Foot & Ankle International - July 2018 - 823

823

Humbyrd et al
regression, including age, race, sex, BMI value, cause of
ESRD, and year of study entry, to estimate the adjusted
odds ratio (aOR) for receiving operative vs nonoperative
treatment among nonwaitlisted, waitlisted, and transplanted
groups and to identify factors associated with receipt of
operative treatment.
We also reviewed the Medicare inpatient claims of
patients who sustained ATR to ascertain whether they were
admitted to the hospital within 14 days of diagnosis. We
conducted multivariate logistic regression using the same
set of covariables as the previous model to compare the
odds of hospital admission among groups.

Fluoroquinolone and Corticosteroid Use
To characterize the associations of fluoroquinolone and corticosteroid use with ATR, we reviewed Medicare Part D
Prescription Drug Coverage claims of the study cohort.
Because Medicare Part D coverage started in January 2006,
this analysis was restricted to 544 185 individuals (50%)
with confirmed Medicare Part D enrollment until their last
day of follow-up. We also reviewed the immunosuppression records of the transplanted group because corticosteroids are often used for maintenance immunosuppression,
which is usually covered by Medicare Part A and B.
Use of fluoroquinolones and corticosteroids was treated
as a time-varying exposure. That is, individuals were treated
as unexposed until the first day of prescription and as
exposed thereafter. Using the same method described
above, we estimated the aIRR of ATR among nonwaitlisted,
waitlisted, and transplanted groups for those with Medicare
Part D coverage. Then, we added fluoroquinolone/corticosteroid exposure as a covariable to examine whether exposure to the medications explains the differences in ATR
incidence among nonwaitlisted, waitlisted, and transplanted
groups. Last, we stratified the population by fluoroquinolone/corticosteroid exposure to compare the aIRRs among
those who were exposed to fluoroquinolones/corticosteroids vs those who were unexposed.

Operative Site Infection
Among patients who underwent operative treatment for
ATR, we ascertained the incidence of SSI using ICD-9-CM
code 998.59 (other postoperative infection). Patients whose
records indicated this code within 30 days of the initial ATR
diagnosis were considered to have developed a SSI.

Statistical Analyses
We compared baseline characteristics using Student t tests
for continuous variables and χ2 tests for categorical variables. BMI values were missing for 8.7% of patients. These
missing values were treated with a missingness indicator.

Alpha level was set to .05. All analyses were conducted
using Stata/MP for Unix, version 14.2, software (StataCorp
LP, College Station, TX).

Results
Study Population
We identified 1091 ATRs during 2 870 817 person-years of
follow-up. Patients who sustained an ATR were significantly younger (61 vs 68 years) and more likely to be white,
to have a higher BMI (29 vs 27 kg/m2), to have diabetes or
glomerulonephritis as the cause of ESRD (53% vs 45% and
10% vs 7.8%, respectively), and to be ambulatory (5.7% vs
2.6% nonambulatory) compared with uninjured patients
(Table 1). Patients with ATR had fewer comorbidities than
those without, with a significantly lower prevalence of congestive heart failure, cancer, cerebrovascular accident,
chronic obstructive pulmonary disease, hypertension,
smoking, and atherosclerosis.

Incidence of Achilles Tendon Rupture
Overall, the crude incidence of ATR was 3.80 per 10 000
person-years (95% CI, 3.58-4.03). The nonwaitlisted group
had a lower crude incidence (2.91; 95% CI, 2.68-3.16) compared with that of the waitlisted (6.07; 95% CI, 5.35-6.90)
and transplanted groups (5.22; 95% CI, 4.66-5.85) (Table
2). After adjusting for confounders, the incidence was significantly lower in the nonwaitlisted group compared with
the transplanted group (aIRR, 0.44; 95% CI, 0.37-0.53).
The incidence was similar between the waitlisted and transplanted groups (aIRR, 0.94; 95% CI, 0.78-1.12).

Risk Factors for Achilles Tendon Rupture
Nonwhite race, male sex, osteoporosis, and current smoker
status were factors associated with a lower risk of ATR.
Compared with underweight/normal BMI (<25 kg/m2),
higher BMI was associated with ATR in a dose-dependent
manner. Patients with ESRD caused by diabetes mellitus
were at higher risk for ATR compared with those with
ESRD caused by glomerulonephritis. Peripheral vascular
diseases and drug dependence were also risk factors for
ATR (Table 3).

Treatment Modality
Of the 1091 patients who sustained ATR, 282 (26% of all
ATRs) were admitted to the hospital within 14 days after
ATR diagnosis. Of these, 217 patients (77% of admitted
patients) had an ICD-9-CM or CPT code related to ATR,
indicating that the reason for admission was related to the
ATR as opposed to another cause. The nonwaitlisted group

Table of Contents for the Digital Edition of Foot & Ankle International - July 2018