The economic burden to the public health system of treating non-viral injecting-related injury and disease in Australia (a cost of illness analysis)
Campbell K. Aitken, Centre for Population Health, Burnet Institute, GPO Box 2284, Melbourne, Victoria Australia 3001. Fax: (03) 9282 2100; e-mail: firstname.lastname@example.org
Objective: We estimated the cost to the public health system of treating Injecting-Related Injuries and Diseases (IRIDs) in the three most populous states in Australia in the 12 months over 2005/06.
Methods: We conducted a cost of illness analysis from the perspective of the public health system. Costs of treating IRIDs in the community were estimated from health service utilisation surveys of injecting drug users and physicians (yielding data on Government subsidised physician visits, medicines prescribed and emergency department presentations). Data on admitted hospitalisations in public hospitals due to IRIDs were extracted from State Government databases. Appropriate costs were attached to all Government-borne services and prescriptions to estimate the total cost to the public health system of treating IRIDs in 2005/06 in Queensland, NSW and Victoria.
Results: Our estimate of the cost to the public health system of treating IRIDs in Queensland, NSW and Victoria in 2005/06 was $20 million.
Conclusion: IRIDs are an under-recognised harm resulting from injecting drug use, but the economic burden of IRIDs in Australia are non-negligible. Research is needed to identify cost effective programs to reduce the clinical and economic burden caused by IRIDs, particularly to reduce hospitalisations due to IRIDs.
Implications: General practitioners, clinicians and other health workers need to be alert to IRIDs in their injecting drug user clients to prevent progression to more serious disease and consequent elevation of the associated economic costs.
Those working with injecting drug users (IDUs) have long been aware of non-viral injecting-related injury and disease (IRIDs).1 Injecting drugs (such as heroin, opioid pharmacotherapies, amphetamine-type substances, cocaine, benzodiazepines, and anabolic-androgenic steroids) can give rise to both vascular and soft tissue injuries and also expose drug users to a range of organisms resulting in the development of localised or systemic infections.2–4 Some of these non-viral IRIDs are anecdotally common and many have been described in the scientific literature.5–7 Some (e.g. infective endocarditis, tetanus) are extremely serious and require urgent medical attention, others (e.g. abscesses, cellulitis) are easily treated but potentially life-threatening if neglected, and others still (e.g. swelling, hives) are minor conditions but very common and may contribute significantly to reduced health and well-being in the IDU population.
Existing literature on the public health impact of injecting drug use is focused on blood-borne viral infections, such as HIV and hepatitis B and C, and drug-related overdoses. Research on non-viral IRIDs other than overdose is scarce, confined largely to case reports8–10 or outbreak investigations.11–13 There is little awareness of the extent of the problem in Australia or internationally, and we know of no published estimates of the economic burden associated with the treatment of IRIDs in Australia or elsewhere. To address this gap, we estimated the cost to the public health system of treating IRIDs over a 12 month period in Australia's three most populous states.
A cross-sectional, quantitative survey of a convenience sample of current injecting drug users (IDU Survey) was conducted in one metropolitan and one regional site in Queensland, NSW and Victoria (six sites in total) during November and early December 2006. IDUs attending the recruitment sites were invited to participate, and recruited if eligible, until state quotas of 140 or pre-set survey end dates were reached. To be eligible, participants must have injected at least monthly in the preceding six months, and have been aged 18 years or older. We also conducted an on-line quantitative survey of medical practitioners (MP survey), all either pharmacotherapy prescribers or emergency department physicians (more details about our data collection methods can be found in Dwyer et al. 2009).14 The survey data were used in conjunction with hospital admissions information to estimate the economic impact of treating IRIDs. Costs were estimated for three treatment settings in which the public health system faces costs of treating IRIDs, the community, public hospital emergency departments (ED) and public hospitals admissions. For all three treatment settings, data collection was confined to the three states from which we collected IDU data.
1. Treatment in the community
In the IDU survey, 393 participants were asked to estimate the number of IRIDs (by IRID type) they had experienced in the previous 12 months (January 2006 to January 2007). IRIDs were defined as non-viral injuries or diseases resulting from the physical act and associated biochemical effects of injecting illicit drugs; drug overdose was specifically excluded. (Note that our interviewers were provided with an overview of IRIDs and their symptoms and received training in administration of the questionnaire, but were not medically trained and were not required to diagnose IRIDs; instead, they were provided with lists of symptoms and non-technical names of IRIDs to help them determine, in conjunction with the interviewee, the likely identities of the IRIDs experienced.) They were also asked what treatments they sought and received for the most recent case of each IRID. These data were used to estimate the resources required to treat IRIDs in the community. A review by Darke found self reporting of drug related problems by injecting drug users was reliable and valid.15 The MP survey on IRID types and prescribed treatments was also consistent with the IDU survey responses.
The total number of health worker consultations (GP, nurse, community health worker) in Queensland, NSW and Victoria was extrapolated by assuming that the treatment sought and received for the most recent episode (including where no treatment was sought), was the same as for previous episodes in the 12 month period.
The Australian Government's Medicare Benefits Scheme (MBS) schedule16 was sourced to value all health services received in the community.
Some IDU survey participants indicated drug therapy had been prescribed; in these cases, we sought expert medical opinion and the Therapeutics Guidelines for Antibiotics17 to determine the most likely type and course of pharmaceutical(s) prescribed to participants. The MP IRID survey provided validation of these assumptions, as there was considerable consistency in the types of treatments prescribed. The Pharmaceutical Benefits Schedule18 was sourced to value the costs to the Federal Government of all pharmacotherapies. Costs of prescribed treatments deemed more likely to be borne by individuals were not included in our estimate (e.g. pressure bandages, analgesics, anticoagulants).
The total costs to the public health system borne in the community generated by the 393 surveyed IDUs was extrapolated to estimate the total cost borne treating the entire IDU population in those three states. The IDU population size was estimated by multiplying the IDU prevalence estimated by Razali et al.19 for 2005 by population statistics from the Australian Bureau of Statistics.20 The upper and lower limits from Razali et al. were used for sensitivity analysis. Table 1 presents estimated IDU populations for each state in our analysis.
Table 1. Injecting drug user population estimates (2005).
2. Public hospital emergency departments
Emergency department (ED) physicians were invited to participate in the online Medical Practitioners' survey to estimate the number of IRID presentations to EDs. Only seven ED physicians responded from a total of 64 hospitals contacted across the three states. The number of ED Physicians working in each of the hospitals to which introductory letters were sent was unknown.
To extrapolate survey data to estimate the number of IDU contacts for all ED physicians across the three states requires knowledge of the total number of full time equivalent ED physicians. There were 614 full-time Fellows of Emergency Medicine working in accredited emergency departments in Victoria, QLD and NSW in 2007 (information supplied by the Australasian College of Emergency Medicine (ACEM), September 2007); thus the number of emergency department IRID episodes observed by the seven ED respondents was extrapolated to 614 ED physicians assuming the same rates. This is likely to be an underestimate of actual contacts, because not all emergency department physicians are Fellows of the ACEM, and 37 part-time fellows were excluded from the calculation due to lack of precision about their equivalent full time status.
The analysis considered IRID episodes treated in an ED setting. Expert opinion indicated ED presentations tend to receive wound care. It is not appropriate to include the cost and resource use that may result from an overdose presentation in an ED (e.g. counselling, behavioural management, pathology). The cost of an emergency department presentation was set at $38 ($47 when adjusted for inflation from August 2002) - the Australian Ambulatory Class 28 Emergency Department cost for “skin with procedure” from the Manual of Resource Items and their Associated Costs for use in submissions to the Pharmaceutical Benefits Advisory Committee involving economic evaluation.21 More serious episodes that resulted in admission were captured in the public hospital admissions estimates.
3. Public hospitals admissions
Victorian, NSW and Queensland Government Admitted Episodes databases were used to identify relevant IRID public hospital separations in the 12 months of 2005/06. There were no specific hospitalisation codes for IRIDs or a specific IDU identifier for each hospitalisation. This made it difficult to identify which hospitalisations were due to IRIDs among IDUs. Two database search strategies using International Classification of Diseases Version 10 Australian Modification (ICD-10-AM)22 codes were used to provide lower and upper estimates of hospitalisations likely to be due to IRIDs. Table 2 lists the diagnoses included in each search strategy. The strategies differed in that the first required the listed infections to be a principal diagnosis while the upper limit search allowed endocarditis, abscess and cellulitis to be both a principal and secondary diagnosis. For both searches, cases presumed to be IRID-related were identified using secondary diagnoses criteria of relevant drug use disorders. The search was limited to people aged between 15 and 60 years as it was believed hospitalisations outside of this range were unlikely to be due to illicit drug injection.
Table 2. Hospitalisation data search strategies.
|Principal diagnosis for:||Injury of blood vessels at shoulder and upper arm level OR||Search 1 minus principle diagnoses for:|
|Tetanus OR||Injury of blood vessels at forearm level OR||Endocarditis OR|
|Streptococcal sepsis OR||Injury of blood vessels at wrist and hand level OR||Abscess OR|
|Other sepsis OR||Injury of blood vessels at hip and thigh level OR||Cellulitis)|
|Gas gangrene OR||S85 Injury of blood vessels at lower leg level OR|| |
|Endophthalmitis OR||Injury of blood vessels at ankle and foot level OR||plus|
|Endocarditis OR||Injury of blood vessels of unspecified body region)|| |
|Phlebitis and Thrombophlebitis OR||AND secondary diagnosis for:||Search 3b|
|Other disorders of veins OR||Opioids OR||Principal or secondary diagnosis for:|
|Abscess OR||Cocaine OR||Endocarditis OR|
|Cellulitis OR||Stimulants OR||Abscess OR|
|Pyogenic arthritis OR||Multiple/Other OR||Cellulitis)|
|Osteomyelitis of vertebra OR||Drug use)||AND secondary diagnosis for:|
|Osteomyelitis OR|| ||Opioids OR|
|Necrosis OR|| ||Cocaine OR|
|Gangrene, NEC OR|| ||Stimulants OR|
| || ||Multiple/Other OR|
| || ||Drug use)|
Each separation identified is assigned an ARDRG (Australian Refined Diagnostic Related Group) code by the treating hospital, each of which (ARDRG) has a Government cost weight attached. All identified hospitalisations were costed using public hospital cost weights from the National Hospital Cost Data Collection (NHCDC) Round 9 cost weights, 2004/05.23 These were adjusted for inflation and presented in 2006/07 Australian dollars using the Health Sector Consumer Price Index from the Australian Bureau of Statistics.24
Within each of the three analysis settings, assumptions were made about several variables that have a potential impact on total public health costs.
- • In the community setting it was assumed that the treatment sought and prescribed in the most recent IRID event was the same as for all previous episodes. It is possible that in earlier events, more or less treatment was sought and received. Given the lack of sampling frame, it is impossible to know if the sample size was large enough to account for this by picking up people across the range of events in time.
- • In the emergency department costing, it was assumed that the seven respondents were representative of all emergency department physicians across the three states in terms of their IDU contacts. With only seven of at least 614 possible respondents providing data, there is a chance that their data are not highly representative, reducing the reliability of our estimate of total emergency department cost for IRID treatment.
- • IRID-related hospitalisations tended to be longer than the average length of stay (ALoS) of hospitalisations reported for each relevant ARDRG presented in the National Hospital Cost Data Collection (NHCDC) Round 9 cost weights, 2004/05.23 This suggests that the average cost per stay for each ARDRG is likely to be an underestimate of the true cost for IRID hospitalisations. No data were available to assess the impact of this difference.
Our sample contained 393 IDUs, 73% of whom were male, of average age 33 years (range 18-65). Most (74%) were unemployed at interview, and 33% reported unstable accommodation in the previous 12 months; most (66%) were born in Australia, and 8% identified as Indigenous Australians. Forty-six per cent reported a diagnosed mental health problem (most commonly depression, 22%). Self-reported hepatitis C exposure prevalence was 54%; self-reported HIV prevalence was below 1%. Nearly 60% of interviewees had been injecting drugs for 10 years or more, and 14% for five years or less. Nearly half reported injecting daily or more frequently in the four weeks preceding interview. More details about our study group are available in Dwyer et al. (2009).14
Table 3 presents the estimated costs due to IRIDs borne by the public health system in the community. Results are presented for survey participants and then extrapolated for all IDUs in Victoria, NSW and Queensland.
Table 3. Costs of treatment in the community.
|Total for surveyed population (n = 393)||$20,566|
|Extrapolated total for all IDUs in Vic, NSW, Qld||$8,706,591|
|Extrapolated (lower 95% CI)||$5,194,757|
|Extrapolated (upper 95% CI)||$11,906,171|
Tables 4 shows the estimated costs faced by the public health system in public hospital emergency departments due to IRID-related treatment, extrapolated from our survey of treatment providers.
Table 4. Costs of public hospital emergency departments.
Table 5 presents the estimated costs to the public health system as a result of public hospital admissions due to IRIDs. The cost estimates are presented for the hospitalisations identified using each of the search strategies presented in Table 2.
Table 5. Costs of public hospital admissions.
Table 6 presents a summary of the total estimated costs of treating IRIDs in the public health systems of Victoria, Queensland and NSW in 2005/06.
Table 6. Total estimated cost to public health system (with sensitivity analysis).
The final two rows in Table 6 present the results of the sensitivity analysis. The lower estimate utilises the lower bound of the IDU population estimate and the lower bound of hospitalisation costs (Search 1 – see Table 2 for search details), while the upper estimate utilises the upper bound of the IDU population estimate and the lower bound of hospitalisation costs (Search 2).
It is estimated that between $16.3 million and $27.6 million (2006) was spent by State and Federal Governments treating injecting related injuries and diseases in Victoria, NSW and Queensland over the 12 months of 2006/07. If we assumed similar rates of IRID events and treatments received in other states, then this would equate to between $21 million and $36 million across all of Australia over 12 months.
The $19.8 million estimate is a small proportion of the total Government health budgets of those states (approximately 0.05%).25 Nevertheless, in real terms the costs of treating IRIDs are considerable, including between 8,496 and 14,044 hospital bed days in the public hospitals of Queensland, NSW and Victoria. For comparison, this estimate exceeds the estimated annual public expenditure on needle and syringe programs in Australia in the early years of this century ($17 million).26 The Australia wide estimate of $25.6 million is also approximately one-third of the annual costs of treating hepatitis C related conditions across Australia in public and private settings ($78,931,278).27 Note that estimated hospitalisation costs are at least as great as the costs of treatment in the community, yet result from far fewer episodes of care; this outcome points to a need to prevent non-serious and potentially serious IRIDs from progressing to serious IRIDs, which are the most expensive to treat. There is an important role here for clinicians who are the point of contact for injecting drug users to make IDUs aware of hygienic injecting practices,1 and also to encourage early treatment of future IRID events to help reduce hospitalisations.
Several limitations of our data and unavoidable assumptions must be borne in mind. The physician sample was small and while the IDU sample size was relatively large, there is no sampling frame for this population making it impossible to determine the appropriate sample size. Further, both physician and IDU surveys relied on convenience samples, therefore findings may not be generalisable to the broader populations. Studying IDUs in three eastern Australian states may not permit any generalisation to IDUs in other states, due to differences in drugs of choice, and models of needle syringe program (NSP) and other service delivery.
Hospitalisation data were retrieved for the 12 months of 2005/06, while the IDU and physician surveys were based on the 12 months of 2006. It is not expected that this misalignment would affect the results.
Data were not collected for several items that have real costs to the Federal and State Governments. These costs (listed below) were excluded from the analysis, again suggesting that our work has produced a conservative estimate of the cost to the public health system of treating IRIDs.
- • The Physician Survey revealed that some presentations were referred to specialists. The cost to the public health system of specialist referrals has not been included as we were not able to determine the number of IDUs covered by responding Physicians. Nevertheless, the Physician survey did reveal that the number of specialist referrals as a proportion of actual diagnoses for specific IRIDs was relatively small.
- • Some patients may experience resistance to antibiotics, which could require follow up visits to the GP and the prescription of a second line antibiotic therapy, but this was not modelled in the analysis due to lack of data.
- • Ambulance costs were not included in this analysis as we did not have data on the means of transport for attending hospital. It is very likely that some attendances utilised an ambulance service.
- • Due to a lack of data, government contributions to private hospitalisations have not been included in the analysis.
A paucity of data around IRIDs in the Australian setting meant that numerous assumptions had to be built into the costing analysis (as described in the methods section and discussed here). Those assumptions introduced uncertainty into the cost estimates; more extensive data collection is required to improve the robustness of this costing analysis. In addition, it should be noted that numerous potentially significant costs associated with IRIDs were unable to be measured in our study as they were broader than the public health system perspective of the study. These include costs to IDUs such as loss of employment or reduced income due to time off work, and costs associated with care outside the formal sector and consequent family disruption. Nevertheless, our analysis has shown that there were substantial costs (in the tens of millions of dollars, at least) borne by the State and Federal Governments as a result of treating IRIDs in Victoria, Queensland and NSW in 2005/06, and that these costs add significantly to those of HIV, hepatitis C and overdose that are conventionally associated with injecting drug use. It has also shown that IRIDs likely to result in hospitalisations generate significant costs.
Implications for public health
Efforts should be made to increase early treatment seeking behaviour among IDUs to reduce the occurrence and severity of IRIDs. General practitioners, clinicians and other health workers need to be alert to IRIDs in their injecting drug user clients to prevent progression to more serious disease and consequent elevation of the associated economic costs.
Several avenues exist for further study of IRIDs and their associated economic costs. The most obvious is via clinical assessment of IRIDs and tracking of treatment through record linkage (with Medicare and associated databases), which would allow more accurate ascertainment of IRIDs and better measurement of the associated costs. Greater validity – in terms of application to all Australian IDUs – could be achieved by extending a future study into all Australian states and territories, which would capture the full range of IRIDs associated with different drug types and injecting practices.
This research was funded by the Commonwealth Department of Health and Ageing and auspiced through Queensland Health. The authors are grateful to the people who participated in the IDU survey and contributed their experiences of injecting-related injuries and diseases.