Longitudinal variation in pressure injury incidence among long-term aged care facilities

Longitudinal variation in pressure injury incidence among long-term aged care facilities Abstract Objective To examine variation in pressure injury (PI) incidence among long-term aged care facilities and identify resident- and facility-level factors that explain this variation. Design Longitudinal incidence study using routinely-collected electronic care management data. Setting A large aged care service provider in New South Wales and the Australian Capital Territory, Australia. Participants About 6556 people aged 65 years and older who were permanent residents in 60 long-term care facilities between December 2014 and November 2016. Main Outcome Measure Risk-adjusted PI incidence rates over eight study quarters. Results Incidence density over the study period was 1.33 pressure injuries per 1000 resident days (95% confidence interval (CI) = 1.29–1.37). Funnel plots were used to identify variation among facilities. On average, 14% of facilities had risk-adjusted PI rates that were higher than expected in each quarter (above 95% funnel plot control limits). Ten percent of facilities had persistently high rates in any three or more consecutive quarters (n = 6). The variation between facilities was only partly explained by resident characteristics in multilevel regression models. Residents were more likely to have higher-pressure injury rates in facilities in regional areas compared with major city areas (adjusted incidence rate ratio = 1.25, 95% CI = 1.04–1.51), and facilities with persistently high rates were more likely to be located in areas with low socioeconomic status (P = 0.038). Conclusions There is considerable variation among facilities in PI incidence. This study demonstrates the potential of routinely-collected care management data to monitor PI incidence and to identify facilities that may benefit from targeted intervention. pressure ulcer, long-term care, aged, quality indicators, medical informatics Introduction A pressure injury (PI)—also known as bedsore, pressure ulcer or decubitus ulcer—is a localised injury to the skin and/or underlying tissue resulting from sustained pressure, shear or a combination of these factors [1]. PIs are a major burden for individuals, their caregivers and the healthcare system [2]. In addition to the multi-billion dollar direct costs associated with treating PIs each year [2], PIs have significant social costs in terms of reduced quality of life, pain, and increased mortality [3–5]. PIs remain a problem in both acute and long-term care settings. Recent estimates of the proportion of people in long-term aged care who have a PI at a given point in time (prevalence) range between 1% and 46% [6]. In Australia, the prevalence of PIs in long-term care has been reported between 9% and 31% [7–9]. Because there is widespread consensus that the majority of PIs are avoidable [10], approaches to reduce the impact of PIs place a strong focus on prevention. For the individual, pressure-reducing support surfaces appear to be the most effective prevention strategy [11]. At an organisational level, PI rates are frequently used as a performance indicator in an attempt to drive changes in the quality of care provided by facilities [12, 13]. Clinical practice guidelines recommend targeting residents who are at high risk of developing a PI for prevention activities [14]. Across 54 studies, the most frequently cited factors associated with the development of PIs were reduced mobility or activity, variables related to perfusion (e.g. diabetes) and skin or PI status [15]. These factors are consistent with conceptual aetiological frameworks [16], which highlight the underlying conditions that contribute to the critical determinants of PI development—that is, prolonged and intense pressure and poor tissue tolerance. For example, a person with reduced mobility will have a greater likelihood of being exposed to sustained pressure due to their diminished ability to change body position, and their tissue tolerance for this pressure may be reduced due to friction from dragging during a position change [16]. Other factors commonly associated with PI development are based on clinical observation, for example, previous PIs as a warning of potential further deterioration [15]. The pervasive nature of PIs and the large number of studies examining their development certainly indicates that there is a complex interplay of factors, which increase the risk of PI incidence. A number of risk assessment tools have been developed to supplement clinical judgement of an individual’s risk [17]. Despite the focus on PIs as an indicator of care quality, there is much less research exploring how the risk of developing PIs varies between facilities. In a recent study, Hartmann et al. [18] described wide variation in PI prevalence rates among 132 nursing homes in the USA, although the authors did not examine influences at a facility level. A small number of studies have reported associations between facility-level factors and PI development [19–22], identifying factors that may preclude best practice PI prevention or quality improvement activities such as lack of clinical staff [20, 21] or resources [19]. However, these studies do not often account for clustering of residents within facilities. Traditional conceptual frameworks that are used for organising the knowledge base and identifying research gaps [16] also ignore the role of contributors to PI development that are extrinsic to the individual. More research is needed to quantify and understand the factors that are associated with variation in PI rates at the places where direct care is provided. Currently, the majority of research studies and many quality indicator programs rely on surveys of PI prevalence [6, 13]. While prevalence provides valuable information about the ongoing burden of PIs, capturing all new cases that develop over a period of time within the care setting (incidence) provides a better indicator of quality of care [23]. Comparing outcomes between facilities can also be difficult due to differences in resident acuity and facility size. Without adjusting for residents with known PI risk factors, facility performance will not be accurately ascertained [24]. Greater variability in PI rates is also more likely to arise by chance in facilities with small numbers of residents, and so the precision of the estimates should be taken into account [24, 25]. The aged care sector has lagged behind other parts of the health sector in adopting electronic records systems [26]. Where these systems are used as part of routine practice, more accurate and timely measurement of PI rates may be possible without the burden of periodic form-based surveys. The aim of this study was to use routinely-collected electronic care management data to examine variation in PI incidence among long-term aged care facilities and to identify resident- and facility-level factors that explain this variation. Methods Data source This study utilised routinely-collected longitudinal data from 60 long-term aged care facilities managed by Uniting, a not-for-profit organisation and the single largest provider of aged care services in New South Wales (NSW) and the Australian Capital Territory (ACT), Australia. Long-term aged care is largely government-funded in Australia [27]. Data were extracted from the clinical and care management platform used by Uniting, iCareHealth [28]. Information recorded in iCareHealth includes resident demographics, facility of residence, admission and discharge information, assessed care needs, medical diagnoses and wound-related data. Full adoption of mature and sophisticated electronic record systems across the sector were in early stages leading up to and across the study period. Study population The study population comprised people aged 65 years and older who were permanent residents in Uniting long-term aged care facilities between 1 December 2014 and 30 November 2016. Non-permanent residents excluded from the study included people receiving short-stay respite and transition care, as well those who did not have a completed Aged Care Funding Instrument (ACFI) to confirm their permanent status (see Appendix A). ACFI is used to assess the relative care needs of all permanent residents and is the mechanism for allocating government subsidies to aged care facilities for delivering care [29]. Facilities with more than two gaps of 7 days or more in recording any wound care information were also excluded (n = 16). Pressure injury incidence Information about each recorded PI included date of occurrence, anatomical location, qualification of person reporting the PI and dated entries describing the care management or status of the wound over time. Date of occurrence is the date the PI was first observed by a care provider. Stage of PI was not available. PIs reported within 72 h of a resident’s admission into the facility were excluded from all analyses. The incidence rate of PIs for every facility was calculated for each 3-month period (quarter) between 1 December 2014 and 30 November 2016. The primary measure of incidence used was incidence density, which is the number of new PIs per 1000 days of care. Incidence density is the recommended measure of PI incidence [14], as using the denominator of days of care allows easier comparison across different populations. Facility incidence density was adjusted for case-mix using negative binomial regression models. Rates were adjusted for age, sex, diabetes diagnosis, any new PI reported in a Uniting facility in the year prior to the start of the quarter, and need for physical assistance with mobility as defined by the ACFI. These clinical characteristics represent the most consistently reported predictors of PI development [15]. Each facility’s expected incidence rate was obtained by summing their residents’ predicted PI rates from the model, given their covariate values. The observed rate of PIs was divided by the expected rate and then multiplied by the mean study population incidence rate to obtain the risk-adjusted rate for each facility [30]. Determining facility variation Funnel plots and multilevel regression models were used to quantify facility variation in PI incidence density. Funnel plots are regularly used to identify outlying performers while accounting for the precision of the estimates [25]. In the funnel plots, risk-adjusted incidence rates for each facility were plotted against the number of permanent residents in the facility. For each quarter, 95% and 99.8% ‘control limits’ were calculated based on the mean study population incidence rate. Facilities with rates outside of these limits were considered outliers. Facilities with three or more consecutive quarters of rates outside the upper 95% limits were considered ‘persistent outliers’ [31]. Multilevel regression models take into account the clustering of residents within facilities and allow the partitioning of variance between levels. The following three-level models (facility–resident–quarter) were fitted using incidence data from across the study period: A null model (facility and resident random effects but no explanatory variables); A model with fixed resident effects (e.g. age) and facility/resident random effects; A model with fixed resident and facility-level effects (e.g. number of residents) and facility/resident random effects. Poisson models were fitted using a Markov chain Monte Carlo approach in MLwiN, and included an additional random effect to allow for overdispersion [32]. The variance partition coefficient (VPC) was calculated to compare the proportion of variability attributable to facilities after accounting for known resident and facility factors, using the exact formulae developed by Stryhn et al. [33]. Adjusted incidence rate ratios (IRRs) for resident and facility factors were obtained for the full model using a mixed-effects negative binomial model in Stata, and included a categorical random slope for study quarter. The median incidence rate ratio (MIRR) for the full model was also obtained to translate inter-facility variation into risk differences. The MIRR is the median of the rate ratios of pair-wise comparisons of incidence rates between any two facilities. Resident and facility variables used in the models are listed in Table 1. Assessed care needs were derived from the ACFI, which consists of 12 questions rated on a four-point scale [29]. Because of collinearity between ACFI scores, each of the 12 questions was dichotomised based on the highest level of need (see Appendix B). ACFI funding was calculated using the basic daily subsidy rates based on each resident’s raw ACFI scores [34]. Facility postcode was used to classify remoteness and socioeconomic status based on the Accessibility/Remoteness Index of Australia and the Index of Relative Socioeconomic Advantage and Disadvantage (IRSAD) [35]. People living in inner and outer regional areas have restrictions to accessibility of goods, services and opportunities for social interaction [35]. In the models, inner and outer regional facilities were combined as there was only one facility located in an outer regional area. Facilities in the lowest IRSAD quintile were compared with those in all other quintiles to determine whether more disadvantaged areas had higher PI incidence. Table 1 Characteristics of 6556 older permanent residents in 60 aged care facilities Resident characteristics  n (%)  Age at study start   65–74  718 (11.0)   75–84  2088 (31.8)   85–94  3233 (49.5)   ≥95  517 (7.9)  Sex   Female  4484 (68.4)   Male  2072 (31.6)  Marital status   Married/de facto  1566 (23.9)   Divorced/widowed/single  4642 (70.8)   Missing  348 (5.3)  Country of birth   Australia  4557 (69.5)   Other main English speaking country  721 (11.0)   Other country  925 (14.1)   Missing  353 (5.4)  Level of care as classified by aged care provider   High (24-h nursing care)  4881 (74.5)   Low (more independent)  1675 (25.5)  Alzheimer’s disease   Yes  2839 (43.3)   No  3717 (56.7)  Diabetes     Yes  1317 (20.1)   No  5239 (79.9)  Highest level of need on Aged Care Funding Instrument for   Personal hygiene  5427 (82.8)   Continence  5251 (80.1)   Toileting  4529 (69.1)   Verbal behaviour  3693 (56.3)   Mobility  3611 (55.1)   Medication  3431 (52.3)   Physical behaviour  2719 (41.5)   Complex healthcare  2186 (33.3)   Cognitive skills  2034 (31.0)   Nutrition (eating)  1451 (22.1)   Depression  1311 (20.0)   Wandering  607 (9.3)  Facility characteristics  N permanent residents in facility—median [IQR]  64 [44–96]  Remoteness of facility postcodea   Major city  43 (71.7)   Inner regional  16 (26.7)   Outer regional  1 (1.7)  Socioeconomic status of facility postcodeb   Lowest quintile  9 (15.0)   Other quintiles  51 (85.0)  % of facility PIs reported by Registered Nurse—median [IQR]  67 [35–94]  Care level of residents   >50% high care  48 (80.0)   ≤50% high care  12 (20.0)  Residents with Alzheimer’s disease (AD)   >50% with AD  20 (33.3)   ≤50% with AD  40 (66.7)  Resident characteristics  n (%)  Age at study start   65–74  718 (11.0)   75–84  2088 (31.8)   85–94  3233 (49.5)   ≥95  517 (7.9)  Sex   Female  4484 (68.4)   Male  2072 (31.6)  Marital status   Married/de facto  1566 (23.9)   Divorced/widowed/single  4642 (70.8)   Missing  348 (5.3)  Country of birth   Australia  4557 (69.5)   Other main English speaking country  721 (11.0)   Other country  925 (14.1)   Missing  353 (5.4)  Level of care as classified by aged care provider   High (24-h nursing care)  4881 (74.5)   Low (more independent)  1675 (25.5)  Alzheimer’s disease   Yes  2839 (43.3)   No  3717 (56.7)  Diabetes     Yes  1317 (20.1)   No  5239 (79.9)  Highest level of need on Aged Care Funding Instrument for   Personal hygiene  5427 (82.8)   Continence  5251 (80.1)   Toileting  4529 (69.1)   Verbal behaviour  3693 (56.3)   Mobility  3611 (55.1)   Medication  3431 (52.3)   Physical behaviour  2719 (41.5)   Complex healthcare  2186 (33.3)   Cognitive skills  2034 (31.0)   Nutrition (eating)  1451 (22.1)   Depression  1311 (20.0)   Wandering  607 (9.3)  Facility characteristics  N permanent residents in facility—median [IQR]  64 [44–96]  Remoteness of facility postcodea   Major city  43 (71.7)   Inner regional  16 (26.7)   Outer regional  1 (1.7)  Socioeconomic status of facility postcodeb   Lowest quintile  9 (15.0)   Other quintiles  51 (85.0)  % of facility PIs reported by Registered Nurse—median [IQR]  67 [35–94]  Care level of residents   >50% high care  48 (80.0)   ≤50% high care  12 (20.0)  Residents with Alzheimer’s disease (AD)   >50% with AD  20 (33.3)   ≤50% with AD  40 (66.7)  aAccessibility/Remoteness Index of Australia [35]. bIndex of Relative Socioeconomic Advantage and Disadvantage [35]. Table 1 Characteristics of 6556 older permanent residents in 60 aged care facilities Resident characteristics  n (%)  Age at study start   65–74  718 (11.0)   75–84  2088 (31.8)   85–94  3233 (49.5)   ≥95  517 (7.9)  Sex   Female  4484 (68.4)   Male  2072 (31.6)  Marital status   Married/de facto  1566 (23.9)   Divorced/widowed/single  4642 (70.8)   Missing  348 (5.3)  Country of birth   Australia  4557 (69.5)   Other main English speaking country  721 (11.0)   Other country  925 (14.1)   Missing  353 (5.4)  Level of care as classified by aged care provider   High (24-h nursing care)  4881 (74.5)   Low (more independent)  1675 (25.5)  Alzheimer’s disease   Yes  2839 (43.3)   No  3717 (56.7)  Diabetes     Yes  1317 (20.1)   No  5239 (79.9)  Highest level of need on Aged Care Funding Instrument for   Personal hygiene  5427 (82.8)   Continence  5251 (80.1)   Toileting  4529 (69.1)   Verbal behaviour  3693 (56.3)   Mobility  3611 (55.1)   Medication  3431 (52.3)   Physical behaviour  2719 (41.5)   Complex healthcare  2186 (33.3)   Cognitive skills  2034 (31.0)   Nutrition (eating)  1451 (22.1)   Depression  1311 (20.0)   Wandering  607 (9.3)  Facility characteristics  N permanent residents in facility—median [IQR]  64 [44–96]  Remoteness of facility postcodea   Major city  43 (71.7)   Inner regional  16 (26.7)   Outer regional  1 (1.7)  Socioeconomic status of facility postcodeb   Lowest quintile  9 (15.0)   Other quintiles  51 (85.0)  % of facility PIs reported by Registered Nurse—median [IQR]  67 [35–94]  Care level of residents   >50% high care  48 (80.0)   ≤50% high care  12 (20.0)  Residents with Alzheimer’s disease (AD)   >50% with AD  20 (33.3)   ≤50% with AD  40 (66.7)  Resident characteristics  n (%)  Age at study start   65–74  718 (11.0)   75–84  2088 (31.8)   85–94  3233 (49.5)   ≥95  517 (7.9)  Sex   Female  4484 (68.4)   Male  2072 (31.6)  Marital status   Married/de facto  1566 (23.9)   Divorced/widowed/single  4642 (70.8)   Missing  348 (5.3)  Country of birth   Australia  4557 (69.5)   Other main English speaking country  721 (11.0)   Other country  925 (14.1)   Missing  353 (5.4)  Level of care as classified by aged care provider   High (24-h nursing care)  4881 (74.5)   Low (more independent)  1675 (25.5)  Alzheimer’s disease   Yes  2839 (43.3)   No  3717 (56.7)  Diabetes     Yes  1317 (20.1)   No  5239 (79.9)  Highest level of need on Aged Care Funding Instrument for   Personal hygiene  5427 (82.8)   Continence  5251 (80.1)   Toileting  4529 (69.1)   Verbal behaviour  3693 (56.3)   Mobility  3611 (55.1)   Medication  3431 (52.3)   Physical behaviour  2719 (41.5)   Complex healthcare  2186 (33.3)   Cognitive skills  2034 (31.0)   Nutrition (eating)  1451 (22.1)   Depression  1311 (20.0)   Wandering  607 (9.3)  Facility characteristics  N permanent residents in facility—median [IQR]  64 [44–96]  Remoteness of facility postcodea   Major city  43 (71.7)   Inner regional  16 (26.7)   Outer regional  1 (1.7)  Socioeconomic status of facility postcodeb   Lowest quintile  9 (15.0)   Other quintiles  51 (85.0)  % of facility PIs reported by Registered Nurse—median [IQR]  67 [35–94]  Care level of residents   >50% high care  48 (80.0)   ≤50% high care  12 (20.0)  Residents with Alzheimer’s disease (AD)   >50% with AD  20 (33.3)   ≤50% with AD  40 (66.7)  aAccessibility/Remoteness Index of Australia [35]. bIndex of Relative Socioeconomic Advantage and Disadvantage [35]. Results Table 1 presents the characteristics of the 6556 older people who were permanent residents in 60 aged care facilities between 1 December 2014 and 30 November 2016. The majority of residents were female (n = 4484, 68.4%), aged 85 and over (n = 3750, 57.2%), and receiving 24-h nursing care (n = 4881, 74.5%). Of the 60 aged care facilities, 43 were located in a major city (71.7%) and the median number of beds was 64 (IQR = 44–96). Compared with the wider long-term aged care population of Australia [36], the study population had a similar proportion of women (P = 0.70) and people born in Australia (P = 0.83), but a greater proportion living in major cities (74.4% vs. 69.5%, P < 0.001). Pressure injury incidence There were 3984 recorded pressure injuries over the 24-month study period, excluding those identified within 72 h of admission. The majority of residents did not experience any PIs over the study period (n = 4785, 72.4%). Among those who did experience a PI, the median number of PIs was two (IQR = 1–3). The most common location for PIs was the gluteus maximus (35.2%), followed by sacrum/coccyx (13.5%), foot (12.8%), toe (11.8%), heel (8.6%), leg (3.9%) and ankle (3.6%). Registered Nurses reported the majority of PIs (70.6%), with Care Service Employees (19.0%) and Enroled/Endorsed Enroled Nurses (9.6%) reporting nearly all remaining PIs. The incidence density over the study period was 1.33 pressure injuries/1000 resident days (95% CI = 1.29–1.37). The lower-extremity incidence density was 0.54/1000 resident days (95% CI = 0.51–0.57). Incidence density fluctuated over the study period quarters (see Table 2), and was highest in the third quarter. Quarterly PI incidence ranged between 7.4% and 9.0%. Table 2 Incidence and variation in pressure injury rates by quarter for 60 aged care facilities Quarter  Incidence density (95% CI)a  Incidence (95% CI)b  N high outliers (%)c  N low outliers (%)c  1  1.19 (1.09–1.31)  7.39 (6.63–8.15)  9 (15.0)  9 (15.0)  2  1.33 (1.21–1.44)  7.78 (7.00–8.56)  10 (16.7)  7 (11.7)  3  1.54 (1.42–1.67)  8.99 (8.15–9.82)  7 (11.7)  13 (21.7)  4  1.39 (1.28–1.52)  8.38 (7.57–9.19)  5 (8.3)  8 (13.3)  5  1.28 (1.17–1.40)  8.31 (7.49–9.13)  7 (11.7)  12 (20.0)  6  1.25 (1.14–1.37)  7.98 (7.17–8.78)  13 (21.7)  16 (26.7)  7  1.33 (1.21–1.45)  8.21 (7.39–9.03)  9 (15.0)  14 (23.3)  8  1.31 (1.19–1.43)  7.89 (7.07–8.70)  8 (13.3)  11 (18.3)  Quarter  Incidence density (95% CI)a  Incidence (95% CI)b  N high outliers (%)c  N low outliers (%)c  1  1.19 (1.09–1.31)  7.39 (6.63–8.15)  9 (15.0)  9 (15.0)  2  1.33 (1.21–1.44)  7.78 (7.00–8.56)  10 (16.7)  7 (11.7)  3  1.54 (1.42–1.67)  8.99 (8.15–9.82)  7 (11.7)  13 (21.7)  4  1.39 (1.28–1.52)  8.38 (7.57–9.19)  5 (8.3)  8 (13.3)  5  1.28 (1.17–1.40)  8.31 (7.49–9.13)  7 (11.7)  12 (20.0)  6  1.25 (1.14–1.37)  7.98 (7.17–8.78)  13 (21.7)  16 (26.7)  7  1.33 (1.21–1.45)  8.21 (7.39–9.03)  9 (15.0)  14 (23.3)  8  1.31 (1.19–1.43)  7.89 (7.07–8.70)  8 (13.3)  11 (18.3)  aNumber of new PIs per 1000 resident days. bPercentage of people developing a new PI over quarter. cAbove (high) or below (low) 95% control limits on funnel plots. Table 2 Incidence and variation in pressure injury rates by quarter for 60 aged care facilities Quarter  Incidence density (95% CI)a  Incidence (95% CI)b  N high outliers (%)c  N low outliers (%)c  1  1.19 (1.09–1.31)  7.39 (6.63–8.15)  9 (15.0)  9 (15.0)  2  1.33 (1.21–1.44)  7.78 (7.00–8.56)  10 (16.7)  7 (11.7)  3  1.54 (1.42–1.67)  8.99 (8.15–9.82)  7 (11.7)  13 (21.7)  4  1.39 (1.28–1.52)  8.38 (7.57–9.19)  5 (8.3)  8 (13.3)  5  1.28 (1.17–1.40)  8.31 (7.49–9.13)  7 (11.7)  12 (20.0)  6  1.25 (1.14–1.37)  7.98 (7.17–8.78)  13 (21.7)  16 (26.7)  7  1.33 (1.21–1.45)  8.21 (7.39–9.03)  9 (15.0)  14 (23.3)  8  1.31 (1.19–1.43)  7.89 (7.07–8.70)  8 (13.3)  11 (18.3)  Quarter  Incidence density (95% CI)a  Incidence (95% CI)b  N high outliers (%)c  N low outliers (%)c  1  1.19 (1.09–1.31)  7.39 (6.63–8.15)  9 (15.0)  9 (15.0)  2  1.33 (1.21–1.44)  7.78 (7.00–8.56)  10 (16.7)  7 (11.7)  3  1.54 (1.42–1.67)  8.99 (8.15–9.82)  7 (11.7)  13 (21.7)  4  1.39 (1.28–1.52)  8.38 (7.57–9.19)  5 (8.3)  8 (13.3)  5  1.28 (1.17–1.40)  8.31 (7.49–9.13)  7 (11.7)  12 (20.0)  6  1.25 (1.14–1.37)  7.98 (7.17–8.78)  13 (21.7)  16 (26.7)  7  1.33 (1.21–1.45)  8.21 (7.39–9.03)  9 (15.0)  14 (23.3)  8  1.31 (1.19–1.43)  7.89 (7.07–8.70)  8 (13.3)  11 (18.3)  aNumber of new PIs per 1000 resident days. bPercentage of people developing a new PI over quarter. cAbove (high) or below (low) 95% control limits on funnel plots. Facility variation Unadjusted and risk-adjusted PI rates by facility for the first quarter are presented in Fig. 1 using funnel plots. There were fewer facilities above the 95% control limits when using risk-adjusted PI rates (n = 8, 13.3%) compared with unadjusted PI rates (n = 11, 18.3%). On average, 14% of facilities had risk-adjusted rates above the upper 95% control limits in each quarter. Figure 1 View largeDownload slide Funnel plots of unadjusted (A) and risk-adjusted (B) pressure injury rates for all facilities in first study quarter. Figure 1 View largeDownload slide Funnel plots of unadjusted (A) and risk-adjusted (B) pressure injury rates for all facilities in first study quarter. Figure 2 demonstrates how the performance of individual facilities can be tracked over time using funnel plots. Six facilities had persistently high outlying rates in any three or more consecutive quarters (10%). If three consecutive quarters of high rates only occurred by random chance, then persistently high outliers would be expected in less than half a percent of facilities (average 14% outliers ^ 3 quarters = 0.27%). It is therefore unlikely that persistently high outliers are being driven by chance alone. The PI incidence density for the six persistently high outlying facilities was nearly double that of other facilities (2.07 vs. 1.08/1000 resident days). Facilities with persistently high outlying rates were more likely to be located in areas with socioeconomic status in the lowest IRSAD quintile (Fisher’s exact P = 0.038). No other known facility characteristics were associated with persistently high outlying facilities. Figure 2 View largeDownload slide Risk-adjusted pressure injury rates following two facilities across study quarters. 95% and 99.8% limits reflect the funnel plot control limits based on number of residents of displayed facility. Figure 2 View largeDownload slide Risk-adjusted pressure injury rates following two facilities across study quarters. 95% and 99.8% limits reflect the funnel plot control limits based on number of residents of displayed facility. Resident and facility-level predictors of PI incidence in the full multilevel model are summarised in Table 3. Residents were more likely to have higher rates of PIs in facilities in regional areas (adjusted IRR = 1.25, 95% CI = 1.04–1.51). Adjusted incidence rates were higher in the middle of the study period (quarters 3, 4 and 5) compared with the first quarter. Table 3 Resident- and facility-level predictors of PI incidence in multilevel modela   Adjusted incidence rate ratio (95% CI)  Resident factors (n = 5947 residents)   Age at start of quarter (ref = 65–69)  **    70–74  1.07 (0.73–1.56)    75–79  1.15 (0.81–1.64)    80–84  1.19 (0.85–1.65)    85–89  1.58 (1.14–2.19)    90–94  1.86 (1.34–2.58)    95+  1.78 (1.24–2.56)   Male (ref = female)  1.32 (1.16–1.51)**   Married/de facto (ref = no partner)  1.12 (0.97–1.29)   Born in English speaking country (ref = non-English speaking)  1.02 (0.86–1.21)   High care (ref = low care)  1.16 (0.99–1.35)   Alzheimer’s disease (ref = no Alzheimer’s disease)  0.93 (0.82–1.06)   Diabetes (ref = no diabetes)  1.16 (1.01–1.33)*   Previous PI recorded (ref = no recorded PI)  1.66 (1.43–1.89)**   Length of stay at start of quarter  1.00 (0.99–1.00)   Mobilityb  1.87 (1.59–2.19)**   Complex healthcareb  1.68 (1.47–1.93)**   Nutrition (eating)b  1.54 (1.31–1.80)**   Medicationb  1.37 (1.20–1.55)**   Continenceb  1.30 (1.08–1.56)*   Cognitive skillsb  1.13 (0.97–1.31)   Depressionb  1.03 (0.88–1.21)   Physical behaviourb  0.98 (0.85–1.13)   Wanderingb  0.93 (0.75–1.17)   Verbal behaviourb  0.89 (0.78–1.02)   Study quarter (ref = 1)  **    2  1.20 (1.01–1.42)    3  1.53 (1.27–1.85)    4  1.54 (1.14–2.10)    5  1.46 (1.06–2.02)    6  1.35 (0.97–1.88)    7  1.44 (1.02–2.02)    8  1.24 (0.87–1.77)  Facility factors (n = 60 facilities)   Number of permanent residents in facility  1.00 (0.99–1.01)   Mean ACFI funding per resident day  0.99 (0.96–1.02)   Inner/outer regional area (ref = major city)  1.25 (1.04–1.51)*   Lowest quintile socioeconomic status (ref = other quintiles)  1.20 (0.81–1.78)   Proportion of PIs reported by Registered Nurse  1.00 (0.99–1.00)   >50% residents high care (ref ≤ 50%)  1.02 (0.79–1.31)   >50% residents with Alzheimer’s disease (ref ≤ 50%)  1.18 (0.96–1.46)    Adjusted incidence rate ratio (95% CI)  Resident factors (n = 5947 residents)   Age at start of quarter (ref = 65–69)  **    70–74  1.07 (0.73–1.56)    75–79  1.15 (0.81–1.64)    80–84  1.19 (0.85–1.65)    85–89  1.58 (1.14–2.19)    90–94  1.86 (1.34–2.58)    95+  1.78 (1.24–2.56)   Male (ref = female)  1.32 (1.16–1.51)**   Married/de facto (ref = no partner)  1.12 (0.97–1.29)   Born in English speaking country (ref = non-English speaking)  1.02 (0.86–1.21)   High care (ref = low care)  1.16 (0.99–1.35)   Alzheimer’s disease (ref = no Alzheimer’s disease)  0.93 (0.82–1.06)   Diabetes (ref = no diabetes)  1.16 (1.01–1.33)*   Previous PI recorded (ref = no recorded PI)  1.66 (1.43–1.89)**   Length of stay at start of quarter  1.00 (0.99–1.00)   Mobilityb  1.87 (1.59–2.19)**   Complex healthcareb  1.68 (1.47–1.93)**   Nutrition (eating)b  1.54 (1.31–1.80)**   Medicationb  1.37 (1.20–1.55)**   Continenceb  1.30 (1.08–1.56)*   Cognitive skillsb  1.13 (0.97–1.31)   Depressionb  1.03 (0.88–1.21)   Physical behaviourb  0.98 (0.85–1.13)   Wanderingb  0.93 (0.75–1.17)   Verbal behaviourb  0.89 (0.78–1.02)   Study quarter (ref = 1)  **    2  1.20 (1.01–1.42)    3  1.53 (1.27–1.85)    4  1.54 (1.14–2.10)    5  1.46 (1.06–2.02)    6  1.35 (0.97–1.88)    7  1.44 (1.02–2.02)    8  1.24 (0.87–1.77)  Facility factors (n = 60 facilities)   Number of permanent residents in facility  1.00 (0.99–1.01)   Mean ACFI funding per resident day  0.99 (0.96–1.02)   Inner/outer regional area (ref = major city)  1.25 (1.04–1.51)*   Lowest quintile socioeconomic status (ref = other quintiles)  1.20 (0.81–1.78)   Proportion of PIs reported by Registered Nurse  1.00 (0.99–1.00)   >50% residents high care (ref ≤ 50%)  1.02 (0.79–1.31)   >50% residents with Alzheimer’s disease (ref ≤ 50%)  1.18 (0.96–1.46)  *P < 0.05; **P < 0.001. aVariance partition coefficient = 0.17, median incidence rate ratio = 1.55. bHighest level of need on Aged Care Funding Instrument for each question (ref = other levels of need). Table 3 Resident- and facility-level predictors of PI incidence in multilevel modela   Adjusted incidence rate ratio (95% CI)  Resident factors (n = 5947 residents)   Age at start of quarter (ref = 65–69)  **    70–74  1.07 (0.73–1.56)    75–79  1.15 (0.81–1.64)    80–84  1.19 (0.85–1.65)    85–89  1.58 (1.14–2.19)    90–94  1.86 (1.34–2.58)    95+  1.78 (1.24–2.56)   Male (ref = female)  1.32 (1.16–1.51)**   Married/de facto (ref = no partner)  1.12 (0.97–1.29)   Born in English speaking country (ref = non-English speaking)  1.02 (0.86–1.21)   High care (ref = low care)  1.16 (0.99–1.35)   Alzheimer’s disease (ref = no Alzheimer’s disease)  0.93 (0.82–1.06)   Diabetes (ref = no diabetes)  1.16 (1.01–1.33)*   Previous PI recorded (ref = no recorded PI)  1.66 (1.43–1.89)**   Length of stay at start of quarter  1.00 (0.99–1.00)   Mobilityb  1.87 (1.59–2.19)**   Complex healthcareb  1.68 (1.47–1.93)**   Nutrition (eating)b  1.54 (1.31–1.80)**   Medicationb  1.37 (1.20–1.55)**   Continenceb  1.30 (1.08–1.56)*   Cognitive skillsb  1.13 (0.97–1.31)   Depressionb  1.03 (0.88–1.21)   Physical behaviourb  0.98 (0.85–1.13)   Wanderingb  0.93 (0.75–1.17)   Verbal behaviourb  0.89 (0.78–1.02)   Study quarter (ref = 1)  **    2  1.20 (1.01–1.42)    3  1.53 (1.27–1.85)    4  1.54 (1.14–2.10)    5  1.46 (1.06–2.02)    6  1.35 (0.97–1.88)    7  1.44 (1.02–2.02)    8  1.24 (0.87–1.77)  Facility factors (n = 60 facilities)   Number of permanent residents in facility  1.00 (0.99–1.01)   Mean ACFI funding per resident day  0.99 (0.96–1.02)   Inner/outer regional area (ref = major city)  1.25 (1.04–1.51)*   Lowest quintile socioeconomic status (ref = other quintiles)  1.20 (0.81–1.78)   Proportion of PIs reported by Registered Nurse  1.00 (0.99–1.00)   >50% residents high care (ref ≤ 50%)  1.02 (0.79–1.31)   >50% residents with Alzheimer’s disease (ref ≤ 50%)  1.18 (0.96–1.46)    Adjusted incidence rate ratio (95% CI)  Resident factors (n = 5947 residents)   Age at start of quarter (ref = 65–69)  **    70–74  1.07 (0.73–1.56)    75–79  1.15 (0.81–1.64)    80–84  1.19 (0.85–1.65)    85–89  1.58 (1.14–2.19)    90–94  1.86 (1.34–2.58)    95+  1.78 (1.24–2.56)   Male (ref = female)  1.32 (1.16–1.51)**   Married/de facto (ref = no partner)  1.12 (0.97–1.29)   Born in English speaking country (ref = non-English speaking)  1.02 (0.86–1.21)   High care (ref = low care)  1.16 (0.99–1.35)   Alzheimer’s disease (ref = no Alzheimer’s disease)  0.93 (0.82–1.06)   Diabetes (ref = no diabetes)  1.16 (1.01–1.33)*   Previous PI recorded (ref = no recorded PI)  1.66 (1.43–1.89)**   Length of stay at start of quarter  1.00 (0.99–1.00)   Mobilityb  1.87 (1.59–2.19)**   Complex healthcareb  1.68 (1.47–1.93)**   Nutrition (eating)b  1.54 (1.31–1.80)**   Medicationb  1.37 (1.20–1.55)**   Continenceb  1.30 (1.08–1.56)*   Cognitive skillsb  1.13 (0.97–1.31)   Depressionb  1.03 (0.88–1.21)   Physical behaviourb  0.98 (0.85–1.13)   Wanderingb  0.93 (0.75–1.17)   Verbal behaviourb  0.89 (0.78–1.02)   Study quarter (ref = 1)  **    2  1.20 (1.01–1.42)    3  1.53 (1.27–1.85)    4  1.54 (1.14–2.10)    5  1.46 (1.06–2.02)    6  1.35 (0.97–1.88)    7  1.44 (1.02–2.02)    8  1.24 (0.87–1.77)  Facility factors (n = 60 facilities)   Number of permanent residents in facility  1.00 (0.99–1.01)   Mean ACFI funding per resident day  0.99 (0.96–1.02)   Inner/outer regional area (ref = major city)  1.25 (1.04–1.51)*   Lowest quintile socioeconomic status (ref = other quintiles)  1.20 (0.81–1.78)   Proportion of PIs reported by Registered Nurse  1.00 (0.99–1.00)   >50% residents high care (ref ≤ 50%)  1.02 (0.79–1.31)   >50% residents with Alzheimer’s disease (ref ≤ 50%)  1.18 (0.96–1.46)  *P < 0.05; **P < 0.001. aVariance partition coefficient = 0.17, median incidence rate ratio = 1.55. bHighest level of need on Aged Care Funding Instrument for each question (ref = other levels of need). Resident factors explained some of the inter-facility variation in PI incidence rates, as seen by the drop in the variance partition coefficient (VPC) from the null model (VPC = 0.75) to the model with resident factors (VPC = 0.35). Adding facility characteristics to the model partly explained the remaining inter-facility variation (VPC = 0.17). The median incidence rate ratio after accounting for known resident and facility factors was 1.55. The model with both resident and facility factors had a better fit (deviance information criterion [DIC]=17 361) than the model with resident factors only (DIC = 18 169) or the null model (DIC = 20 073). Discussion Routinely-collected electronic care management data were used in this study to examine variation in PI incidence among 60 long-term aged care facilities between December 2014 and November 2016. Both funnel plots and the results of multilevel models demonstrated that there was substantial variation between facilities in PI rates that was only partly explained by resident characteristics. This study is one of very few to examine incidence density of PIs in long-term aged care. Despite an international consensus that incidence provides a clearer measure of quality of care than prevalence [37], a recent systematic review identified only four peer-reviewed studies describing PI incidence in long-term care [6]. One of these studies reported the incidence density of lower-extremity PIs in Japan as 0.46 per 1000 resident days [38], which is comparable to the lower-extremity rate for our study population of 0.54 per 1000 resident days. While incidence studies have traditionally been more time-consuming and expensive to conduct than prevalence studies, our study demonstrates the potential of using routinely-collected electronic care management data to benchmark and meaningfully track this care indicator over time. This could provide a less burdensome and more methodologically-sound alternative for quality indicator programs where periodic surveys of PI prevalence are still often used [12, 13]. The majority of the research literature has focused on individual risk factors for PI development. Consistent with these studies [15], we found that older age, being male, having diabetes, previous PIs, incontinence, and poor mobility, were associated with higher PI rates. The reduction in the number of outlying facilities between the unadjusted and risk-adjusted funnel plots confirms the importance of accounting for commonly reported individual-level PI risk factors to ensure valid comparison of facility performance [24]. However, as in a recent longitudinal analysis of PI prevalence in nursing homes in the US [18], there remained considerable variation between facilities even after adjusting for resident risk factors. We have advanced the work of this previous study by using multilevel models to partition the total variability in PI incidence between levels, finding that 35% of the inter-facility variation remained attributable to the facility level after accounting for known resident factors (VPC = 0.35). The high MIRR (1.55) in the final model confirmed these findings, suggesting that the facility in which a person resides is as important as a risk factor for PI incidence as many individual-level risk factors. When an outcome is relatively rare and sample size small, high outcome rates are more likely to occur by chance. This study utilised funnel plots to identify outlying facilities while accounting for the increased variability in performance expected from small facilities [25]. We also measured persistence of high PI rates over time, as a way of distinguishing entrenched poor performance from random or intermittently high rates that later regress toward the mean [31]. Persistently high outlying facilities in our study had a PI incidence density almost twice that of other facilities. Targeting these facilities for intervention activities could therefore have the greatest impact in reducing the total number of PIs. Previous research has identified a number of common characteristics of facilities with higher rates of PIs, including having no nurses [20], no medical director or director of nursing [21], lower Medicaid reimbursement rates [19], and higher proportion of black residents [22]. In our study, persistently high outlying facilities were more likely to be located in areas with low socioeconomic status, and residents were more likely to have higher rates of PIs in facilities in regional areas. While our study did not find higher rates of PIs for those born in a non-English speaking country or in facilities with a higher proportion of PIs reported by a registered nurse, socioeconomic disparities appear to be a consistent underlying factor across both our study and previous research. This suggests more clinical support and greater resources may be needed to ensure all facilities are able to undertake best practice PI prevention and quality improvement activities. This study has a number of limitations. The functionality of the software platform, together with its implementation and usage across all facilities and workforce within the study period, presented some challenges. PI stage was not available, and so it is unknown at what severity PIs were detected by facility staff. However, our quarterly PI incidence of between 7% and 9% is consistent with another small study of four Australian facilities, where 8% of residents experienced a PI over a 3-month period [39]. While the ACFI was able to provide information about a broad range of resident care needs, other resident-level factors that have been associated with PI development, such as skin moisture or albumin levels [15], were not able to be accounted for. We were also not able to examine additional facility-level factors that may help to explain the remaining variation in PI rates, including the use of pressure-reducing support surfaces, direct nurse ratios and other quality improvement activities. The socioeconomic status variable was dichotomised to reduce the potential for classifying now gentrified areas as advantaged where the resident population may have lived in the area for a long period of time. However, this limits further exploration of the relationship between socioeconomic status and PI development. Although we excluded facilities with gaps in wound recording and were able to identify facilities with consistently high PI rates, some of the facility variation may also be explained by differences in PI recording, as will be the case wherever care providers are responsible for reporting. A strength of this study is the use of data from the largest provider of aged care in NSW/ACT, which contained a representative proportion of people born in non-English speaking countries and a sizable proportion of people living outside major city areas. However, given the generally low levels of digital maturity in Australian long-term aged care [40] and research suggesting facilities with higher levels of IT sophistication have better performance on quality measures [41], further work needs to be undertaken to determine whether PI incidence is poorer in less digitally-mature facilities, and whether the facility-level factors identified in this study are reflected across other Australian providers. Finally, an emphasis on the occurrence of PIs, in this study and much of the literature, applies a narrow focus to measuring the impact of PIs. Incorporating additional indicators such as the effect of PIs on quality of life [42], or the rate of PI healing [43], may be useful to evaluate the quality of care provided after a PI has occurred. Routinely-collected electronic care management data were able to be used in this study to track and compare PI incidence among facilities over time. Further exploration is needed to identify the elements that have made prevention efforts successful in facilities with low PI rates, and the reasons why facilities with certain characteristics are at risk of poor performance. The wide variation between facilities in PI rates certainly indicates that there is scope for targeted intervention to reduce the incidence of PIs in long-term care. Supplementary material Supplementary material is available at International Journal for Quality in Health Care online. Acknowledgements The authors thank members of the project Steering Committee, particularly Uniting staff who provided access to the data, assisted in the interpretation of the dataset, and provided their expertise to the project. Funding This work was supported by an Australian Research Council Linkage Grant undertaken in partnership with Uniting [grant number LP120200814]. Author contributions Study design—M.J., J.S., A.G., J.I.W.; Data acquisition—A.G., J.I.W.; Data analysis—M.J.; Data interpretation—all authors; Manuscript drafting—M.J.; Critical revision—J.S., A.G., J.I.W.; Final approval—all authors. Ethical approval The study was approved by the Macquarie University Human Research Ethics committee [reference number 5201401031]. References 1 National Pressure Ulcer Advisory Panel, European Pressure Ulcer Advisory Panel and Pan Pacific Pressure Injury Alliance. Haesler E (ed). Prevention and treatment of pressure ulcers: quick reference guide . Osborne Park, Australia: Cambridge Media, 2014. https://www.npuap.org/wp-content/uploads/2014/08/Updated-10-16-14-Quick-Reference-Guide-DIGITAL-NPUAP-EPUAP-PPPIA-16Oct2014.pdf 2 Sen CK, Gordillo GM, Roy S et al.  . Human skin wounds: a major and snowballing threat to public health and the economy. Wound Repair Regen  2009; 17: 763– 71. Google Scholar CrossRef Search ADS PubMed  3 Berlowitz DR, Brandeis GH, Anderson J et al.  . Effect of pressure ulcers on the survival of long-term care residents. J Gerontol Series A  1997; 52: M106– M10. Google Scholar CrossRef Search ADS   4 Gorecki C, Brown JM, Nelson EA et al.  . Impact of pressure ulcers on quality of life in older patients: a systematic review. J Am Geriatr Soc  2009; 57: 1175– 83. Google Scholar CrossRef Search ADS PubMed  5 Pieper B, Langemo D, Cuddigan J. Pressure ulcer pain: a systematic literature review and National Pressure Ulcer Advisory Panel white paper. Ostomy Wound Manage  2009; 55: 16– 31. Google Scholar PubMed  6 Hahnel E, Lichterfeld A, Blume-Peytavi U et al.  . The epidemiology of skin conditions in the aged: a systematic review. J Tissue Viability  2017; 26: 20– 8. Google Scholar CrossRef Search ADS PubMed  7 Santamaria N, Carville K, Prentice J et al.  . Reducing pressure ulcer prevalence in residential aged care: results from phase II of the PRIME trial. Wound Pract Res  2009; 17: 12– 22. 8 Banks M, Bauer J, Graves N et al.  . Malnutrition and pressure ulcer risk in adults in Australian health care facilities. Nutrition  2010; 26: 896– 901. Google Scholar CrossRef Search ADS PubMed  9 Courtney M, O’Reilly MT, Edwards H et al.  . Benchmarking clinical indicators of quality for Australian residential aged care facilities. Aust Health Rev  2010; 34: 93– 100. Google Scholar CrossRef Search ADS PubMed  10 Black JM, Edsberg LE, Baharestani MM et al.  . Pressure ulcers: avoidable or unavoidable? Results of the National Pressure Ulcer Advisory Panel consensus conference. Ostomy Wound Manage  2011; 57: 24– 37. Google Scholar PubMed  11 Chou R, Dana T, Bougatsos C et al.  . Pressure ulcer risk assessment and prevention: a systematic comparative effectiveness review. Ann Intern Med  2013; 159: 28– 38. Google Scholar CrossRef Search ADS PubMed  12 Castle NG, Ferguson JC. What is nursing home quality and how is it measured? Gerontologist  2010; 50: 426– 42. Google Scholar CrossRef Search ADS PubMed  13 Australian Government Department of Health. National Aged Care Quality Indicator Programme. Available at: https://agedcare.health.gov.au/ensuring-quality/quality-indicators/about-the-national-aged-care-quality-indicator-programme. Accessed 26 September, 2017. 14 National Pressure Ulcer Advisory Panel. Pressure ulcer incidence density as a quality measure. Press Release: 11 March 2014. Available at: http://www.npuap.org/pressure-ulcer-incidence-density-as-a-quality-measure/. Accessed 26 September, 2017. 15 Coleman S, Gorecki C, Nelson EA et al.  . Patient risk factors for pressure ulcer development: systematic review. Int J Nurs Stud  2013; 50: 974– 1003. Google Scholar CrossRef Search ADS PubMed  16 Braden B, Bergstrom N. A conceptual schema for the study of the etiology of pressure sores. Rehabil Nurs  2000; 25: 105– 10. Google Scholar CrossRef Search ADS   17 Moore ZE, Cowman S. Risk assessment tools for the prevention of pressure ulcers. The Cochrane Libr  2014; 5: CD006471. 18 Hartmann CW, Shwartz M, Zhao S et al.  . Longitudinal pressure ulcer rates after adoption of culture change in veterans health administration nursing homes. J Am Geriatr Soc  2016; 64: 151– 5. Google Scholar CrossRef Search ADS PubMed  19 Grabowski DC, Angelelli JJ. The relationship of medicaid payment rates, bed constraint policies, and risk‐adjusted pressure ulcers. Health Serv Res  2004; 39: 793– 812. Google Scholar CrossRef Search ADS PubMed  20 Kwong EWy, Pang SMc, Aboo GH et al.  . Pressure ulcer development in older residents in nursing homes: influencing factors. J Adv Nurs  2009; 65: 2608– 20. Google Scholar CrossRef Search ADS PubMed  21 Kang Y, Tzeng H-M, Miller NA. Facility characteristics and risk of developing pressure ulcers in US nursing homes. J Nurs Care Qual  2016; 31: E9– E16. Google Scholar CrossRef Search ADS PubMed  22 Li Y, Yin J, Cai X et al.  . Association of race and sites of care with pressure ulcers in high-risk nursing home residents. JAMA  2011; 306: 179– 86. Google Scholar CrossRef Search ADS PubMed  23 Berlowitz D. Incidence and prevalence of pressure ulcers. In: Thomas DR, Compton GA (eds). Pressure ulcers in the aging population: a guide for clinicians . Springer: Humana Press, 2014. 24 Berlowitz DR, Anderson JJ, Ash AS et al.  . Reducing random variation in reported rates of pressure ulcer development. Medical Care  1998; 36: 818– 25. Google Scholar CrossRef Search ADS PubMed  25 Spiegelhalter DJ. Funnel plots for comparing institutional performance. Stat Med  2005; 24: 1185– 202. Google Scholar CrossRef Search ADS PubMed  26 Towne SD Jr, Lee S, Li Y et al.  . Assessment of eHealth capabilities and utilization in residential care settings. Health Informatics J  2016; 22: 1063– 75. Google Scholar CrossRef Search ADS PubMed  27 Australian Government, Aged Care Financing Authority. Fourth Report on the Funding and Financing of the Aged Care Sector. July 2016. Available at: https://agedcare.health.gov.au/sites/g/files/net1426/f/documents/10_2016/acfa_annual_report_on_funding_and_financing_of_the_aged_care_industry_2016.pdf. Accessed 26 September, 2017. 28 Telstra Health. iCareHealth – Clinical and Care Management Software. Available at: http://www.icarehealth.com.au/. Accessed 26 September, 2017. 29 Australian Government, Department of Health. Aged Care Funding Instrument. User Guide 2017. Available at: https://agedcare.health.gov.au/acfi-user-guide. Accessed 26 September, 2017. 30 Iezzoni LI. Risk adjustment for medical effectiveness research: an overview of conceptual and methodological considerations. J Investigat Med  1995; 43: 136. 31 Duckett S, Griffiths K Perils of place: identifying hotspots of health inequalities. Grattan Institute; 2016. Available at: https://grattan.edu.au/wp-content/uploads/2016/07/874-Perils-of-Place.pdf. Accessed 26 September, 2017. 32 Gelman A, Hill J. Data analysis using regression and multilevel/hierarchical models . New York, NY, USA: Cambridge University Press, 2007. 33 Stryhn H, Sanchez J, Morley P et al.  , editors. Interpretation of variance parameters in multilevel Poisson regression models. Proceedings of the 11th International Symposium on Veterinary Epidemiology and Economics; 2006. 34 Australian Government, Department of Health. Basic Subsidy Amount (Aged Care Funding Instrument). Available at: https://agedcare.health.gov.au/aged-care-funding/residential-care-subsidy/basic-subsidy-amount-aged-care-funding-instrument. Accessed 26 September, 2017. 35 Australian Bureau of Statistics. Australian Statistical Geography Standard (ASGS): Correspondences, July 2011. Cat. no. 1270.0.55.006. Canberra: ABS. 36 Australian Institute of Health and Welfare. Residential aged care and home care 2014–15. Cat. no. WEB ; 133. Canberra: AIHW, 2016. 37 Baharestani MM, Black JM, Carville K et al.  . Dilemmas in measuring and using pressure ulcer prevalence and incidence: an international consensus. International Wound. Journal  2009; 6: 97– 104. 38 Okuwa M, Sanada H, Sugama J et al.  . A prospective cohort study of lower-extremity pressure ulcer risk among bedfast older adults. Adv Skin Wound Care  2006; 19: 391– 7. Google Scholar CrossRef Search ADS PubMed  39 Moore KJ, Doyle CJ, Dunning TL et al.  . Public sector residential aged care: identifying novel associations between quality indicators and other demographic and health-related factors. Aust Health Rev  2014; 38: 325– 31. Google Scholar CrossRef Search ADS PubMed  40 Flinders University & the Aged Care Industry IT Council. A technology roadmap for the Australian aged care sector. June 2017. Available at: http://aciitc.com.au/wp-content/uploads/2017/06/ACIITC_TechnologyRoadmap_2017.pdf 41 Alexander G, Madsen R. IT sophistication and quality measures in nursing homes. J Gerontol Nurs  2009; 35: 22– 7. Google Scholar CrossRef Search ADS PubMed  42 Gorecki C, Nixon J, Lamping DL et al.  . Patient-reported outcome measures for chronic wounds with particular reference to pressure ulcer research: a systematic review. Int J Nurs Stud  2014; 51: 157– 65. Google Scholar CrossRef Search ADS PubMed  43 Bates‐Jensen BM, MacLean CH. Quality indicators for the care of pressure ulcers in vulnerable elders. J Am Geriatr Soc  2007; 55: S409– 16. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press in association with the International Society for Quality in Health Care. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png International Journal for Quality in Health Care Oxford University Press

Longitudinal variation in pressure injury incidence among long-term aged care facilities

Loading next page...
 
/lp/ou_press/longitudinal-variation-in-pressure-injury-incidence-among-long-term-hUKU6ToB6F
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press in association with the International Society for Quality in Health Care. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
1353-4505
eISSN
1464-3677
D.O.I.
10.1093/intqhc/mzy087
Publisher site
See Article on Publisher Site

Abstract

Abstract Objective To examine variation in pressure injury (PI) incidence among long-term aged care facilities and identify resident- and facility-level factors that explain this variation. Design Longitudinal incidence study using routinely-collected electronic care management data. Setting A large aged care service provider in New South Wales and the Australian Capital Territory, Australia. Participants About 6556 people aged 65 years and older who were permanent residents in 60 long-term care facilities between December 2014 and November 2016. Main Outcome Measure Risk-adjusted PI incidence rates over eight study quarters. Results Incidence density over the study period was 1.33 pressure injuries per 1000 resident days (95% confidence interval (CI) = 1.29–1.37). Funnel plots were used to identify variation among facilities. On average, 14% of facilities had risk-adjusted PI rates that were higher than expected in each quarter (above 95% funnel plot control limits). Ten percent of facilities had persistently high rates in any three or more consecutive quarters (n = 6). The variation between facilities was only partly explained by resident characteristics in multilevel regression models. Residents were more likely to have higher-pressure injury rates in facilities in regional areas compared with major city areas (adjusted incidence rate ratio = 1.25, 95% CI = 1.04–1.51), and facilities with persistently high rates were more likely to be located in areas with low socioeconomic status (P = 0.038). Conclusions There is considerable variation among facilities in PI incidence. This study demonstrates the potential of routinely-collected care management data to monitor PI incidence and to identify facilities that may benefit from targeted intervention. pressure ulcer, long-term care, aged, quality indicators, medical informatics Introduction A pressure injury (PI)—also known as bedsore, pressure ulcer or decubitus ulcer—is a localised injury to the skin and/or underlying tissue resulting from sustained pressure, shear or a combination of these factors [1]. PIs are a major burden for individuals, their caregivers and the healthcare system [2]. In addition to the multi-billion dollar direct costs associated with treating PIs each year [2], PIs have significant social costs in terms of reduced quality of life, pain, and increased mortality [3–5]. PIs remain a problem in both acute and long-term care settings. Recent estimates of the proportion of people in long-term aged care who have a PI at a given point in time (prevalence) range between 1% and 46% [6]. In Australia, the prevalence of PIs in long-term care has been reported between 9% and 31% [7–9]. Because there is widespread consensus that the majority of PIs are avoidable [10], approaches to reduce the impact of PIs place a strong focus on prevention. For the individual, pressure-reducing support surfaces appear to be the most effective prevention strategy [11]. At an organisational level, PI rates are frequently used as a performance indicator in an attempt to drive changes in the quality of care provided by facilities [12, 13]. Clinical practice guidelines recommend targeting residents who are at high risk of developing a PI for prevention activities [14]. Across 54 studies, the most frequently cited factors associated with the development of PIs were reduced mobility or activity, variables related to perfusion (e.g. diabetes) and skin or PI status [15]. These factors are consistent with conceptual aetiological frameworks [16], which highlight the underlying conditions that contribute to the critical determinants of PI development—that is, prolonged and intense pressure and poor tissue tolerance. For example, a person with reduced mobility will have a greater likelihood of being exposed to sustained pressure due to their diminished ability to change body position, and their tissue tolerance for this pressure may be reduced due to friction from dragging during a position change [16]. Other factors commonly associated with PI development are based on clinical observation, for example, previous PIs as a warning of potential further deterioration [15]. The pervasive nature of PIs and the large number of studies examining their development certainly indicates that there is a complex interplay of factors, which increase the risk of PI incidence. A number of risk assessment tools have been developed to supplement clinical judgement of an individual’s risk [17]. Despite the focus on PIs as an indicator of care quality, there is much less research exploring how the risk of developing PIs varies between facilities. In a recent study, Hartmann et al. [18] described wide variation in PI prevalence rates among 132 nursing homes in the USA, although the authors did not examine influences at a facility level. A small number of studies have reported associations between facility-level factors and PI development [19–22], identifying factors that may preclude best practice PI prevention or quality improvement activities such as lack of clinical staff [20, 21] or resources [19]. However, these studies do not often account for clustering of residents within facilities. Traditional conceptual frameworks that are used for organising the knowledge base and identifying research gaps [16] also ignore the role of contributors to PI development that are extrinsic to the individual. More research is needed to quantify and understand the factors that are associated with variation in PI rates at the places where direct care is provided. Currently, the majority of research studies and many quality indicator programs rely on surveys of PI prevalence [6, 13]. While prevalence provides valuable information about the ongoing burden of PIs, capturing all new cases that develop over a period of time within the care setting (incidence) provides a better indicator of quality of care [23]. Comparing outcomes between facilities can also be difficult due to differences in resident acuity and facility size. Without adjusting for residents with known PI risk factors, facility performance will not be accurately ascertained [24]. Greater variability in PI rates is also more likely to arise by chance in facilities with small numbers of residents, and so the precision of the estimates should be taken into account [24, 25]. The aged care sector has lagged behind other parts of the health sector in adopting electronic records systems [26]. Where these systems are used as part of routine practice, more accurate and timely measurement of PI rates may be possible without the burden of periodic form-based surveys. The aim of this study was to use routinely-collected electronic care management data to examine variation in PI incidence among long-term aged care facilities and to identify resident- and facility-level factors that explain this variation. Methods Data source This study utilised routinely-collected longitudinal data from 60 long-term aged care facilities managed by Uniting, a not-for-profit organisation and the single largest provider of aged care services in New South Wales (NSW) and the Australian Capital Territory (ACT), Australia. Long-term aged care is largely government-funded in Australia [27]. Data were extracted from the clinical and care management platform used by Uniting, iCareHealth [28]. Information recorded in iCareHealth includes resident demographics, facility of residence, admission and discharge information, assessed care needs, medical diagnoses and wound-related data. Full adoption of mature and sophisticated electronic record systems across the sector were in early stages leading up to and across the study period. Study population The study population comprised people aged 65 years and older who were permanent residents in Uniting long-term aged care facilities between 1 December 2014 and 30 November 2016. Non-permanent residents excluded from the study included people receiving short-stay respite and transition care, as well those who did not have a completed Aged Care Funding Instrument (ACFI) to confirm their permanent status (see Appendix A). ACFI is used to assess the relative care needs of all permanent residents and is the mechanism for allocating government subsidies to aged care facilities for delivering care [29]. Facilities with more than two gaps of 7 days or more in recording any wound care information were also excluded (n = 16). Pressure injury incidence Information about each recorded PI included date of occurrence, anatomical location, qualification of person reporting the PI and dated entries describing the care management or status of the wound over time. Date of occurrence is the date the PI was first observed by a care provider. Stage of PI was not available. PIs reported within 72 h of a resident’s admission into the facility were excluded from all analyses. The incidence rate of PIs for every facility was calculated for each 3-month period (quarter) between 1 December 2014 and 30 November 2016. The primary measure of incidence used was incidence density, which is the number of new PIs per 1000 days of care. Incidence density is the recommended measure of PI incidence [14], as using the denominator of days of care allows easier comparison across different populations. Facility incidence density was adjusted for case-mix using negative binomial regression models. Rates were adjusted for age, sex, diabetes diagnosis, any new PI reported in a Uniting facility in the year prior to the start of the quarter, and need for physical assistance with mobility as defined by the ACFI. These clinical characteristics represent the most consistently reported predictors of PI development [15]. Each facility’s expected incidence rate was obtained by summing their residents’ predicted PI rates from the model, given their covariate values. The observed rate of PIs was divided by the expected rate and then multiplied by the mean study population incidence rate to obtain the risk-adjusted rate for each facility [30]. Determining facility variation Funnel plots and multilevel regression models were used to quantify facility variation in PI incidence density. Funnel plots are regularly used to identify outlying performers while accounting for the precision of the estimates [25]. In the funnel plots, risk-adjusted incidence rates for each facility were plotted against the number of permanent residents in the facility. For each quarter, 95% and 99.8% ‘control limits’ were calculated based on the mean study population incidence rate. Facilities with rates outside of these limits were considered outliers. Facilities with three or more consecutive quarters of rates outside the upper 95% limits were considered ‘persistent outliers’ [31]. Multilevel regression models take into account the clustering of residents within facilities and allow the partitioning of variance between levels. The following three-level models (facility–resident–quarter) were fitted using incidence data from across the study period: A null model (facility and resident random effects but no explanatory variables); A model with fixed resident effects (e.g. age) and facility/resident random effects; A model with fixed resident and facility-level effects (e.g. number of residents) and facility/resident random effects. Poisson models were fitted using a Markov chain Monte Carlo approach in MLwiN, and included an additional random effect to allow for overdispersion [32]. The variance partition coefficient (VPC) was calculated to compare the proportion of variability attributable to facilities after accounting for known resident and facility factors, using the exact formulae developed by Stryhn et al. [33]. Adjusted incidence rate ratios (IRRs) for resident and facility factors were obtained for the full model using a mixed-effects negative binomial model in Stata, and included a categorical random slope for study quarter. The median incidence rate ratio (MIRR) for the full model was also obtained to translate inter-facility variation into risk differences. The MIRR is the median of the rate ratios of pair-wise comparisons of incidence rates between any two facilities. Resident and facility variables used in the models are listed in Table 1. Assessed care needs were derived from the ACFI, which consists of 12 questions rated on a four-point scale [29]. Because of collinearity between ACFI scores, each of the 12 questions was dichotomised based on the highest level of need (see Appendix B). ACFI funding was calculated using the basic daily subsidy rates based on each resident’s raw ACFI scores [34]. Facility postcode was used to classify remoteness and socioeconomic status based on the Accessibility/Remoteness Index of Australia and the Index of Relative Socioeconomic Advantage and Disadvantage (IRSAD) [35]. People living in inner and outer regional areas have restrictions to accessibility of goods, services and opportunities for social interaction [35]. In the models, inner and outer regional facilities were combined as there was only one facility located in an outer regional area. Facilities in the lowest IRSAD quintile were compared with those in all other quintiles to determine whether more disadvantaged areas had higher PI incidence. Table 1 Characteristics of 6556 older permanent residents in 60 aged care facilities Resident characteristics  n (%)  Age at study start   65–74  718 (11.0)   75–84  2088 (31.8)   85–94  3233 (49.5)   ≥95  517 (7.9)  Sex   Female  4484 (68.4)   Male  2072 (31.6)  Marital status   Married/de facto  1566 (23.9)   Divorced/widowed/single  4642 (70.8)   Missing  348 (5.3)  Country of birth   Australia  4557 (69.5)   Other main English speaking country  721 (11.0)   Other country  925 (14.1)   Missing  353 (5.4)  Level of care as classified by aged care provider   High (24-h nursing care)  4881 (74.5)   Low (more independent)  1675 (25.5)  Alzheimer’s disease   Yes  2839 (43.3)   No  3717 (56.7)  Diabetes     Yes  1317 (20.1)   No  5239 (79.9)  Highest level of need on Aged Care Funding Instrument for   Personal hygiene  5427 (82.8)   Continence  5251 (80.1)   Toileting  4529 (69.1)   Verbal behaviour  3693 (56.3)   Mobility  3611 (55.1)   Medication  3431 (52.3)   Physical behaviour  2719 (41.5)   Complex healthcare  2186 (33.3)   Cognitive skills  2034 (31.0)   Nutrition (eating)  1451 (22.1)   Depression  1311 (20.0)   Wandering  607 (9.3)  Facility characteristics  N permanent residents in facility—median [IQR]  64 [44–96]  Remoteness of facility postcodea   Major city  43 (71.7)   Inner regional  16 (26.7)   Outer regional  1 (1.7)  Socioeconomic status of facility postcodeb   Lowest quintile  9 (15.0)   Other quintiles  51 (85.0)  % of facility PIs reported by Registered Nurse—median [IQR]  67 [35–94]  Care level of residents   >50% high care  48 (80.0)   ≤50% high care  12 (20.0)  Residents with Alzheimer’s disease (AD)   >50% with AD  20 (33.3)   ≤50% with AD  40 (66.7)  Resident characteristics  n (%)  Age at study start   65–74  718 (11.0)   75–84  2088 (31.8)   85–94  3233 (49.5)   ≥95  517 (7.9)  Sex   Female  4484 (68.4)   Male  2072 (31.6)  Marital status   Married/de facto  1566 (23.9)   Divorced/widowed/single  4642 (70.8)   Missing  348 (5.3)  Country of birth   Australia  4557 (69.5)   Other main English speaking country  721 (11.0)   Other country  925 (14.1)   Missing  353 (5.4)  Level of care as classified by aged care provider   High (24-h nursing care)  4881 (74.5)   Low (more independent)  1675 (25.5)  Alzheimer’s disease   Yes  2839 (43.3)   No  3717 (56.7)  Diabetes     Yes  1317 (20.1)   No  5239 (79.9)  Highest level of need on Aged Care Funding Instrument for   Personal hygiene  5427 (82.8)   Continence  5251 (80.1)   Toileting  4529 (69.1)   Verbal behaviour  3693 (56.3)   Mobility  3611 (55.1)   Medication  3431 (52.3)   Physical behaviour  2719 (41.5)   Complex healthcare  2186 (33.3)   Cognitive skills  2034 (31.0)   Nutrition (eating)  1451 (22.1)   Depression  1311 (20.0)   Wandering  607 (9.3)  Facility characteristics  N permanent residents in facility—median [IQR]  64 [44–96]  Remoteness of facility postcodea   Major city  43 (71.7)   Inner regional  16 (26.7)   Outer regional  1 (1.7)  Socioeconomic status of facility postcodeb   Lowest quintile  9 (15.0)   Other quintiles  51 (85.0)  % of facility PIs reported by Registered Nurse—median [IQR]  67 [35–94]  Care level of residents   >50% high care  48 (80.0)   ≤50% high care  12 (20.0)  Residents with Alzheimer’s disease (AD)   >50% with AD  20 (33.3)   ≤50% with AD  40 (66.7)  aAccessibility/Remoteness Index of Australia [35]. bIndex of Relative Socioeconomic Advantage and Disadvantage [35]. Table 1 Characteristics of 6556 older permanent residents in 60 aged care facilities Resident characteristics  n (%)  Age at study start   65–74  718 (11.0)   75–84  2088 (31.8)   85–94  3233 (49.5)   ≥95  517 (7.9)  Sex   Female  4484 (68.4)   Male  2072 (31.6)  Marital status   Married/de facto  1566 (23.9)   Divorced/widowed/single  4642 (70.8)   Missing  348 (5.3)  Country of birth   Australia  4557 (69.5)   Other main English speaking country  721 (11.0)   Other country  925 (14.1)   Missing  353 (5.4)  Level of care as classified by aged care provider   High (24-h nursing care)  4881 (74.5)   Low (more independent)  1675 (25.5)  Alzheimer’s disease   Yes  2839 (43.3)   No  3717 (56.7)  Diabetes     Yes  1317 (20.1)   No  5239 (79.9)  Highest level of need on Aged Care Funding Instrument for   Personal hygiene  5427 (82.8)   Continence  5251 (80.1)   Toileting  4529 (69.1)   Verbal behaviour  3693 (56.3)   Mobility  3611 (55.1)   Medication  3431 (52.3)   Physical behaviour  2719 (41.5)   Complex healthcare  2186 (33.3)   Cognitive skills  2034 (31.0)   Nutrition (eating)  1451 (22.1)   Depression  1311 (20.0)   Wandering  607 (9.3)  Facility characteristics  N permanent residents in facility—median [IQR]  64 [44–96]  Remoteness of facility postcodea   Major city  43 (71.7)   Inner regional  16 (26.7)   Outer regional  1 (1.7)  Socioeconomic status of facility postcodeb   Lowest quintile  9 (15.0)   Other quintiles  51 (85.0)  % of facility PIs reported by Registered Nurse—median [IQR]  67 [35–94]  Care level of residents   >50% high care  48 (80.0)   ≤50% high care  12 (20.0)  Residents with Alzheimer’s disease (AD)   >50% with AD  20 (33.3)   ≤50% with AD  40 (66.7)  Resident characteristics  n (%)  Age at study start   65–74  718 (11.0)   75–84  2088 (31.8)   85–94  3233 (49.5)   ≥95  517 (7.9)  Sex   Female  4484 (68.4)   Male  2072 (31.6)  Marital status   Married/de facto  1566 (23.9)   Divorced/widowed/single  4642 (70.8)   Missing  348 (5.3)  Country of birth   Australia  4557 (69.5)   Other main English speaking country  721 (11.0)   Other country  925 (14.1)   Missing  353 (5.4)  Level of care as classified by aged care provider   High (24-h nursing care)  4881 (74.5)   Low (more independent)  1675 (25.5)  Alzheimer’s disease   Yes  2839 (43.3)   No  3717 (56.7)  Diabetes     Yes  1317 (20.1)   No  5239 (79.9)  Highest level of need on Aged Care Funding Instrument for   Personal hygiene  5427 (82.8)   Continence  5251 (80.1)   Toileting  4529 (69.1)   Verbal behaviour  3693 (56.3)   Mobility  3611 (55.1)   Medication  3431 (52.3)   Physical behaviour  2719 (41.5)   Complex healthcare  2186 (33.3)   Cognitive skills  2034 (31.0)   Nutrition (eating)  1451 (22.1)   Depression  1311 (20.0)   Wandering  607 (9.3)  Facility characteristics  N permanent residents in facility—median [IQR]  64 [44–96]  Remoteness of facility postcodea   Major city  43 (71.7)   Inner regional  16 (26.7)   Outer regional  1 (1.7)  Socioeconomic status of facility postcodeb   Lowest quintile  9 (15.0)   Other quintiles  51 (85.0)  % of facility PIs reported by Registered Nurse—median [IQR]  67 [35–94]  Care level of residents   >50% high care  48 (80.0)   ≤50% high care  12 (20.0)  Residents with Alzheimer’s disease (AD)   >50% with AD  20 (33.3)   ≤50% with AD  40 (66.7)  aAccessibility/Remoteness Index of Australia [35]. bIndex of Relative Socioeconomic Advantage and Disadvantage [35]. Results Table 1 presents the characteristics of the 6556 older people who were permanent residents in 60 aged care facilities between 1 December 2014 and 30 November 2016. The majority of residents were female (n = 4484, 68.4%), aged 85 and over (n = 3750, 57.2%), and receiving 24-h nursing care (n = 4881, 74.5%). Of the 60 aged care facilities, 43 were located in a major city (71.7%) and the median number of beds was 64 (IQR = 44–96). Compared with the wider long-term aged care population of Australia [36], the study population had a similar proportion of women (P = 0.70) and people born in Australia (P = 0.83), but a greater proportion living in major cities (74.4% vs. 69.5%, P < 0.001). Pressure injury incidence There were 3984 recorded pressure injuries over the 24-month study period, excluding those identified within 72 h of admission. The majority of residents did not experience any PIs over the study period (n = 4785, 72.4%). Among those who did experience a PI, the median number of PIs was two (IQR = 1–3). The most common location for PIs was the gluteus maximus (35.2%), followed by sacrum/coccyx (13.5%), foot (12.8%), toe (11.8%), heel (8.6%), leg (3.9%) and ankle (3.6%). Registered Nurses reported the majority of PIs (70.6%), with Care Service Employees (19.0%) and Enroled/Endorsed Enroled Nurses (9.6%) reporting nearly all remaining PIs. The incidence density over the study period was 1.33 pressure injuries/1000 resident days (95% CI = 1.29–1.37). The lower-extremity incidence density was 0.54/1000 resident days (95% CI = 0.51–0.57). Incidence density fluctuated over the study period quarters (see Table 2), and was highest in the third quarter. Quarterly PI incidence ranged between 7.4% and 9.0%. Table 2 Incidence and variation in pressure injury rates by quarter for 60 aged care facilities Quarter  Incidence density (95% CI)a  Incidence (95% CI)b  N high outliers (%)c  N low outliers (%)c  1  1.19 (1.09–1.31)  7.39 (6.63–8.15)  9 (15.0)  9 (15.0)  2  1.33 (1.21–1.44)  7.78 (7.00–8.56)  10 (16.7)  7 (11.7)  3  1.54 (1.42–1.67)  8.99 (8.15–9.82)  7 (11.7)  13 (21.7)  4  1.39 (1.28–1.52)  8.38 (7.57–9.19)  5 (8.3)  8 (13.3)  5  1.28 (1.17–1.40)  8.31 (7.49–9.13)  7 (11.7)  12 (20.0)  6  1.25 (1.14–1.37)  7.98 (7.17–8.78)  13 (21.7)  16 (26.7)  7  1.33 (1.21–1.45)  8.21 (7.39–9.03)  9 (15.0)  14 (23.3)  8  1.31 (1.19–1.43)  7.89 (7.07–8.70)  8 (13.3)  11 (18.3)  Quarter  Incidence density (95% CI)a  Incidence (95% CI)b  N high outliers (%)c  N low outliers (%)c  1  1.19 (1.09–1.31)  7.39 (6.63–8.15)  9 (15.0)  9 (15.0)  2  1.33 (1.21–1.44)  7.78 (7.00–8.56)  10 (16.7)  7 (11.7)  3  1.54 (1.42–1.67)  8.99 (8.15–9.82)  7 (11.7)  13 (21.7)  4  1.39 (1.28–1.52)  8.38 (7.57–9.19)  5 (8.3)  8 (13.3)  5  1.28 (1.17–1.40)  8.31 (7.49–9.13)  7 (11.7)  12 (20.0)  6  1.25 (1.14–1.37)  7.98 (7.17–8.78)  13 (21.7)  16 (26.7)  7  1.33 (1.21–1.45)  8.21 (7.39–9.03)  9 (15.0)  14 (23.3)  8  1.31 (1.19–1.43)  7.89 (7.07–8.70)  8 (13.3)  11 (18.3)  aNumber of new PIs per 1000 resident days. bPercentage of people developing a new PI over quarter. cAbove (high) or below (low) 95% control limits on funnel plots. Table 2 Incidence and variation in pressure injury rates by quarter for 60 aged care facilities Quarter  Incidence density (95% CI)a  Incidence (95% CI)b  N high outliers (%)c  N low outliers (%)c  1  1.19 (1.09–1.31)  7.39 (6.63–8.15)  9 (15.0)  9 (15.0)  2  1.33 (1.21–1.44)  7.78 (7.00–8.56)  10 (16.7)  7 (11.7)  3  1.54 (1.42–1.67)  8.99 (8.15–9.82)  7 (11.7)  13 (21.7)  4  1.39 (1.28–1.52)  8.38 (7.57–9.19)  5 (8.3)  8 (13.3)  5  1.28 (1.17–1.40)  8.31 (7.49–9.13)  7 (11.7)  12 (20.0)  6  1.25 (1.14–1.37)  7.98 (7.17–8.78)  13 (21.7)  16 (26.7)  7  1.33 (1.21–1.45)  8.21 (7.39–9.03)  9 (15.0)  14 (23.3)  8  1.31 (1.19–1.43)  7.89 (7.07–8.70)  8 (13.3)  11 (18.3)  Quarter  Incidence density (95% CI)a  Incidence (95% CI)b  N high outliers (%)c  N low outliers (%)c  1  1.19 (1.09–1.31)  7.39 (6.63–8.15)  9 (15.0)  9 (15.0)  2  1.33 (1.21–1.44)  7.78 (7.00–8.56)  10 (16.7)  7 (11.7)  3  1.54 (1.42–1.67)  8.99 (8.15–9.82)  7 (11.7)  13 (21.7)  4  1.39 (1.28–1.52)  8.38 (7.57–9.19)  5 (8.3)  8 (13.3)  5  1.28 (1.17–1.40)  8.31 (7.49–9.13)  7 (11.7)  12 (20.0)  6  1.25 (1.14–1.37)  7.98 (7.17–8.78)  13 (21.7)  16 (26.7)  7  1.33 (1.21–1.45)  8.21 (7.39–9.03)  9 (15.0)  14 (23.3)  8  1.31 (1.19–1.43)  7.89 (7.07–8.70)  8 (13.3)  11 (18.3)  aNumber of new PIs per 1000 resident days. bPercentage of people developing a new PI over quarter. cAbove (high) or below (low) 95% control limits on funnel plots. Facility variation Unadjusted and risk-adjusted PI rates by facility for the first quarter are presented in Fig. 1 using funnel plots. There were fewer facilities above the 95% control limits when using risk-adjusted PI rates (n = 8, 13.3%) compared with unadjusted PI rates (n = 11, 18.3%). On average, 14% of facilities had risk-adjusted rates above the upper 95% control limits in each quarter. Figure 1 View largeDownload slide Funnel plots of unadjusted (A) and risk-adjusted (B) pressure injury rates for all facilities in first study quarter. Figure 1 View largeDownload slide Funnel plots of unadjusted (A) and risk-adjusted (B) pressure injury rates for all facilities in first study quarter. Figure 2 demonstrates how the performance of individual facilities can be tracked over time using funnel plots. Six facilities had persistently high outlying rates in any three or more consecutive quarters (10%). If three consecutive quarters of high rates only occurred by random chance, then persistently high outliers would be expected in less than half a percent of facilities (average 14% outliers ^ 3 quarters = 0.27%). It is therefore unlikely that persistently high outliers are being driven by chance alone. The PI incidence density for the six persistently high outlying facilities was nearly double that of other facilities (2.07 vs. 1.08/1000 resident days). Facilities with persistently high outlying rates were more likely to be located in areas with socioeconomic status in the lowest IRSAD quintile (Fisher’s exact P = 0.038). No other known facility characteristics were associated with persistently high outlying facilities. Figure 2 View largeDownload slide Risk-adjusted pressure injury rates following two facilities across study quarters. 95% and 99.8% limits reflect the funnel plot control limits based on number of residents of displayed facility. Figure 2 View largeDownload slide Risk-adjusted pressure injury rates following two facilities across study quarters. 95% and 99.8% limits reflect the funnel plot control limits based on number of residents of displayed facility. Resident and facility-level predictors of PI incidence in the full multilevel model are summarised in Table 3. Residents were more likely to have higher rates of PIs in facilities in regional areas (adjusted IRR = 1.25, 95% CI = 1.04–1.51). Adjusted incidence rates were higher in the middle of the study period (quarters 3, 4 and 5) compared with the first quarter. Table 3 Resident- and facility-level predictors of PI incidence in multilevel modela   Adjusted incidence rate ratio (95% CI)  Resident factors (n = 5947 residents)   Age at start of quarter (ref = 65–69)  **    70–74  1.07 (0.73–1.56)    75–79  1.15 (0.81–1.64)    80–84  1.19 (0.85–1.65)    85–89  1.58 (1.14–2.19)    90–94  1.86 (1.34–2.58)    95+  1.78 (1.24–2.56)   Male (ref = female)  1.32 (1.16–1.51)**   Married/de facto (ref = no partner)  1.12 (0.97–1.29)   Born in English speaking country (ref = non-English speaking)  1.02 (0.86–1.21)   High care (ref = low care)  1.16 (0.99–1.35)   Alzheimer’s disease (ref = no Alzheimer’s disease)  0.93 (0.82–1.06)   Diabetes (ref = no diabetes)  1.16 (1.01–1.33)*   Previous PI recorded (ref = no recorded PI)  1.66 (1.43–1.89)**   Length of stay at start of quarter  1.00 (0.99–1.00)   Mobilityb  1.87 (1.59–2.19)**   Complex healthcareb  1.68 (1.47–1.93)**   Nutrition (eating)b  1.54 (1.31–1.80)**   Medicationb  1.37 (1.20–1.55)**   Continenceb  1.30 (1.08–1.56)*   Cognitive skillsb  1.13 (0.97–1.31)   Depressionb  1.03 (0.88–1.21)   Physical behaviourb  0.98 (0.85–1.13)   Wanderingb  0.93 (0.75–1.17)   Verbal behaviourb  0.89 (0.78–1.02)   Study quarter (ref = 1)  **    2  1.20 (1.01–1.42)    3  1.53 (1.27–1.85)    4  1.54 (1.14–2.10)    5  1.46 (1.06–2.02)    6  1.35 (0.97–1.88)    7  1.44 (1.02–2.02)    8  1.24 (0.87–1.77)  Facility factors (n = 60 facilities)   Number of permanent residents in facility  1.00 (0.99–1.01)   Mean ACFI funding per resident day  0.99 (0.96–1.02)   Inner/outer regional area (ref = major city)  1.25 (1.04–1.51)*   Lowest quintile socioeconomic status (ref = other quintiles)  1.20 (0.81–1.78)   Proportion of PIs reported by Registered Nurse  1.00 (0.99–1.00)   >50% residents high care (ref ≤ 50%)  1.02 (0.79–1.31)   >50% residents with Alzheimer’s disease (ref ≤ 50%)  1.18 (0.96–1.46)    Adjusted incidence rate ratio (95% CI)  Resident factors (n = 5947 residents)   Age at start of quarter (ref = 65–69)  **    70–74  1.07 (0.73–1.56)    75–79  1.15 (0.81–1.64)    80–84  1.19 (0.85–1.65)    85–89  1.58 (1.14–2.19)    90–94  1.86 (1.34–2.58)    95+  1.78 (1.24–2.56)   Male (ref = female)  1.32 (1.16–1.51)**   Married/de facto (ref = no partner)  1.12 (0.97–1.29)   Born in English speaking country (ref = non-English speaking)  1.02 (0.86–1.21)   High care (ref = low care)  1.16 (0.99–1.35)   Alzheimer’s disease (ref = no Alzheimer’s disease)  0.93 (0.82–1.06)   Diabetes (ref = no diabetes)  1.16 (1.01–1.33)*   Previous PI recorded (ref = no recorded PI)  1.66 (1.43–1.89)**   Length of stay at start of quarter  1.00 (0.99–1.00)   Mobilityb  1.87 (1.59–2.19)**   Complex healthcareb  1.68 (1.47–1.93)**   Nutrition (eating)b  1.54 (1.31–1.80)**   Medicationb  1.37 (1.20–1.55)**   Continenceb  1.30 (1.08–1.56)*   Cognitive skillsb  1.13 (0.97–1.31)   Depressionb  1.03 (0.88–1.21)   Physical behaviourb  0.98 (0.85–1.13)   Wanderingb  0.93 (0.75–1.17)   Verbal behaviourb  0.89 (0.78–1.02)   Study quarter (ref = 1)  **    2  1.20 (1.01–1.42)    3  1.53 (1.27–1.85)    4  1.54 (1.14–2.10)    5  1.46 (1.06–2.02)    6  1.35 (0.97–1.88)    7  1.44 (1.02–2.02)    8  1.24 (0.87–1.77)  Facility factors (n = 60 facilities)   Number of permanent residents in facility  1.00 (0.99–1.01)   Mean ACFI funding per resident day  0.99 (0.96–1.02)   Inner/outer regional area (ref = major city)  1.25 (1.04–1.51)*   Lowest quintile socioeconomic status (ref = other quintiles)  1.20 (0.81–1.78)   Proportion of PIs reported by Registered Nurse  1.00 (0.99–1.00)   >50% residents high care (ref ≤ 50%)  1.02 (0.79–1.31)   >50% residents with Alzheimer’s disease (ref ≤ 50%)  1.18 (0.96–1.46)  *P < 0.05; **P < 0.001. aVariance partition coefficient = 0.17, median incidence rate ratio = 1.55. bHighest level of need on Aged Care Funding Instrument for each question (ref = other levels of need). Table 3 Resident- and facility-level predictors of PI incidence in multilevel modela   Adjusted incidence rate ratio (95% CI)  Resident factors (n = 5947 residents)   Age at start of quarter (ref = 65–69)  **    70–74  1.07 (0.73–1.56)    75–79  1.15 (0.81–1.64)    80–84  1.19 (0.85–1.65)    85–89  1.58 (1.14–2.19)    90–94  1.86 (1.34–2.58)    95+  1.78 (1.24–2.56)   Male (ref = female)  1.32 (1.16–1.51)**   Married/de facto (ref = no partner)  1.12 (0.97–1.29)   Born in English speaking country (ref = non-English speaking)  1.02 (0.86–1.21)   High care (ref = low care)  1.16 (0.99–1.35)   Alzheimer’s disease (ref = no Alzheimer’s disease)  0.93 (0.82–1.06)   Diabetes (ref = no diabetes)  1.16 (1.01–1.33)*   Previous PI recorded (ref = no recorded PI)  1.66 (1.43–1.89)**   Length of stay at start of quarter  1.00 (0.99–1.00)   Mobilityb  1.87 (1.59–2.19)**   Complex healthcareb  1.68 (1.47–1.93)**   Nutrition (eating)b  1.54 (1.31–1.80)**   Medicationb  1.37 (1.20–1.55)**   Continenceb  1.30 (1.08–1.56)*   Cognitive skillsb  1.13 (0.97–1.31)   Depressionb  1.03 (0.88–1.21)   Physical behaviourb  0.98 (0.85–1.13)   Wanderingb  0.93 (0.75–1.17)   Verbal behaviourb  0.89 (0.78–1.02)   Study quarter (ref = 1)  **    2  1.20 (1.01–1.42)    3  1.53 (1.27–1.85)    4  1.54 (1.14–2.10)    5  1.46 (1.06–2.02)    6  1.35 (0.97–1.88)    7  1.44 (1.02–2.02)    8  1.24 (0.87–1.77)  Facility factors (n = 60 facilities)   Number of permanent residents in facility  1.00 (0.99–1.01)   Mean ACFI funding per resident day  0.99 (0.96–1.02)   Inner/outer regional area (ref = major city)  1.25 (1.04–1.51)*   Lowest quintile socioeconomic status (ref = other quintiles)  1.20 (0.81–1.78)   Proportion of PIs reported by Registered Nurse  1.00 (0.99–1.00)   >50% residents high care (ref ≤ 50%)  1.02 (0.79–1.31)   >50% residents with Alzheimer’s disease (ref ≤ 50%)  1.18 (0.96–1.46)    Adjusted incidence rate ratio (95% CI)  Resident factors (n = 5947 residents)   Age at start of quarter (ref = 65–69)  **    70–74  1.07 (0.73–1.56)    75–79  1.15 (0.81–1.64)    80–84  1.19 (0.85–1.65)    85–89  1.58 (1.14–2.19)    90–94  1.86 (1.34–2.58)    95+  1.78 (1.24–2.56)   Male (ref = female)  1.32 (1.16–1.51)**   Married/de facto (ref = no partner)  1.12 (0.97–1.29)   Born in English speaking country (ref = non-English speaking)  1.02 (0.86–1.21)   High care (ref = low care)  1.16 (0.99–1.35)   Alzheimer’s disease (ref = no Alzheimer’s disease)  0.93 (0.82–1.06)   Diabetes (ref = no diabetes)  1.16 (1.01–1.33)*   Previous PI recorded (ref = no recorded PI)  1.66 (1.43–1.89)**   Length of stay at start of quarter  1.00 (0.99–1.00)   Mobilityb  1.87 (1.59–2.19)**   Complex healthcareb  1.68 (1.47–1.93)**   Nutrition (eating)b  1.54 (1.31–1.80)**   Medicationb  1.37 (1.20–1.55)**   Continenceb  1.30 (1.08–1.56)*   Cognitive skillsb  1.13 (0.97–1.31)   Depressionb  1.03 (0.88–1.21)   Physical behaviourb  0.98 (0.85–1.13)   Wanderingb  0.93 (0.75–1.17)   Verbal behaviourb  0.89 (0.78–1.02)   Study quarter (ref = 1)  **    2  1.20 (1.01–1.42)    3  1.53 (1.27–1.85)    4  1.54 (1.14–2.10)    5  1.46 (1.06–2.02)    6  1.35 (0.97–1.88)    7  1.44 (1.02–2.02)    8  1.24 (0.87–1.77)  Facility factors (n = 60 facilities)   Number of permanent residents in facility  1.00 (0.99–1.01)   Mean ACFI funding per resident day  0.99 (0.96–1.02)   Inner/outer regional area (ref = major city)  1.25 (1.04–1.51)*   Lowest quintile socioeconomic status (ref = other quintiles)  1.20 (0.81–1.78)   Proportion of PIs reported by Registered Nurse  1.00 (0.99–1.00)   >50% residents high care (ref ≤ 50%)  1.02 (0.79–1.31)   >50% residents with Alzheimer’s disease (ref ≤ 50%)  1.18 (0.96–1.46)  *P < 0.05; **P < 0.001. aVariance partition coefficient = 0.17, median incidence rate ratio = 1.55. bHighest level of need on Aged Care Funding Instrument for each question (ref = other levels of need). Resident factors explained some of the inter-facility variation in PI incidence rates, as seen by the drop in the variance partition coefficient (VPC) from the null model (VPC = 0.75) to the model with resident factors (VPC = 0.35). Adding facility characteristics to the model partly explained the remaining inter-facility variation (VPC = 0.17). The median incidence rate ratio after accounting for known resident and facility factors was 1.55. The model with both resident and facility factors had a better fit (deviance information criterion [DIC]=17 361) than the model with resident factors only (DIC = 18 169) or the null model (DIC = 20 073). Discussion Routinely-collected electronic care management data were used in this study to examine variation in PI incidence among 60 long-term aged care facilities between December 2014 and November 2016. Both funnel plots and the results of multilevel models demonstrated that there was substantial variation between facilities in PI rates that was only partly explained by resident characteristics. This study is one of very few to examine incidence density of PIs in long-term aged care. Despite an international consensus that incidence provides a clearer measure of quality of care than prevalence [37], a recent systematic review identified only four peer-reviewed studies describing PI incidence in long-term care [6]. One of these studies reported the incidence density of lower-extremity PIs in Japan as 0.46 per 1000 resident days [38], which is comparable to the lower-extremity rate for our study population of 0.54 per 1000 resident days. While incidence studies have traditionally been more time-consuming and expensive to conduct than prevalence studies, our study demonstrates the potential of using routinely-collected electronic care management data to benchmark and meaningfully track this care indicator over time. This could provide a less burdensome and more methodologically-sound alternative for quality indicator programs where periodic surveys of PI prevalence are still often used [12, 13]. The majority of the research literature has focused on individual risk factors for PI development. Consistent with these studies [15], we found that older age, being male, having diabetes, previous PIs, incontinence, and poor mobility, were associated with higher PI rates. The reduction in the number of outlying facilities between the unadjusted and risk-adjusted funnel plots confirms the importance of accounting for commonly reported individual-level PI risk factors to ensure valid comparison of facility performance [24]. However, as in a recent longitudinal analysis of PI prevalence in nursing homes in the US [18], there remained considerable variation between facilities even after adjusting for resident risk factors. We have advanced the work of this previous study by using multilevel models to partition the total variability in PI incidence between levels, finding that 35% of the inter-facility variation remained attributable to the facility level after accounting for known resident factors (VPC = 0.35). The high MIRR (1.55) in the final model confirmed these findings, suggesting that the facility in which a person resides is as important as a risk factor for PI incidence as many individual-level risk factors. When an outcome is relatively rare and sample size small, high outcome rates are more likely to occur by chance. This study utilised funnel plots to identify outlying facilities while accounting for the increased variability in performance expected from small facilities [25]. We also measured persistence of high PI rates over time, as a way of distinguishing entrenched poor performance from random or intermittently high rates that later regress toward the mean [31]. Persistently high outlying facilities in our study had a PI incidence density almost twice that of other facilities. Targeting these facilities for intervention activities could therefore have the greatest impact in reducing the total number of PIs. Previous research has identified a number of common characteristics of facilities with higher rates of PIs, including having no nurses [20], no medical director or director of nursing [21], lower Medicaid reimbursement rates [19], and higher proportion of black residents [22]. In our study, persistently high outlying facilities were more likely to be located in areas with low socioeconomic status, and residents were more likely to have higher rates of PIs in facilities in regional areas. While our study did not find higher rates of PIs for those born in a non-English speaking country or in facilities with a higher proportion of PIs reported by a registered nurse, socioeconomic disparities appear to be a consistent underlying factor across both our study and previous research. This suggests more clinical support and greater resources may be needed to ensure all facilities are able to undertake best practice PI prevention and quality improvement activities. This study has a number of limitations. The functionality of the software platform, together with its implementation and usage across all facilities and workforce within the study period, presented some challenges. PI stage was not available, and so it is unknown at what severity PIs were detected by facility staff. However, our quarterly PI incidence of between 7% and 9% is consistent with another small study of four Australian facilities, where 8% of residents experienced a PI over a 3-month period [39]. While the ACFI was able to provide information about a broad range of resident care needs, other resident-level factors that have been associated with PI development, such as skin moisture or albumin levels [15], were not able to be accounted for. We were also not able to examine additional facility-level factors that may help to explain the remaining variation in PI rates, including the use of pressure-reducing support surfaces, direct nurse ratios and other quality improvement activities. The socioeconomic status variable was dichotomised to reduce the potential for classifying now gentrified areas as advantaged where the resident population may have lived in the area for a long period of time. However, this limits further exploration of the relationship between socioeconomic status and PI development. Although we excluded facilities with gaps in wound recording and were able to identify facilities with consistently high PI rates, some of the facility variation may also be explained by differences in PI recording, as will be the case wherever care providers are responsible for reporting. A strength of this study is the use of data from the largest provider of aged care in NSW/ACT, which contained a representative proportion of people born in non-English speaking countries and a sizable proportion of people living outside major city areas. However, given the generally low levels of digital maturity in Australian long-term aged care [40] and research suggesting facilities with higher levels of IT sophistication have better performance on quality measures [41], further work needs to be undertaken to determine whether PI incidence is poorer in less digitally-mature facilities, and whether the facility-level factors identified in this study are reflected across other Australian providers. Finally, an emphasis on the occurrence of PIs, in this study and much of the literature, applies a narrow focus to measuring the impact of PIs. Incorporating additional indicators such as the effect of PIs on quality of life [42], or the rate of PI healing [43], may be useful to evaluate the quality of care provided after a PI has occurred. Routinely-collected electronic care management data were able to be used in this study to track and compare PI incidence among facilities over time. Further exploration is needed to identify the elements that have made prevention efforts successful in facilities with low PI rates, and the reasons why facilities with certain characteristics are at risk of poor performance. The wide variation between facilities in PI rates certainly indicates that there is scope for targeted intervention to reduce the incidence of PIs in long-term care. Supplementary material Supplementary material is available at International Journal for Quality in Health Care online. Acknowledgements The authors thank members of the project Steering Committee, particularly Uniting staff who provided access to the data, assisted in the interpretation of the dataset, and provided their expertise to the project. Funding This work was supported by an Australian Research Council Linkage Grant undertaken in partnership with Uniting [grant number LP120200814]. Author contributions Study design—M.J., J.S., A.G., J.I.W.; Data acquisition—A.G., J.I.W.; Data analysis—M.J.; Data interpretation—all authors; Manuscript drafting—M.J.; Critical revision—J.S., A.G., J.I.W.; Final approval—all authors. Ethical approval The study was approved by the Macquarie University Human Research Ethics committee [reference number 5201401031]. References 1 National Pressure Ulcer Advisory Panel, European Pressure Ulcer Advisory Panel and Pan Pacific Pressure Injury Alliance. Haesler E (ed). Prevention and treatment of pressure ulcers: quick reference guide . Osborne Park, Australia: Cambridge Media, 2014. https://www.npuap.org/wp-content/uploads/2014/08/Updated-10-16-14-Quick-Reference-Guide-DIGITAL-NPUAP-EPUAP-PPPIA-16Oct2014.pdf 2 Sen CK, Gordillo GM, Roy S et al.  . Human skin wounds: a major and snowballing threat to public health and the economy. Wound Repair Regen  2009; 17: 763– 71. Google Scholar CrossRef Search ADS PubMed  3 Berlowitz DR, Brandeis GH, Anderson J et al.  . Effect of pressure ulcers on the survival of long-term care residents. J Gerontol Series A  1997; 52: M106– M10. Google Scholar CrossRef Search ADS   4 Gorecki C, Brown JM, Nelson EA et al.  . Impact of pressure ulcers on quality of life in older patients: a systematic review. J Am Geriatr Soc  2009; 57: 1175– 83. Google Scholar CrossRef Search ADS PubMed  5 Pieper B, Langemo D, Cuddigan J. Pressure ulcer pain: a systematic literature review and National Pressure Ulcer Advisory Panel white paper. Ostomy Wound Manage  2009; 55: 16– 31. Google Scholar PubMed  6 Hahnel E, Lichterfeld A, Blume-Peytavi U et al.  . The epidemiology of skin conditions in the aged: a systematic review. J Tissue Viability  2017; 26: 20– 8. Google Scholar CrossRef Search ADS PubMed  7 Santamaria N, Carville K, Prentice J et al.  . Reducing pressure ulcer prevalence in residential aged care: results from phase II of the PRIME trial. Wound Pract Res  2009; 17: 12– 22. 8 Banks M, Bauer J, Graves N et al.  . Malnutrition and pressure ulcer risk in adults in Australian health care facilities. Nutrition  2010; 26: 896– 901. Google Scholar CrossRef Search ADS PubMed  9 Courtney M, O’Reilly MT, Edwards H et al.  . Benchmarking clinical indicators of quality for Australian residential aged care facilities. Aust Health Rev  2010; 34: 93– 100. Google Scholar CrossRef Search ADS PubMed  10 Black JM, Edsberg LE, Baharestani MM et al.  . Pressure ulcers: avoidable or unavoidable? Results of the National Pressure Ulcer Advisory Panel consensus conference. Ostomy Wound Manage  2011; 57: 24– 37. Google Scholar PubMed  11 Chou R, Dana T, Bougatsos C et al.  . Pressure ulcer risk assessment and prevention: a systematic comparative effectiveness review. Ann Intern Med  2013; 159: 28– 38. Google Scholar CrossRef Search ADS PubMed  12 Castle NG, Ferguson JC. What is nursing home quality and how is it measured? Gerontologist  2010; 50: 426– 42. Google Scholar CrossRef Search ADS PubMed  13 Australian Government Department of Health. National Aged Care Quality Indicator Programme. Available at: https://agedcare.health.gov.au/ensuring-quality/quality-indicators/about-the-national-aged-care-quality-indicator-programme. Accessed 26 September, 2017. 14 National Pressure Ulcer Advisory Panel. Pressure ulcer incidence density as a quality measure. Press Release: 11 March 2014. Available at: http://www.npuap.org/pressure-ulcer-incidence-density-as-a-quality-measure/. Accessed 26 September, 2017. 15 Coleman S, Gorecki C, Nelson EA et al.  . Patient risk factors for pressure ulcer development: systematic review. Int J Nurs Stud  2013; 50: 974– 1003. Google Scholar CrossRef Search ADS PubMed  16 Braden B, Bergstrom N. A conceptual schema for the study of the etiology of pressure sores. Rehabil Nurs  2000; 25: 105– 10. Google Scholar CrossRef Search ADS   17 Moore ZE, Cowman S. Risk assessment tools for the prevention of pressure ulcers. The Cochrane Libr  2014; 5: CD006471. 18 Hartmann CW, Shwartz M, Zhao S et al.  . Longitudinal pressure ulcer rates after adoption of culture change in veterans health administration nursing homes. J Am Geriatr Soc  2016; 64: 151– 5. Google Scholar CrossRef Search ADS PubMed  19 Grabowski DC, Angelelli JJ. The relationship of medicaid payment rates, bed constraint policies, and risk‐adjusted pressure ulcers. Health Serv Res  2004; 39: 793– 812. Google Scholar CrossRef Search ADS PubMed  20 Kwong EWy, Pang SMc, Aboo GH et al.  . Pressure ulcer development in older residents in nursing homes: influencing factors. J Adv Nurs  2009; 65: 2608– 20. Google Scholar CrossRef Search ADS PubMed  21 Kang Y, Tzeng H-M, Miller NA. Facility characteristics and risk of developing pressure ulcers in US nursing homes. J Nurs Care Qual  2016; 31: E9– E16. Google Scholar CrossRef Search ADS PubMed  22 Li Y, Yin J, Cai X et al.  . Association of race and sites of care with pressure ulcers in high-risk nursing home residents. JAMA  2011; 306: 179– 86. Google Scholar CrossRef Search ADS PubMed  23 Berlowitz D. Incidence and prevalence of pressure ulcers. In: Thomas DR, Compton GA (eds). Pressure ulcers in the aging population: a guide for clinicians . Springer: Humana Press, 2014. 24 Berlowitz DR, Anderson JJ, Ash AS et al.  . Reducing random variation in reported rates of pressure ulcer development. Medical Care  1998; 36: 818– 25. Google Scholar CrossRef Search ADS PubMed  25 Spiegelhalter DJ. Funnel plots for comparing institutional performance. Stat Med  2005; 24: 1185– 202. Google Scholar CrossRef Search ADS PubMed  26 Towne SD Jr, Lee S, Li Y et al.  . Assessment of eHealth capabilities and utilization in residential care settings. Health Informatics J  2016; 22: 1063– 75. Google Scholar CrossRef Search ADS PubMed  27 Australian Government, Aged Care Financing Authority. Fourth Report on the Funding and Financing of the Aged Care Sector. July 2016. Available at: https://agedcare.health.gov.au/sites/g/files/net1426/f/documents/10_2016/acfa_annual_report_on_funding_and_financing_of_the_aged_care_industry_2016.pdf. Accessed 26 September, 2017. 28 Telstra Health. iCareHealth – Clinical and Care Management Software. Available at: http://www.icarehealth.com.au/. Accessed 26 September, 2017. 29 Australian Government, Department of Health. Aged Care Funding Instrument. User Guide 2017. Available at: https://agedcare.health.gov.au/acfi-user-guide. Accessed 26 September, 2017. 30 Iezzoni LI. Risk adjustment for medical effectiveness research: an overview of conceptual and methodological considerations. J Investigat Med  1995; 43: 136. 31 Duckett S, Griffiths K Perils of place: identifying hotspots of health inequalities. Grattan Institute; 2016. Available at: https://grattan.edu.au/wp-content/uploads/2016/07/874-Perils-of-Place.pdf. Accessed 26 September, 2017. 32 Gelman A, Hill J. Data analysis using regression and multilevel/hierarchical models . New York, NY, USA: Cambridge University Press, 2007. 33 Stryhn H, Sanchez J, Morley P et al.  , editors. Interpretation of variance parameters in multilevel Poisson regression models. Proceedings of the 11th International Symposium on Veterinary Epidemiology and Economics; 2006. 34 Australian Government, Department of Health. Basic Subsidy Amount (Aged Care Funding Instrument). Available at: https://agedcare.health.gov.au/aged-care-funding/residential-care-subsidy/basic-subsidy-amount-aged-care-funding-instrument. Accessed 26 September, 2017. 35 Australian Bureau of Statistics. Australian Statistical Geography Standard (ASGS): Correspondences, July 2011. Cat. no. 1270.0.55.006. Canberra: ABS. 36 Australian Institute of Health and Welfare. Residential aged care and home care 2014–15. Cat. no. WEB ; 133. Canberra: AIHW, 2016. 37 Baharestani MM, Black JM, Carville K et al.  . Dilemmas in measuring and using pressure ulcer prevalence and incidence: an international consensus. International Wound. Journal  2009; 6: 97– 104. 38 Okuwa M, Sanada H, Sugama J et al.  . A prospective cohort study of lower-extremity pressure ulcer risk among bedfast older adults. Adv Skin Wound Care  2006; 19: 391– 7. Google Scholar CrossRef Search ADS PubMed  39 Moore KJ, Doyle CJ, Dunning TL et al.  . Public sector residential aged care: identifying novel associations between quality indicators and other demographic and health-related factors. Aust Health Rev  2014; 38: 325– 31. Google Scholar CrossRef Search ADS PubMed  40 Flinders University & the Aged Care Industry IT Council. A technology roadmap for the Australian aged care sector. June 2017. Available at: http://aciitc.com.au/wp-content/uploads/2017/06/ACIITC_TechnologyRoadmap_2017.pdf 41 Alexander G, Madsen R. IT sophistication and quality measures in nursing homes. J Gerontol Nurs  2009; 35: 22– 7. Google Scholar CrossRef Search ADS PubMed  42 Gorecki C, Nixon J, Lamping DL et al.  . Patient-reported outcome measures for chronic wounds with particular reference to pressure ulcer research: a systematic review. Int J Nurs Stud  2014; 51: 157– 65. Google Scholar CrossRef Search ADS PubMed  43 Bates‐Jensen BM, MacLean CH. Quality indicators for the care of pressure ulcers in vulnerable elders. J Am Geriatr Soc  2007; 55: S409– 16. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press in association with the International Society for Quality in Health Care. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

International Journal for Quality in Health CareOxford University Press

Published: May 4, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off