Risk factors for prolonged length of stay after first single-level lumbar microdiscectomy

Data of patients who underwent first single-level lumbar microdiscectomy between 2018 and 2022 were retrospectively reviewed. Inclusion criteria were a minimal age of 18 years and a symptomatic lumbar disc herniation confirmed radiographically by MRI. Patients with a history of previous lumbar spine surgery, as well as previous traumatic and recurrent lumbar disc herniations, were excluded from this study.

Demographic data, medical records, intraoperative course, and imaging studies were retrospectively analysed. Since the primary outcome of our study relates to the patient’s discharge date and no further observations beyond that date were included in the study, no follow-up or further data were recorded. Narcotic consumption was measured in oral morphine equivalents (OME). Immobility was subjectively attributed to the patients by the physician who made the initial contact based on the ability to stand up and walk without assistance. The factors “spinal stenosis” and “additional lumbar disc herniation (LDH)” refer to pathologies at other levels than the one it was operated on. A complete listing of all factors is presented in Table 1. The patients’ health status was determined by the American Society of Anaesthesiologists (ASA) risk classification, and the Charlson Comorbidity Index (CCI index) was collected and analysed separately from the other risk factors (Table 2).

Significant factors were identified as such, if their presence increased the patients’ median length of stay (LOS) by at least 1 day compared to the mean of nonaffected patients. For our study, LOS referred exclusively to the postoperative number of days the patients stayed in hospital.

The readiness to be discharged from the hospital was judged by the assigned surgeons based on the ability of voluntary bladder control, compensated pain symptoms, and sufficient mobility.

Statistical analysis was carried out using the IBM SPSS® Version 27 for Windows 10 (IBM Corp. Released 2020. IBM SPSS Statistics for Windows, Version 27.0. Armonk, NY: IBM Corp). Metric variables are presented as means and standard deviations (± SD) and medians with interquartile range (± IQR). Categorial variables are presented as number (n) and percentage (%). Bivariate statistical analysis was performed to evaluate the relation of the potential risk factors and prolonged hospital by using the Mann–Whitney U-test and Kruskal–Wallis test. To identify the most probable cause of bias in our statistics, the mean age of all factors that significantly influenced LOS in bivariate analysis was compared using Student’s t-tests. Correlation analyses were performed using Spearman’s correlation with Spearman’s Rho (r_s) for all significant factors in the bivariate analysis. Independent risk factors were identified by using a multivariable linear stepwise regression model with all factors being initially included. Afterwards, the factors were excluded step by step until only those being significant remained. A p-value < 0.05 was considered significant in two-tailed testing.

The decision for operative treatment of lumbar disc herniations in our department is based on the correlation of clinical presentation and imaging. Patients with paresis or conus-cauda-syndrome were primarily operated, other cases only after failure of conservative treatment.

All patients underwent single-level lumbar microdiscectomy. Small variations in the surgical management might have occurred as the patients included in the study were operated on by different surgeons. The primary goal of each operation was pressure relieve of the affected nerve root by removing the herniated disc fragment. If a rupture of the fibrous ring was found intraoperatively, a nucleotomy was performed as well to minimize the risk of recurrence. If the nerve root appeared to be still constricted on subsequent palpation, decompression of the root was performed as a final step.

The patients were operated on by different surgeons. However, each operation was performed by a specialist or under the supervision of a specialist.

A total of 200 patients who were operated on at our clinic between 2018 and 2022 were included in this study. The mean postoperative hospital stay was 4.73 ± 2.22 days (median 4 ± 0, min 1, max 16). Among them, 84 (42%) were female and 116 (58%) were male. 119 (59.5%) patients had a normal LOS, and 81 (40.5%) had an increased LOS. All other factors that were analysed are presented in Table 1.

To create comparability regarding the health status, the median ASA risk classification as well as the mean CCI values of both patient groups were compared before further bivariate and multivariable analysis. We did not include ASA and CCI among the other risk factors but analysed them separately. The reason for this is that the health status of the patients has an enormous influence on the length of hospital stay; a different health status of the two groups would thus generate a fundamental bias for the further analysis.

In our study, there was no statistical difference in the health status, concerning the ASA and the CCI Index, between patients with a normal LOS and patients with a prolonged LOS (Table 2).

Statistical analysis of factors potentially prolonging postoperative hospital stay was done for a total of 24 factors. The results of the bivariate analysis, the median LOS of all factors, and subgroups as well as the correlation coefficients can be seen in Table 1.

Sex (p = 0.002) was a significant demographic factor. Mean age for men was 51.6 ± 13.9 (median 51.5 ± 8.5) in comparison to the mean age for women 55.4 ± 16.1 (median 53.5 ± 16.3). Student’s t-test trended towards a significance in the age distribution (p = 0.10, CI =  − 7.67–0.78). Another demographic factor that was significant was the age of the overall population (p = 0.002).

Regarding preoperative physical status, only preoperative immobility reached statistical significance (p ≤ 0.001). Figure 1 shows the distribution of hospital discharges of preoperative mobile and immobile patients in relation to the LOS. The figure indicates a shift towards a delayed time of discharge for immobile patients. Thus, no patient who was mobile preoperatively left the hospital after the sixth day and no patient that was immobile preoperatively was discharged before the third postoperative day.

The intake of anticoagulation and/or antiplatelet agents (p = 0.045) was significantly related to LOS. The distribution of mean age and Student’s t-test showed that patients under those medications were considerably older than those who did not take anticoagulation and/or antiplatelet agents (mean 70.86 ± 10.9 vs. 51.1 ± 13.9, p < 0.001).

In our model, 40.2% (R² = 0.402) of the scattering in the risk of prolonged LOS after first lumbar microdiscectomy could be explained, resulting in a moderate effect size according to Cohen (1992) (Table 3).

Sex was a potential risk factor as women had a prolonged mean LOS of 1 day. However, the result is questionable in terms of reliability, since the mean age of men and women differs considerably, although not yet statistically significant.

In our study, age over 65 years was significantly correlated with LOS. This result reproduced and supported previous studies that presented a similar finding [7, 17, 22]. Khanaan et al. explained the correlation as a consequence of a greater number of comorbidities and a higher rate of complications after surgery [17]. However, age was not identified as an independent risk factor in the multiple regression presumably because of the restricted correlation and limited effect on LOS resulting in a modest percentage of variation in our regression model.

Regarding comorbidities, the only significant risk factor in bivariate analysis of our study was diabetes mellitus with diabetic patients having a median prolonged stay of 1 day. This finding is consistent with previous studies dealing with lumbar spine surgery [2, 13, 20]. Maloney et al. demonstrated an approximately 1.4-fold increase (1.9 vs 1.4 days) in LOS in a diabetic population undergoing open lumbar microdiscectomy [2]. Furthermore, Guzman et. al showed that controlled diabetes increased the LOS after degenerative lumbar spine surgery only by half a day (1.1-fold), while patients with uncontrolled diabetes mellitus had an increase of about 2.4 days (1.7-fold). This finding demonstrates the importance of glycaemic control in the perioperative period. That is especially relevant in patients with certain oral antidiabetics, which may have to be discontinued and substituted by insulin perioperatively.

The intake of anticoagulation and/or antiplatelet medication was significantly related to LOS; however, the factor seemed to be more of a confounder than an independent predictor for LOS as patients taking anticoagulation and/or antiplatelet medication were significantly older than those who did not.

The results of our study suggest that a postoperative need for opioids is associated with a prolonged LOS. A similar observation was made in other studies in patients with a high postoperative Visual Analogue Scale (VAS) [3]. However, identifying pain as an independent risk factor for prolonged LOS is quite difficult due to the subjective individual experience of pain on the one hand and the various potential confounding factors like an emotional response or anxiety on the other hand [9]. Furthermore, the prevalence of patients with chronic pain and with increased narcotic demand is especially high among those with spinal disorders [5]. Under consideration of possible aggravating factors that may intensify the patients’ experienced pain after surgery, adequate pain management is imperative to improve the functional outcome after surgery. As inadequately treated postsurgical pain contributes to longer hospital stays, slower progress in ambulation and development of chronicity of functional deficits [8]. Overall, the need for opioids after an operation appears to be more useful as an indirect measure of postoperative pain than as a predictive factor in itself.

Comparing surgical techniques, performance of nucleotomy plus sequestrectomy increased the median LOS by one day in contrast to simple sequestrectomy. A previous meta-analysis by Huang et. al from 2015 identified three studies comparing LOS after these two surgical techniques [1]. LOS in these studies ranged from 0.9 to 6.4 days in the sequestrectomy group and from 1.17 to 6.94 days in the nucleotomy group. While all three studies concluded shorter LOS after sequestrectomy, the difference was not statistically significant. Though, it can be concluded that while there might be some difference between the two surgical techniques, the correlation to LOS was modest (r = 0.16). Thus, our data supports the findings of previous studies that the choice between the two surgical techniques has no statistically relevant effect on the length of postoperative hospital stay.

The reasons that may influence preoperative immobility or decreased physical function are widespread. Among these are enhanced pain sensation, psychological, and psychiatric pre-existing conditions [19]. In addition, Bernstein et al. showed that socioeconomic disadvantage can have a negative impact on physical function during the initial clinical presentation in lumbar disc herniations (p ≤ 0.001) [21].

The strength of the survey of preoperative immobility as a predictor therefore results from the fact that it indirectly combines many other factors to one single factor. Apart from that, the assessment of preoperative immobility to predict prolonged LOS is easy and quick, as it does not need additional information to calculate the result. Our study has shown that the determination of preoperative immobility is particularly accurate in identifying patients who had a significantly prolonged LOS. All patients with a postoperative LOS of more than 6 days were classified with preoperative immobility and could be identified as at-risk patients for prolonged LOS during the initial assessment of the physical status.

Patients whose LOS is well above average are particularly at risk of complications and require significantly higher level of resources. The fact that the assessment of preoperative immobility was able to precisely identify these patients increases the clinical benefit of this factor. By using preoperative immobility as predictor, healthcare providers can tailor interventions and develop targeted strategies, such as better conservative pain management or periradicular infiltrations, to optimize postoperative recovery.

Limitations of the results arise, because in its current binominal form and therefore without the possibility of gradations, the factor loses informational content as it cannot determine the degree of severity in comparison to established clinical scores. Due to its limitations, the preoperative immobility as a predictor needs to be graded and be implemented as a score to make this factor comparable in future studies.

Beside preoperative immobility, no other factor of the initial physical status at the time of admission was significantly associated with LOS. This finding is consistent with other studies that were using different outcome measures like the Short Form (36) Health Survey (SF-36), Visual Analogue Scale (VAS), or Oswestry Disability Index (ODS) to predict outcome after lumbar microdiscectomy [15, 18].