Identifying Surgical and Trauma Patients in New Zealand for Opioid-Related Pharmacoepidemiological Research: A Descriptive Study

Round 1

Reviewer 1 Report

The authors summarised a process of identifying and describing cohorts of patients who have either undergone surgery or were admitted to a hospital for trauma in NZ between 2007 and 2019. The authors aimed to use these cohorts to describe the prevalence of persistent opioid use in surgical and trauma patients and to assess the clinical outcomes of morbidity and mortality related to persistent opioid use. The authors claimed that their study would allow future studies to explore other influences on the outcomes of surgery and trauma in NZ.

However, how the description of the prevalence of persistent opioid use in surgical and trauma patients and the assessment of the clinical outcomes of morbidity and mortality related to persistent opioid use in NZ would benefit future studies to explore other influences on the outcomes of surgery and trauma in NZ are unclear. The authors did not highlight, nor provide, the scientific merit of their summary/findings. The research question is unclear. The purpose of the study is hard to identify. 

Author Response

Thank you for your review. We are sorry the purpose of our study was not clearer. It was to provide data on a clearly defined large group of patients undergoing surgery or trauma admitted to hospital for use in our own studies on persistent opioid usage and potentially for future studies by other investigators on various questions related to surgery and trauma in New Zealand. We have updated the Discussion to reflect this (lines 165-170):

Discussion section lines 165 to 170:

‘Our database provides a valuable source of information on the characteristics of patients admitted to hospital for trauma or surgery. In addition, it will assist with the design of potential future prospective studies of various interventions to improve patient outcomes, notably by facilitating estimates of the sample size required for such studies in particular subgroups of the patients in our cohort. Moreover, the information could potentially be used for comparison with other countries, or (with ongoing collection and analysis) for tracking opioid use and associated adverse effects in New Zealand.’

Reviewer 2 Report

Thank you for the opportunity to review this manuscript titled “Identifying surgical and trauma patients in New Zealand for opioid-related pharmacoepidemiological research: A descriptive study”. The authors conducted a population-based retrospective descriptive study involving all hospital centres in NZ among patients admitted to a NZ hospital with trauma or who underwent interventional surgical procedures between 1st January 2007 to 31st December 2019. Overall, the manuscript is well written and provides detailed description of the data source, variables used in the analysis, results, and discussion. However, this study only provides a descriptive summary of the baseline variables and does not actually compare the two cohorts. I have few minor comments for the authors to address.

 

1.     Line 111: When was length of stay measured? When were other event variables such as admission type measured?

2.     Line 143, Table 1: The authors should provide p values to determine if there was a significant difference between the two cohorts with respect to baseline sociodemographic and event data.

3.     Line 154: The authors should add text regarding the hypothesis for differences in the two cohorts with respect to clinical outcomes. Otherwise, this study does not have a strong rationale and relevance for the readers.

 

 

 

Author Response

Thank you for your constructive feedback and suggestions. At your suggestion, we have added the new results of chi-squared tests to facilitate comparison between the three cohorts. (see below and lines 101-103, 138-143 and table 1 in the manuscript).

Method section lines 101-103

‘This resulted in three independent cohorts for further analysis: surgical only, trauma only and trauma with surgery. At the request of a reviewer, we then undertook statistical tests to facilitate comparisons between these three cohorts of patients.’

Method section lines 138 to 143 and table 1

‘Variable reporting and data analysis

Categorical variables were reported as percentages. Continuous data were converted to categorical variables based on clinical relevance and data distribution by discussion and consensus with a multidisciplinary team. Chi-square tests were used to assess differences in baseline characteristics between the different cohorts. We designated a two-tailed P value < 0.05 as statistically significant. No correction was made for multiple testing.  Results are presented in Table 1. The analyses were done using SPSS v28 IBM corporation, Armonk NY, USA.’

1. Line 111:When was length of stay measured? When were other event variables such as admission type measured?

The variables were measured at the time of hospitalisation as part of routinely collected health data during the hospital visit (Method section lines 121-122).

Method section 121-122

‘These event variables were recorded at the time of a patient’s hospital admission as part of routinely collected health data.’

 

2. Line 143, Table 1:The authors should provide p values to determine if there was a significant difference between the two cohorts with respect to baseline sociodemographic and event data.

To determine if significant differences exist between the three cohorts, we have added the P-values to determine if there were differences between the different cohorts. See discussions lines 171-172, 252-253 and Table 1.

Discussion section lines 171-172:

‘We found significant differences between the three primary cohorts across all variables (P<0.001 in each case). Clinically significant differences between cohorts were seen…’

Discussion Section 252-253

‘Differences between the three cohorts should be interpreted cautiously noting that allocation to cohorts was not randomized and that a large number of statistical tests were undertaken, making type I errors quite likely’.

 

3. Line 154:The authors should add text regarding the hypothesis for differences in the two cohorts with respect to clinical outcomes. Otherwise, this study does not have a strong rationale and relevance for the readers.

We thank the reviewer for this suggestion but note that we did not plan any hypothesis testing in this study. Thus, we did not register the study or designate a primary outcome variable or any other way of definitely deciding whether to accept or reject any particular hypothesis. This is because the aim of the study was to generate a database of information on two large, defined cohorts of patients to use in describing the characteristics of surgical and trauma patients admitted to hospitals in NZ and in in our own research on persistent postoperative opioid usage. It is somewhat self-evident that the three cohorts would be different. However, we have now undertaken statistical testing and included P values to facilitate a general comparison between the three groups. Because this was not planned, we have noted (for transparency) that we have done this at your suggestion.

Please see Methods section (lines 101-103, 138-143 and table 1) and changes to the abstract (lines 18-30), discussion (lines 165-170, 189-190, 202-205) and conclusion (lines 255-261) to reflect the differences identified and implications for future research.

Abstract section lines 18-30

Unique aspects of New Zealand’s (NZ) health system allow for a novel pharmacoepidemiologic approach to conducting population-based clinical research. Identifying suitable surgical and trauma patients is crucial for opioid-utilisation and adverse outcome studies. We aimed to describe all patients admitted to a NZ hospital with trauma or who had undergone surgery between 1st January 2007 to 31st December 2019. This was a retrospective population-based study involving all hospital centres in NZ. No age limit was placed on the cohorts. We excluded patients with hospitalisation episodes for surgery or trauma one year before the event. We identified 1.78 million surgical only patients, 633,386 trauma only, and 250,800 trauma with surgery patients. A total of 18 comorbidities were extracted for each cohort. The cohorts had notable differences in ethnicity, domicile region, length of stay, previous opioid use and baseline comorbidities. We have identified and described the cohorts of patients who were admitted to hospital either for surgery or trauma in NZ between 2007-2019. These cohorts will be used for studies describing the prevalence and outcomes related to persistent opioid use. Our findings of significant differences in baseline characteristics across the cohorts suggests studies should treat surgical and trauma patients separately.

Abstract word count 198/200

Discussion section lines 165 to 170:

‘Our database provides a valuable source of information on the characteristics of patients admitted to hospital for trauma or surgery. In addition, it will assist with the design of potential future prospective studies of various interventions to improve patient outcomes, notably by facilitating estimates of the sample size required for such studies in particular subgroups of the patients in our cohort. Moreover, the information could potentially be used for comparison with other countries, or (with ongoing collection and analysis) for tracking opioid use and associated adverse effects in New Zealand.’  

Discussion section lines 189-190:

‘Across the three cohorts, the trauma only cohort had the greatest proportion of patients with both pre-trauma opioid use and pre-trauma diagnosis of opioid use disorder.’

Discussion section lines 202-205:

‘Trauma only patients may have more risk factors than surgical only, or trauma with surgery patients for developing persistent opioid use. The trauma only cohort had a higher rate of pre-trauma opioid use, pre-trauma diagnosis of opioid use disorder, comorbidity burden, mental health disorder and chronic pain; all these variables have been shown to have a significant association with persistent opioid use.’

Conclusion now edited to read as follows - lines 255-261:

‘We have identified and described cohorts of patients who have either undergone surgery or were admitted to a hospital for trauma in NZ between 2007 and 2019. We plan to use these cohorts in a series of studies to describe the prevalence of persistent opioid use in surgical and trauma patients and to assess the clinical outcomes of morbidity and mortality related to persistent opioid use. Significant differences in baseline characteristics in relation to persistent opioid use were observed across the two cohorts, indicating that future studies on opioid-related adverse outcomes should treat surgical and trauma patients separately. These patient cohorts are also available for other researchers wishing to explore other influences on the outcomes of surgery and trauma in NZ.’

Minor edits in text as follows:
Lines 8-9
‘University of Health Sciences and Pharmacy, St Louis’
Line 53
‘The first step towards this type of study is to identify from…’
Line 107
‘… domicile region in NZ, and the region of the healthcare facility…’
Lines 145-146
‘Between 2007-2019, we identified 1.78 million surgical only patients, 633,386 trauma only, and 250,800
trauma with surgery.’
Lines 158-160
‘… 415,926 patients (23.2%) had an ASA score of one, 365,440 (20.4%) of two, 98,031 (5.5%) three
and 16,181 (0.9%) four. The urgency of the surgery for the surgical only cohort was elective for 191,066
patients (10.7%) and acute for 1,598,341 patients (89.3%).’
Line 164
‘…substantial cohorts for our group’s planned opioid-related research…’
Lines 177-178
‘The NZ national census in 2018 reported the population of European ethnicity to be 70.2%, Māori
16.5%, Pacific 8.1%, Asian 15.1%, and MELAA 1.5%.
Line 179
‘…trauma with surgery but underrepresented in terms of surgery only…’
Line 180
‘In all three cohorts Auckland was the patient domicile region with the most patients.’
Lines 187-188
‘For our subsequent studies on persistent opioid use and its related harm, other important
confounding…’
Line 192
‘Of the 18 comorbidities examined,’
Line 200
‘… event at 17.8%, compared to 9.2% in surgical only patients and 8.8% in trauma with surgery
patients.’
Line222
‘…most severe.’’
Lines 238-240
‘Thus, medication dispensing in small quantities and for shorter duration courses may be underreported. Data before 2006 are not recorded comprehensively. Hence our cohort identification only
encompassed the years 2007-2019.’
Line 250
‘, it is mandatory to report this variable to the MoH.’ 

Round 2

Reviewer 1 Report

The authors added 5 lines (165-170) of text without fully addressing the core of my first comment: "The authors claimed that their study would allow future studies to explore other influences on the outcomes of surgery and trauma in NZ. However, how the description of the prevalence of persistent opioid use in surgical and trauma patients and the assessment of the clinical outcomes of morbidity and mortality related to persistent opioid use in NZ would benefit future studies to explore other influences on the outcomes of surgery and trauma in NZ are unclear. The authors did not highlight, nor provide, the scientific merit of their summary/findings. The research question is unclear. The purpose of the study is hard to identify. " in which major revision was implied and expected.

Author Response

Thank you for your review. The core purpose and aim of our paper were to describe the methods used to identify surgical and trauma cohorts of patients in New Zealand and to summarise their relevant baseline characteristics. We intend to use these cohorts of patients in future research on the persistent use of opioids (and other medicines) after surgery and trauma. We believe other investigators may find these cohorts of patients useful for various potential investigations into different aspects of surgery and trauma, particularly but perhaps not only in New Zealand. We have further revised our paper to clarify and highlight the purpose, and the scientific merit of the study. To be absolutely clear, the study did not have any hypotheses. It was a cohort identification study using routinely collected health data in NZ, and the paper reports on the methods used to identify the cohort and the characteristics of this cohort.

The scientific merit of our work lies in the description of this important group of patients, and in the creation of an accessible dataset for use in future research by ourselves and (potentially) others. This information is novel, and this group of patients is important from many perspectives, including those of policy.

Note in particular that the current paper does not include work aimed at assessing the prevalence of persistent opioid use or other outcomes related to surgical and trauma patients. We plan to do this in future research using this dataset. That is why we have described the potential applications of cohort identification which include exploring influences on the outcomes of surgery and trauma as an example. To further clarify this, we have revised our text more extensively, specifically the abstract (lines 19-30), introduction (lines 39-40, 43-45, 55-61), and discussion (lines 167-181, previously lines 165-170 as per reviewer comment). In addition to these changes, we have updated the method section and table 1 at the request of reviewer 2 to facilitate a comparison of the three cohorts (lines 105-107, lines 143-147). The following section in the discussion (lines 213-216, lines 261-263) and conclusion (lines 268-270) were also updated to elaborate on our findings and how they may inform future studies.

Abstract section lines 19-30:

‘Unique aspects of New Zealand’s (NZ) health system allow for a novel pharmacoepidemiologic approach to conducting population-based clinical research. A defined cohort of surgical and trauma patients would facilitate future studies into opioid utilisation, outcomes, and other questions related to surgery and trauma. We aimed to describe all patients admitted to a NZ hospital with trauma or to undergo surgery between 1st January 2007 to 31st December 2019. This was a retrospective population-based study involving all hospital centres in NZ. We excluded patients with hospitalisation episodes for surgery or trauma one year before the event. We identified 1.78 million surgical only patients, 633,386 trauma only, and 250,800 trauma with surgery patients. A total of 18 comorbidities were extracted for each cohort. These cohorts were distinct, with differences in ethnicity, domicile region, length of stay, previous opioid use and baseline comorbidities. We plan to use this dataset for future research into the prevalence and outcomes of persistent opioid use, and to make our dataset available to other researchers. Our findings of significant differences between cohorts suggest studies should treat surgical and trauma patients separately.’

Word count 182/200

Introduction section lines 39-40

‘Opioid dependence and the harm directly and indirectly associated with this are a major public health problem in many countries’

Introduction section lines 43-45

‘One way of doing this would be through the retrospective analysis of large datasets collected for administrative or other purposes, but such datasets are not available in all regions of the world.’

Introduction section lines 55-61:

‘The purpose of our current study was to use routinely collected health data from the NZ MoH to identify and describe a cohort of patients admitted to any NZ hospital for trauma or to undergo a surgical procedure between 1st January 2007 to 31st December 2019. We aim to then use these cohorts to investigate the prevalence and outcomes of persistent opioid use after surgery and trauma in NZ. The demonstration that it is feasible to identify and collect considerable information on large cohorts of patients admitted to NZ hospitals for surgery or trauma, and access to our methods and datasets, may also be of interest to other investigators and other stakeholders for research into various aspects of surgery and trauma in NZ.’

Method section lines 105-107

‘At the request of a reviewer, we then undertook statistical tests to facilitate comparisons between these three cohorts of patients.’

Method section lines 143-147

‘Categorical variables were reported as percentages. Continuous data were converted to categorical variables based on clinical relevance and data distribution by discussion and consensus with a multidisciplinary team. Chi-square tests were used to assess differences in baseline characteristics between the different cohorts. We designated a two-tailed P value < 0.05 as statistically significant. No correction was made for multiple testing. Results are presented in Table 1. The analyses were done using SPSS v28 IBM corporation, Armonk NY, USA.’

Discussion section lines 167 to 181:

‘These represent substantial cohorts for our group’s planned opioid-related research. The fact that such a large number of patients undergoing hospitalisation for surgery or trauma can be identified, and that substantial information can be collected about them, may also be of value to others conducting studies of surgical outcomes in NZ. Even as they stand, our results provide a valuable (and, we believe, novel) picture of the characteristics of patients admitted to hospitals for trauma or surgery in NZ.

Our proposed research into the prevalence and outcomes of persistent opioid use in these cohorts of patients will provide hitherto unavailable information on the extent of this public health problem in NZ that will inform health policy. It will also allow comparisons to be made between NZ and other countries, noting that there are considerable socio-political and healthcare differences between NZ and many other high-income countries. Given that the datasets used in this study are maintained on an ongoing basis, our approach opens the opportunity to monitor changes in the prevalence and outcomes of opioid use after surgery or trauma and thus evaluate any interventions arising from changes in healthcare policy (for example). Furthermore, our data could inform the design of potential future prospective studies of investigator-initiated interventions to improve outcomes after trauma or surgery, including but not only informing estimates of the required sample size for such studies.’

Discussion section lines 213-216

‘Patients admitted with trauma may have more risk factors for developing persistent opioid use than those admitted for surgery only, or trauma with surgery. For example, the trauma only cohort had a higher rate of pre-trauma opioid use, pre-trauma diagnosis of an opioid use disorder, comorbidity burden, mental health disorder and chronic pain, and all these factors have been shown to have a significant association with persistent opioid use.’

Discussion section lines 261-263

‘Differences between the three cohorts should be interpreted cautiously noting that allocation to cohorts was not randomised and that a large number of statistical tests were undertaken, making type I errors quite likely.’

Conclusion section lines 268-270:

‘Significant differences in baseline characteristics in relation to persistent opioid use were observed across the three cohorts, suggesting that future studies on opioid-related adverse outcomes should treat surgical and trauma patients separately.’