Vitamin D deficiency during the perioperative period increases the rate of hardware failure and the need for revision fusion in adult patients undergoing single-level lumbar spine instrumentation surgery

Abstract

Background

Vitamin D has been shown to play important roles in both calcium homeostasis and bone healing. Only three studies have directly examined the relationship between vitamin D deficiency and hardware failure, nonunion, and/or revision surgery. Results are contradictory and none were large enough to provide the statistical power necessary to make definitive conclusions.

Methods

A retrospective analysis was performed utilizing the PearlDiver national insurance claims database consisting of 91 million individual patient records. Patients aged 30 and over who underwent a non-segmental posterior lumbar fusion procedure (CPT-22840) in 2012-2019 were included. Data collected included, hardware failure, revision surgery occurrence, and vitamin D deficiency. Hardware failure and revision rates were compared between vitamin D deficient and non-deficient groups. We ran a logistic regression analysis using the following variables: age, Charlson Comorbidity Index (CCI), gender, vitamin D deficiency, obesity, tobacco use, diabetes, osteoporosis, rheumatoid arthritis, and Crohn's disease.

Results

108,137 patients matching inclusion criteria were identified, with an overall hardware failure rate of 2.7% and revision rate of 4.1%. Failure rates were significantly higher for patients diagnosed with vitamin D deficiency during the full queried period (3.3% vs. 2.6%, OR = 1.26; p < 0.0001), as were revision rates (4.3% vs 3.5%, OR = 1.25; p < 0.0001). Patients diagnosed with deficiency pre-surgery, higher failure (3.1% vs 2.6%, OR = 1.19; p < 0.01) and rates of revision (4.4% vs 3.5%, OR = 1.27; p < 0.0001) were increased compared to the non-deficient group. In the logistic regression analysis, vitamin D deficiency remains a significant contributor to hardware failure and revision surgery.

Conclusions

These results demonstrate that pre- and/or post-operative vitamin D deficiency is independently correlated with risk for hardware failure and revision surgery in single-level lumbar fusion patients.

Introduction

For patients undergoing a lumbar spine fusion, optimizing physiological conditions to mitigate risk of hardware failure and the need for future revision is a critical step in preoperative planning and postoperative care. Vitamin D plays an important role in promoting calcium absorption in the digestive and renal tracts and is a critical factor in the pathways of bone formation and healing [1]. Therefore, it is hypothesized that adequate levels of vitamin D would be important in the successful outcome of spinal fusion surgery.

A high incidence of vitamin D deficiency has been reported in those undergoing lumbar spine procedures, with prevalence rates ranging from 23 – 76% [2], [3], [4], [5], [6]. It has also been shown that those with pre- and post-operative deficiencies have poorer patient reported outcome scores [2,7]. However, only three studies have directly examined the effect of hypovitaminosis D on lumbar fusion nonunion, hardware failure, and need for revision surgery [3,4,8].

Donnally III et al., in a retrospective single-center study, published an analysis of 150 lumbar spine fusion patients and found that neither pre- or post-operative vitamin D levels were associated with a change in risk for pseudoarthrosis, hardware failure, or revision surgery [8]. In another retrospective single-center study, Stoker et al. evaluated 85 patients undergoing 4+ level fusions, and found that vitamin D deficiency was not a risk factor for increased rates of revision surgery [3]. Conversely, Ravindra et al. found in a prospective observational single-center study of 133 spinal fusion patients that vitamin D deficiency was associated with an increase in nonunion rates in multivariate analysis [4]. The results of these studies and the conclusions that can be drawn are limited by their small sample sizes and the non-homogenous surgical treatments received [3,4,8]. In the second and third studies, the cohorts analyzed were also not limited to lumbar patients.

Our study was a retrospective observational study designed to be sufficiently powered utilizing PearlDiver's national claims databases to study the association between vitamin D deficiency and hardware failure/revision surgery in lumbar fusion patients. In order to decrease the variability in the study population, it was designed to include only patients undergoing a single segment lumbar fusion.

Methods

Study database

This study was a retrospective cross-sectional review of longitudinally followed lumbar fusion patients from 2012 to 2019. Institutional IRB approval was obtained as a low-risk non-human study. The PearlDiver database includes standardized and longitudinal HIPAA-compliant medical record claims data from the years 2010 to 2020. The specific database used, M91, includes anonymized medical records of 91 million individual patients randomly selected from all payers except Kaiser and Tricare. The database includes Medicare and Medicaid patient populations. The data can be interrogated for individual patients in a longitudinal manner for medical diagnoses, complications, and treatments.

Patient cohort

The study population was defined as patients aged 30-85 who underwent a lumbar non-segmental single-level instrumented fusion procedure, identified through CPT code, 22840, between the years 2012 and 2019. We excluded patients who had the 22840 procedure code along with a cervical or thoracic surgery on the same day to ensure that the study cohort was only lumbar surgery patients.

Variable definitions

Subsequent hardware failure was identified through ICD-9 and ICD-10 diagnosis codes. The codes used were: Breakdown of internal fixation device of vertebrae (ICD-10-D-T84216D; ICD-10-D-T84216A), displacement of internal fixation of vertebrae (ICD-10-D-T84226A; ICD-10-D-T84226D), other mechanical complication of internal fixation device of vertebrae (ICD-10-D-T84296A; ICD-10-D-T84296D), and Other mechanical complication of internal orthopedic device or implant (ICD-9-D-99649).

Revision surgery occurrence was identified by querying patient records for CPT codes signifying a lumbar spine fusion surgery which occurred later than three months after the index surgery. The patients who were considered to have had revision fusion surgery had any spinal fusion procedure at least 3 months after the index procedure. The codes used were: posterior interbody fusion (CPT-22630, CPT-22633), anterior interbody fusion (CPT-22558), posterior fusion (CPT-22612), and posterior instrumentation (CPT-22840, CPT-22842, CPT-22843, CPT-22844). The posterior instrumentation was determined to be associated with thoracic or lumbar surgery if there is no cervical posterior fusion code (CPT-22600) was associated with it.

We defined vitamin D deficiency during the time of surgery and perioperative period by identifying patients with ICD diagnostic code (ICD-9-CM-268.9, ICD-10-CM-E55.9) in the period within 730 days before and 365 days after the surgery. The hardware failure and revision rates were compared between two cohorts defined as vitamin D deficient and not-deficient.

Statistical analysis

The rates of hardware failure and revision fusion surgery between patients with and without vitamin D deficiency were compared using the Chi-squared method. Logistic regression analysis of hardware failure and revision surgery was done for vitamin D deficiency with other variables known to contribute to postoperative complications (age [9,10], gender [11], obesity [12,13], diabetes [14], smoking [15], and osteoporosis [16]). Specifically, two separate logistic regression analyses were conducted. The first multivariable logistic regression included hardware failure as the dependent variable and age, Charlson Comorbidity Index (CCI), gender, vitamin D deficiency, obesity, tobacco use, diabetes, osteoporosis, rheumatoid arthritis, and Crohn's disease as predictor variables. The same predictor variables were used in the second multivariable logistic regression with revision as the dependent variable. Rheumatoid arthritis and Crohn's Disease were added as predictor variables to account for patients with chronic conditions.

Results

The total number of patients identified as having a non-segmental lumbar instrumentation was 108,137 (Table 1). Of these, 11.2% of the patients had vitamin D deficiency identified during the preoperative period and 15.2% were diagnosed in the combined pre- and post-operative period as defined by the search.

Table 1.

Sample Demographics

Age (Mean, SD)	All patients (n = 108,137)	60.3 (12.0)
	Vitamin D deficient (n = 16,408)	61.0 (11.3)
	Not vitamin D deficient (n = 91,729)	60.1 (12.1)
Gender	Male (n = 45,201)	41.8%
	Female (n = 62,936)	58.2%
Obesity	Yes (n = 32,549)	30.1%
	No (n = 75,588)	69.9%
Diabetes	Yes (n = 39,362)	36.4%
	No (n = 68,775)	63.6%
Smoking	Yes (n = 10,706)	9.9%
	No (n = 97,431)	90.1%
Osteoporosis	Yes (n = 17,410)	16.1%
	No (n = 90,727)	83.9%

Overall hardware failure was identified in 2,949 patients at a rate of 2.7%. Failure rates were significantly higher for patients diagnosed with vitamin D deficiency during the full queried period than those never diagnosed (3.3% vs. 2.6%, OR = 1.26; p < 0.0001). Univariate chi-squared analysis showed that vitamin D supplementation in patients with vitamin D deficiency compared to those without supplementation had no association with either revision (p = 0.57) or hardware failure (p = 0.47).

Complications following lumbar surgery may result in the physician ordering vitamin D tests, leading to a post-operative identification of vitamin D deficiency. To address this potential sampling bias, we also ran separate analyses with patients whose deficiencies were diagnosed preoperatively. Those with pre-operative diagnoses of vitamin D deficiency had a higher likelihood of hardware failure compared to those who never received a deficiency diagnosis throughout the query period (3.1% vs 2.6%, OR = 1.19; p < 0.01). The Kaplan-Meier Survival Curve (Fig. 1) demonstrates the differential rate of failure which increases with time. There was no statistical difference in failure rates between patients with vitamin D deficiency diagnosed in the pre- vs. post-operative periods (3.1% vs 3.6%, OR = 0.87; p = 0.10).

Fig. 1.

Kaplan Meier Survival Curve for Implant Survival. Solid line denotes those without vitamin D deficiency; dash line denotes those with vitamin D deficiency.

Revision fusion surgery was identified in 4,417 patients at a rate of 4.1%. For patients interrogated for vitamin D status during the entire pre- and post-operative period, those patients with vitamin D deficiency had a higher likelihood of revision surgery compared to those without deficiency (4.3% vs 3.5%, OR = 1.25; p < 0.0001). Kaplan-Meier Survival Curve (Fig. 2) demonstrates increasing revision fusion rate for vitamin D deficient patients with the passing of time. Those with pre-operatively diagnosed vitamin D deficiencies had higher revision rates than those never diagnosed (4.4% vs 3.5%, OR = 1.27; p < 0.0001). There was no difference in revision rates found between patients with vitamin D deficiency diagnosed in the pre- vs. post-operative periods (4.4% vs 4.2%, OR = 1.03; p = 0.69).

Fig. 2.

Kaplan Meier Survival Curve for Not Undergoing Revision Surgery. Solid line denotes those without vitamin D deficiency; dash line denotes those with vitamin D deficiency.

A logistic regression analysis of hardware failure and revision surgery rates demonstrated that vitamin D deficiency remained significantly associated with hardware failure (p = 0.0076; OR = 1.14; 95% CI = 1.10-1.28) (Table 2) and revision surgery (p < 0.0001; OR = 1.17; 95% CI = 1.10-1.25) (Table 3).

Table 2.

Lumbar Fusion Hardware Failure Logistic Regression Analysis. Lack of hardware failure used as referent for Odds Ratio.

	z-value	Significant	Odds Ratio (95% CI)
Age (Years)	-11.912	p < 0.0001	0.98 (0.98-0.98)
CCI	-1.046	p = 0.2953	0.99 (0.97-1.01)
Gender (Male	4.304	p < 0.0001	1.19 (1.10-1.28)
Hypovitaminosis D	2.669	p = 0.0076	1.14 (1.04-1.26)
Obesity	8.159	p < 0.0001	1.38 (1.28-1.49)
Tobacco Use	8.663	p < 0.0001	1.43 (1.32-1.55)
Diabetes	5.123	p < 0.0001	1.24 (1.14-1.34)
Osteoporosis	9.896	p < 0.0001	1.65 (1.49-1.82)
Rheumatoid Arthritis	7.16	p < 0.0001	1.54 (1.37-1.74)
Crohn's Disease	0.58	p = 0.5617	1.09 (0.81-1.47)

Table 3.

Lumbar Fusion Revision Surgery Logistic Regression Analysis. Lack of revision surgery used as referent for Odds Ratio.

	z value	Significant	Odds Ratio (95% CI)
Age (Years)	-1.001	p = 0.3170	1.00 (1.00-1.00)
CCI	-7.865	p < 0.0001	0.95 (0.94-0.96)
Gender (Male	4.195	p < 0.0001	1.11 (1.06-1.17)
Hypovitaminosis D	4.796	p < 0.0001	1.17 (1.10-1.25)
Obesity	10.165	p < 0.0001	1.3 (1.23-1.36)
Tobacco Use	7.804	p < 0.0001	1.25 (1.18-1.32)
Diabetes	7.251	p < 0.0001	1.22 (1.15-1.28)
Osteoporosis	8.035	p < 0.0001	1.31 (1.22-1.40)
Rheumatoid Arthritis	9.373	p < 0.0001	1.47 (1.35-1.59)
Crohn's Disease	3.922	p < 0.0001	1.44 (1.2-1.72)

Discussion

Vitamin D is recognized as an important hormone in maintaining bone health and prevention of osteoporosis [1]. Because of its direct action in bone mineralization, deficiencies in vitamin D have been theorized to increase the chance of developing pseudoarthrosis after spinal fusion and subsequent hardware failure and need for revision. Despite the recognition of high prevalence of vitamin D deficiency in the general US adult population [17], [18], [19], [20], [21] and serious potential complications that this deficiency can cause in spinal fusion patients, there is a surprisingly low number of scientific inquiries of this topic. The few papers published are all based on small series of patients from single institutions which mainly describe a high prevalence of deficiency but offer no conclusive evidence of its effect on surgical failure [3,4,8].

Relative paucity of existing literature on this topic is most likely due to the relatively uncommon rates of hardware failure and pseudarthrosis, especially in single level spinal fusion, requiring a large cohort of patients for a sufficiently powered study to determine the effects of any variables. Khalid et al. demonstrated that for 1,723 patients undergoing a single level lumbar fusion surgery with normal bone mineral density, rates of pseudoarthrosis were found at 4.1% and revision surgery at 2.4% [16]. Given these relatively low rates of pseudarthrosis and revision surgery, detecting a clinically relevant and statistically significant difference would require a cohort of more than 12,000 and 30,000 patients, respectively. Therefore, the analysis of this type of effect cannot be done through a single institutional study and may be only be answerable by a large population-based study, as in the present study.

In this study, we used the PearlDiver database consisting of longitudinally recorded medical insurance claims of 91 million individual patients. We queried the database for concurrent diagnoses of single level lumbar spinal fusion, hardware failure, revision surgery, and vitamin D deficiency.

This study demonstrates that there was a 26% increase in the incidence of hardware failure in patients who are vitamin D deficient compared with non-deficient patients. Similarly, revision surgery incidence was higher by 25%. When we analyzed those patients who had vitamin D deficiency diagnosed before the surgery and compared them to patients without vitamin D deficiency, we found similarly increased rates of hardware failure and revision surgery. This analysis was done to eliminate the post-surgery bias of physicians ordering vitamin D tests after complications such as fusion failure or need for revision surgery.

The correlation between low vitamin D levels and low bone mineral density and osteoporosis has been well established [1,22,23], as well as the correlation between osteoporosis and hardware failure in lumbar fusion [16,24,25]. However, direct correlation between osteoporosis and the need for revision remains unsettled [16,25,26]. While it would be reasonable to assume the effect of vitamin D deficiency on hardware failure and possibly revision in this study may be due to osteoporosis acting as a mediating variable, logistic regression analysis demonstrated vitamin D deficiency acts as an independent predictor of these outcomes regardless of osteoporosis status (Tables 2 and 3).

Vitamin D deficiency has also been associated with non-osseous conditions such as cardiovascular disease [27], acute myocardial infarction [28], diabetes [29], autoimmune disease [30], upper respiratory tract infection [31], several types of cancer [32], and multiple sclerosis [33]. Many of these factors could also play an important role in outcomes of spinal surgery, increasing risk for medical complications and delaying recovery. Depression [34] and anxiety [35] also have demonstrated associations with vitamin D deficiency, and with worsened patient reported outcomes (PROs) after lumbar fusion [36]. As previously mentioned, Vitamin D deficiency itself has been associated with lower PROs scores after lumbar surgery [2,7]. These lower scores may influence physicians to consider revision surgery more readily than they might for a patient with the same imaging who is subjectively doing well. Similarly, vitamin D deficiency has been associated with chronic body pain [37], muscle weakness [38], poor bone quality [39], and poor bone healing [40]; all these factors can affect the need for future revision surgery. Regardless of the specific pathway, it is clear from these associations and the present study's results that vitamin D deficiency has the potential to negatively impact lumbar fusion outcomes in a significant way for the individual patient.

Although severe Vitamin D deficiency diseases such as Rickets are relatively rare [41], the high prevalence of deficiency in the United States adult population has been extensively reported in literature. The prevalence rates of reported deficiency in the US are in the range of 28.9-41.6% for all adults [17], [18], [19] and 40-100% in the elderly [20,21]. In this study, the identified cohort of vitamin D deficient patients in the perioperative period was 15% of total undergoing fusion; which is clearly lower than these previous estimates. We determined it is likely our series underestimates the true rate of vitamin D deficiency in our sample by 10-25%, possibly denoting a significant lack of screening for hypovitaminosis D by spine surgeons, which has also been reported for osteoporosis and osteomalacia [42]. Since this unidentified cohort will be embedded in both deficient and non-deficient groups, it is likely that our findings underrepresent the true effect of vitamin D deficiency on hardware failure or revision surgery. However, those who carry the diagnosis of vitamin D deficiency in this analysis are likely to be truly deficient, as the only way to make this diagnosis is with laboratory values.

This study was limited in that the identification of vitamin D deficiency relied on documentation of vitamin D deficiency diagnostic coding and was not based on laboratory values, since these were not available in the database. This methodology therefore could not determine the severity of deficiency and also missed those patients who were not tested for vitamin D deficiency, so the reported rate in this paper most likely under-represents the deficiency rate in the population which may alter the result. There is no firm diagnostic criteria for hardware failure other than the physician's impression and recording of the diagnosis, which may lead to decreased interrater reliability in this variable. However, by using hardware failure codes rather than pseudoarthrosis codes, we felt that we could most accurately capture these complications as hardware failure following spine fusion is more easily recognized in the database than pseudoarthrosis. Because the need for additional surgery can have various causes, such as adjacent segment disease, trauma, or other issues unrelated to hardware failure, revision procedures may include surgery at other levels than the index surgery level and therefore, we cannot attribute this exclusively to poor bone healing/hypovitaminosis D. Regardless, we do not expect this to be a significant subset of the cohort as adjacent segment disease diagnoses typically occur several years after the initial surgery [43,44]. In this study, we did not control for vitamin D supplementation or other osteoporosis treatments. Finally, this study was limited by the fact that it is a retrospective inquiry into an administrative database and therefore is reliant on accuracy of inputted data. A larger prospective study is most likely needed to answer the true prevalence of this deficiency in the spinal surgery population and to determine the true incidence of these associated complications.

Conclusion

This paper is the first that uses a large cohort of patients, sufficiently powered to examine the effects of Vitamin D deficiency on hardware failure and revision surgery rates in patients undergoing lumbar spinal fusion. The study demonstrates that there is a clinically meaningful increase in the rates of these complications with high statistical significance in patients diagnosed with vitamin D deficiency.

Declaration of Competing Interest

One or more of the authors declare financial or professional relationships on ICMJE-NASSJ disclosure forms.