The association between sociodemographic characteristics and the event of undergoing first-time, simple lumbar discectomy: A case-control study.

Background: In disc herniation, nonsurgical treatments are recommended prior to elective discectomy but are often associated with consultation fees, whereas the discectomy itself may be without user payment. This may affect individual preferences in the choice of treatment. This retrospective case-control study examined the association between socioeconomic characteristics and the event of undergoing a first-time, single-level, simple lumbar discectomy.
Methods: The consecutively formed study population comprised patients undergoing elective lumbar discectomy at a Danish public hospital between 2010 and 2013. A national authority identified three gender- and age-matched controls per case for comparison. Measures investigated in this study were marital status, ethnicity, socioeconomic classification, educational level, the extent of sick leave 52 weeks prior to surgery, personal income, and equivalized disposable household income. All measures were provided by national registries. The associations were examined using uni- and multivariate logistic regression analysis.
Results: In a study population of 888 operated patients (age (SD) 46 (14); ODI (SD) 47 (18); leg pain intensity (VAS) (iqr) 74 (33), EQ-5D (iqr) 0.26 (0.62)) compared to 2664 controls, the probability of undergoing lumbar discectomy was significantly associated with lower vs. higher educational levels ((OR 1.98-2.53), and with periods of sick leave exceeding two weeks within one year prior to surgery (OR 9.47 (95% CI 7.68-11.68)). In the multivariate analysis, the event of undergoing discectomy was insignificantly associated with any other socioeconomic characteristics, whereas the personal income was of significant importance in the univariate analysis.
Conclusion: The event of undergoing free-of-fee elective first-time, single-level, simple lumbar discectomy is more common among individuals with low educational levels and unstable labor market attachment when examined in a case-control study. Being a multifactorial challenge, this calls upon the active engagement of several policy sectors.

Introduction

Among patients with low back pain in primary spine care, radiculopathy is present in 11-12% [1,2], with disc herniation being the most frequent cause [3].

When absolute indications for spine surgery are absent, the first line of treatment for symptomatic lumbar disc herniation (LDH) is nonsurgical. Low-level evidence-recommended nonsurgical treatment includes, among others, adaptation to or normalization of physical activity, supervised exercise therapy, manual joint mobilization, and education about pain mechanisms [4,5]. Nonsurgical prognosis is considered good, and pain relief and improved physical function should be expected within one to two months after symptom debut.

However, if pain or activity-compromising disability persists or worsens despite relevant nonsurgical treatment, clinical guidelines suggest referral for surgical assessment within 12 weeks of symptom debut [4,5]. In such cases, elective discectomy can be offered if the diagnosis is confirmed by magnetic resonance imaging (MRI) or computed tomography (CT).

According to Bush et al. 1992, around 14% of UK patients assessed with symptomatic LDH in primary care undergo discectomy [6]. However, the proportion of individuals with LDH undergoing discectomy could be somewhat higher today because the use of some spine surgery procedures has increased. This is without being explained by higher frequencies of spinal disorders [7]. Increased use of private or employment-based private health insurance or changed patient demands seems to be a contributing mechanism behind changed patterns in health care utilization of specialist spinal care, including surgery [8]. The use of private health insurance does not seem to affect the overall use of health care services but is associated with an increased probability of hospital contacts. Thus promoting unequal access to specialized healthcare [8]. Without insurance coverage, user fees or co-payments for services within the healthcare systems may pose a barrier to recommended medical treatments [9]. The scope of this study is however not to examine the consequences of user fees or co-payments.

In a Danish context, health care services provided by licensed practitioners are generally reimbursed in full or in part by the health authorities. Primary care services other than the services of a general practitioner e.g. the conservative management of spinal disorders are associated with some degree of user fee or co-payments, typically 30-50% of the total expense. Specialized services like spinal surgery are fully reimbursed and thus at no direct cost to the patient. Conversely, hospital services are only accessible upon referral, whereas most primary care services are direct access.

Compensation policies during spine-related sick leave are likewise important for the actions of the individual, such as sickness absence or –presence [10,11]. In Denmark, individuals on sick leave are legally guaranteed sickness benefits. During the study period, the maximum length of sickness benefits allowed by the Ministry of Employment reached 156 weeks [12]. In the sick leave period, the municipality makes person-oriented efforts for a quick return to the labor market to avoid long-term absence and unemployment.

Inequality in health is widely identified and associated with socioeconomic status (SES) that is typically defined by income, educational level, insurance status, race, and labor market attachment [13,14]. The inequalities are evident in both higher rates of morbidity, including chronic diseases and mortality [13]. In previous settings, individuals of low SES more frequently use prolonged free-of-fee hospitalizations, whereas individuals of higher SES to a greater extent use preventive and specialized services with possible user fees or co-payments [9,15,16]. The increased tendency towards the use of low-value treatments among individuals with low SES is an expression of the inequality in health [16,17].

Within the area of musculoskeletal disorders, the societal and health-related consequences of low SES recognized in the literature are extensive (Figure 1), which will complicate clinical evaluation and treatment of LDH, and in general. Moreover, low SES itself is a risk factor for spinal degenerative diseases [16] confounded by factors such as being a smoker or having diabetes [18,19]. These findings could potentially induce inequality in the treatment of LDH.

Figure 1

Illustration of societal and health-related consequences to a low socio-economic status (SES), defined by educational level, income, and labor market attachment when compared to individuals with higher SES

This case-control study, based on secondary national administrative data, aims to explore whether sociodemographic and -economic characteristics are associated with the likelihood of undergoing first-time, single-level, simple lumbar discectomy at a free-of-fee, public hospital.

Material and methods

Study design

A retrospective, case-control study based on secondary data from national administrative registers. The study adheres to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [20].

Specification of key study dates and settings

The study population was identified as part of a previous study [21] and comprised a consecutive series of patients who, according to a research database, underwent first-time, single-level, simple lumbar open discectomy or micro-endoscopic discectomy during the inclusion period between June 1, 2010, and December 31, 2013, at a public hospital.

All patients were referred from primary or secondary care after insufficient nonsurgical treatment or otherwise presenting with a specific indication for elective discectomy. The diagnosis was verified by magnetic resonance imaging or computed tomography. Patients were 18 years or older and had minor degrees of comorbidities (American Society of Anesthesiologists (ASA) Classification Score <3 [22]).

Once scheduled for surgery, the patients enrolled in the research database DaneSpine [23], which identified the study population (described below).

Exclusion criteria were previous spine surgery at any level or other spinal pathology at the time of surgery such as malignancy, infection, spondylolysis, spondylolisthesis, inflammatory arthritis, notable scoliosis, or metabolic bone disease.

Statistics Denmark, a government authority described below, identified the control population as a random sample of the Danish population, with three controls per surgical patient, matched by gender, age (month and year of birth), and the municipality of residence at the time of surgery [24]. The control population was not a clinical population, and having, by chance, undergone LDH surgery according to the Danish National Patient Register [25] disqualified for participation in the control cohort.

The identification of the controls was based on the Danish Civil Registration System, which stores current and historical personal information on all Danish residents based on a personal identification number [26].

Data sources

The Danish research database DaneSpine is maintained by the Danish Society of Spine Surgeons. It holds Patient-Reported Outcome Measures (PROMs) and clinical data prospectively collected pre-, peri- and postoperatively [23].

Within the study period, 98-99% of the patients who underwent lumbar discectomy at the public hospital enrolled in the database according to estimates based on data provided by the Danish society of Spine Surgeons and the National Patient Register.

Statistics Denmark is an independent governmental authority that collects and publishes statistics on the Danish society based on civil registration numbers and related data collected from several national authorities and registers. Of particular interest to this study are data on employment and social payment transfers at individual levels retrieved from the Danish Register Based Evaluation of Marginalization (DREAM) database [27,28]. The DREAM database is described in the Appendix.

Variables

The outcome of interest was the event of undergoing lumbar discectomy.

The following variables retrieved from the DaneSpine database described the surgical group: age, body mass index (BMI), gender, smoking habits, duration of leg and back pain, intensity of leg and back pain (Visual Analogue Scale (VAS)) [29], mental and physical health (Short Form Health Survey version 1 (SF-36)) [30], functional disability (Oswestry Disability Index (ODI)) [31,32] and health-related quality of life (EuroQoL (EQ-5D)) [33].

Sociodemographic variables of interest retrieved from Statistics Denmark included weekly data on employment within one year prior to surgery and data on ethnicity, marital status, highest educational level, annual personal and equivalent disposable family income, and socioeconomics. Functional disabilities related to the disc herniation at the time around surgery are likely to affect labor market attachment as evidenced by changes in income and socioeconomic classification. Thus, the values of these variables are included as recorded one year prior to surgery to describe placement in the labor market unaffected by the disease.

See the Appendix for a detailed description and definition of the socioeconomic variables and the initial data management.

Statistical analysis methods

Normally distributed variables were described using means and standard deviations, and non-parametric data were described using medians and interquartile ranges. Categorical data were expressed as proportions. Independent variables were tested for intercorrelation, and variables with moderate or strong correlation (R2 > .5) were excluded from further analysis.

The associations between the event undergoing lumbar discectomy and the independent variables were estimated using univariate logistic regression analysis. All associations expressed as Odds Ratio (OR).

Using stepwise backward elimination based on a Likelihood ratio test, all variables were further tested in a multivariate regression analysis. All estimates were reported with a 95% confidence interval (CI).

Valid CIs and p values were ensured using the Hosmer and Lemeshow rule of thumb; the number of events (operated patients) exceeded the suggested 10 per independent variable [34]. Bootstrapping (1000 reps) further computed bias-corrected CIs and p-values [35].

Few missing observations were expected due to the overall completeness of the datasets as described in the Data Sources section. Therefore no handling was planned except assessing and presenting the completeness of data per variable in the Results section.

Analyses were performed using Stata 16 (StataCorp, College Station, Tx, USA). An alpha level of .05 was used in all tests.

Ethics and Data Control

National agencies being the Danish Health Safety Authority (#3-3013-1174/1), and the Danish Data Protection Agency (#14/26345) approved the study. According to Danish law, ethical approval from the Regional Scientific Ethics Committee for Southern Denmark was not required [36]. When completing the DaneSpine questionnaires, the patients gave written informed consent for the use of their data in research. All data were stored following the Danish Open Administration Act, the Danish Act on Processing of Personal Data, and the Health Act.

Results

Participants

According to the research database DaneSpine, 1206 patients underwent first-time lumbar discectomy within the study period. However, a survey of the electronic medical records revealed inaccurate notation in the research database as 318 patients did not fulfil the inclusion criteria, leaving 888 patients and subsequently 2664 controls in the final study population (Figure 2). In particular, previous spine surgery or surgery at more levels disqualified from participation. Sociodemographic and clinical baseline characteristics describing the operated cohort are presented in Table 1.

Figure 2

Flowchart describing the selection of the study population

Table 1

Sociodemographic and clinical characteristics for patients undergoing first-time, single-level, simple lumbar discectomy (n=888).

Mean (SD)

Age, n=888an (%)Male, n=888Married/Cohabiting, n=888Smoker, n=888Ethnicity, n=887DanishImmigrant/descendantEducational level, n=871Primary and lower secondary schoolHigher general and preparatory examination programsShort- and medium cycle higher educationLong-cycle higher educationPre-operative sick leave, n=888≤2 weeks>2 weeksSocioeconomic classificationb, n=888EmployerEmployeeReceiving social benefitsEquivalent disposable family income bc, n=7961st quintile2nd quintile3rd quintile4th quintile5th quintilePersonal income bd, n=8701st quintile2nd quintile3rd quintile4th quintile5th quintileDuration of back pain, n=888No painPain <3 monthsPain 3 months till <12 monthsPain ≥ 1 yearDuration of leg pain, n=888No painPain <3 monthsPain 3 months till <12 monthsPain ≥ 1 yearMean (SD)Body Mass Index, n=882Mental health (SF-36 MCS), n=882Physical health (SF-36 PCS), n=882Disability (ODI), n=876Median (iqr)Quality of life (EQ-5D), n=884Back pain intensity (VAS), n=880Leg pain intensity (VAS), n=881

46 (13.70)477 (54)491 (55)295 (33)827 (93)60 (7)265 (30)419 (48)167 (19)20 (2)511 (58)377 (43)38 (4)581 (65)269 (30)258 (32)184 (23)153 (19)124 (16)77 (10)311 (36)193 (22)156 (18)127 (15)83 (10)69 (8)144 (16)372 (42)303 (34)10 (1)240 (27)473 (53)165 (19)27 (5)27.75 (7.34)40.55 (11.94)47.29 (18.24)0.26 (0.62)50 (50.5)74 (33)

SD, standard deviation; iqr, Interquintile range; SF-36 MCS, The Short Form Health Survey version 1, mental health score; SF-36 PCS, The Short Form Health Survey version 1, physical health score; EQ-5D, European Quality of life – 5 Dimensions.

age range: 18-87 years

as assessed one year prior to surgery

Income range in Danish kroners: 0-2.599.181 DKK

Income range in Danish kroners: 0-1.250.712 DKK

The association between sociodemographic characteristics and undergoing first-time, single-level, simple lumbar discectomy

Overall, very few data were missing, less than 10% depending on the variables assessed (Table 1 and 2), and they were considered of no significance to the results [37].

Table 2

Associations between patient socioeconomics and the probability of undergoing first-time, single-level, simple lumbar discectomy (n=888) compared to the control cohort (n=2664), N=3552. Associations presented as Odds Ratios with the 95% confidence intervals.

	Univariate analysis	Multivariate analysis
	OR (95% CI)	OR	95% CI
Marital status (vs. Married/Cohabiting), N=3443
Single/Non-cohabiting	1.06 (0.91-1.23)	1.10	(0.93-1.31) N(0.94-1.31) BC
Ethnicity (vs. Danish), N=3542
Immigrant/Descendant	0.80 (0.60-1.08)	0.72	(0.48-1.10) N(0.47-1.08) BC
Highest completed educational level (vs. Long-cycle higher education), N=3431
Primary and lower secondary school	3.06 (1.88-4.98)	2.53	(1.46-4.39) N(1.52-4.47) BC
Higher general and preparatory examination programs	2.70 (1.67-4.35)	2.01	(1.17-3.47) N(1.26-3.59) BC
Short- and medium cycle higher education	2.44 (1.49-4.01)	1.98	(1.14-3.47) N(1.20-3.63) BC
Socioeconomic classificationc (vs. Employer), N=3523
Employee	1.13 (0.78-1.64)	0.92	(0.58-1.46) N(0.60-1.50) BC
Receiving social benefits	1.28 (0.87-1.88)	1.47	(0.91-2.37) N(0.91-2.42) BC
Sick leave (vs. ≤2 weeks), N=3543
>2 weeks	8.68 (7.15-10.54)	9.47	(7.68-11.68) N(7.67-11.58) BC
Equivalent disposable family income a (vs. 5th quintile), N=3190
1st quintile	1.25 (0.96-1.62)
2nd quintile	1.37 (1.06-1.77)
3rd quintile	1.25 (0.96-1.62)
4th quintile	1.07 (0.82-1.40)
Personal incomeb (vs. 5th quintile), N=3435
1st quintile	1.34 (1.04-1.74)	1.12	(0.81-1.54) N(0.81-1.57) BC
2nd quintile	1.59 (1.23-2.04)	1.00	(0.73-1.37) N(0.74-1.39) BC
3rd quintile	1.55 (1.20-1.99)	1.00	(0.75-1.33) N(0.77-1.34) BC
4th quintile	1.46 (1.13-1.89)	1.14	(0.86-1.52) N(0.85-1.51) BC

OR, Odds Ratio; CI, Confidence Interval; N, Normal confidence interval; BC, Bias-corrected confidence interval

Bold indicates p values<0.05.

In Danish kroners as assessed one year prior to surgery, 1st quintile: <161.410, 2nd quintile: 161.410-200.570, 3rd quintile: 200.570-243.920, 4th quintile: 243.920-300.300, and 5th quintile: >300.300.

In Danish kroners as assessed one year prior to surgery, 1st quintile: <131.770, 2nd quintile: 131.770-184.400, 3rd quintile: 184.400-235.920, 4th quintile: 235.920-310.710, and 5th quintile: >310.710.

As assessed one year prior to surgery

Among categorical variables, small cell sizes (n≤5) were addressed using aggregation to secure statistical reliability. For further description of initial data handling, see the Appendix.

Testing for intercorrelation between independent variables led to no exclusions as no moderate or strong correlations were found (R2 < 0.5).

Results from the univariate and multivariate analysis are presented in Table 2. For the latter, both uncorrected and bias-corrected CIs are reported.

In univariate analysis, being on sick leave for more than two weeks within one year prior to the outcome event was associated with an 8.7 OR (95% CI 7.2 - 10.5) of undergoing discectomy. Also, educational level and personal income were statistically significantly associated with the outcome. As such, patients with lower levels of education and patients with a low income were 2-3 times and around 1.5 times, respectively, more likely to undergo discectomy than patients with the highest educational level and personal income. The lower the educational level, the higher OR compared to the highest level of education with OR=1.

The independent variables marital status, ethnicity, duration of sick leave, educational level, socioeconomic classification, and personal income constituted the final model in the stepwise backward multivariate analysis, which included all variables in the initial model. Sick leave for more than two weeks within one year prior to the outcome event and educational levels remained statistically significantly associated with the event of undergoing discectomy. This with ORs largely unchanged.

The resampling through bootstrapping did not indicate multiplicity. The size of CIs reflected the uneven variation of both the educational levels with less highly educated individuals and the duration of sick leave. Estimates of the statistically significant associations found should therefore be used cautiously.

Discussion

This study, based on secondary national administrative data, aimed at exploring whether sociodemographic and -economic characteristics were associated with the likelihood of undergoing first-time, single-level, simple lumbar discectomy at a free-of-fee, public hospital.

Compared to an age-, gender- and municipality-matched control having not undergone spine surgery, the event of undergoing lumbar discectomy (n=888) was statistically significantly associated with educational levels and the duration of preoperative sick leaves.

The lower the educational levels and the longer duration of the sick leaves, the higher probability of discectomy. This is in line with the results of Lurie et al., who found preferences towards discectomy compared to nonsurgical care in patients with lower levels of education and higher levels of unemployment [38].

Workability can be regarded as a proxy for the balance between functional capacity and physical work demands, and individuals with low educational levels and more physically demanding work are likely to be less able to adapt to functional disabilities, e.g. when afflicted by LDH [39,40]. This may negatively affect workability. A preference for care that immediately results in an ability to remain attached to the labor market could further be anticipated. Moreover, for individuals with a low education level, which is associated with lower income [41], user fees or co-payments for primary care health services are previously established as a significant barrier towards high-value health care [17].

In this study, we had no access to data on the degree of primary care health services used, and the presence of a low-value pattern should therefore not be presumed. Future studies should focus on retrieving data on both primary and secondary health care services to more fully describe the pattern of health care use among patients with LDH.

Low SES is associated with fewer psychosocial resources for coping with work conditions and medical issues [17,40]. Navigating treatments and labor market attachment while being in pain or disabled might be difficult and of possible importance to treatment preferences. ORs associated with the socioeconomic classification as presented in table 2 may illustrate these relationships. Within the area of lumbar spine surgery, the PREPARE study found that pre-surgical psychosocial interventions, such as a fear-avoidance-reducing approach, had a positive effect on self-efficacy [42]. Psychosocial resources are thus modifiable and of evident relevance to interventions targeted the non-medical conditions that may influence treatment preferences.

The magnitude of the association between income and the outcome event (discectomy or not) is modest, although significant, in the univariate analysis (OR 1.34-1.59), and without significance in the multivariate analysis. The ORs are of almost similar magnitudes regardless of the quintiles compared to the highest quintile. Denmark, like most other Nordic countries, has a low degree of income inequality compared to other European countries [43], but mechanisms, such as employer-paid health insurance, complicate the interpretation of the associations found. We suggest that in a Danish context, the educational level will be a better estimate of the SES than income, as the educational value is more stable over time [44], but this may be context-specific.

Lurie et al. found that the disease-related factors such as patient-reported pain and disability levels were important to the surgery preference in favor of nonsurgical treatment [38]. In clinical guidelines, the duration of symptoms, and by extension, the likelihood of extended sick leave, is included as indicators for considering surgical evaluation. As such, the positive association between extended sick leaves and undergoing discectomy found in this study was expected.

As described in the Appendix, the cut point of two weeks was estimated from a heavily right-skewed histogram. Due to limitations in the national administrative data (see the Appendix), only sick leaves exceeding the employer-paid period (3-4 weeks within the study period) are included in the study. This affects the cut point. Including sick leaves of shorter duration would affect the distribution within the variable and likely the cut point. Such conditions are particular to a Danish context, and extrapolation should not be uncritical. Durations of sick leaves are, however, strongly associated with SES [45] and should be accounted for and should be accounted for during clinical assessment.

Immigrants or descendants were found less likely to undergo lumbar discectomy. The finding reflects the clinical perception of immigrants/descendants being less willing to undergo discectomy in the surgical department participating in the study. This although surgery is offered which possibly illustrates cultural differences. This relationship seems of greater importance to the probability of undergoing discectomy, than does the negative labor market attachment otherwise associated with this population.

Strengths and limitations

A clear strength of the study was the use of comprehensive secondary data from national registries.

Since 1968, all individuals living in Denmark have been registered in the Danish Civil Registration System and assigned a unique personal identification number at birth or upon immigration to Denmark, allowing accurate cross-reference between all national registries and datasets [26]. This permits register studies that typically have high degrees of data completeness and minimal risks of selection bias. However, as a single-center study, the validity and generalizability of the results are limited and the findings should be replicated in other and extended populations, and wide confidence intervals found call upon larger study samples.

The exclusion of patients with an increased degree of comorbidity (ASA classification score ≥ 3) might have led to an underestimation of the association of SES and surgical preference because high comorbidity is associated with low SES [16], including an underrepresentation in the labor market [46]. However, a stable labor market attachment at the time around spine surgery predicts good outcomes after surgery [21,47] and may dilute the effect of excluding high-comorbidity patients by affecting the decision-making disfavouring surgery.

The increasing use of health insurance has offered many patients the possibility to circumvent waiting times for treatments at public hospitals. As a result, more discectomies are conducted at private hospitals, and on the shorter duration of symptoms [48]. The present study did not include the patients who underwent lumbar discectomy at private hospitals, therefore it might overestimate the association between SES and the outcome event, as the surgical group examined is likely less educated and attached to the labor market when compared to surgical patients at private hospitals [48]. Thus, an extrapolation of the present results should be adjusted for the framework of local health care systems. However, as the scope of this study was explorative towards whether sociodemographics and -economics was of importance to the probability of surgical treatment of symptomatic LDH, the free-access tax-financed health care system is suitable. But limits, however, the comparison of the present results with findings in private supplementary or employment-based private health insurance-driven health care systems.

The study population was identified for previous research, hence the aging data from 2010-2013. However, neither the basic structures within the health care system nor the guidelines for the treatment of lumbar disc herniation have changed significantly within the last decade.

Conclusion

This explorative case-control study found associations between socioeconomic factors and the probability of undergoing first-time, single-level, simple lumbar discectomy at a public free-of-fee hospital.

Compared to an age- and gender-matched non-clinical control population, the ORs of undergoing discectomy among 888 operated patients increased from 1.98 to 2.53 with decreasing educational levels compared to the highest educational level. Likewise, sick leaves exceeding two weeks within one year prior to surgery showed a statistically significant association with an increased probability of undergoing surgery (OR 9.47 (95% CI 7.68-11.68)).

Although the associations should be tested in larger samples and data on the use of non-surgical health care should be included, the findings indicate patterns of low values healthcare use among individuals holding low socioeconomic status. The treatment of LDH-related pain and disabilities should be recognized by authorities and health care providers for its complexity among patients with low SES.

Data availability

Access to data may be obtained by contacting the corresponding author.

Declaration of Competing Interest

None.