Cost-benefit analysis

Several economic evaluation techniques can contribute to calculating ROI but the one selected for this study was a cost-benefit analysis (CBA). A CBA is the basis of an encompassing ROI analysis and is suggested as an appropriate technique for understanding the returns on research investment using FAIT [22]. It takes a broader societal perspective of the range of benefits captured and reported, both costs and consequences are reported in monetary values, and the reportable metric of CBA is a ratio of benefit per dollar of cost [23]. The single ratio form of reporting ROI allows direct comparison with ROI calculated for other programmes and can also be calculated using simulations or projections; both reasons for the selection of a CBA over other techniques for this study. These projections can then be compared to subsequent recalculations using actual data as the VALID BP programme progresses and delivers outputs. All monetary values used refer to the Australian dollar unless otherwise specified.

Inclusions and exclusions

The CBA included:

The costs of VALID BP’s initial research and ongoing translational activities

The costs associated with validating BPMDs and training healthcare practitioners in proper BP measurement technique

The benefits of improved diagnostic accuracy for hypertension across the Australian healthcare system including cost savings from elimination of unnecessary medication use and prevention of disease sequalae and averted losses to economic productivity through better BP measurement and management.

The CBA excluded:

The costs of replacing nonvalidated BPMDs

Consumer price index (CPI) increases (inflation) in future research and implementation costs

The potential price differences between validated and non-validated devices

The cost of treatment (mainly medications) for previously undiagnosed patients

Program Logic Model

To inform the costs and benefits for the CBA, a retrospective program logic model for the entire VALID BP Project was developed collaboratively by the project lead (JS) and the independent subject-matter assessors (ZD and SR). Project documentation and publications were used as references. The program logic model maps the need for the VALID BP project through to its anticipated impacts, helping to determine the costs (activities and outputs) and benefits (achieved and aspirational impacts) for the VALID BP project (Fig. 1).

Fig. 1

Retrospective Program Logic Model for the VALID BP Project.

Measuring costs

The broad cost categories for the VALID BP project included research-related labour, research-related non-labour costs to cover a range of research and translational activities as previously described, and the cost of implementing a comprehensive BPMD validation programme and training health providers on the proper use of BPMDs. Research-related costs were based on historic recollection of actual time expended and other expenses relating to the VALID BP project till end 2020.

Research costs Labour

Contributions for all the key VALID BP investigators, collaborators, and students were costed using a micro-costing methodology. Although activities were global, separating the costs by country was impossible, so the costs for all researcher time (globally) was included in the model, even though the benefits being monetised only relate to Australia. The approximate hours of researcher time (2014–2020) and the approximate wage level (converted to Australian academic wages for all positions) was used to determine the value of research labour. All researcher time for the VALID BP project was covered by the host institutions in conducting their specific tasks and not by specific grant income related to the project. Academic wages were costed using the Australian National University Academic Staff Salary Schedule[24]. Additional costs of 30% to cover superannuation contributions, payroll tax, workers compensation, and leave loading [25] and overheads of 27% to cover the in-kind contribution from the various institutions where team members were based, were added to the wage costs. These overheads covered utilities, infrastructure, and research support such as access to information technology and services provided by academic institutions. The historical research hours for the VALID BP project were allocated to systematic reviews, writing of manuscripts, development of the online BP validation course and other resources, media and communications, project meetings, training sessions and advocacy activities. Table 1 contains details of all historical labour costs.

Table 1 Research labour costs.

Labour assumptions:

1.

Considering this work is ongoing, it is assumed that future research costs will be required to realise the benefits of VALID BP. Given that most of the historical research costs were expended between 2018 and 2020 (3-year period) it is assumed that the equivalent of an additional $141,052 per year in labour costs would have to be invested to maintain the momentum from this work and allow time for the realisation of 100% validation of BPMDs in Australia by 2028.

2.

It is also assumed that it will take 8 more years to achieve universal validation in Australia. For each additional year, the research costs are expected to increase proportionally based on the consumer price index (CPI), but for the purposes of this preliminary model, these increases have been excluded.

Non-labour

Two grants totalling $48,680.62 were awarded to the VALID BP project between 2018 and 2020 and these were used to cover non-labour expenses, including travel costs (valued at $19,420), incidentals, and publishing costs.

These costs do not account for the VALID BP project work after 2020 so projections to 2028 have added an additional $48,680 per year (inflated for the purposes of remaining conservative) in non-labour costs to achieve the goal of 100% validation. This brings the project research costs up to $189,733 per year for 8 years.

Implementation costs BPMD validation

The costs of the scientifically recommended method for validating BPMDs involve a strict research protocol which include a sample of 85 people and a cross-section of patients with different arm sizes, BPs, and gender balance [5]. The estimated cost for this method of validation was obtained from a US study published in the Journal of Clinical Hypertension [26]. The highest estimated costs were used, converted into Australian dollars and adjusted to 2022 values using the GDP deflator taken from the Australian National Accounts [27]. The total cost of a validation study was estimated at $59,340 per device model. Some manufacturers have already validated their main devices, so the burden on individual manufacturers will not be evenly spread, with some manufacturers being affected more significantly than others. Due to a lack of publicly available data on industry costs, sales, and profits of BPMDs and for the purposes of this protocol, it was assumed that the total costs to the industry for the change to full validation are included in the estimated validation cost per model. An additional 10% was added to cover any potential regulatory costs applied by the Therapeutics Goods Administration (TGA) for checking and approving device validation.

Regulatory change

A key translation pathway to enact effective change in the use of clinically validated BPMDs is through regulation and legislation. This would involve making BPMD validation part of the regulatory and legal requirement for registration of such devices for sale in Australia. It was unfeasible to micro-cost this activity, so existing data was used to estimate the potential cost, recognising that the journey to regulatory and legislative change is unique in every instance and unlikely to be identical in cost. A Health Research Council of New Zealand study on the cost to governments of enacting public health legislation (and deemed to be “generally applicable to other developed countries”) found that the average cost of a new regulation in New Zealand from the proposal stage through every step in the legislature (converted to Australian dollars at 2022 values) is $526 816 (95% UI: 386,602 to 1,163,500) [28] For this analysis, we have assumed that the cost in Australia is similar given the two countries have very similar constitutional and governance systems. Due to a lack of data and potential variability, we have assumed that the cost of lobbying for legislation is covered under the ongoing research costs we have projected to 2028. This legislation cost was only accounted for once in the modelled projections, in 2025, when it is feasible that advocacy efforts may have created sufficient traction for this new legislation to be enacted. This would allow a further 3 years to achieve 100% validation of BPMDs in Australia and phase out nonvalidated BPMDs already in circulation.

Regulatory infrastructure

The Australian TGA operates through a cost recovery framework associated with the registration and listing of medicines and medical devices. The TGA currently lists all BPMDs irrespective of whether they are validated or not. This framework generates the bulk of the agency’s funding through fees and charges paid by the manufacturers of products that are requesting approval[29]. The assumption was made that any increases in costs to the TGA or the Australian government associated with a requirement to approve only clinically validated BPMDs, would be passed on to the manufacturer and already included in the addition to validation costs for these manufacturers.

Costs to consumers

It is likely that BPMD manufacturers will pass on the costs of BPMD validation to Australian users, with these costs being borne by practices and consumers. However, these costs remain in the initial validation cost per model plus 10% estimate for industry and given the societal perspective, are not included again to avoid double-counting. In Australia, BPMDs (upper arm or wrist cuff devices) vary in price from $53.00 to over $630.00, with the more expensive devices generally targeted at the clinical facilities rather than personal use. Unfortunately, there is no data on the number of these devices currently on the market and potential phasing-out pathway for nonvalidated devices (compulsory versus optional and subsidised versus un-subsidised). Without this data, the costs of replacing all nonvalidated BPMDs cannot be accurately projected and were therefore omitted from the current analysis.

Training costs

To maximise the benefit from BPMD validation in Australia, healthcare team members need to be educated and upskilled on proper BP measurement technique. The online certified training course developed as part of the VALID BP project activities [17] takes roughly an hour to complete (including the test) and certification can be updated every six months. We have used the Australian average weekly earnings of $1713.90 per week or $55.70 per hour, including 30% additional costs and allocated two hours to complete the certification yearly. Around 565,753 health practitioners (excluding dental practitioners) were registered with the Australian Health Practitioner Regulation Agency in 2018. From 2013 to 2018, there was a 16.3% increase in the number of health practitioners, which equates to an average increase of 3.26% per year. We applied these labour costs to an estimate of the cohort requiring training in each subsequent year.

Scenario analysis

To estimate the downstream benefits of improved accuracy from BPMDs, we modelled and projected costs and benefits based on currently available and comparable data sources. Three different impact scenarios were tested, based on a projected trajectory for the VALID BP project from 2020 to 2028. These scenarios are all aspirational and based on estimates.

● Scenario A: A modest increase in the percentage of validated BPMDs caused by modest impact with policymakers, physicians, pharmacists, and consumers. Increase in the rate of validated BPMDs from 14.5% to 20% by 2021.

●Scenario B: Validation of all BPMDs carried by major Australian pharmacy chains due to work with the Pharmacy Guild. Validation of 60% of all BPMDs by 2025 when it is assumed that a new regulation requiring universal validation of all BPMDs on the Australian market would come into play.

●Scenario C: Validation of 100% of devices in Australia by 2028 including replacement of all non-validated devices currently in use.

Together, these scenarios create a form of sensitivity analysis that provides a projected minimal impact, a projected mid-point, and a maximum possible impact occurring at 100% validation, given the likelihood that there will still be nonvalidated devices in use despite the best efforts by government and industry to phase them out. It is also important to note that any changes to the accuracy of BPMDs will not rest solely with this project and that only a proportion of the benefits can be attributed to the VALID BP project.

Modelling benefitsChanges to hypertension prevalence based on testing

From interviews with healthcare providers in Australia and official Australian population data [30] we derived an informed estimate of how many people over the age of 18 present for a GP consult and have a BP test every year. At baseline, this is ~18.5 million. We used a decision tree model to vary the number of people who would be under- and over-diagnosed at the different levels of device validation, and the estimated societal benefit from improving validation rates based on a reduction in unnecessary medication use and savings from avoided downstream health care costs and reduced productivity.

Currently, only 14.5% of clinically relevant upper arm and wrist cuff BPMDs are validated out of a total of 312–446 models sold in Australia. Based on the results of Akpolat et al., we also assumed that only 15% of non-validated BP devices provide sufficiently accurate measurements compared with more rigorously validated devices [10].

Assumptions underpinning the benefit analysis:

● If models are not within 4 mmHg, then they are under or overestimating BP by at least 5 mmHg [5, 6].

● In Australia, 85% of all nonvalidated models and 32% of all validated models are measuring blood pressure with an error of 5 mmHg or greater [14].

● Based on available data, the rate of overestimation (56%) is slightly higher than the rate of underestimation (44%)[31].

● All 440 models have an equal market share

● That the proportion of validated and non-validated devices currently in use is the same in both home and clinical settings

● That BPMD devices used in both home and clinical settings have an equal contribution to the overall burden of misdiagnosis

● The total number of models will remain constant with no new models entering the market between 2020 and 2028.

Supplementary file 1 details how the number of Australians being over and under-diagnosed were calculated based on the projected proportion of validated devices.

Medication costs due to overdiagnosis

In Australia, the annual cost of a common daily anti-hypertensive medication, Lisinopril, is listed as a subsidised medication on the Pharmaceutical Benefits Scheme[32]. The cost to consumers of the lowest dose (5 mg) assumed to be the most likely dose for someone who has normal BP (<140/90 mmHg) and is over diagnosed by up to 5 is $20.15 per pack of 30 tablets. The cost to the Government for subsidising the medication is $14.44 per pack of 30 tablets which is $175.69 per year and the cost to the patient for 365 tablets is $245.16 per year totalling $420.85 per year to both the Australian government and patients. Based on population health survey data, the assumption is that only 35% of patients diagnosed with hypertension take daily anti-hypertensive medication[33]. The assumption is also made that once commenced, patients would take these medications for life.

Health system savings by reducing missed diagnoses (underdiagnosis)

In 2009, it was estimated that hypertension and its consequences generated direct costs to the Australian health care system (medications, hospitalisations, emergency care etc.) of $1.8 billion over the life cycle of a given cohort, which translates to $2.27 billion in 2020 dollars (adjusted for inflation) [1]. To reduce ambiguity in the concept of a “life cycle cohort” that is found in the data, a simplified assumption was made that a cohort spanned 64 years (from 18 to 82— average life expectancy of an Australian). This means that with a highly conservative estimate, hypertension and related consequences in Australia costs the health system approximately $35,468,750 per year [34].

A Mexican study estimated that 37% of the healthcare cost of hypertension is spent in direct treatment and 63% on treatment of complications and sequalae such as myocardial infarctions and stroke [35]. When applied to our example, $22.3 million of the costs per year can be attributed to downstream costs from untreated hypertension, as a highly conservative estimate. We know from a life table monitoring study that in Australia, 61.4% of people with hypertension are left untreated [36]. Data from the AIHW confirms that 5.9 million Australians have hypertension[1]. Of these, 3.7 million (61.4%) are untreated costing the health system an average of $6.08 per person in downstream costs. This is the figure we have used to calculate the downstream healthcare costs for the cohort with undiagnosed hypertension for whom treatment is delayed.

Productivity costs due to hypertension

Over the working lifetime of a given Australian population cohort (ages 20–69), hypertension causes an estimated loss in economic productivity of $137.2 billion by reducing people’s capacity for work, while optimal treatment of all patients would save $76.4 billion over the working lifetime of the cohort [36]. These estimates were generated in 2009, so the savings adjusted to 2022 dollars would equal $96.15 billion over the working lifetime of the cohort. Given the working lifetime is for ~49 years from age 20–69, that equates to a saving of $1.96 billion annually. The potential lost productivity from underdiagnosis of hypertension equates to $556 million gross domestic product (GDP) or $736 per person per year. A key assumption is that of those who receive an accurate diagnosis and are identified as being hypertensive, only 35% are adequately treated and controlled. This reduces the potential savings by 65%.