Preoperative Noninvasive Cardiac Testing in Older Adults With Hip Fracture

A Multi-Site Study

Liron Sinvani, MD; Daniel A. Mendelson, MD; Ankita Sharma, DO; Christian N. Nouryan, MA; Joanna S. Fishbein, MPH; Michael G. Qiu, PhD, MD; Roman Zeltser, MD; Amgad N. Makaryus, MD; Gisele P. Wolf-Klein, MD

Disclosures

J Am Geriatr Soc. 2020;68(8):1690-1697. 

In This Article

Results

In the 1,079 patients, mean age was 84.2 years (SD = 8.1); most (75% [n = 807]) were female and white (82%). In terms of marital status, most patients were widowed (43%) or married (36%). The mean CCI was 6.26. The median TTS was 1.1 days (IQR = .8–1.8 days), median LOS was 5.3 days (IQR = 4.2–7.2 days), and in-hospital mortality was 3% (n = 32) (Table 1).

Surgeries were performed most frequently on Fridays (16%), followed by Sundays, Tuesdays, and Wednesdays (each 15%), Mondays and Thursdays (both 14%), and least frequently on Saturdays (11%). Overall, 61% were admitted to tertiary hospitals and 39% to community hospitals. For service category, 54% were admitted to the surgical service and 45% to the medical service. Most (85%) were discharged to skilled nursing facilities, 10% were discharged home, and 3% died (Table 1).

Of those patients who underwent noninvasive cardiac testing, 34.3% (N = 370) had a TTE, .7% (N = 8) had noninvasive stress testing (NST), and none had a dobutamine stress echo. In terms of facility, 47% of patients in community hospitals had a preoperative TTE as opposed to 26% in tertiary centers. The top-three TTE indications were preoperative clearance (54%), abnormal electrocardiogram (17%), and valvular disease (5%) (Figure 1). NST indications were preoperative clearance (n = 6 [75%]), acute coronary syndrome (n = 1 [13%]), and coronary artery disease (n = 1 [13%]).

Figure 1.

Indications for transthoracic echocardiogram before surgery (n = 370). CHF, congestive heart failure; ECG, electrocardiogram.

Patients who had a preoperative TTE (vs those who did not) had a median TTS of 1.55 days versus 1.02 days (P < .01), as well as a median LOS in days of 5.78 versus 5.08 (P = .62).

Factors Associated With Preoperative TTE

The following factors were found to be independently associated with preoperative TTE status: hospital type, admission day, service category, marital status, insurance status, ethnicity, and CCI (P < .0001, .004, <.0001, <.0001, <.0001, .001, and .036, respectively). Age category was also associated with preoperative TTE status at P = .1 level with P = .08. Sex and race were not associated with preoperative TTE status (P > .1) (Supplementary Table S1).

The final multivariable model predicting odds of preoperative TTE included hospital type (tertiary vs community), services (medical vs surgical), marital status, and ethnicity. Specifically, patients in community hospitals had almost three times greater odds of having preoperative TTE performed compared with those cared for at a tertiary hospital (P < .001; 95% CI = 2.22–3.94), after adjusting for service, marital status, and ethnicity status. Patients on the medical service had 3.5 times greater odds of having preoperative TTE performed compared with those on the surgical service (P < .001; 95% CI = 2.60–4.60), after covariate adjustment (Figure 2).

Figure 2.

Multivariate model for factors associated with preoperative transthoracic echocardiogram (TTE) status among hip fracture patients (n = 1,079). CI = confidence interval.

Clinical Outcomes Associated With TTE

Time to Surgery. Results from univariate Cox regression analysis show the following factors were significantly associated with TTS at α = .1 level: TTE status before surgery, sex, hospital type, service category, marital status, insurance status, and CCI (P < .001, P < .001, P = .007, P < .001, P = .005, P = .047, and P < .001, respectively).

The final multivariable extended Cox model predicting TTS included time-varying TTE status before surgery, sex, hospital type, service, CCI, and marital status (P < .001, P = .002, P < .001, P < .001, P = .013, and P = .002, respectively) (Figure 3). Hazard ratios (HRs) greater than 1 suggest a shorter TTS. The results show that having TTE by any given time t before surgery appears to be associated with a decrease in TTS compared with a (similar) subject at the same time t who had not yet had a TTE (HR = 1.3; 95% CI = 1.1–1.5). This finding may suggest that getting a preoperative TTE early in admission compared with later in admission does not cause a delay to procedure, but it may the later the TTE is performed.

Figure 3.

Multivariable extended Cox model to predict time to surgery (TTS) among hip fracture patients (n = 1,079). CI = confidence interval.

In addition, a patient admitted to a tertiary hospital was more likely to have surgery performed sooner compared with a similar patient admitted to a community hospital (HR = 1.3; CI = 1.1–1.5). A patient admitted to the surgical service had shorter TTS compared with a similar patient admitted to the medical service (HR = 1.7; CI = 1.5–1.9), after covariate adjustment). Further, a patient with a higher comorbidity score was associated with a delay to surgery (HR for a 5-point increase in CCI = .80; 95% CI = .67-.96), and a married patient was more likely to have a procedure sooner than a similar patient who was single, separated, divorced, or widowed (Figure 3).

Length of Stay Outcome (Defined as Time Between Surgery and Discharge). Hospital type, service category, insurance category, race, and CCI were each independently associated with LOS (P < .001, P < .001, P < .001, P = .012, and P < .001, respectively). Age, day of the week for admission, and ethnicity were not significantly associated with LOS after surgery (P > .1). TTS was associated with LOS after surgery (P = .01). Results demonstrated that preoperative TTE status was not significantly associated with LOS after surgery (P = .39) univariately or after adjusting for postoperative TTE (P = .65); postoperative TTE was associated with LOS (P < .001).

The final multiple extended Cox regression model included postoperative TTE status, hospital type, service, insurance status, and CCI alongside preoperative TTE (which was forced to remain in the model because it was the main predictor of interest). A patient admitted to a tertiary hospital or to the surgical service was more likely to have longer LOS postsurgery compared with a similar patient admitted to a community hospital or to the medical service, after covariate adjustment (HR = .76; P < .001; and HR = .81; P < .001, respectively). Further, patients with Medicare and Medicaid or private coverage were more likely to be discharged alive sooner compared with patients covered under just Medicaid (HR = .45; P < .001). Finally, patients with a higher comorbidity score were associated with a delay to discharge (HR = .92; P < .001) (Figure 4).

Figure 4.

Multivariate model predicting length of stay (LOS) after surgery among hip fracture patients (n = 1,079). CI = confidence interval.

In-hospital Mortality Outcome. Thirty-two (3%) patients died following hip fracture repair. Results from the logistic regression failed to show an association between having a TTE performed before the procedure and in-hospital mortality (P = .7). However, stay in the intensive care unit and cardiac care unit were each independently associated with in-hospital mortality (P < .001 and P = .042, respectively). The other factors were not significantly associated with in-hospital mortality risk in univariate analyses (P > .1).

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....