Prison population: social care needs

5. Results

The central, and preferred, estimate produced by this study is for 7-10% of the prison population in Scotland to have social care needs. This is based on extrapolating from the non-prison population based on characteristics including age, deprivation levels, gender and the premature ageing of the prison population.

A wide range of estimates are available outside of the central estimate. Extrapolating from the non-prison population receiving care produces an estimate that around 3% of the Scottish prison population would have met eligibility criteria thresholds for support outside of prison. Instead, focusing on social care data in the SPS PR2 extract provides the highest estimate in the range, with a result of 13%.

On average, people with longer sentences have a higher probability of having social care needs. Additionally, the probability of having social care needs is higher in older age groups. Women seem to have a slightly higher probability of having social care needs than men.

Sensitivity analysis shows that the percentage variability of estimates of social care need is greater for any given individual prison compared to variability in the overall prison population, i.e. total social care needs across all prisons are likely to fluctuate less than the needs of each individual prison. At any given time, some prisons will experience higher-than-expected levels of need (given the characteristics of their prison population), and others will experience belowaverage levels.

This chapter brings together the results from the statistical modelling explained in Chapter 4. It further includes the results from the Monte Carlo analysis for the first 2 methodologies, as well as the caveats and limitations for using the statistical modelling to estimate the social care support needs of the prison population in Scotland.

Estimates of social care support needs 

The results from the methodologies we developed in the previous section are summarised in the table below. In this table, we also include the SPS estimate referenced in the Dundee Report (Levy et al., 2018).

Table 1. Estimates of social care support needs of people in Scottish prisons

	Methodologies / Data Points	Central estimate for the percentage of prison population with social care needs
1	Methodology 1 – Applying support levels of the general population	2.7%
2a	Methodology 2a – Self-reporting social care support needs (model A)	10.1%
2b	Methodology 2b – Self-reporting social care support needs (model B)	7.1%
2c	Methodology 2c – Self-reporting social care support needs (model C)	10.2%
3	Data Point 1 – Using the SPS survey	8.3%
4	Data Point 2 – Using the PR2 data extract	13.0%
5	Data Point 3 – Using preliminary Test of Change screening results (for one prison only)	12.2%
6	Memorandum item: SPS (2018)	1.9%

Methodology 1 provides an estimate of 2.7% of people in Scottish prisons having social care support needs and where, outside of prison, they would be likely to meet eligibility criteria thresholds. This methodology is based on extrapolating from the number of individuals in the wider population who have been assessed and receive social care support from the Local Authorities to cover their social care support needs. This will therefore exclude cases where individuals receive social care support from family members, which will not be possible for members of the prison population, or people who have social care needs but are not eligible for receiving social care support in the community. Consequently, this methodology may result in an under-estimation of social care support needs in prisons.

Methodologies 2a, 2b and 2c provide estimates of 10.1%, 7.1% and 10.2% respectively, and they are based on extrapolating from the number of individuals in the wider population who self-report having social care support needs. Methodologies 2a and 2c control for premature ageing of the prison population by controlling either for deprivation or the different effective ageing rates between the prison and non-prison population. These methodologies include people who have social care support needs, irrespective of the kind of social care support they need and who will provide it. These results should be viewed as the central estimates of this study, giving a preferred range of 7-10%, albeit with indications from other data (discussed below) that a higher proportion of the prison population in Scotland may have support needs.

Data Point 1 relies on people in custody self-reporting their personal care needs. This methodology shows that 8.3% of people in Scottish prison settings have personal care needs. As explained in Chapter 4, this methodology calculates the percentage of survey respondents who required assistance in managing i) transferring/moving around the prison, (ii) washing/personal care, (iii) toilet use, (iv) dressing, (v) collecting meals, and/or (vi) eating meals. The group of people that need assistance with at least one of these daily activities also includes people with physical and/or mental health conditions, hidden disabilities, as well as people with drug or alcohol problems. People who have these kinds of conditions but who have reported that they can deal with everyday activities are not considered as having personal care needs. 

Data Point 2 suggests that an estimated 13.0% of people in Scottish prisons have conditions expected to cause social care needs based on PR2 data. Data Point 3, which uses preliminary Test of Change screening results, provided an estimate of 12.2% of people having social care needs in one Test of Change site. These results provide some tentative evidence that the proportion of the prison population in Scotland with social care needs may be slightly higher than the estimates presented in this study, although neither data source should be considered to be robust. The social care element of PR2 data is not thought to be fully reliable because it is thought there are both some missing active cases and some missing resolved cases, i.e. a mix of false negatives and false positives, in this data source. Data Point 3 is recent and reliable but unfortunately covers only one prison, making it difficult to draw inferences about the overall prison population in Scotland from this source.

The results in the table above show that there is no single answer to the question of the proportion of the prison population in Scotland that have social support needs, which is consistent with messages from stakeholder interviews in Section 3. Providing social care support only to those who would meet eligibility criteria thresholds outside of prison results in the relatively low estimate of 2.7% generated by Methodology 1. Focusing primarily on personal care rather than social care generates the 1.9% estimate from SPS (2018). Under these definitions, the level of current unmet needs is relatively low. 

In contrast, Methodologies 2a, 2b and 2c suggest a range of 7-10% based on extrapolating from social care needs in the non-prison population to the prison population. This range is notably higher than the 2.7% generated by Methodology 1 and the 1.9% estimate from SPS (2018). Going further, PR2 data and early Test of Change results suggest even higher estimates of up to 13.0%. 

Disaggregating results to prisoner status, age groups and gender

The results for methodologies 1 and 2a are disaggregated by prisoner status (unsentenced, short-term, long-term, life sentences), age groups and gender. Sentences are either determinate or indeterminate. The latter category does not have a set end point – for example, a life sentence is considered an indeterminate sentence. Determinate sentences are categorised as short-term and long-term sentences. A short-term sentence is less than four years, while a person given a long-term sentence will be in prison for four years or more.^[30] The ages of people in custody are categorised to the following age groups: under 30, 30-39, 40-49, 50-59, 60-69, and over 70 years old. The results are also broken down by gender.

Table 2. Disaggregation of estimates of having social care needs to prisoner status, age groups and gender

	Methodology 1	Methodology 2a
Prisoner status (%)
Unsentenced	2.2%	8.9%
Short-term sentence	2.5%	9.6%
Long-term sentence	3.9%	11.4%
Life sentence	3.3%	12.9%
Age groups (%)
Under 30 years old	1.0%	4.7%
30-39 years old	1.9%	8.0%
40-49 years old	2.8%	12.6%
50-59 years old	3.7%	19.8%
60-69 years old	8.0%	22.3%
Over 70 years old	32.0%	27.9%
Gender (%)
Women	3.2%	10.8%
Men	2.7%	10.0%

Methodology 1 provides lower estimates, on average, than methodology 2a. The results disaggregated by prisoner status show that, on average, the longer the sentence, the higher the probability of having social care needs. For example, the results from methodology 2a show that 10% of people who are serving a short-term sentence have social care needs, while 13% of people with a life sentence need social care support. Moreover, the probability of having social care needs is higher in older age groups. In particular, based on methodology 1, 1% of people aged less than 30 years old have social care needs, while 32% of people aged 70 years old or more have social care needs. When results are disaggregated by gender, it is estimated that women are more likely than men to have social care needs.

Understanding uncertainty around estimates

This section includes probability distributions for the level of need by performing a technique known as "Monte Carlo analysis" on the first two models (methodology 1 – Applying support levels of the general population and Methodology 2a – Self-reporting social care support needs). The models above estimate a probability for each individual of having social care support needs rather than assigning definite support needs to some individuals and a definite lack of support needs to others. However, in reality, these probabilities will translate into some individuals with social care support needs and some without. For example, if ten people in custody each have a 10% chance of having social care support needs, then on average 1 in 10 will be assessed to have needs, but by running simulations sometimes this will be zero out of 10 and sometimes 2 or more out of the 10.

Monte Carlo analysis simulates possible realities based on predicted probabilities a large number of times (1,000 times in this case), similar to producing a distribution around flipping a biased coin a large number of times, recording whether it was heads or tails (1 if the prisoner has social care support needs and 0 otherwise). The bias in the coin is the estimated probability of having social care support needs. This is done to provide an understanding of the potential variability around predicted total numbers of people in custody with social care support needs.

Applying this technique based on Methodology 1, which produced a central estimate of 2.7%, provided a range of 2-3.5% of people in custody having social care support needs, suggesting a lower bound of 2% and an upper bound of 3.5% of individuals. Applying this technique based on Methodology 2a produced an estimate that 8.5-11.5% of people in Scottish prisons having social care support needs, noting that the central estimate from Methodology 2a is 10.1%.

Moreover, a random prison was selected to show how variability for a single prison can be quite different to the whole prison population level so that at any given time there will be some prisons with above-average social care support needs and some below. Methodology 1 produced an estimate of 2.3% of people in the randomly selected prison having social care support needs, but Monte Carlo analysis estimated a range of 0.5-4.5% of people in this specific prison having social care support needs. Methodology 2a provided an estimate of 9.1%, and the analysis estimated a range of 6-13%.

The percentage variability for an individual prison is greater than for the whole prison population, noting that e.g. the 6-13% range for an individual prison is larger than the 8.5-11.5% range for the overall prison population. This has an important policy implication in that it demonstrates the importance of enabling flexibility in how total social care funding for Scottish prisons is allocated between prisons. At any given time, each prison is likely to be facing a larger-than-normal level of requirements to provide social care support or a lower-than-normal level of requirements. An ideal funding system would accommodate these fluctuations.

Limitations of a statistical approach

Although the use of a statistical approach in this study can provide a picture of the prevalence of social care support needs among people in custody, it does not provide a complete representation of the diversity and complexity of needs. A potential limitation of using statistical approaches in this research is that it cannot, or it would be highly complex to, differentiate between drivers of need and their impact on need (e.g. which types of drivers result in what kinds of need, given the variation within drivers of need on impact of that need). Moreover, statistical methodologies do not fully identify which drivers of social care support need are most acute across the prison estate (recognising that this will be a dynamic phenomenon) and in what ways - in terms of the extent and nature of social care support required. One person may have needs emerging from an interplay of drivers, but in terms of identifying the extent of need, what needs they manifest in, who they predominantly affect in terms of demographic, and the extent of required provision as shaped by length of prison sentence, statistical approaches are limited in what they can say about this. Given the number of drivers/variables – or the complex causative and consequential nature of the phenomena shaping the need for social care support in prison, it would be challenging, if not impossible, to accurately capture this using a solely statistical approach.

Thus, statistical models are limited in their capacity to reveal the full nature of a phenomenon and how need is experienced at the individual level, as individual experiences might differ markedly depending on age, gender, nature and driver of need, experience of living with or coping with need, sentence length, etc. Statistically identifying prevalence or extent of social care support need cannot, then, fully capture the diversity and complexity of social care support needs, or how they are affected by the institutional environment.

Limitations of extrapolating from the general population 

The estimation of social care support needs of people in prison through extrapolating from the general population's needs may not take full account of the extent and effects of health inequities and social deprivation typically faced by people in prison (Levy et al., 2018). Despite controlling for the average characteristics of deprivation, the complexity and concentration of needs among the prison population might be under-estimated. For example, the degrees of mental illness, substance misuse and physical ill health are higher among people in custody when compared to the general population (Harris et al., 2007). The incidence of mental ill health is also higher among the prison population – particularly for women. Moreover, while there are higher rates of disability in the prison population, such as those with a learning disability, the extent of their prevalence, and therefore potential associated level of need, is widely under-estimated (Levy et al., 2018). Another issue when extrapolating from the general population is that, even when controlling for the age of people in custody, people in prison tend to be biologically older than their chronological age.

Anomalies of self-reporting

Key to not only identifying the level of need for social care in prisons, but also to reconciling anomalies between official statistics and self-reporting data, could be clarifying understandings of social care, the parameters surrounding what counts as social care and, thus, the identification of need, and how it can be measured. There is a difference in the literature between social care support as simply personal care and social care as a broader term, which includes supporting wellbeing and participation (in a prison context) in purposeful activities. Levy et al. (2018) suggest that limiting the identification of social care support need to personal care assistance reflects a 'reductionist approach to social care to solely health needs, leading possibly to false or narrow strategies about needs and the goals and desires of people in prison'.

Moreover, both self and social stigma are likely to play an influential role in self-reporting. Stigmatised or hidden disabilities (such as learning disabilities or mental health) are at risk of going undetected, resulting in an under-estimation of both need and lack of provision of specialist support.