Research and analysis

Better Outcomes through Linked Data: Rough sleeping and substance use treatment

Published 11 December 2025

Applies to England

Foreword

This report has been produced by the Ministry of Housing, Communities and Local Government (MHCLG) Homelessness Pilot Team, in collaboration with the Office for Health Improvement and Disparities (OHID), as part of the Better Outcomes through Linked Data (BOLD) programme.

The project linked the Rough Sleeping Questionnaire (RSQ) and National Drug Treatment Monitoring System (NDTMS) to provide unique insights into the relationship between homelessness and rough sleeping and drug and/or alcohol misuse treatment. This report provides the foundation for future work in this area, exhibiting that linking data between these datasets is possible. Moreover, it gives some conclusions which are worth exploring in more depth, such as that for these respondents, substance misuse often came before homelessness for people who didn’t experience homelessness in childhood.

I would like to thank the OHID team for their hard work, guidance and invaluable knowledge in undertaking this project and the wider MHCLG Homelessness Team for their input and quality assurance. 

MHCLG is continuing to develop its evidence base to help combat homelessness and rough sleeping. This report provides groundwork for future data linking projects to gain further insights into homelessness and rough sleeping – thereby taking us a step closer to the aims of reducing homelessness and ending rough sleeping.

Stephen Aldridge 

Chief Economist & Director for Analysis and Data 

Ministry of Housing, Communities and Local Government

List of acronyms and abbreviations

BOLD – Better Outcomes through Linked Data

DHSC – Department for Health and Social Care

DK – Don’t Know

DWTS – Don’t Want To Say

MHCLG – Ministry of Housing, Communities and Local Government

MoJ – Ministry of Justice

NDTMS – National Drug Treatment Monitoring System

OHID – Office for Health Improvement and Disparities

RSQ – Rough Sleeping Questionnaire

Executive summary

This report is a research publication led by the Ministry of Housing, Communities and Local Government (MHCLG) incorporating data from, and processed by the Office for Health Improvement and Disparities (OHID), which is part of the Department of Health and Social Care (DHSC). MHCLG and DHSC entered into a data sharing agreement as part of the Better Outcomes through Linked Data (BOLD) programme, which is led by the Ministry of Justice (MoJ).

Background

This report focuses on people with experiences of homelessness and rough sleeping who reported having received treatment for substance misuse in England. A linked dataset was created by matching NDTMS (National Drug Treatment Monitoring System) data to the RSQ (Rough Sleeping Questionnaire) data collection, using personal identifiers within the NDTMS and RSQ records. The aim was to assess the relationship between substance misuse treatment outcomes and participants’ history of homelessness and rough sleeping, and the role of the services. In addition, participant’s declarations of substance usage were compared between rough sleeping and substance misuse datasets to assess consistency in reporting

Summary of key findings

• Linking individuals’ data held separately on NDTMS and collected through the RSQ is possible, as evidenced by the successful linking of the 171 out of 354 records submitted for this process.
• There are discrepancies in the recording of this data, and in the format it was recorded, making it difficult to draw firm conclusions. However, from the data available, the key indicative results from the sample of 171 linked participants are:
  • People tended to experience substance misuse before housing difficulties, except for participants who experienced homelessness in childhood.
  • The substance use of 26% (23 of 89) of people escalated^{[footnote 1]} between completing the RSQ and starting treatment.
  • People who were long-term homeless were more commonly found to have either previously injected or are currently injecting
  • It is possible to compare housing situations across datasets, however without consistent housing categories conclusions are difficult to draw out.

Summary of limitations

Data linking between the RSQ and NDTMS enabled improved understanding of the links between rough sleeping and substance misuse from government data, such as:

• Determining the order in which people tend to first have housing difficulties and first use their problem substance.
• Comparing housing situations for those who took the RSQ before and those who took it after entering substance misuse treatment.

The findings in this report are limited due to the size of the linked data set. As well as limiting the generalisability of findings presented, it also prevented some of the planned analysis. This also means that, in presenting the findings we have suppressed some data in tables to protect individuals’ information and have, in some cases, grouped categories together.

Other important factors to consider when reading this report are:

• The RSQ is self-reporting and can therefore contradict records in the NDTMS.
• Housing categories in the NDTMS and RSQ do not fully match limiting the comparability of housing situations.

1. Introduction

Background

1.1 This report focuses on people with experiences of homelessness and rough sleeping who reported having received treatment for substance misuse in England. A linked dataset was created by matching NDTMS (National Drug Treatment Monitoring System) data to the RSQ (Rough Sleeping Questionnaire) data collection, using personal identifiers within the NDTMS and RSQ records. The aim was to assess the relationship between substance misuse treatment outcomes and participants’ history of homelessness and rough sleeping, and the role of the services. In addition, participant’s declarations of substance usage were compared between rough sleeping and substance misuse datasets to assess consistency in reporting

Rough Sleeping Questionnaire

1.2 Rough Sleeping Questionnaire fieldwork took place across six waves between February 2019 and March 2020. 991 respondents completed the survey, including details of their homelessness experience, support needs and vulnerabilities, and their use of public services.

1.3 A link to the RSQ summary pack, which includes the responses for the 991 respondents can be found at Summary pack: The Rough Sleeping Questionnaire findings. See the initial findings publication.

1.4 Participation in the RSQ was voluntary and participants could also specify what types of data linking they consented to. 424 participants from the RSQ reported both having had drug and/or alcohol treatment at some time and indicated they consented to their information being shared with other government departments for linking with data on substance misuse treatment. 70 of these were excluded from linking because there was insufficient personal information for linking.^{[footnote 2]}

National Drug Treatment Monitoring System (NDTMS)

1.5 The Office for Health Improvement and Disparities manages the National Drug Treatment Monitoring System (NDTMS), which collects data on all people accessing publicly funded structured drug and alcohol treatment in England. It captures sociodemographic information as well as data on the: substances people have problems with; interventions people receive; and treatment outcomes. People accessing treatment are asked for consent to share their information.^{[footnote 3]}

Overview of data sharing and linkage

1.6 Our aim was to link the available Rough Sleeping Questionnaire records with the NDTMS to estimate how many of the questionnaire participants had started structured drug or alcohol treatment between 1 August 2018 and 31 March 2022, and conduct analysis on this subgroup of both data sets.

1.7 There is no unique identifier which enables an individual’s drug treatment data to be perfectly linked to the same person who completed the questionnaire. However, both datasets include information (separately stored) about: a person’s forename and surname initials; date of birth; and the postcode of where a person lives (or has lived recently), though this was missing for 70 (17%) RSQ records.

1.8 This information was used to link these data using deterministic and probabilistic approaches. Deterministic linkage is defined by a clear set of rules. The simplest method to apply in this project is to require all individuals on both NDTMS and RSQ to share the same: set of initials; date of birth; the postcode of where a person lives (or has lived recently). Where a pair of records exist across the two data sets where this information matches, the deterministic interpretation is that these are the same individuals.

1.9 In contrast, probabilistic linkage is a more flexible approach which can lead to a higher linkage rate. Fellegi-Sunter is a probabilistic model for linking two data sets over several fields. This model has been incorporated into the Splink Python package by MoJ, which enables probabilistic record linkage on a large scale. The model allows for different fields to have higher discriminatory power than others (for example, if two records have the same sex, it is less discriminatory for linking purposes than the two records having the same set of first name and surname initials).

1.10 When the data was received it was checked for incorrect data (for example dates of birth in correct formats) and cleaned. Using both data sets, and a prior assumed probability of matching randomly between the two data sets, Splink determined the discriminatory power for each variable. Then a linkage model was produced and manual checks were carried out on all matches due to the small number. If the analysts found errors in the matching, then the probabilities were adjusted and the model was re-run until the analysts undertaking the linkage were satisfied and an acceptable record probability linkage rate was chosen so matches which didn’t reach the likelihood were rejected. Where records in one data set matched to multiple records in other data sets the record with the highest probability was selected. If there were multiple matches with the same matching rate so the highest probability couldn’t be determined, both matches were removed.

1.11 Where an individual from the RSQ had multiple links to individuals in the NDTMS data, if only one link was above the 98.7% probability matching threshold, the link with the highest probability of being a true match was taken. Where there were multiple links with a probability of being a true match above the threshold, these records were removed from the analysis as the appropriate match could not be determined.

1.12 Values of less than 5 people were suppressed in the data tables to prevent plausible identification of the participants (unless in a row/column which gives no personal information, e.g. Don’t Know /Don’t Want To Say/No answer). If the table included totals meaning the suppressed number could be calculated, the next lowest number in that column/row was also suppressed (see Handbook on Statistical Disclosure Control for Output for more information).

Results of data linkage

1.13 424 questionnaire records were sent to DHSC. Of these 70 records were removed due not having enough data for data linkage. This resulted in 354 records being used in the linkage process.

1.14 171 Rough Sleeping Questionnaire participant records were linked to the individual’s drug treatment records with at least one treatment episode starting between 1 August 2018 and 31 March 2022. Of these, 36 were identified through deterministic matching, and 135 were linked through probabilistic matching with a probability of 0.987 or above. These were manually checked and all matches were retained in the analytical dataset.

Figure 1.1- From 991 RSQ participants, 171 people were linked to NDTMS records with at least one treatment episode starting between 1 August 2018 and 31 March 2022

^{Flowchart showing the process of linking 171 records from 991 RSQ records.}

1.15 As shown in Table 1.1 37 participants had completed the RSQ at least 28 days before starting a period of structured drug and/or alcohol treatment. 93 participants had completed the RSQ during a period of structured drug and/or alcohol treatment. 89 participants had completed the RSQ at least 28 days after completing a period of structured drug and/or alcohol treatment.

Table 1.1- 82% (140) of linked records had more than one NDTMS episode between 1 August 2018 and 31 March 2022, with the maximum being 12

Number of NDTMS Episodes	RSQ Records	Percentage of total
1	31	18
2	66	39
3 - 6	66	39
7 - 12	8	5

^{Table showing number of NDTMS records linked to individual RSQ records for NDTMS episodes between 1 August 2018 and 31 March 2022.}

1.16 It is important to note the RSQ was a questionnaire which was only completed once by each person, whereas the NDTMS contains multiple treatment episodes. People may have had treatment before, after and at the time of completing the questionnaire. Therefore, although 171 individual’s treatment records were linked to the questionnaire, there are people who had multiple treatment episodes within a record. Hence the below table, Table 1.2, does not sum to the total number of records linked (171).

Table 1.2- Participants took the RSQ either before, after or in the same 28 days as starting substance misuse treatment

RSQ Completion	RSQ Records
At least 28 days after substance misuse treatment	37
In the same 28 day period as substance misuse treatment	93
At least 28 days before substance misuse treatment	89

^{Table showing the order in which the questionnaire was completed and substance misuse treatment was undertaken for all participants.}

Research questions

1.17 This data linking project between MHCLG and OHID aimed to address the following questions:

• What are the different housing situation and substance misuse histories reported?
• How do previous housing situation and substance misuse impact substance misuse treatment journeys and outcomes?

1.18 A question relating changes in housing situation following substance misuse treatment was proposed. However due to a mis-match between the definitions used in the questionnaire and the treatment database these could not be compared.

2. Results

All linked participants: Relationship between development of substance dependency and first experience of rough sleeping

2.1 This cohort includes all 171 participants which were linked. The aim for this group is to determine which generally occurs first: rough sleeping (or sofa surfing/homelessness) or substance misuse. The tables in this section will compare the age these participants first slept rough with the age of the first use of their problem substance. This is split into tables with and without alcohol in the substance misuse category, allowing us to understand the impact people who misuse alcohol have on the statistics in comparison to drug only users. The final two tables compare how the age of the first use of problem substance affects housing outcomes.

2.2 Only the tables involving rough sleeping information is presented in this section. Those involving homelessness and sofa surfing can be found in Annex 1 (tables A1-A4). They have been omitted from this report as the results follow a very similar pattern to those found for rough sleeping.

2.3 Table 2.1 shows a comparison of age of first sleeping rough as reported in the questionnaire, with age of first substance use as reported to treatment services. Ages have been clustered to preserve participant anonymity.

Table 2.1- People who first sleep rough over 25 were more commonly found to first use their problem substance before first rough sleeping

	Age of First Use of Problem Substance
Age of first time Rough Sleeping (RSQ)	Under 16	16-20	21 and over	N/A	Total
25 and under	22	36	14	4	76
Over 25	19	21	22	1	63
Not slept rough	<5	<10	11	1	24
DK/DWTS/No answer	<5	<5	1	1	8
Total	48	68	48	7	171

^{Age of instance of rough sleeping compared to age of first use of their problem substance.}

2.4 Most people had started their problem substance by the age of 20, as reported to their drug treatment service.

2.5 People who first sleep rough over the age of 30 were more commonly found to have the first use of their problem substance prior to experiencing rough sleeping.

2.6 It is unclear whether the first use of their problem substance or rough sleeping comes first for people who first sleep rough in the age range 16-30. This is further demonstrated by figure 2.1 below.

Figure 2.1- People that first sleep rough under the age of 16 were more commonly found to have the first use of their problem substance after experiencing rough sleeping

^{Box plots showing the age of first substance use (from treatment records) for first age rough sleeping categories (from questionnaire).}

2.7 Figure 2.1 shows for people who first slept rough under the age of 16, the median age for time of first rough sleeping (as shown on the x-axis) is lower than the median age of first use of problem substance (as shown on y-axis).

2.8 When alcohol is excluded from the substance misuse category, as shown in table 2.2 and figure 2.2, the same conclusions can be made for people that first sleep rough under the age of 16, and older than 30.

2.9 It is less clear for the ages in the range 16-30, though people were more commonly found to have had their first use of the problem substance occur before sleeping rough for people that first sleep rough 25 and under.

Table 2.2 - People that first sleep rough under 16 are more commonly found to have first used their problem substance (excluding alcohol) after first rough sleeping

	Age of First Use of Problem Substance (Excluding alcohol)
Age of first time Rough Sleeping (RSQ)	Under 16	16-20	21 and over	N/A	Total
25 and under	13	31	15	17	76
Over 25	8	11	24	20	63
Not slept rough	<5	5	10	<10	24
DK/DWTS/No answer	<5	3	1	<5	8
Total	24	50	50	47	171

Figure 2.2- People that first slept rough over the age of 30 are more commonly found to have already used their problem substance

^{Box plots showing the age of first substance use (excluding alcohol) for first age rough sleeping categories.}

2.10 The shift in median age when removing alcohol implies problems with non-alcohol substances may have been more common at older ages (over 25) for RSQ participants than alcohol.

Pre-treatment RSQ participants

2.11 There were 89 participants who underwent a course of substance misuse treatment after completing the RSQ. The tables below compare substance misuse categories and injection status between the RSQ and NDTMS.

2.12 Additional tables for this cohort which were not included in this report can be found in Annex 1 (tables A7-A13).

2.13 As shown in table 2.3, for people who attended treatment after completing the RSQ, it appears there is a greater prevalence of injecting amongst people that are homeless in the long term. Of those homeless for a year or more, 62% (23 people) of those had injected. In comparison, which those who were homeless for less than a year, 40% or lower (<5 people) had injected.

Table 2.3- For people who attended treatment after completing the RSQ, those that were long-term homelessness had a greater likelihood of either currently injecting or had previously injected

	NDTMS Injection Status
Time Spent Homeless (RSQ)	Currently Injecting/Previously Injected	Never Injected	Total
A year or more	23	14	37
Less than a year	<5	<10	10
DK/DWTS/Not asked	<25	<20	42
Total	50	39	89

^{NDTMS injection status compared to time spent homeless as reported in the RSQ.}

2.14 When comparing substance misuse categories from the RSQ to the NDTMS, 26% (23 participants) experienced an escalation in their reported substance misuse from completing the RSQ to going into substance treatment. This was composed of 14 participants who reported no drug or alcohol use in the RSQ, plus 9 who reported no alcohol or non-opioid drug use in the RSQ, but were treated for opiates. This is shown in table 2.4

Table 2.4- 26% of participants saw an escalation^{[footnote 4]} in their reported substance misuse before entering treatment 

	NDTMS Substance Reporting Group
Drug or Alcohol Type (RSQ)	Either alcohol and/or non-opiates	Opiates	Total
Any opioid use	<5	<40	42
Either alcohol and/or non opioid use	21	9	30
No drug/alcohol use	8	6	14
DK/DWTS/Missing	<5	<5	3
Total	33	56	89

^{Self-reported substance misuse in the RSQ compared to NDTMS substance reporting group.}

Post-treatment RSQ participants

2.15 Only 37 participants completed the RSQ post-treatment. To maintain participant’s privacy, the data tables for this cohort contained a substantial amount of suppression for smaller values. Hence the analysis for this section is not publishable.

Co-occurring treatment and RSQ participants

2.16 There were 93 participants who underwent substance misuse treatment during the same time period as completing the RSQ. As shown in Table 2.5 there was not consistency between substance use reported in the questionnaire and what was reported to treatment services.

Table 2.5- 36% (25 people) of participants that underwent treatment for opiate misuse reported not using opiates in the last 3 months in the RSQ

	NDTMS Substance Reporting Group Category
Drug or Alcohol Type (RSQ) used in last 3 months	Opiates	Either Alcohol and/or non-opioid	Total
Any Opioid Use	45	<5	<50
Either alcohol and/or non-opioid	13	15	28
No drug/alcohol	12	<5	<17
DK/DWTS/Missing	0	2	2
Total	70	23	93

^{NDTMS substance reporting group compared to the drug or alcohol type used in the last 3 months reported in the RSQ.}

2.17 Although these 93 participants completed the RSQ and started substance misuse treatment in the same 28 day period, only 65% (60 from 93) had a consistent reporting of their problem substance between the RSQ and NDTMS.

2.18 12 out of 70 people who had an NDTMS problem substance of opiates reported, had reported not using any substances In the last three months when completing the Rough Sleeping Questionnaire.

Summary

2.19 Linking homelessness and rough sleeping data to substance misuse records is possible. With a larger and more representative dataset, the findings could be drawn with a greater degree of confidence.

2.20 For the cohort of people in this report, it was more common for rough sleeping to occur before substance misuse for people first rough sleeping under the age of 16. This partially answers the research question “What are the different housing situation and substance misuse histories reported?”

2.21 With aligned definitions future data linking projects will be better able to answer the research question “How do previous housing situation and substance misuse impact substance misuse treatment journeys and outcomes?” (see 1.17)

2.22 There are differences in the reporting of substance use between the RSQ and NDTMS, even when the RSQ is taken in the same time period as substance treatment.

2.23 When looking substance use outcomes of rough sleepers, escalations in misuse were observed. This implies there could be opportunities to help rough sleepers through preventative approaches to their substance usage when engaging with other services. However, this difference may also be due to participants under-reporting their substance misuse in the RSQ.

3. Strengths, limitations and future work

Strengths

3.1 Valuable insights were still obtained despite the need for the data to be in suppressed summary tables.

3.2 The report helps improve understanding from the limited evidence base on links between experiences of homelessness, substance misuse and pathways to and from substance misuse treatment services.

3.3 This linking is based on the RSQ, which represents one of the largest survey data collections on people who sleep rough ever attempted in the UK.

Limitations

3.4 The small cohort size for analysis (which was 171 participants at most) is the biggest limitation to this research. After sub-grouping, some data was supressed for privacy reasons, restricting the possible analysis. Due to this small cohort size, this report is focussed on the associations in the available data, but does not draw conclusions. Generalising these findings is not recommended.

3.5 In addition, responses within the RSQ were self-reported. Some of the responses contradict information participants gave to the NDTMS. This is understandable and a useful finding when considering the use of administrative data. However it limits the conclusions which can be drawn from differences between the datasets.

3.6 Specific limitations to the amount of data available include:

• Only approximately 50% of the available records were successfully linked, and the linked dataset will exclude people who started treatment before 2018, including those still in treatment since then. This is due to consent for data linking only given to records made in this timeframe.
• Differences in housing categories limit the direct comparability of housing situations (see 1.17).
• Data concerning the age of first experience of homelessness, rough sleeping, and sofa surfing is grouped into 25 years or under and over 25. This was to ensure individuals’ personal information is protected. If age groups were more granular, more data would have needed to be suppressed.

Future work

3.7 This report has shown data linkage between rough sleeping data and substance misuse treatment data is possible, and can give valuable insights into user pathways to and from substance misuse treatment. Hence the next logical step is to expand the data MHCLG collect on rough sleeping to increase the potential for future linkage. Renewed rough sleeping data could provide more scope for future linking with the NDTMS. This will allow for more certain conclusions than those which have been proposed in this report.

3.8 To gain further insights into the impact substance misuse treatment has on homelessness and rough sleeping, a more consistent approach to defining housing situations is needed across departments. Work on this has already started within MHCLG, informed by this report.

3.9 The analysis from this report could be replicated for a larger, more representative linked dataset. It would provide additional evidence long term impacts rough sleeping, such as the costs incurred from substance misuse treatment programmes.

3.10 MHCLG is undertaking a second Rough Sleeping Questionnaire in 2024/25 and this will include variable options which align with datasets such as the NDTMS for ease of comparison.

3.11 When designing survey collections intending to link data, early conversations with dataset owners would enable better understanding of variables to match, for example use of NHS numbers. This could lead to improvements in both deterministic and probabilistic matching, whilst also protecting participant privacy.

3.12 Future analysis would also further benefit from linking to other datasets such as wider health datasets relating to the experiences of those rough sleeping. Disparities in self-reported experiences are to be expected, and therefore would need to be part of the research questions formed in the analytical planning process.

Annex 1

This section includes tables generated from the data linkage for sofa surfing and homelessness variables. As discussed in 2.2 these follow similar patterns to rough sleeping so have not been included in the body of the report.

All linked records

Table A1 - Age of first use of problem substance compared to age of first time sofa surfing

	Age of First Use of Problem Substance (NDTMS)
Age of first time Sofa Surfing (RSQ)	Under 16	16-20	21 and over	N/A	Total
25 or under	22	27	14	4	67
over 25	14	19	20	1	54
Not Sofa Surfed	10	17	13	2	42
DK/DWTS/No answer	2	5	1	0	8
Total	48	68	48	7	171

Table A2 - Age of first use of problem substance (excluding alcohol) compared to age of first time sofa surfing

	Age of First Use of Problem Substance (exc alcohol)
Age of first time Sofa Surfing (RSQ)	Under 16	16-20	21 and over	N/A	Total
25 and under	13	24	16	14	67
Over 25	<10	11	19	<20	54
Not Sofa Surfed	<5	11	14	<15	42
DK/DWTS/No answer	0	4	1	3	8
Total	24	50	50	47	171

Table A3 - Age of first use of problem substance compared to age of first time homeless

	Age of First Use of Problem Substance
Age of first time Homeless (RSQ)	Under 16	16-20	21 and over	N/A	Total
25 and under	27	33	19	4	83
Over 25	20	32	28	2	82
DK/DWTS/No answer	1	3	1	1	6
Total	48	68	48	7	171

Table A4 - Age of first use of problem substance (excluding alcohol) compared to age of first time homeless

	Age of First Use of Problem Substance (Excluding alcohol)
Age of first time Homeless (RSQ)	Under 16	16-20	21 and over	N/A	Total
25 and under	16	31	19	17	83
Over 25	8	17	30	27	82
DK/DWTS/No answer	0	2	1	3	6
Total	24	50	50	47	171

Table A5 - RSQ housing situation compared to age of first use of problem substance

	Age of First Use of Problem Substance (NDTMS)
Main Accommodation Last Month (RSQ)	Under 16	16-20	21 and over	N/A	Total
Long term/short term/Secure accommodation	31	46	36	3	116
Homeless/ Rough Sleeping	15	20	12	3	50
DK/DWTS/No answer	2	2	0	1	5
Total	48	68	48	7	171

Table A6 - RSQ housing situation compared to age of first use of problem substance (excluding alcohol)

	Age of First Use of Problem Substance (excluding alcohol) (NDTMS)
Main Accommodation Last Month (RSQ)	Under 16	16-20	21 and over	N/A	Total
Long term/short term/Secure accommodation	17	36	36	27	116
Homeless/ Rough Sleeping	6	13	14	17	50
DK/DWTS/No answer	1	1	0	3	5
Total	24	50	50	47	171

Pre-treatment RSQ participants

Table A7 - Main accommodation stayed in the month leading up to the RSQ compared with housing situation whilst entering substance treatment

	Housing Problem Record (NDTMS)
Main Accommodation Last Month (RSQ)	Housing Problem/Urgent Housing Problem	No Housing Problem	Total
Homeless/ Rough Sleeping	18	13	32
Long term/short term/Secure accommodation	24	30	54
DK/DWTS/Other	1	2	3
Total	44	45	89

Table A8 - Main accommodation stayed in the month leading up to the RSQ compared with NDTMS substance reporting group

	NDTMS Substance Reporting Group
Main Accommodation Last Month (RSQ)	Either alcohol and/or non-opiates	Opiates	Total
Homeless/ Rough Sleeping	14	18	32
Long term/short term/Secure accommodation	18	36	54
DK/DWTS/Other	1	2	3
Total	33	56	89

Post-treatment RSQ participants

Table A9 - Main accommodation stayed in the month leading up to the RSQ compared with housing situation whilst entering substance treatment

	Housing Problem Record (NDTMS)
Main Accommodation Last Month (RSQ)	Housing Problem/Urgent Housing Problem	No Housing Problem	Total
Homeless/Rough Sleeping	<10	<5	<10
Long term/short term accommodation	14	12	26
DK/DWTS/Other	<5	<5	<5
Total	22	15	37

Table A10 - Main accommodation stayed in the month leading up to the RSQ compared to NDTMS substance reporting group

	NDTMS Substance Reporting Group
Main Accommodation Last Month (RSQ)	Either alcohol and/or non-opiates	Opiates	Total
Homeless/ Rough Sleeping	<5	<5	<10
Long term/short term accommodation	16	10	26
DK/DWTS/Other	<5	<5	<5
Total	22	15	37

Table A11 - Main accommodation stayed in the month leading up to the RSQ compared to NDTMS Injection Status

	NDTMS Injection Status
Main Accommodation Last Month (RSQ)	Currently Injecting/Previously Injected	Never Injected	Total
Homeless/ Rough Sleeping	6	<5	<10
Long term/short term accommodation	9	17	26
DK/DWTS/Other	1	<5	<5
Total	16	21	37

Co-occurring treatment and RSQ participants

Table A12 - Main accommodation stayed in the month leading up to the RSQ compared to NDTMS housing record

	Housing Problem Record (NDTMS)
Main Accommodation Last Month (RSQ)	Housing Problem/Urgent Housing Problem	No Housing Problem	Total
Homeless/ Rough Sleeping	20	8	28
Long term/short term/Secure accommodation	33	30	63
DK/DWTS/Other	1	1	2
Total	54	39	93

1. An escalation is defined by either moving from no drug or alcohol use to substance use, or from using either alcohol or non-opioids to opioid use. ↩

2. See published privacy notices describing information use for this research. ↩

3. See current privacy notice. ↩

4. An escalation is defined by either moving from no drug or alcohol use to substance use, or from using either alcohol or non-opioids to opioid use. ↩