Supplementary material

Appendix Table 1. Naloxone distribution model parameters.

Input name Base case (range) Justification Source

Markov model annual transition parameters for UK adaptations

Heroin use to nonoverdose

death (in excess of

background mortality)

0.0075 (0.0025 to

0.0125)

Heroin users are at excess risk for death in addition to overdose. [1]

Heroin use to overdose

First overdose

Second overdose

Subsequent

overdoses

0.09 (0.02 to 0.12)

0.22 (0.05 to 0.30)

0.34 (0.27 to 0.60)

Findings from international studies in North America and Australia demonstrate a high rate of

overdose among young heroin users, that less than 75% of users ever overdose, and that there is a

higher risk for overdose among those who have overdosed previously. A UK study demonstrated

23% of heroin users had ever overdosed and that the mean number of overdoses was 3.6 [2].

[1]

1

Input name Base case (range) Justification Source

Annual relative reduction in

risk for first overdose

0.933 (0.900 to

1.000)

An additional approach used to calibrate the model to extant Australian epidemiologic findings was

to decrease the likelihood of a first-time overdose such that for those who had never overdosed, risk

for first-time overdose was half after 10 years of use; this does not affect the likelihood of repeat

overdoses. The parameter was exponentiated by Coffin and Sullivan to the years elapsed and the

result multiplied by its reference parameter to reduce the likelihood of the event over time.

[1]

Heroin use to discontinuation

of heroin use

0.06 (0.01 to 0.10) The rate of discontinuing heroin use was taken by Coffin and Sullivan from large prospective North

American cohorts.

[1]

Discontinuation of heroin use

to heroin use

0.070 (0.056 to

0.084)

The risk for relapse to heroin use was based on a US study showing 50% relapse over 5 years, with

an age-based reduction in risk for relapse such that relapse was half as likely after 10 years,

resulting in a median duration of heroin use of 15 years.

[1]

Annual relative reduction in

risk for relapse

0.933 (0.900 to

1.000)

This parameter acted as the age-based reduction in the risk for relapse. The parameter was

exponentiated to the preceding number of model cycles and the result multiplied by its reference

parameter to reduce the likelihood of the event over time.

[1]

2

Input name Base case (range) Justification Source

Overdose to discontinuation

of heroin use

0.062 (0.028 to

0.113)

A United States study, examining rates of discontinuing drug use after overdose, found that 26% of

injection drug users sought substance-abuse treatment within 30 days of an overdose event, 75% of

whom enrolled, suggesting an increase in discontinuation of heroin use related to the overdose

event. Coffin and Sullivan thus assumed a relative 10% increase in the likelihood of discontinuing

heroin use after an overdose, ranging on sensitivity analysis from half to double the rate of

discontinuation if there was not an overdose.

[1]

Decision-tree parameters (proportions) used in the UK adaptations

Joint probability that

distributed naloxone is used

each year

0.172 This parameter is calculated from parameters in the next five rows.

Proportion of heroin users

prescribed naloxone

0.30 (0.05 to 0.60) In Scotland (where a publicly-funded naloxone take-home policy is in place) the target is for one

third coverage of injecting drug users, we therefore assumed a distribution of naloxone that reached

30% of heroin users.

Assumption

based on

Reference 3

3

Input name Base case (range) Justification Source

Proportion of overdoses

witnessed

0.85 (0.32 to 0.94) Coffin estimated a base case of 85% of overdoses being witnessed based on United States and

Australian evidence [1]. A UK study demonstrated that 81% of those in the community and not in

treatment and 85% of those in treatment had a witnessed overdose [4]. The lower limit was based

on a US study that showed a lower rate of witnessed overdose (32%) [5] and the upper limit was a

10% increase on the base case.

[1,4,5]

4

Input name Base case (range) Justification Source

Proportion of witnessed

overdoses where naloxone is

available (i.e. hasn’t been lost

or not with the victim at the

time of the overdose)

0.75 (0.40 to 0.85) Evidence from Scotland’s national take-home naloxone program suggests that in 2014/2015 11% of

all naloxone kits supplied was because the previous kit supplied had been lost (739 kits issued out

of 6498 were due to previous kits lost) [6] and from the Welsh national take-home naloxone

program 12% of kits supplied in 2013/2014 were due to lost kits (538 resupplied of which 41%

were due to lost kits – 221 kits out of a total supply of 1802) [7]. Furthermore, a UK prospective

cohort study showed that only 18.5% of heroin users prescribed naloxone actually carry it with

them [8]. Based on a review of evidence and likely real-life scenarios, the N-ALIVE trial being

conducted in the United Kingdom has assumed that only 75% of current or former heroin injectors

on release from prison would still have their naloxone after 4 weeks [9]. Based on these data we

assume that 75% of heroin users who have been prescribed naloxone would be in possession of it at

an overdose.

Assumption

based on

References

[6–9]

5

Input name Base case (range) Justification Source

Proportion of witnesses with

naloxone who attempt to use it

0.90 (0.77 to 0.99) A UK survey of heroin users showed that 89% would use naloxone if they witnessed an overdose

and it had been available [4]. A UK prospective cohort study showed that 77% of heroin users

would administer naloxone in an overdose situation (after training in overdose management and

naloxone provision this figure rose to 99%) [8]. The evaluation of the Welsh naloxone take-home

program found that 88% of drug users and friends and family would be willing to administer

naloxone pre-training and 99% post training [10]. Based on these data we assumed 90% of

witnesses would attempt to use naloxone.

Assumption

based on

References

[4,8,10]

Social network modifier (used

in sensitivity analysis)

1.0 (0.5 to 1.5) Coffin and Sullivan used this assumption to reflect other data points. The point estimate for this

parameter relies on the other parameters, and the range allows for a broader range of error in the

related variables (that is, if heroin users in a given community are more or less likely to use in

groups, this variable would be higher or lower, respectively).

[1]

Proportion who call an ambulance

Proportion who call an

ambulance (no naloxone use)

0.60 (0.30 to 0.80) The evaluation of the Welsh naloxone take-home program found that in the absence of naloxone

distribution an ambulance was called in approximately 60% of overdoses [10].

[10]

6

Input name Base case (range) Justification Source

and then go to accident and

emergency

1.0 (0.50 to 1.0) Table 3.69 of the ‘Overdose and Poisoning’ chapter of the Joint Royal Colleges Ambulance Liaison

Committee United Kingdom Ambulance Services Clinical Practice Guidelines 2016 states that

‘Transfer to further care’ should occur for ‘All patients suffering an opioid overdose whether or not

they have responded to naloxone – the effects of respiratory depression opioid overdose can last 4-5

hours’ [11].

[11]

Proportion who call an

ambulance call after naloxone

use

0.85 (0.55 to 0.95) The evaluation of the Welsh naloxone take-home program found that after naloxone administration

an ambulance was called in approximately 85% of overdoses [10].

[10]

and then go to accident and

emergency

1.0 (0.50 to 1.0) Table 3.69 of the ‘Overdose and Poisoning’ chapter of the Joint Royal Colleges Ambulance Liaison

Committee United Kingdom Ambulance Services Clinical Practice Guidelines 2016 states that

‘Transfer to further care’ should occur for ‘All patients suffering an opioid overdose whether or not

they have responded to naloxone – the effects of respiratory depression opioid overdose can last 4-5

hours’ [11].

[11]

Proportion who survive overdose

7

Input name Base case (range) Justification Source

Proportion who survive an

unwitnessed overdose

0.90 (0.80 to 0.94) Based on the evidence that 90% of overdoses result in survival without emergency medical care or

administration of naloxone by nonmedical first responders.

[13]

Reduction in survival for

second overdose

0.015 (0.000 to

0.020)

To account for the decrease in survival for subsequent overdoses. [1]

Reduction in survival for

subsequent overdoses

0.015 (0.000 to

0.020)

To account for the decrease in survival for subsequent overdoses. [1]

Proportion who survived a

witnessed overdose (no

naloxone or ambulance)

0.918 (0.800 to

0.940)

To account for the increase in survival when an overdose is witnessed the model set survival to

91.8% for a witnessed overdose where no ambulance was called or naloxone used [1]. This was in

line with UK studies suggesting a fatality rate of witnessed overdose of 7% [3].

[1]

Relative increase in survival

during a witnessed overdose

when an ambulance was

called (no naloxone)

1.013 (0.980 to

1.035)

This accounts for the relative increase in survival from the baseline of surviving an unwitnessed

overdose (i.e. 90%). The absolute value is 0.930, in line with UK studies suggesting a fatality rate

of witnessed overdose of 7% [3].

Assumption

based on

Reference

[3]

8

Input name Base case (range) Justification Source

Relative increase in survival

during a witnessed overdose

with naloxone

1.067 (1.035 to

1.089)

This accounts for the relative increase in survival from the baseline of surviving an unwitnessed

overdose (i.e. 90%). The absolute value is 0.980. Studies in the United Kingdom and Germany

suggest that 95% (21 out of 22) [14] to 100% [8,12] of cases where naloxone is used result in

survival. This is in line with assumptions made by Coffin and Sullivan.

Assumption

based on

References

[1,8,12,14]

UK model costs

IM naloxone (1 mg/mL, 2 mL

prefilled syringe of

Prenoxad®)

£15.30 is the price

for one

Kit costs incurred after each overdose where naloxone was administered and biannually among

active heroin user to account for naloxone going out of date and for losses.

Prenoxad used as it is licensed for use in the community.

[15]

Distribution £8.50 Distribution into drug services. Coffin and Sullivan assumed in addition to drug costs were other

component costs ($3) and staff time and distribution ($10).

[1]

Training costs for users,

family and friends

£124 per kit for

first-time

administration

Training the individuals (one to one about 10 to 30 mins): Drug services – community (per-care

contact).

[16]

9

Input name Base case (range) Justification Source

Ambulance £233 Cost of see, treat, convey. [17]

Accident and emergency visit £278 VB03Z: emergency medicine, category 3 investigation with category 1–3 treatment. [17]

Utility values

Heroin user 0.80 (0.73 to 0.90) Estimates from Coffin and Sullivan derived from a United States paper involving clients at a

midwestern United States substance-abuse treatment-intake centre, which included a 6-month

follow-up survey that may have captured those not actively seeking treatment and documented a

utility of 0.80 by SF-6D methodology and a 6.5% improvement for those who achieved interim

abstinence.

[1]

Relative increase in utility for

heroin user in recovery

1.07 (1.00 to 1.13) To be consistent and conservative, Coffin and Sullivan used the 6.5% improvement in utility from

the study outlined above, although the SF-6D methodology has been found to be poorly responsive

to changes in drug use and a study of opioid maintenance therapy found a relative improvement of

16% in utility for a heroin user. Utility was coded to never exceed 1.0 when values were set to

upper extremes simultaneously.

[1]

SF-6D, short form-6 domains

10

References

[1] Coffin PO, Sullivan SD. Cost-effectiveness of distributing naloxone to heroin users for lay overdose

reversal. Ann Intern Med 2013;158:1–9.

[2] Gossop M, Griffiths P, PowisB, etal. Frequency of non-fatal heroin overdose: survey of heroin users

recruited in non-clinical settings. BMJ 1996;313:402.

[3] Bird SM, Parmar MKB, Strang J. Take-home naloxone to prevent fatalities from opiate-overdose:

protocol for Scotland’s public health policy evaluation, and a new measure to assess impact. Drugs

2015;22:66–76.

[4] Strang J, Powis B, Best D. Preventing opiate overdose fatalities with take-home naloxone: pre-launch

study of possible impact and acceptability. Addiction 1999;94:199–204.

[5] Davidson PJ, McLean RL, Kral AH, et al. Fatal heroin-related overdose in San Francisco, 1997–2000:

a case for targeted intervention. J Urban Health 2003;80:261–73.

[6] National Health Service Scotland Information Services Division. National Naloxone Programme

Scotland—Monitoring Report 2014/15. Edinburgh, Scotland: Scotland Information Services Division,

2015.

[7] Public Health Wales. Take home naloxone 2013–14. 2014. Available from:

http://www2.nphs.wales.nhs.uk:8080/SubstanceMisuseDocs.nsf/Public/

ECF9E1D7909A8ABC80257E0A0059E887/$file/Naloxone%20report%202014%20FINAL.pdf?

OpenElement. [Accessed August 25, 2016].

[8] Strang J, Manning V, Mayet S, et al. Overdose training and take-home naloxone for opiate users:

prospective cohort study of impact on knowledge and attitudes and subsequent management of overdoses.

Addiction 2008;103:1648–57.

[9] Strang J, Bird SM, Parmar MKB. Take-home emergency naloxone to prevent heroin overdose deaths

after prison release: rationale and practicalities for the N-ALIVE randomized trial. J Urban Health

2013;90:983–96.

11

[10] Bennett T, Holloway K. The impact of take-home naloxone distribution and training on opiate

overdose knowledge and response: an evaluation of the THN Project in Wales. Drugs (Abingdon Engl)

2012;19:320–8.

[11] Joint Royal Colleges Ambulance Liaison Committee, Association of Ambulance Chief Executives.

UK Ambulance Services Clinical Practice Guidelines. London: Joint Royal Colleges Ambulance Liaison

Committee, 2016.

[12] Dettmer K, Saunders B, Strang J. Take home naloxone and the prevention of deaths from opiate

overdose: two pilot schemes. BMJ 2001;322:895–6.

[13] Coffin PO, Sullivan SD. Cost-effectiveness of distributing naloxone to heroin users for lay overdose

reversal in Russian cities. J Med Econ 2013;16:1051–60.

[14] Bennett T, Holloway K. Evaluation of the Take Home Naloxone Demonstration Project. Glamorgan,

Wales: University of Glamorgan, 2011.

[15] British Medical Association, Royal Pharmaceutical Society. British National Formulary (BNF) 67.

London: BMJ Group and Pharmaceutical Press, 2016.

[16] Curtis L. Unit Costs of Health and Social Care 2015. Kent, UK: Personal Social Services Research

Unit, University of Kent, 2015.

[17] Department of Health. NHS reference costs 2014 to 2015. 2015.Available from:

https://www.gov.uk/government/publications/nhs-reference-costs-2014-to-2015. [Accessed August 25,

2016].

12