Follow-up visits are a vital part of orthopaedic postoperative care to monitor patient progress and modify treatment plans to maximize patient recovery. Missed appointments and noncompliance are challenging topics that prevent the surgeon from adequately addressing postoperative complications and have been associated with poorer outcomes including worse function, greater pain, strain on the patient-physician relationship, higher costs, and delays in recognition of complications such as surgical site infection or fracture nonunion.1–14

Previous studies have examined factors associated with patients with orthopaedic trauma missing follow-up appointments at urban level 1 trauma centers with conflicting results regarding distance.1,2 Influence of distance on attendance to follow-up visits in a more rural setting has yet to be investigated. Orthopaedic care in nonurban environments is already limited, and populations trend toward higher poverty rates and lower general educational achievement, which have also been associated with poor outcomes.15

The purpose of this study was to investigate the relationship between orthopaedic patient attendance at postoperative follow-up visits and distance to care at a level 1 trauma center in a midsized city serving a largely rural population crossing several state lines. Of secondary interest are other variables that may affect patient compliance, such as the patient's age, demographics, insurance, pathology, mechanism of pathology, and surgical intervention.

Methods

We conducted a retrospective chart review of all adult patients who underwent surgery between January 2, 2019, and December 23, 2019, by a single fellowship-trained orthopaedic surgeon. All surgeries took place at a single level 1 trauma center. Patients who required several (≥ 2) procedures over the course of the year were excluded because they represent a unique demographic unrepresentative of the general population. Patients were also excluded if they did not have a follow-up appointment scheduled.

At our institution, patients are commonly scheduled for their first postoperative appointment at their preoperative history and physical appointment for elective procedures and at the time of discharge for patients with trauma. Patients who have a cellphone listed in the electronic medical record (EMR) are sent a reminder text message before their appointment and a prearrival survey. If the patient does not complete the prearrival survey, the patient receives a second reminder text message 24 hours before his or her appointment. If the patient does not have a cellphone number in the EMR or if they are a new patient, he or she will also get a letter in the mail listing the appointment date and time. After a patient no-shows an appointment, the nursing staff makes an attempt to call and reschedule the patient and a letter is mailed as an attempt to reschedule the patient if they do not answer the phone call.

Data collected included age, sex, patient-reported race, and ethnicity as recorded in the EMR, pathology, pathology mechanism, surgical intervention, insurance type, return visit dates, and if the patient attended the appointments. We collected patients' addresses at the time of surgery. Driving distance to care and driving time to care were calculated using Google Maps Application Programming Interface between each patient's address and the clinic address. Of note, the clinic, emergency center, surgical center, and hospital are all on the same campus.

All statistics were done using Stata (StataCorp. 2013; Stata Statistical Software: Release 13: StataCorp LP). Univariate logistic regression was done to obtain an odds ratio (OR) representing the effect of various preoperative factors on likelihood of patients missing one of their first two scheduled follow-up visits after orthopaedic surgery. Univariate logistic regression was also done for missing the first appointment and again for missing the second appointment. Multivariate logistic regression was then done to identify predictors of patients missing one of their first two scheduled follow-ups using purposeful selection with P-values <0.200.16–18P-values of <0.05 were considered significant.

Results

We identified 518 patients with demographics, which is summarized in Table 1. The average age was 48.7 ± 17.1 years (range 18.1 to 91.2). Our population was 55% men (n = 287). Our population was predominately White, non-Hispanic/Latino (n = 315, 61%) followed by White Hispanic or Latino (n = 172, 33%), Black/African American (n = 26, 5%), and Native American (n = 2) with three patients without race or ethnicity data. Most of the patients had either private insurance (n = 128, 25%), Medicare (n = 115, 22%), or Medicaid (n = 113, 22%). We additionally had 57 patients (11%) on workers' compensation (WC), 22 (4%) covered by accident insurance, 28 (5%) Veteran's Administration, and 55 (11%) self-pay. More than half of the pathologies were traumatic in nature/mechanism (n = 291, 56%). The average distance from care was 70 ± 132 miles (range 1.3 to 1,365). We had 196 (37%) of our patients living more than 50 miles and 140 (27%) living more than 100 miles from the clinic. The average time traveled to care was 69 ± 118 minutes (range 5 to 1,216).

Table 1 - Demographics by Attendance Demographic Total n = 518 Attended n = 439 Missed n = 79 P Age (years) 48.7 ± 17.2 49.6 ± 17.0 43.6 ± 1.9 0.001 Male 287 (55%) 236 (54%) 51 (65%) 0.075 Race 0.008  White 484 (93%) 416 (95%) 68 (86%)  Black/African American 26 (5%) 16 (4%) 10 (13%)  Native American 2 (<1%) 2 (<1%) 0  No Response 6 (1%) 5 (1%) 1 (1%) Ethnicity 0.461  Not Hispanic or Latino 333 (64%) 281 (64%) 52 (66%)  Hispanic or Latino 174 (34%) 150 (34%) 24 (30%)  No response 11 (2%) 8 (2%) 3 (4%) Insurance <0.001  Private 128 (25%) 115 (26%) 13 (16%)  Veteran's Administration (VA) 28 (5%) 26 (6%) 2 (3%)  Medicare 115 (22%) 105 (24%) 10 (13%)  Medicaid 113 (22%) 82 (19%) 31 (39%)  Accident coverage 22 (4%) 18 (4%) 4 (5%)  Workers' compensation (WC) 57 (11%) 55 (13%) 2 (3%)  Self-pay 55 (11%) 38 (9%) 17 (22%) Mechanism of pathology  Trauma 291 (56%) 240 (55%) 51 (65%) 0.103 Distance (miles) 70 ± 132 61 ± 108 120 ± 217 <0.001  Within city limits 265 (51%) 230 (52%) 35 (44%)  <30 295 (57%) 258 (59%) 37 (47%)  30-49 27 (5%) 23 (5%) 4 (5%)  50-99 56 (11%) 52 (12%) 4 (5%)  100-150 82 (16%) 63 (15%) 18 (23%)  >150 58 (11%) 42 (10%) 16 (20%)  Time (minutes) 69 ± 118 61 ± 97 115 ± 194 <0.001

P-values in this table are based on the t-test and chi-squared. Bolded P-values are considered significant.

Thirty-two patients (6%) did not attend their first scheduled follow-up appointment. An additional 47 patients (10%) did not attend their second scheduled follow-up for a total of 79 patients (15%) who did not attend one of their appointments.

Figures 1–3 present univariate logistic regression for missing the first, second, or either postoperative follow-up appointment, respectively. The significance of age, ethnicity, insurance, traumatic mechanism, and distance did not change based on first, second, or either follow-up appointment. Male sex was only notable for missing the second follow-up appointment. Being Black or African American race was only notable for missing the second or either follow-up appointment. Both patients of Native American heritage attended their postoperative visits.

Figure 1:

Graph showing univariate logistic regression of each variable for missing the first appointment. Bolded odds ratios are considered significant with a P-value <0.05.

Figure 2:

Graph showing univariate logistic regression of each variable for missing the second appointment. Bolded odds ratios are considered significant with a P-value <0.05.

Figure 3: Graph showing univariate logistic regression of each variable for missing either appointment. Bolded odds ratios are considered significant with a P-value P-values are given in Table 2. All variables with a P-value Figure 7.

Each minute of travel time and each mile of distance from care increased the risk of missing one of the first two postoperative visits. An increase in 10 minutes of travel time led to an OR 1.03 (P = 0.001). In other words, every 10 minutes was a 3% increase in the odds of missing one of the appointments. An increase in 10 miles of distance led to an OR 1.02 (P = 0.001). In other words, every 10 miles was a 2% increase in the odds of missing one of the appointments. A failure curve of missing one of the first two appointments by distance from clinic was constructed, which is shown in Figure 4. The risk of missing an appointment became significant at 50 miles with an OR 1.65 (P = 0.042) but continued to increase at 70 miles (OR 1.92, P = 0.008), 100 miles (OR 2.37, P = 0.001), and 150 miles (OR 2.40, P = 0.007).

Figure 4:

Graph showing the failure curve of missing appointment by distance.

Figures 5–7 present multivariate logistic regression using purposeful selection with P-values <0.200 for missing the first, second, or either postoperative follow-up appointment, respectively. The significance of age, ethnicity, traumatic mechanism, and distance did not change based on first, second, or either follow-up appointment. Male sex and Black or African American race were only notable for missing the second or either follow-up appointment. Self-pay and Medicaid insurance were notable for missing the first, second, and either postoperative follow-ups, while WC was only notable for the second or either follow-up.

Figure 5:

Graph showing the final multivariate logistic regression model using purposeful selection for missing first appointment. Bolded odds ratios are considered significant with a P-value <0.05. Reference variables are as follows: † private insurance and ‡ distance <70 mi.

Figure 6:

Graph showing the final multivariate logistic regression model using purposeful selection for missing second appointment. Bolded odds ratios are considered significant with a P-value <0.05. Reference variables are as follows: α female, β White, † private insurance, and ‡ distance <70 mi.

Figure 7: Graph showing the final multivariate logistic regression model using purposeful selection. Bolded odds ratios are considered significant with a P-value P-values are given in Table 2.

Comparison of the univariate and multivariate values for the first, second, and either follow-up visit is presented in Tables 2–4, respectively.

Table 2 - Univariate and Multivariate Logistic Regression for Missing First Appointment Variable OR* (CI) P OR P Age 0.97 (0.95-0.99) 0.008 Male 1.58 (0.75-3.35) 0.225 Race 0.081  White Ref  Black/African American 3.08 (0.99-9.56) 0.052  Native American Perfect attendance Ethnicity 0.995  Not Hispanic or Latino Ref  Hispanic or Latino 1.00 (0.47-2.13) 0.995 Insurance 0.003  Private Ref Ref  Self-pay 7.09 (1.80-27.87) 0.005 6.33 (1.60-25.10) 0.009  Veteran's Administration 1.54 (0.15-15.41) 0.712 1.57 (0.16-15.76) 0.701  Workers' compensation 0.74 (0.08-7.31) 0.800 0.67 (0.07-6.62) 0.731  Accident coverage 1.98 (0.20-19.99) 0.561 1.56 (0.15-15.98) 0.708  Medicare 1.50 (0.33-6.86) 0.600 1.55 (0.34-7.09) 0.573  Medicaid 5.89 (1.65-21.08) 0.006 6.08 (1.69-21.85) 0.006 Mechanism of pathology  Trauma 1.53 (0.72-3.24) 0.261 Distance (per 10 miles) 1.02 (1.01-1.04) 0.003  >70 miles 2.06 (1.01-4.22) 0.049 2.10 (1.01-4.44) 0.042

OR* (CI) = Univariate Odds Ratios found in Figures 1-3, OR = Multivariate Odds Ratios found in Figures 5-7.

Bolded P-values are considered significant.

Table 3 - Univariate and Multivariate Logistic Regression for Missing Second Appointment Variable OR* (CI) P OR P Age 0.98 (0.96-0.99) 0.001 Male 1.70 (1.01-2.84) 0.041 1.84 (1.05-3.25) 0.035 Race 0.007  White Ref Ref  Black/African American 3.47 (1.49-8.13) 0.004 2.54 (1.01-6.40) 0.048  Native American Perfect attendance Perfect attendance Ethnicity  Not Hispanic or Latino Ref  Hispanic or Latino 0.90 (0.53-1.54) 0.713 Insurance <0.001  Private Ref Ref  Self-pay 3.96 (1.76-8.90) 0.001 3.10 (1.34-7.14) 0.008  Veteran's Administration 0.68 (0.14-3.20) 0.626 0.59 (0.12-2.84) 0.512  Workers' compensation 0.32 (0.07-1.47) 0.800 0.24 (0.05-1.10) 0.049  Accident coverage 1.97 (0.58-6.70) 0.280 1.38 (0.40-4.84) 0.612  Medicare 0.66 (0.26-1.66) 0.378 0.70 (0.28-1.77) 0.451  Medicaid 2.91 (1.43-5.96) 0.003 2.66 (1.26-5.61) 0.010 Mechanism of pathology  Trauma 1.53 (0.91-2.55) 0.100 Distance (per 10 miles) 1.02 (1.01-1.04) 0.001  >70 miles 1.95 (1.18-3.21) 0.009 1.99 (1.16-3.41) 0.012

OR* (CI) = Univariate Odds Ratios found in Figures 1-3, OR = Multivariate Odds Ratios found in Figures 5-7.

Bolded P-values are considered significant.

Table 4 - Univariate and Multivariate Logistic Regression for Missing Either Appointment Variable OR* (CI) P OR P Age 0.98 (0.96-0.99) 0.004 Male 1.57 (0.95-2.58) 0.073 1.74 (1.01-3.01) 0.047 Race 0.003  White Ref Ref  Black/African American 3.82 (1.67-8.77) 0.002 2.78 (1.12-6.91) 0.027  Native American Perfect attendance Perfect attendance Ethnicity 0.461  Not Hispanic or Latino Ref  Hispanic or Latino 0.86 (0.51-1.46) 0.585 Insurance <0.001  Private Ref Ref  Veteran's Administration 0.68 (0.14-3.20) 0.626 0.60 (0.13-2.89) 0.528  Medicare 0.84 (0.35-2.00) 0.698 0.90 (0.37-2.16) 0.806  Medicaid 3.34 (1.65-6.78) 0.001 3.05 (1.46-6.37) 0.003  Accident coverage 1.97 (0.58-6.70) 0.041 1.38 (0.40-4.85) 0.608  Workers' compensation 0.32 (0.07-1.48) 0.144 0.23 (0.05-0.96) 0.045  Self-pay 3.96 (1.76-8.90) 0.001 3.12 (1.35-7.21) 0.008 Mechanism of pathology  Trauma 1.51 (0.92-2.48) 0.100 Distance (per 10 miles) 1.02 (1.01-1.04) 0.001  >70 miles 1.92 (1.18-3.12) 0.008 2.02 (1.19-3.42) 0.009

OR* (CI) = Univariate Odds Ratios found in Figures 1-3, OR = Multivariate Odds Ratios found in Figures 5-7.

Bolded P-values are considered significant.

The final multivariate logistic regression model for missing either appointment consisted of male sex (OR 1.74, P = 0.047), Black or African American race (OR 2.78, P = 0.027), self-pay (OR 3.12, P = 0.008), WC (OR 0.23, P = 0.045), Medicaid (OR 3.05, P = 0.003), and traveling more than 70 miles to clinic (OR 2.02, P = 0.009).

Discussion

Patient attendance at follow-up appointments can be a major obstacle in improving patient outcomes and sustainability of health care. The recovery timeline varies from patient to patient, and adjustments to postoperative visits are made accordingly. Orthopaedic postoperative appointments often require suture removal, changes in wound management, review of pain management, changes in mobility or weight-bearing status, adjustments to physical therapy protocols, and addressing of patient concerns.19 Missing appointments can result in detrimental effects on patient outcomes and need for additional visits or procedures.5,6,20 Existing literature approximates that 27 to 33% of orthopaedic patients miss their first postoperative visit and that 70% of patients will miss at least one appointment in the first 6 months.1,2,8,13 By comparison, our study found that 6% of patients did not attend their first scheduled follow-up appointment, an additional 10% missed their second follow-up appointment, and a total of 15% did not attend one of their appointments. To address this, many clinics will overbook appointments to retain productivity in the setting of missed visits; however, there are increased opportunity costs to patients through prolonged queues in clinic and decreased patient satisfaction, which may further exacerbate future no-show rates.21,22

Distance and Travel Time

Distance from care is thought to have a negative effect on patient care and follow-up across many healthcare specialties.20,23,24 Despite this, few studies have cited distance as a risk factor for orthopaedic surgery patients in not attending their follow-up appointments.1,