Profiles of exercise adherence in late pregnancy: A latent profile analysis among Chinese women.

HeMa1

HuinaChen1

LifenOuyang2

ShuruXu1

KeqingLi1

ZhaomeiXie1

DongmeiDuan3

ChenhuiZhou1✉Phone86- 15916889908Emailchenhuizh6@126.com

1School of NursingGuangdong Medical University523808DongguanGuangdongChina

2Shunde Women and Children’s Hospital of Guangdong Medical University528300ShundeGuangdongChina

3Dongguan maternal and child health care hospital523110DongguanGuangdongChina

He Ma¹, Huina Chen¹, Lifen Ouyang², Shuru Xu¹, Keqing Li¹, Zhaomei Xie¹, Dongmei Duan³, Chenhui Zhou^1*

¹School of Nursing, Guangdong Medical University, Dongguan, Guangdong, 523808, China

²Shunde Women and Children's Hospital of Guangdong Medical University, Shunde, Guangdong, 528300, China

³Dongguan maternal and child health care hospital, Dongguan, Guangdong, 523110, China

*Corresponding author at: Chenhui Zhou, School of Nursing, Guangdong Medical University, Dongguan, Guangdong, 523808, China, Phone: 86-15916889908, e-mail address: chenhuizh6@126.com

Abstract

Background

The benefit of physical activity and exercise interventions for pregnant women crucially depends on adherence. The study aimed to identify latent categories of exercise adherence among pregnant women in the third trimester and explored the influence of these distinct profiles.

Methods

Convenience sampling was used to recruit participants from three maternal and child health hospitals in China, between November 2024 and June 2025. Latent profile analysis (LPA) was used to identify potential classes of exercise adherence among pregnant women in the third trimester; multinomial logistic regression was used to explore the factors associated with these profiles.

Results

A total of 531 participants were included in this study and were classified as high (n = 59), moderate (n = 239), and low adherence (n = 233) to exercise. Compared to the low-adherence group, the influencing factors for the high-adherence group were weekly exercise habit, pelvic floor muscle training (PFMT), exercise self-efficacy, social support, enjoyment of exercise, other commitments, lack of time, and fatigue. Compared to the mid-adherence group, the influencing factors for the high-level group were education level, weekly exercise habit, exercise self-efficacy, social support, lack of time, and fatigue.

Conclusion

Exercise adherence among pregnant women appeared heterogeneous and presented in three distinct categories. Health workers should develop targeted interventions based on the socio-psychological characteristics of pregnant women in the third trimester to improve their adherence to exercise.

Keywords:

pregnant women

exercise adherence

exercise self-efficacy

social support

latent profile analysis

Background

Pregnancy is a physiologically and psychologically unique period that is crucial for the health of pregnant women and their offspring. Regular physical activity and exercise during pregnancy are widely recognized for enhancing maternal well-being and optimizing offspring development [1]. Although the World Health Organization (WHO) recommends that all women with healthy pregnancies engage in at least 150 minutes of moderate-intensity aerobic exercise per week [2], adherence remains a primary determinant of the effectiveness of exercise during pregnancy. Evidence shows that exercise adherence to recommended guidelines is often low, thereby limiting the benefits that could be gained [3].

Previous studies have indicated that physical inactivity during pregnancy increased the risk of excessive gestational weight gain, gestational diabetes mellitus, pre-eclampsia, gestational hypertension, macrosomia, instrumental delivery, urinary incontinence, and depressive disorders [4, 5]. Sarno et al. [6] recruited women with a singleton, uneventful pregnancy during their third trimester; 37% of them undertook sports/exercise activities. A systematic review of 11,323 Chinese women during pregnancy reported that 21.0% of them met the recommended level of exercise [7]. A cross-sectional study involving 1,636 Chinese pregnant women found that the prevalence of physical inactivity was 47.5%, and that walking was the most common form of physical activity [8].

The benefits of physical activity and exercise for both mother and their child are dependent on long-term adherence. Walasik et al. [9] examined physical activity patterns among 9000 pregnant women in Poland. During the first and second trimesters, 90% of participants exercised, whereas in the last pregnancy, almost 13% of respondents discontinued physical activity. Kókai et al. [10] investigated the effectiveness of two 8-week app-based moderate to vigorous physical activity interventions for pregnant women. This study began in October 2021 with 663 participants. At week 21, 254 women discontinued this intervention. Shang et al. [11] reported that less than half of the women undertook 150 minutes of exercise every week before pregnancy.

40.5% of participants kept regular exercise according to the guidelines during pregnancy.

Adherence, defined by the WHO, is the extent to which a person’s behavior complies with agreed recommendations from health care professionals [12]. Exercise adherence is a complex and dynamic phenomenon during pregnancy, influenced by demographic, psychological, and environmental factors [11]. Previous studies employed a variable-centered approach to categorize pregnant women based on whether their physical activity and exercise met the standards recommended for exercise [6]. Researchers classified participants into “exercisers” and “non-exercisers” based on whether they performed recommended physical activity to assess adherence to physical activity and exercise [3, 6, 10, 13]. Apparently, these analysis methods appeared inadequate for individualized intervention. Knowledge regarding reasons for adherence to exercise during pregnancy remains limited.

Exercise adherence, along with reasons for both adherence and non-adherence, was assessed by adopting the Exercise Adherence Rating Scale (EARS) [14], a reliable and valid self-reported outcome measure consisting of 6 items assessing adherence to exercise and 10 items evaluating reasons for exercise adherence [15]. EARS has been translated and used in several countries, including Denmark [17], Brazil [18], and Sweden [16]. In the current study, we evaluated adherence to physical activity among low-risk pregnant women by adapting a Chinese version of the EARS.

A firm understanding of person-centered factors that influence physical activity during pregnancy is crucial for developing effective individualized exercise programs. We further applied latent LPA, a method particularly suited for identifying individual latent characteristics from a person-centered perspective. The results could facilitate the development of tailored exercise interventions for pregnant women.

To our knowledge, no study has used LPA to examine patterns of exercise adherence among pregnant women. The present study, therefore, applied LPA to delineate adherence profiles in pregnant women from three Guangdong cities—Shenzhen, Dongguan, and Shunde (Foshan)—and to identify related factors of these profiles.

Methods

Study design and participants

An observational study was conducted from November 2024 to June 2025 at three maternal and child health hospitals in Shenzhen, Dongguan, and Shunde, Guangdong Province, China. We used convenience sampling to recruit pregnant women in their third trimester of pregnancy. The inclusion criteria: aged ≥ 18 years and voluntarily participating in the study. Exclusion criteria included: aged < 18 years; significant chronic conditions that could affect Exercise during pregnancy (e.g., pre-eclampsia, cervical insufficiency, unexplained persistent vaginal bleeding); mental disorders; or unwillingness to participate.

Six trained researchers collected the data in person. After confirming eligibility, the researchers explained the study’s purpose, risks, and benefits to the participants. Those who met the inclusion criteria completed the questionnaire in real-time, thereby minimizing the likelihood of invalid responses.

Sampling method and sample size

According to the literature[17], a minimum sample size of 250 to 500 participants is required for LPA.

Hence, this study employed the LPA method and adhered to established guidelines for sample size calculations.

We conveniently recruited 531 participants from 558 eligible pregnant women.

Measures

A generic questionnaire was developed based on a review of relevant literature and the study’s objectives. This questionnaire collected information on various demographic factors, including age, pre-pregnancy body mass index (BMI), residence, education level, personal monthly income, and other relevant variables. Among these, pre-pregnancy BMI was calculated as body weight divided by the square of height. Using the Chinese classification, women with a BMI < 18.5 kg/m² are considered underweight, those with a BMI of 18.5 ~ 23.9 kg/m² as normal weight, those with a BMI of 24 ~ 27.9 kg/m² as overweight, and those with a BMI ≥ 28 kg/ m² as obese [18]. The obstetric parameters included history of gravidity and parity.

Exercise adherence

The exercise adherence was assessed using the exercise adherence rating scale (EARS) developed by Newman in 2017 [14]. This scale consists of three parts: the EARS-A scale consists of 6 qualitative questions to provide information about their adherence behaviour for individuals; the EARS-B scale consists of 6 items evaluating adherence to prescribed home exercise, with items 1, 3 and 5 are reverse scored; the EARS-C scale consists of 10 questions regarding the reasons for adherence or non-adherence, with items 3, 7, 8 and 10 are reverse scored [14, 15]. Translated and validated across several countries, it has shown good reliability and validity (Cronbach’s α = 0.77 to 0.94) [16, 19, 20]. The Chinese version of the EARS was modified by Wu Yuxuan [15]. The total score range of the EARS-B scale is between 0 and 16, with higher scores indicating better adherence. This study utilized the EARS-B scale to evaluate exercise adherence in pregnant women and the EARS-C scale to identify its influencing factors. In this study, Cronbach’s alpha of the scale was 0.817.

Pregnancy exercise social support

The present study used the Physical Activity Social Support Scale (PASSS) to measure social support for physical activity [21]. The PASSS has 24 items. Items are scored on a 5-point Likert scale, with one indicating “strongly disagree” and five indicating “strongly agree.” The total scores range from 24 to 120, with higher scores indicating greater social support for physical activity. The reported Cronbach’s α was 0.95 [21]. The Cronbach’s α of the PASSS was 0.967 in the present study.

Pregnancy exercise self-efficacy

Prenatal exercise self-efficacy was measured using the Pregnancy Exercise Self-Efficacy Scale (P-ESES). Kroll et al. devised an Exercise Self-Efficacy Scale for spinal cord injury in 2007 [22]. Bland et al. adapted and validated it among pregnant women in 2013, with a Cronbach’s α of 0.838 [23]. The Chinese version of the P-SESE was modified by Yang Hongmei et al., Cronbach’s α = 0.804. The P-ESES includes 10 items, which are divided into three domains: overcoming exercise barriers, emotional barriers, and support barriers. Items are scored on a 5-point Likert scale, with one indicating “strongly disagree” and five indicating “strongly agree.” The total P-ESES score ranges from 10 to 50. A total P-ESES score < 20 indicates a low level of exercise self-efficacy, a score 21 to 40 suggests a moderate level, and a score >40 is regarded as indicative of a high level of exercise self-efficacy. In this study, Cronbach’s alpha of the scale was 0.963.

Data analysis

Descriptive analysis

SPSS software version 27.0 was used for statistical analysis. Necessary normality tests were performed with kurtosis and skewness − 3.29 to 3.29 [24]. In the study, descriptive statistics were made with the SPSS 27.0 package program. Demographic characteristics and scale scores now explicitly include calculated values of N (%), mean (±), and standard deviation (SD) to improve clarity and interpretability.

Latent profile analysis

Mplus software, version 8.3, was used to analyze the latent profiles of the pregnancy exercise self-efficacy. The best model was selected based on model fit indices and clinical significance. The fit index included the Akaike information criterion (AIC), Bayesian information criterion (BIC), adjusted BIC (aBIC), entropy, Lo-Mendell-Rubin likelihood ratio test (LMR), and Bootstrap-based likelihood ratio test (BLRT). Lower AIC, BIC, and aBIC values indicate a better model fit. Entropy, ranging from 0 to 1, reflects classification accuracy, with values closer to 1 indicating higher accuracy. The LMR and BLRT were used to compare models with k and (k-1) classes; p < 0.05 suggests that the k-class model fits the data significantly better than the (k-1)-class model [25].

Ethical considerations

This study was conducted in accordance with the guidelines of the Declaration of Helsinki.

It was approved by the Ethics Committee of the Dongguan Maternal and Child Health Hospital (Approval No. 2024 − 155); Shunde Women and Children's Hospital of Guangdong Medical University (KY-2024-070); and Baoan Central Hospital of Shenzhen (BYL20240628). Before the study, both oral and written consents were obtained from all eligible participants, ensuring that participation or non-participation would not affect their work performance or future employment opportunities. Additionally, all collected information would be anonymous and de-identified. Furthermore, participants were informed that they could withdraw from the study at any time.

Results

Demographic information

A total of 558 questionnaires were distributed in this study. Among them, 27 questionnaires had incomplete responses and were deemed invalid; therefore, they were eliminated. Complete responses were considered valid, resulting in a total of 531 valid questionnaires. The effective response rate of the questionnaire was 95.16%. All participants were in the third trimester. Most resided in urban areas and had a mean age of 31.81 ± 3.99 years. Only 23.4% achieved 150 minutes or more of exercise per week. The participants' average total PASSS score was 34.47 ± 8.10, and that of P-ESES was also 34.47 ± 8.10. Other demographic information was detailed in Table 1.

Table 2
presents the average scores for each item of the EARS-C scale (reasons for adherence / non-adherence). Statistical results indicate that most items have high F-values and p < 0.05, demonstrating statistically significant differences among exercise adherence levels.
Table 1 Socio-demographic characteristics of the sample (N = 531)
Variable			Number	N (%) / Mean (SD)
Age			-		31.81 ± 3.99
Pre-pregnancy BMI	Abnormal		169		31.8
	Normal		362		68.2
Gestational weight gain	Abnormal		286		53.9
	Normal		245		46.1
Gravidity	Primigravid		237		44.6
	Multigravid		294		55.4
Parity	Primipara		307		57.8
	Multipara		224		42.2
Planned pregnancy	Intended		402		75.7
	Unplanned		129		24.3
Work nature	Sedentary		281		52.9
	Moderately active		97		18.3
	Home convalescence		153		28.8
Education level	High School and below		89		16.8
	College		138		26.0
	Bachelor's degree or above		304		57.3
Partner's education level	High School and below		103		19.4
	College		132		24.9
	Bachelor's degree or above		296		55.7
Marital Status	Married		515		97.0
	Other		16		3.0
Personal monthly income (RMB)	< 10000		63		11.9
	≥ 10000		468		88.1
Caregiver	Partner		324		61.0
	Other		207		39.0
Residence	Urban		386		72.7
	Other		145		27.3
Weekly exercise habit	< 30min		201		37.9
	< 150min		206		38.8
	≥ 150min		124		23.4
PFMT	Never		355		66.9
	Regularly		176		33.1
PASSS			-		78.76 ± 18.57
P-ESES			-		34.47 ± 8.10
PFMT = Pelvic Floor Muscle Training; PASSS = Physical Activity Social Support Scale; P-ESES = Pregnancy Exercise Self-Efficacy Scale
Table 2: Differences of the EARS-C scale items across pregnancy exercise adherence level. (N = 531)
	Total (531)	Class 1 (233)	Class 2 (239)	Class 3 (59)		F	P
1. I adjust the way I do my exercises to suit myself	1.36 ± 0.85	1.55 ± 0.90	1.21 ± 0.65	1.20 ± 1.19		11.124	< 0.001*
2. Other commitments prevent me from doing my exercises	1.76 ± 0.96	1.53 ± 0.85	1.77 ± 0.91	2.63 ± 1.03		34.881	< 0.001*
3. I feel confident about doing my exercises	2.45 ± 0.90	1.86 ± 0.84	2.80 ± 0.62	3.36 ± 0.52		153.58	< 0.001*
4. I don't have time to do my exercises	2.42 ± 0.99	2.21 ± 1.00	2.39 ± 0.94	3.32 ± 0.60		33.061	< 0.001*
5. I'm not sure how to do my exercises	2.22 ± 1.05	2.02 ± 1.05	2.21 ± 0.97	3.08 ± 0.92		26.712	< 0.001*
6. I don't do my exercises when I am tired	1.19 ± 0.80	1.01 ± 0.67	1.19 ± 0.71	1.90 ± 1.17		31.896	< 0.001*
7. I do my exercises because I enjoy them	2.16 ± 0.96	1.67 ± 0.92	2.48 ± 0.79	2.76 ± 0.84		68.478	< 0.001*
8. My family and friends encourage me to do my exercises	2.50 ± 0.95	2.00 ± 1.01	2.79 ± 0.69	3.24 ± 0.60		78.438	< 0.001*
9. I stop doing my exercises when my pain is worse	0.82 ± 0.71	0.73 ± 0.69	0.86 ± 0.61	1.00 ± 1.03		4.261	0.015*
10. I do my exercises to improve health	2.66 ± 0.99	2.25 ± 1.10	2.90 ± 0.77	3.32 ± 0.63		46.399	< 0.001*
Class 1 = Low-adherence profile; Class 2 = Moderate-adherence profile; Class 3 = High-adherence profile. *p < .05

Results of latent profile analysis

The study conducted a profile analysis based on the EARS-B scale scores. It used AIC, BIC, aBIC, LMR (P-value), and BLRT (P-value) as evaluation indices to select the best model from 1 to 4 potential profile models established in sequence. The results of the model fit metrics for each profile are detailed in Table 3. The number of model categories increased from 1 to 4, and the AIC, BIC, and aBIC continued to decrease, with entropy being the highest among the four profile models. The 4-profile model LMR was excluded(P > 0.05). Therefore, the model with three profiles was finally chosen as the optimal potential profile model for this study, as shown in Table 3.

Table 3
Latent profile analysis model fit for exercise adherence in pregnant women (N = 531).
Classes	AIC	BIC	aBIC	LMR(P)	BLRT(P)	Entropy	Categorical probability
class1	9007.198	9058.495	9020.403	-	-	-	1
class2	8440.484	8521.705	8461.393	< 0.001	< 0.001	0.816	0.461/0.539
class3	8288.664	8399.808	8317.276	0.0018*	< 0.001*	0.802	0.439/0.450/0.111
class4	7757.399	7898.466	7793.714	0.2101	< 0.001	1	0.280/0.403/0.266/0.051
The selected models' AIC, BIC, and ABIC are information criteria; P-LMR and P-BLRT are model fit statistics. Depiction: Class 1 = Low-adherence profile; Class 2 = Moderate-adherence profile; Class3 = High-adherence profile; *p < .05

Naming of latent profile

Figure 1 presents the mean scores of each item for the three exercise adherence profiles among pregnant women. The categories were named: “low-adherence” group, “moderate-adherence” group, and “high-adherence” group based on the characteristics of the mean scores across different categories. The “moderate-adherence” profile had the highest percentage, at 45.0%, followed by the “low-adherence” profile at 43.9%, and the “high-adherence” profile at 11.1%.

Fig. 1

Schematic diagram of the 3-category model of adherence to exercise in pregnant women.

Description of the selected LPA profiles of exercise adherence in pregnant women. Depiction: Class 1, Low-adherence profile; Class 2, Moderate-adherence profile; Class 3, High-adherence profile
Note. Q1 = I do my exercises as often as recommended; Q2 = I don't get around to doing my exercises; Q3 = I do most, or all, of my exercises; Q4 = I do less exercise than recommended by my healthcare professional; Q5 = I fit my exercises into my regular routine; Q6 = I forget to do my exercises in the horizontal coordinate.

Inter-profile characteristic differences

The chi-square test and the one-way analysis of variance were used to compare differences in the presence of influencing factors among pregnant women in different potential exercise adherence categories. The results showed that gravidity, parity, education level, Exercise habit, PFMT, exercise self-efficacy, and social support were statistically significant (P < 0.05). The rest of the categorical differences were not statistically significant (P > 0.05), as seen in Table 4.

Table 4: Differences in demographics, pregnancy exercise self-efficacy, and pregnancy physical activity support in pregnancy exercise adherence level. (N = 531)
Variable			Class 1 (233)	Class 2 (239)	Class 3 (59)	c2 / F	P
Age			31.47 ± 4.12	32.09 ± 3.85	32.02 ± 3.95	1.538	0.216
Pre-pregnancy BMI	Abnormal		81 (34.80%)	73 (30.50%)	15 (25.40%)	2.223	0.329
	Normal		152 (65.20%)	166 (69.50%)	44 (74.60%)
Gestational weight gain	Abnormal		133 (57.10%)	127 (53.10%)	26 (44.10%)	3.300	0.192
	Normal		100 (42.90%)	112 (46.90%)	33 (55.90%)
Gravidity	Primigravid		120 (51.50%)	89 (37.20%)	28 (47.50%)	9.928	0.007*
	Multigravid		113 (48.50%)	150 (62.80%)	31 (52.50%)
Parity	Primipara		151 (64.80%)	118 (49.40%)	38 (64.40%)	12.706	0.002*
	Multipara		82 (35.20%)	121 (50.60%)	21 (35.60%)
Planned pregnancy	Intended		168 (72.10%)	185 (77.40%)	49 (83.10%)	3.751	0.153
	Unplanned		65 (27.90%)	54 (22.60%)	10 (16.90%)
Work nature	Sedentary		123 (52.80%)	125 (52.30%)	33 (55.90%)	0.998	0.910
	Moderately active		43 (18.50%)	42 (17.60%)	12 (20.30%)
	Home convalescence		67 (28.80%)	72 (30.10%)	14 (23.70%)
Education level	High School and below		40 (17.20%)	42 (17.60%)	7 (11.90%)	11.353	0.023*
	College		60 (25.80%)	71 (29.70%)	7 (11.90%)
	Bachelor's degree or above		133 (57.10%)	126 (52.70%)	45 (76.30%)
Partner's education level	High School and below		50 (21.50%)	46 (19.20%)	7 (11.90%)	2.793	0.593
	College		57 (24.50%)	59 (24.70%)	16 (27.10%)
	Bachelor's degree or above		126 (54.10%)	134 (56.10%)	36 (61.00%)
Marital Status	Married		223 (95.70%)	235 (98.30%)	57 (96.60%)	2.800	0.247
	Other		10 (4.30%)	4 (1.70%)	2 (3.40%)
Personal monthly income (RMB)	< 10000		31 (13.30%)	27 (11.30%)	5 (8.50%)	1.184	0.553
	≥ 10000		202 (86.70%)	212 (88.70%)	54 (91.50%)
Caregiver	Partner		153 (65.70%)	133 (55.60%)	38 (64.40%)	5.297	0.071
	Other		80 (34.30%)	106 (44.40%)	21 (35.60%)
Residence	Urban		159 (68.20%)	179 (74.90%)	48 (81.40%)	5.142	0.076
	Other		74 (31.80%)	60 (25.10%)	11 (18.60%)
Exercise habit weekly	< 30min	117 (50.20%)		78 (32.60%)	6 (10.20%)
	< 150min		81 (34.80%)	107 (44.80%)	18 (30.50%)	62.962	0.001*
	≥ 150min		35 (15.00%)	54 (22.60%)	35 (59.30%)
PFMT	Never		177 (76.00%)	152 (63.60%)	26 (44.10%)	23.697	< 0.001*
	Regularly		56 (24.00%)	87 (36.40%)	33 (55.90%)
PASSS			67.12 ± 16.25	86.14 ± 13.60	94.83 ± 17.45	128.479	< 0.001*
P-ESES			29.10 ± 7.56	37.90 ± 5.52	41.80 ± 5.18	151.931	< 0.001*
PFMT = Pelvic Floor Muscle Training; PASSS = Physical Activity Social Support Scale; P-ESES = Pregnancy Exercise Self-Efficacy Scale; *p < .05

Multinomial logistic regression of exercise adherence profiles

Demographic information

When the low-adherence profile was the reference group, relative to the moderate-adherence profile, the weekly exercise habit < 30 min (OR = 0.218, p < 0.001) for the moderate-adherence profile was significantly harmful; the scores of PASSS (OR = 1.060, p < 0.001) and P-ESES (OR = 1.189, p < 0.001) for the moderate-adherence profile were significantly positive. The results presented in Table 5 suggest that, first, women’s weekly exercise habit < 30 minutes, the less likely they are to be categorized into the moderate-adherence profile. Second, the higher scores of PASSS and P-ESES were more likely they were to be classified as moderate-adherence.

Relative to the high-adherence profile: the exercise habit weekly < 30 min (OR = 0.021, p < 0.001) and < 150 min (OR = 0.174, p < 0.001), and never PFMT (OR = 0.319, p < 0.05) for the high-adherence profile was significantly negative; the scores of PASSS (OR = 1.091, p < 0.001) and P-ESES (OR = 1.347, p < 0.001) for high-adherence were significantly positive. Suggests that, first, the weekly exercise habit < 30 min and < 150 min, and women who have never engaged in PFMT, are less likely they are to be categorized into the high-adherence profile. Second, the higher scores of PASSS and P-ESES were more likely to be classified as a high adherence profile.

When the moderate-adherence profile was the reference group, relative to the high-adherence profile: the college education level (OR = 0.299, p < 0.05), weekly exercise habit < 30 min (OR = 0.097, p < 0.001) and < 150min (OR = 0.259, p < 0.001) for high-adherence profile were significantly negative; PASSS (OR = 1.029, p < 0.05) and P-ESES (OR = 1.133, p < 0.001) scores for high-adherence profile were significantly positive. Suggests that, first, college education level and weekly exercise habit (< 30min and < 150min) in pregnant women are less likely to be classified as a high-adherence profile. Second, the higher the PASSS and P-ESES scores, the more likely pregnant women are to be classified as a high-adherence profile.

Table 5
Multinomial logistic regression analysis of the pregnancy exercise adherence among three classes (N = 531)
Variables	Class 1^a vs. Class 2				Class 1^a vs. Class 3				Class 2^a vs. Class 3
	β	SE	P	OR	β	SE	P	OR	β	SE	P	OR
	-9.155	0.995	< 0.001*		-16.82	1.800	< 0.001*		-7.666	1.555	< 0.001*
Gravidity
Primigravid	-0.652	0.405	0.107	0.521	-0.862	0.616	0.162	0.422	-0.209	0.507	0.680	0.811
Multigravid
Parity
Primipara	0.057	0.406	0.889	1.059	0.805	0.632	0.202	2.237	0.748	0.521	0.151	2.113
Multipara
Education level
High School and below	0.729	0.380	0.055	2.073	0.678	0.615	0.271	1.969	-0.051	0.518	0.921	0.950
College	0.285	0.301	0.343	1.330	-0.924	0.553	0.095	0.397	-1.208	0.484	0.013*	0.299
Professional or graduate
Weekly exercise habit
< 30min	-1.525	0.384	< 0.001*	0.218	-3.855	0.636	< 0.001*	0.021	-2.330	0.539	< 0.001*	0.097
< 150min	-0.397	0.366	0.279	0.673	-1.747	0.493	< 0.001*	0.174	-1.351	0.375	< 0.001*	0.259
≥ 150min
PFMT
Never	-0.554	0.286	0.053	0.574	-1.142	0.434	0.009*	0.319	-0.587	0.358	0.100	0.556
Regularly
PASSS	0.059	0.010	< 0.001*	1.060	0.087	0.015	< 0.001*	1.091	0.028	0.013	0.025*	1.029
P-ESES	0.174	0.024	< 0.001*	1.189	0.298	0.042	< 0.001*	1.347	0.124	0.037	< 0.001*	1.133
^aReference group; Class 1 = Low-adherence profile; Class 2 = Moderate-adherence profile; Class3 = High-adherence profile; PFMT = Pelvic Floor Muscle Training; PASSS = Physical Activity Social Support Scale; P-ESES = Pregnancy Exercise Self-Efficacy Scale; *p < .05

Reasons for adherence / non-adherence

When the low-adherence profile was the reference group, relative to the moderate-adherence profile: item 1: “I adjust the way I do my exercises to suit myself” (OR = 0.689, p < 0.05), for moderate-adherence were significantly negative; item 2: “Other commitments prevent me from doing my exercises” (OR = 1.520, p < 0.05), item 3: “I feel confident about doing my exercises” (OR = 3.280, p < 0.001), item 7: “I do my exercises because I enjoy them” (OR = 1.981, p < 0.001), item 8: “My family and friends encourage me to do my exercises” (OR = 1.513, p < 0.05), and item 10: “I do my exercises to improve my health” (OR = 1.412, p < 0.05), for moderate-adherence was significantly positive; This suggests that, first, the higher scores of item 3, item 7, item 8 and item 10 more likely they were to be categorized as moderate-adherence profile; second, the lower scores of item 1 of scores more likely they was to be categorized as moderate-adherence profile. The results of the study are presented in Table 6. Relative to the high-adherence profile: item 2: “Other commitments prevent me from doing my exercises” (OR = 2.005, p < 0.05), item 3: “I feel confident about doing my exercises” (OR = 9.089, p < 0.001), item 4: “I don’t have time to do my exercises” (OR = 4.505, p < 0.001), item 6: “I don’t do my exercises when I am tired” (OR = 2.692, p < 0.001), item 7: “I do my exercises because I enjoy them” (OR = 2.141, p < 0.05) and item 8: “My family and friends encourage me to do my exercises” (OR = 3.743, p < 0.05) were significantly positive; This suggests that, the higher scores of item2, 3, 4, 6, 7 and 8 more likely they were to be categorized the high-adherence profile.

When the moderate-adherence profile was the reference group, relative to the high-adherence profile: item 3: “I feel confident about doing my exercises” (OR = 2.522, p < 0.05), item 4: “I don’t have time to do my exercises” (OR = 3.916, p < 0.001), item 6: “I don’t do my exercises when I am tired” (OR = 2.211, p < 0.001), item 8: “My family and friends encourage me to do my exercises” (OR = 2.474, p < 0.05) for high-adherence were significantly positive. Suggests that the higher scores of items 3, 4, 6, and 8 are more likely to be categorized as high-adherence.

Table 6
Multinomial logistic regression analysis of the EARS-C scale items among three classes (N = 531)
Variables	Class 1^a vs. Class 2				Class 1^a vs. Class 3				Class 2^a vs. Class 3
	β	SE	P	OR	β	SE	P	OR	β	SE	P	OR
	-7.180	0.854	< 0.001*		-21.887	2.417	< 0.001*		-14.707	2.271	< 0.001*
1. I adjust the way I do my exercises to suit myself	-0.372	0.170	0.028*	0.689	-0.496	0.288	0.085	0.609	-0.124	0.246	0.616	0.884
2. Other commitments prevent me from doing my exercises	0.418	0.167	0.012*	1.520	0.695	0.262	0.008*	2.005	0.277	0.218	0.205	1.319
3. I feel confident about doing my exercises	1.166	0.193	< 0.001*	3.208	2.091	0.445	< 0.001*	9.089	0.925	0.413	0.025*	2.522
4. I don't have time to do my exercises	0.140	0.153	0.359	1.150	1.505	0.439	< 0.001*	4.505	1.365	0.419	0.001*	3.916
5. I'm not sure how to do my exercises	0.112	0.140	0.424	1.119	0.445	0.283	0.116	1.561	0.333	0.258	0.196	1.395
6. I don't do my exercises when I am tired	0.197	0.192	0.307	1.217	0.990	0.289	< 0.001*	2.692	0.794	0.234	< 0.001*	2.211
7. I do my exercises because I enjoy them	0.684	0.145	< 0.001*	1.981	0.761	0.267	0.004*	2.141	0.077	0.238	0.745	1.080
8. My family and friends encourage me to do my exercises	0.414	0.165	0.012*	1.513	1.320	0.423	0.002*	3.743	0.906	0.401	0.024*	2.474
9. I stop exercising when my pain is worse	0.161	0.196	0.414	1.174	0.631	0.379	0.096	1.880	0.471	0.342	0.168	1.602
10. I do my exercises to improve my health	0.345	0.157	0.027*	1.412	0.564	0.325	0.083	1.757	0.218	0.300	0.466	1.244
^aReference group; Class 1 = Low-adherence profile; Class 2 = Moderate-adherence profile; Class3 = High-adherence profile, *p < .05

Discussion

This study is the first to examine latent profiles of exercise adherence among women in their third trimester and to identify subtypes of exercise adherence using a person-centered approach. Exercise adherence among pregnant women was classified into three distinct profiles: low-adherence (43.9%), moderate-adherence (46%), and high-adherence (11.1%). The average scores for these groups were 9.32, 13.77, and 19.66, respectively. This result was consistent with the finding from Lu [26]. The mean score on the EARS-B scale among pregnant women was 12.47 ± 4.01, which fell below the established cut-off point [27].

The current study demonstrated that education level, weekly exercise habit, and PFMT were associated with exercise adherence among pregnant women. Consistent with previous studies, low education level was identified as a negative influence on meeting recommended physical activity and exercise during pregnancy [28, 29]. Compared to the high-adherence group, pregnant women who had a lower education level were more likely to be classified into the mid-adherence group. Zhang et al. [30] found that pre-pregnancy exercise habit was a significant predictor of intent to initiate physical activity. A systematic review also demonstrated that regular physical activity was associated with higher self-efficacy. Pregnant women had a more positive attitude toward physical activity and exercise during pregnancy; ultimately, they adhered to exercise [31]. In this study, women who performed PFMT tended to have higher exercise adherence. Pelvic floor disorders, such as stress urinary incontinence (SUI) and pelvic floor pain during the last pregnancy, may impair women's motivation and ability to exercise. PFMT is most effective in treating SUI by strengthening the pelvic floor muscles. Physical activity programs that incorporate PFMT may be more motivating for women to engage in general aerobic exercise during pregnancy [32, 33]. Additionally, information about the benefits of exercise and the dangers of inactivity during pregnancy should be incorporated into tailored exercise programs for pregnant women with less formal education.

Multinomial logistic regression analysis indicated that exercise self-efficacy was a crucial factor influencing exercise adherence among pregnant women; higher scores on the P-ESES significantly increased the odds of membership in the moderate or high adherence profiles. Aligning with social cognitive theory [23], exercise self-efficacy refers to the confidence in one's ability to overcome more challenging tasks and achieve desired results[34, 35]. Similarly, social support has been identified as a predictor of exercise adherence, as confirmed by a systematic review [36]. Although the present study also observed a comparatively modest effect of social support on exercise adherence, this suggests that the influence of social support on exercise adherence may be moderated by internal and external factors [36]. Indeed, partners, family members, friends, and health-care professionals have repeatedly been identified as key sources of social support for promoting physical activity during pregnancy [37]. Moreover, studies by Verloigne and Lu [38, 39] indicated that physical activity self-efficacy mediates the relationship between social support and physical activity. Therefore, interventions designed to enhance exercise self-efficacy, leverage multiple sources of social support, and encourage regular physical activity may help pregnant women achieve better adherence outcomes.

We used the EARS-C scale to investigate the reasons for adherence in those who exhibited good and poor adherence to exercise. These included items 3 and 8 as two major protective factors in the mid- and high-adherence groups (Table 6), indicating that exercise self-efficacy and social support play crucial roles in exercise adherence [23]. Other facilitating factors included enjoyment of exercise and self-adjustment. In line with self-determination theory, the enjoyment of exercise, as an inherent motivator, was positively associated with exercise behavior [40]. Empirical evidence further confirmed that intrinsic motivation supports both sustained exercise and psychological well-being[40]. Accordingly, prenatal exercise programs should emphasize enjoyment, skill mastery, and accomplishment, while enhancing maternal well-being and vitality. Mobile Health exercise programs with attractive course visuals and motivating music enhance the overall workout experience, making it more enjoyable [41]. Furthermore, healthcare professionals should focus on developing personalized and gamified mobile health (mHealth) exercise programs for pregnant women.

Negative influences to exercise adherence included competing commitments, lack of time, and fatigue, which align with findings from Shang and Koleilat [11, 42]. Prior research attributes these barriers to the dual responsibilities of caregiving and the workforce that these women faced [37]. Since the implementation of China’s “three-child policy”, the proportion of multiparous women has increased; over half of the participants who were multiparas in our study faced challenges of balancing competing commitments and physical exercise. For this reason, it is essential to consider time constraints when designing physical activity interventions for pregnant women. In response, we propose a community-based program that will facilitate peer interaction and the sharing of challenges and concerns [43]. The program will include weekly group classes tailored to varying fitness levels, as well as workshops on time management and strategies for balancing responsibilities [44].

This study has two strengths. First, it uses a multicenter design, which reduces the bias that single-center studies often have. Second, it takes a person-centered analytical approach to capture varied adherence patterns among individuals. This study has several limitations. First, all measures were self-reported questionnaires, excluding weight and height. Nevertheless, these scales have all shown good reliability and validity in this study, ensuring the accuracy of the results. Its cross-sectional design limits causal inference and does not allow for investigating how exercise adherence changes during the entire pregnancy period. Future studies should follow pregnant women across the three trimesters of pregnancy to facilitate a more comprehensive exploration of the multi-trajectory mechanisms underlying exercise adherence.

Conclusion

This study showed that exercise adherence among pregnant women appeared heterogeneous and presented in three distinct categories. Exercise self-efficacy, social support, weekly exercise habit, PFMT, enjoyment of exercise, other commitments, lack of time, and fatigue emerged as primary factors influencing sustained exercise. This evidence lays a crucial foundation for future tailored physical activity programs.

Abbreviations

WHO	World Health Organization
GDM	Gestational Diabetes Mellitus
EARS	Exercise Adherence Rating Scale
LPA	Latent Profile Analysis
BMI	Body Mass Index
PASSS	Physical Activity Social Support Scale
P-ESES	Pregnancy Exercise Self-Efficacy Scale
SD	Standard Deviation
AIC	Akaike Information Criterion
BIC	ayesian Information Criterion
aBIC	adjusted Bayesian Information Criterion
LMR	Lo-Mendell-Rubin
BLRT	Bootstrapped Likelihood Ratio Test
OR	Odds Ratio
RMB	Renminbi
PFMT	Pelvic floor muscle training

Author Contribution

ZCH and MH design and write articles. MH, CHN, XSR, OYLF, LKQ, and XZM completed the data collection. MH and CHN completed the data input and analysis. ZCH completed the revision and check of the article. All the authors read and approved the final version of the manuscript.

Funding

This work was supported by the Guangdong Province Graduate Education Innovation Program Project (2025JGXM_083).

Data Availability

The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the guidelines of the Declaration of Helsinki. It was approved by the Ethics Committee of the Dongguan Maternal and Child Health Hospital (Approval No. 2024 − 155); Shunde Women and Children's Hospital of Guangdong Medical University (KY-2024-070); and Baoan Central Hospital of Shenzhen (BYL20240628). Before the study, both oral and written consents were obtained from all eligible participants, ensuring that participation or non-participation would not affect their work performance or future employment opportunities. Additionally, all collected information would be anonymous and de-identified. Furthermore, participants were informed that they could withdraw from the study at any time.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Reference

Gascoigne EL, Webster CM, Honart AW, Wang P, Smith-Ryan A, Manuck TA. Physical activity and pregnancy outcomes: an expert review. Am J Obstet Gynecol MFM. 2023;5(1):100758.

Bull FC, Al-Ansari SS, Biddle S, Borodulin K, Buman MP, Cardon G. World Health Organization 2020 guidelines on physical activity and sedentary behaviour. Br J Sports Med. 2020;54(24):1451–62.

Silva-Jose C, Sánchez-Polán M, Barakat R, Gil-Ares J, Refoyo I. Level of Physical Activity in Pregnant Populations from Different Geographic Regions: A Systematic Review. J Clin Med 2022, 11(15).

Harris JE, Baer LA, Stanford KI. Maternal Exercise Improves the Metabolic Health of Adult Offspring. Trends Endocrinol Metab. 2018;29(3):164–77.

Wu YN, Wang HB, Xu H, Qian ZT, Wang MM, Xu M. Efficacy of non-pharmacological interventions for the prevention of antenatal depression: A systematic review and network meta-analysis. Neurosci Biobehav Rev. 2025;172:106088.

Sarno L, Borrelli P, Mennitti C, Gentile A, Calvanese M, Orlandi G. Adherence to physical activity among pregnant women in Southern Italy: results of a cross-sectional survey. Midwifery. 2024;137:104102.

Zhang W, Zhang L, Xu P, Guo P, Mao M, Zhao R. Physical activity levels and influencing factors among pregnant women in China: A systematic review and meta-analysis. Int J Nurs Stud. 2024;158:104841.

Zhou T, Lin Y, Xu F, Ma X, Wang N, Ding Y. Factors influencing physical inactivity status among chinese pregnant women: a cross-sectional study. BMC Public Health. 2022;22(1):2310.

Walasik I, Kwiatkowska K, Kosińska Kaczyńska K, Szymusik I. Physical Activity Patterns among 9000 Pregnant Women in Poland: A Cross-Sectional Study. Int J Environ Res Public Health 2020, 17(5).

10.

Kókai L, Ceallaigh DÓ, Wijtzes AI, Cawley J, Hagger MS, Burdorf A. App-Based Physical Activity Intervention Among Women With Prior Hypertensive Pregnancy Disorder: A Randomized Clinical Trial. JAMA Netw Open. 2025;8(4):e252656.

11.

Shang X, Ye L, Gu X, Zhou A, Xu Y, Zhang Y. Attitudes and Barriers to Physical Activity and Exercise Self-Efficacy Among Chinese Pregnant Women: A Cross-Sectional Study. J Multidiscip Healthc. 2023;16:3561–73.

12.

Frost R, Levati S, McClurg D, Brady M, Williams B. What Adherence Measures Should Be Used in Trials of Home-Based Rehabilitation Interventions? A Systematic Review of the Validity, Reliability, and Acceptability of Measures. Arch Phys Med Rehabil. 2017;98(6):1241–e12561245.

13.

Zhang JY, Zhang W, Zhang L, Shen XS, Liu Y, Feng NN. Enhancing physical activity during pregnancy using a multi-theory model: a study protocol for a randomised controlled trial in China. BMJ Open. 2025;15(1):e085910.

14.

Newman-Beinart NA, Norton S, Dowling D, Gavriloff D, Vari C, Weinman JA. The development and initial psychometric evaluation of a measure assessing adherence to prescribed exercise: the Exercise Adherence Rating Scale (EARS). Physiotherapy. 2017;103(2):180–5.

15.

Wu YX, Yu XY, Luo XC, Lin ST, Du HM, Liu D. Translation of the Exercise Adherence Rating Scale and its reliability and validity test in cancer patients. Chin Nurs Res 2024, 38(21).

16.

Söderberg J, Sallfeldt E, Ribom E, Urell C. Cross-cultural adaptation and test-retest reliability assessment of a Swedish version of the exercise adherence rating scale in patients after shoulder surgery. Ann Med. 2024;56(1):2409962.

17.

ein JY, Coxe S, Cham H. Statistical Power to Detect the Correct Number of Classes in Latent Profile Analysis. Struct equation modeling: multidisciplinary J. 2013;20(4):640–57.

18.

Department of Medical Administration, National Health Commission of the People's Republic of China. Guideline for diagnosis and treatment of obesity (2024 edition). Chin J Dig Surg. 2025;40(01):6–30.

19.

Adhikari SP, Dev R, Shrestha JN. Cross-cultural adaptation, validity, and reliability of the Nepali version of the Exercise Adherence Rating Scale: a methodological study. Health Qual Life Outcomes. 2020;18(1):328.

20.

Takasaki H, Kawazoe S, Miki T, Chiba H, Godfrey E. Development and validity assessment of a Japanese version of the Exercise Adherence Rating Scale in participants with musculoskeletal disorders. Health Qual Life Outcomes. 2021;19(1):169.

21.

Xiang Z, Han R, Zhao Q, Yang X, Gao L. Adaptation and psychometric evaluation of the social support scale for exercise among pregnant women. Chin Nurs Manage. 2022;22(7):998–1002.

22.

Kroll T, Kehn, Matthew H, Groah PS. Suzanne: The SCI Exercise Self-Efficacy Scale (ESES): development and psychometric properties. Int J Behav Nutr Phys Act. 2007;4:34.

23.

Bland HW, Melton BF, Marshall ES, Nagle JA. Measuring exercise self-efficacy in pregnant women: psychometric properties of the pregnancy-exercise self-efficacy scale (P-ESES). J Nurs Meas. 2013;21(3):349–59.

24.

Kim HY. Statistical notes for clinical researchers: assessing normal distribution (2) using skewness and kurtosis. Restor dentistry endodontics. 2013;38(1):52–4.

25.

Muthén. Latent variable analysis. The Sage handbook of quantitative methodology for the social sciences 2004. 345(368): pp. 106–109.

26.

Lu J, Wang J, You Z, Wang Q, Sun G. A study on the influencing factors of exercise adherence in patients with chronic heart failure: from a configuration perspective. Front Psychol. 2025;16:1536349.

27.

de Lira MR, de Oliveira AS, França RA, Pereira AC, Godfrey EL, Chaves TC. The Brazilian Portuguese version of the Exercise Adherence Rating Scale (EARS-Br) showed acceptable reliability, validity and responsiveness in chronic low back pain. BMC Musculoskelet Disord. 2020;21(1):294.

28.

Ceprnja D, Chipchase L, Liamputtong P, Gupta A. Physical activity and associated factors in Australian women during pregnancy: A cross-sectional study. Health promotion J Australia: official J Australian Association Health Promotion Professionals. 2024;35(4):1217–23.

29.

Chen X, Xiang Z, Chen L, Sun K, Deng Y, Gao L. Physical activity among Chinese pregnant women in the first trimester: A cross-sectional study. Int J Nurs Sci. 2025;12(3):261–7.

30.

Chen X, Xiang Z, Chen L, Sun K, Deng Y, Gao L. Predicting Physical Activity in Chinese Pregnant Women Using Multi-Theory Model: A Cross-Sectional Study. Int J Environ Res Public Health, 2022. 19(20).

31.

Yang C, Wu Q, Lv Q, Hou X, Ye X, Yang Y. Efficacy of physical exercise on the physical ability, cardiac function and cardiopulmonary fitness of patients with atrial fibrillation: a systematic review and meta-analysis. Front Cardiovasc Med. 2024;11:1352643.

32.

Zhang DF, Kari B, Roc M, Migue SP, Cristina SJ, Montse P. Influence of pelvic floor muscle training alone or as part of a general physical activity program during pregnancy on urinary incontinence, episiotomy and third- or fourth-degree perineal tear: Systematic review and meta-analysis of randomized clinical trials. Acta Obstet Gynecol Scand. 2024;103(6):1015–27.

33.

Stephanie JW, Peter L, Rhianon B, June DC, Siv M, Ashleigh K. Pelvic floor muscle training for preventing and treating urinary and faecal incontinence in antenatal and postnatal women. Cochrane Database Syst Rev, 2020. 5(5): p. Cd007471.

34.

Xiang Z, Han R, Chen L, Gao L. Predictors of physical activity among Chinese pregnant women during the first trimester: A cross-sectional study. J Sports Sci. 2023;41(20):1883–91.

35.

Yang X, Han R, Song Y, Zhang J, Huang H, Zhang J. The Mediating Role of Physical Activity Self-Efficacy in Predicting Moderate-Intensity Physical Activity in Pregnant People at High Risk for Gestational Diabetes. J Midwifery Women's Health. 2024;69(3):403–13.

36.

Thompson EL, Vamos CA, Daley EM. Physical activity during pregnancy and the role of theory in promoting positive behavior change: A systematic review. J sport health Sci. 2017;6(2):198–206.

37.

Shum KW, Ang MQ, Shorey S. Perceptions of physical activity during pregnancy among women: A descriptive qualitative study. Midwifery. 2022;107:103264.

38.

Chen L, Han RR, Chen X, Fu BL, Nogueira B, Gao LL. Evaluation of the mediating role of physical activity self-efficacy in the relationship between knowledge, social support, and physical activity in pregnant women with a high risk for gestational diabetes. BMC Pregnancy Childbirth. 2024;24(1):857.

39.

Verloigne M, Cardon G, De Craemer M, D' Haese S, De Bourdeaudhuij I. Mediating Effects of Self-Efficacy, Benefits and Barriers on the Association between Peer and Parental Factors and Physical Activity among Adolescent Girls with a Lower Educational Level. PLoS ONE. 2016;11(6):e0157216.

40.

Teixeira PJ, Carraca EV, Markland D, Silva MN, Ryan RM. Exercise, physical activity, and self-determination theory: a systematic review. Int J Behav Nutr Phys Act. 2012;9:78.

41.

Verloigne M, Cardon G, De Craemer M, D' Haese S, De Bourdeaudhuij I. Effectiveness of a Smartphone App to Promote Healthy Weight Gain, Diet, and Physical Activity During Pregnancy (HealthyMoms): Randomized Controlled Trial. JMIR Mhealth Uhealth. 2021;9(3):e26091.

42.

Koleilat M, Vargas N, vanTwist V, Kodjebacheva GD. Perceived barriers to and suggested interventions for physical activity during pregnancy among participants of the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) in Southern California. BMC Pregnancy Childbirth. 2021;21(1):69.

43.

Chen YP, Hou LY, Li Y, Lou Y, Li W, Yang H. Barriers and motivators to promotion of physical activity participation for older adults with mild cognitive impairment or dementia: An umbrella review. Int J Nurs Stud. 2023;143:104493.

44.

Modolo VB, Antunes HK, Gimenez PR, Santiago ML, Tufik S, Mello MT. Negative addiction to exercise: are there differences between genders? Clin (Sao Paulo Brazil). 2011;66(2):255–60.

Yes