Universal Health Visiting Pathway evaluation: phase 1 report - routine data analysis - baseline outcomes

Appendix 1

A. Outcomes recorded in child health reviews

Nine outcome measures were recorded in two child health reviews:

• infant feeding (Outcomes 1a and 1b) in the 6-8 week review, and
• in the 27-30 month review:
• parental / carer smoking (Outcomes 4a and 4b),
• developmental concerns (Outcomes 5a, 5b and 5c), and
• child's BMI (Outcomes 6a and 6b).

For all these outcomes, data quality was explored initially. First, the percentage of each birth cohort with the subsequent appropriate review was calculated. For example, for the infant feeding outcomes:

Percentage of children in specific birth cohort with subsequent record of 6-8 week review = (Number of children in that birth cohort with subsequent record of 6-8 week review / Number of live births in that quarter) x 100%

The percentage of each birth cohort with a 27-30 month review was similarly calculated for parental/carer smoking, developmental concern, and BMI outcomes.

Second, the percentage of the birth cohort with 'meaningful' data was computed. Again, for the infant-feeding outcomes:

Percentage of children in specific birth cohort with meaningful data for 'current feeding' = [Number of children in that birth cohort with meaningful data for 'current feeding' (i.e. B or F or M or O or U) / Number of live births in that quarter] x 100%

'Meaningful' or 'valid' data for these four outcome groups are defined as follows:

• 'Meaningful' data for the infant feeding outcomes are defined as either B (breast milk only), F (formula milk only), M (mixed breast and formula milk), O (other) or U (unknown).
• 'Meaningful' data for the smoking outcomes are responses 'yes' or 'no' to 'primary carer current smoker' (Outcome 4a) and 'child exposed to second hand smoke' (Outcome 4b).
• For the two BMI outcomes, 'valid' data was defined as the recording of both height and weight, to enable the child's BMI at age 27-30 months to be calculated. BMI is calculated as:

(weight in kg) / [(height in m)²]

• See subsection 'Developmental concerns' below for a description of the developmental-concern outcomes.

Infant feeding: outcomes 1a and 1b

Data source for all child health data is CHPS-PreSchool May 2020, Public Health Scotland, and for births is NRS.

The two outcome measures for infant feeding are:

• exclusive breast milk feeding at 6-8 weeks (Outcome 1a), and
• any breast milk feeding at 6-8 week (Outcome 1b).

They are calculated thus:

Outcome 1a = [(Number of children in specified birth cohort with 'current feeding' = B) / (Number of children in that birth cohort with meaningful data for 'current feeding')] x 100%

Outcome 1b = {[(Number of children in specified birth cohort with 'current feeding' = B) + (Number of children in specified birth cohort with 'current feeding' = M)] / (Number of children in that birth cohort with meaningful data for 'current feeding')} x 100%

Parental smoking: outcomes 4a and 4b

Data source for all child health data is CHPS-PreSchool May 2020, Public Health Scotland, and for births is NRS.

The two outcome measures for parental/carer smoking are:

• primary carer current smoker at 27-30 months (Outcome 4a), and
• child exposed to second hand smoke at 27-30 months (Outcome 4b).

These outcomes were calculated using a similar method to the feeding outcomes above.

Developmental concerns: outcomes 5a, 5b and 5c

Data source for all child health data is CHPS-PreSchool May 2020, Public Health Scotland, and for births is NRS.

The three developmental-concern outcome measures relate to a number of developmental domains, which changed slightly in 27-30 month reviews recorded from April 2017 onwards. Appendix Table 1 displays how the domains in the earlier reviews (undertaken before April 2017) map to the domains in the later reviews. The three developmental-concern outcome measures looked at as part of this evaluation are:

• any developmental concern at 27-30 months (Outcome 5a),
• any concern about speech, language and communication development at 27-30 months (Outcome 5b), and
• any concern about social and emotional development at 27-30 months (Outcome 5c).

Health visitors recorded N (no concerns), C (concern newly suspected), P (concern/disorder previously identified) or X (assessment incomplete) on the 27-30 month review form.

As with the other three groups of outcomes derived from CHSP-PS data, first the level of data recorded in each birth cohort for each developmental outcome was explored, and then the percentage of children with 'meaningful' data who have the outcome was calculated for each birth cohort.

For Outcome 5a (any developmental concern at 27-30 months), first the percentage with complete data for all the developmental concerns (that is, N, C or P recorded for each developmental domain) was calculated:

Percentage of children in birth cohort with complete data for 'development' = [Number of children in specified birth cohort with complete data for 'development' (i.e. N or C or P) for each development domain / Number of live births in that quarter] x 100%

The percentage of the birth cohort with any developmental concern at 27-30 months is:

Outcome 5a = [Number of children in specified birth cohort with C or P recorded against any developmental domain / Number of children in that birth cohort with complete data for 'development' (i.e. N or C or P) for each development domain] x 100%

It should be noted that for some children, no developmental concern is noted, but some of the developmental-concern domains (see Appendix Table 1) in the 27-30 month review are incomplete or missing:

Percentage of birth cohort who have no concern recorded, but data in some domains are incomplete or missing = (Number of children in specified birth cohort without a concern, but some domains incomplete or missing / Number of live births in that quarter) x 100%

For Outcome 5b, any concern about speech, language and communication (S, L & C) development at 27-30 months, the two calculations are:

Percentage of birth cohort with meaningful data for 'S, L & C development' = [Number of children in specified birth cohort with meaningful data for 'S, L & C development' (i.e. N or C or P) / Number of live births in that quarter] x 100%

Outcome 5b = [Number of children in specified birth cohort with C or P recorded against S, L & C domain / Number of children in that birth cohort with meaningful data for 'S, L & C development' (i.e. N or C or P)] x 100%

For Outcome 5c, any concern about social and emotional (S & E) development at 27-30 months, the relevant domains for this outcome measure are:

• Social, Emotional, Behavioural and Attention domains for reviews provided between April 2013 and March 2017 (i.e. 16 birth cohorts from January-March 2011 to October-December 2014); and

• Personal/Social and Emotional/Behavioural domains for reviews provided from April 2017 onwards (i.e. 5 birth cohorts from January-March 2015 to January-March 2016).

Supplementary Table 1 How developmental domains in the earlier version of the 27-30 month child health review (April 2013 to March 2017) map to domains in the later version (from April 2017 onwards)

Reviews provided April 2013 – March 2017 (i.e. approximate birth cohorts January 2011 – December 2014)	Reviews provided from April 2017 onwards (i.e. approximate birth cohorts January 2015 – March 2016 in Phase 1)
Social*	Personal/Social*
Emotional*	Emotional/Behavioural*
Behavioural*
Attention*
Speech, Language & Communication	Speech, Language & Communication
Fine Motor	Fine Motor
Gross Motor	Gross Motor
Vision	Vision
Hearing	Hearing
n/a	Problem Solving

Note

*The domains proceeded by an asterisk are those relating to Outcome 5c, any concern about social and emotional development at 27-30 months. There are four domains in the earlier version of the review form, and two in the later version.

Hence the two calculations for Outcome 5c are:

Percentage of birth cohort with meaningful data for 'S & E development' (each relevant domain) = [Number of children in specified birth cohort with meaningful data for 'S & E development' (i.e. N or C or P for each S & E domain) / Number of live births in that quarter] x 100%

Outcome 5c = [Number of children in specified birth cohort with C or P recorded against S & E domains / Number of children in that birth cohort with meaningful data for 'S & E development' (i.e. N or C or P)] x 100%

It should be noted that Greater Glasgow and Clyde (GGC) did not change to use the ASQ in line with other Health Boards. For this reason, developmental data for children from GGC is incomplete on the data systems.

BMI: outcomes 6a and 6b

Data source for all child health data is CHPS-PreSchool May 2020, Public Health Scotland, and for births is NRS.

Height and weight measurements recorded on the 27-30 month review form where the values were greater or less than 7 standard deviations from the mean value are considered 'extreme' values and have been excluded. BMI is calculated as:

BMI = Body mass in kg / (Body height in m)²

The two BMI outcomes are:

• Child at risk of overweight or obesity (BMI ≥85^th centile) at 27-30 months (Outcome 6a); and
• Child clinically obese (BMI ≥98^th centile) at 27-30 months (Outcome 6b).

These outcomes are calculated using a similar method to the feeding outcomes.

B. Other outcomes

Immunisations: outcome 2

Data source is Scottish Immunisation and Recall System (SIRS).

Longitudinal data at individual-child level are required to construct the immunisation-coverage outcome; hence the number of live births per quarter (which forms the denominator for Outcome 2) has been taken from SIRS and is restricted to children who had been registered for the Child Health Programme (CHP) between birth and second birthday.

Complete coverage of universal primary and end infancy immunisations by second birthday (Outcome 2) is defined as the child having received all of the following vaccines:

• 3 doses of the 5-in-1 (DOB up to 31 July 2017) or 6-in-1 (DOB from 1 August 2017) vaccine (provided at any point);
• 1 dose of pneumococcal (provided on or after 1^st birthday);
• 1 dose of HiB/MenC booster (provided on or after 1^st birthday); and
• 1 dose of MMR (provided on or after 1^st birthday).

Outcome 2 = Percentage of children in specified birth cohort with complete immunisation coverage recorded on SIRS by 2^nd birthday

= (Number of children in birth cohort with complete immunisation coverage recorded on SIRS by 2^nd birthday / Number of live births in that quarter) x 100%

The Scottish Index of Multiple Deprivation (SIMD) has been used to categorise the data into population-weighted quintiles, based on the child's home postcode at the time of their second birthday. The SIMD version has been used according to the point in time when the children reached their second birthday. A small number of SIMD quintiles are missing/unknown due to missing postcode information.

The following versions of SIMD have been used:

• 2011 births (assessed in 2013): SIMD 2012;
• 2012-2014 births (assessed in 2014-2016): SIMD 2016;
• 2015-2016 births (assessed in 2017-2018): SIMD 2020v2.

Health Board boundaries change over time; Health Board of residence is based on the 2019 Health Board configurations for all time periods presented. A small number of Health Boards are missing or unknown due to missing postcode information and therefore the Health Board of residence could not be mapped for these records.

Dental care: outcome 3

Data sources are MIDAS, Public Health Scotland and NRS for births.

Outcome 3 is attendance at a dentist by the child's second birthday, and includes attendance at any NHS dentist providing general dental services: independent contractor 'high-street' dentists and the Public Dental Service are both included.

Outcome 3 = Percentage of children in the birth cohort who attended a dentist before 2^nd birthday

= (Number of patients in the birth cohort who attended a dentist before 2^nd birthday / Number of live births in that quarter) x 100%

Where the data are stratified by SIMD quintile, some children are excluded as their SIMD status could not be identified (see SIMD data subsection above in main text).

Unintentional injuries: outcomes 7a, 7b and 7c

Data sources are SMR01, ISD Scotland, PHS, and NRS for births.

Data for hospital admission for unintentional injuries were collected on discharge from non-obstetric and non-psychiatric hospitals (SMR01) in Scotland. The definitions for each of the three outcomes using SMR01 emergency admission type and ICD10 codes for each outcome are listed in Appendix 2 below.

The data are a count of individuals, not a count of admissions: a patient could have had several emergency admissions in any year, but would only be counted once in our data. Data for NHS patients treated in non-NHS locations (for example, private hospitals, hospices, etc.) are incomplete.

Where the data are stratified by SIMD quintile, some children are excluded as their SIMD status could not be identified. SMR01 completeness varies across health boards, and where fewer than 5 children have an outcome, the values have been suppressed due to the potential risk of disclosure and to help maintain patient confidentiality.

The three outcomes are calculated as follows:

Outcome 7a = (Number of children in the birth cohort with any hospital admission for an unintentional injury by 3^rd birthday / Number of live births in that quarter) x 100%

Outcome 7b = (Number of children in the birth cohort with any hospital admission for an unintentional poisoning, burn or scald by 3^rd birthday / Number of live births in that quarter) x 100%

Outcome 7c = (Number of children in the birth cohort with any hospital admission for an unintentional long-bone fracture or head injury by 3^rd birthday / Number of live births in that quarter) x 100%

Child Protection Register: outcomes 8a and 8b

Data source is Scottish Government Child Protection data, 2012-13 to 2018-19.

The two CPR outcomes are:

• Child placed on Child Protection Register at any point between birth and 3^rd birthday (Outcome 8a); and
• Child placed on CPR for ≥ 6 months between birth and 3^rd birthday (Outcome 8b).

They are calculated for each birth cohort as a percentage of live births within that quarter.

There are three important points to note, when looking at the results for Outcomes 8a and 8b. First, individual-level child protection data only began on 1 August 2012; no individual level records are held centrally before this date. As a result of this, recorded values for the number of children placed on the CPR for each of the first seven birth cohorts will be lower than the actual values, because children could have been registered before individual level data collection started.

Second, because of the start date for data collection of the individual level data, sometimes the first record is a de-registration – i.e. the child was registered before 1 August 2012, but only appears in the data when they are de-registered after that date. For these children, it is known that they were registered at some point, but the date of registration is unknown. Thus for calculating Outcome 8b, Placed on child protection register for ≥6 months between birth and third birthday, it has been assumed that all these children were registered on 1 August 2012, even though it is probable that some were registered some time prior to that date.

The third point refers to the data in the final year, i.e. from April-June 2015 to January-March 2016. Individual level data relating to children placed on the CPR are available for Scotland, and have also been broken down to council (rather than health board) level. However, data values less than 5 are suppressed to avoid identification of children, and as a result, data are unavailable for many of the smaller councils. Therefore it has not been possible to remove from the dataset children who were born in Shetland and Tayside in the final four quarters of the time period or children born in Orkney in the final two quarters. Thus, approximately 4,256 of the 55,258 children born during this final year (7.7%) may have been partially exposed to the UHVP.

In summary, it should be borne in mind that the data values for the first seven birth cohorts for Outcomes 8a and 8b are almost certainly underestimated, and data values for the final year (i.e. four quarters) may include a small number of children who have been partially exposed to the UHVP.

For those children with multiple episodes on the CPR, the sum of the periods has been calculated in days. Thus for Outcome 8b, Placed on CPR for ≥6 months, the cumulative time the child has spent on the register may be more than 6 months (183 days), although they will not necessarily have spent a single consecutive six-month period on the register.

Children can be on the child protection register and looked after at the same time: for example, of the 2,599 children on the CPR at 31 July 2019, 551 had LAC status. It is probable that some of these instances may be due to the time-lag between a child becoming looked after and being de-registered from the CPR.

Data stratified by SIMD quintile are not available for child protection interventions, due to the potential risk of identification (see SIMD data section above).

Looked After Children: outcomes 9a and 9b

Data source is Scottish Government Children Looked After statistics, 2008-09 to 2018-19.

The two LAC outcomes are:

• Registered as having Looked After Child (LAC) status at any point between birth and 3^rd birthday (Outcome 9a); and
• Registered as having LAC status for ≥ 6 months between birth and 3^rd birthday (Outcome 9b).

Each outcome is calculated for each birth cohort as a percentage of live births within that quarter.

A small proportion of the records for Looked After children (LAC) show children as being in care at 31 July year X, but they do not appear in the following data collection (year X + 1); for example, some children are shown as having LAC status at 31 July 2015, but do not appear in the 2015-2016 data return. This seems to happen when there is a lag between the child's records being updated and the data being submitted to ScotXed, suggesting that they ceased to have LAC status prior to 31 July. When calculating the time in care in such instances, it has been assumed that the child had LAC status until 31 July of the last data extract in which they appeared.

For those children with multiple episodes of care, the sum of the periods has been calculated in days. Again, for Outcome 9b, LAC status for ≥6 months, the cumulative time the child has been looked after may be more than 6 months (183 days), although they will not necessarily have spent a consecutive six-month period in care.

Children can be on the child protection register and looked after at the same time: for example, of the 2,599 children on the CPR at 31 July 2019, 551 had LAC status. It is probable that some of these instances may be due to the time-lag between a child becoming looked after and being de-registered from the CPR.

As with Outcomes 8a and 8b, data relating to children registered with LAC status are available for Scotland, and have also been broken down to council (rather than health board) level. However, data values less than 5 are suppressed to avoid identification of children, and as a result, data are unavailable for many of the smaller councils. Consequently, it has not been possible to remove from the dataset children who were born in Shetland and Tayside in the final four quarters of the imte period or children born in Orkney in the final two quarters. Thus, approximately 4,256 of the 55,258 children born during this final year (7.7%) may have been partially exposed to the UHVP. Therefore it should be borne in mind that the data values for the final year (i.e. four quarters) may include a small number of children who have been partially exposed to the UHVP.

Data stratified by SIMD quintile are not available for Looked After Child interventions, due to the potential risk of identification (see SIMD data section above).