Prolonged opioid use after distal radius fracture

Prolonged opioid use (more than 90 days) after injury puts the patient at risk for adverse effects. We investigated the patterns of opioid prescription after distal radius fracture and the effect of pre‐ and post‐fracture factors on the risk for prolonged use.


| INTRODUCTION
Opioids have a proven effect on acute pain and are commonly prescribed to treat pain after traumatic injury or surgery (Chou et al., 2014). However, the evidence on its effect on persistent non-cancer pain (over 3 months) is weak (Bialas et al., 2020;Chou et al., 2014). Instead, many studies show risk for adverse effects (Chou et al., 2014;Dowell et al., 2016;Kotlińska-Lemieszek & Żylicz, 2022), including addiction, overdose (Dowell et al., 2016), possible sensitization to, or increased, pain (Kotlińska-Lemieszek & Żylicz, 2022;Rivat & Ballantyne, 2016) and increased all-cause mortality (Kotlińska-Lemieszek & Żylicz, 2022;von Oelreich et al., 2020). Long-term opioid use is therefore a concern not only for the individual but also for the health care systems in many countries. Over the past decade, the long-term use of prescribed opioids, especially synthetic opioids, has risen dramatically in the USA (Meyer et al., 2020) leading to misuse, addiction and overdoses in what is commonly referred to as the opioid epidemic (Shipton et al., 2018). This rapid rise has not been seen in Europe thus far, but the use of the same synthetic opioids has led to concerns that a similar situation could develop (Heilig & Tägil, 2018;Meyer et al., 2020).
One of the most common fractures in Western Europe is distal radius fracture, which affects people of all ages (Court-Brown & Caesar, 2006;Porrino et al., 2014;Rundgren et al., 2020). This fracture normally causes acute pain immediately after both the injury and the initial treatment (reduction/cast/splint/surgery). This acute pain is normally severe but decreasing over time, with most patients having mild or no pain after 3 months, although around 11% report persistent moderate to severe pain (MacDermid et al., 2003). Standard treatment for distal radius fracture in Sweden is either a splint or surgical reduction and fixation followed by occupational and or physical therapy. The acute pain caused by the fracture is typically treated with paracetamol, non-steroid analgesic drugs (NSAID) or, in more severe cases, opioids (Heilig & Tägil, 2018). Patients with fracture types requiring surgery for reduction are also at risk for postsurgical pain which is commonly treated with opioids. Therefore, for some patients, distal radius fracture could potentially act as a catalyst for prolonged opioid use with associated risks.
To better identify patients that are at risk for prolonged opioid use, before treatment initiation, is of great importance to prevent the potentially harmful side effects. Therefore, this study aims to investigate the pattern of opioid prescription after distal radius fracture. We assess the proportion of patients that continue opioid treatment after 3 months, since this constitutes a risk of iatrogenic dependency. We specifically study the importance of prior pain, prior prescription of opioids and prior comorbid mental illness, on the risk for prolonged opioid use.

| METHOD
We used a register-based cohort design and linked information on all individuals residing in the southernmost region of Sweden, Skåne, during 2015-2018, from four population registers. The linkage was executed by using the ten digit personal identity number which all Swedish residents are assigned (Ludvigsson et al., 2009).

| Data sources
2.1.1 | Skåne health care register The Skåne health care register contains information on all health care visits in the region of Skåne, Sweden (population 1.38 million, 13.4% of the national population). Data from all consultations in primary, secondary and tertiary care are automatically registered. Recorded variables include diagnosis, coded by International Classification of Disease 10th revision (ICD 10), codes for surgical and medical procedures as well as date of visit (Löfvendahl et al., 2020).

| Longitudinal integration database
for health insurance and labor market studies (LISA) LISA is held by Statistics Sweden, the national authority for official statistics. The register contains socioeconomic data on the Swedish population, including income, type of work and highest achieved level of formal education.

| The total population register
The total population register, governed by Statistics Sweden, contains vital data on the Swedish population such as births, deaths and information on municipality, including relocation or migration (Ludvigsson et al., 2016).

| Swedish prescribed drug register
This register is held by the Swedish National Board of Health and Welfare and contains information on all dispensed prescription medications for the Swedish population, including, type of drug, dosage, pack size and time of prescription as well as time of purchase (Wettermark et al., 2007).

| Study population
The cohort includes all patients with a distal radius fracture diagnosis recorded in the Skåne Health Care Register between the years 2015 and 2018. Additionally, the patients were required to be 18 years or older and residents of the region at the time of injury. Distal radius fracture was defined as an ICD-10 code of S52.5 (fracture of lower end of radius) or S52.6 (fracture of lower end of both ulna and radius), from either a department of emergency medicine, orthopaedics, or hand surgery, where distal radius fractures are treated. Patients with a diagnosis of distal radius fracture in the year 2014 were excluded from analysis to avoid misclassification, i.e., classifying follow-up visits for a previous fracture as a new fracture. The first registered diagnosis of distal radius fracture 2015-2018 was defined the index fracture. Patients were then followed for 1 year to determine opioid use. Patients who suffered a new fracture within the following 1 year after the index fracture were excluded from the study to avoid misclassification. Patients who had surgery for distal radius fracture more than 3 months after index fracture were also excluded to avoid the risk to include unknown new fractures, i.e., the first fracture was handled without surgery, but a second fracture required surgery. Patients who died or moved from the region within the follow-up period (1 year from fracture) were excluded (N = 514, 5.2%).

| Prolonged opioid use after fracture
Data on prescribed and dispensed opioids were collected from the Swedish prescribed drug register. Purchase of prescribed opioids was assigned to four different time periods depending on date of purchase. The time period 1-3 months after fracture was defined as the treatment period, month 4-6 after fracture was defined as follow-up period 1, month 7-9 as follow-up period 2 and month 10-12 as follow-up period 3. Having at least one purchase in both the treatment period and in follow-up period 1 was defined as having prolonged use. To be defined as having prolonged use in, e.g. follow-up period 2 a patient would have to have a purchase in all the preceding three-month periods from fracture (see Figure 1).

| Prior opioid use
Any prior opioid use was defined as any purchase of prescribed opioids in the period 365 days before the fracture up to 14 days before fracture. The 14 days before fracture were not counted since there was a risk of patients receiving a prescription for the pain of distal radius fracture before the diagnosis was registered, potentially leading to misclassification.
In a sensitivity analysis, any prior opioid use was stratified and analysed separately for those with regular use in the last year and those with some use in the last year. Regular use was defined as purchase of opioids in three out of four quarters of a year. Years were counted as multiples of 365 days from the fracture, except for the year before fracture which was defined as 365 days to 14 days before fracture to avoid misclassification (see previous paragraph). Some use was defined as any use during the last year before fracture that did not reach the criteria for prior regular use.
Among those with no regular use in the last year, we additionally studied presence or absence of regular use in years 2-5 years before fracture. This was done for those patients who were adults (18+) and living in | 851 LARROSA PARDO et al. Skåne during that entire period (N = 8704, % of study population) the rest of the study population was excluded from this analysis. For those that had a regular use of opioids in this period, the latest year with regular use was registered.

| Prior illnesses
Mental Illness was defined as a registered ICD-10diagnosis of any mental and/or behavioural disorder except substance addiction in the Skåne Healthcare Register (ICD-10: F0 or F2-F9) 365 days before the day of fracture up until the day before fracture.
Addiction was defined as a registered ICD-10 diagnosis of any substance addiction (ICD-10: F1) anytime between 365 days before the day of fracture up until the day before fracture.

| Treatment
Using the Swedish Classification of Care Measures, surgery for distal radius fracture was defined as Surgical codes starting with NDJ, 'surgery for fracture on wrist hand or fingers', excluding code NDJ09 (closed reduction of fracture), registered in the Skåne Health Care Register, from date of fracture up until 90 days after fracture.
Occupational and physical therapy treatment was defined as a visit registered in the Skåne Health Care Register to an occupational therapist or physiotherapist in any clinic, but related to distal radius fracture, (ICD-10 S52.5 or distal radius and ulna ICD-10 S52.6) within 3 months of the initial fracture.

| Socioeconomic factors
Age at time of fracture was categorized into 18-49, 50-59, 60-69, 70-79 and 80 years or older. Information on sex was coded binary as male/female. The highest achieved level of formal education (seven categories) was collected from the LISA register 2010. We classified education into three groups based on years of formal education, 9 years or less, 10-12 years and more than 12 years. Education in Sweden is free at all levels, with the first 9 years of school being mandatory, corresponding to the first category, upper secondary school is 3 years, corresponding to the second category and the third category corresponds any form of higher education, i.e., college/university level. We used education as a proxy for socioeconomic status.

| Statistical analysis
Descriptive statistics are presented as number and percent. The proportions of patients with prolonged opioid use during the different follow-up periods were calculated for all patients and separately by exposure status (any opioid use in the last year, prior mental illness, prior pain, occupational/physical therapy treatment, surgery), the Cochrane Armitage trend test was used to test for trends between groups across the different follow-up periods. Finally, descriptive data were calculated restricted to patients who had no opioid use, defined as opioid-naïve the year before fracture.
Relative risk for prolonged opioid use associated to the different exposures was calculated through modified Poisson regressions (Zhao, 2013), each exposure was tested separately. First, each model was adjusted for sex and age. Second, all models, except for surgery, were adjusted for education (since education was not considered a confounder to the association surgery-prolonged opioid use). In the third model, different adjustments were made depending on the potential confounders of the exposure under study. Specifically, the effect of prior opioid use was additionally adjusted for prior mental illness, addiction and pain symptoms. The effect of prior mental illness was additionally adjusted for prior pain and prior addiction. The effect of prior pain was additionally adjusted for prior addiction. The effect of occupational or physical therapy treatment was additionally adjusted for surgery and prior pain. We also performed the same analysis on patients with no opioid use in the year before fracture, with the same adjustments as previously described.
In a separate sensitivity analysis, we estimated the effect of regular previous opioid use in the last year compared to some use or no opioid use.
To examine the association between previous regular use of opioids in the past 5 years before fracture and prolonged opioid use after fracture, we included the exposure as latest year with regular use (i.e. 2,3, 4 or 5 years) before fracture and adjusted for sex, age group and education level. No regular use during the last 5 years was used as the reference level. Latest regular use in the last year preceding the fracture was not analysed here since it was analysed separately in the sensitivity analysis above.
All statistical analyses were conducted using the SAS software, Version: Enterprise 8.1. (SAS Institute Inc.). The study was approved by the Regional ethical review board In Lund (Dnr: 2011/432).

| RESULTS
In total, 9369 adult patients living in Skåne with a registered diagnosis of distal radius fracture in the years 2014-2018 met the inclusion criteria and were included in the study. 7116 (76%) were female and 2084 (22.2%) were treated with surgery, Table 1.

| Opioid treatment for the fracture
During the treatment period (month 1-3), 35.2% purchased any prescribed opioids. Among the patients who were treated with surgery for their distal radius fracture, the proportion of patients who purchased opioids in the treatment period was larger compared to those with no surgery, 77.0% versus 23.2% (Table 1). Among those with any prior opioid use the year before the fracture, 58.8% were treated with opioids as compared to 31.1% of those with no opioids last year. Of the patients with prior mental illness, 40.96% purchased opioids in the treatment period compared to 33.6% of those without prior mental illness. Regarding patients with prior pain symptoms, 43.5% purchased opioids during treatment as compared to 34.0% among those with no prior pain symptoms. The proportion of patients using opioids during treatment was larger for patients who also attended physical or occupational therapy during treatment as compared to those who did not, 38.8% versus 31.4 (Table 2).

| Opioid-naïve patients
Among the opioid-naïve patients, i.e., those with no opioid prescribed the year before fracture, 31.09% purchased any prescribed opioids during the treatment period. For those that were treated with surgery 75.9% purchased opioids in the treatment period compared to 18.1% in those who did not have surgery. Among the opioid-naïve patients with prior mental illness, 33.1% purchased opioids in the treatment period compared to 30.6% among those without prior mental illness Opioid-naïve patients with previous pain symptoms had a proportion of 34.6% who purchased opioids in compared to those without previous pain, 30.8%. The opioid-naïve patients who attended physical or occupational therapy had a higher proportion of opioid purchasers, 35.6%, in the treatment period compared to those who did not attend, 26.3% (Table 3). The period between date of fracture and the following 3 months.

| Prolonged opioid use
In total, 664 patients (7.1%) purchased opioids both in the treatment period and in the first follow-up period (months 4-6 after fracture) and were thus classified as prolonged users. Moreover, 4.9% and 4.2% continued purchasing opioids in the second and third follow-up period, respectively. Among patients with any opioid use in the last year before the fracture, 33.7% were prolonged users as compared to 2.5% among those with no opioid use last year (p < 0.01) (Tables 2 and 3). 55 (0.69%) of the patients with no opioid use the year before fracture continued purchasing opioids in the third follow-up period 10-12 months after fracture.
In total, 13.3% of patients with prior mental illnesses (other than addiction) had prolonged opioid use and among those with prior pain 15.5% had prolonged use.
Physical-or occupational therapy treatment was associated with lower opioid use, 6.4% were classified as prolonged users as compared to 7.8% among patients without physical-or occupational therapy, p < 0.01 (Tables 2 and  3). Patients undergoing surgery had higher prevalence of prolonged use of opioids as compared to those not undergoing surgery (8.3% vs. 6.7% in follow-up period 1), but in the third follow-up period, the proportion of prolonged use was lower in those who underwent surgery (3.8% vs. 4.2%) (Tables 2 and 3).

| Opioid-naïve patients
Among the opioid-naïve patients, 199 (2.49%) purchased opioids both in the treatment period and in the first follow-up period (months 4-6 after fracture) and were thus classified as prolonged users. Moreover, 1.03% and 0.69% continued purchasing opioids in the second and third follow-up period, respectively. In total, 3.6% of patients with prior mental illnesses (other than addiction) had prolonged opioid use compared to 2.2% without prior mental illness, and the proportion continues to be higher in the following periods.
Opioid-naïve patients with previous pain symptoms had a higher proportion who purchased opioids in the months 4-6, 3.5% compared to those without previous pain, 2.4% but the difference is lower in subsequent periods and the Cochrane Armitage test for trend was not significant.
Physical-or occupational therapy treatment was associated with slightly lower opioid use, 2.4% were classified T A B L E 2 Number and percentage with prolonged opioid use in each follow-up period by exposure. as prolonged users as compared to 2.6% among patients without physical-or occupational therapy, and the proportions remain lower in subsequent periods compared those who did not attend. p < 0.01 (Tables 2 and 3). Patients undergoing surgery had higher prevalence of prolonged use of opioids as compared to those not undergoing surgery (3.8% vs. 2.1%) in follow-up period 1 and remained higher than those who did not have surgery in the two following periods.

| Effect of different exposures on prolonged opioid use
The effect of any opioid use last year on the adjusted relative risk of prolonged opioid use was 10.8 (9.12-12.8) as compared to those unexposed. Prior mental illness was significantly associated with prolonged opioid use (RR = 1.29 [95% CI: 1.14- In the subgroup analysis where we studied only the opioid-naïve patients (no opioids in the year before fracture), prior mental illness gave a risk ratio of 1.38 (95% CI: 1.02-1.87). Prior pain was not significantly associated with prolonged use (RR = 1.29 95% CI: 0.81-2.06). Occupational or physical therapy was not significantly associated with lower prolonged use (RR = 0.83 95% CI: 0.63-1.10). Surgery was associated with an increased risk of prolonged use 2.20 (95% CI: 1.63-2.97).
In the sensitivity analysis where we subdivided any prior use the year before fracture, regular opioid use the year before fracture rendered a risk ratio of 24.1 (95% CI 20.4-28.6) while some, but not regular, use was associated with an RR of 4.93 (95% CI 3.97-6.13) both compared to those without any use in the last year (Table 5).
Regular use of opioids before the fracture was associated with prolonged use after the fracture; regular use 2 years before fracture (but not the last year) had a risk ratio of 9.88 (95% CI 7.33-13.3), the risk diminishes gradually for regular use in earlier years but remains significant with previous use 5 years before giving a risk ratio was 3.09 (95% CI 1.33-7.18) compared to those who had no regular use in last 5 years before fracture (Table 6)

| DISCUSSION
We find that opioid treatment related to distal radius fracture is a risk factor for prolonged opioid use. A history of opioid use increased the risk for prolonged opioid use T A B L E 3 Number and percentage in the opioid-naïve study population (no opioids the year before fracture) with prolonged opioid use in each follow-up period by exposure. further and we show that a regular use as far back as 5 years before the fracture (i.e. 4 years without regular use of opioids) significantly increases the risk for prolonged use. A more recent use, even if occasional, the year prior to the fracture was also associated with increased risks as was prior mental illness. However, being treated by an occupational therapist or physiotherapist was associated with lower risk. Prior pain was not a statistically significant risk factor when adjusting for confounding factors. Patients treated with surgery had an increased risk of prolonged use in month 4-6 but the proportion of surgery patients with prolonged use in months 10-12 after surgery was lower compared to those without surgery. In the subgroup analysis restricted to opioid-naïve patients the proportions with prolonged use over multiple follow-up periods were overall smaller although the proportion of prolonged opioid use remained higher trough all periods of follow-up compared to the group without surgery. Overall, most patients did not fill any prescriptions for opioids during the treatment period, indicating that for most the pain is manageable without opioids. The Swedish national guidelines for treatment of distal radius fracture do not mention pharmacological treatment of pain ( patients who were ≥18 years old 5 years before fracture were included in the analysis n = 8704. b Adjusted for sex, age group and education. T A B L E 4 Effect of the exposures on prolonged opioid use (having purchased opioid during month 4-6 post fracture). RR (95% CI).

Risk ratios
Model Model 2: adjusted for sex age group and education level 2010.
Model 3 and opioid naïve. a Additionally adjusted for psychiatric diagnosis. addiction and consultation for pain in the year before distal radius fracture. b Additionally adjusted for diagnosis of addiction, consultation for pain in the year before distal radius fracture and latest regular use.
c Additionally adjusted for psychiatric diagnosis, diagnosis of addiction in the year before distal radius fracture and latest regular use. d Additionally adjusted for surgery for distal radius fracture and consultation for pain in the year before distal radius fracture.
sparing and multimodal pain management strategies should be considered (American Academy of Orthopaedic Surgeons., 2020). A Swedish enquiry showed that 10/15 centers prescribed opioids after distal radius fracture, most commonly short-and long-acting oxycodone (Heilig & Tägil, 2018). The lack of guidelines however allows for differences in prescription practices between physicians. The follow-up of opioid treatment is also left to individual physicians or clinics and the patient's expectations. Our finding that opioid use in the previous year was associated with an increased risk for prolonged opioid use has been shown before in patients who had surgery for distal radius fracture, for trauma patients and patients undergoing elective hand surgery among others (Johnson et al., 2016;Qin et al., 2021;von Oelreich et al., 2020). In our study, we additionally show the risk was increased even though there had been several years without regular use in between, i.e., a regular use that has ended still act as a potential risk factor for later prolonged use. Those with regular opioid use in the year before fracture might be continuing a previous consumption after fracture and would then not truly be at risk for developing prolonged use. They are, however, presented here for transparency and for their clinical relevance. The interesting results are that even those with no regular use in the last year before fracture have an increased risk for prolonged use. There are several mechanisms that could lead to the risk increase after previously terminated opioid use. Opioids are addictive and repeated exposure can increase the risk of addiction. Patients who have had a good pain-relieving effect of opioids before might be more prone to ask for opioids when in pain after fracture. Doctors' willingness to prescribe opioids to those already exposed might be higher compared to introducing opioids to opioid-naïve patients, either because the patient has proven they can quit before or because the perceived difficulties in discontinuing prescriptions (Ekelin & Hansson, 2018). Non-medical use of prescription opioids (where they are used as self-treatment of a perceived medical condition using medication without a prescription or used in a manner not intended by the prescriber, such as use of medication to achieve euphoric states) has an estimated one-year prevalence of 3.8% in Sweden and having been prescribed opioids have been shown to largely increase the risk for nonmedical use (Novak et al., 2016). Previous regular use can be a sign of (most often undiagnosed) opioid dependency and the prolonged use constitute a relapse in iatrogenic addiction. Those patients with ongoing non-medical use before fracture might also have an 'opportunity' to extend their prescription opioid use, in part by having a new condition to treat (the fracture) and in part through contact with a new prescribing clinic (orthopaedics, ER).
Among the patients with no opioid use in the last year before fracture 55 (0.69%) had a prolonged use in months 10-12 after fracture. Although not a large number these cases are important since they indicate a probable iatrogenic dependency, where individuals who the year before fracture did not use any prescription opioids now have at least 10 months of continuous use, indicating that the health care has caused harm.
The relation between prior mental illness and the risk for prolonged opioid use has been shown in previous studies which have reported increased risk for both prolonged opioid use after surgery and risk of opioid addiction (Kent et al., 2019;Klimas et al., 2019). A high prevalence of mental illness has also been found among patients with nonmedical use of opioids (Novak et al., 2016) and opioid use disorder (Jones & McCance-Katz, 2019), while prescription opioid use was common in patients with serious mental illness (Spivak et al., 2018) where a potential mechanism might be through the anxiolytic effect of opioids.
To avoid causing harm in the form of prolonged use, it seems reasonable for clinicians to ask not only for ongoing opioid treatment but also for self-medication and prior use as well as history of mental illness before considering a new prescription and to plan for an adequate follow-up and discontinuation of opioid treatment.
Rehabilitation, e.g., information and instructions on therapeutic exercise, provided by physical and/or occupational therapist is already suggested in the local clinical guidelines for all patients with distal radius fracture and seems to have a protective effect against prolonged opioid use according to our study, however, only 50% of the patients do attend. To motivate patients to partake in the recommended rehabilitation and have a follow-up routine in place could possibly decrease the number of patients with prolonged opioid use.
Surgery increased the risk for prolonged use, but only in the short term, since the proportion of patients undergoing surgery with prolonged use in months 10-12 was lower than those without surgery. Since the indication for surgery depends on the degree of dislocation of the fracture after closed reduction it might be that the patients requiring surgery have more painful injuries to begin with, post-surgical pain might also lead to prescription of opioids in the treatment period, a pre-requisite for prolonged use, and there is indeed a large difference in proportion of filled prescriptions in the treatment period (77.02% for surgery patients vs. 23.2%). It is interesting to note that this difference diminishes greatly in the first follow-up period. The lower proportion of prolonged use 10-12 months after fracture might depend on a more structured follow-up after surgery, for instance over 70% of patients attended occupational-or physical therapy compared to 45% of those who did not have surgery, and guidelines include a post-surgical follow-up visit to the orthopaedic outpatient clinic. In the opioid-naïve group, surgery increased the risk for prolonged use and the proportions of prolonged use remained slightly higher trough all periods of follow-up, (0.9% vs. 0.6% in months 10-12). In opioid-naïve patients, exposure to opioids is a prerequisite for prolonged use and the proportion using opioids in the treatment period is several times higher after surgery (75.9% compared to 18.1%). A recent systematic review found that opioid prescribing after elective minor and moderate surgical procedures did not reduce self-reported pain compared to opioid-free analgesia but was associated with increased risk of vomiting and other adverse events (Fiore et al., 2022). Limiting unhelpful opioid prescription could be a way to lower prolonged use.

| Strengths and limitations
A strength of this study is the large population under study, that also had access to affordable publicly funded health care. Moreover, all health care data were registered into a single register where we can identify virtually all relevant cases of incident distal radius fracture during the study period and follow them across health care providers. Another important strength is the ability to follow patients over time, both after fracture but also before, to identify risk factors, and in the case of latest regular use, study the effect of exposure as far back as 5 years before fracture.
The registers also give rise to limitations, such as the lack of possibility of follow-up questions to patients; we did not have access to patients charts for more detailed information on, e.g. pain intensity. This lack of detail in individual cases is a trade-off against the possibility of instead including a much larger population. Regarding the data on opioids, we can only be certain that the patient made the purchase, but not that the opioids were consumed. However, using data on actual purchases, as opposed to only prescriptions, is one step closer to the real consumption of opioids. There is of course also a risk that some individuals have an undiagnosed opioid dependency before fracture maintained with illegal opioids which we cannot detect, giving risk for some misclassification in for instance latest regular use, this is however as close to the actual use we can come with the available data.

| CONCLUSIONS/ IMPLICATIONS
Distal radius fracture can be a gateway to prolonged use of opioids, especially among patients with previous history of opioid use or mental illness. To avoid iatrogenic opioid dependency, a more structured follow-up and planning of opioid treatment, including questioning the use of opioid versus non-opioid pain relief might be beneficial.
Rehabilitation seems to have a protective effect; however, only 50% of the patients attend. Thus, to motivate and facilitate more patients to participate in rehabilitation, while also informing about the risks and lack of beneficial effects of prolonged opioid use, could be a first step towards decreasing prolonged opioid use in ours and similar health care systems.

AUTHOR CONTRIBUTIONS
All authors have contributed as: FLP, EB, IFP MECS and AJ participated in study design, interpretation and discussion of results and commented the manuscript, FLP and AJ acquisition of data. FLP analysed, drafted and revised the manuscript. AJ (guarantor). All authors approved the final manuscript.

ACKNO WLE DGE MENTS
The study was financed by grants from the Swedish state under the agreement between the Swedish government and the county councils, the ALF-agreement (grant number: 0238).

FUNDING INFORMATION
This study received funding from governmental 'Avtal om. Läkarutbildning och Forskning' grants to university health care in. Region Skåne, Sweden (grant number: 0238), regional PhD project grants from Southern Healthcare Region, Sweden.