The effect of early knee symptom on stages of change for exercise and its characteristic among young college students: application of the transtheoretical model

Abstract

Objectives

To explore the effect of early knee symptoms on the stages of change for regular exercise among young college students and its underlying mechanism using the transtheoretical model (TTM).

Methods

Electronic questionnaires on chronic and frequent knee symptoms, four constructs of TTM for regular exercise, and other variables were designed and distributed in 2022. Valid data was analyzed for 1111 young college student using descriptive statistics, chi-square tests, independent sample t-tests, MANOVA with post-hoc, and logistic regression models.

Results

The prevalence of knee symptoms was 17.37% (95% CI: 15.21 to 19.76). The distribution of exercise stages between students with and without knee symptoms was significantly different (χ² = 13.57, p = 0.01), and the total proportion of action and maintenance in students with knee symptoms (28.50%) was relatively higher than that in those without (19.83%). MANOVA showed ten processes of change, self-efficacy, pros and cons were significantly different across stages in students with knee symptoms (Pillai’s Trace = 0.618, F = 2.516, p < 0.001) and in those without (Pillai’s Trace = 0.317, F = 5.993, p < 0.001), and large effects size were only found for counterconditioning (η² = 0.15), self-liberation (η² = 0.17), self-efficacy (η² = 0.15), and pros (η² = 0.14) among students with knee symptom.

Conclusion

This study implied that early knee symptoms might prompt individuals to engage in and maintain regular exercise behaviors, that specifically enhanced the power of psychological and behavioral variables, especially of self-liberation, counterconditioning, self-efficacy and pros. Insights into the different patterns of constructs of TTM for exercise behavior among students with knee symptoms would be valuable for developing interventions. However, considering the relationship was associational rather than causal, and that reverse causality was possible, that was those who exercised regularly was more aware of their symptoms.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13102-025-01520-9.

Introduction

Knee osteoarthritis (KOA) is one of the most common degenerative diseases leading to disability in middle-aged and elderly individuals worldwide and poses a substantial and increasing health burden [1, 2]. With changes in modern lifestyle, the prevalence KOA has increased indicating an accelerated incidence [3], and a growing trend of knee problems among young people has been reported [4]. A population-based healthcare database in England showed that the annual consultation rate for osteoarthritis in individuals aged 35–44 years increased from 0.3 to 2.0/1000 persons between 2003 and 2010, representing nearly a sevenfold increase [3]. Another study in England found that the prevalence of knee problems among university staff and students aged 18–39 was 31.8% [5]. Similarly, a study in Denmark reported a 23% prevalence of knee or hip pain among individuals aged 29–39 years [6]. The latest data from 17.7 million adults covered by Chinese health insurance indicated that compared to other age groups, the prevalence of KOA increased more significantly in individuals under 35 years of age, with an annual change of approximately 36.6% [7]. Above evidence from studies confirmed and implied that with the changes in modern work, transportation, diet, and other lifestyle factors, joint problems are not limited to middle-aged and elderly; aging is not the sole factor in joint degeneration, and the trend of knee joint problems among young individuals should be paid attention to. The knee joint, being the most important weight-bearing joint in the body, plays a vital role in mobility and quality of life. With increasing life expectancy, protecting joint health and preserving joint function are the cornerstones for healthy aging. Therefore, it is essential to address modifiable risk factors related to knee problems early in life.

Physical activity (PA) is an important but modifiable factor in knee health, with both inactivity and overactivity being risk factors for the development and progression of KOA and knee pain [5, 8–10]. Appropriate levels PA is essential for maintaining joint homeostasis and functions [11, 12]. However physical inactivity (PIA) is prevalent worldwide. Data from the World Health Organization showed the global prevalence of PIA among adults aged 18 years and older was approximately 27.5% in 2016 [13]. PIA is more prevalent among younger adults, particularly in college students. A survey across 23 countries found that the prevalence of PIA among young people aged 16–30 was 41.1% [14]. In Spain, nearly 30% of college students reported never engaging in physical exercise [15]. In contrast, the average physical exercise score of Chinese college students was only 27.4 (± 17.72) out of 100 points [16]. Among young people, there is a strong correlation between PIA and sedentary behavior due to learning and internet use [15, 17]. A survey conducted in 20 countries showed that individuals between the ages of 18–39 were 20% more likely to spend a significant amount of time sitting (540 min or more per day) compared to those between the ages of 40–65 [18]. A recent systematic review suggested that sedentary time has increased over the last 10-year among university students, with self-reported sedentary time averaging 7.29 h per day, and accelerometer-based measurement indicating 9.82 h per day [17]. In China, the mean sedentary time among Chinese college students was 7.51 h per day [19], with 42.65% sitting for over 8 h. Similarly, severe sedentary behavior was observed among Korean undergraduates who sat for an average of 7.96 h daily [20]. Considering the effect of IPA on knee health, improving the PA level among young college students is the key to ensuring knee health. Hence, it is important to analyze the exercise behavior of young people to enhance targeted interventions.

The transtheoretical model (TTM) is a stage-based model for behavior change that incorporates four related constructs considered central to behavior change: stages of change (SOCs), processes of change (POCs), decisional balance, and self-efficacy [21, 22]. TTM has been widely applied to interventions for various behavioral problems, such as smoking cessation, weight control, and exercise adoption [22]. Systematic reviews have synthesized evidence showing that stage-matched interventions can effectively promote behavior change [23, 24]. Wang et al. [25] reported that TTM-based intervention could improve exercise adherence in older adults with KOA in long term and help participants progress through the stages of change [26]. Specifically, the application of TTM led 86.60% of older adults who were in the contemplation or preparation stages to progress to the action stage after 6 months of exercise intervention [27]. As participants progressed through the stages, exercise improved muscle strength [27], walking speed, and joint function [26], which are important for symptom relief and joint function improvement. Similar effectiveness could be found in adolescents and young adults [28, 29]. Research of Gutiérrez-Higuera et al. reported that a 22-week educational program based on TTM significantly increased PA levels in Mexican adolescents, while Huang et al. demonstrated that a stage-matched messages intervention improved PA volume among young Taiwanese female freshmen [28].

In TTM-based exercise research, regular exercise is consistently defined as at least 20 min of exercise per session, performed on at least 3 days per week [30, 31], with the aim of promoting health regardless of PA intensity. Research showed that a considerable proportion of college students were in the ‘lower’ stages of regular exercise, namely precontemplation, contemplation, and preparation stages, while only a small proportion of them were in the more active stages, such as action and maintenance [32]. Previous studies found that 62.5% of college students in Hong Kong were in the ‘lower’ stages with 37.5% in the action or maintenance stages [32]. In Chinese mainland, these figures were 74.9% and 25.1%, respectively, among college students [33], with a similar trend observed among Korean students [34] and Japanese undergraduates [35]. Therefore, the TTM can be used to analyze the psychological and behavioral characteristics of young students regarding exercise behavior, providing valuable insights for further targeted interventions.

Several studies [33–35] whether have described the stages for regular exercise and examined its correlations with various constructs and potential factors among young students, or the prevalence of knee symptoms in them [5]. However, few of these studies have explored whether early knee symptoms would affect exercise behavior among young college students, especially the mechanism by which joint symptoms affect physical exercise. Therefore, in the context of the rapid development of joint symptoms, this study aimed to explore the effect of early knee symptoms on the SOC for regular exercise and its characteristics among young students by application TTM. The core hypotheses were as follows: (1) knee symptoms would hinder exercise behavior and and thereby influence the SOCs. That was, compared with the asymptomatic group, the percentage of students in the action or maintenance stage would be lower in the symptomatic group. (2) The discriminative power of TTM constructs (e.g., pros/self-efficacy) would be stronger in the symptomatic group than in the asymptomatic group. We expected the findings from this study will provide insights into understanding exercise behaviors among young adults and offer valuable evidence for early targeted interventions for joint symptoms.

Methods

Design

This study employed a cross-sectional design.

Participants

In this study, we primarily used convenience sampling from five universities in Zhejiang Province and Beijing, China. Recruitment was conducted via advertisements placed on noticeboards or via electronic posters distributed through student groups. Students who were interested in participating could complete an electronic questionnaire. Additionally, all participants were allowed to share the questionnaire link with their friends or classmates. As remuneration, each participant received a reward ranging from 5 to 10 RMB. To avoid potential selection bias, where symptomatic individuals might be more likely to complete the questionnaire than asymptomatic individuals, we designed the name of the questionnaire to emphasize its focus on PA characteristics but not targeting any special levels of physical activity. All the students who participated voluntarily completed the questionnaire. The inclusion criteria for participants were: (I) students who were ≥ 18 years old; (ii) participants who were students in any colleges nationally. The exclusion criteria were as follows: participants suffered from acute knee injury within the past months or had any surgery which was recommended inactivity by doctors. The questionnaire data were collected between June and November 2022.

Sample size

Power Analysis and Sample Size software (PASS) (version 2008, NCSS Corporation) determined that a sample size of 1060 was required for logistic regression analysis of a binary response variable, with an expected α = 0.05, β = 0.1. Given a non-response rate of 5%, the target sample size was set at 1113.

A total of 1123 questionnaires were completed, of which 1111 were valid after verification of the responses. Twelve questionnaires were excluded due to incorrectly reported ages and some student’s refusal to provide their actual school names.

Variables and instruments

Demographic characteristics of the college students, including sex, age in years, and school year, were collected. Furthermore, body mass index (BMI) was calculated using height and weight and categorized as underweight (< 18.5 kg/m²), normal (18.5–23.9 kg/m²), overweight (24.0–27.9 kg/m²), and obese (≥ 28.0 kg/m²) in accordance with the recommended classifications for Chinese individuals [36].

Knee symptom

A self-reported chronic and frequent knee symptom was defined in this study, which was evaluated using the following question: “Have you often (i.e., three or more times per week) experienced pain, stiffness, or limitation of activity in or around the knees in the past 6 months? “, which was adapted from the item developed by O’Reilly et al. to evaluate the prevalence of knee pain in the population [37]. And similar questions have been widely employed in studies involving participants with early-stage KOA [38–40]. The participants answered: “Neither,” “Yes, only the left,” “Yes, only the right,” “Yes, both”. Those who responded with any affirmative answer were all considered to have knee symptoms. Before our survey, we had invited two clinicians and professors in the field of osteoarthritis to assess whether the question was appropriate in practice and subsequently modified the questions based on their comments, confirming the face validity. We also invited 40 students to the pre-test to verify the suitability, readability and retest reliability within two weeks. The test-retest reliability was 0.98.

Regular exercise

In accordance with the definition used in questionnaires and scales of TTM [30, 31, 41, 42], regular exercise was defined as engaging in at least 20 min of exercise per session, for at least three days per week, with the goal of promoting health, regardless of the intensity of PA.

Stages of change of regular exercise

The SOCs for regular PA was assessed using the Chinese version of a-five items measure [41]. The items are as follows: “I currently do not exercise and do not intend to exercise in the next 6 months” (precontemplation). “I currently do not exercise, but I intend to start to exercise in the next 6 months” (contemplation). “I plan to exercise in the coming month and have prepared for it " or “I currently get some exercise, but not regularly” (preparation). “I currently exercise regularly but for less than 6 months (action).” “I currently exercise regularly and have been doing so for 6 months or longer (maintenance).” From the five alternatives, each participant selected the statement that best presented their current thoughts and behavior.

The exercise processes of change questionnaire

The POCs were measured using the Chinese version of the POC questionnaire [41]. Ten different POCs were identified that are thought to facilitate progression through the stages of change and can be grouped into two high-order factors: experiential and behavioral processes, as thoroughly described in a previous study [41]. As the original questionnaire was designed for workers, two items were deemed unsuitable for college students and were therefore deleted from the Chinese version. In addition, two items with similar meanings in Chinese were combined into one. The Chinese POC questionnaire consisted of 36 items, with four items for each of the ten processes, except for dramatic relief (three items), social liberation (two items) and self-liberation (three items). Participants were asked to rate the frequency of each behavioral occurrence during the past month on a 5-point Likert scale from 1 (never) to 5 (repeatedly). The average of the items in each process was calculated as scores of processes. The Cronbach’s α of the Chinese questionnaire in this study was 0.97, and the Cronbach’s α for individual processes ranged from 0.82 to 0.92.

Self-efficacy to regulate exercise

Self-efficacy refers to people’s beliefs about their ability to achieve specific outcomes [42]. Self-efficacy to regulate exercise was defined as the confidence individuals have in their ability to manage high-risk situations without relapsing into PIA. This was measured using the Chinese version of the Self-Efficacy to Regulate Exercise Questionnaire [43]. This questionnaire consisted of 18 items, and participants used 11-point numeric rating scale to rate the strength of their efficacy beliefs ranging from 0 (“Cannot do”) to 10 (“Highly certain can do”) with 1-unit intervals. The average of the items was calculated as scores of self-efficacy.The Cronbach’s α of the Chinese questionnaire in this study was 0.97.

Decisional balance scale for regular exercise

Decisional balance reflects an individual’s perceived positive (pros) and negative (cons) aspects of regular exercise; it assumes that a person will not adopt or maintain regular exercise unless their pros exceed their cons. In this study, decisional balance for regular exercise was measured using the Chinese version of the Decisional Balance Scale [44]. The 16-item questionnaire consisted of two scales: one measuring the pros (n = 10 items) and the other measuring the cons (n = 6 items) of regular exercise. Respondents rated the importance of each item in their decision to exercise, using a scale from 1 (“not at all agree”) to 5 (“extremely agree”). Scores were summed, and means were computed for each scale, with higher scores reflecting higher perceived pros or cons. The Cronbach’s α of the Chinese questionnaire in this study was 0.92, along with 0.96 and 0.89 for pros and cons respectively.

Statistical analysis

Descriptive statistics (means and standard deviations (SD) for continuous variables and counts and percentages for categorical data) were used to report key variables. Multivariate analysis of variance (MANOVA) was conducted using SOCs as the independent variable and the POCs, self-efficacy, and pros and cons as dependent variables to identify differences across stages for the symptomatic group and the asymptomatic group. The assumption tests for MANOVA showed: (1) Normality or near normality could be met for most variables (except for social liberation, pros) in the symptomatic group with the absolute values of other variables’ Z-score within 1.96; for the asymptomatic group, half of variables, including social liberation, pros, and environmental reevaluation, social liberation, and reinforcement management, exhibited slightly negative skewed distribution, as judged from histograms, skewness, and kurtosis. As noted, the original scores of the TTM variables (processes of change, pros, cons, and self-efficacy) are discrete rather than continuous; therefore, the normality assumption could not be fully satisfied. According to the central limit theorem (CLT) of probability theory, in the case of a large sample size, data that does not follow a normal distribution can also be considered to follow a normal distribution. As recommended by Hair et al., for sample sizes of 200 or more, the detrimental effects of non-normality may be negligible [45]. Hence, for the asymptomatic group (N = 918), the effect of non-normality could be negligible.(2) For the homoscedasticity test, the Levene’s test indicated equal variances for the symptomatic group, except for helping relationships (p = 0.002), stimulus control(p = 0.002) and self-efficacy (p = 0.03). In contrast, robust tests of equality of means (Welch’s test) were applied for the asymptomatic group due to unequal variances, which revealed significant differences in means between stages (all p-values < 0.001). For both groups, the Box’s M test indicated unequal covariance matrices because this method is especially sensitive to the presence of non-normal variables and unequal cell size. The homoscedasticity violation in this study might be attributed to the skewed distribution for some variables. Hair et al. suggest that some MANOVA techniques are robust to assumption violations, particularly with large sample sizes [45]. Therefore, the effect of heteroscedasticity on the results is likely limited in this study. (3) To test linearity (but not multicollinearity), Pearson’s and Spearman’s correlations were used to examine the associations between dependent variables in the symptomatic and asymptomatic groups, respectively. Results indicated significant correlations between variables without multicollinearity: for the asymptomatic group, r ranged from 0.19 to 0.79 (all p-values < 0.001) and variance inflation factors (VIF) ranged from 1.24 to 4.10; for the symptomatic group, r ranged from 0.23 to 0.79 (all p-values < 0.001) and VIF ranged from 1.10 to 3.95. Based on the above three assumption test results, Pillai’s Trace was reported in MANOVA for both groups because this method is considered more robust especially when unequal cell sizes appear, or homogeneity of covariances is violated [45]. The post-hoc tests following MANOVA were Dunnett’s T3 and Tukey’s HSD for the asymptomatic and symptomatic groups, respectively. The effects of stages on the variance of these dependent variables, expressed as proportions of variance accounted for (η²), were evaluated, and small, medium, and large effects were defined as proportions of variance of 0.01, 0.06, and 0.14, respectively [46]. In addition, chi-square test and independent-sample t-tests were used to assess the difference between participants with and without knee symptoms.

To identify variables that distinguish potential stages, multiple logistic regression models (forward selection via Wald) were achieved using SOCs as dependent variables and ten POCs, knee symptoms (no symptoms as reference), age, BMI, sex (male as reference), self-efficacy, and pros and cons as predictors. According to the spiral pattern of the SOC [47], individuals typically do not progress (i.e., moving from a given stage to any subsequent stage, e.g., from precontemplation to contemplation, preparation, action, or maintenance) linearly through the stages, nor do they regress (i.e., moving from a given stage to any prior stage, e.g., from action to precontemplation, contemplation, or preparation) in a uniform manner. Based on the longitudinal shifts from previous studies [48, 49], we analyzed individuals in each stage and the potential stages that they might progress or regress to in the future, and finally set up ten logistic regression models to identify factors that could be used to predict individuals’ potential stages after interventions in the future and to better understand regular exercise behaviors across stages in a long time period.

These models were established based on the following analysis: (1) individuals in precontemplation might progress to contemplation, preparation, action or maintenance four potential stages; (2) individuals in contemplation might progress to preparation, action or maintenance three stages, or regress to precontemplation; (3) individuals in preparation might progress to action or maintenance, or regress to precontemplation and contemplation two potential stages; and (4) individuals in action might progress to maintenance or regress to precontemplation, contemplation, and preparation three potential stages; (5) individuals in maintenance might regress to precontemplation, contemplation, preparation or action four potential stages. In logistic regression models, the former stage was defined as reference. For example, in “Precontemplation/Contemplation” model, we recorded “Precontemplation” as “0” and “Contemplation” as “1”. After deleting duplicate models with identical stages, ten models were retained as shown in Table 4.

Table 4.

Variables that distinguish potential stages of change (N = 1111)

Stages a)	Variables	B	SE	Wald	p-value	OR	PCP
Precontemplation/Contemplation	Counter-conditioning	0.70	0.22	10.51	< 0.001	2.02	67.2%
	Helping relationships	−0.78	0.18	18.37	< 0.001	0.46
	Self-liberation	0.40	0.18	4.89	0.03	1.50
	Self-efficacy	0.13	0.05	6.30	0.01	1.14
	Cons	−0.39	0.14	7.65	0.01	0.68
	Pros	0.62	0.15	16.79	< 0.001	1.85
Precontemplation/Preparation	Environmental reevaluation	−0.50	0.17	8.88	< 0.001	0.61	67.8%
	Counter- conditioning	0.79	0.18	19.31	< 0.001	2.21
	Pros	0.70	0.16	18.61	< 0.001	2.01
Precontemplation/Action	Sex	−1.49	0.29	26.72	< 0.001	0.23	78.9%
	BMI	−0.09	0.04	5.18	0.02	0.91
	Consciousness raising	0.43	0.20	4.64	0.03	1.54
	Dramatic relief	−0.66	0.19	11.59	< 0.001	0.52
	Counter- conditioning	1.85	0.31	35.99	< 0.001	6.35
	Helping relationships	−1.00	0.25	16.34	< 0.001	0.37
	Cons	−0.74	0.18	16.06	< 0.001	0.48
	Pros	0.93	0.22	18.19	< 0.001	2.53
Precontemplation/Maintenance	Sex	−1.35	0.35	15.15	< 0.001	0.26	86.0%
	Social-liberation	−0.74	0.34	4.70	0.03	0.48
	Counter-conditioning	1.86	0.47	15.45	< 0.001	6.41
	Helping relationships	−0.99	0.33	8.77	< 0.001	0.37
	Cons	−0.59	0.20	8.41	< 0.001	0.55
	Pros	1.18	0.28	17.65	< 0.001	3.26
	Self-Efficacy	0.28	0.09	9.51	< 0.001	1.33
Contemplation/Preparation	Consciousness raising	0.32	0.15	4.42	0.04	1.38	65.8%
	Environmental reevaluation	−0.65	0.16	16.11	< 0.001	0.52
	Helping relationships	0.40	0.14	7.79	0.01	1.49
	Cons	0.25	0.12	3.91	0.05	1.28
Contemplation/ Action	Sex	-0.73	0.22	11.05	<0.001	0.48	75.6%
	Environmental reevaluation	-0.67	0.18	14.38	<0.001	0.51
	Counter- conditioning	1.05	0.19	31.60	<0.001	2.85
	Self-efficacy	0.14	0.07	3.93	0.05	1.15
	Cons	-0.35	0.14	6.43	0.01	0.70
Contemplation/ Maintenance	Sex	-1.05	0.29	12.93	<0.001	0.35	85.7%
	Social-liberation	-0.75	0.29	6.80	0.01	0.47
	Counter-conditioning	0.91	0.33	7.76	0.01	2.49
	Reinforcement management	-0.82	0.32	6.68	0.01	0.44
	Self-liberation	0.96	0.36	7.09	0.01	2.60
	Cons	-0.33	0.16	4.43	0.04	0.72
	Pros	0.63	0.25	6.29	0.01	1.87
	Self-Efficacy	0.28	0.09	10.13	<0.001	1.33
Preparation/ Action	Sex	-1.12	0.24	21.83	<0.001	0.33	67.3%
	Counter- conditioning	0.79	0.21	14.84	<0.001	2.20
	Stimulus control	-0.52	0.21	6.31	0.01	0.60
	Cons	-0.47	0.14	10.83	<0.001	0.62
Preparation/ Maintenance	Sex	-1.45	0.31	22.25	<0.001	0.23	79.6%
	Knee symptom	0.94	0.37	6.55	0.01	2.55
	Cons	-0.53	0.17	9.70	<0.001	0.59
	Pros	0.52	0.22	5.73	0.02	1.68
	Self-Efficacy	0.37	0.09	17.39	<0.001	1.45
Action/ Maintenance	Knee symptom	0.67	0.33	4.06	0.04	1.95	70.2%
	Self-efficacy	0.36	0.08	18.25	<0.001	1.43

^a)The first stage was set as reference in each couple stages. PCP Predicted correct percent

A statistical significance level of 0.05 was set for all analyses (two-sided), and pairwise comparisons were performed at an adjusted significance level of 0.005. All data were analyzed using SPSS version 25.0 (IBM Corporation).

Results

General data

The general characteristics of the participants and the differences between students with and without knee symptoms are presented in Table 1. Among the 1111 college students, 193 (17.37% [95% confidence interval [CI]: 15.21% to 19.76%]) had knee symptoms. The proportions of students at the five stages were as follows: 24.84% in precontemplation, 34.74% in contemplation, 18.99% in preparation, 13.50% in action, and 7.92% in maintenance. The total proportion of action and maintenance among students with knee symptoms (28.50%) was higher than that of those without symptoms (19.83%). From precontemplation to maintenance, the proportion of students with knee symptoms was relatively high in action (19.33%) and maintenance (29.54%), followed by 18.84% in precontemplation, 14.50% in contemplation, and 14.22% in preparation. Students with knee symptoms were more likely to be older (t = 2.14, p = 0.03) and had a higher BMI (t = 2.25, p = 0.03), with Cohen’s d value of 0.17, indicating a small effect.

Table 1.

General characteristic of the participants (n/(%))

Characteristic	Total (N = 1111)	No symptom (N = 918)	Knee symptom (N = 193)	χ2/t	p-value
Age, mean (SD) years	20.86(2.63)	20.78(2.57)	21.25(2.87)	2.14 a)	0.03
Sex				2.27 b)	0.13
Male	391(35.20)	314(34.20)	77(39.90)
Female	720(64.80)	604(65.80)	116(60.10)
BMI, mean (SD) kg/m2	20.66(3.46)	20.55(3.37)	21.17(3.83)	2.25 a)	0.03
The stage of change				13.57 b)	0.01
Precontemplation	276(24.84)	224(24.40)	52(26.94)
Contemplation	386(34.74)	330(35.95)	56(29.02)
Preparation	211(18.99)	181(19.72)	30(15.54)
Action	150(13.50)	121(13.18)	29(15.03)
Maintenance	88(7.92)	62(6.75)	26(13.47)

^a)Independent samples t-test and t-value. ^b)Chi-square or Fisher’s Exact tests were used and χ²-value

Differences in processes of change, self-efficacy, and decision balance by stages of change and by knee symptom

Tables 2 and 3 present the means (SDs), results of univariate F-tests along with post-hoc analyses, and independent-sample t-tests for POCs, self-efficacy, and pros and cons across the five stages of exercise behavior in students with and without knee symptoms. MANOVA indicated that ten POCs, self-efficacy, pros, and cons were significantly different across stages in students with knee symptoms (Pillai’s Trace = 0.618, F = 2.516, p < 0.001) and in those without (Pillai’s Trace = 0.317, F = 5.993, p < 0.001). Compared to students without knee symptoms, variances of the 10 POCs, self-efficacy, and pros and cons across SOCs in students with knee symptoms were relatively small, resulting in lower F-values. Among students with knee symptoms, F-values of 10 POCs across stages ranged from 1.01 for dramatic relief to 9.77 for self-liberation, and post-hoc showed significant differences primarily between precontemplation and maintenance; however, no significant differences were observed in consciousness raising, dramatic relief, environmental re-evaluation, helping relationships and stimulus control (all ps > 0.005 for these five POCs). Among students without knee symptoms, students in precontemplation reported using significantly fewer of the 10 POCs than those in the other four stages; the rare significance of 10 POCs was found between the latter four stages, except for consciousness-raising, counterconditioning, and self-liberation. For both groups, no significant differences were observed in cons across the stages (all ps > 0.005).

Table 2.

Differences in processes of change by stages and knee symptom (mean ± SD)

Process	Groups	Change of Stages a)					F	p-value	Post hocb) (p < 0.005)	η2
		PC	C	PR	A	M
Consciousness raising	No symptom	2.77(0.95)	3.09(0.72)	3.24(0.83)	3.41(0.82)	3.48(1.03)	17.01	< 0.001	PC < All; C < A	0.07
	Knee symptom	2.89(1.07)	3.3(0.87)	3.44(0.72)	3.1(0.87)	3.63(1.08)	3.47	0.009	n.s.	0.07
	t	−0.78	−2.03	−1.25	1.75	−0.60
	p-value	0.44	0.04	0.21	0.08	0.55
Dramatic relief	No symptom	2.99(0.94)	3.22(0.80)	3.33(0.91)	3.21(0.99)	3.37(1.05)	4.62	< 0.01	PC < PR	0.02
	Knee symptom	3.11(1.00)	3.36(0.87)	3.03(0.88)	3.07(0.94)	3.37(1.28)	1.01	0.40	n.s.	0.02
	t	−0.82	−1.20	1.68	0.72	−0.02
	p-value	0.42	0.23	0.09	0.47	0.98
Environmental reevaluation	No symptom	2.74(0.96)	3.1(0.70)	3.09(0.84)	3.1(0.86)	3.24(1.09)	8.27	< 0.001	PC < C, PR, A	0.04
	Knee symptom	2.92(1.05)	3.29(0.78)	2.86(0.87)	3.03(0.96)	3.52(1.12)	2.79	0.03	n.s.	0.06
	t	−1.19	−1.78	1.40	0.40	−1.09
	p-value	0.24	0.08	0.16	0.69	0.28
Self reevaluation	No symptom	2.96(0.89)	3.36(0.68)	3.41(0.78)	3.61(0.81)	3.72(1.03)	20.28	< 0.001	PC < All	0.08
	Knee symptom	3.08(0.99)	3.51(0.84)	3.42(0.79)	3.57(0.66)	4.08(0.97)	5.94	< 0.001	PC < M	0.11
	t	−0.84	−1.29	−0.05	−0.28	−1.50
	p-value	0.40	0.20	0.96	0.78	0.14
Social liberation	No symptom	3.19(0.94)	3.51(0.72)	3.62(0.84)	3.73(0.87)	3.67(1.12)	11.41	< 0.001	PC < C, PR, A	0.05
	Knee symptom	3.24(0.99)	3.73(0.81)	3.42(0.97)	3.78(0.59)	4.02(1.08)	4.37	< 0.01	PC < M	0.09
	t	−0.38	−2.10	1.19	−0.33	−1.35
	p-value	0.71	0.04	0.24	0.74	0.18
Counter- conditioning	No symptom	2.74(0.92)	3.20(0.69)	3.39(0.83)	3.63(0.84)	3.72(0.97)	34.42	<0.001	PC< All; C<A	0.13
	Knee symptom	3.00(1.04)	3.35(0.80)	3.38(0.80)	3.66(0.62)	4.13(0.91)	8.18	<0.001	PC, C < M	0.15
	t	-1.79	-1.45	0.06	-0.21	-1.86
	p-value	0.07	0.15	0.95	0.83	0.07
Helping relationships	No symptom	2.70(0.98)	2.90(0.82)	3.13(0.91)	3.12(0.94)	3.27(1.03)	9.41	<0.001	PC<PR, A, M	0.04
	Knee symptom	2.89(1.06)	3.11(0.90)	3.13(0.80)	3.07(0.96)	3.40(1.43)	1.11	0.36	n.s.	0.02
	t	-1.30	-1.79	-0.004	0.25	-0.45
	p-value	0.20	0.08	0.10	0.80	0.66
Reinforcement management	No symptom	3.00(0.91)	3.36(0.69)	3.43(0.79)	3.60(0.83)	3.60(0.90)	15.21	<0.001	PC< All	0.06
	Knee symptom	3.08(0.97)	3.44(0.87)	3.48(0.79)	3.52(0.61)	3.89(1.09)	3.88	0.01	PC<M	0.08
	t	-0.55	-0.69	-0.32	0.50	-1.32
	p-value	0.58	0.49	0.75	0.62	0.19
Self liberation	No symptom	2.86(0.90)	3.32(0.70)	3.48(0.82)	3.7(0.85)	3.78(0.93)	31.80	<0.001	PC< All; C<A, M	0.12
	Knee symptom	3.01(0.99)	3.45(0.85)	3.4(0.85)	3.49(0.61)	4.29(0.81)	9.77	<0.001	All <M	0.17
	t	-1.12	-1.12	0.46	1.21	-2.43
	p-value	0.27	0.27	0.64	0.23	0.02
Stimulus control	No symptom	2.67(0.92)	3.03(0.77)	3.26(0.80)	3.25(0.88)	3.47(0.96)	19.77	<0.001	PC< All	0.08
	Knee symptom	2.93(1.12)	3.32(0.82)	3.13(0.74)	2.99(0.70)	3.51(1.32)	2.22	0.07	n.s.	0.05
	t	-1.73	-2.52	0.86	1.50	-0.13
	p-value	0.09	0.01	0.39	0.14	0.90

^a)PC Precontemplation, C Contemplation, PR Preparation, A Action, M Maintenance, All All other stages

^b)Mean differences were tested using Dunnett T3 and Turkey HSD pairwise comparison for no-symptom and knee-symptom respectively, because the former was with unequal variance (p<0.05)

Table 3.

Differences in self-efficacy and decision balance by SOC and knee symptom (mean ± SD)

Variables	Group	Change of Stages a)					F	p-value	Post hoc b) (p < 0.005)	η2
		PC	C	PR	A	M
Self-efficacy	No symptom	3.7(2.26)	4.55(1.78)	4.87(1.94)	5.53(1.50)	6.23(2.01)	31.31	< 0.001	PC < All; C < A, M; PR < M	0.12
	Knee symptom	4.06(2.55)	4.74(2.07)	4.16(1.50)	4.53(1.51)	6.78(2.04)	8.45	< 0.001	All < M	0.15
	t	−1.02	−0.72	1.92	3.22	−1.18
	p-value	0.31	0.48	0.06	< 0.01	0.24
Pros	No symptom	3.25(0.82)	3.67(0.65)	3.78(0.70)	3.91(0.75)	4.23(0.80)	32.19	< 0.001	PC < M	0.12
	Knee symptom	3.22(0.92)	3.74(0.78)	3.83(0.79)	3.76(0.65)	4.25(0.87)	7.81	< 0.001	PC < All; C, PR < M	0.14
	t	0.25	−0.63	−0.38	1.01	−0.15
	p-value	0.81	0.53	0.70	0.31	0.88
Cons	No symptom	3.08(0.78)	3.06(0.70)	3.24(0.76)	2.94(0.94)	3.02(1.16)	2.94	0.02	n.s.	0.01
	Knee symptom	2.96(0.98)	3.25(0.77)	3.03(0.97)	2.64(0.75)	3.14(1.12)	2.28	0.06	n.s.	0.05
	t	0.90	−1.83	1.35	1.58	−0.43
	p-value	0.37	0.07	0.18	0.12	0.67

^a)PC Precontemplation, C Contemplation, PR Preparation, A Action, M Maintenance, All All other stages

^b)Mean differences were tested using Turkey HSD pairwise comparison except Dunnett T3 for pros in no-symptom and self-efficacy for two groups, because the latter were with unequal variance (p < 0.05)

A larger gradient across the stages was observed in students with knee symptoms compared to those without, reflected in the larger effect sizes (i.e., η²) for self-efficacy, pros, cons and 10 POCs except helping relationships, and stimulus control. Large effect size were found for counterconditioning (η² = 0.15), self-liberation (η² = 0.17), self-efficacy (η² = 0.15), and pros (η² = 0.14) in students with knee symptom, implying that these TTM constructs would become significantly better discriminators for stage progression when knee symptom were present. Notably, both groups had equal medium and small effect sizes for consciousness raising (η² = 0.07) and dramatic relief (η² = 0.02), respectively. A statistical discrepancy with lower F-values but larger η2 was found in symptomatic group was attributed to that η2 was the critical measure of effect magnitude, and the lower F-values likely stemed from the smaller sample size (N = 193) not a weak effect.

For students who were at the same stage, the analysis revealed that contemplators with knee symptoms employed consciousness raising (t=−2.03, p = 0.04), social liberation (t=−2.10, p = 0.04), and stimulus control (t=−2.52, p = 0.01) significantly more frequent than those without symptoms. Additionally, maintainers with knee symptoms used self-liberation (t=−2.43, p = 0.02) significantly more frequently than those without symptoms. The trend chart of POCs across groups presented in Supplementary Fig. 1 showed much differences. Compared to students without knee symptoms, those with knee symptoms used ten POCs more frequently during the precontemplation, contemplation, or maintenance stages (except dramatic relief and stimulus control at the maintenance stage), but used them less frequently in the preparation or action stages (except for consciousness raising at preparation stage). A similar pattern was observed for self-efficacy, as shown in Supplementary Fig. 2, with a significant difference observed at the action stage (t = 3.22, p < 0.01). In both groups, self-reevaluation, counterconditioning, helping relationships, and reinforcement management were used by students at similar frequencies during the preparation or action stages, as shown in Supplementary Fig. 1. Regarding decision balance, pros gradually increased as participants in both groups progressed in each stage, with similar pros across stages, although a relative difference was noted in the action stage. Students with knee symptoms had relatively insignificant lower cons at the action stage (t = 1.58, p = 0.12) but higher cons at the contemplation stage (t=−1.83, p = 0.07) compared to those without symptoms (Supplementary Fig. 2).

Variables that distinguished potential stages

There was no potential multicollinearity among POCs, self-efficacy, pros, cons, with variance inflation factors (VIF) ranging from 1.48 for self-efficacy to 4.14 for counterconditioning. The results of the logistic regression analysis is presented in Table 4. For students in precontemplation, counterconditioning and pros were the most common variables that predicted whether they would progress to subsequent stages, followed by helping relationships and cons. For students in contemplation, cons, counterconditioning and self-efficacy were the relatively common variables that predicted whether participants would progress or regress to potential stages. For students in preparation, environmental reevaluation was the key common variable that predicted whether they would regress to precontemplation or contemplation, while sex and cons were common variables that predicted whether they would progress to action or maintenance; additionally, knee symptoms was a key variable (OR = 2.55, p = 0.01) for predicting progression to maintenance. For students in the action stage, counterconditioning and cons were common variables that predicted whether they would regress to other stages, while knee symptoms (OR = 1.95, p = 0.04) and self-efficacy (OR = 1.02, p < 0.001) were the only two significant predictors that prompted participants to progress to maintenance. For maintenance students, self-fficacy was the most critical key common variable that predicted the individuals regression to any other stage; additionally, sex, cons, pros were the other three common variables that predicted whether they would regress to the three stages of precontemplation, contemplation, or preparation, while knee symptoms were another common variable that predicted whether they would regress to preparation or action.

Discussion

In the context of the rapid development of joint issues, especially knee problems, which is becoming more common among the youth, our study analyzed the effect of early knee symptoms on SOCs for regular exercise behavior and its influencing factors in young college students using the TTM. Although significant differences observed between students with and without knee symptoms were rare, the findings suggest important insights that could be explored in further studies.

Our study revealed that the prevalence of knee symptoms among the Chinese college students who participated was 17.37%, indicating that knee problems affect a significant portion of young people. Although several studies [3, 6, 7] have reported a rapid and increasing prevalence of knee problems among younger adults, few studies have reported this problem in young college students. A study conducted at a university in the United Kingdom reported a 31.8% prevalence of knee problems among staff and students [5], with participants having a mean age of 22.6 years (SD = 5.4), similar to our study population. In addition, a cohort study that followed 1064 women aged 25 years at baseline for ten years reported that 33% of them experienced knee pain [50]. Athough self-reported knee discomfort could not confirmed clinically early joint pathology and there was the likelihood of heterogeneity in symptom etiology, this evidence suggestsed the need for early attention to knee problems in young adults.

In this study, a high percentage (78.57%) of the students were in the ‘lower’ stages (i.e., precontemplation, contemplation, and preparation) of regular exercise, while only 21.42% of the participants engaged in regular exercise or maintained exercise behavior. These findings are consistent with previous studies on mainland Chinese college students (74.9% and 25.1%, respectively) and Korean college students (80.2% and 19.8%, respectively) but differ significantly from those of American college students (23.0% and 77.0%, respectively). Although PA is essential for maintaining homeostasis and joint function [11, 12], PIA is widely prevalent [13], especially among young college students [14, 15], due to their long hours of engagement in learning and Internet use [15, 17]. Hence, targeted interventions and strategies should be implemented to encourage young students to exercise regularly.

A higher percentage of PIA [14, 15] and long hours of sedentary behavior [17, 19, 20] observed in previous studies among college students, along with a high percentage of young students who did not engage in regular exercise in our study, might explain the high prevalence of knee symptoms [5]. Mueller and Maluf proposed a “Physical Stress Theory” that describes how biological tissues adapt to physical stress and concluded that biological tissues exhibit five adaptive responses to physical stress: decreased stress tolerance (e.g., atrophy), maintenance (e.g., homeostasis), increased stress tolerance (e.g., hypertrophy), injury, and death [12]. Based on Physical Stress Theory, knee symptoms might often occur in young college students due to decreased stress tolerance in joints resulting from prolonged inactivity. Our research revealed that the proportion of students in the more active stages (i.e., action and maintenance) of knee symptoms was relatively higher than that in those without symptoms (28.50% versus 19.83%). In addition, across the total students’ population, the proportions of students with knee symptom were relatively higher in the action (19.33%) and maintenance (29.54%) stages compared to any ‘lower’ stage. There are two possible explanations for the relatively high proportion of students in the active stage of exercise in the symptom group and for the high proportion of students with knee symptoms across the stages of action and maintenance. The first one is that long-term regular exercise might cause knee symptoms owing to a mismatch between joint tolerance and physical stress during PA. Prolonged inactivity before the students begin an exercise can lead to an atrophic and hypofunctional state of the joint tissue, reducing tolerance to mechanical forces; thus, joint tissues may become prone to injury during activities [11, 12], and those who have engaged in regular exercise, whether in the action or the maintenance stage, might develop knee symptoms resulting from chronic injury due to mismatch stress. The second one is that students with knee symptoms may engage in exercise aiming at relieving symptoms. Though inactivity is recommended for acute onset of knee pain [51], exercise therapy is considered one of the primary non-pharmacological treatments for KOA and other chronic musculoskeletal pain conditions [52, 53], and exercise is also recommended for relieving chronic knee pain [51]. The logistic regression findings highlighted that knee symptoms significantly predicted the progression to the maintenance stage among individuals in the preparation stage (OR = 2.55) and action stage (OR = 1.95), respectively. Combined with the relatively large η² values for TTM constructs, these fingdings implied knee symptoms acted as a powerful motivational trigger for long-term adherence of regular exercise and therapeutic exercise mechanism mignt be the primary driver that knee symptoms prompted the regular exercise in young students.

The factors distinguishing the stages in the logistic models were the core factors for developing and implementing interventions to encourage and maintain exercise behavior. Our study developed ten models to identify individuals’ potential stages after the intervention in future. A few previous studies have developed models for adjacent stages [34, 35]. However, in practice, atypical patterns of the SOCs often occur from precontemplation to maintenance [47]. Previous studies have proved that individuals do not progress or regerss step by step [49, 54]. Hence, multiple models would be necessary to identify potential candidate targets for personalized interventions. Knee symptoms was a factor in both preparation/maintenance and action/maintenance model. When individuals at preparation and action stage progressed to the maintenance stage, those with knee symptoms were 2.5 times and nearly twice as likely to progress as those without repectively. This suggests that regular exercise might be a possible method that young students adopt to relieve and manage knee symptoms, and symptoms may prompt them to engage in exercise and maintain it for a long time. Knee pain is the most common reason for seeking medical care [39], and exercise therapy is the first-line therapy for chronic symptoms [51]. Therefore, paying attention to knee discomfort among young adults is crucial, and essential support should be provided to those who experience knee discomfort.

There was a significant difference in the constructs of the TTM for regular exercise behavior between students with and without knee symptoms. First, compared to those without knee symptoms, students with knee symptoms showed relatively small variance across the stages in 10 POCs, self-efficacy, pros, and cons. Significant differences across stages were primarily found between the lowest stage (i.e., precontemplation) and the most active stage (i.e., maintenance), with no significant difference in consciousness raising, dramatic relief, environmental re-evaluation, helping relationships, stimulus control, and cons between these two stages. This implies that knee symptoms might encourage individuals to seek information through personal actions and environmental events, resulting in fewer significant differences between stages. Second, from contemplation to maintenance, that is, during the period from the time when individuals intend to exercise to the time they could exercise regularly for a long time, students with knee symptoms used POCs less frequently, had low self-efficacy and pros, but reported the lowest cons when they started exercising or engaged in regular exercise (i.e., preparation and action). This may be due to the high perceived cons during the contemplation and maintenance stage. On the one hand, knee symptoms may improve contemplators’ intention to engage in exercise to relieve symptoms and enhance joint function; hence, they may improve their self-cognition, seek beneficial information from their social environment, and gain confidence in starting exercise. Meanwhile, individuals in contemplation may also experience much cons to exercise, as knee symptoms can limit their free motion and cause distress. Therefore, targeted exercise guidance may be helpful for contemplator with knee symptoms. On the other hand, maintaining regular exercis over the long term is particularly challenging for individuals with knee symptoms, requiring substantial effort and self-discipline. Individuals in the maintenance stage may experience a high level of cons, and to maintain regular exercise, they need to improve their cognition and find more beneficial information from the social environment again to further improve self-efficacy. Hence, for maintainers with knee symptoms, multiple interventions and effective support are essential to prevent a relapse from exercising to not exercising (i.e., precontemplation), especially through strategies such as consciousness raising and stimulus control. Third, relatively large effects of stages on the adoption of regular exercise and a significant gradient across stages were observed in students with knee symptoms, with the effects of counterconditioning, self-liberation, self-efficacy, and perceived pros being particularly noticeable. These variables may be crucial in motivating students with knee symptoms to progress through the stages, and tailored interventions should be considered.

The conclusion drawn from the above findings is that knee symptoms indeed influence the psychological and behavioral key variables of regular exercise in young college students, which might be the potential mechanism by which joint symptoms affect physical exercise. Although knee pain and diagnosis of KOA cannot be equated, a strong association between knee pain and radiographic osteoarthritis has been proven [39, 55], and self-reported knee pain has been used as an indicator of KOA. As a chronic condition, KOA requires prevention and comprehensive care models across a continuum from early to late stages [56]. Hence, identifying risk factors and developing comprehensive care for early symptoms in young individuals is valuable, as it can potentially prevent individuals from many years of discomfort [56]. For students, engaging in exercise is affected by substantial factors, such as self-efficacy, pros and cons of exercise, and the experimental and behavioral strategies for obtaining related information, i.e., the four constructs of the TTM [35, 57], motivation and enjoyment [54], and knowledge [33]. Our study suggests that knee health could be another factor that explains the mechanism of exercise behavior.

Notably, this study provides insights into understanding the effect of knee symptoms on exercise behavior and its underlying mechanism by applying the TTM, and offers some valuable inspiration for further intervention. However, several limitations should be improved in the future study. First, due to the inherent limitation of the cross-sectional study design in establishing causality, the inability to determine the temporal sequence (i.e., whether knee symptoms preceded the initiation of regular exercise or vice versa among students was limited and further clarification was warranted. This could be achieved by collecting detailed symptom-related data at the time of participant recruitment or by adopting a longitudinal study design. Second, while our study provided several insights into the differences in the constructs of TTM for exercise behavior in students with and without knee symptoms, these findings should be confirmed by further studies. Finally, although our study could help raise awareness of knee health in youths, a self-reported symptom definition may capture non-pathological pain, which may influence the strength and specificity of observed associations.

Conclusion

Attention should be paid to the knee health of young students. This study implied that early knee symptoms acted as a motivational cue that specifically enhanced the power of psychological and behavioral variables, especially of self-liberation, counterconditioning, self-efficacy and pros, in driving long-term exercise adherencemight prompt individuals to engage in and maintain regular exercise behaviors. Gaining an understanding of the different patterns of TTM constructs related to exercise behavior in students with knee symptoms would be valuable for the development of appropriate interventions. However, considering the relationship was associational rather than causal, and that reverse causality was possible, that was those who exercised regularly was more aware of their symptoms. The exact mechanism would be further explored in future research.

Authors’ contributions

All authors were involved in revising the article, and all authors approved the final version to be published. Study conception and design: LMW. Data collection: LMW, QQL, HYQC. Data analysis and interpretation: HYQC, LMW. Draft of the manuscript: HYQC, LMW.

Funding

This work was supported by the Provincial Health Science and Technology Plan of Zhejiang Province (No. 2023RC038) and the Traditional Chinese Medicine Science Plan of Zhejiang Province (No. 2023ZL033).

Data availability

All the data used during the study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

In accordance with the Declaration of Helsinki, the study was conducted after being approved by Zhejiang Chinese Medical University Biomedical Ethics Committee (IRB20220607-8) and electronic informed consent was obtained from all participants involved in the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.