Counts injuries, not persons, i.e., some patients have more than one injury. NA: Not applicable

Table 2 displays the principal diagnosis (i.e., primary reason) for the readmission after the index TSI hospitalisation within 28 and 90 days; hence the denominator is the number of admission episodes, not the number of persons. Allied Health Services collectively apportion the majority of primary reasons for admission (56.3% and 59.6%). While this is a broad-reaching category, most individuals requiring readmission in this category were coded as "Care involving use of rehabilitation procedure, unspecified" (88%). All of these individuals had previously been discharged from an acute-care admission. Greater than 10% of the cohort of persons with TSI required readmission for complications of internal fixation devices (including infection), open wound infections and other complaints; 16 persons (13.4%) were readmitted within 28 days and 35 (14%) within 90 days.

The measure of resources required to accommodate these readmissions was counted as bed days and costs. The mean ± SD length of stay for readmissions within 28 days was 3.1 ± 7.0 days, compared with the length of stay for readmissions within 90 days at 4.8 ± 20.3 days. The total number of acute care bed days used by the 102 persons readmitted in 250 episodes was 1232 days at a total cost of \$708,464 (95%CI: \$417,325–\$999,603). The mean ± SD per patient readmission cost was \$6946 ± \$14,532. The mean costs for 90-day readmissions were highest for patients with spinal cord injuries (\$24,558), persons aged 16–29 years (\$21,947) and transport incidents occurring on a street/highway (\$15,492).


**Table 2.** Readmissions within 28 and 90 days.

\* Principal ICD10-AM diagnosis group in brackets (15). **#** Of total readmission numbers.

#### **4. Discussion**

In a cohort of 740 patients who had sustained work-related incident related traumatic spinal injuries during a three-year period, we found that around 8% (*n* = 61) of these patients experienced 119 inpatient readmissions within 28 days post-acute care discharge. Reasons for unplanned readmissions included requiring allied health interventions (56.3%), injury repair and operative complications (13.4%), circulatory problems including embolus (5%) and mental health problems (4.2%). The mean (SD) per patient cost for acute readmission was \$6,946 (\$14,532). Total costs of unplanned readmissions in the 90 days post-acute discharge were \$708,464 (95%CI: \$417,325–\$999,602).

Although the recently reported NSW average of unplanned readmissions was 6.8% [13], we found a (disease-specific) readmission rate of 16.1% by 28 days post the index discharge among people who sustained work-related TSI, resulting predominantly from spinal fractures. Hospital performance monitoring in Australia measures a range of indicators including the incidence of hospital-acquired complications and rates of unplanned readmissions. These indices are now linked to government funding; for example, admissions where a hospital acquired complication incurs a payment reduction to hospital reimbursement [18]. Unplanned hospital readmissions can be a signal of issues with the effectiveness, continuity and integration of care provided to patients. As such, the use of patient clinical data in this way can help to drive safety and quality improvement. In this study, it may be that patients discharged to home may have rather benefited from a subsequent rehabilitation admission, avoiding the need to return to an acute care hospital. While rehabilitation admissions still impose a system cost, it is unlikely to be to the same extent. This information is therefore helpful in the discharge planning stage for patients with a traumatic spinal injury.

Our study had some limitations. We did not have access to various important variables about the injured worker, such as ethnicity, level of education and experience, the employment situation (e.g., whether permanent/part-time/casual) and specific occupation. Indigenous status was available but not identified from the APDC collection for this study. We have assumed all the patients to be of non-indigenous status for the estimation of costs using the NWAU-based approach; while we have not provided estimate of the cost impact this may have had, it is not anticipated to vary the current estimates significantly. Furthermore, the APDC District Network Return (DNR) does not include hospitalisation cost data for patients who were admitted to private hospitals across NSW. The degree of under-representation that this presents is uncertain, however such severe injuries are much more likely to be treated within the public hospital system [21]. We used NWAU-based costing approach to estimate the costs over the DNR data-based estimation to include all the public hospital separations in the costing analysis. The cost estimates presented are an under-estimation of the true costs as the separations at private hospitals were excluded from the analysis.

Braaf et al. [22] described the experiences of seriously injured patients within the Victorian state trauma service regarding communication from the health professionals caring for them after their injury. Many patients reported insufficient information, confusion and lack of clear communication to have hampered their discharge from acute care. While we did not interview patients involved in the current study, the high rate of readmissions for care involving rehabilitation suggests an area for improvement in the discharge planning process. Readmission can reflect the underuse of recommended care, adverse events and complications of hospital care, inadequate discharge planning or problems with coordination and integration of care across hospital, primary care and community settings. Patients discharged without sufficient information for continuity of care [23] can deteriorate once out of the acute care setting. Readmissions with infections, open wounds and device complications could likely be avoided with education, support and a general practitioner liaison.

#### **5. Conclusions**

Work-related traumatic spinal injuries create a significant burden of cost and disability for the Australian workforce but are preventable and also fall under a current focus of the Safe Work Australia policy to reduce serious injury compensation claims by 30% by 2022. This study demonstrates that the ongoing burden of work-related spinal trauma is not insignificant in the acute period post primary discharge. This study offers evidence of unmet needs after acute TSI and can assist trauma care coordinators' comprehensive assessments of these patients prior to discharge. Improved quantification of the ongoing personal and health service after work-related injury is a vital part of the information needed to improve recovery after major work-related trauma.

**Author Contributions:** Conceptualisation, L.N.S. and T.D.; Methodology, L.N.S., H.M. and B.V.; Software, H.M. and L.N.S.; Formal Analysis, L.N.S., H.M. and B.V.; Data Curation, L.N.S., H.M. and B.V., Writing—Original Draft Preparation, L.N.S.; Writing—Review and Editing, J.T.Y., J.W.M., R.Q.I., H.M. and T.D.; Supervision, J.W.M.; Project Administration, L.N.S.; Funding Acquisition, L.N.S.

**Funding:** L.N.S. acknowledges and is supported by funding from icare™.

**Acknowledgments:** The authors would like to thank the Meteor coding assistant team at the Australian Institute of Health and Welfare for their strategic advice in the clarification of search and usage of activity codes.

**Conflicts of Interest:** The authors declare no conflict of interest. The founding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **Appendix A**

#### A. TSI Identification

The dataset combined all patients for whom one of the following ICD-10-AM (International Classification of Diseases—Tenth Revision, Australian Modification) codes have been identified in any separation, and in any position of the diagnostic code list within the NSW Admitted Patient Data Collection (APDC):

S12, S12.0, S12.1, S12.2, S12.21, S12.22, S12.23, S12.24, S12.25, S12.7, S12.8, S12.9, S13.1, S13.10, S13.11, S13.12, S13.13, S13.14, S13.15, S13.16, S13.17, S13.18, S13.2, S13.3, S14.0, S14.10, S14.11, S14.12, S14.13, S14.70, S14.71, S14.72, S14.73, S14.74, S14.75, S14.76, S14.77, S14.78, S22.0, S22.00, S22.01, S22.02, S22.03, S22.04, S22.05, S22.06, S22.1, S24.0, S24.1, S24.10, S24.11, S24.12, S24.7, S24.70, S24.71, S24.72, S24.73, S24.74, S24.75, S24.76, S24.77, S32, S32.0, S32.00, S32.01, S32.02, S32.03, S32.04, S32.05, S34.0, S34.1, S34.3, S34.70, S34.71, S34.72, S34.73, S34.74, S34.75, S34.76, T06.0, T06.1, T09.3.
