You are here

Regulation impact statement: Codeine re-scheduling

Version 1.1, December 2016

20 December 2016

Book pagination

Regulatory and health economic impact models

A regulatory costing model and an economic and social impact model (health economic model) have been developed by an independent consultancy (KPMG). Each model was informed by a range of data sources, including public submissions, scientific literature, information provided by other government agencies, as well as industry and peak body consultations.

To determine and test the assumptions of the regulatory impact and economic impact models, data were sourced from the following relevant business units in the Department of Health:

  • Pharmaceutical Benefits Scheme (PBS). Potential increased costs due to additional prescription of PBS listed medicines in the case of up-scheduling to Schedule 4.
  • Medicare Benefits Schedule (MBS). Increased costs due to additional medical practitioner visits to obtain prescriptions in the case of up-scheduling to Schedule 4.

Primary data sources are referenced throughout, and include:

  • Australian Register of Therapeutic Goods (ARTG, August 2016): product name, formulation, sponsor (company) and manufacturer details of all therapeutic goods lawfully supplied in Australia
  • IMS Health (June 2013): product sales data
  • Pharmaceutical Benefits Scheme (PBS, August 2016): details of the medicines subsidised by the Australian Government
  • Medicare Benefits Schedule (MBS, August 2016): details of the Medicare services subsidised by the Australian Government
  • Public submissions to the TGA (May 2015 – January 2016): from interested members of the public, individual specialists (i.e. pharmacists and medical practitioners), pharmaceutical companies, representative / peak bodies, and other agencies[77]
  • Confidential submissions to the TGA (May 2015 – January 2016, supplied by TGA): submissions as above, but not permitted for public release
  • Interviews (August 2016): see 'Overview of consultation activities' (p. 95) for more detail
  • MedsASSIST (PGA, 10 August 2016): the real-time recording and monitoring program established by the PGA to support patient safety and improve the clinical outcomes for medicines containing codeine
  • Published scientific literature concerning health effects of long-term codeine use, and the comparative effectiveness of pain relief compared to Paracetamol and ibuprofen.

Regulatory impact model

The development of the regulatory model was undertaken in accordance with the Office of Best Practice Regulation (OBPR) Guidance.[78] As outlined in this framework, regulatory costs were estimated for administrative compliance costs and substantive compliance costs only. Delay costs (application and approval delays) were determined to be out of the scope of the model as it was envisaged that any changes to scheduling for codeine products would incorporate sufficient time for industry to respond without experiencing any stock-outs (which occur when existing pharmaceutical stock is withdrawn or exhausted prior to new stock being available).

The model identified the key regulatory compliance processes that would arise from the implementation of each option. Component elements of each of the identified regulatory processes were then broken down into their respective time, cost and frequency components and the value of the respective inputs sourced from previous regulation impact statements prepared by the TGA, as well as information provided via consultations with industry and peak bodies. Cost models were developed in Microsoft Excel, with the summary for each option also presented in the standard Regulatory Burden Measure (RBM) format.

Inputs and assumptions

The development of the regulatory cost estimates was informed by targeted consultation with sponsors who currently produce codeine-based products in the OTC market. The target sponsors were Sandoz Pty Ltd, Sanofi-Aventis Australia Pty Ltd, GlaxoSmithKline Consumer Healthcare Pty Ltd, Soul Pattinson Manufacturing Pty Ltd and Johnson & Johnson Pacific. These companies occupy both distinct and overlapping segments of the OTC and prescription market and were able to provide a range of perspectives given the different incentives and risks that are intrinsic to their business models.

Sponsors were provided with a list of questions (Appendix A) which were subsequently used to structure conversations in meetings and focus on issues directly related to the modelling and implementation implication of the different options.

Broadly, the interviews were structured around six topics outlined in the questionnaire:

  • product strategy
  • market response
  • labelling
  • packaging
  • updated listing and regulatory approvals
  • implementation.

The baseline assumptions used to support the regulatory model are reported in Table 4.

Business-as-usual (BAU) variations to existing medicines

There is high variability between how often sponsors change an aspect of their product (e.g. update label, Product Information (PI) etc.). Some sponsors vary their ARTG entry regularly (even more than once a year), whereas other sponsors will not vary their products for several years. The majority of ARTG variation applications are for prescription products.

  • A 2014 survey of industry revealed that companies will update their labels as part of BAU, on average, every 3 years.[79] Therefore, with an assumed 18 month implementation timeframe which also aligns with the RASML implementation cycle, it is assumed that at least half of the affected sponsors will have the opportunity to roll the labelling changes into already scheduled updates.
  • Consultations with sponsors and manufacturers have identified the cost of implementing a minor labelling change ranges between $2,000 and $6,500 per product. This incorporates the costs of artwork and internal processes to quality assure and implement the change on the production line. For the purposes of the costing, drawing from the TGA RIS on General Requirements for Labels for Medicines, the average cost to implement a labelling change (per OTC product) is estimated at $4,171.[80] We have assumed that this incorporates the aspects of multiple labels per product.
  • These minor label change costs include pre-production costs (such as label redesign and approval, artwork and proofing) and production costs (new printing plates for conventional printing processes, changes to the digital printing process). The costs also cover any potential changes to the Product Information (PI)/ Consumer Medicines Information (CMI).
  • A minor label change is defined as a small change to the phrasing of text on a label that does not necessitate a change to, or rearrangement of, other label graphics.
Regulatory costs associated with any transition
  • The compliance costs for stock recalls has not been estimated as the assumed implementation timeframe should provide adequate time for turn-over of stock across the supply chain. PGA have indicated that the codeine products have a short shelf life (i.e. rapid turnover) within pharmacies[81] due to fast sales associated with these products.
  • A default FTE wage rate of $37.40 per hour and an on-cost multiplier of 1.75 have been adopted to account for non-wage labour on-costs as per OBPR guidance.[82] This results in a scaled up rate of $65.45 per hour.
  • An individual's time has been costed at $29.00 per hour as per OBPR guidance. A pharmacists wage has been estimated at $35.90 per hour[83] (therefore $62.83 per hour when the on-cost multiplier of 1.75 is applied). A doctor's wage has been estimated at $97 per hour[84] (therefore $169.75 per hour when the on-cost multiplier of 1.75 is applied).
  • The amendment of current Schedule 4 and Schedule 8 entries in the Poisons Standard will have administrative Government impact only (legislative change) but will not have any impact from a regulatory cost perspective on businesses, community organisations or individuals.

Baseline assumptions

Number of codeine products entered on the ARTG (as at August 2016)

Table 4 details the current distribution of codeine products across Schedule 2, Schedule 3, Schedule 4 and Schedule 8.[85]

  • Seventy-three (73) entries on the ARTG are for codeine products that currently have Annual Charge Exemption (ACE) status. This means they have $0 turnover and are not being actively marketed in Australia. These entries have been excluded from Table 4.
  • As some ARTG entries cover more than one medicine unit (e.g. different pack sizes), a multiplier has been applied to ARTG entries for each of the following types of medicines:
    • Prescription (Schedule 4) – 2.3 medicines per ARTG entry; and
    • OTC (Schedule 2 and Schedule 3) – 2.5 medicines per ARTG entry.[86]
  • The result of applying this multiplier is shown as the 'adjusted' value in brackets in Table 4.
  • The ARTG data indicates the majority of products with codeine as an active ingredient within the Schedule 3 category. Furthermore, an analysis of sponsors with a product presence in multiple categories finds that:
    • all 15 sponsors with Schedule 2 products also have products in the Schedule 3 category;
    • of the 15 sponsors with Schedule 2 products, three also have products in the Schedule 4 and/or Schedule 8 categories;
    • of the 22 sponsors with Schedule 3 products, five also have products in the Schedule 4 and/or Schedule 8 categories; and
    • No sponsors have products across all four categories.
  • For many medicines, there is more than one label associated with a product. For example, a medicine in a blister pack is assumed to be associated with two labels (the backing of the blister pack and the outside carton). Based on an analysis of ARTG entries, a multiplier is applied to the number of medicine products to estimate the number of associated labels:
    • Prescription – 1.89 labels per medicine product; and
    • OTC – 1.85 labels per medicine product.[87]
Table 4: Number of products containing codeine phosphate [actual and (adjusted)] listed in the ARTG by Schedule
Number of products [actual (adjusted)] [a]
Sponsor Schedule 2 Schedule 3 Schedule 4 Schedule 8 No. of ARTG entries
Alphapharm Pty Ltd 1(2.5) 5(12.5) 1(2.3) 7 (17.3)
Amneal Pharma Australia Pty Ltd 3(7.5) 3 (7.5)
Apotex Pty Ltd 9(22.5) 30(75) 39 (97.5)
Arrow Pharma Pty Ltd 2(5) 7(17.5) 1 10 (23.5)
Aspen Pharma Pty Ltd 14(35) 4(9.2) 1 19 (45.2)
Aspen Pharmacare Australia Pty Ltd 1(2.3) 1 (2.3)
Aurobindo Pharma Australia Pty Ltd 4(10) 4 (10)
Bayer Australia Ltd 2(5) 1(2.5) 3 (7.5)
Biotech Pharmaceuticals Pty Ltd 1(2.5) 3(7.5) 1 5 (11)
Care Pharmaceuticals Pty Ltd 2(5) 2 (5)
Cipla Australia Pty Ltd 2(5) 12(30) 14 (35)
Generic Health Pty Ltd 2(5) 7(17.5) 9 (22.5)
GlaxoSmithKline Consumer Healthcare Australia Pty Ltd 2(5) 7(17.5) 9 (22.5)
Johnson & Johnson Pacific Pty Ltd 5(12.5) 1(2.5) 6 (15)
Orion Laboratories Pty Ltd T/A Perrigo Australia 1(2.5) 5(12.5) 6 (15)
Pharmacare Laboratories Pty Ltd 5(12.5) 12(30) 17 (42.5)
Pharmacor Pty Ltd 6(15) 6 (15)
Phebra Pty Ltd 1 1
Reckitt Benckiser Pty Ltd 3(7.5) 3 (7.5)
Sandoz Pty Ltd 2(5) 3(7.5) 5 (12.5)
Sanofi-Aventis Australia Pty Ltd 8(20) 4(9.2) 12 (29.2)
Sigma Company Limited 4(10) 11(27.5) 15 (37.5)
Soul Pattinson Manufacturing Pty Ltd 4(10) 10(25) 14 (35)
Symbion Pty Ltd 4(10) 14(35) 18 (45)
Total ARTG entries [actual (adjusted)] 46 (115) 168 (420) 10 (23) 4 (4) 228 (562)
Actual ARTG entries affected (%) 20 74 4 2
No. of sponsors affected (n=24) 15 22 4 4
Sponsors affected (%) 63 92 17 17

a Note not all columns will sum due to rounding; Source: ARTG extract 1 August 2016.

Health economic impact model

This part of the document provides an overview of the health economic model. More information is available in the KPMG Report (Annex 1) which provides clear illustrations of the potential sources of bias in the analysis and how these were resolved.

How the health economic model works

The health economic model works in five main steps, as follows:

Step 1: Simulation. The simulation is based on holding constant the total amount of codeine currently used by each of the five patient/consumers groups (see 'Five groups of consumers') (though possibly in different pack sizes) and then adjusting the variables including the following to reflect the post-regulatory change environment:

  • packs of low-dose codeine
  • out of pocket cost
  • pharmacy purchases
  • GP consultations.

Step 2: Plausibility analysis. The next step assesses, for each consumer group, the plausibility that this change in consumer level resource use will occur (holding the consumption of low-dose codeine constant) when considered relative to alternative pathways. Each of these alternative pathways has different uses of resources and health outcomes associated with them. The proposed regulatory change does not add pathways; it removes one and hence changes which of the existing pathways that patients will take. These alternative pathways will often incorporate discussions with both GPs and pharmacists, and include options such as:

  • use of an OTC analgesic (e.g. paracetamol and/or ibuprofen) without codeine
  • non-pharmacotherapy
  • prescribed low or high dose codeine
  • alternative prescribed pharmacotherapy
  • GP or self-referrals to allied healthcare providers
  • GP referrals to specialist pain clinics.

Step 3. Allocation of population across pathways. This step entails identifying the proportion of consumers in each group (cohort) that will pursue each pathway. This proportion varies across groups and also depends on factors such as the capacity for the existing pain clinics to see additional patients, including those who are referred by their GP and are eligible for a Medicare rebate.

Step 4. Determination of resource use and health outcomes. This step assigns changes in resource use to each cohort, where these changes in resource use and health outcomes are a consequence of changes in behaviour.

Step 5. Project changes over time. The final step projects these changes over the specified ten-year period, taking into account changes in cohorts, the ongoing needs for additional services and the longer term impact on health outcomes.

Sources of complexity – interdependency and data paucity

The first source of complexity in developing this model relates to the interdependency between the responses by the three main stakeholder groups for this model (consumers, GPs and pharmacists) to the proposed regulatory change. In summary, potential actions by these groups are:

  • Consumers will need to decide whether they go to their GP, consider the treatment options provided by the GP, and/or use OTC non-codeine analgesics such as paracetamol and/or ibuprofen
  • GPs will have a new patient group for whom they can suggest a range of additional options (appropriate for the patient), including referring patients to specialist pain units or allied healthcare providers; providing scripts of low or high dose codeine or other prescription only medicines; and/or suggesting OTC pain relief medications
  • Pharmacists will continue to interact with consumers who are seeking pain relief, such as those who require prescription medications, some of which will be new to the consumer, and consumers continuing to attend the pharmacy seeking pain relief without any prescriptions for medications.

The second source of complexity was the paucity of data that could inform the estimate of the key benefit of the proposed regulatory change; that being the proportion of current users of OTC low-dose codeine will experience a health gain, including preventing deaths, as a consequence of this change. No data was available to inform the following estimates:

  • the number of people who are currently dependent on low-dose codeine (note high dose specifically excluded)
  • the number of adverse events attributable to low-dose codeine (excluding high dose codeine)
  • the number of people who use low-dose codeine chronically and, while currently not dependent, are at risk of dependence.

The critical issue in developing the economic model is the combined effect of these two complexities (interdependency between responses and paucity of estimate data). Unless there are changes in treatment (the cumulative result of changes in activity and/or decisions made by the three groups identified above) there will be very limited changes in health outcomes. The proposed regulatory change in itself will not produce the expected benefits; rather, it will be the changes in people's activity and behaviour that realises the benefits.

The first way in which complexity described above is accommodated within the simulation model is by the use of five separate consumer groups, rather than relying on the concept of an 'average' consumer. This segmentation of the population allows the plausibility of responses to the proposed re-scheduling to be assessed more meaningfully and also reduces sources of unintentional systematic error that result from working with a model based on the characteristics of an 'average' consumer. However, this approach does increase the number of required assumptions, an important consideration in this data-poor area. A possible consequence of these additional assumptions is that a very wide range for any resultant metric, such as a net cost or benefit, emerges as a consequence of sensitivity analyses that use multiple variables. The model incorporates a set of analyses that identify the most plausible range of each parameter and interdependencies between parameter, before conducting the sensitivity analyses, hence addressing this risk.

The second way in which these complexities accommodated in the simulation model is its conservative approach to estimating the size of the projected benefits of the proposed regulatory change (See 'Sources of uncertainty' below ). This approach entails ensuring that the base case parameters for estimating these benefits is conservative, and that the key mechanisms by which this gain is achieved, as articulated by stakeholders, is clearly mapped to these explained. Potential losses in health outcomes to some consumers were also identified by stakeholders are also incorporated into the model.

Sources of uncertainty

In general, simulation modelling has two main types of uncertainty, those associated with model structure and those associated with model inputs. Further information regarding each uncertainty is described in the Annex 1 of this report (KPMG report – Annex E).

In this model the main drivers of input uncertainty are related to the potential health benefits. These are:

  • the proportion of consumers that could potentially benefit from changing their pathway
  • the proportion of consumers who will change their pathway in an optimal way
  • the extent of additional health benefits that will be achieved if consumers with the potential to benefit change their pathway and receive improved treatment
  • the time period that the health benefit is maintained without the need for additional investments in treatment and therapy.

As a general principle, the economic model sought to take conservative approaches to the estimation of health gains.

The main reason for this conservative approach was that health gains were the primary driver of benefits and also the most contested benefit (there was wide disagreement across various stakeholders); and for either case (low or high benefits), there was minimal supporting data and evidence.

As a general principle, the economic model sought to take conservative approaches to the estimation of health gains, for two reasons.

The first reason for taking a conservative approach was that the health benefits arising from the improved therapeutic pathways taken by patients who would otherwise be chronic users of low-dose codeine were the primary driver of benefits in the model, but are likely to be the most contested benefit. There was wide disagreement across the range of stakeholders as to the proportion of current chronic or acute users who would benefit from using different therapeutic pathways, and the extent of this benefit. Stakeholders who were supportive of the up-scheduling highlighted the benefits in terms of patients who would have an improved diagnosis of chronic or acute pain and also a shift to high dose codeine to reduce the risks related to paracetamol and/or ibuprofen use. These stakeholders also referred to the evidence from the Cochrane[88] systematic reviews regarding the evidence of limited effectiveness of low-doses of codeine compared to paracetamol and/or ibuprofen alone. However, no stakeholder provided supporting evidence or data, other than the Cochrane reviews, and instead gave specific examples. Stakeholders in favour of the scheduling proposal gave specific examples of patients who suffered from migraines and were prescribed sumatriptan rather than codeine medications as a result of GP advice, or alternatively patients that suffered knee pain were referred to weight-loss clinics to better manage their specific symptoms. Stakeholders who were not supportive of the up-scheduling gave examples of patients (>65 years of age) who could not tolerate NSAIDs such as ibuprofen and would now have no options available to them.

The second reason for taking a conservative approach was that the protocol[89] to be applied to the valuation of a statistical life year is the use of an 'unconstrained' willingness to pay $182,000. This statistical life year is intended to be a year at full health and therefore is also the value of a Quality Adjusted Life Year (QALY). The unconstrained willingness to pay approach is the preferred option under this protocol; however, it results in a much higher value being placed on a QALY compared to that used in health economics and health technology assessment in Australia and internationally.[90]

Avoided deaths were an additional benefit. It was not possible to develop an accurate estimate of deaths that could be prevented as a consequence of Option 6. A published study provided an estimate of the annual deaths attributable specifically to OTC codeine medicines. However, the authors indicated that when deaths involving OTC codeine medicine abuse occurred, it was likely that there were multiple influencing factors and changed access to OTC codeine medicines would not necessarily prevent these deaths. [10] The base case of the model assumed conservatively that 5 deaths would be prevented and this assumption was varied in the sensitivity analysis (see 'Sensitivity analyses' p. 61).

Additionally, the only OTC low-dose codeine medicine specific data that was available to support the model (IMS data for sales of codeine-based products) was for 24 months to September 2013). An extrapolation of the IMS sales data was used to estimate the number of sales in 2017 (base year). Some externally sourced inputs (such as categorisation and fee structure for MBS GP consultations and the discount rate) were not specific to the options being modelled but rather relate to the broader health system. Initial assumptions were formed for the remaining inputs and then, in the absence of data or relevant literature, they were tested during interviews with peak bodies (i.e. expert opinion was sought).

Five main considerations for economic model

Listed below are five key considerations that the model addresses and which collectively reduce the systematic bias that may arise from a simpler 'average-consumer' model. These considerations are:

  • Why is the concept of an 'average' consumer potentially misleading?
  • At what point does using an 'average' response to the proposed regulatory change produce a biased estimate of the increase in GP visits?
  • What are the five consumer groups and what are their differences?
  • What are the drivers of the model's benefits and costs? and
  • How are the model's key parameters determined?
Is there an average consumer of codeine?

To test the extent of any systematic bias that could emerge if the economic model were based on an average consumer rather than differentiated groups of consumers, a "pre-modelling" exercise using hypothetical data was constructed. Table 5 shows the average use of codeine per customer, which was derived for hypothetical sales data associated with products in the market (Table 4). Considering a market with 14 million sales and 1.1 million consumers, the average use per consumer is 12.7 packs per year. From interviews with stakeholders, there appears to be at least two broad types of consumers. First, acute users, who were assumed for this pre-modelling exercise to use an average of two packs per year. Second, chronic users who were assumed for this exercise to use an average of 120 packs per year (two to three packs a week). When these patterns of use are combined with the total number of consumers and total volume used, then it is possible to solve for the proportion of users in each groups. In this hypothetical case, chronic users represent 9% of the market and 86% of the sales.

Table 5: Consumer share of market and share of sales by two hypothetical consumer groups
Low use consumers High use consumers Total use
Number of consumers 1,000,000 100,000 1,100,000
Packs per consumer per year 2 120 12.7
Packs per year 2,000,000 12,000,000 14,000,000
Share of consumers 91% 9% 100%
Share of packs 14% 86% 100%

Source KPMG report Table E1

The implications for the economic modelling are as follows:

  • Historical information and experience suggests that a different behavioural response in the high use consumer compared to the low use consumer when regulatory changes are made that place tighter restrictions on availability. If schedule 2 and 3 medicines were up-scheduled to schedule 4 requiring all codeine-containing medicines to be prescribed by a medical practitioner only, then following implications are likely:
    • The demand for the medicines by the low use consumer may reduce use if required to take time to see a GP and possibly the incur the additional cost of the co-payment.
    • The high use group may be willing to visit a GP 12 times a year, but they are far less likely to make 50 visits per year.
  • Large changes in the behaviour of a small group of consumers, the 9% with high use, will lead to large changes in overall demand.
  • Health benefits are most likely to accrue in the smaller group of high use consumers; although a minority low use consumers may also gain some health benefits due to exploration of alternatives in therapy or medicines to codeine.
  • Adverse health outcomes are still considered likely, even though average use is only 12.7 packs a year; there are consumers who are using substantially more packs per year and could develop dependence over time.
Five groups of consumers

The five different groups of consumers are set out in Figure 2 and are stratified according to type of use, type of pain and level of dependency.

Figure 2

Figure 2: Patient/consumer groups used in the health economic modelling

(Source: KPMG Report: Figure E1)

Codeine use can be classified as either therapeutic or non-therapeutic. The classifications and the associated characteristics used in the economic model were derived from stakeholder consultations.

Therapeutic use can be stratified into chronic or acute use. Consumers that self-treat acute conditions are likely to non-dependent. These consumers might purchase only one or two packs per year and use it in a way consistent with the medical advice: i.e. for no more than 3 consecutive days without medical advice with no more than eight tablets a day.[91] Consumers that suffer chronic pain conditions could be one of two types: non-dependent or dependent. Some chronic users are non-dependent and use only up to the maximum daily dose but use at this level for most days in a year. Frequent use can lead to dependence and indeed consumers that self-treat chronic pain with codeine are at greater risk of becoming dependent in comparison the consumers that self-treat acute pain conditions. In the economic model, dependent chronic consumers use more than the recommended maximum dose each day, whereas non-dependent consumers use the maximum recommended dose each day. Both types of chronic users use the low-dose codeine for the majority of days in a year, which in the base case is assumed to be 250 days> they are also more likely to consume the largest pack size (40 tablets) compared to acute users. Consequently, a regulatory change that restricts pack size to three days' supply will impact this group of consumers more so than acute users.

Chronic therapeutic use is more likely to result in adverse events such as gastrointestinal bleeds. They are also more likely to have an accidental death due to an overdose of codeine, although this is considered to be a very rare event.

While non-therapeutic use is often referred to in the media[92], it is likely to be only a very small share of consumers in this group. Nonetheless, a conservative approach was used and this group was included in the total volume of sales. However, the model did not account for any benefits that might result from the proposed up-scheduling for this group. It is likely that use of low-dose codeine medicines will be limited substantially if they are required to go to a GP to obtain a script.

How many additional GP visits?

One approach to calculating the number of additional GP visits is to start with the estimated total current volume of packs sold and make an assumption regarding the rate at which consumers will go to a GP to obtain a prescription for this medicine, for example, on 40% of occasions. However this approach is based on an "average consumer approach" which, as discussed previously, can lead to a systematically biased estimate of the post-regulatory change if there are distinct types of consumers with different current use and different potential responses.

Some reasonable assumptions about the proposed regulatory changes that were made are:

  • There will be additional GP consultations associated with ongoing utilisation of the current Schedule 3 codeine-containing medicines and these are considered separately from the additional consultations regarding different treatment pathways'.
  • The GP will be able to provide up to five repeats (based on the level of repeats associated with current Schedule 4 codeine-containing medicine, and the option to use private scripts).
  • Not all trips to the GP will be solely for the purpose of obtaining a prescription for codeine as some patients will request a script as part of a consultation that would otherwise have occurred.
  • Different consumers will respond differently to the up-scheduling of codeine and to restrictions on the availability of codeine-containing medicines, depending upon whether they have acute or chronic pain.
  • GPs will also respond to the actions of their patients. I.e. if their patients come every two weeks for an additional supply then they may refer them to a service such as a pain clinic.[93]
  • There is a limit to the number of additional consultations a consumer can have in a month, simply due to the financial and opportunity cost of attending a GP clinic.

By incorporating these reasonable assumptions in base case, the projected volume of additional GP visits in the base case is substantially lower than it would otherwise have been (refer to Annex 1, KPMG report, Table E3 for further information on the disaggregated approach to determine additional GP consultations.)

The number of additional GP visits is significantly lower than predicted by alternative models[94] due to two key differences; the use of a segregated population (see Figure 2 ) rather than an average consumer to predict additional consultations and the assumption that five repeats are available. The former brings a number of other factors into consideration, e.g., the additional constraints of the maximum additional visits any consumer could have in a week.

The Macquarie University Centre for the Health Economy (MUCHE) reported the results of a study on the value of OTC medicines.[95] The report stated that their survey found that, in the case of analgesics/pain relievers, when asked what they would do if the medicines for their condition became unavailable over the counter: 63% of respondents said they would see their doctor, 24% said they would do nothing, 15% said they would use a home remedy and 7% said they would 'supplement'. More options were explored in this economic model, due to the segregated population. If codeine was re-scheduled to schedule 4, 48% of all current users would use OTC paracetamol and/or ibuprofen medicines. Another 12% were assumed to visit their GP who would advise them to use OTC analgesics and/or non-pharmaceutical pain management options.

A report prepared for the Pharmacy Guild by Cadence[96] projected that there would be an additional 8.7 million GP visits as a consequence of up-scheduling low-dose codeine combination medicines at a cost of $316.44 million as a result of patients attending a doctor for these scripts. The Cadence model assumes that 53% of all current users will continue if low-dose codeine medicines if they were up-scheduled and the additional GP costs correspond to one consultation per prescription. Targeted stakeholder consultations indicated that patients can have up to five repeat scripts (as is currently the case for prescription codeine products on private prescription), that is, a ratio of up to six scripts per GP visit, reducing the figure of $316.44 million to as low as $52 million.

Sensitivity analysis associated with GP visits

In the current economic model, the additional number of GP visits will vary between consumer subgroups (Figure 2), as different cohorts will consume different amounts of low-codeine medicines (Table 6).

Table 6: Type of current consumers and their pattern of use for Schedule 3 medicines only
Therapeutic use Non-therapeutic use
Number of consumers Acute Chronic dependent Chronic not dependent Chronic Acute Total
Total baseline use by type of consumer As % of all users 80.0% 3.8% 15.2% 0.1% 0.9% 100%
Number of packs 4,624,637 3,347,226 12,595,103 214,340 63,421 20,844,727
As % of all packs sold 22.2% 16.1% 60.4% 1.0% 0.3% 100%
Total baseline expenditure Total expenditure ($M) $23.1 $31.3 $105.5 $2.0 $0.3 $151.7

Of note:

  • Around 6% of the Australian population over the age of 12 (19,874,413 at June 2015) purchase at least one pack of low-dose codeine per year. Given that there will be an estimated 20.8 million packs sold in 2017 and that many consumers purchase more than one pack, this result of 6% of the population purchasing at least one pack has face validity.
  • 80 per cent of all consumers are 'acute' users; however, as a group they purchase 22.2% of all packs.
  • 19% of all consumers are chronic therapeutic users and they purchase 76.5% of all packs sold.

For acute therapeutic users of low-dose codeine medicines (assumed to be 20% population and consume 5 packs/year), the up-scheduling of codeine will result in greater costs to the consumer and the government due to increased GP visits required for this group. However, increasing this population from 20% to 25% (acute therapeutic users with 50% increase in number of GP visits) results in approximately 0.3% increase in the net health benefit. That is, the model was insensitive to a change in the GP visits associated with acute users of low-dose codeine medicines.

Similarly, changing the percentage of consumers who use low-dose codeine medicines therapeutically, chronically and are dependent from 20% to 50%, and thus would require a GP visit to source codeine did not produce a significant change in the net health benefit (1% increase). Despite the higher treatment costs associated with chronic users that are dependent, due to the likely referrals to pain clinics, no significant change in net health benefit was noted when the percentage of users were changed. That is, the economic model was insensitive to the increase in number GP visits for this group.

Changing the percentage of consumers who use low-dose codeine medicines therapeutically, chronically but are non-dependent from 10% to 1%, reduced the net health benefit (-3.5% decrease). Even acknowledging that these patients are likely to visit the GP on a number of occasions, the economic model was relatively insensitive to the increase in number of GP visits for this group, despite this cohort consuming the greater number of medicine packs (Table 6).

The above sensitivity analysis indicates that additional GP visits associated with a up-scheduling of codeine does not significantly change the net benefits predicted from the economic model (that is GP visits is not a key input for this model). The key inputs were QALYs, the number of prescription repeats and the number of deaths prevented. Further sensitivity analysis is provide in the following pages of this RIS.

What are the health benefits?

The identified health benefits are of the following types (see KPMG report Annex E):

  • prevention of accidental death;
  • improved quality of life in the event that the patient is exposed to and benefits from, alternate pain treatment options that the patient would have not have otherwise used;
  • For consumers taking combination therapy, potential for prevention of adverse events related to unintentional overdose of paracetamol or ibuprofen; and
  • reduced dependence and risk of dependency.

If low-dose codeine is only available by prescription, then the health gains compared to the existing situation (low-dose codeine OTC) are driven by changes in treatment and therapy that result from changes in the availability of low-dose codeine OTC as an option. This change may arise as a result of the patient discussing alternative treatment options with their pharmacist or doctor (Figure 3). These alternative treatment options are available currently. The removal of the option of low-dose codeine OTC results in consumers exploring other options; it does not result in other options being introduced.

Figure 3

Figure 3: Potential patient actions and therapy options in response to pain

(Source: KPMG report Figure E2)

Consultation with a GP may result in the GP:

  • assisting the patient in maintaining their current codeine usage by prescribing low/high dose codeine medicines;
  • recommending the use of paracetamol or ibuprofen (or combinations) without codeine;
  • recommending non-pharmacotherapy (such as exercise, physiotherapy etc.);
  • prescribing alternative pharmacotherapy;
  • referring the patient to allied health providers; and/or
  • referring the patient to specialist pain clinics.

Furthermore, as a result of the patient visiting their GP, the cause of the patient's pain may be subsequently diagnosed and treated, ultimately reducing the patient's need for analgesics. When consumers are referred to allied healthcare professionals or specialist pain clinics by their GPs, then in many situations, patients will be able to obtain a Medicare rebate for the service. However, there are currently waiting lists for pain clinic services, so capacity constraints must also be considered.

Minor changes in consumer behaviour – is this a health benefit?

As discussed previously, when Schedule 2 codeine-containing analgesics were re-scheduled to Schedule 3 in 2010, no significant public health benefit was reported ('Historical scheduling of codeine' p. 33). While there are likely to be some minor changes in consumer purchasing behavior with reduction of pack sizes; these behaviors are unlikely to have any significant public health benefits (Refer to 'What are the health benefits?' for further details on p. 57). Health benefits as shown in the model will only be realized from improved therapeutic pathways taken by patients who would otherwise be chronic therapeutic users of low-dose codeine.[97] For this improved therapeutic pathway to be realized the key enabler is a visit to a GP. However, this is unlikely to occur under options that retain OTC codeine-containing medicines, regardless of pack size or whether up-scheduled from Schedule 2 to Schedule 3.

The MedsASSIST program was not included in the modelling. However, the PGA has suggested that a combination in a reduction in pack size together with the use of mandated MedsASSIST program (which contains referral pathway modules that the program does not currently have [PGA, personal communication 2016]) could also lead to improved therapeutic pathways for consumers. While theoretically possible, there appear to be a number of operational limitations that are presented by this option. These limitations include:

  • The increased instances of 'pharmacy shopping' to source codeine.[98] This is inconsistent with the limited ability of pharmacists to actively engage with 'challenging' patients to manage the use of codeine in OTC medicines, noting that the pharmacy environment does not usually allow for private conversations in the way that doctors' rooms do (as outlined on p. 39 'Real-time monitoring programs');
  • States and Territories will need to agree to support MedsASSIST with mandatory reporting. Such changes require uniform adoption at the jurisdiction level and changes to relevant jurisdiction legislation, which would take time;
  • Absence of a pharmacist-initiated patient referral system to pain management specialists or clinics for greater oversight and intervention (as is currently available for GPs);
  • Pharmacist do not have the patients' full medical history, and thus are not best placed to advise on other therapeutic pathways that may be more beneficial, such as non-pharmacological or alternative pharmacological treatments;
  • A recording and monitoring system is inconsistent in principle with scheduling of OTC medicines (Schedules 2 and Schedule 3);
  • The voluntary MedsASSIST program, implemented by 65% of pharmacies, reports data that shows that approximately 2% of transactions have been denied, although about 6% sales have been made under duress in pharmacies (following a real or perceived threat to the safety of pharmacy staff);
  • In the July and September reporting periods, approximately 20% of consumers purchased codeine products between 4 - 5 times and approximately 10% purchased products more than 5 times within a 7-month period; and
  • In one particular example, an individual received 660 tablets over 3.5-month period despite their purchasing behaviour being tracked in MedsASSIST. This brings into doubt whether MedsASSIST is actually deterring access to codeine by consumers with codeine-dependence problems. The data does suggest that consumers with addiction problems will change their behaviour to source their codeine.

Noting the above issues, a key assumption in the model is that significant health benefits will be realized only from improved therapeutic pathways taken by patients after consultation with GPs as compared to continued chronic use of low-dose codeine products (see Figure 5, p. 87).[99] For further details refer to 'What are the health benefits?' (p. 57). Other key assumptions and their base case values underpinning the economic analysis are:

  • 99% of people who used low-dose codeine at least once are using it for therapeutic purposes with the remainder of consumers using it for non-therapeutic purposes.
    • Of all users, 80% are using it therapeutically for acute conditions
    • Of the 19% using it therapeutically for chronic conditions, 20% are dependent on low-dose codeine (i.e. of all users, 3.8% are dependent on low-dose codeine)
  • For cough and cold products (currently Schedule 2): if up-scheduled to Schedule 4 then the prescribing behavior of GPs will be to provide zero repeats. As a liberal estimate, 20% of consumers are estimated to continue to use cough and cold products containing codeine and it is estimated that only 30% of these patients will make an additional visit to the GP to source these products compared to the GP visits they would have made anyway in the absence of the regulatory change.
  • For Schedule 3 low-dose codeine-containing products: if up-scheduled to Schedule 4 then the prescribing behavior of GPs will be to provide up to 5 repeats (through authorized prescribing processes) if they assess that it is appropriate for this patient to continue with this pharmacotherapy, given the patient's symptoms and medical history.
  • Projected number of prevented deaths from up-scheduling of codeine to Schedule 4 is 5 per year.
  • Pharmaceutical companies are unlikely to seek PBAC approval for PBS listing of low-dose codeine; however, consumers are currently using low-dose codeine without a PBS subsidy and its price is currently below the general beneficiary co-payment. Prescribers of low-dose codeine can use a 'private script' (i.e. non PBS) with up to 5 repeats compared to PBS option with zero repeats. Therefore, it is unlikely that concessional consumers will choose a low-dose codeine PBS option with zero repeats, in order to access the PBS subsidy should one be available (which in itself is unlikely, noting a 30 mg codeine medicine is currently available through PBS). There is no financial advantage (no PBS subsidy) for a general beneficiary (non-concessional).
  • Substitute low-dose codeine analgesics with paracetamol and/or ibuprofen products, that are cheaper than existing codeine products and readily available at supermarkets (which introduce further price competition).[100]
What are the health costs?

For consumers and the Government the main additional costs relate to:

  • net out-of-pocket costs to consumer (extra time spent to visit doctor);
  • additional costs to MBS due to additional GP (exploration of alternative treatment pathways or to obtain access to low-dose codeine medicines) and referral to pain management clinics; and
  • additional costs to the PBS due to additional scripts for PBS listed pain medications (alternatives to low-dose codeine medicines).

The additional costs to the MBS are the primary driver of additional costs to government (refer to Table ES8 of KPMG report).

Key modelling parameters – current use pattern

Four steps are taken to determine the key modelling parameters given available data (sales data from IMS), reasonable assumptions about the number of tablets consumed per day, days of use for each of the five types of consumers, and the share of total consumers in each group (Figure 4). These four steps are outlined below, with the inputs to steps 1, 2 and 3 having been informed by discussions with stakeholders:

  • The total number of sales by pack size (dollar value and packs sold) was calculated from the IMS data. The total retail sales value by pack size was calculated by assuming a 44 per cent retail mark up.[101]
  • 6-20 tablets per day over 12-365 days per year depending on the type of consumer.
    1. a) Therapeutic, acute pain, non-dependent – 8 tablets/day over 12 days/year
    2. b) Therapeutic, chronic pain, non-dependent – 8 tablets/day over 250 days/year
    3. c) Therapeutic, chronic pain, dependent – 12 tablets/day over 250 days/year
    4. d) Non-therapeutic, occasional use, non-dependent –6 tablets/day over 20 days/year
    5. e) Non-therapeutic, regular use, dependent –20 tablets/day over 365 days/year
  • The proportion of consumers in each group
    1. a) Therapeutic – 99%, of whom:
      1. a. Therapeutic, acute pain, non-dependent – 80% of all users
      2. b. Therapeutic, chronic pain – 19%% of all users of whom:
        1. 1. 40% are therapeutic, chronic pain, non-dependent
        2. 2. 60% are therapeutic, chronic pain, dependent
    2. b) Non-therapeutic – 1%, of whom:
      1. a. 10% are non-therapeutic, regular use, dependent
      2. b. 90% are non-therapeutic, occasional use, non-dependent
  • The pack sales were then allocated by the model across consumers, starting with the smaller packs allocated to the lower use consumers (20 tablets), and with the maximum pack sizes (40 tablets) allocated last to the most frequent users.

This approach enabled the average pack size and expenditure per consumer group to be calculated. In turn, this allowed the assessment of the plausibility of changes to current behaviour based on a more accurate picture of the current pattern of use.

Figure 4

Figure 4 : Determining key modelling parameters

Sensitivity analyses

Sensitivity analyses were performed to test the robustness of the key summary statistic, the net benefit, to the model's inputs (refer to Annex 1, KPMG Report Table E4 for further details).

The sensitivity analysis indicated that a positive net benefit is maintained under all plausible assumptions regarding the possible values of the model's inputs. The net benefit remained positive even when the following (highly unlikely) values for inputs were assumed:

  • Reducing the benefit
    • No deaths are prevented (compared to 5 per year)
    • The average QALY benefit per treated patient decreases by 80%  
  • Increasing the costs
    • There are no repeats for low-dose codeine (increases the number of GP consultations)
    • The cost per patient who is treated increases by 80 %

Therefore, it can be concluded that the model's key statistic, the net benefit, is very likely to be positive.


The overall results for regulatory costs, economic costs and benefits are presented in Table 20 (refer to 'Net benefit for each scenario' p. 89). The main result is the net benefit, which is presented using two subset results for the first year and tenth year net benefit. These yearly results are further subdivided into: (1) first year benefit, cost and net benefit; and (2) the ten year NPV benefit, cost and net benefit. Option 6 is the only option with a net benefit hence all other options have a net economic cost, driven by additional out-of-pocket costs to consumers as well as the cost attributed to the administrative burden for businesses and individuals.

In many scenarios there was a net reduction in out of pocket costs to the consumer. When this occurs, the model adjusts the resultant saving to be included as a benefit, not a negative cost. This saving to consumers is the result of a combination of factors, including:

  • the reduction in use of low-dose codeine
  • the substitution of low-dose codeine with cheaper supermarket products such as paracetamol or ibuprofen
  • patients who continue with prescription medicines (whether containing codeine or not) in most cases will pay the same or less than their current expenditure on low-dose OTC codeine medicines
  • if patients substitute low-dose codeine medicine with high dose prescription codeine medicine via script, they can be provided with up to five repeats by their GP, thus reducing the need for visits to their GP
  • the high bulk-billing rate for GP consultations
  • the rate at which pain-related GP consultations can be accommodated within visits that would otherwise have occurred in the absence of the proposed regulatory change.

One potential issue raised by the modelling is that the predicted demand for additional consultations at pain clinics is unlikely to be accommodated within existing capacity.

Most stakeholders indicated that additional face-to-face education[102] for prescribers and pharmacists was unlikely to be necessary. With an estimated one million people using at least one Schedule 3 low-dose codeine product a year, the need to invest in an education and awareness campaign, particularly for consumers, is apparent. This issue was raised with stakeholders. The question of how an education campaign for consumers would be funded, and what form it might take, is still to be determined and is dependent on the regulatory process changes, if any. The cost of this campaign was not included in these estimates.

Book pagination