Understanding the patterns of use, motives, and harms of new psychoactive substances in Scotland

Appendices

A. Technical Appendix 1: NEO Data

Estimating the number of injecting NPS takers by applying statistical models for incomplete count data to needle exchange data

As part of this study possible sources of administrative data relating to NPS use that could be used to estimate the prevalence of NPS use in Scotland were examined. When estimating the prevalence of opiate use mark re-capture methods have been used with some success in Scotland and England but these require data from a number of different sources and detailed information at the individual level. Sufficient numbers of NPS takers are not currently appearing in these data sources to make this multi-source method an option but researchers have met with some success using single source estimation methods for areas with sparse data or only one available data set (Hay & Smit, 2003). With this in mind the research team approached NHS Greater Glasgow and Clyde and NHS Lothian with a view to obtaining data from their Neo 360 ⁰ database and assessing the feasibility of using this data to produce estimates of injecting NPS use for these two areas. The following section describes the needle exchange data used in the analysis and gives the resultant estimates of injecting NPS use for NHS Greater Glasgow and Clyde and NHS Lothian areas.

Methods

The method used to produce population estimates from a single data set is known as truncated Poisson. In our case we will be examining the number of visitors to needle exchange services over a 12-month period. When applying this method, we note the frequency for visits for every individual over the duration of the study. The frequency pattern follows a Poisson distribution but our data is incomplete as we cannot observe individuals that appear zero times, therefore the distribution is truncated below one. An estimate of the total population is given by adding all observed individuals to an estimate of those that appear zero times, the hidden population. As part of our analysis we have used two different estimators, the Zelterman (1988) and Chao estimators (1998). Both estimators can be calculated using the total number of individuals along with the lower case frequencies and given their simplicity the formulae are given below.

Zelterman's estimator of the unknown population size est(N) is given by:

est(N) = n/ [1 - Probability(f ₀)]

Where the Probability (f ₀) is e ^λ

λ = 2 * (f ₂/f ₁)

Chao's estimator is given by:

est(N) = n + ( f ₁) ²

2(f ₂)

where,

f ₁ = the number of individuals appearing just once in the data set

f ₂ = the number of individuals appearing twice in the data set

n = all individuals appearing in the data set

Both estimators are based on the lower frequencies, as it is thought that those that are observed only once or twice in a data set resemble closely those that do not appear in the data set at all. This dependence on the lower frequencies is also helpful when addressing some possible heterogeneity in the data set as those that appear a huge number of times may not reflect the 'typical' service user so a greater reliance on the earlier frequency classes lessens the impact of these groups. It should be borne in mind that if the frequency patterns for NPS users differ greatly from this description then this could impact on the estimates. Another positive aspect of the method is its ability to cope with sparse data. As with all estimation methods there are certain assumptions that must be met. These are:

1. the population is 'closed'
2. the population is homogeneous (no heterogeneity across individuals)
3. the individual probabilities of observation and re-observation are stable over time.

Closed population

The closed population assumption stresses that the true population is unaffected by births, deaths or migrations over the study period. In order to minimise the potential for this assumption to be violated we have chosen to use a slice of data covering a 12 month period.

Homogeneous population

This assumption asserts that the probability of observation should not differ greatly across groups of individuals. Both estimators used are robust in relation to heterogeneity and are known to underestimate the true population. In order to assess any heterogeneity we have attempted to stratify the estimates by gender and/or age group where the data was available. This will help us to model any heterogeneity in relation to these characteristics.

Stable (re)capture probability

This assumption would mean that attending the needle exchange on one occasion wouldn't necessarily impact the probability of future attendance. In order to lessen the impact of violating this assumption we have confined the data slice to a short 12-month period.

Data

Data is routinely collected from pharmacies/needle exchange services and entered into Neo 360 ⁰ database. Although the data is collated from a number of different sources around both health board areas, as it is collated to health board level in this database, we treat it as a single source. Every service user is assigned a unique identifier composed of initials and date of birth, given the sensitive nature of this data the reference code was 'blurred' before passing to research team so that we could identify unique individuals without having access to these identifiers.

Data and estimates for NHS Greater Glasgow and Clyde area

There were 1,896 transactions by NPS takers at NHS Greater Glasgow & Clyde needle exchange services between 01/01/15 and 31/12/15. This data corresponds to 148 individuals. Twelve individuals were from outside the NHS GGC area and so were removed from the analysis. The remaining sample were overwhelming male (88%) and ranging in age from 18 to 61 with a mean age of 38. Over half (56%) of the NPS takers accessing needle exchange services in the NHS GGC area are poly drug users with 43% reporting using heroin. When asked if they were in structured treatment for their drug use only 131 individuals responded with twenty-one (6%) confirming that they were in structured treatment. Table 1 shows the frequency data required to produce estimates of the hidden population of injecting NPS takers.

Table 1: Frequency of contact at Greater Glasgow & Clyde needle exchange services by gender

	All individuals in data set	Individuals appearing once	Individuals appearing twice
Male	120	22	14
Female	16	5	3
Total	136	27	17

Table 2 lists the estimates of injecting NPS takers. There are two estimates relating to each estimator, the unstratified estimate and an estimate stratified by gender. The Zelterman estimator produces a figure of 190 injecting NPS takers. The direct, unstratified estimate is the same as the sum of the stratified gender estimates. We can see that the stratified estimates have a slightly wider confidence interval running from 114-265. The lower bound of this confidence interval is lower than the total number of observed individuals which indicates that the observed number of women was too small for the asymptotic estimation of the 95% confidence intervals. As we would anticipate the Zelterman estimates are slightly higher than those produced by the Chao estimator which gives an estimate of 157 injecting NPS users. Again the sum of the stratified or pooled gender estimate is the same as the unstratified estimate but for the Chao estimator the confident intervals are narrower than those for the Zelterman estimates. It should be noted that the 95% confidence intervals overlap for both sets of estimates.

Table 2: Population size estimates for injecting NPS takers in NHS Greater Glasgow & Clyde area using Zelterman's (1988) and Chao's (1989) estimators

	n	Est (N)	95% CI	Hidden population	Known/ Hidden
Zelterman's unstratified estimate	136	190	130-250	54	2.52
Zelterman's stratified estimate (gender)	136	190	114-265	54	2.52
Chao's unstratified estimate	136	157	144-190	21	6.48
Chao's stratified estimate (gender)	136	157	143-210	21	6.48

Data and estimates for NHS Lothian area

There were 7,717 visits by injecting NPS takers to needle exchange services in the NHS Lothian area between 01/01/15 and 31/12/15. These visits were made by 447 individuals. Seven users were from outside the NHS Lothian area and were removed from the analysis to leave a final sample of 440 individuals. The majority of users visiting needle exchanges during this period were male (79%) and aged between 18 and 60, with a mean age of 35. Housing status was recorded for 320 of the sample and just over half owned or rented their accommodation, 41% were in temporary accommodation, the remainder were sleeping rough (6%). When asked about accessing structured treatment only 194 (44%) responded with 25% indicating they attend a specialist drug treatment service and a further 21% receive support from their GP.

Table 3: Frequency of contact at Lothian needle exchange services by gender & age group

	All individuals in data set	Individuals appearing once	Individuals appearing twice
Male	349	77	37
Female	91	14	13
18 - 24	14	7	2
25 - 34	202	38	26
35 - 64	224	46	22
Total	440	91	50

Table 3 lists the number of individuals visiting NHS Lothian needle exchanges by gender and age group. As stated previously the majority of users are male. There are few injecting NPS takers under the age of 25 attending needle exchanges. Table 4 lists the estimates of injecting NPS takers for both the Zelterman and Chao estimators with corresponding confidence intervals. There are three estimates given for each method, the unstratified estimate plus gender and age stratifications.

Table 4: Population size estimates for injecting NPS takers in NHS Lothian area using Zelterman's (1988) and Chao's (1989) estimators

	n	Est (N)	95% CI	Hidden population	Known/ Hidden
Zelterman's unstratified estimate	440	660	572-748	220	2.00
Zelterman's stratified estimate (gender)	440	673	562-784	233	1.89
Zelterman's stratified estimate (age group)	440	667	527-807	227	1.94
Chao's unstratified estimate	440	523	457-842	83	5.30
Chao's stratified estimate (gender)	440	528	454-994	88	5.00
Chao's stratified estimate (age group)	440	528	450-1357	84	5.24

The unstratified Zelterman estimate gives a similar if slightly lower result than both the stratified gender and age estimates. This would be expected as the unstratified estimate is considered an underestimate and the stratified estimates attempt to model any heterogeneity in the sample resulting in a larger estimate. As anticipated the Chao estimates are lower than those produced by the Zelterman estimator, however the age group estimate has a wider confidence interval indicating some uncertainty in the model. When the separate stratified estimates are examined this occurs in the older age group (35-64).

References Cited

Chao, A. (1989) 'Estimating population size for sparse data in caputure-recapture experiments', Biometrics 45: pp.427-438.

Hay, G. & Smit, F. (2003) 'Estimating the number of drug injectors from needle exchange data', Addiction Research & Theory 11:4, pp.235-243.

Zelterman, D. (1988) 'Robust estimation in truncated distributions with application to capture-recapture experiments', Journal of Statistical Planning and Inference 18, pp.225-237.

B. Technical Appendix 2: Prevalence estimate

Estimating the prevalence of NPS use in Scotland using a treatment multiplier

The multiplier method is a simple way of estimating unknown populations such as prevalence of drug use. The method uses the available information on the population in question as a benchmark, e.g. number of drug users in treatment, and applies a multiplier that is related to the population and has normally been derived from a small scale study.

Construction of the treatment multiplier

From our survey we know that 125 NPS users are in contact with drug treatment services. Only 194 NPS survey participants answered this question giving us a treatment rate of 64.43%. Therefore we can say that for every 1 NPS user in treatment there are 1.56 users (100/64.43).

Using the treatment multiplier to produce an estimate for the number of NPS users in Scotland

We sought data on the numbers of NPS users in treatment from the Scottish Drug Misuse Database. The data dashboard's latest available update in May 2016 gives treatment numbers for 2014/15. This data set does not currently provide detailed information on NPS use. For the year 2014/15, they reported 191 people in Scotland receiving treatment for Mephedrone use and a further 636 receiving treatment for use of 'other' drugs. This gives us a total of 827 users in treatment. Using the multiplier, this would generate an estimate of 1284 NPS users in Scotland (827 * 1.56).

However, given the limited nature of the treatment data referring to NPS users, and comparing this estimate with the two injecting estimates calculated using the Neo data, we would question the robustness of this Scottish estimate as an under-estimate. As a result, this figure is not reported as a finding in the main report.

C. Interview Participant Demographics

	Pseudonym	Location (based on NHS board)	Gender	Age	MSM	PWID	Young Person	MH Service User	Homeless Person
1	Michael	Ayrshire & Arran	M	26				X
2	Gary	Greater Glasgow & Clyde	M	39				X	X
3	Kieran	Lothian	M	46				X
4	Steven	Ayrshire & Arran	M	39				X	X
5	Debbie	Ayrshire & Arran	F	22					X
6	Alistair	Ayrshire & Arran	M	28				X
7	Nick	Ayrshire & Arran	M	36					X
8	Tracey	Ayrshire & Arran	F	26		X		X
9	John	Ayrshire & Arran	M	26		X		X
10	James	Greater Glasgow & Clyde	M	19		X			X
11	Daniel	Greater Glasgow & Clyde	M	17		X			X
12	Peter	Greater Glasgow & Clyde	M	17		X			X
13	Jacob	Lanarkshire	M	49				X
14	Paula	Greater Glasgow & Clyde	F	32				X	X
15	Andrea	Fife	F	34		X		X
16	Moira	Fife	F	39		X
17	Jessica	Fife	F	35		X		X
18	Tiffany	Tayside	F	28				X
19	Nicola	Tayside	F	32		X
20	Claire	Tayside	F	30		X		X	X
21	Christina	Tayside	F	39		X
22	Luke	Grampian	M	20			X
23	Samuel	Dumfries & Galloway	M	19			X		X
24	William	Lothian	M	55	X			X
25	Hugh	Greater Glasgow & Clyde	M	61	X	X
26	Thomas	Lothian	M	45	X	X
27	Graeme	Lothian	M	44	X	X		X
28	Kevin	Lothian	M	35		X
29	Kimberly	Lothian	F	32		X		X
30	Alexander	Greater Glasgow & Clyde	M	42	X
31	Michelle	Highland	F	16			X
32	Chloe	Highland	F	17			X
33	Colin	Lanarkshire	M	27				X

D. Qualitative Data Collection: Topic guides

Interview Topic Guide

Background Information Age: Ethnicity: Gender: Location: What I want to do now is focus on the first time you used a legal high... What was it? PROMPT: powder, name on packet, effects told it would have What did you know about it before taking it? PROMPT: source, e.g. online, friends PROMPT: What effects were you expecting from it? PROMPT: did you know what was an active dose? Where did you get it from? Why that source? How did you take it? PROMPT: swallowed, snorted, dabbed, smoked, injected? Why that method? How much did you take? PROMPT: all/portion? Why? Where were you and who were you with? What happened? PROMPT: Using anything else (inc. alcohol) before/during/after? PROMPT: Effects of drug on body (ability to distinguish effects?) PROMPT: Effects of drug on behaviour (ability to distinguish effects?) What led up to you trying it that first time? PROMPT: peers, trauma, curiosity, legality e.t.c. PROMPT: motives about a legal high, or that legal high Had you tried any illegal drugs before you tried it? If yes, was it like any other drug you have tried? If no, did it prompt you to try any illegal drugs? If so, what? What did you particularly enjoy about it that first time? Was there anything you didn't like? PROMPT: During PROMPT: After What did you do to help you manage that? Have you been to see anyone for any help with the effects? PROMPTS: A&E, GP, CAT, NSP, friends, family, dealer, internet Have you tried it since then? Why/not? Have you tried any other legal highs since then? GENERATE LIST OF LEGAL HIGHS TRIED How often would you say you take legal highs? Do you take with other substances (inc. alcohol)? And now I want to focus on the last time you tried a legal high What was it? Why did you take it? What did you know about it before taking it? PROMPT: source, e.g. friends, internet PROMPT: did you know what was an active dose? Where did you get it from? How did you take it? PROMPT: swallowed, snorted, dabbed, smoked, injected? How much did you take? Where were you and who were you with? What happened? PROMPT: Using another substances before/during/after? PROMPT: Effects of drug body/behaviour (ability to distinguish effects?) What did you particularly enjoy about it this last time? Was there anything you didn't like? PROMPT: During PROMPT: After PROMPT: longer term PROMPT FOR ALL: physical, mental, relationships, hobbies, job, £ What did you do to help you manage that? Have you been to see anyone for any help with the effects? PROMPT: A&E, GP, CAT, NSP, friends, family, dealer, internet Ending Questions What help would you have liked to have been available to you that wasn't? If you could give one piece of advice to someone about to try a legal high for the first time, what would it be?

Focus Group Topic Guide

Background Information Role: Type of service: Time in Role: Geographical location: Prevalence How common do you think NPS use is among your service users? PROMPT: particular groups? PROMPT: age ranges? Gender? PROMPT: changes over time? PROMPT: specific substances/drug categories? How related to certain groups? where use? Motives How do you know service users use NPS? PROMPT: deliberate disclosure? Accidental? What reasons do they give for use? PROMPT: story of a recent client (anonymised) Have any tried NPS and then not carried on? Why not? Have any active drug users you are working with never tried NPS? Why not? Have any transitioned on to NPS and moved away from traditional drugs? Why? In general terms, how have your service users' patterns of use changed over the last few years? PROMPT: quantity, route of admin, type of NPS used, polydrug use What is different about motivations for NPS compared to controlled drugs? Harms What, if any, negative effects have you seen as a result of NPS use? PROMPT: harm to self, close others (fam/friends), wider community? PROMPT: What do you think has caused this? PROMPT: legal harms (e.g. using a substance that was legal but that has since been controlled?) PROMPT: have you had experience of users becoming dependent on NPS? PROMPT: quantity, route of admin, type of NPS used, polydrug use Has NPS use had an impact on reducing the harms caused by illegal drugs? If yes, what ways? How do you support NPS-using service users? What resources would make supporting NPS-using service users easier? Ending If you could give one piece of advice to someone (e.g. friend, rather than service user) contemplating NPS use, what would it be?

E. Focus Group demographics

NHS Board	N	%
Ayrshire & Arran	1	2%
Borders	1	2%
Dumfries & Galloway	0	-
Fife	1	2%
Forth Valley	1	2%
Grampian	9	21%
Greater Glasgow & Clyde	11	26%
Highland	4	10%
Lanarkshire	2	5%
Lothian	8	19%
Orkney	0	-
Shetland	0	-
Tayside	2	5%
Western Isles	0	-
National Role	2	5%
Total of participants	42	100%

F. Online Surveys

Links to PDF versions of the surveys used in this study are below.

It should be noted that question logic was built in to survey questions, so certain questions would only be displayed if participants answered yes to an initial question e.g. Have you injected NPS, would then bring up a set of NPS injecting questions.

NPS Survey

Online at: https://issuu.com/scottishdrugsforum/docs/nps_user_survey

NPS Survey Draw

Online at: https://issuu.com/scottishdrugsforum/docs/nps_prize_draw

Staff survey

Online at: https://issuu.com/scottishdrugsforum/docs/nps_survey_staff