictors of a high need rating were identified by univariate analysis and significant predictors added to a logistic regression model, with non-significant variables successively removed until all remaining were significant at p<0.05. Predictors of (log) first year contact rates were identified using linear regression in the same manner.

The commonest or strongest predictors of contact rates were identified and used to produce a post hoc risk rating, with families positive for any one of the predictors/risk markers deemed to be high risk. Ethical approval was not required, but Caldicott guardian approval was obtained for this use of the data.


There were 1266 Starting Well families with only one birth occurring at least 12 months before the end of the project (fig 1). Of these, 64 families (all from one centre and with consecutive study numbers) had a large gap in their contact record in the first year, presumptively the result of failed data input. Because their contact counts were thus known to be inaccurate they were excluded altogether, leaving 1202 families for this analysis.

Figure 1 How study families were selected for this analysis.

Of the 1202 families, 38% had no paid income, 17% were of South Asian origin, and in 5% a parent had been in care as a child (table 1). Most family data were complete for >90% of families, with the exception of SIMD decile derived from the postcode (80% complete) and the Edinburgh Postnatal Depression Scale scores (59% complete) (table 1). Families with these variables missing showed no difference in the proportion who were high risk, but rates of missing data did vary significantly by project area.

Table 1 Prevalence of demographic characteristics and their association with initial assessment as being high need

An FNS was allocated to 1039 (86%) families during their baby’s first year: the proportion of families with no FNS again varied significantly by area, with only a weak non-significant trend for those with scores to be high risk. Only 686 families (57% of the total, 66% of those ever rated) had their first need rating recorded by the time their baby was 2 months of age (the age recommended in the Starting Well protocol) and only 761 (63% of the total, 73% of those ever rated) had been rated by the age of 4 months. Of the 1039 families who received an FNS, 133 (13%) were initially rated high need (FNS of 3) and 302 (29%) were rated high need at some point. These ratings of high need were recorded for only 99/302 (33%) by age 2 months and for 143/302 (47%) by 4 months, while 103/302 (34%) were made after the age of 9 months.

A wide range of predictors related to being rated high need, but the significant independent predictors in this analysis mainly related to parental socio-emotional vulnerability (table 1). Families identified as high need later had very similar patterns of risk markers to those identified earlier.

The families had a median of 14 (interquartile range (IQR) 9–20) visits in the first year. Visiting rates in the first year among those initially rated high need were nearly double those for the remainder (high need geometric mean (95% CI) 24 (22 to 27), average need 13 (13 to 14), low need 11 (10 to 13)). Fifty three per cent of families initially rated high need had high contact rates and 38% were referred out at some time to social work services (table 2). However, only 32% of all families with high contact rates and 35% of all families referred to social work services were initially rated high need and only 61% and 72%, respectively, were ever rated high need (table 2).

Table 2 Relationship between family needs score or presence of risk markers, and subsequent high contact rates or referral to social work services

Six family characteristics proved the commonest or strongest predictors of contact rates (no income, baby born preterm or multiple pregnancy, South Asian, prior social work/criminal justice involvement, either parent in care as a child). If families had been defined as high need on the basis of having one or more of these risk markers, 60% children would have been deemed high need, but this would still have identified only 73% of those with high contact rates and 82% of those referred to social work services. If the additional ecological risk factor of living in a post code area within the most deprived SIMD decile was added, this would have identified 93% of those with high contact rates and 96% of those referred to social work services, but 83% of the cohort would have been defined as high need (table 2).


This analysis has revealed that during an intensive family support programme in an inner city area, health visitors varied their contact patterns substantially in relation to varying need, but their early prediction of need identified only about a third of those who would ultimately prove most vulnerable. Moreover, although a majority of families had been allocated an FNS by age 4 months, only half of those families ultimately judged high need (FNS 3) had been so rated by the time their child was 4 months old, suggesting either that it took longer for health visitors to arrive at this judgement or that families’ circumstances changed. While individual risk markers were significantly related to vulnerability, for many families the only risk marker was residence in an area of extreme poverty.

Starting Well was a large project with dedicated resources for project database development and the project data used in this analysis were largely complete for most variables used. We looked at all variables with significant degrees of missing data to test whether missingness related to individual factors (using the post hoc risk rating) or area factors, and in all cases it was area factors that proved significant, but the lower performing area varied by variable. This led us to suspect that there were form filling and data entry quirks in each area that led to missing data, rather than poor compliance with clinical protocols by field staff. All data were collected by field staff in a service setting, so they represent the real assessments that would be used to form the basis of the service given to families. Accuracy could not be directly assessed, but assessments correlated in the expected manner with both the FNS and visiting rates.

The Starting Well project was deliberately located in a deprived, multi-cultural, urban area. This means that, while these results cannot be simply generalised to the whole population, they are highly informative about vulnerable inner city populations. For this analysis, actual visiting rates provided to families by health visitors, and referral of families to social work services, were taken as the key outcome indicators against which health visitors’ assessments of family needs were validated. These indicators are not ideal, as assessment of need was intended to influence visiting rates, and more frequent visiting increased the likelihood that problems requiring referral to social work services would be uncovered. However, this would tend to make health visitors’ judgements appear more accurate than they actually were, suggesting that health visitor prediction may be even less accurate than these results suggest.

There are important similarities and differences between the service models used in Starting Well and current child health promotion policy. In Starting Well even families assessed as low need still received many contacts, with multiple opportunities for their assessed level of need to be revised, and health visitors ultimately identified two thirds of families requiring frequent visiting or social work services referral as being high need, with this revised rating commonly made late in a child’s first year. In contrast to the Starting Well model, current child health promotion guidelines do not recommend any routine health visitor contacts between the 4- and 12-month immunisations, and this may well mean that health visitors are more cautious when making their initial assessments. A range of factors may influence the threshold for a family being assessed as high need, including the professional’s level of experience and risk aversion, the level of need in the population served, and the resources available.14 One-off family needs assessments are based on the assumption that needs and vulnerability are fixed traits: in practice many families’ needs change considerably over time in response to internal and external stressors and supports.15 However, when there are no other opportunities to routinely see families, there will be no opportunity to revise this assessment unless families themselves seek help.

These results could be used to suggest that assessment of family needs based on a simple checklist of risk markers is potentially more sensitive than assessment based on health visitors’ professional judgement. However, checklists are potentially damaging to health visitor–family relationships, particularly when professionals are trying to engage the most disenfranchised families,1618 although some authors remain cautiously positive about the flexible use of structured family needs assessment tools.19 Another interpretation would be that that in very deprived areas child health promotion resources should be targeted at the whole population, rather than solely on the basis of individual assessment, an approach advocated in HFAC4. This avoids stigmatisation and ensures that families who are unable to actively seek help still receive the service. In this setting only families with no indicators of vulnerability might occasionally be left for core input. However, this approach is potentially unpopular as it requires the active re-allocation of health visitors, who currently tend to be distributed according to historical accident rather than population need,20 although ways of ensuring their more equitable distribution have been sought.2124

There is growing interest in targeted service provision across the UK and elsewhere.25 26 In general, universal services are simple to administer, but they may be preferentially accessed by low risk individuals with the least need.27 Targeted services have thus been advocated as being more efficient and potentially more effective in reducing inequalities. Pilot schemes based on the US Nurse-Family Partnership programme are currently underway in England which aim to target high intensity supportive services to families with children at risk of social exclusion identified in the antenatal period.25 28 Models of targeted service which are offered to families who would otherwise be unserved may be efficient, but those missed are no worse off than before the service was introduced. However, targeted services have a very different implication when they replace a universal service and this analysis suggests that in practice a third of the most vulnerable families would be missed if targeting relied only on individual assessments carried out repeatedly over the course of a child’s first year, with two thirds missed if the assessment process stopped when the child was 4 months of age.

This paper is not intended as a criticism of either health visitors or Health for All Children. Health visitors have an excellent track record as autonomous professionals carrying heavy and complex caseloads and the proposal by HFAC4 that they should be encouraged to use their professional judgment to offer tailored, flexible care is admirable. What does need to be challenged is the interpretation by others that HFAC4 dictates that most families need little or no service from health visitors after the age of 4 months. These data would suggest that it is not reasonable to expect health visitors to robustly identify needs by the age of 4 months and that in very deprived populations some sort of ongoing universal service should be offered for at least the first year.



  • Funding: This analysis was supported by the Starting Well Programme, funded by the Scottish Executive and commissioned through NHS Health Scotland. RW is currently funded by the Chief Scientist Office of the Scottish Government Health Directorate to undertake an evaluation of the implementation of Health for All Children 4 in Scotland (Fellowship reference CAF/06/05).

  • Competing interests: The researchers were independent from funders at all times with regard to study design, analysis and interpretation of data, deciding to submit the results for publication and writing the paper.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.