Introduction

Back pain is one of the most common health problems; an estimate shows that up to 95% of all people will experience back pain at some point in their lifetime [1, 2]. It is the most common cause of chronic pain and is a leading cause of activity limitation and work disability worldwide [3]. Back pain has an enormous impact on individuals, healthcare systems, and national economies, and treatment approaches have important consequences for patients, clinicians, and society [4, 5]. According to the national clinical guidelines, the first choice of treatment for patients with back pain is 'stay active' and 'stay at work'. As the second-line treatment, manual therapy and the use of medication for short time periods are suggested, and the use of opioids is specifically recommended as a treatment for the few having specific needs [6], but only a few studies report on long time use of prescribed opioids [7, 8].

During the last decades, opioid therapy has become one of the standard pharmaceutical approaches to manage chronic back pain, despite multiple side effects associated with opioids [9]. Long-term opioid use induces a substantial risk of misuse [10]. Studies of patients prescribed opioids on a long-term basis indicate that as many as 45% could be engaging in aberrant drug-taking behaviors [11]. Observational studies have shown that treatment with long-term opioid therapy is associated with poor pain outcomes, functional impairment, and lower return to work rates [12]. Other studies have demonstrated that long-term opioid use is strongly associated with opioid misuse, accidental opioid overdose, addiction, polydrug abuse, and a cluster effect of addiction behaviors [13]. These concerns and restricted efficacy, have resulted in clinical guidelines recommending that opioid therapy should be used only in carefully selected patients, for a short duration, and with appropriate monitoring [5, 14].

Surgical treatment usually addresses patients having severe radiating pain and specifically patients not responding to protocolled non-surgical treatment. Naturally, surgery is associated with a high risk of short-term postoperative opioid use, but the patients are expected to discontinue opioid use after surgery [15]. However, studies on long-term opioid use after surgery are often limited by reliance on the patients' self-report or small sample sizes [16, 17].

The Danish healthcare sector is tax-financed and provides free coverage for all Danish residents. Everyone who lives in Denmark has the same access to any treatment, which usually follows national clinical guidelines. The primary entry point to the Danish healthcare system is the General Practitioners who act as gatekeepers to specialized services and who are responsible for medical prescriptions. This gives us a unique opportunity to investigate the association between surgical treatment for a back disorder and long-term use of opioids.

The main objective of this register-based study is to evaluate whether the risk of long-term opioid use in patients with degenerative back disorder is reduced in patients undergoing surgery. The hypothesis is that for patients with degenerative back disorder, the risk of using opioids one and two years after diagnosis is not associated with having a surgical treatment.

Methods

Study population

In Denmark, all citizens are assigned a unique civil registration number, which allows the linkage of individuals between the Danish Central Data Registries (CDR). We define the study population using the Danish National Patient Register (DNPR), which contains information on all patients' hospital contacts since 1970 [18]. From the DNPR we identified all adult patients diagnosed with degenerative back disorders (DBD) such as a herniated disc, spinal stenosis, and radiculopathy (ICD10 codes: M43*, M47*, M480, M500-2, M510-2, M96*, M990-7) at the Spine Centre of Southern Denmark (SCSD) in the period from 1 January 2013 until 31 December 2017. To evaluate the long-term use of opioids we followed all patients in the study population for two years after the diagnosis date. The diagnosis date was identified as the date of the first contact with the Spine Centre without a previous contact for 12 months [19]. Patients who died or emigrated within two years after the diagnosis date were excluded. Information about the patients' living and emigration status was obtained from the CDR. Moreover, to avoid the influence of postoperative opioid use on the two-year follow-up, all patients who had undergone surgery more than six months after the diagnosis date were also excluded from the study.

Definition of variables

Surgery was defined as 'yes' if the patient had spinal surgical treatment within six months after the first contact at the SCSD and otherwise as 'no'. Spinal surgical procedures were defined using surgical procedure codes for decompression (KABC1*, KABC(20*-26*), KABC(3*-5*)), and fusion surgery codes (KNAG(0*,3*,4*,6*,7*), KNAK4).

The use of opioids was defined on an individual level using the Danish Prescription Register [20]. The registry contains information on all prescribed drugs that have been dispensed in Denmark since 1995. We identified opioids by the ATC code—N02A. In previous studies on prescribed drugs, opioid users were defined as individuals who have used at least one prescription per month [21]. To avoid bias due to seasonal variation of redemption of prescribed drugs we defined the patients' use of opioids in three months intervals. The primary outcome of the study was the use of opioids two years after the diagnosis date. The variable was defined as 'yes' if the patients have redeemed at least one prescription of opioids, within a period from 25 to 27 months after diagnosis date and otherwise as 'no'. The secondary outcome – use of opioids one year after the diagnosis date was defined similarly: It was defined as 'yes' if the patients have redeemed at least one prescription of opioids, within a period from 13 to 15 months after the diagnosis date and otherwise as 'no'.

Some patients used opioids before the diagnosis date. This could be due to different levels of disease or different severity levels of pain or even due to some non-DBD related diseases. To take this into account we calculated the use of opioids within three months before the diagnosis date for each patient. Visualizing of the prior description can be seen in Fig. 1.

Fig. 1
figure 1

Schedule for monitoring the opioid treatment strategy

Full size image

Confounders: Age was defined as 0—'young' if the patient's age at the start of the course of the disease was below or equal to 60 and 1—'old' if the age was above 60 years. This cut point was used since in Denmark early retirement pension is possible from the age of 60 for all citizens. Sex was defined as 0—'male' and 1—'female'. The patient’s socio-economic status was calculated using the information about personal equivalent disposable income in the year of the DBD which was retreated from [22].The socio-economic status was defined as 0—'low' if the patient's income was lower than the median of 213,993 DKK, and 1- high if the income was equal to or higher than the median. Information about the patient’s highest completed education was retreated from the [22]. Education was defined in three categories: (i) 0—low (municipal primary and lower secondary school), (ii) 1—medium (upper secondary school), and (iii) 2—high (higher education including bachelor, masters, and doctoral levels). Variables of adults in the home and children in the home were constructed using the household information available in [23]. Variables were defined as 0—`yes’ if another adult/child was registered at the same address or 1- otherwise as `no`. Information about the patient's working status in the year of the diagnosis was defined as the patient’s social group (i) 0—employed, (ii) 1—unemployed, (iii) 2—senior citizen. Using the DNPR records we calculated the patient’s somatic comorbidity (CCI) according to the Charlson comorbidity index [24]. CCI was classified as 0 for CCI equal to zero, 1 for CCI equal to 1 and otherwise 2.

Statistical analyses

We used an approximation of the Fisher exact test to explore the association between previously described confounders and outcomes. Since the patient's use of opioids before hospital contact was positively associated with both outcomes, all analyses were performed stratified according to the status of opioid use before the diagnosis date. We performed descriptive analyses showing the distribution of patients for each confounder in numbers and percentages. We conducted logistic regression analyses adjusting for all previously described confounders to estimate the Odds Ratio for opioid use one and two years after the diagnosis date.

All previously described analyses were also performed as sub-analyses to compare use of opioids for patients who underwent surgery less than three months after the diagnosis date with the use of opioids for non-surgically treated patients.

Additionally, we performed analyses comparing the use of opioids after surgery for patients who underwent surgery more than six months after the diagnosis date and the use of opioids after surgery for patients who underwent surgery less than six months after the diagnosis date.

All analyses were conducted using Stata 16 (Stata Corp, College Station, Texas).

Results

We extracted a total of 16,056 patients diagnosed with a DBD from the DNPR. Of those 649 were excluded because they had more than one back disorder course during the study period. The study population was reduced to 15,231 patients due to the exclusion of patients who died or emigrated during the first two years after the diagnosis date with DBD. Among 140 patients who died within the two years, 39 had undergone DBD surgery and four of those died within 30 days after surgery. Further 769 patients were excluded because they had undergone surgery more than six months after the DBD. A study population flowchart is presented in Fig. 2. The total study population comprises 14,462 patients, 3,518 of who received surgical treatment. The median duration from the diagnosis date to surgery was 53 days (25% of the patients' duration was less than 22 days and 75% of the patients' duration was less than 92 days).

Fig. 2
figure 2

Selection of the study population among patients seen at a specialized spine center

Full size image

The histogram shown in Fig. 3 illustrates the use of opioids. The histogram indicates that for both groups of patients, the proportion of patients who had used opioids declined with time after the diagnosis date.

Fig. 3
figure 3

Quarterly use of opioid in 14,462 patients seen at a specialized spine center stratified by surgical status and by use of opioids before the first BPD hospital contact

Full size image

Patients who did not use opioids before

The baseline characteristics of the study population are shown in Table 1. Several of the surgically treated patients were older than 60 years (50% were aged > 60 years in the surgical group vs 38% in the non-surgical group). Consequently, fewer of the surgically treated patients have children living at home (27% in the surgical group vs 35% in the non-surgical group), and several were senior citizens (45% in the surgical group vs 34% in the non-surgical group). The results shown in Table 2 illustrate that there is no difference in the proportion of patients who used long-term opioids after the diagnosis date. Only 11% of the surgically treated patients and 9% of the non-surgically treated patients used opioids one year after and for both groups of non-surgically and surgically treated patients only 9.5% of patients used opioids two years after the diagnosis date. The results of the adjusted logistic regression analysis (see Table 3) show that the ratio of surgically treated patients who used opioids one year after the diagnosis date is higher than for non-surgically treated patients (OR = 1.21, 95%CI (1.01; 1.45)), whereas the ratio of surgically treated patients who used opioids two years after the diagnosis date does not differ from the ratio of non-surgically treated patients (OR = 1.01, 95%CI (0.84; 1.22)).

Table 1 Baseline characteristics in 14,462 patients with degenerative back pain disorders seen at a specialized spine center stratified by previous use of opioids and surgical status
Full size table
Table 2 Long-term use of opioids in 14,462 patients seen at a specialized spine center stratified by surgical treatment less than six months after the first visit
Full size table
Table 3 Estimated effect of surgery on the use of opioids in 14,462 patients seen at a specialized spine center. Results from the logistic regression analyses, OR for use of opioids versus not use
Full size table

Patients who used opioids before

There are no significant differences in the baseline characteristics between surgically and non-surgically treated patients in that group of patients (see Table 1). The proportion of long-term opioid usage is lower for surgically treated patients compared to the non-surgically treated one and two years after the first visit (31.4% vs. 36.6%; 28.4% vs. 34.2%). The results of the adjusted logistic regression analysis point in the same direction; the ratio for surgically treated patients using opioids one year after the diagnosis date is significantly lower than for non-surgically treated patients (OR = 0.78, 95%CI (0.68; 0.85)). The same applies to the use of opioids two years after the diagnosis date (OR = 0.75, 95%CI (0.66; 0.86)).

The pattern of opioid use after surgery for patients who underwent surgery more than six months after the diagnosis date was very similar to the pattern of the use of opioids after surgery for patients who underwent surgery less than six months after the diagnosis date (results are not shown).

Discussion

The main objective of this study is to investigate whether surgically treated patients with DBD have the same risk of being long-term opioid users as DBD patients who did not have surgery. This study based exclusively on register data broadly elucidates surgery and provides a scientific basis for further research on the use of opioids in patients with degenerative back disorder.

Patients who did not use opioids before

We found that almost 10% of the patients diagnosed with degenerative back pain disorders at the Spine Center of Southern Denmark, who did not use opioids before DBD, ended up being long-term opioid users. Moreover, we found that surgical treatment demonstrates no advantage over non-surgical treatment in the group of patients who did not use opioids one year before surgery regarding the risk of being long-term opioid users.

Patients who did use opioids before

Furthermore, we found that at least 32% of the patients who used opioids before the first hospital contact with DBD continued opioid use two years after. These findings are consistent with other studies, indicating that more patients with previous opioid use became long-term users compared to patients who did not use opioids before. In this group of patients, surgical treatment was statistically associated with less long-term opioid use. Long-term opioid use is common both for surgically and non-surgically treated patients [25].

Strengths and limitations

The strengths of the study are that it is based on a large register-based data set with detailed medication information, and the potential effect of recall- and information bias is reduced by the use of register-based data.

This study also has several limitations. Firstly, as the present study only analyzed register-based data on dispensed medicine, it is not possible to assess compliance and adherence. Secondly, the definition of long-term opioid use is not standardized, although our results are consistent with studies where other definitions are used. Thirdly, there was an imbalance in the observation time between surgically treated and non-surgically treated patients, although we tried as far as possible to avoid it by excluding patients who had undergone surgery more than six months after the DBD. We do not think that this imbalance has a major impact on our results, but it needs to be examined more closely. As weakness of our study the lack of finer details could also be mentioned, such as primary indication for surgery, type of surgery, and more specific diagnosis. Moreover, we do not know whether the long-term use of opioids is due to addiction or poor pain relief. In the current study, we researched the effect of any kind of DBD surgery on long-term opioid use. In the following studies, we will explore the separate effect of fusion and no fusion surgery, and research whether DBD surgery influences the dosage reduction of opioid use.

Our results indicate that surgical treatment does not always result in discontinuation of long-term opioid use. One explanation could be that many patients do not achieve the expected pain relief from surgery. However, the fact that previous opioid use is a major predictor of long-term use suggests that opioid dependency may be equally or more important, making discontinuation more difficult. Patients should be informed that discontinuation of opioid use could be challenging and the clinics need to be aware that it is important to focus on phasing out opioids, possibly in collaboration with the GP.

Conclusions

This study based on the Danish registers provides information on the association between surgery and long-term use of opioids. We found that patients with degenerative back disorder who used opioids before their first visit at a specialized spine center are more often non-opioid users one year later, if they were surgically treated, compared to patients who did not have surgical treatment. Whereas previous non-opioid users have the same risk of being opioid users after surgical treatment as patients who did not have surgery. The study provides a scientific basis for further research on the use of opioids in patients with degenerative back pain.