Prison population: social care needs

4. Data and methodology

This study makes use of detailed publicly available data from the Scottish Household Survey on a representative sample of individuals in Scotland, alongside individual-level data provided by the Scottish Government on individuals receiving social care, a snapshot of individual-level data in the current Scottish prison population, as well as preliminary Test of Change screening results from a Test of Change site. This is augmented by SPS prisoner survey results.

This data is then used to create probability models to extrapolate from the nonprison population in Scotland to the prison population while controlling for age, gender, and premature ageing, controlling either for deprivation levels, or differential ageing between prison and non-prison population. Different methodologies are employed to assess needs based on both levels of need in administrative data and also based on self-reporting of needs.

This section presents the methodology that was developed to identify the extent of social care support needs in the Scottish prison population as well as the data used to feed into the statistical modelling. As there is no precedent for statistical modelling of social care support need in prison settings, a novel methodology was developed to arrive at central estimates of social care support need as well as upper and lower bounds. The limited data on the prison population prevents the direct calculation of the social care needs of people in Scottish prisons, and it makes the need for developing and following a novel modelling route imperative. 

With model choice guided by the availability of data, models described in this section have been designed to produce estimates of social care support needs and can produce new estimates based on future data updates. The results can be used to inform policies on how to support people in Scottish prisons with social care support needs, which will guide future assessments of funding requirements.

Data

A variety of datasets were collated and analysed in order to better understand the likely social care support needs of the Scottish prison population and to produce a range of potential estimates of this. The data sources used were a mix of publicly available data and data provided by the Scottish Government:^[26]

• Scottish Household Survey (2017):^[27] this was used to estimate an individual-level probability model for the non-prison population in Scotland of having social care support needs. The annual survey covers a sample of 1,000 households, and it asks participants questions on personal characteristics, such as gender and age, the deprivation quantile, as well as whether the individual has a condition that prevents him/her from carrying out activities in his/her every-day life. The self-reported data was accessed via the UK Data Service,
• Social Care Survey (2015-2017): this dataset was provided by the Scottish Government and provides information on the demographic characteristics of people who receive social care support. All 32 Scottish Local Authorities provide information about people being assessed for social care support, including home care services (such as re-ablement services), meals services, housing support services, community alarms and other telecare services, self-directed support (including direct payments), as well as social worker/support workers services. People receiving social care support are people with dementia, mental health problems, learning disability, physical disability, drug- and/or alcohol related problems, people who need palliative care or people who are carers, people who have problems arising from infirmity due to age, as well as other vulnerable groups, including people with HIV/AIDS, or acquired brain injury,
• A data extract from the SPS PR2 system (2020): this dataset was provided by the Scottish Government and includes information on the prison population's demographic characteristics and potential social care support needs. PR2 is an operational system used by the SPS to manage the prison population – it is not designed as a way of monitoring health or social care needs,
• SPS Prisoner Survey (2017-2018): an anonymised dataset of the Scottish Prisoner Survey was provided by the Scottish Government, including self-reported information on demographic characteristics of people in custody and questions covering potential social care support needs. The 16^th wave of this survey includes 3,137 observations across the 15 prisons in Scotland. Some of the caveats of the SPS Prisoner Survey is that it is not representative of the prison population, meaning that the most marginalised and disadvantaged individuals might not be taken into account. Additionally, the questions in the SPS survey focus on personal care needs, rather than a broader definition of social care support needs,
• Preliminary Test of Change screening results: an anonymised dataset on preliminary Test of Change screening results was provided by the Scottish Government. The dataset included results of screened people in custody in one of the Test of Change sites, which adopted a wide-ranging approach to assessing needs among people in custody. The results flagged various social care needs related to one of the following categories: alcohol dependency, alcohol related brain injury, autism/Asperger, blood-borne virus, dementia/memory/incapacity, drug dependency, hearing impairment/deaf, learning difficulties, mental health, neurological condition (not dementia), frailty, physical disability or long term condition, visual impairment or "other". 

Based on the Scottish Household Survey, the non-prison population is on average, as expected, in the middle quintile for the Scottish Index of Multiple Deprivation (SIMD).^[28] Although the PR2 extract did not provide deprivation information for people in custody, according to ScotPHO (2010), people in Scottish prisons are most likely to come from the more deprived areas of Scotland; on average, they are in the bottom two SIMD quintiles. More recent data shows that this picture has not changed in the last decade; people in custody are three times more likely to come from the 10% most deprived areas of Scotland (The Scottish Government, 2020).

Methodology

Using the data described above, two different methodologies have been implemented to estimate the social care support needs of the Scottish prison population, supported by presenting three additional data points to inform the assessment. These are described below in non-technical terms, but further technical detail on the methodologies is provided in an annex.

Methodology 1 – Applying support levels of the general population

Methodology 1 uses data on the non-prison population to statistically estimate a probability model^[29] that can then be applied to each individual in the current Scottish prison population. This model was based on calibrating social care support needs in the Scottish Household Survey (2017) to the level of social care provision reported in the Social Care Survey data extract (2015-2017). It should be noted that this calibration implies that this methodology will therefore identify the proportion of people in prison in Scotland who not only have social care needs but would have met eligibility criteria thresholds for support. This is not a feature of the other methodologies below.

The two datasets provided information on age, gender, and deprivation of non-prison population. The probability model that has been estimated controls for the aforementioned demographic characteristics. This estimated relationship is applied to each individual in the PR2 extract, which provided information on the age and gender of the prison population.

This results in assigning each member of the prison population a percentage probability of having social care support needs that would be covered and supported in the community. These probabilities can then be summed across the prison population to produce an estimate of the proportion of the prison population with needs. For example, if four individuals each had an estimated 25% probability of having a social care support need then summing across the four individuals would produce an estimate of one person in this population of four having a social care support need.

Methodology 2 – Self-reporting social care support needs

The second methodology is similar to the first, but instead of creating a model calibrated to the Social Care Survey data extract (2015-2017), the model relies on self-reporting of social care needs in the Scottish Household Survey (2017). The model controls for the same demographic characteristics as in Methodology 1. Then, the estimated relationship is again applied to each person in the PR2 extract. 

Three versions of Methodology 2 have been implemented. Methodology 2a is exactly as per Methodology 1 except for using self-reported assessments of need to estimate probabilities. Methodology 2b excludes the deprivation adjustment. Methodology 2c also excludes the deprivation adjustment, but instead adjusts directly for premature ageing. Following evidence from the literature on accelerated ageing of people in custody, and adopting the 10 year-differential ageing between prison and non-prison population, the age of the prison population is shifted by a decade to better match the age of the non-prison population. For example, under Methodology 2c, an individual in prison aged 50-59 is treated as 60-69 years of age for extrapolating from the non-prison population.

Data Point 1 – Using the SPS survey

In contrast to the first 2 methodologies, which extrapolate from the non-prison population to the prison population, Data Point 1 presents the estimate of the percentage of people in custody having personal care needs based on the SPS Survey 2017. 

The survey includes questions about requiring assistance in managing: (i) transferring/moving around the prison, (ii) washing/personal care, (iii) toilet use, (iv) dressing, (v) collecting meals, (vi) eating meals, as well as questions on physical and mental health conditions, hidden disabilities, and substance use. This survey data was used to calculate the percentage of survey respondents who required assistance in managing at least one of the aforementioned activities of daily living. This sub-sample includes people who have physical and/or mental health conditions, hidden disabilities, as well as people with drug or alcohol problems. People who have these types of conditions but who report that they can deal with activities of daily living are not categorised as people with personal care, and some of them might need healthcare support instead of social care.

Data Point 2 – Using the PR2 data extract

Data Point 2 uses information provided in the data extract from the SPS PR2 system (2020). This data source contains information on the percentage of people in custody that have potential social care needs, broken down by category. The potential social care needs of people in custody include needs stemming from at least one of the following: physical issues, mental health, drug use, alcohol use, learning disabilities, neurological problems, autism, dementia, palliative care issues, and age-related frailty. It should be noted that this data source is not considered to be fully reliable on social care data, which is why it is not the central focus of determining the total level of need. It is thought there are both missing cases and resolved cases, i.e. a mix of false negatives and false positives, in this data source. 

Data Point 3 – Using preliminary Test of Change screening results

As with Data Point 1 and Data Point 2, Data Point 3 contains information on the percentage of people with social care needs in one of the Test of Change sites. The preliminary Test of Change screening results are used in this methodology: if an individual is flagged as having at least one condition (alcohol dependency, alcohol related brain injury, autism/Asperger, blood-borne virus, dementia/memory/incapacity, drug dependency, hearing impairment/deaf, learning difficulties, mental health, neurological condition (not dementia), frailty, physical disability or long term condition, visual impairment and "other"), then this person is considered to have social care needs.

An important limitation of this approach is that the single Test of Change site may have had a level of social care needs unrepresentative of other prisons in Scotland.