Pressure injury in Australian public hospitals: a cost-of-illness study

Additional keywords: opportunity cost, pressure ulcer, prevalence rate, treatment cost.

Pressure injury (PI) is considered one of the most common causes of iatrogenic harm to patients.¹ It is associated with sustained pain, discomfort and increased immobility and mortality rates in addition to decreased quality of life in both acute and long-term care settings.^2,3 It also carries a substantial financial burden associated with ongoing care incurred by individuals and families, the healthcare system and society. As a preventable condition, PI prevalence is being measured nationally and internationally as an indicator of quality of nursing care in health facilities, and complainants in litigation can be awarded substantial costs.^4–6 In Australia, The National Safety and Quality Health Service Standards provide health service leaders guidance on areas to target in improvement strategies. Standard 8, Preventing and Managing Pressure Injury, requires health service organisations to implement evidence-based systems to prevent PIs and to manage them when they do occur.⁷ Yet, hospital-acquired PI remains an unsolved problem.

The treatment for PIs is known to be costly; however, there is little precise information on prevalence and costs.⁸ The prevalence rate, and subsequently estimated number of PI cases, varies significantly from one study to the next. Studies have reported prevalence rate estimates ranging from 2 to 23%^2,9–17 and incidence rates between 1.5 and 38%.^18–22 In Australia, the overall reported PI prevalence between 1983 and 2002 ranged between 3 and 36.7%.²³ The variation is associated with healthcare setting (acute care vs long-term care vs home care), disease specific (e.g. spinal cord injury, cardiovascular etc.) and data collection methods (e.g. hospital surveys, patient-level data).³⁴ More specifically, state-wide audits estimate PI prevalence in hospitals ranges from 9.5 to 17.6%.¹⁰ Studies in nursing home and long-term care settings estimate the prevalence of PI to be around 8.9%.^25,26

Similarly, studies investigating the financial burden associated with PIs have presented a wide range of estimates along several dimensions, such as degree of severity (Stage I to IV),²⁷ additional length of stay (LOS) attributable to PIs and whether PIs occurred in medical or surgical patients.^16,19,22,28 The variation in cost per case is substantial, with estimates ranging between US$500 and US$40 000 in the US^9,29 and from £1214 for Stage I to £14 108 for Stage IV in the UK.¹⁷ Findings suggest that personnel costs, such as nursing and carers’ time, contribute a large proportion of the total treatment cost, whereas the use of medical materials, special beds and mattresses only make a minor contribution.³⁰ For severe cases, complications that lead to delayed healing, additional diagnostic tests and monitoring and extended LOS are a major determinant of cost.¹⁶ Subsequently, there is a wide range of cost estimates associated with PI treatment across countries (from millions to billions of dollars),^{2,8,9,16,30,31} representing between 0.4 and 6.6% of a country’s health expenditure.

There have been studies on the prevalence and economic losses of PI in Australia. Graves et al. estimated the impact of PI on in-patient LOS, as well as the opportunity cost of bed days lost using 2001–02 data.^19,32 Jackson et al.³³ reported that PI ranked among the top five hospital-acquired complications (by total additional system cost). Based on data from 2005–06 (Victoria) and 2006–07 (Queensland), Jackson et al. reported the total number of PI cases to be 2873, representing a prevalence of 0.2% (much lower than the prevalence estimated in other studies^1,34,35) and a total cost of A$22.9 million for the public hospitals in these two states.³³ It is noted that this prevalence rate, and thus estimated cost, was likely to be underestimated. Jackson et al.³³ derived the estimates from data collected retrospectively using the Diagnosis-Related Group (DRG) system, whereas the state reports^1,34,35 estimated their prevalence rates from prospective prevalence surveys where PI was recorded through direct skin inspection, which was more likely to be more accurate. Antonio and Conrad³⁶ presented an evaluation of the Wound Care Improvement program in the Ballarat Health Service (Victoria), in which prevalence data by PI stage were collected for 2009, 2011 and 2012. That study also estimated cost saving attributable to the reduction of PI; however, the sample was small and not representative of Victoria or wider Australia. The most recent study, by Graves and Zheng,³⁷ estimated the direct healthcare costs of chronic wounds in Australia to be A$1.13 billion (±0.72 billion) for the 2010–11 financial year. None of these studies estimated the costs disaggregated by PI stage.

The states of Victoria, Queensland, Western Australia and South Australia have conducted regular PI audits and introduced hospital-acquired PI as a quality control indicator, rationalising that PI is largely preventable in the hospital setting if appropriate and timely screening, skin assessment and prevention strategies are applied. These efforts have generated some data on PI for Australian public facilities.^{1,10,34,35,38–40} Yet, the annual PI cost to the public hospital system and, more broadly, to Australian society, as well as the impact of PI severity on costs, remain largely unknown.

In the present study, we investigate the direct treatment cost related to PI, as well as the indirect opportunity cost of bed days lost due to preventable PIs, by PI severity and Australian states and territories, in the context of public hospitals. We retain our focus on the costs of treatment, and do not estimate the cost of prevention, which has been the focus of cost-effectiveness studies.^41,42

A cost-of-illness analysis was conducted using a prevalence approach and a 1-year time horizon, based on data from the existing literature extrapolated using simulation methods. The PI treatment cost was estimated in the public hospital setting, disaggregated by stage (from Stage I to IV)⁷ and by state and territory. This cost was a direct health system cost that included nursing time for risk assessment, monitoring and repositioning, skin dressings, moisturiser, antibiotics and analgesics and supporting surfaces. The indirect opportunity costs associated with extended LOS represented the value that should have been produced (in terms of the value of benefits for patients with other illness who could have been treated) if PI were to be completely prevented. This cost was disaggregated by state and territory. The direct and indirect costs were then summed to give an overall estimate of the costs to the Australian public hospital system and society more broadly.

We used existing data from the published and grey literature to estimate: (1) the number of patients with PIs, by stage and state; (2) the average treatment cost for each PI stage; and (3) the total treatment cost, by stage and by state.

The estimation of PI cases was based on the number of in-patient cases at risk of PI and the prevalence rate. A wide range of prevalence rates has been reported in Australia, from as low as 0.2% in the study of Jackson et al.³³ to 17.6% in the Western Australia Wound Prevalence Survey.¹⁰ To incorporate uncertainty around the true prevalence, we fitted a beta distribution, which is suitable for parameters bounded by 0 and 1, using the minimum (0.2%) and maximum (17.6%) values reported in the Australian literature.^{1,7,32–34,38,39} The number of cases with PI was then calculated as the product of the estimated prevalence rate and the number of in-patient cases (discharges), sourced from Australian hospital statistics 2012–13.⁴³

We used the average treatment cost per case by stage (from Stage I to IV) estimated in a UK study by Dealey et al.¹⁶ Although data derived from Australia would be preferred, to our knowledge these data are not available. Jackson et al.³³ only provide the average cost per case, not disaggregated by stage, whereas Graves and Zheng³⁷ used estimated cost per case from the international literature. The system of PI staging in the UK is comparable to that used in Australia, as defined by the Pan Pacific Clinical Practice Guidelines.²⁷ In addition, the translation of costs related to use of dressings, medication, equipment and nursing time for the management of PIs is considered to be a reasonable approximation, given the similarities in the Australian and UK public health systems.

The treatment cost in Dealey et al.³² was estimated by the bottom-up costing method using 2011 data. It took into account daily resources for PI treatment, such as nursing time, special mattresses, dressings and medication. The marginal cost of bed day (£300) was included only for PI associated with cellulitis and osteomyelitis. To avoid double counting the bed day cost (which was included in the estimation of opportunity cost PI discussed below), we removed it from the treatment cost per case. All costs in UK 2011 prices were converted to Australian 2012–13 prices using the web tool by Shemilt et al.⁴⁴ (see Table 1). The stage-specific cost per case was generated from a uniform distribution with the minimum and maximum values specified as ±10% of the average cost per stage.

Table 1.  Treatment cost per case, by pressure injury stage and health state and mean cost per patients
Data were sourced from Dealey et al.¹⁶ Costs in A$ were calculated using the web-based tool developed by Shemilt et al.⁴⁴ PI, pressure injury

Total treatment cost was calculated as the product of the number of PI cases and the cost per case, and disaggregated by PI stage (using shares of PI cases by severity) and state (using number of discharges). The associated standard deviations were computed from a simulated sample generated by a mixed distribution of beta (the number of PI cases) and uniform (treatment cost per case) distributions.

From a societal perspective, the indirect opportunity cost of PI is represented in part as the costs associated with bed days lost due to preventable cases. That is, if PI is completely prevented, there would be more bed days available for treatment of other illness. In the literature, this value is approximated by ‘willingness to pay’ for those bed days.³² Other societal costs, such as those related to lost productivity for carers or the intangible cost arising from reduced quality of life, were not included due to a lack of data that would support their valuation.⁴⁵

The calculation of the opportunity cost associated with lost bed days due to PI involved: (1) identifying the number of bed days wasted due to PI (extended LOS); (2) calculating average opportunity cost per bed day; and (3) estimating the total opportunity cost of extended bed days.

The independent effect of PI on LOS estimated by Graves et al.³¹ was used in the analysis. The mean (± s.d.) estimate of 4.31 (± 1.26) days for additional LOS was used to specify the gamma distribution parameters (α and β) for the simulation. The disaggregation of LOS by PI stage was based on Dealey et al.¹⁶ due to a lack of data from Australian studies. This choice ensures consistency because treatment cost per case by stage was also sourced from the same study.

In the Australian public hospital context, the opportunity cost (or willingness to pay) for a bed day can be approximated by the average casemix-adjusted cost per bed day. The average cost per overnight discharge for each state of Australia was extracted from the Australian Hospital Statistic collection 2012–13.⁴³ Overnight discharge cost is used because PI is associated with extended LOS.

The total value of bed days lost due to PIs for each PI stage was then calculated as the product of the number of bed days lost and average cost per bed day. The former was estimated from the number of PI cases and the average extended LOS per case. Because the extended LOS was generated from a gamma distribution and the number of PI cases was generated from a beta distribution, the opportunity cost, as the product of extended LOS, PI case and cost per bed days, was generated from a mixed distribution.

The variables and assumptions underlying the estimation process are summarised in Table 2. We obtained point estimates and confidence intervals (CI) for the prevalence rates, LOS and treatment costs using appropriate probability distributions under the Monte Carlo simulation with 10 000 draws. Stata 13 (StataCorp, College Station, TX, USA) was used for all estimations.

Table 2.  Key variables for the estimations
MDC, major diagnostic category; PI, pressure injury; LOS, length of stay

Our estimates suggest that the prevalent number of cases of PI for Australian public hospitals for 2012–13 was 121 645 (95% CI 100 846–142 444). The resulting treatment cost across all states and PI stages was estimated to be A$983 million (95% CI A$815–1151 million) per annum, representing approximately 1.9% of all public hospital expenditure (A$42 billion in the same period)⁴⁶ or 0.6% of the public recurrent health expenditure (A$132 billion).⁴⁶

Tables 3 and 4 give the number of cases and associated treatment costs disaggregated by state and PI stage. As expected, the three largest states of New South Wales, Victoria and Queensland shouldered approximately 75% of the total cases and cost for Australia. The estimates also indicate that the large cost burdens came from Stage II and IV (30% of cost), the former due to its largest share of PI cases and the latter due to its high treatment cost per case.

Table 3.  Number of pressure injury (PI) cases per annum, by state and PI stage (2012–13)
NSW, New South Wales; Vic., Victoria; Qld, Queensland; WA, Western Australia; SA, South Australia; Tas., Tasmania; ACT, Australia Capitol Territory; NT, Northern Territory

Table 4.  Total cost of pressure injury (PI) treatment per annum by state and PI stage (2012–13)
NSW, New South Wales; Vic., Victoria; Qld, Queensland; WA, Western Australia; SA, South Australia; Tas., Tasmania; ACT, Australia Capitol Territory; NT, Northern Territory

The total number of bed days lost for Australia was estimated at 524 661 (95% CI 366 067–683 254) per annum and the resulting opportunity cost was valued at A$819 million per annum (95% CI A$572–1067 million). The disaggregation of extra bed days and opportunity costs by states are presented in Tables 5 and 6. Again, the greatest numbers of lost bed days and opportunity cost accrued to New South Wales, Victoria and Queensland.

Table 5.  Total number of extra bed days per annum, by state and pressure injury stage (2012–13)
NSW, New South Wales; Vic., Victoria; Qld, Queensland; WA, Western Australia; SA, South Australia; Tas., Tasmania; ACT, Australia Capitol Territory; NT, Northern Territory

Table 6.  Total annual opportunity cost of pressure injury (PI; A$ million), by state and PI stage (2012–13)
NSW, New South Wales; Vic., Victoria; Qld, Queensland; WA, Western Australia; SA, South Australia; Tas., Tasmania; ACT, Australia Capitol Territory; NT, Northern Territory

The total economic burden of PI to Australia, estimated by the sum of treatment and opportunity costs, amounted to $1.8 billion per annum, of which 55% was attributable to treatment cost. Because most PI cases were of Stage I and II (which are not generally anticipated to extend LOS), the overall estimated treatment costs were higher than the opportunity costs.

The present study provides the most comprehensive estimates of the economic burden of PI in Australia to date, including the treatment cost per case by PI stage, total cost of PI treatment by PI stage and for individual states, and the opportunity cost associated with bed days lost due to PI. We show that the estimated overall economic burden of PI for society is substantial.

Although our findings are generally consistent with the costs associated with PI treatment estimated in the literature, there are a few major differences between our estimates and those from comparable Australian studies. Compared with Graves and Zheng,³⁷ the number of PI cases per annum was smaller in the present study (121 645 vs 236 295). This was driven by assumptions regarding the distribution of the data and the range of prevalence rates. We used Australian data for the prevalence rates, whereas Graves and Zheng³⁷ used the rates derived from the international literature applied to Australian hospital figures for the number of discharges. Compared with state audit reports,¹⁰ our prevalence estimates by state were lower, representing a conservative approach to cost estimation. Our estimate for the average cost per PI case, based on Dealey et al, (A$8513) was lower than that reported in Jackson et al. ¹⁶ (A$9297), and higher than that of Graves and Zheng³⁶ (range $A2371–$A7139). Although Graves and Zheng ³⁶ extracted the average cost per PI case from the international literature, they did not include the estimates from Dealey et al. ¹⁶ and Jackson et al. ³³ This resulted in lower estimated treatment costs.

The cost estimate presented here is within the range estimated for other countries. Our estimate of total cost for PI treatment was lower than other high-income countries, both as a percentage of public hospital (1.9%) and public recurrent health expenditure (0.6%). Studies have indicated that treatment for PI costs range between 0.4 and 3.2% of the UK national health budget,^8,31 1.2% for The Netherlands and 5.2% for Spain.³⁰ Similarly, various US studies indicate that PI accounts for approximately 3.9% of in-patient hospital costs per annum.^9,41

Our cost estimate is likely to understate the total burden of PI for society because we did not include the treatment cost in long-term and home care settings, or the less tangible costs associated with quality of life lost and the opportunity cost of PI to patients and families. Nonetheless, our results indicate that PIs impose a large economic and importantly avoidable burden to Australia (although, still substantially less than the top diseases by expenditure). It would be possible to have a more comprehensive understanding of the economic burden of this adverse event if better data (e.g. prevalence rates, resources required for PI treatment and quality of life associated with PI) were available in both long-term and home care settings.

Although PI is not a completely avoidable adverse event, some jurisdictions have recently introduced financial penalties to incentivise hospitals to improve quality and avoid hospital-acquired PIs. Well-known examples include the US implementation of the hospital-acquired condition reduction program and the exclusion of certain PI cases considered preventable in the hospital setting from Medicare payment (since early 2008).^47,48 In Australia, Queensland’s activity-based funding model for 2012–13 also includes pressure injury Stage III and IV as adverse events for which there is a reduced payment for hospitals.⁴⁹ Because the penalty is designed in the form of a reduction in government reimbursement to public hospitals, it does not represent a cost from a health system perspective (although it does represents a cost from an individual hospital perspective). As such, these payments are not included in the current analysis. The extent to which these payments have an impact on improved quality of care and either the prevalence rates or reported cases of PI across different stages will become apparent with appropriate evaluation over the next few years.

The present study has several limitations, most of which relate to data availability. First, the data were derived from several different Australian and UK studies, and we assumed these data are generalisable to the Australian public hospital context. Nevertheless, we incorporated a range of values and distributions to capture the uncertainty. Second, we made various assumptions with regard to the uniformity of prevalence rates, extended LOS and treatment cost per case across different states of Australia. Subsequently, there is some degree of uncertainty involved in our final estimations. The total cost estimates fall well within the range indicated in the international literature.^8,9,30,31,41

The present study, together with the paucity of data available to support cost estimates, highlights an under-researched area. Further robust data are required to support estimates of the economic burden associated with this prevalent and costly adverse event and, importantly, comparative evaluation of the cost-effectiveness of strategies to prevent PIs. For instance, facilities should develop a systematic collection of PI incidence and prevalence data to inform health managers on strategic planning, resource allocation and to track improvement, especially when hospital-acquired PI is regarded as a quality indicator (Standard 8, Preventing and Managing Pressure Injury, of the National Safety Health Quality Standards is an example⁴²). In the present study, the impact of PI on the quality of life of patients, and their families, has been largely unexplored. However, the estimates available for time to heal in the literature, with even a Stage I PI estimated to take on average 28 days to heal,¹⁶ suggest the impact on quality of life may be expected to be considerable. Finally, data collection for PI in the long-term care setting (home and nursing home), such as prevalence, outpatient visits, medications, social services and informal care, should receive more funding attention. This would enable future research to fully capture the cost of PI outside the public hospital setting, which no doubt represents a substantial additional economic burden for the health system and for society as a whole.

None declared.