Unit costs for non-acute care in Ireland 2016—2019

Background: This paper presents detailed unit costs for 16 healthcare professionals in community-based non-acute services in Ireland for the years 2016—2019. Unit costs are important data inputs for assessments of health service performance and value for money. Internationally, while some countries have an established database of unit costs for healthcare, there is need for a more coordinated approach to calculating healthcare unit costs. In Ireland, detailed cost analysis of acute care is undertaken by the Healthcare Pricing Office but to date there has been no central database of unit costs for community-based non-acute healthcare services. Methods: Unit costs for publicly employed allied healthcare professionals, Public Health Nurses and Health Care Assistant staff are calculated using a bottom-up micro-costing approach, drawing on methods outlined by the Personal Social Services Research Unit in the UK, and on available Irish and international costing guidelines. Data on salaries, working hours and other parameters are drawn from secondary datasets available from Department of Health, Health Service Executive and other public sources. Unit costs for public and private General Practitioner, dental, and long-term residential care (LTRC) are estimated drawing on available administrative and survey data. Results: The unit costs for the publicly employed non-acute healthcare professionals have changed by 2–6% over the timeframe 2016–2019 while larger percentage changes are observed in the unit costs for public GP visits and public LTRC (14-15%). Conclusions: The costs presented here are a first step towards establishing a central database of unit costs for non-acute healthcare services in Ireland. The database will help ensure consistency across Irish health costing studies and facilitate cross-study and cross-country comparisons. Future work will be required to update and expand on the range of services covered and to incorporate new data and methodological developments in cost estimation as they become available.


Overview
This paper presents a set of unit costs for non-acute care services in Ireland for the years 2016-2019 1 .
A unit cost refers to the value of resources used to produce a single good or service (Creese & Parker, 1994). For example, in healthcare, a unit cost can refer to the cost of a general practitioner (GP) visit, an outpatient appointment, a laboratory test, an episode of inpatient hospital care, a week in a nursing home etc. (Conteh & Walker, 2004;Curtis & Burns, 2019). Unit costs are important data inputs for assessments of health service performance and value for money (Curtis & Burns, 2019).
Internationally, while some countries have an established database of unit costs for healthcare (e.g., the UK), the need for a more coordinated approach to calculating healthcare unit costs has been recently highlighted (Mayer et al., 2020a). In Ireland, detailed cost analysis of acute care is undertaken by the Healthcare Pricing Office (HPO) but there is no central database of unit costs for community-based non-acute healthcare services (Whyte et al., 2018). This paper seeks to address this gap and presents unit costs for 16 healthcare professionals including publicly employed allied healthcare professionals, public health nurses, and a range of health care assistant staff, as well as GPs, dentists, and public and private long-term residential care (LTRC) for the years 2016-2019. It is important to acknowledge at the outset that this is a first step towards establishing a central database of unit costs for non-acute healthcare services in Ireland and future work will be required to update and expand on the range of services covered and to incorporate new data and methodological developments in cost estimation as they become available. The work presented here is the result of ongoing consultation with an advisory group and other stakeholders, drawing on their expertise in order to adopt a consensus approach in the decisions around prioritising data collection and in shaping the methodology.
In this introduction we discuss first the rationale for developing a set of unit costs for non-acute care (section 1.2 and section 1.3), examine background literature on established methods for estimating unit costs (section 1.4), and provide a brief overview of international and national guidelines for, and examples of, healthcare unit cost databases (section 1.5 and section 1.6).

Why unit costs for non-acute care in ireland?
Resource allocation decisions in healthcare rely on valid and consistent data (e.g., on efficacy, efficiency, equity) including good cost data based on "methodologically sound unit cost information" (Mayer et al., 2020a(Mayer et al., :1142. Unit costs are used in a range of health economic studies including cost-of-illness studies, economic evaluations of alternative healthcare services and technologies, budget impact analyses, and other costing studies. Unit costs are also required for projection models that estimate future demand for, and cost of, healthcare services; for example, the Hippocrates projection model of Irish healthcare demand and expenditure (Wren et al., 2017).
In Ireland, the absence of a national unit cost database is frequently mentioned in the Irish health economic literature (e.g., Gillespie et al., 2019) and the disadvantages include inconsistencies across studies with potential for misinterpretation of cross-study comparisons, challenges in making international comparisons because of inconsistent/incomplete reporting of methods and time-consuming and costly duplication of efforts by researchers to estimate unit costs.
The development of activity-based funding for acute public hospital care by the HPO has improved the availability of consistent unit cost estimates for inpatient and day cases in public hospitals. Inpatient and day case hospital costs are calculated using detailed bottom-up methods and there are plans to expand activity-based funding to acute outpatient services 2 . Several Irish costing studies have used the casemix costs published by the HPO (e.g., Butler et al., 2016;Connolly et al., 2015;O'Sullivan et al., 2016).
Many Irish costing studies also incorporate costs of communitybased non-acute services in their analysis, from cost-of-illness studies (e.g., stroke costs, Smith et al., 2012;dementia costs, Connolly et al., 2014, costs of managing wound-care, Gillespie et al., 2019), to economic evaluations (e.g., palliative care, , extending HPV vaccination to boys, HIQA, 2018a. However, there is no single repository of unit costs for these non-acute services and researchers are required to draw on multiple data sources and methods. Moreover, recent analysis has shown that non-acute care has a very uneven distribution across the country with considerable geographic inequity in supply (Smith et al., 2019). Further investment in the non-acute care sector has been acknowledged as an important priority in the Sláintecare reform programme if more care is to be delivered, where appropriate, in non-acute rather than acute settings (Government of Ireland, 2018; Houses of the Oireachtas Committee on the Future of Healthcare, 2017). Identifying unit costs for non-acute services will also facilitate research on how non-acute care can integrate with, or substitute for, acute care, which are important questions for the current reform programme.
Thus, this paper aims to provide a set of unit costs for a range of community-based non-acute services using standardised and transparent methods for use in health costing studies in the Irish context. At the outset, there are important contextual factors that need to be taken into account in the development of these unit costs.
-First, public service salaries have undergone several changes over recent years and thus it is useful to look at the trends in unit costs. This paper presents unit costs for the years 2016-2019. The economic recession of 2008/2009 led to public service salary reductions and other cost saving measures (e.g., increased working hours and reductions in annual leave), detailed in the Financial Emergency Measures in the Public Interest (FEMPI) Acts (Government of Ireland, 2009a;Government of Ireland, 2009b;Government of Ireland, 2013), public service agreement (DPER, 2010), and national recovery plan (Government of Ireland, 2010). In the subsequent years, as the economy started to recover, public service salaries have been revised upwards at different time points, sometimes mid-year, and these adjustments are taken into account in the unit cost calculations.
-Second, the analysis for this paper has taken place in the context of the coronavirus disease 2019  pandemic, resulting in an inevitable interruption to normal channels of communication with, and data access from, the Health Service Executive (HSE) and other government departments. There are some data gaps and these are highlighted where relevant. As mentioned above, the set of unit costs included in this paper should be seen as a fluid set of data where improvements can, and should, be made over time if and when more resources are put into establishing and maintaining an Irish database of healthcare unit costs. Moreover, unit costs for additional services should be added over time.
A key advantage of this paper is that costing studies in the Irish context will have a single port of call for a range of unit costs for non-acute care thereby ensuring consistency and transparency in methods making cross-study and cross-country comparisons easier in the future.

Which unit costs?
For the purposes of this paper, "non-acute" refers to care services provided outside of an acute hospital setting (e.g., in a health centre, home, long-term residential care facility). Healthcare professionals working in community-based non-acute settings can include GPs, allied health professionals, nurses, dentists, and a range of health care assistants. These professionals include private practitioners (e.g., GPs) and publicly employed professionals.
The recent geographic profile of non-acute supply in Ireland provides a useful overview of the non-acute care sector, focusing on the most central non-acute healthcare services in Ireland, representing the key professions that make up primary care teams and wider community healthcare networks (Smith et al., 2019;Wren et al., 2017).
The analysis by Wren et al. (2017) and Smith et al. (2019) provide a starting point for identifying which services to focus on for the development of a set of unit costs for non-acute care, namely allied health, public health nursing, GP care, and LTRC. In addition, this paper includes unit costs for a range of health care support staff (see Table 1 for the full list of unit costs) and dentists. The majority of unit costs are for publicly employed or publicly financed services, with private fees for GPs, Dentists, and LTRC also included.

Unit cost methods
There are two broad methodological approaches for estimating unit costs in healthcare: the top-down approach (also known as the step-down, gross costing or average costing approach) and the bottom-up approach (encompassing micro-costing, activitybased costing, and patient-level costing) (Batura et al., 2014;Olsson, 2011;Whyte et al., 2018). There are advantages to each and it is generally accepted that different costing methods are appropriate depending on the specific service under investigation (Mayer et al., 2020a).
The top-down methodology assigns aggregated healthcare expenditure to individual units based on some measure of use (Chapko et al., 2009) (e.g., total hospital outpatient expenditure divided by the total number of outpatient consultations to give a cost per consultation). This approach assumes that costs are equally distributed across patients and can be suitable for relatively homogenous services that have relatively similar material and personnel use and similar utilisation patterns across patients (Beecham, 1995;Edbrooke & Hibbert, 1999;Mogyorosy & Smith, 2005;Oostenbrink et al., 2002;Waters & Hussey, 2004). The approach is less suitable for complex services where there is considerable variation in resource type and/ or intensity of use across patients (e.g., hospital inpatient care) (Whyte et al., 2018). A more detailed top-down approach has also been developed whereby patients with shared characteristics are divided into sub-groups so that an average cost is generated for each sub-group (Whyte et al., 2018).
In the bottom-up methodology, and in particular, micro-costing, each component of resource used to produce a given service (e.g., staff, equipment, office space) is identified, measured, valued, summed and then divided by a specific unit of analysis (e.g., per hour, per contact) to give the unit cost (Mayer et al., 2020a). For example, a visit to a physiotherapist in a local health centre is likely to require several resource components including direct (i.e., physiotherapist) salary costs, indirect salary costs (e.g., health centre office staff, cleaning, etc.), indirect administrative costs (e.g., lighting, heating, office supplies), capital overhead costs (i.e., building costs), and others. These individual resource components are identified, measured, valued, summed and then divided by an agreed metric, such as annual working hours, to give a unit cost per hour, or annual number of patient contacts, to give a unit cost per contact, and so on.
Bottom-up costs can be more precise than costs based on the top-down methodology and they also allow for greater analysis of variations in costs across patients, but are more data and time intensive (Cunnama et al., 2016;Mogyorosy & Smith, 2005;Whyte et al., 2018). In practice, many studies use a mix of top-down and bottom-up approaches depending on data availability and the importance of each cost item to the overall analysis (Hendriks et al., 2014) and this is in line with available international costing guidelines for health economic analysis (Drummond et al., 2015) as discussed in the next section.
1.5 Guidelines for unit cost databases 1.5.1 International guidelines. Guidelines for conducting economic evaluation and other costing studies in healthcare include broad guidelines for measuring costs (Drummond et al., 2015;McPake et al., 2020). For example, there are guidelines on which costs should be included in economic evaluations and these decisions are influenced by the perspective of the study (e.g., patient travel could be an important cost component where a societal perspective is adopted, but not perhaps where a provider perspective is adopted, Drummond et al., 2015). There are also general guidelines on how costs should be measured (e.g., opportunity cost, market price, valuations for non-market items, etc.).
The preferred cost in economics is the opportunity cost, that is, the value of the benefit foregone when a resource is consumed and not available for its best alternative use (McPake et al., 2020). The preference for opportunity cost in health economic evaluations was recently confirmed in a Delphi study among European health economists, with country-specific standard costs being a recommended proxy measure (Mayer et al., 2020a). In a comparison of alternative methods for valuing GP care in four European countries (Austria, Germany, Netherlands, UK), Mayer et al. (2020a) noted that the UK unit cost for GP care most closely resembled the opportunity cost concept, consistent with a societal perspective. In contrast, the German and Dutch methods were closer proxies of the payer perspective than of a societal perspective.
However, there is general consensus in the broad costing guidelines that different unit costing approaches can be justified depending on the analytical perspective of the study. It is also acknowledged that costing exercises require considerable effort and are time-consuming and that there is a degree of judgement required to determine how precise cost estimates need to be within a given study (Drummond et al., 2015). The decision to embark on micro-costing or to use a more aggregated cost estimate for a specific service depends to a large extent on the quantitative and qualitative importance of that service to the intervention being analysed. In other words, the more important the cost item is to the analysis, the greater the effort should be made to estimate it accurately (Drummond et al., 2015;Mayer et al., 2020a;McPake et al., 2020).
There are also country-specific guidelines for estimating unit costs in healthcare. For example, the Canadian Agency for Drugs and Technologies in Health (CADTH) has outlined costing methods for commonly used healthcare services in Canada with some guidelines on when different costing methods are to be used (CADTH, 2016). In Australia, the Department of Health periodically publishes a unit costs guideline for hospital and community-based services including residential and home care (Australian Government DOH, 2016). In Europe, the PECUNIA 3 project was established in 2018 to ensure greater consistency and comparability of cost (and outcome) data across studies, sectors and countries. Pilot tests of preliminary standardised unit cost calculation templates have demonstrated general applicability and validity and have highlighted areas for improvement (Mayer et al., 2020b).
Internationally standardised unit cost calculation templates will assist the process of establishing databases of transparent country-specific unit costs for healthcare in European countries. In advance of these becoming finalised, this paper follows closely the methods and guidelines outlined by the Personal Social Services Research Unit (PSSRU) in the UK as well as Irish guidelines where available. The PSSRU has maintained a unit cost database for a wide range of acute and non-acute health and social care services since 1992. The PSSRU unit costs have been used extensively in costing studies and economic evaluations both in the UK and in applications to other countries including Ireland (Whyte et al., 2018).
The PSSRU outlines core requirements for unit costs and these are summarised here (Curtis & Burns, 2019): -Unit costs should be consistent across different economic analyses.
-Unit costs should be comprehensive and consider long-run marginal costs (e.g., staff qualification costs, building costs) as well as direct costs such as salaries.
-Unit costs should be clearly documented so that "it is clear what judgements have been made in constructing them, so that they can be used in an informed way" (Curtis & Burns, 2019:1).
More details on the PSSRU guidelines and methods are outlined in Section 2. The next section introduces available Irish guidelines for unit cost estimation.

Irish guidelines. The Public Spending Code (PSC) in
Ireland is a set of rules and procedures that public bodies are required to follow when evaluating, planning, and managing public resources (both current and capital expenditures) 4 . The PSC also contains detailed technical guidelines for assistance in cost-benefit analyses, regulatory impact analyses, financial assessments, etc. In particular, there are detailed descriptions of key central technical references and parameter values for use in financial and economic appraisal, including guidelines for the estimation of salary costs (encompassing pension costs and overheads) (DPER, 2019).
The Health Information and Quality Authority (HIQA) produces healthcare-specific guidelines. These include guidelines for budget impact analyses of health technologies (HIQA, 2018b) and for economic evaluations of health technologies (HIQA, 2020). While there are detailed guidelines for the estimation of drug costs, and the guidelines for labour costs follow closely those outlined under the PSC, both reports note that the methods for identifying other cost data in the Irish context are not well defined (HIQA 2018b; HIQA 2020). HIQA acknowledges that the absence of a central medical costs database in Ireland means that generating Irish healthcare cost data is "challenging and time consuming" (HIQA, 2020:35). In this context, HIQA recommends the need for flexibility regarding cost valuation and that assumptions and cost estimates should be clearly reported and subjected to sensitivity analysis.
This paper draws on these Irish guidelines where possible to inform the methodology for estimating unit costs for non-acute healthcare, outlined in detail in Section 2. In particular, the paper draws on available Irish recommendations for estimating labour costs. However, where there is strong justification for modifying the parameters set out in the Irish guidelines (e.g., on pensions and overheads), these are discussed in detail and clearly reported. The overall aim is to use available data and methodological guidelines to develop unit costs that are consistent with the definition of opportunity costs.
1.6 International and national unit cost data 1.6.1 International unit cost data. To assist health service performance assessment (e.g., economic evaluations, cost-of-illness studies, etc.) and ultimately to support resource allocation decisions in healthcare, some European countries (e.g., UK, the Netherlands) have developed national unit cost databases (Mayer et al., 2020a) although these vary from built-for-purpose, regularly updated databases (e.g., UK) to systematic reviews of existing cost studies (e.g., Austria).
As noted above, in the UK, the PSSRU publishes unit costs in detailed tables with transparent methods and data sources that are updated on an annual basis (Curtis & Burns, 2019). Different methods are applied depending on the service in question (Whyte et al., 2018). Of direct relevance to this study, a micro-costing approach is used to generate unit costs for community-based healthcare professionals and this approach is described in more detail in Section 2.
In the Netherlands, costing manuals are published periodically by the Dutch National Health Care Institute (2000, 2004, 2010, and 2017) (Kanters et al. 2017;Oostenbrink et al. 2000;Tan et al., 2012). Bottom-up and top-down methodologies are applied to different healthcare services (e.g., bottom-up approach for hospital care costs, top-down approach for primary care physicians, elderly care, home care).
In Austria, the Medical University of Vienna established a unit cost library in 2016. The publicly accessible online database is a compilation of unit costs retrieved from Austrian costing studies and economic evaluations for the years 2004-2019 (Mayer et al., 2020a).
In countries where there are no established unit cost databases, there is evidence of considerable research time spent on sourcing and/or estimating unit cost data. Several costing studies and economic evaluations contain detailed methods and data on healthcare costs using both top-down and bottom-up approaches. Costs can be country-specific or area-specific, or specific to a particular disease or healthcare intervention. Recent examples of top-down analyses range from estimates of unit costs of public hospital costs in Myanmar (Than et al., 2017) to estimates of hospital and other costs in a cost-effectiveness analysis of treatments for severe gastroesophageal reflux disease (GERD) in Korea (Park et al., 2020). Recent examples of bottom-up analyses include detailed estimates of costs of community-based cardiovascular disease prevention care in the USA (Wang et al., 2019), and costs of facilitating a very early infant diagnosis (VEID) test of HIV in Lesotho (Tchuenche et al., 2018); see also a recent review of the unit cost literature by Whyte et al. (2018).
However, there are also many economic analyses that source unit cost information from administrative sources with little attention paid to the costing approach (Mayer et al., 2020a). Mayer et al. (2020a) demonstrate the importance of thinking about, and using the most appropriate, methodological approach when estimating and using unit costs. The authors applied six different unit costing methodological approaches to Austrian data on GP care and found that the cost per GP consultation varied by more than 170% across the different methods. The authors emphasise the importance of a standardised cost database for countries that do not have a set of national unit costs in order to increase the quality of cost data used in health economic analyses and also to "potentially improve the acceptability of such evidence in policy making" (Mayer et al., 2020a(Mayer et al., :1146.

Irish data on unit costs.
In the absence of a national database of unit costs for healthcare, Irish costing studies have used various data sources and methods for cost estimation and many of the costs included are specific to the healthcare service or technology under examination (Whyte et al., 2018).
However, despite the absence of a central database, researchers have applied, where possible, consistent methods and estimates for community-based non-acute care services. For example for publicly employed allied healthcare visits, many Irish costing studies reference HSE salary scale data (Connolly et al., 2014;Connolly et al., 2015;Gillespie et al., 2019). This indicates that the Irish guidelines on labour costs have been applied in these studies although detailed methods (e.g., pension cost, overhead estimates) are not presented. Similarly, a common estimate of the fee paid for a private GP visit has been applied in a number of studies with some limited explanation of the methods (e.g., Connolly et al., 2014;Gillespie et al., 2013;Smith et al., 2012).
Other studies have relied on international cost estimates, for example using UK costs where no Irish data are available (e.g., Gillespie et al., 2011;Gillespie et al., 2019;Manca et al., 2003) and authors have emphasised the need for standardised national unit costs (e.g., Gillespie et al., 2019).
In a recent economic evaluation of palliative care in Ireland, the PSSRU methodology was applied to Irish data for a range of non-acute allied health services . Where Irish-specific data were unavailable (e.g., public sector pension rates, overhead rates), UK data were applied. In this paper, the PSSRU methodology is also applied but we expand on the approach by  by applying Irishspecific data throughout (with the exception of capital costs), presenting unit costs for a wider range of services and for multiple years.
The rest of the paper is structured as follows: Section 2 outlines the methods and data used to calculate unit costs for non-acute services in Ireland. Section 3 presents unit cost tables for selected non-acute services in Ireland for the period 2016-2019. Section 4 presents some concluding comments including recommended next steps for further development of a database of unit costs for non-acute care in Ireland.

Methods and data 2.1 Introduction
The majority of the unit costs included in this paper are calculated using the micro-costing methodology outlined by the PSSRU. The methods are presented in detail in this section along with the data sources used to generate Irish-specific unit costs. These methods are applied to Irish data for the publicly employed healthcare professionals in community-based healthcare services in Ireland listed in Table 1.
Alternative methods are applied to generate public and private unit costs for GP care, dental care, and LTRC and these are also outlined in this section. As outlined in Section 1, these represent the most central non-acute healthcare services in Ireland (Smith et al., 2019;Wren et al., 2017). All analyses are undertaken in Microsoft Excel (Version 2008).
2.2 Micro-costing methods 2.2.1 Overview. The cost estimation approach developed by the PSSRU aims to be transparent and flexible. This paper focuses on the bottom-up approach that is used by the PSSRU to estimate unit costs for community professional staff, community-based nurses and other healthcare professionals 5 . The approach is outlined in Table 2, giving examples of unit costs for different healthcare professionals in the UK, and shows that the unit costs are comprised of separate cost components. These components include wages, salary oncosts (encompassing employers' national insurance and superannuation), qualification costs (where available), overheads (staff and non-staff), capital overheads, and travel (where available). Total cost (i.e. the sum of the components) is divided by the estimated number of hours worked per year to give a unit cost per hour.
Where a user has more up-to-date information on a specific cost component they can substitute their own data for any component (Curtis & Burns, 2019), and this lends flexibility to the unit costs. Sources of information are provided for each cost component thereby ensuring transparency (Curtis & Burns, 2019).
Each of the cost components in Table 2. is described in more detail in the following sections. For each component, the availability of Irish data is also outlined in order to generate Irish-specific unit costs for the healthcare professionals listed in Table 1 Each Grade is given a grade code which is linked with a Grade Group, Staff Group and Staff Category. In line with the approach adopted by  for generating unit costs for palliative care, we have selected the grade code with the largest number of whole-time equivalents (WTEs) employed, in other words, the modal group. For example, Table 3 outlines the percentage breakdown of WTE community-based OTs by grade code based on data from the Health Service Personnel Census, HSE. The majority of WTE OTs employed in community services are working at the grade of Senior OT. Thus, the unit cost for OT in this paper is based on the salary that applies to the grade of Senior OT.
Economic evaluations and other costing studies that apply unit costs to utilisation rates typically rely on high-level data on utilisation (e.g., number of physiotherapy visits, number of OT visits, number of GP visits), rather than on detailed utilisation disaggregated by grade code (e.g., number of visits with an OT; number of visits with an OT, Senior; number of visits with an OT, Manager). In the absence of utilisation data disaggregated by grade code, assumptions need to be made as to which grade codes, and their associated costs, to focus on. Thus, this paper aims to assist this process by identifying, and calculating the unit costs for the modal grade codes based on a reasonable working assumption that these are the most likely grade codes that patients encounter in community-based services. However, it is important to note that the same methods for calculating unit costs that are outlined here can be applied to any grade code. The HSE consolidated salary scales are publicly available (HSE, 2020a) if unit costs for alternative grade codes are needed.
For the majority of healthcare professionals included in this section of the paper, one grade code accounts for 50-100% of the WTEs working in community services.

2.2.2.2
Basic salary or total earnings. The second decision to make regarding the salary component in the micro-costed unit cost is whether to include basic salary or total earnings. Basic salary refers to the contracted salary before any additional allowances and other benefits (in cash/in kind) are added. Total earnings refer to the total payments made after allowances and other benefits are added to basic salary. For HSE employees, these additional payments can include payments for overtime, on-call work, allowances, weekend and night-duty work, and any other arrears .
Basic salary has the advantage of being both publicly available (HSE, 2020a) and consistent with the PSSRU approach and is adopted for this analysis 6 . Alternatively, examination of earnings would give a more accurate picture of the total HSE staffing costs of running public community-based services. However, it is more difficult to calculate the average earnings for a given grade and would require additional analysis in conjunction with the HSE but this should be considered for future analysis of unit costs.
Basic salary scales for each grade code are available from the HSE for each of the years included in this paper (HSE, 2020a). Where salaries are revised mid-year, a weighted average salary for the year is calculated, as described below.
Following the recession of 2008/2009, as the economy began to recover, salaries to public servants were revised upwards on a number of occasions over the period 2016-2020. Some of these revisions occurred mid-year. For example, on 1 st January, 2019, salaries for public servants earning less than €30,000 were increased (by 1%) in line with requirements of the Public Service Pay and Pensions Act 2017 (Government of Ireland, 2017). On 1 st April, 2019, salaries for higher earning public servants were increased to restore pay to pre-recession levels in line with the FEMPI Act 2015 (Government of Ireland, 2015), and on 1 st September, 2019, salaries for all public servants were increased by 1.75% in accordance with the Public Service Pay and Pensions Act 2017 (Government of Ireland, 2017). Where the salary for a given grade code is revised mid-year, a weighted average salary for that year is calculated based upon the month where the revision was implemented. This is illustrated in Equation 1 using the example of OT, Senior in 2019 where the salary (at point 5, the mid-point of the scale) was revised from €56,695 to €57,687 on the 1 st September of that year.
For each grade code, there are different salary points and the number of points varies by grade code. Employees move up these points on an annual basis subject to performance assessment by the HSE 7 . The salary at the mid-point of the scale is  , 2020). Class A is divided into sub-classes based on weekly pay bands and the contribution rates vary across these bands (see Table 4). For each healthcare professional, the annual mean basic salary (see above) is divided by 52 (i.e., to give weekly salary) to determine the appropriate weekly PRSI rate. The appropriate PRSI rate is multiplied by the annual mean basic salary to give the annual employer PRSI contribution.

Superannuation
Irish guidelines on public sector pension costs Employer contribution to superannuation is included in the unit costs as part of the salary oncosts.
Healthcare professionals employed by the HSE are eligible for a public sector pension. The full cost of a public sector pension is difficult to identify ex ante because the Government operates a 'pay as you go' scheme whereby public sector retirement benefits are paid as and when the costs arise. In contrast, private sector pension costs are much more visible because the pension benefits must be funded in advance via employee and employer contributions to a pension fund (DPER, 2017a).
HIQA guidelines for budget impact and economic evaluations of health technologies suggest that a pension rate of 4% (of basic salary) captures the employer contribution to superannuation, citing 2012 guidelines from DPER as the source (HIQA, 2018b; HIQA, 2020).
More recent DPER guidelines for economic appraisals (DPER, 2019) recommend higher pension rates based on an actuarial review of the cost of public sector pensions (DPER, 2017a). The review estimated a notional contribution, expressed as a percentage of pensionable salary, that would be required to be paid throughout the working life of an average employee in order to generate the pension (lump sum plus gratuity) owing to that employee (DPER, 2017a). Estimated employee pension contributions were deducted from this notional contribution to give the net value of the Government's contribution to the employee's pension.
Benefit regimes for public sector pensions have been adjusted over time and there are different cohorts of pension scheme members depending on their date of joining the public service. The most notable difference is between members who joined the public service prior to 2013, and those who joined on or after 1 st January 2013 when the Single Public Service Pension Scheme (the "Single Scheme") was introduced. The benefit regime in the Single Scheme has a very different structure compared with pre-2013 public sector pensions (e.g., retirement benefits are based on career average pensionable salary rather than final salary; later minimum pension age; Consumer Price Index linking rather than pay parity increases).
Thus, the actuarial review reported different notional employer pension contribution rates for pre-and post-2013 entrants to selected posts such as nurses, as well as an overall average contribution rate for public sector posts that have broadly similar benefit structures and salary progression (including Civil Servants, National School Teachers, Nurses, and Engineers), see Table 5.
The figures in Table 5 do not take into account the Pension Related Deduction (PRD). The PRD was a deduction from the salary of pensionable public service employees between 2009 and 2018 as part of the wider package of emergency financial measures aimed at stabilising public finances following the economic recession of 2008/2009 (DPER, 2017a). Inclusion of PRD rates would reduce the notional employer contribution but these were not applied to the calculation of net notional employer contribution rates in the 2017 review because the PRD was not considered a pension contribution under legislation (DPER, 2017a).
However, the PRD has since been replaced by the Additional Superannuation Contribution (ASC) which is a permanent pension contribution linked with legislation (DPER, 2017b; Government of Ireland, 2017). Thus, although the PRD was a temporary measure, it was replaced rather than discontinued and it is important to take the PRD and its successor, the ASC, into account in order to calculate a more realistic estimate of the notional employer public pension contribution rate.
The DPER actuarial review estimated that the average rate of PRD for the posts included in Table 5 was 5%. Taking PRD into account for the years 2016-2018 would reduce the average notional employer contribution rate to 24% (pre-2013 cohort) and 4% (post-2013 cohort) 8 . The ASC rates for 2019 and 2020 are lower than the PRD rates, particularly for the post-2013 cohort. For 2019, the average ASC rates are estimated to be 4% and 2% for pre-2013 and post-2013 cohorts respectively, bringing the average notional employer contribution rates to 25% (pre-2013) and 7% (post-2013) respectively 9 . The notional employer contribution rates for 2016-2020 with and without the PRD/ASC are outlined in Table 6.
In this analysis, the pension rates with PRD/ASC that are outlined in Table 6 are adopted, combined with assumptions about the proportion of health service employees in each of the pre-and post-2013 cohorts.

Pension cohorts within HSE
Application of the separate pre-and post-2013 DPER public sector pension employer contribution rates in the unit costs for publicly employed community-based staff requires data on the proportion of HSE employees in each pension cohort.
Data on the number of active HSE (including Section 38) Single Scheme members at December of each year are deducted from the total number of HSE (including Section 38) employees to estimate the proportion of employees in the pre-and post-2013 pension cohorts for the years 2016-2019. Data on Single Scheme membership were provided by the DOH 10 and DPER 11 . HSE headcount data for the years 2016-2019 were also provided by the DOH 12 .
The pre-and post-2013 employer pension contribution rates (with PRD/ASC) from Table 6 are weighted by the estimated proportion of HSE staff in each cohort, to give the overall average pension contribution rate, presented in  changes in service delivery that involve changing the staff mix (Curtis & Burns, 2019). For example, in the UK, including estimated qualification costs can increase unit costs for allied health professionals by 8-10% (for Bands 5-6) (Curtis & Burns, 2019). Calculation of qualification costs would have required extensive analysis that was beyond the resources available to this project but should be considered in future unit cost estimation work in the Irish context.

Cost component D: overheads 2.2.5.1 Irish guidelines on overheads.
Overhead costs refer to the costs of running the background infrastructure within which the healthcare professionals work. These can include non-pay costs such as utilities (e.g., light, heat, telephone, internet), accommodation costs, office facilities, and general supplies as well as administrative and management staff costs.
HIQA guidelines for budget impact and economic evaluations of health technologies suggest that an overhead rate of 25% (of basic salary) can be applied as a general rule of thumb in the absence of more detailed information on overheads (HIQA, 2018b; HIQA, 2020). This estimate covers "rent, light and heat, office facilities, telephone, general supplies, and so on" (HIQA, 2020: 66). The 25% estimate is taken from the well-cited textbook on methods for conducting economic evaluations by Drummond et al. (2015).
DPER also issues guidelines for economic appraisals and indicates that an overhead rate of 25% can be applied to cover costs of "accommodation, utilities, support and back-office staff, training, travel, etc." (DPER, 2019: 8) although no source is documented for this estimate 13 .
In both cases, HIQA and DPER acknowledge the importance of using more detailed estimates of overhead costs if available (DPER, 2019; HIQA, 2018b; HIQA, 2020). However, there are no published estimates of the overhead costs associated with professionals working in non-acute healthcare settings (e.g., health centres) in Ireland.

PSSRU overhead estimates.
The rate of 25% for overheads is low compared with estimates used by the PSSRU.  The PSSRU overhead rates for community-based professionals are based on 2013/2014 financial accounts for 10 community trusts (Curtis & Burns, 2019) and include management and other non-care staff costs (e.g., administration) at 24.5% of salary costs (basic plus oncosts) plus non-staff costs at 38.2% of salary costs (basic plus oncosts). The non-staff costs include costs for "office, travel/transport, publishing, training courses and conferences, supplies and services (clinical and general), and utilities such as water, gas and electricity" (Curtis & Burns, 2019: 112). Together these indicate an overhead rate of more than 60% of salary costs (basic plus oncosts). Previous application of PSSRU methods to Irish unit costs adopted these overhead rates .

Selection of overhead rate and sensitivity.
The HIQA/DPER description of overheads does not include management staff and this may account for some of the difference between the HIQA/DPER and PSSRU overhead estimates. However, the current HIQA overhead rate is also lower than what was previously recommended for staff cost calculations in Ireland. The HIQA guidelines indicate that the non-pay overhead costs should be estimated 'in accordance with the methods outlined in the Regulatory Impact Analysis (RIA) guidelines issued by the Department of the Taoiseach" (HIQA, 2020: 66). The guidelines from the Department of the Taoiseach were issued in 2009 and recommended an overhead rate of 40% of basic salary, although no source was given for the estimate (DOT, 2009).
There is a clear gap in evidence around the overhead costs for community-based services in the Irish setting. The expansion of activity-based funding to acute outpatient services by the HPO could produce useful data on cost breakdowns in outpatient departments in due course although this will still be second-best to detailed cost breakdowns for services in the community setting.
Given the uncertainty around the overhead estimates, three overhead scenarios are drawn from the available Irish guidelines and PSSRU data to undertake some sensitivity analysis: -

Cost component F: travel.
Similar to qualification costs, estimation of travel costs for publicly employed communitybased health professionals in Ireland would have required extensive analysis that was beyond the resources available to this project and there are no up-to-date estimates from the PSSRU for these professionals (Curtis & Burns, 2019), but these costs should be considered in future analysis.

Number of hours worked per annum.
For a given publicly employed healthcare professional, each of the cost components outlined above are summed and divided by the estimated total number of hours worked per annum to give the unit cost per hour of service. The total number of hours worked per annum is calculated as follows: estimated total number of days worked per annum (number of working days in a given year minus annual leave entitlement and estimated sickness absence days) multiplied by the number of hours worked per day. Table 8 outlines the data sources for each of these elements using the example of Occupational Therapist.

Ratio of direct to indirect time.
There are some data on time use by healthcare professionals in the UK and Ireland but they are typically out of date and/or based on small sample sizes. For these reasons they are not included in this paper in line with the PSSRU approach of focusing on data sources that are less than 10 years old (Curtis & Burns, 2019). However, the reader is directed to section V in the unit cost report by Curtis & Burns (2019) and to  for further discussion of time use patterns by community-based allied health professionals and nurses (including palliative care nurses).

GP unit costs.
2.3.1.1 Structure of General Practice in Ireland. GPs in Ireland are self-employed private practitioners and are free to set their own prices for services provided to private patients .
Many GPs hold a General Medical Services (GMS) contract with the HSE to provide GP care that is free at the point of use to Medical Card (MC) holders (HSE, 1989) and to GP Visit Card (GPVC) holders (HSE, 2005). MCs and GPVCs are granted mainly on the basis of an income-based means test but some are also granted on a discretionary basis where paying for healthcare would cause undue financial hardship. Since 2015, GPVCs are available for adults aged 70+ years. GPs can also hold the 'Under 6s' contract to provide care that is free at the point of use to children under the age of 6 years (HSE, 2015). GPs receive an annual capitation payment (adjusted for age and sex) for each MC and GPVC holder on their list, as well as fees for out-of-hours and special items of service provided to

Methods for GP unit costs Background literature on measuring GP costs in Ireland
Given that both public (i.e., MC and GPVC holders) and private patients are treated in private GPs practices, the most obvious approach would be to estimate a single unit cost for GP care to reflect the average cost of all (i.e., public and private) GP care across GP practices.
However, the bottom-up micro-costing PSSRU method has not been applied to GP unit costs in Ireland given the difficulty in obtaining accurate and reliable estimates of GP earnings as well as detailed estimates of the average overhead and capital costs of running a GP practice.
Rather, there are detailed data on HSE payments to GPs for treatment provided to public patients (including practice support payments), and survey estimates of average payments made to GPs for a private GP visit. Thus, there are estimates of: -the average cost (to the HSE) of a public GP visit; -and the average price of a private GP visit.
The general approach to estimating the cost of publicly funded GP care in Ireland has been to include capitation (mean, or age and sex specific) plus mean additional payments to GPs in respect of MC and/or GPVC holders. There are variations in which additional payments are included depending on the focus of the study.
For example, the most recent estimates of the costs of public GP care have focused on the marginal costs of extending eligibility for publicly funded GP care. Wren et al. (2015) examined the costs of introducing universal health insurance in Ireland, including estimates of the additional costs of providing universal access to GP care free at the point of use. Connolly et al. (2018) expanded on this analysis to focus on how to set a reimbursement price for universal GP care in Ireland and estimated the impact of universal GP care on healthcare expenditure in Ireland. In both pieces of work, the authors examined alternative remuneration rates that would be paid to GPs for existing non-cardholders in the event of universal access to publicly funded GP care. The remuneration rates applied to non-cardholders included: • Age and gender-specific GMS capitation rate plus the average payment made to GPs in respect of existing cardholders for out-of-hours and special service fees. This scenario assumed that the remaining payments made by the HSE under the GMS contracts (e.g., secretarial and nursing payments, annual leave, rostering, rural practice, study leave, etc.) are related to practice overheads and would not apply for marginal additional payments (Wren et al., 2015).
• Age and gender-specific GMS capitation rate plus the average payment made to GPs in respect of existing cardholders for out-of-hours and special service fees, secretarial/nursing, annual leave, rostering/out-of-hours allowances, and superannuation . Other fees (such as dispensing and asylum seekers) were excluded on the basis that they are specific to certain practices and do not extend to the total population  For private GP prices, researchers have relied on available survey data, detailed below.

Public GP visit cost -methods
This analysis draws on the methods adopted in the existing literature and the public GP visit cost is estimated as follows.
In Equation 2, the average annual payment to GPs per person eligible under the MC/GPVC schemes is calculated. ( & )

Average annual payment toGPs per eligible person MC GPVC Weighted average annual no of visits per cardholder MC GPVC
There is an underlying assumption here that GP visits are all of equal length but further data are needed to examine variations in the duration of GP visits and the factors that drive these variations. Recent analysis by Pierse et al. (2019) demonstrate a useful method for collecting data on visit duration in Irish GP practices.

Private GP visit cost -methods
The most recent survey data available on private GP fees are identified. To calculate Equation 1, in the baseline scenario (Scenario 1), total payments to GPs include almost all of the payments that are made by the PCRS to GPs in respect of MC and GPVC holders (including under 6s), see Table 9. The inclusion of the allowances for practice-based overhead supports, as well as the direct patient costs of capitation and fees, is consistent with the micro-costing PSSRU methodology. In line with Connolly et al. (2018), fees for asylum seekers and dispensing fees are excluded on the basis that these payments apply to specific practices and do not extend to the total MC/GPVC population.  Key caveats to note about the GP utilisation data include:

Data for public GP unit costs
• For GP utilisation rates aged 15 years and under, medical card status refers to that of the survey respondent and not the child. Medical card status of the child was not asked in the survey and thus there may be some mis-classification of child medical card status (e.g., where a child has a MC or GPVC but the adult survey respondent does not, or vice versa). GP visiting rates are on average 3-5 visits per annum for children aged 0-15 years in the 2019 survey. These rates are higher than earlier data on child GP utilisation rates (e.g., Wren et al., 2015) but this is the case for other age groups too when comparing HI data with earlier data on GP use.
• For GP utilisation rates aged 15 years and under, sample sizes were too small to examine GP utilisation for MC and GPVC holders separately.
• GP visiting rates were not included in the 2017 HI survey and 2018 rates have been applied.
• There is potential underestimation of visits vis-àvis the payments to GPs. The HI survey questions on GP utilisation specify that a range of visits should be excluded, such as visits to perform prescribed and scheduled procedures (e.g., injections) and this is potentially out of line with the range of allowances included in the payments to GPs under the baseline and sensitivity scenarios outlined in Table 9.
Using the HI data,  implementation of community health networks, advances in eHealth and data management, medicines optimisation and other measures to improve contractual issues (e.g., GP practice profile, complaints procedures etc.). To support this package, the capitation fees (for ages 6+ years) have been set to increase in a phased basis over a 4-year period, 2019-2022, for GPs who sign up to the 2019 Agreement. This strand will see improvements in the delivery of primary care but does not in itself imply changes in visiting rates -i.e., payments increase, visits stay the same. Table 11 outlines the capitation fees for 2020 under the 2019 Agreement.
The Integrated Model of Chronic Disease Prevention and Management aims to provide a structured treatment programme for those with specified chronic diseases (including Diabetes Type 2, Asthma, Chronic Obstructive Pulmonary Disease, Cardiovascular Disease), as well as other measures to identify undiagnosed cases of chronic disease (or cases at high risk of Cardiovascular Disease or Diabetes), and provide annual preventive visits for patients identified with high risk of Cardiovascular Disease or Diabetes. Specific fees for the treatment, case finding, and preventive elements of the programme have been agreed and the programme will be rolled out over the period 2020-2023. This strand will see improvements in chronic disease prevention, diagnosis and management and implies changes in visiting rates (and possibly the duration of visits also), i.e., payments increase and visits increase.
The agreement also includes provisions for paying GPs for additional Special Items of Service. This includes payments for Therapeutic Phlebotomy for Haemochromatosis. Current admissions of MC and GPVC holders to acute/out-patient care for this treatment will be transferred to General Practice (approx. 3 visits per patient per year). GPs will also be paid for their role in handling involuntary admissions to acute mental health facilities, and for attending virtual consultations between GPs and Consultant Cardiologists to discuss patients with heart failure. This strand will see both increases in payments as well as visits.

GP Practice Nurse
Further analysis is also needed to examine the role of the Practice Nurse to avoid reporting duplicate/overlapping unit costs. A separate unit cost for a Practice Nurse is not included in this paper but it should be noted that some of the costs associated with Practice Nurse services are incorporated into the GP unit costs outlined above (e.g., allowances for nursing support, fees for services more likely to be provided by Practice Nurses). Further data on payments and roles and responsibilities within GP practices are required to estimate the unit cost of a GP Practice Nurse visit. The fifth wave of the HI survey (2018-2019) reported that 29% of adults paid up to €50 per GP visit and a further 24% paid up to €75 per visit (Healthy Ireland, 2019). The average payment made by non-medical card holders in the fifth wave of the HI survey was €49.78 21 indicating that an average private fee of €50 is still a reasonable estimate.

Dentist unit costs 2.3.2.1 Structure of dental practice in Ireland.
Dental services in Ireland are delivered in a mixed public-private The majority of dental care is provided by private general dental practitioners and financed by out-of-pocket payments by individuals (Nolan, 2019).
Publicly funded dental care is available through three different schemes: the Dental Treatment Services Scheme (DTSS), the Treatment Benefit Scheme (TBS) and the Public Dental Service (PDS) (DOH, 2019; Nolan, 2019). The DTSS finances dental healthcare services for adult medical cardholders, the TBS subsidises a limited range of dental healthcare for non-medical cardholder adults with eligibility linked to their PRSI contributions, and the PDS finances dental care mainly for children of primary school age (DOH, 2019; Nolan, 2019). The HSE reimburses private dental practitioners for services provided under the DTSS on a fee-for-service basis, and directly employs public sector dental practitioners to deliver the PDS. The TBS is administered separately by the Department of Employment Affairs and Social Protection (Nolan, 2019).

Data for public and private fees.
The schedule of fees payable to dentists under the DTSS is publicly available in the PCRS annual reports (PCRS, 2016; PCRS, 2017; PCRS, 2018; PCRS, 2019) but it is important to note that these fees may not reflect actual costs.
There are limited survey estimates of the private fees charged by dentists with the average price for a dental consultation/ examination close to €45 for the years 2010 to 2014 (National Consumer Agency, 2010; WhatClinic.com, 2014). A more recent survey of private dental fees is underway and preliminary results for 2020 prices are reported here for the services of dental examination (excluding x-rays), cleaning (scale and polish by the dentist), and routine extraction. The survey covers a nationally representative sample of c. 100 private dentists in the Republic of Ireland 22 .

Long-term residential care unit costs 2.3.3.1 Structure of long-term residential care in Ireland.
Long-term residential care is delivered in a range of settings including private nursing homes, public facilities including extended care units, public nursing homes, and a small number of welfare homes, and in non-statutory/voluntary agencies.
The LTRC sector has been changing rapidly over recent years and approximately 75% of LTRC beds are now provided by private nursing homes (HIQA, 2014).
Although the majority of care takes place in private LTRC beds, the majority of residents are financed through the public Nursing Homes Support Scheme (NHSS), known as the 'Fair Deal' scheme (Wren et al., 2017). Applicants to the Fair Deal scheme receive a care needs assessment to confirm that long-term nursing home care is appropriate. The Fair Deal participant makes a contribution towards the cost of his/her care (based on an assessment of income and assets) and the HSE pays the balance (Citizens Information Board, 2020).

Data sources on costs.
Data are available on the costs and charges for care in the public and private LTRC facilities participating in the Fair Deal Scheme. These data are assumed to be representative of the LTRC sector given that in 2018, 94% of all LTRC facilities that were registered with HIQA participated in the Fair Deal Scheme (C&AG, 2020) 23 .
An individual weekly charge is calculated for each LTRC facility participating in the Fair Deal Scheme. The methods for calculating the weekly charge are different for public and private facilities.
Charges for public LTRC facilities are published by the HSE 24 and are based on costs of care including: pay (management, nursing and support staff) excluding superannuation costs, plus operating expenses to cover minor capital works, general equipment and furniture, training and education costs. However, it is important to note that the cost components and operating expense guidelines for calculating these weekly charge rates have not been updated since 2009 and this has been recently criticised (C&AG, 2020).
The charges for private LTRC facilities are based on an agreed maximum price negotiated between each facility and the National Treatment Purchase Fund (NTPF) 25 .

Unit cost tables
3.1 Micro-costed unit costs 3.1.1 Overview. The tables in this section show the estimated unit costs for 11 publicly-employed, community-based, healthcare professionals calculated using micro-costing methods based on the PSSRU approach. These include allied health, public health nursing, and care support staff. While the baseline provides a useful mid-range estimate, there may be justification for applying the low/high sensitivity estimates depending on the focus of the study. For example, in a study where unit costs for a range support staff are being included, it may be reasonable to adjust the overheads downwards accordingly assuming that some of these support services are captured in the overhead costs and to avoid double counting. In these cases, transparency in reporting which unit cost is adopted is crucial to ensuring comparability across studies.                            Table 23 for estimated unit costs for public GP visits over the period 2016-2019. Table 24 for estimated unit costs for private GP visits over the period 2016-2019. Table 25 for the schedule of fees payable to private dentists under the DTSS over the period 2016-2019 for public dental costs. Table 26 for estimated average private dental prices for specific services in the year 2020.     Table 27 for estimated unit costs of LTRC in public and private facilities over the period 2016-2020.

Conclusions & next steps 4.1 Concluding comments
This paper presents detailed unit costs for 16 healthcare professionals in community-based non-acute services in Ireland. A central national database of unit costs for non-acute healthcare has not previously been available for Ireland. The unit costs set out in this paper can be used in a wide range of health costing studies in the Irish context (e.g., economic evaluations, cost-of-illness studies). The availability of a set of transparent costs will ensure consistency across Irish health costing studies and facilitate cross-study and cross-country comparisons.

Limitations
The unit costs included here are based on the best available data for 2016-2019 32 . There are important limitations in currently available data that need to be addressed in future analyses 32 With the exceptions of LTRC and private dental care, for which unit costs for 2020 are also provided. of unit costs and these have been highlighted where relevant throughout the paper.
First, for the micro-costed unit costs, there remain data gaps around some of the core cost components that have been included in this paper. In particular, there are no published detailed data on overhead costs in community-based healthcare facilities in Ireland. While the paper has presented alternative overhead rates, the baseline of 40% provides a reasonable benchmark between the high rates applied in the UK and the comparatively low rate recommended in current Irish costing guidelines.
Second, there is need to advance data collection for some key services (e.g., additional nursing cadres working in the community) and cost components that have not been included in this paper, in particular, qualification costs, and travel costs.
Third, there is need to advance analysis around the core salary cost component. Assessment of total earnings rather than basic salary together with analysis of actual rather than contracted working hours would give important insight into the extent of overtime that staff are working. The cost of agency staff is also not considered in this paper and is an important consideration in the context of considerable reliance on agency staff in the Irish healthcare workforce (Houses of the Oireachtas Committee on the Future of Healthcare, 2017).
Fourth, this paper has focused on providing a unit cost per hour of service while in practice there is variability in duration of visits both within and across services. For example, an initial physiotherapy assessment might last 1 hour while subsequent visits might last 30 minutes (Brick et al., 2010), or a GP visit can last a few minutes or much longer depending on the complexity of the presenting complaint (Pierse et al., 2019). As discussed earlier, the increased role for GPs in chronic disease prevention, diagnosis and management as part of the new 2019 GP Agreement (DOH et al., 2019) may have implications for the duration of GP visits.
Thus, as noted at the outset, the set of costs included in this paper should be seen as a fluid set of data where improvements need to be made over time. In the meantime, transparency in methods and data sources ensures that users of these unit costs are better able to make cross-study and cross-country comparisons.

Next steps
This paper brings Ireland in line with other countries that already have a central database of unit costs for non-acute care (e.g., UK) and will facilitate future engagement with international efforts to standardise cost-calculation methods across Europe.
The unit costs for the publicly employed non-acute healthcare professionals presented in this paper have changed by 2-6% over the timeframe 2016-2019. Larger percentage changes are observed in the unit costs for public GP visits and public LTRC (14-15%) over the period 2016-2019. As discussed, future patterns in GP costs are difficult to project because of the implications of the new 2019 GP Agreement (DOH et al., 2019) and further examination of the public LTRC costs will be required in light of the outcomes of the LTRC value-for-money review. While the patterns observed in the unit costs presented in this paper provide reasonable estimates of the likely costs for these services for 2020-2021, it is important to recognise that for these cost data to make useful contributions to decisionmaking in healthcare research, they need to be updated on a regular basis with specific need for more in-depth analysis of changes to GP and LTRC costing structures. Moreover, the implications of the COVID-19 pandemic for the costs of delivering community-based non-acute healthcare (e.g., increased phone and online consultations) will need to be assessed.
The analysis also suggests that current Irish health costing guidelines need to be revised, particularly in terms of pension costs.
As outlined above, there is need for more detailed data and analysis in the following areas relating to the costs of delivering community-based non-acute care: -Overhead costs; -Qualification costs for community-based non-acute healthcare professionals; -Travel costs; -Additional salary costs (over and above basic salary) associated with overtime, agency staff, etc.; -Duration of consultations (average duration for each service as well as examination of sources of variation in durations).
The joint tasks of addressing these data gaps as well as updating the unit costs will require dedicated resources for the development of an ongoing database of unit costs for non-acute care. Linking in with the research by the PECUNIA project will also ensure consistency in methods and comparability in costs in cross-country studies within Europe.

Source data
The analysis in this paper is based on secondary data sources that are available from the DOH, the HSE, other public administrative bodies, and publicly available survey data.
-For the micro-costed unit costs, the primary data points include basic salaries for selected healthcare professionals, at selected grades, and these are drawn from the HSE consolidated salary scales and informed by data provided by the HSE Personnel Census. Superannuation estimates are generated using data on the proportion of HSE staff on pre-and post-2013 pension schemes. Data on the number of active HSE (including Section 38) Single Scheme members at December of each year are deducted from the total number of HSE (including Section 38) employees to estimate the proportion of employees in the pre-and post-2013 pension cohorts for the years 2016-2019. Data on Single Scheme membership were provided by the DOH and DPER. HSE headcount data for the years 2016-2019 were also provided by the DOH. Annual leave entitlements are identified from the HSE as well as from national agreements. Absenteeism rates for HSE staff working in community services, disaggregated by staff category, are available in HSE Management Data Reports. -For dental care, the schedule of fees payable to dentists under the DTSS is publicly available in the PCRS annual reports. Preliminary findings from a 2020 survey of private dental fees by the authors of this paper are also included and the final results from the survey are due to be made publicly available this year.
-For LTRC, charges for public LTRC facilities are published by the HSE and the charges for private LTRC facilities are published by the NTPF.