Sonographic screening of distance runners for development of future Achilles and patellar tendon pain

Abstract

Objective:

The two primary aims of this study were to identify ultrasonographic tendon abnormalities in asymptomatic runners and to examine the likelihood of developing pain in runners with ultrasound abnormalities compared to those without abnormalities.

Design:

Longitudinal, prospective cohort study

Setting:

2019 Salt Lake City Marathon

Participants:

Recreational half- and full marathon runners

Assessment of Risk Factors:

The Achilles and patellar tendons of asymptomatic runners were examined with ultrasound imaging prior to a running event. Runners were monitored for self-reported outcomes of pain in the examined tendons at 1, 3, 6 and 12 months after the event.

Main Outcome Measures:

Development of pain based on presence of asymptomatic tendon abnormalities.

Results:

138 runners (36.2 ± 12.0 years of age, 49.3% males, 31.2% full marathon runners) were included. Ultrasound abnormalities of the Achilles and patellar tendons were identified in 24.6% and 39.1% of the runners before the race, respectively. Ultrasound abnormalities were significantly associated with about a three-fold increase (hazard ratio = 2.55, p = 0.004) in the hazard of developing pain in the Achilles tendon and patellar tendon (hazard ratio = 1.67, p = 0.042) over the year following the race. Positive and negative predictive values of developing pain over the year were 34.1% and 87.2%, respectively, for abnormal findings in the Achilles tendon, and 22.9% and 85.0%, respectively, for the patellar tendon.

Conclusion:

Presence of ultrasonographic abnormalities is associated with increased development of pain in the Achilles and patellar tendons within one year of a marathon or half-marathon.

Introduction

Patellar and Achilles tendinopathy are common injuries in athletes^1–8 and non-athletes^9,10; specifically, they are frequent in runners^3,4,6,7. They both clinically present with pain and structural change leading to decreased function. Both can result in significant disability and a prolonged recovery period. For patellar tendinopathy, approximately 30% of affected individuals are unable to return to full sport participation within 6 months.¹¹ In some populations, patellar tendinopathy is cause for retirement from sport in over 50% of affected athletes.¹² Achilles tendinopathy has a lifetime prevalence of 7.5% in high school runners,⁶ and an overall incidence in runners of approximately 10%, with prevalence ranging from 6.2% to 9.5% reported in a systematic review.¹³

Tendinopathies to these two tendons are believed to be related to altered loading patterns from fatigue and biomechanical changes.^14–16 Ultrasonography provides a rapid method to assess for tendon structural abnormalities, and is a non-invasive, rapid, low-cost, radiation-free instrument. It has repeatedly been used in assessment of tendinopathic pain, with similar sensitivity to magnetic resonance imaging (MRI).^17–21 Alteration in loading or other factors result in structural tendon changes, such as increased tendon thickness, hypoechoic changes, paratenon changes, neovascularization and others, which can be assessed with ultrasound imaging.^15,18,22,23 These tendon structural changes are thought to either represent a pathologic disease spectrum, a form of active adaptation to loading, or an incidental asymptomatic finding without association with loading or subsequent injury.^24,25 Additionally, although similar studies have been performed on a variety of other sports,²⁶ it remains unclear whether structural findings carry a different prognosis for runners. Investigations on this topic would benefit both clinicians and runners by promoting early diagnosis and intervention for future injuries.

The authors published a previous study with a similar cohort,²⁷ but was limited by poor follow-up. While maintaining a high followup rate, the primary aims of this study were: 1) to understand the prevalence and type of ultrasonographic abnormalities in half-marathon and full marathon runners and 2) to examine if half-marathon and marathon runners who have pre-race ultrasonographic abnormalities are more likely to progress to pain in the abnormal-appearing tendon over the following year, compared to those who did not have these abnormalities. We also aimed to assess the positive (PPV) and negative predictive values (NPV) of this method as a screening test for development of tendon pain over one year.

Methods

Study Design:

Runners 18 years of age or over who were signed up to participate in the 2019 Salt Lake City marathon or half-marathon were recruited at a booth at the race’s pre-race exposition; no prior notifications were made. All subjects reported no current or recent pain (in the last week) in their patellar or Achilles tendons. Subjects with prior rupture (partial or complete) of, or surgery to, these tendons were excluded. Age, sex, height (m), weight (kg), smoking history, and running history (years running, average miles run per week in the past year) were obtained for each participant. As marathons have been demonstrated to acutely induce ultrasonographic change in runners,^28–30 ultrasound images were collected prior to subjects running the race. To verify completion of the race, an email was sent to participants two days after the race. As decided prior to study initiation, all runners who did not respond at this timepoint were excluded from the study. This was decided a priori as prior data suggest asymptomatic sonographic changes in runners are unlikely to preclude pain within a month of the scan.²⁷ Subjects were then contacted at specific timepoints (1, 3, 6, and 12 months after the race) via email to complete an online self-reported injury survey. The follow-up surveys described the locations, both verbally and pictorially, of the Achilles and patellar tendons for the subjects, as well as collecting a VISA-A³¹ and VISA-P³² for symptomatic subjects (see supplemental data files A–C for surveys). Pilot data, performed on 15 recreational runners, demonstrated that 100% of subjects were able to accurately identify their patellar and Achilles tendons based on these instructions (gauged by experienced sports medicine physicians).

Ultrasound image acquisition:

Evaluation of the patellar and Achilles tendons was performed with a 15-MHz linear transducer (Sonosite X-porte). Images were saved as static images without the subject’s information. While scanning, the sonographer was not actively identifying areas of abnormalities (i.e. they did not make note of abnormalities, to maintain blinding), and the subjects were not informed of potential abnormalities. To limit inter-rater variability, this was performed by only one of the study physicians with several years of experience with musculoskeletal ultrasound, having performed over 400 scans of patellar and Achilles tendons, who was blinded to the subjects’ information. The patellar tendons were examined with the runner in a supine position, with the quadriceps muscle activated so the tendon just became taut. Runners were placed prone with the feet hanging over the edge of the table with the ankles passively flexed to 90° for evaluation of the Achilles tendons. Transverse images were saved at the tendon location at its greatest width while longitudinal assessment was made in the midline tendon, centered over the area of maximum thickness, making sure to include the patellar attachment.

Ultrasound image analysis:

At a later time-point, all previously-stored tendon images (transverse and longitudinal) were reviewed by an experienced sonographer in a blinded manner. Previously described measurement techniques and definitions were used to standardize measurements.^27,33–35 Each imaged tendon was classified in a binary manner based on the sonographic appearance: normal or abnormal. Abnormal findings were subcategorized into the following²⁷: 1) focal hypoechogenicity, 2) paratenon blurring³⁴ (indistinct posterior aspect of tendon with focal thickening), 3) presence of intratendinous calcification (punctate hyperechoic area with acoustic shadowing), 4) presence of an osteophyte at the enthesis, and 5) tendon thickening (defined as barrel-shaped thickening greater than 0.5mm in the Achilles or greater than 1.0mm in the patellar tendon as compared to the normal distal portion³⁶). Inter-rater reliability for image analysis was tested on a randomly-selected subset of subjects’ images (25% of the full cohort for both tendons) by a second experienced musculoskeletal ultrasonographer; the resulting kappa value for the measurements was 0.66 (95% CI 0.48–0.83), considered substantial agreement.³⁷

Data Analysis:

The primary outcome variable for the study was the development of pain after the race in the patellar or Achilles tendon. Descriptive statistics were computed (demographics, pre-race ultrasound findings, and post-race Achilles and patellar tendon pain). Because of missing data from non-survey responses, runners’ demographics were compared across the time points (baseline and each follow-up) using a one-way analysis of variance for continuous variables and Pearson’s χ² test for categorical variables, in order to ensure that the samples from the different time points did not significantly differ, and that the samples were derived from the same population. A two-by-two contingency table analysis was performed to examine the association between ultrasound abnormality and development of pain in the Achilles and patellar tendons immediately after the race, and at 1, 3, 6, and 12 months post-race. In addition, the association of the specific type of ultrasound abnormality to post-race pain at any time point was examined using a two-by-two contingency table analysis. Data from both right and left tendons were obtained from each runner and used for the analysis. As expected, post-race pain between the right and left sides in both Achilles and patellar tendons were highly correlated. In order to account for the dependence of observations within runners, significance tests for the two-by-two contingency table analysis above, including the calculations of a relative risk (RR) and its 95% confidence interval (CI), were performed, using appropriate generalized linear models.^38,39 In particular, the models were constructed, specifying the binomial distribution and log link function, along with calculating cluster-robust standard errors^40,41 with the cluster being each runner. Further, PPV and NPV were calculated from a two-by-two contingency table comparing ultrasound abnormalities to the development of post-race pain at any point in the same tendon. Additionally, associations between type of ultrasound abnormality and pain in the Achilles and patellar tendons were examined using a two-by-five contingency table analysis with Fisher’s exact test. As a multivariate analysis, Cox proportional hazards (PH) models were used to examine the risk of developing pain in the Achilles and patellar tendons after the race by presence of ultrasound abnormality, while using gender, age, years of running, and average miles per week of training over a year as covariates. A hazard ratio (HR) was calculated and a survival curve was constructed to plot the proportion of runners with injury-free intervals, separately for the Achilles and patellar tendons. Similar to the contingency table analysis above, cluster-robust standard errors^40,41 were used to adjust for the correlated observations between the right and left tendons. The assumption of PH was tested based on Schoenfeld residuals⁴² and reestimation.⁴³ There was a possibility that the PH assumption was not met in the Achilles tendon model for a covariate, average miles per week of training. Similarly, in the Cox PH model for the patellar tendon, ultrasound abnormality showed a potentially non-proportional hazards. Consequently, each of these covariates was entered in the Cox PH model as a time-varying covariate to account for non-proportional hazards.^44,45 Adjustments for missing data points during statistical analysis were treated as follows: Non-survey responses on post-race pain at each time point were treated as missing data; whereas, for the variable on pain at any time point, non-survey responses were treated as absence of pain.

Power analysis:

Based on prior data, proportions of runners with abnormal and normal ultrasound findings developing pain in the Achilles tendon were 23.5% and 7.5%, respectively.²⁷ These percentages for the patellar tendon were 34.4% and 13.6%. With an alpha level of 0.05, a power of 0.80, and a two-sided test, we estimated required sample sizes of 160 and 130 for the Achilles and patellar tendons, respectively. Since right and left tendons were to be evaluated from each runner, the finalized, total samples sizes needed for this study were calculated as 80 and 65 for the Achilles and patellar tendons, respectively.

Ethical Considerations

This study was reviewed and approved by the Institutional Review Board at the primary author’s institution (IRB #00098658) and all subjects provided informed consent.

Results

A total of 166 runners were scanned (28 were immediately excluded for not responding to the two-day post-race email), of which 138 runners were included and followed up to 1 year after the race. Survey response rates were: 100.0% (immediately after the race), 96.4% (1 month post-race), 92.8% (3 months post-race), 89.1% (6 months post-race), and 81.2% (1 year post-race). Demographic information of the subjects is summarized in Table 1.

Table 1:

Demographic information of the 138 runners included in the study.

		N=138
Variable		Mean (SD)	f (%)
Age		36.2 (12.0)
BMI		23.7 (3.2)
Years of running		11.5 (10.5)
Weekly running mileage (mi)		19.8 (11.7)
Number of half marathons		6.9 (8.6)
Number of full marathons		3.1 (6.3)
Gender	Male		68 (49.3%)
	Female		70 (50.7%)
Race	Half marathon		95 (68.8%)
	Full marathon		43 (31.2%)
Completed race	Yes		135 (97.8%)
	No		3 (2.2%)

Overall, 24.6% and 39.1% of the runners had pre-race ultrasound abnormalities in the Achilles and patellar tendons, respectively, with 5.1% and 11.6 % of the runners having bilateral abnormalities. At any point after the race, 15.9% and 17.0% of the runners developed self-reported pain in the Achilles and patellar tendons, respectively. There were no significant differences in the runners’ demographics across the time points (baseline and each survey follow-up; p > 0.05), ensuring that the loss of follow-up introduced little bias to the analysis results. Frequencies of ultrasound abnormality in the Achilles and patellar tendons, and RR of developing pain at each interval time point are described in Table 2. The PPV and NPV for the Achilles tendon to develop pain at any point over the year were 34.1% and 87.2%, respectively. For the patellar tendon, those values were 22.9% and 85.0%. Sensitivities/specificities were 31.8%/88.4% for the Achilles, and 34.0%/76.4% for the patellar tendons, respectively.

Table 2:

Risk of developing pain in the affected tendon, based on the finding of pre-race ultrasound (US) abnormalities.

Relative risk (RR)		N = 138 with N=276 tendons (left & right)
			Post-race pain [f (%)]
Follow-up time	Tendon	Pre-race US abnormality	Yes	No	RR (95% CI)	p *
Immediately after race	Achilles	Yes	5 (12.2)	36 (87.8)	1.69 (0.75, 3.79)	0.207
		No	17 (7.2)	218 (92.8)
	Patellar	Yes	3 (4.3)	67 (95.7)	0.49 (0.11, 2.24)	0.357
		No	18 (8.7)	188 (91.3)
1 month post-race	Achilles	Yes	4 (10.0)	36 (90.0)	2.83 (0.94, 8.48)	0.064
		No	8 (3.5)	218 (96.5)
	Patellar	Yes	5 (7.4)	63 (92.6)	1.46 (0.59, 3.58)	0.413
		No	10 (5.1)	188 (94.9)
3 months post-race	Achilles	Yes	3 (7.5)	37 (92.5)	2.31 (0.76, 7.02)	0.138
		No	7 (3.2)	209 (96.8)
	Patellar	Yes	4 (6.1)	62 (93.9)	3.84 (1.10, 13.41)	0.035
		No	3 (1.6)	187 (98.4)
6 months post-race	Achilles	Yes	7 (19.4)	29 (80.6)	8.17 (2.77, 24.06)	< 0.001
		No	5 (2.4)	205 (97.6)
	Patellar	Yes	5 (8.2)	56 (91.8)	2.17 (0.73, 6.45)	0.165
		No	7 (3.8)	178 (96.2)
1 year post-race	Achilles	Yes	3 (8.8)	31 (91.2)	3.35 (0.98, 11.50)	0.054
		No	5 (2.6)	185 (97.4)
	Patellar	Yes	8 (14.3)	48 (85.7)	4.80 (1.59, 14.50)	0.005
		No	5 (3.0)	163 (97.0)
Any time point	Achilles	Yes	14 (34.2)	27 (65.8)	2.67 (1.57, 4.57)	< 0.001
		No	30 (12.8)	205 (87.2)
	Patellar	Yes	16 (22.9)	54 (77.1)	1.52 (0.88, 2.62)	0.134
		No	31 (15.1)	175 (84.9)

^*From generalized linear model (binomial distribution with log link function) with cluster-robust standard errors.

The association of type of ultrasound abnormality to pain in the Achilles and patellar tendons at any time point are summarized in Table 3. No significant relationship was identified between type of ultrasound abnormality and development of pain in either tendon (p > 0.05).

Table 3:

Specific ultrasound findings for all tendons (right and left Achilles and patellar tendons, n = 552 tendons [138 runners]), and development of pain over the following year. Values represented as n (%), where n refers to the number of tendons with that particular abnormality, and the percentage relates to the proportion of that particular tendon abnormality that did or did not develop pain.

		Hypoechogenic focus	Paratenon blurring	Intratendinous calcification	Osteophyte	Thickening
Achilles	Develop pain	12 (38.7%)	2 (40%)	1 (16.7%)	1 (14.3%)	11 (50%)
	Do not develop	19 (61.3%)	3 (60%)	6 (83.3%)	6 (85.7%)	11 (50%)
Patellar	Develop pain	8 (16.7%)	0 (0%)	8 (28.6%)	4 (26.7%)	7 (17.5%)
	Do not develop	40 (83.3%)	1 (100%)	20 (71.4%)	11 (73.3%)	33 (82.5%)

Survival estimates (pain-free period) for the Achilles and patellar tendons by ultrasound findings derived from the Cox PH models are shown in Table 4 and Figures 1 and 2. The Cox PH model revealed that the presence of an ultrasound abnormality was significantly associated with about a three-fold increase (HR = 2.55 [95% CI = 1.35, 4.81], p = 0.004) in the hazard of developing pain in the Achilles tendon after the race, adjusting for gender, age, type of race, years of running, and miles of running per week. In addition, females were significantly less likely to develop pain (HR = 0.40 [95% CI = 0.18, 0.87], p = 0.021).

Table 4:

Results of Cox Proportional Hazards model, examining the development of pain for each tendon, with covariates included.

Tendon	Predictor		HR (95% CI)	z	p
Achilles	Ultrasound abnormality (reference: no)	Yes	2.55 (1.35, 4.81)	2.90	0.004
	Gender (reference: male)	Female	0.40 (0.18, 0.87)	−2.31	0.021
	Age		1.00 (0.98, 1.03)	0.33	0.738
	Race (reference: half marathon)	Full marathon	1.76 (0.87, 3.53)	1.58	0.114
	Years of running		1.00 (0.96, 1.04)	0.01	0.996
	Miles of running per week*		1.01 (0.99, 1.03)	0.91	0.365
Patellar	Ultrasound abnormality (reference: no)*	Yes	1.67 (1.02, 2.73)	2.04	0.042
	Gender (reference: male)	Female	0.80 (0.39, 1.63)	−0.62	0.537
	Age		0.99 (0.95, 1.02)	−0.73	0.465
	Race (reference: half marathon)	Full marathon	0.76 (0.36, 1.60)	−0.72	0.474
	Years of running		0.99 (0.95, 1.04)	−0.43	0.667
	Miles of running per week		0.99 (0.96, 1.02)	−0.61	0.542

HR = hazard ratio.

^*Included as a time-varying covariate to account for non-proportional hazards in Cox proportional hazards model.

Figure 1:

Survival curve for Achilles tendon pain. The vertical axis represents the proportion of athletes who did not develop pain in the Achilles tendon. The horizontal axis reflects each survey time-point following the race. Immediate refers to pain within the first two days following the race.

Figure 2:

Survival curve for patellar tendon pain. The vertical axis represents the proportion of athletes who did not develop pain in the patellar tendon. The horizontal axis reflects each survey time-point following the race. Immediate refers to pain within the first two days following the race.

There was also a statistically significant association between the presence of an ultrasound abnormality and developing pain in the patellar tendon (HR = 1.67 [95% CI = 1.02, 2.73], p = 0.042), adjusting for the same covariates above. However, the effect size was much smaller and its lower-bound 95% CI was close to 1.00, potentially indicating a clinically nonsignificant effect on the patellar tendon. Severity of symptoms (VISA-A and VISA-P values) in symptomatic runners are demonstrated in table 5.

Table 5:

Mean pain, VISA-A, and VISA-P scores in symptomatic subjects at each follow-up time point. No subjects had Achilles or patellar tendon pain on the day before the race.

Time point	1 month	3 month	6 month	1 year
Mean pain score (0–10), Achilles	3.6 ± 0.9	3.1 ± 0.8	3.3 ± 1.0	4.2 ± 1.0
Mean pain score (0–10), patellar	2.9 ± 1.0	2.1 ± 0.5	3.1 ± 1.0	2.9 ± 1.0
Mean VISA-A	54.2 ± 7.3	55.9 ± 12.1	54.9 ± 16.6	58.8 ± 9.9
Mean VISA-P	45.8 ± 10.5	46.1 ± 8.5	46.5 ± 11.9	51.2 ± 9.7

Discussion

This study demonstrates that asymptomatic Achilles and patellar tendon abnormalities in runners are associated with a higher risk of developing pain in that tendon in the year following a half-marathon or marathon. Prospective studies in a variety of sports have found conflicting results on the clinical importance of asymptomatic structural tendon findings. According to a systematic review of prospective studies conducted by McAuliffe et al. in 2016, ultrasonographic tendon abnormalities are associated with future patellar (RR = 4.35) and Achilles (RR = 7.33) tendon pain in numerous sports.⁴⁶ Overall, the highest specific structural abnormalities associated with subsequent injury were presence of neovascularization,^47,48 tendon thickening^34,36,49, and hypoechoic tendon abnormalities.^36,50,51 Most of the sports studied primarily included jumping-landing-pivoting sports, and thus represent a considerably different biomechanical loading pattern than in pure distance running, which consists of lower forces in a repetitive cyclic loading pattern. Tendon changes, including changes in echogenicity and thickening, have been found to change asymptomatically with loading in runners,^52–54 with increased running training loads,⁵⁵ and more years running.⁵⁶ Runners specifically appear to have more tendon collagen disorganization, increased tendon thickness, and reduced ultrasonographic echogenicity of tendons,⁵⁷ which may be a normal tissue adaptation to running load. Thicker tendons have not been found to be significantly associated with increased risk of pain on the limited prospective studies on runners available.^29,47

Prior prospective studies looking at runners are scarce, and have focused on the Achilles tendon, with limited prospective data available for the patellar tendon (see table 6).^27,29,47 These studies suggest that an ultrasonographic evaluation has an ability to predict pain in the Achilles tendon (and potentially in the patellar tendon, to a lesser extent) at certain time-points during the year after the race.

Table 6 –

prior studies examining the patellar and Achilles tendons of runners to subsequent development of pain.

Study	Number subjects; tendons examined	Findings	Study limitations
Cushman27	104 runners; Achilles and patellar tendons	Runners with Achilles abnormalities on ultrasound were 3.14 times (p = 0.010) more likely to develop pain over the following year. Patellar tendon abnormalities were similar with a relative risk of 2.52 (p = 0.008).	Included pre-existing tendon pain, low response rate (53.6% at 6 months), large confidence intervals
Ooi29	21 runners, 20 controls; Achilles tendon	Within 3 days of a marathon, there was a significant reduction in tendon stiffness, as measured by compression elastography. Reduced tendon stiffness was associated with increased risk of development of pain (p = 0.016) at 4–6 weeks post-race.	Smaller sample size, no significant B-mode findings (likely due to sample size)
Hirschmuller47	427 runners; Achilles tendon	Runners contacted 6 and 12 months after a race – neovascularization (OR = 6.9 [2.6–18.8], p = 0.0001) was the only predictive factor for future pain.	Response rate (67%), recorded after race (but with 2hr rest period), potential recall bais

Similar to the aforementioned running studies, our study also finds the NPV to outperform the PPV. For the Achilles tendon, the statistically significant findings suggest that those with a pre-race normal ultrasound are less likely to develop pain within the next year. The patellar tendon, though less robust in its statistical significance, also demonstrated similar findings – a relatively low PPV compared to the relatively high NPV. We did not identify any specific type of ultrasonographic pre-race abnormality that was more likely to develop pain at a later time-point. Asymptomatic ultrasound abnormalities have a relatively low positive predictive value; hence, having an abnormality does not mean a runner will surely develop pain. The prevalence of asymptomatic abnormalities identified in this study fit within the range seen within other sports, though it is quite variable, ranging from 11–52%.^{26,33,36,49–51,59–62} It should be noted that we did not examine neovascularization, which has been demonstrated to be a key abnormality that may predict future tendon pain.^26,47,58

Interestingly, the development of Achilles tendon pain was correlated with male gender. Despite the smaller effect sizes compared with the association with the presence of an ultrasound abnormality, these findings are somewhat different from those reported in prior studies. Studies looking at Achilles and patellar tendinopathy in marathon runners have found an association with age⁶³, as well as with using a training schedule, compression socks, or having a prior Achilles injury.⁶⁴ Studies looking at Achilles and patellar tendinopathy in masters track and field runners found no influence of age, gender, or weight.^65,66 More studies are needed to conclude whether any of the runners’ demographics variables can be used to aid in the prediction of future symptomatic pain, with the presence of ultrasonographic abnormalities.

Our study has limitations. First, self-reports were used to collect the data on the development of pain in the area of the tendon. We attempted to minimize bias introduced by self-report of pain by: 1. performing a pilot study of our pictorial and verbal descriptions to ensure patients could accurately assess location, and 2. we collected VISA-A and VISA-P data. Second, objective evaluations of tendons were performed only prior to the race, and no objective physical examinations or imaging of tendon pathology was performed at follow-up. Since no physical examination was performed for the participants reporting pain on follow-up, a possibility exists that pain in the reported region was due to abnormalities in adjacent structures. Third, although participants were not told of the results of the scan, there is a possibility of participant suggestion bias for reporting pain in the previously scanned area due to participation in the study. Fourth, although we were able to limit the loss to follow-up, there were still some missing data, which was most prominent at 12 months after the race. Despite the loss of follow up, the subjects’ demographic variables were comparable between pre-race and any time point at post-race. Further, ultrasonographic evaluation of tendon neovascularization was not performed. Although tendon neovascularization assessment has been shown to be a reliable additional method for evaluating tendon pathology^46,67, the trade-off for time to perform neovascularization assessments was considered against the decrease in number of subjects able to be recruited and scanned at the time of the race. Though it would seem that neovascularization measurement time would be negligible, pilot data suggested that addition of this test would reduce total sample size by over 25% (increasing the time-per-subject from 3 minutes to 4 minutes). Fifth, although runners tend to reduce their running leading up to a race, runners could have had altered tendon appearance if they had been running significantly leading up to the race;^28,29,68 their pre-race training load (or cross-training regimen) was not ascertained. Similarly, subjects may have been more likely to sign up if they had prior patellar or Achilles tendon injuries (more than a week out), which could have affected selection. Lastly, a single experienced ultrasonographer performed the studies and a single experienced ultrasonographer blindly interpreted the studies, which could limit generalizability, but did allow for consistency. In general, intra-rater and inter-rater reliability of tendon ultrasonographic evaluation has been shown to be high,²⁵ and a second experienced ultrasonographer reviewed a subset of the images and tested the inter-rated reliability, with substantial agreement. The dynamic nature of ultrasound does limit the ability to accurately assess static (saved) ultrasound images, and may introduce additional bias. After our study was performed, more recent literature suggests that thickness 30mm from the insertion may have been better than our measurement of maximal thickness.⁵⁷

Despite the limitations above, the current study has strengths that need to be highlighted. A large number of runners were examined using ultrasonography by a single clinician with ample musculoskeletal ultrasound experience. These factors contributed to the consistency of the findings and provided an appropriate statistical power for the data analysis. The subjects were half- and full marathon runners in a marathon event, making the study findings extremely applicable to the target population of interest. The subject response rates were high for a longitudinal study of one year’s duration. Our study has demonstrated feasibility of performing ultrasonographic evaluations for a large number of subjects and following up on them in a longitudinal (1 year) fashion, with minimal subject attrition. Lastly, although PPVs for the Achilles and patellar tendons were low in this study (34.1% and 22.9%, respectively), relatively high NPVs for these tendons (≥ 85.0%) indicate that ultrasonographic evaluations could provide reassurance of the low possibility of developing pain in these tendons for someone with normal sonographic tendon structure. This is an important attribute for a screening test, and the current study has shown that ultrasonographic evaluation could function as a screening test for future pain in the Achilles and patellar tendons.

In conclusion, this study offers additional evidence that asymptomatic abnormal sonographic tendon structure increases the risk of the subsequent development of Achilles tendon pain in runners over the year following the race. Evaluation of the patellar tendon reached statistical significance but may not have been clinically significant. Male runners and those running the full marathon, as opposed to the half-marathon, were more likely to develop Achilles tendon pain. The PPV and NPV for the sonographic evaluation described in this study were 34.1% and 87.2% at the Achilles tendon.