Documented Cannabis Use Is a Risk Factor for Nonunion After Nonoperative Management of Scaphoid Fractures: A Retrospective Review of 159 998 Patients

Abstract

Background:

The purpose of the present study was to assess whether an association exists between cannabis use and nonunion among patients with acute scaphoid fractures initially treated nonoperatively.

Methods:

Using the PearlDiver national data set (2010-2020), adult patients with a scaphoid fracture were identified. Patients initially treated with nonoperative management were selected, and 3 subgroups were defined: non-cannabis/non-tobacco (non-users), tobacco-only, cannabis-only. Matched cohorts for each subgroup controlled for age, sex, and comorbidities. Nonunion rates at 6 months, 12 months, and 2 years after fracture diagnosis were determined for each subgroup’s matched cohort. Rates of subsequent nonunion surgery were also determined. Multivariable analysis was then done to determine potential associations.

Results:

A total of 159 998 scaphoid fracture patients were identified. The non-users’ group exhibited a nonunion rate of 7.7%. The nonunion rates for the tobacco-only and cannabis-only groups (11.8% and 10.1% respectively) were significantly greater than that of the non-users group. Based on multivariable analysis, cannabis use and tobacco use were individually associated with a greater risk of scaphoid nonunion as well as subsequent nonunion surgery at 6 months, 12 months, and 2 years after index fracture.

Conclusion:

Documented cannabis use in the electronic medical record is associated with an increased risk of scaphoid nonunion and subsequent nonunion surgery. Further studies are necessary to clarify if this association can be contributed solely to cannabis use, to mixing cannabis or tobacco, or to other variables. In the meantime, physicians can counsel their patients on this potential association and increased risk.

Introduction

Scaphoid fractures account for 50% to 80% of carpal bone fractures in young, active individuals. ¹ Typically resulting from a fall onto an outstretched hand, these fractures have a risk of nonunion due to retrograde blood supply, substantial cartilage surface area, and instability. ² Scaphoid nonunion can lead to further wrist pathology including carpal instability and arthritis. As such, prompt evaluation, appropriate treatment, and close monitoring are essential to maximize patient outcomes and limit the risk of nonunion. ¹

Several risk factors for nonunion following scaphoid fracture have previously been identified, including fracture displacement > 1 mm, proximal pole fractures, nicotine use, male sex, comorbid psychiatric conditions, younger age, and surgical treatment.^{3 -5} Nicotine use is of particular interest because it is the only one of these risk factors that is potentially modifiable. Smoking is known to negatively impact bone mineral density, affect microvascularity, and alter calcium and vitamin D metabolism.^6,7 In fact, past studies have identified union rates over 80% in non-smokers, compared with 46% in non-smokers. ³

Despite recent widespread medical and recreational legalization of cannabis across the United States, ⁸ limited data exist regarding the potential impact of cannabis use on bone healing. Sophocleous et al reported decreased bone mineral density in cannabis users.^9,10 However, Bourne et al ¹¹ in 2017 found no association between cannabis use and bone mineral density. In 2020, Heath and associates found, through a retrospective study, an association between cannabis use and increased time to union among surgically treated pediatric fracture patients. ¹²

With the increasing prevalence of cannabis use due to recent legalization efforts, it is important to gain further insight into how cannabis may impact scaphoid fracture healing. The present study aimed to assess whether an association exists between documented cannabis use and nonunion among acute scaphoid fractures initially treated nonoperatively using the PearlDiver M151 database, a comprehensive national billing claims database. We hypothesize that documented cannabis use would not be a risk factor for the development of a scaphoid nonunion.

Methods

Study Cohort

Data was abstracted from 2010 to 2021 M151 Ortho PearlDiver Mariner Patient Claims Database (PearlDiver Technologies, Colorado Springs, CO, USA); this is a large, national multi-insurance administrative claims database which includes records for approximately 151 million orthopedic patients and is well established in the orthopedic/hand literature.^{13 -15} Information in this data set output is in de-identified aggregated form, which is Health Insurance Portability and Accountability Act-compliant. For this reason, our institutional review board determined studies using this data set to be exempt from review.

Adult patients (above the age of 18) who sustained an acute scaphoid fracture were identified from the data set using International Classification of Diseases -9 and -10 (ICD-9 and ICD-10) codes. Scaphoid fractures that occurred between January 1, 2010, and January 1, 2020, were included. Exclusion criteria included infection, neoplasms, trans-scaphoid perilunate dislocations, and pediatric patients (age < 18). To exclude patients who did not undergo initial nonoperative management, patients treated with surgery within the 1 month of their fracture diagnosis were excluded using Current Procedural Terminology (CPT) codes.

Patient Outcome Metrics

Once the final exclusion criteria had been applied, patient demographics including age, sex, and Elixhauser Comorbidity Index (ECI) were collected. Elixhauser Comorbidity Index is an estimate of comorbidity burden using ICD diagnosis codes; ECI has been extensively used in the past for statistical matching in several administrative database studies.^14,16,17

We then used ICD-9/ICD-10 codes to divide the patients into 3 groups: (1) cannabis-only users; (2) tobacco-only users; and (3) non-users (neither cannabis nor tobacco). Concomitant cannabis and tobacco users were not included in the present analysis due to limited patient numbers leading to insufficient power to draw adequate comparisons. Statistical matching was performed to control for age, sex, and ECI in a 1:1:1 ratio (Figure 1). Users who used both cannabis and tobacco were not included to allow for an independent assessment of isolated cannabis and isolated tobacco use.

Figure 1.

Schematic diagram depicting process of categorizing scaphoid fracture patients, followed by statistical matching to arrive at 3 matched cohorts.

For each of the 3 matched groups, nonunion rates were determined based on ICD coding at 6 months, 1 year, and 2 years after injury. Rates of subsequent surgery for nonunion were then determined based on CPT codes 25628, 25440, and 25430 at 6 months, 1 year, and 2 years after injury. As such, the primary outcome metrics for the present study were incidence and odds of scaphoid nonunion surgery among initially nonoperatively managed scaphoid fractures among non-smokers, cannabis smokers, and tobacco users.

Data Analysis

Patient demographics were first compared among the 3 unmatched and matched groups through univariate analysis using chi-square analysis. Among the matched groups, nonunion rates as well as rates of surgical intervention for nonunion were compared using chi-square at 6 months, 1 year, and 2 years after initial fracture. Subsequent multivariable analysis was then performed to determine whether cannabis use and tobacco use were independently associated with (1) scaphoid nonunion diagnosis; (2) surgery for scaphoid nonunion at each of the studied time points.

Results

Patient Cohort

Of the identified groups, non-users were 145 467 (88.8%), tobacco-only users were 13 132 (8.0%), and cannabis-only users were 1399 (0.9%) (Table 1). Before statistical matching was performed (Table 1), the mean age (±standard deviation) was 46.8 ± 19.2 year for the non-users group, 44.3 ± 15.6 years for the tobacco-only group, and 33 ± 13.4 years for the cannabis-only group. The ECI for the non-users group (±standard deviation) was 2.57 ± 3.04, 4.20 ± 3.5 for the tobacco-only group, and 5.18 ± 3.2 for the cannabis-only group. Sex distribution varied between the 3 groups. Matching based on age, sex, and ECI was then done 1:1:1. After matching, each group had 1343 patients and were no longer different based on age, sex, or ECI (Table 1).

Table 1.

Univariate Analysis of Unmatched and Matched Scaphoid Fracture Patients, Stratified by Tobacco/Cannabis Use.

	Unmatched				Matched
Variable	Non-users	Tobacco only	Cannabis only	P value	Non-users	Tobacco only	Cannabis only	P value
N (%)	145 467 (88.8%)	13 132 (8.0%)	1399 (0.9%)		1343 (33.3%)	1343 (33.3%)	1343 (33.3%)
Age (mean ± st dev)	46.8 ± 19.2	44.3 ± 15.6	33.0 ± 13.4	< .0001	33.3 ± 13.3	33.4 ± 13.3	33.1 ± 13.3	1.000
Sex
Female (N, %)	76 292 (52.4%)	6636 (50.5%)	373 (26.7%)	< .0001	367 (27.3%)	367 (27.3%)	367 (27.3%)	1.000
Male (N, %)	69 175 (47.6%)	6496 (49.5%)	1026 (73.3%)		976 (72.7%)	976 (72.7%)	976 (72.7%)
ECI (mean ± st dev)	2.57 ± 3.04	4.20 ± 3.50	5.18 ± 3.32		4.99 ± 3.06	4.99 ± 3.06	4.99 ± 3.06	1.000
0-3	105 705 (72.7%)	6755 (51.4%)	524 (37.5%)	< .0001	515 (38.3%)	515 (38.3%)	515 (38.3%)
4-6	24 027 (16.5%)	3490 (26.6%)	514 (36.7%)		503 (37.5%)	503 (37.5%)	503 (37.5%)
7-9	9745 (6.7%)	1725 (13.1%)	208 (14.9%)		196 (14.6%)	196 (14.6%)	196 (14.6%)
10 +	5990 (4.1%)	1162 (8.8%)	153 (10.9%)		129 (9.6%)	129 (9.6%)	129 (9.6%)

Note. Univariate analysis of unmatched and matched scaphoid fracture patients stratified by use of tobacco or cannabis. ECI = Elixhauser Comorbidity Index.

Incidence of Scaphoid Nonunion

In the non-users group, the scaphoid nonunion rate was 6.5% at 6 months, 7.4% at 1 year, and 7.7% at 2 years after fracture. In the tobacco-only group, scaphoid nonunion rates at these time intervals were 10.4%, 11.6%, and 11.8%, respectively. In the cannabis-only group, scaphoid nonunion rates at 6 months, 1 year, and 2 years after fracture were 8.7%, 9.8%, and 10.1%, respectively. When performing chi-square, each of the 3 groups’ scaphoid nonunion rates significantly differed from each other (P < .05).

When performing multivariable analysis, documented cannabis use was found to be a risk factor for scaphoid nonunion at 6 months (odds ratio [OR]: 1.45, 95% confidence interval [CI] [1.08, 1.96], P = .015), 1 year (OR = 1.45, 95% CI [1.09, 1.92], P = .0107), and 2 years (OR = 1.44; 95% CI [1.09, 1.90], P = .0107) after initial fracture (Table 2, Figure 2). Tobacco use was also noted to be a risk factor for nonunion at 6 months (OR = 1.68, 95% CI [1.27, 2.25], P = .0004), 1 year (OR = 1.65, 95% CI [1.26, 2.18], P = .0003), and 2 years (OR = 1.62, 95% CI [1.24, 2.13], P = .0004) after initial fracture.

Table 2.

Multivariate Analysis: Risk of Scaphoid Nonunion Diagnosis Among Matched Cohorts.

N = 3588	OR [95% CI]	P value
Nonunion diagnosis at 6 months
Non-users (Referent)
Tobacco only	1.68 [1.27, 2.25]	.0004
Marijuana only	1.45 [1.08, 1.96]	.0151
Nonunion diagnosis at 12 months
Non-users (Referent)
Tobacco only	1.65 [1.26, 2.18]	.0003
Marijuana only	1.45 [1.09, 1.92]	.0107
Nonunion diagnosis at 2 years
Non-users (Referent)
Tobacco only	1.62 [1.24, 2.13]	.0004
Marijuana only	1.44 [1.09, 1.90]	.0107

Note. Multivariate analysis: Risk of nonunion diagnosis following acute scaphoid fracture among matched cohorts. OR = odds ratio; CI = confidence interval.

Figure 2.

Forrest plot depicting risk of scaphoid nonunion (multivariable analysis) among tobacco users and cannabis users at time intervals.

Incidence of Surgery for Scaphoid Nonunion

In the non-users group, the rate of surgical intervention rate for nonunion was 1.9% at 6 months, 2.4% at 12 months, and 2.4% at 2 years after fracture. In the tobacco-only group, scaphoid nonunion rates at these time intervals were 4.4%, 5.4%, 5.5%, respectively. In the cannabis-only group, scaphoid nonunion rates at 6 months, 1 year, and 2 years after fracture were 3.6%, 4.5%, and 4.5%, respectively. When performing chi-square, each of the 3 groups’ surgical intervention rates significantly differed from each other (P < .05).

When performing multivariable analysis, documented cannabis use was noted to be a risk factor for scaphoid nonunion surgery at 6 months (OR = 2.14, 95% CI [1.32, 3.54], P = .0025), 1 year (OR = 2.01, 95% CI [1.28, 3.18], P = .0025), and 2 years (OR = 2.05, 95% CI [1.32, 3.24], P = .0018). Tobacco use was also noted to be a risk factor for scaphoid nonunion surgery at each time point studied (P < .05, Table 3, Figure 3).

Table 3.

Multivariate Analysis: Risk of Scaphoid Nonunion Surgery Among Matched Cohorts.

N = 3588	OR [95% CI]	P value
Surgery for scaphoid nonunion at 6 months
Non-users (Referent)
Tobacco only	2.43 [1.52, 3.98]	.0003
Cannabis only	2.14 [1.32, 3.54]	.0025
Surgery for scaphoid nonunion at 12 months
Non-users (Referent)
Tobacco only	2.43 [1.59, 3.80]	< .0001
Cannabis only	2.01 [1.28, 3.18]	.0025
Surgery for scaphoid nonunion at 2 years
Non-users (Referent)
Tobacco only	2.52 [1.65, 3.93]	< .0001
Cannabis only	2.05 [1.32, 3.24]	.0018

Note. Multivariate analysis: Risk of surgery for scaphoid nonunion following acute scaphoid fracture among matched cohorts. OR = odds ratio; CI = confidence interval.

Figure 3.

Forrest plot depicting risk of scaphoid nonunion repair (multivariable analysis) among tobacco users and cannabis users at time intervals.

Discussion

Cannabis use nationwide continues to increase due to recent legalization efforts. US medical and recreational cannabis sales were estimated to be approximately $33.6 billion by the end of 2023 and subsequently $56.9 billion by the end of 2028. ¹⁸ Despite the projected increases in usage, data on the impact of cannabis in the field of orthopedics remains limited. While one study reported that cannabis use may be associated with decreased bone mineral density, another study found no such association.^10,11 Copeland-Halperin et al ¹⁹ recently reported in 2021 that “little is known about [cannabis]’ perioperative effects on wound healing, or on cardiovascular, pulmonary, and hematologic physiology.”

The present study investigates a potential association between documented cannabis use in the health record and scaphoid fracture nonunion. We specifically focused attention on acute scaphoid fractures that were initially treated nonoperatively. Using a national, administrative claims data set, records of 159 998 scaphoid fracture patients were identified for review. The overall rate of nonunions within 2 years after fracture was 9.9%, which is similar to reported rates of 5% to 12% in the literature.^{20 -23} The nonunion rate of cannabis-only users was 10.1%, significantly greater than that of the non-users group (7.7%).

On multivariable analysis, documented cannabis use was found to be independently associated with scaphoid nonunion after controlling for age, gender, and comorbidities. This relationship maintained statistical significance (P < .05) at 6 months, 1 year, and 2 years from fracture. One previous retrospective investigation also found an association between cannabis use and delayed union in a different fracture population (surgically treated pediatric fractures). ¹² Furthermore, the present study found that documented cannabis users were approximately 2 times more likely than non-users to undergo surgical intervention for a scaphoid nonunion (OR = 2.01, P = .0025).

In the present study, documented tobacco users exhibited the greatest incidence of scaphoid nonunion (11.8%), when compared with cannabis-only users (10.1%) and non-users (7.7%). Tobacco users were also at highest risk to have surgery for scaphoid nonunion. Tobacco use (specifically nicotine) is a well-known risk factor for nonunion after a scaphoid fracture.^21,24 The present findings reaffirm that tobacco use is a significant predictor of scaphoid nonunion.

There are several limitations to this study, some of which are inherent to the use of national datasets. First, though our methodology enabled us to study a large cohort of scaphoid fracture patients, the data set is susceptible to entry and coding mistakes. We were only able to identify cannabis users, tobacco users, and presence of a nonunion only if the patients had an associated ICD diagnosis code. However, it is important to note that the present study was able to identify an overall scaphoid nonunion rate of 9.9%, which matches that of previous studies. Tobacco use was identified as a significant risk factor for nonunion, which was an anticipated finding. Second, the limitations of the PearlDiver data set precluded a more in-depth analysis of avascular necrosis or location of fracture (ie, waist vs proximal pole); these are important variables that should guide further investigations on this topic. It was also not possible to differentiate mode of cannabis consumption (smoking, vaping, edibles, etc). Furthermore, this study is retrospective in nature, which limits the generalizability of the data sample when compared with prospective studies and limits our ability to determine cause-and-effect of cannabis use on scaphoid nonunion—specifically due to our inability to determine temporality and dose-response relationship. Finally, the discrepancy between nonunion diagnoses and surgery is likely due to several factors, including coding errors within the database, patient loss-to-follow-up, patients declining surgery, and patients pursuing nonunion repair surgery at a different site than where they were diagnosed (which falls outside the capture net of the PearlDiver database).

Conclusion

Despite these limitations, the authors assessed the relationship between documented cannabis use and nonunion among scaphoid fractures initially treated nonoperatively. Multivariable analyses show that documented cannabis use is independently associated with a nonunion diagnosis as well as surgical intervention for a scaphoid nonunion. Future studies will be instrumental in further characterizing the association between cannabis use and scaphoid nonunion; namely, whether this can be attributed solely to cannabis use, to mixing cannabis or tobacco, or to other psychosocial factors. In the interim, physicians can counsel patients accordingly on this potential association and increased risk.