You are here

  • Home
  • Archive
  • Volume 24, Issue 6
  • Injury upon injury: a prospective cohort study examining subsequent injury claims in the 24 months following a substantial injury

Article Text

Article menu
Download PDFPDF
Original article
Injury upon injury: a prospective cohort study examining subsequent injury claims in the 24 months following a substantial injury
  1. Helen Harcombe1,
  2. Gabrielle Davie1,
  3. Emma Wyeth2,
  4. Ari Samaranayaka3,
  5. Sarah Derrett1
  1. 1Injury Prevention Research Unit, Department of Preventive and Social Medicine, University of Otago, Dunedin, New Zealand
  2. 2Ngāi Tahu Māori Health Research Unit, Department of Preventive and Social Medicine, University of Otago, Dunedin, New Zealand
  3. 3Department of Preventive and Social Medicine, University of Otago, Dunedin, New Zealand
  1. Correspondence to Dr Helen Harcombe, Injury Prevention Research Unit, Department of Preventive and Social Medicine, University of Otago, PO Box 56, Dunedin 9054, New Zealand; helen.harcombe{at}otago.ac.nz

Abstract

Objectives This study examines subsequent injuries reported to the Accident Compensation Corporation (ACC), New Zealand’s universal no-fault injury insurer, in the 24 months following an ACC entitlement claim injury event. Specific aims were to determine the: (1) 12 and 24 month cumulative incidence of at least one ACC-reported subsequent injury (ACC-SUBS-Inj), (2) characteristics of participants with and without ACC-SUBS-Inj, (3) frequency of ACC-SUBS-Inj, (4) time periods in which people are at higher risk of ACC-SUBS-Inj and (5) types of ACC-SUBS-Inj.

Methods Interview data collected directly from participants in the Prospective Outcomes of Injury Study (POIS) were combined with ACC-SUBS-Inj data from ACC and hospital discharge datasets. A subsequent injury was defined as any injury event resulting in an ACC claim within 24 months following the injury event for which participants were recruited to POIS (the sentinel injury). All ACC-SUBS-Inj were included irrespective of whether they were the same as the sentinel injury or not.

Results Of 2856 participants, 58% (n=1653) experienced at least one ACC-SUBS-Inj in 24 months; 31% (n=888) had more than one ACC-SUBS-Inj. The time period of lowest risk of ACC-SUBS-Inj was the first 3 months following the sentinel injury event. Spine sprain/strain was the type of injury with the greatest number of ACC-SUBS-Inj claims per person.

Conclusions More than half of those with an ACC entitlement claim injury incurred further injury events that resulted in a claim in the following 24 months. Greater understanding of these subsequent injury events provides an avenue for injury prevention.

  • cohort study
  • recidivism
  • descriptive epidemiology

https://doi.org/10.1136/injuryprev-2017-042467

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    The burden of injury is considerable worldwide.1 Contributing to this burden are multiple injury events over time, incurred by the same individuals. In New Zealand, the Accident Compensation Corporation (ACC), New Zealand’s universal no-fault injury insurer, accepted 1.9 million injury claims in 2016,2 yet the population of New Zealand was 4.7 million.3 As well as contributing to the overall injury burden, it has been suggested that subsequent injuries may be more costly, both financially and in terms of work disability, compared with ‘initial’ injuries.4 5 Understanding the epidemiology of subsequent injuries may provide an avenue for injury prevention strategies and rehabilitation support.

    Internationally, there is a lack of detailed knowledge about subsequent injuries among general injury populations, that is, injury populations that involve both work-related and non-work-related injury, a broad range of injury types and injuries that do and do not require hospitalisation. A number of studies have examined subsequent injuries among particular injury populations such as workers’ compensation claimants,6–8 people attending trauma centres or emergency departments9–13 or among children.14–16 Others have focused on the recurrence of specific injuries such as low back pain,17 ankle sprains18 or hamstring strains.19 20 Previously, we examined self-reported subsequent injuries among a general injury population21 who had had an injury involving an ACC entitlement claim (a type of claim for injuries likely to warrant compensation for requiring more than a week off work or other rehabilitative assistance such as home-help).22 At least one self-reported subsequent injury event was reported by 28% of participants over a 9-month period (between 3 and 12 months following their entitlement claim injury).21 However, information about the subsequent injuries reported in our previous paper was limited to self-report via interviews.23

    This paper examines data from the same cohort of participants,21 but extends this by examining subsequent injuries that involved an ACC claim in the 24 months following the sentinel injury event. The aims are to determine the: (1) 12 and 24 month cumulative incidence of having at least one ACC-reported subsequent injury (ACC-SUBS-Inj), (2) characteristics of participants with and without ACC-SUBS-Inj, (3) frequency of ACC-SUBS-Inj, (4) time periods in which people are at higher risk of ACC-SUBS-Inj and (5) types of ACC-SUBS-Inj.

    Methods

    This Subsequent Injury Study (SInS), combined data from three sources. First, interview data from the Prospective Outcomes of Injury Study (POIS),23 a longitudinal cohort study (n=2856) previously undertaken in New Zealand. Participants aged between 18 and 64 years were recruited between 2007 and 2009 and interviewed (on average) 3, 12 and 24 months following an ACC entitlement claim injury event (the ‘sentinel’ injury).23 24 Second, administrative data from the ACC about injury claims lodged by these participants in the 24 months following their sentinel injury. Third, hospital discharge data from the National Minimum Dataset (NMDS) was obtained for participants who were hospitalised within 7 days of their sentinel or subsequent injury event (figure 1). The SInS protocol has been described previously.25 Ethical approval is from the New Zealand Health and Disability Multi-Region Ethics Committee (MEC/07/07/093).

    Figure 1
    Figure 1

    Data linkage in the Subsequent Injury Study (SInS): Accident Compensation Corporation claims (ACC), National Minimum Dataset of hospital discharges (NMDS) and interviews from the Prospective Outcomes of Injury Study (POIS).

    Demographic information including age, sex, ethnicity and occupation were collected during POIS interviews.24 Age, sex and ethnicity were obtained using questions from the New Zealand Census.26 Participants could identify as having more than one ethnicity. In our analyses, ethnicity was prioritised27 into four groups (Sole New Zealand European, Māori, Pacific or ‘none of the above’) with Māori, the indigenous population of New Zealand, given the highest priority. This meant that if someone was Māori and any other ethnicity they were classified as Māori. If participants identified as Pacific and any other ethnicity other than Māori they were classified as Pacific. Participants in paid employment at the time of their sentinel injury were asked about their main job, with occupations classified as ‘professional’, ‘technical’, ‘trade or manual’ or ‘unclassifiable’.28 Data about pre-sentinel injury health characteristics were also collected during POIS and have been described previously.29 30 Participants were asked a range of questions in their POIS interviews about the sentinel injury event including: whether the injury was due to an accident or assault; if they perceived the sentinel injury event to be a threat to their life or to long-term disability at the time of injury, and whether or not the injury was work-related.

    The ACC accepts claims regardless of the injury setting (eg, workplaces, sport and recreational settings, on the road and in the home), intentionality and whether people are workers or non-workers.31 Subsequent injuries from any mechanism, any body region and any nature of injury were included in the study, regardless of the type of sentinel injury. To align with ACC’s classification of claims known as ‘late lodgements’, ACC-SUBS-Inj claims were excluded if they were lodged >12 months after the subsequent injury event.32 Unlike the sentinel injury, which was an entitlement claim, all ACC claim types were included in our analyses of subsequent injuries. These were classified as ‘entitlement claims’, ‘medical fees only claims’ (for injuries that required treatment by a health professional but no additional rehabilitation support),22 ‘other claims’ (eg, those involving assessments) and ‘unclassified’ (those that did not have a claim type specified, eg, where funding was associated with a District Health Board’s bulk funding).33

    The nature, body region and severity of sentinel and subsequent injuries were derived from ACC’s injury diagnosis codes. Diagnoses in ACC data had been  coded to READ codes,34 International Classification of Diseases (ICD)-9-Clinical Modification35 or ICD-1036 codes. Where necessary, READ and ICD-9 codes were mapped to ICD-10 codes.29 Participants could have incurred more than one injury type in the sentinel injury event and in each subsequent injury event. For example, an MVC (one injury event) may result in a lower limb fracture, an upper limb fracture and an intracranial injury. A subsequent injury event could have a similar scenario and participants could have more than one subsequent injury event. Twelve variables were used to describe the types of injury involved. These variables were derived from the most common nature of injury and body region groupings of sentinel injuries (eg, intracranial injury; lower extremity fracture).30 The same variables were used to classify subsequent injuries. The severity of each injury event was estimated by mapping ICD-10 diagnoses to the AIS and then calculating the New ISS (NISS).37 38 Hospitalisation was defined as being admitted to hospital, or treated at an emergency department for >3 hours, within 7 days of the injury event.29 To determine which ACC-SUBS-Inj resulted in hospitalisation ACC claims resulting from subsequent injury events for our cohort were probabilistically linked (ie, matched using software-computed weights that calculate the probability that records from each data source relate to the same injury event) to hospital discharge data in the NMDS.

    Statistical analyses used in this paper were descriptive statistics, χ2 tests and calculating 95% CIs for proportions. Participants’ pre-sentinel injury sociodemographic and health characteristics, and sentinel-injury characteristics, were examined according to whether or not participants had an ACC-SUBS-Inj and p values were obtained from χ2 tests. The percentage (and 95% CIs) of ACC-SUBS-Inj within each 3-month interval within the 2 years of follow-up was calculated to identify time periods of higher ACC-SUBS-Inj incidence. The types of ACC-SUBS-Inj were described and the number of claims per person determined for each of the 12 body region/nature of injury combinations. Analyses were carried out using Stata v.14.39

    Results

    Of the 2856 participants, 38% (n=1087) had at least one ACC-SUBS-Inj in the 12 months following their ACC entitlement claim injury event (the sentinel injury); by 24 months, 1653 participants (58%) had incurred at least one ACC-SUBS-Inj.

    Younger participants were more likely to have ACC-SUBS-Inj with 62% of participants aged 18–34 years having at least one ACC-SUBS-Inj compared with 55% and 56% of those aged 35–49 and 50–64 years, respectively (table 1). There were differences by ethnicity, with 62% of both Māori and Pacific participants having at least one ACC-SUBS-Inj compared with 58% of New Zealand European participants and 51% of participants in the ‘none of the above’ ethnicity category. Sixty per cent of males had at least one ACC-SUBS-Inj compared with 55% of females. Sixty-four per cent of trade or manual workers had at least one ACC-SUBS-Inj compared with 54% of both technical workers and professionals (table 1).

    View this table:
    Table 1

    Presentinel injury sociodemographic characteristics of participants according to ACC-reported subsequent injury (ACC-SUBS-Inj) status

    Participants who had problems with their usual activities, or with pain or discomfort prior to their sentinel injury, were more likely to have ACC-SUBS-Inj compared with those who had no problems with these aspects. (table 2). Those that had depressive-type episodes or two or more chronic conditions prior to their sentinel injury were also more likely to have at least one ACC-SUBS-Inj compared with those without depressive episodes or with fewer chronic conditions, respectively. Of those that had greater pre-sentinel injury disability, 66% had at least one ACC-SUBS-Inj compared with 57% of those that did not have pre-sentinel injury disability. Of participants who were physically active on five or more days of the week, 60% had at least one ACC-SUBS-Inj compared with 56% of those that were physically active on <5 days. Of those that had a high alcohol intake, 63% had at least one ACC-SUBS-Inj compared with 55% of those with no or low alcohol intake; 63% of those that used recreational drugs had at least one ACC-SUBS-Inj compared with 57% of participants that did not use recreational drugs.

    View this table:
    Table 2

    Pre-sentinel injury health characteristics of participants according to ACC-reported subsequent injury (ACC-SUBS-Inj) status

    Participants who had a sentinel injury of lower severity (NISS 1–3), were not hospitalised for their sentinel injury or had no trouble accessing healthcare services for their sentinel injury were more likely to have ACC-SUBS-Inj (table 3). Participants were less likely to have an ACC-SUBS-Inj if their sentinel injury was a lower extremity fracture.

    View this table:
    Table 3

    Sentinel injury-related characteristics according to ACC-reported subsequent injury (ACC-SUBS-Inj) status

    While 27% of participants (n=765) experienced only one ACC-SUBS-Inj event, 31% of participants (n=888) experienced more than one ACC-SUBS-Inj event in the 24 months; 16% (n=468) had two, 7% (n=197) had three, 4% (n=110) had four and 4% (n=113) had five or more ACC-SUBS-Inj events. During the first 12 months following the sentinel injury event, 1087 of the 2856 participants had a total of 1617 ACC-SUBS-Inj events. Over 24 months, 1653 participants had a total of 3444 ACC-SUBS-Inj events (table 4). Of the 3444 ACC-SUBS-Inj events, 4% (n=149) resulted in hospitalisation; 12% (n=418) an ACC entitlement claim (table 4).

    View this table:
    Table 4

    Claim type and hospitalisation status of ACC-reported subsequent injury (ACC-SUBS-Inj) events over 12 and 24 months following the sentinel injury event

    There was a stepped increase in the frequency of ACC-SUBS-Inj from the first 3-month period following the sentinel injury event to the second and third 3-month periods. Thereafter, the frequency of ACC-SUBS-Inj remained similar to 24 months following the sentinel injury event (figure 2).

    Figure 2
    Figure 2

    Distribution of ACC-reported subsequent injury (ACC-SUBS-Inj) events over the 24 months following the sentinel injury event. ACC, Accident Compensation Corporation.

    One injury event could result in more than one injury diagnosis and the 3444 ACC-SUBS-Inj events involved 4470 injury diagnoses. The most common types of ACC-SUBS-Inj were spine sprain/strains (25%, n=1136), followed by lower extremity dislocation/sprain/strains comprising 17% (n=775) of the 4470 injuries (table 5, column 1).

    View this table:
    Table 5

    Frequency of ACC-reported subsequent injury (ACC-SUBS-Inj) in the 24 months following the sentinel injury event by the type of injury

    Table 5 (column 2) presents the number of ACC-SUBS-Inj events that involved at least one of each type of injury. Multiple injury diagnoses from the same injury event could potentially be in the same injury type category, for example, someone could incur a cervical spine strain and a thoracic spine strain in the same injury event, both of which would be classified as spine sprain/strains explaining why the 1136 spine sprain/strains occurred in 905 ACC-SUBS-Inj events.

    Furthermore, table 5 (column 3) presents the number of participants with each type of ACC-SUBS-Inj. Participants could have multiple ACC-SUBS-Inj events involving the same type of injury over the 24-month period, for example, the 905 ACC-SUBS-Inj events involving spine sprain/strains occurred among 638 participants (ie, some of the same participants had multiple spine sprain/strains over time). The type of ACC-SUBS-Inj with the highest number of claims per person were spinal sprain/strains with 1.42 claims per person over the 24 months followed by lower extremity dislocation/sprain/strains with 1.34 claims per person.

    Discussion

    A substantial proportion of participants (38% and 58%, respectively) had at least one ACC-SUBS-Inj event in the 12 and 24 months following their sentinel injury event. There were differences in pre-sentinel sociodemographic and health characteristics for participants that had at least one ACC-SUBS-Inj compared with those that had no ACC-SUBS-Inj. For example, those that had at least one ACC-SUBS-Inj were more likely to be trade or manual workers compared with technical or professional workers. Those participants classified as ‘none of the above’ ethnicity (ie, not Māori, Pacific or sole NZ European) were least likely to have ACC-SUBS-Inj. Potentially, a greater proportion of that group may not have been born in New Zealand and it is possible more recent migrants (despite having already an ACC entitlement claim injury) may be less familiar with the ability to register ACC claims for subsequent injuries. Sentinel injury characteristics also differed, for example, participants who were hospitalised because of their sentinel injury were less likely to have ACC-SUBS-Inj. Some of these findings align with previous research, for example, our previous research21 and others4 have also found subsequent injury to be more common among trade or manual workers, and our finding that those whose sentinel injury was a lower extremity fracture were less likely to have ACC-SUBS-Inj aligns with a previous study among children and adolescents.14 However, the analyses in the current study were not adjusted for potential confounders and other factors may explain the findings. Spine sprains/strains were the most frequently occurring ACC-SUBS-Inj. Previous literature has also found high rates of recurrence of these types of injury, for example, a systematic review reporting a pooled estimate (although from two studies) of 73% of people experiencing a recurrence of low back pain within 12 months.17

    Strengths of this study were the inclusion of a range of injury types for both sentinel and subsequent injury events, the inclusion of sentinel and subsequent injuries that involved hospitalisation and non-hospitalisation and both work-related and non-work-related injuries. The study combined data derived directly from participants, as well as from administrative routinely collected datasets. A limitation was that the ‘sentinel’ injury defined at the participants’ recruitment was unlikely to have been the first-ever injury reported to ACC for these individuals. However, the intent of the study was not to capture individuals’ first-ever injury event, rather it was to examine injuries following a sentinel injury event that was serious enough to warrant an ACC entitlement claim. While the study is focused on a general injury cohort, it is limited to people who were aged between 18 and 64 years (inclusive) at the time of their sentinel injury. A further limitation is that analyses have not accounted for participants who died during the period of follow-up. Five participants died within the first 12 months following their sentinel injury and a further six died in the 12–24 months following their sentinel injury event. This paper aims to provide an overview of a number of aspects that illustrate the complexity of research into subsequent injury over 24 months among a general injury cohort. As such, the statistical analyses in this paper are limited to descriptive statistics, χ2 tests and calculating 95% CIs for proportions. Accounting for potential confounding factors in the analyses of sociodemographic, health and sentinel injury characteristics was beyond the scope of this paper and further work is currently underway to understand this aspect.

    Our previous research relating to subsequent injury was limited to self-reported subsequent injuries based on asking participants to recall whether they had experienced at least one subsequent injury since their last interview. The current study linking ACC and NMDS data about subsequent injury to the interview data has now allowed us to determine aspects such as the timing of the subsequent injury, the number of subsequent injuries per person and the nature of the subsequent injuries. The current study’s focus on subsequent injuries reported to ACC is also identifying those injuries that came to the attention of healthcare providers and enables us to distinguish between straightforward treatment-only claims and entitlement claims where it is expected more rehabilitation and health services will be required, whereas self-reported injuries did not allow us to consider that aspect. In our previous research, we found a cumulative incidence of 28% for at least one self-reported subsequent injury between 3 and 12 months following the sentinel injury event.21 Although examining the same cohort of participants, it is problematic to make direct comparisons between that research and the current study as the time periods examined differ (9 months in the previous research and 12 and 24 months in the current study). Nevertheless, the 28% cumulative incidence of at least one self-reported subsequent injury in 9 months is comparable with 38% having at least one ACC-reported subsequent injury in 12 months. Unlike in our previous research, the current study enabled the number of subsequent injury claims per person to be determined. The finding that 31% of participants experienced more than one ACC-SUBS-Inj in the 24 months following their sentinel injury is concerning.

    The lower frequency of ACC-SUBS-Inj in the first 3 months (and to a lesser extent 3–6 months) following participants’ sentinel injury event may relate to a number of factors. We can only speculate, but people still recovering from their sentinel injury in these phases may not have resumed all their usual activities so may be less at risk of subsequent injury. However, it may be that participants have recently received injury prevention advice (at the time of their sentinel injury) or people may be more cautious in this initial period following injury. The timing of subsequent injury has been investigated among people hospitalised for an injury event in the USA.12 However, in that study they examined participants’ first subsequent injury and only those subsequent injuries involving hospitalisation or death. They examined the timing of subsequent injuries in 6-month time periods from 6 months to 5 years following a sentinel injury and found the highest rate ratio of subsequent injury was 6–12 months following the sentinel injury.

    People who have already experienced one injury that has come to the attention of the ACC (which occurs via a health professional) are an identifiable group for which targeted interventions to address injury prevention may be possible. This paper provides an overview of ACC-SUBS-Inj among such a cohort and highlights areas that, in order to address the issue of subsequent injury among general injury populations, need to be understood in greater depth. Future work will focus on understanding to what extent characteristics of sentinel and subsequent injuries are consistent (eg, whether sentinel and subsequent injuries occur in the same or different body regions, are the same or different types of injury and whether the mechanism of injury and place of occurrence are the same or different). Future work will also examine predictors of ACC-SUBS-Inj with the aim of contributing to the development of secondary injury prevention initiatives to reduce the overall burden of injury.

    What is already known on this subject?

    • The burden of injury is substantial and includes injury events occurring to the same individuals over time (subsequent injuries).

    • The epidemiology of subsequent injury is little understood in general injury populations.

    What this study adds?

    • Among a general injury cohort, the cumulative incidence of at least one subsequent injury claim submitted to Accident Compensation Corporation was 38% in 12 months and 58% in 24 months.

    • A substantial proportion of participants (31%) had more than one subsequent injury claim in 24 months.

    Acknowledgments

    The authors are grateful to the study participants for sharing their information. The authors thank Sue Wilson, Dave Barson and Brandon de Graaf for their work in preparing the data for analysis, and Daniela Aldabe and Amy Richardson for providing helpful comments on an earlier draft of this paper.

    References

    1. ↵
      1. Vos T,
      2. Flaxman AD,
      3. Naghavi M, et al
      . Years lived with disability (YLDs) for 1160 sequelae of 289 diseases and injuries 1990-2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 2012;380:2163–96.doi:10.1016/S0140-6736(12)61729-2
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      The Accident Compensation Corporation. Annual report 2014. Wellington: The Accident Compensation Corporation, 2014.
    3. ↵
      Statistics New Zealand. National population estimates: At 20 June 2014. [updated 14 Aug 2014, 20 Sep 2016] http://www.stats.govt.nz/browse_for_stats/population/estimates_and_projections/NationalPopulationEstimates_HOTPAt30Jun14.aspx.
    4. ↵
      1. Ruseckaite R,
      2. Collie A
      . Repeat workers’ compensation claims: risk factors, costs and work disability. BMC Public Health 2011;11:492.doi:10.1186/1471-2458-11-492
      OpenUrlCrossRefPubMed
    5. ↵
      1. Wasiak R,
      2. Kim J,
      3. Pransky G
      . Work disability and costs caused by recurrence of low back pain: longer and more costly than in first episodes. Spine 2006;31:219–25.doi:10.1097/01.brs.0000194774.85971.df
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Ruseckaite R,
      2. Collie A
      . The incidence and impact of recurrent workplace injury and disease: a cohort study of WorkSafe Victoria, Australia compensation claims. BMJ Open 2013;3:e002396.doi:10.1136/bmjopen-2012-002396
      OpenUrlAbstract/FREE Full Text
    7. ↵
      1. Galizzi M
      . On the recurrence of occupational injuries and workers' compensation claims. Health Econ 2013;22:582–99.doi:10.1002/hec.2829
      OpenUrlCrossRefPubMed
    8. ↵
      1. Cherry NM,
      2. Sithole F,
      3. Beach JR, et al
      . Second WCB claims: who is at risk? Can J Public Health 2010;101(Suppl 1):S53–7.
      OpenUrlPubMed
    9. ↵
      1. Hedges BE,
      2. Dimsdale JE,
      3. Hoyt DB, et al
      . Characteristics of repeat trauma patients, San Diego County. Am J Public Health 1995;85:1008–10.doi:10.2105/AJPH.85.7.1008
      OpenUrlCrossRefPubMedWeb of Science
    10. ↵
      1. Kaufmann CR,
      2. Branas CC,
      3. Brawley ML
      . A population-based study of trauma recidivism. J Trauma 1998;45:325–32.doi:10.1097/00005373-199808000-00019
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      1. Smith RS,
      2. Fry WR,
      3. Morabito DJ, et al
      . Recidivism in an urban trauma center. Arch Surg 1992;127:668–70.doi:10.1001/archsurg.1992.01420060034006
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. Worrell SS,
      2. Koepsell TD,
      3. Sabath DR, et al
      . The risk of reinjury in relation to time since first injury: a retrospective population-based study. J Trauma 2006;60:379–84.doi:10.1097/01.ta.0000203549.15373.7b
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Williams JM,
      2. Furbee PM,
      3. Hungerford DW, et al
      . Injury recidivism in a rural ED. Ann Emerg Med 1997;30:176–80.doi:10.1016/S0196-0644(97)70139-2
      OpenUrlCrossRefPubMedWeb of Science
    14. ↵
      1. Keays G,
      2. Swaine B,
      3. Ehrmann-Feldman D
      . Association between severity of musculoskeletal injury and risk of subsequent injury in children and adolescents on the basis of parental recall. Arch Pediatr Adolesc Med 2006;160:812–6.doi:10.1001/archpedi.160.8.812
      OpenUrlCrossRefPubMed
    15. ↵
      1. Johnston BD,
      2. Martin-Herz SP
      . Correlates of reinjury risk in sibling groups: a prospective observational study. Pediatrics 2010;125:483–90.doi:10.1542/peds.2009-1594
      OpenUrlAbstract/FREE Full Text
    16. ↵
      1. Braun PA,
      2. Beaty BL,
      3. DiGuiseppi C, et al
      . Recurrent early childhood injuries among disadvantaged children in primary care settings. Inj Prev 2005;11:251–5.doi:10.1136/ip.2004.006510
      OpenUrlAbstract/FREE Full Text
    17. ↵
      1. Pengel LH,
      2. Herbert RD,
      3. Maher CG, et al
      . Acute low back pain: systematic review of its prognosis. BMJ 2003;327:323.doi:10.1136/bmj.327.7410.323
      OpenUrlAbstract/FREE Full Text
    18. ↵
      1. van Rijn RM,
      2. van Os AG,
      3. Bernsen RM, et al
      . What is the clinical course of acute ankle sprains? a systematic literature review. Am J Med 2008;121:324–31.doi:10.1016/j.amjmed.2007.11.018
      OpenUrlCrossRefPubMedWeb of Science
    19. ↵
      1. Prior M,
      2. Guerin M,
      3. Grimmer K
      . An evidence-based approach to hamstring strain injury: a systematic review of the literature. Sports Health 2009;1:154–64.doi:10.1177/1941738108324962
      OpenUrlCrossRefPubMed
    20. ↵
      1. Sherry MA,
      2. Best TM,
      3. Silder A, et al
      . Hamstring strains: basic science and clinical research applications for preventing the recurrent injury. Strength Cond J 2011;33:56–71.doi:10.1519/SSC.0b013e31821e2f71
      OpenUrl
    21. ↵
      1. Harcombe H,
      2. Derrett S,
      3. Samaranayaka A, et al
      . Factors predictive of subsequent injury in a longitudinal cohort study. Inj Prev 2014;20:393–400.doi:10.1136/injuryprev-2014-041183
      OpenUrlAbstract/FREE Full Text
    22. ↵
      The Accident Compensation Corporation. Explanation of terms, 2008. http://www.acc.co.nz/about-acc/statistics/acc-injury-statistics-2008/0-introduction/IS0800006 (accessed 14 Dec 2016).
    23. ↵
      1. Derrett S,
      2. Langley J,
      3. Hokowhitu B, et al
      . Prospective outcomes of injury study. Inj Prev 2009;15:e3.doi:10.1136/ip.2009.022558
    24. ↵
      1. Derrett S,
      2. Davie G,
      3. Ameratunga S, et al
      . Prospective Outcomes of Injury Study: recruitment, and participant characteristics, health and disability status. Inj Prev 2011;17:415–8.doi:10.1136/injuryprev-2011-040044
      OpenUrlAbstract/FREE Full Text
    25. ↵
      1. Derrett S,
      2. Harcombe H,
      3. Wyeth E, et al
      . Subsequent Injury Study (SInS): Improving outcomes for injured New Zealanders. Inj Prev 2011;17:415–8.
      OpenUrlAbstract/FREE Full Text
    26. ↵
      Statistics New Zealand. 2006 Census of Population and Dwellings. 2006.
    27. ↵
      Ministry of Health. Ethnicity data protocols for the health and disability sector. Wellington: Ministry of Health, 2004.
    28. ↵
      Statistics New Zealand. New Zealand standard classification of occupations. Wellington: Statistics New Zealand, 2001.
    29. ↵
      1. Derrett S,
      2. Samaranayaka A,
      3. Wilson S, et al
      . Prevalence and predictors of sub-acute phase disability after injury among hospitalised and non-hospitalised groups: a longitudinal cohort study. PLoS One 2012;7:e44909.doi:10.1371/journal.pone.0044909
    30. ↵
      1. Langley J,
      2. Davie G,
      3. Wilson S, et al
      . Difficulties in functioning 1 year after injury: the role of preinjury sociodemographic and health characteristics, health care and injury-related factors. Arch Phys Med Rehabil 2013;94:1277–86.doi:10.1016/j.apmr.2013.02.012
      OpenUrl
    31. ↵
      The Accident Compensation Corporation. Am I covered? [updated 20 Jan 2016, 1 Apr 2016]. http://www.acc.co.nz/making-a-claim/am-i-covered/index.htm.
    32. ↵
      The Accident Compensation Corporation. Lodge and manage claims, 2014. [updated 20 Feb 2014, 2 Mar 2017]. http://www.acc.co.nz/for-providers/lodge-and-manage-claims/PRV00030.
    33. ↵
      The Accident Compensation Corporation and The Ministry of Health. Accident services: a guide for DHBs and ACC staff, 2014. http://www.acc.co.nz/PRD_EXT_CSMP/groups/external_providers/documents/guide/wpc120337.pdf.
    34. ↵
      The Accident Compensation Corporation. Using read codes. Updated Oct 2015. http://www.acc.co.nz/for-providers/lodge-and-manage-claims/PRV00037 (cited Jan 2017).
    35. ↵
      Medicode (Firm). ICD-9-CM: International classification of diseases, 9th revision. Clinical Modification Salt Lake City, Utah: Medicode, 1996.
    36. ↵
      NCCH (National Centre for Classification in Health). The international statistical classification of diseases and related health problems, tenth revision, Australian Modification. 3rd edn. Sydney: NCCH, Faculty of Health Sciences, The University of Sydney, 2002. http://meteor.aihw.gov.au/content/index.phtml/itemId/270546
    37. ↵
      1. Stevenson M,
      2. Segui-Gomez M,
      3. Lescohier I, et al
      . An overview of the injury severity score and the new injury severity score. Inj Prev 2001;7:10–13.doi:10.1136/ip.7.1.10
      OpenUrlAbstract/FREE Full Text
    38. ↵
      1. Wilson SJ,
      2. Derrett S,
      3. Cameron ID, et al
      . Prevalence of poor outcomes soon after injury and their association with the severity of the injury. Inj Prev 2014;20:57–61.doi:10.1136/injuryprev-2012-040690
      OpenUrlAbstract/FREE Full Text
    39. ↵
      StataCorp. Stata statistical software: release 14. College Station, TX: StataCorp LP. 2015.
      1. Cole TJ
      . Too many digits: the presentation of numerical data. Arch Dis Child 2015;100:608–9.doi:10.1136/archdischild-2014-307149
      OpenUrlFREE Full Text

    Footnotes

    • Contributors HH led the preparation of this paper and wrote the draft manuscript, HH, AS and GD undertook the analyses, SD and HH lead the Subsequent Injury Study project and SD leads the Prospective Outcomes of Injury Study project. All authors contributed to the design of the study, the interpretation of data and the writing and editing of the manuscript. All authors have read and approved the final manuscript.

    • Funding The Subsequent Injury Study was funded by the Health Research Council of New Zealand (2015-2017). The Prospective Outcomes of Injury Study was funded by the Health Research Council of New Zealand (2007-2013) and cofunded by the Accident Compensation Corporation, New Zealand (2007-2010). The views and conclusions in the article are of the authors and may not represent those of the funders.

    • Competing interests None declared.

    • Ethics approval New Zealand Health and Disability Multi-Region Ethics Committee (MEC/07/07/093).

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data sharing statement Due to ethical constraints, the data cannot be shared but anyone interested in pursuing collaborative research should contact