Bioecological attributes of learning adaptation among nursing freshmen during pandemic-induced isolation: a latent profile analysis

Abstract

Background

The COVID-19 pandemic has significantly disrupted educational style, potentially affecting the learning adaptation of nursing freshmen who are integral to the future nursing workforce.

Objective

This study aimed to identify distinct subgroups of nursing freshmen based on their bioecological attributes related to learning adaptation during the pandemic.

Methods

A multicenter, cross-sectional study was conducted of 1170 first-year nursing students from six higher education institutions in China. Learning adaptation, resilience, parental attachment, interaction anxiety, and mobile phone addiction, were investigated. Latent Profile Analysis (LPA) was utilized to identify distinct profiles.

Results

Descriptive statistics indicated a positive level of learning adaptation among participants, with an overall mean score of 3.51 ± 0.57. LPA revealed four distinct profiles: ‘Struggling Learners’ (5.47%), ‘Moderate Engagers’ (70.60%), ‘Adaptable Strivers’ (18.29%), and ‘Optimal Adapters’ (5.64%), which demonstrated significant differences in adaptation, resilience, parental attachment, interaction anxiety, and mobile phone addiction tendencies (P < 0.05).

Conclusion

The study’s findings emphasize the heterogeneity in learning adaptation among nursing freshmen and the importance of considering bioecological attributes when developing educational interventions during crisis. Recognizing these profiles can guide the development of targeted strategies to enhance student adaptation and academic achievement.

Introduction

The COVID-19 pandemic, emerging in late 2019, caused unprecedented disruptions in global education systems, with nursing education facing particular challenges due to its practice-based nature [1]. For nursing freshmen—who represent the future frontline healthcare workforce—the pandemic introduced unique bioecological stressors that threatened their academic and psychological adaptation [2]. Learning adaptation, defined as the dynamic process through which students adjust their psychological states and behaviors to achieve equilibrium with academic environments [3], extensive research demonstrates that adaptive capacity serves as both a predictor of autonomous learning and a protective factor for mental health [4, 5] and academic performance [6, 7]. Thus, understanding the global mechanisms of nursing students’ adaptation has emerged as an internationally urgent priority in post-pandemic educational reconstruction [8].

Bronfenbrenner’s bioecological model [9] provides a valuable framework for understanding the multidimensional impacts of the pandemic on nursing students. At the microsystem level, the pandemic fundamentally altered students’ immediate learning environments. Lockdowns disrupted traditional peer interactions and clinical training [10]. Simultaneously, the swift transition to virtual classrooms, albeit essential, led to excessive mobile phone use [11]. This not only posed new challenges to students’ academic focus but also affected their mental well being. In response to these microsystem changes, greater individual resilience defined as the ability to recover from adversity [12] - became necessary for students to succeed in the transformed academic landscape [13]. During the COVID − 19 pandemic, this attribute proved to be of particular significance [14]. Within the mesosystem, the connections between microsystems became strained, as home-school transitions became fragmented [15, 16], placing additional strain on familial support systems. Parental relationships, as the most significant social ties for first-year students [17], played a pivotal role in adaptation. Empirical studies confirm that parent-child relationships [18] and perceived parental involvement [19, 20] strongly predict learning adaptation, fostering virtues such as wisdom, courage, and emotional stability [21]. Meanwhile, at the macrosystem level, broader societal and technological changes manifested through the widespread adoption of digital learning platforms [22]. This macrosystem shift, while enabling continued education, also increased interaction anxiety in virtual environments [23].

The storm of stressors resulted in concerning adaptation rates - studies reported 27.13% of college freshmen exhibited maladaptation during peak pandemic periods [24]. Traditional approaches often examine risk factors in isolation [25], while life course theory (LCT) posits that cognitive development, familial connections, and social interactions are interwoven throughout an individual’s lifespan [26]. For nursing freshmen, the pandemic constituted a critical period—social isolation disrupted academic microsystems (e.g., peer learning) and strained mesosystem linkages (e.g., family-school communication), potentially triggering cumulative disadvantage. Developmental cascade theory (DCM) further suggests that such disruptions may propagate across domains (e.g., behavioral coping deficits exacerbating cognitive struggles) [27]. However, few studies have employed person-centered approaches to examine how these factors collectively shape nursing students’ adaptation during crises.

To address these limitations, we employ Latent Profile Analysis (LPA) [28], a method that identifies naturally occurring subgroups based on multivariate patterns, overcoming the limitations of traditional variable-centered analyses. LPA has proven valuable in medical education research [29–31] and is recognized as an effective tool for implementing differentiated interventions [32]. By applying LPA, we hypothesize that nursing freshmen’s pandemic adaptation will manifest as distinct subgroups characterized by differential combinations of: (a) individual resilience and mobile phone addiction (microsystem), (b) parental attachment (mesosystem), and (c) interaction anxiety (macrosystem), consistent with Bronfenbrenner’s bioecological framework.

Methods

Setting and participants

This study adopted a multi-center cross-sectional design. Given the geographical restrictions during the pandemic, a multi-stage sampling strategy was employed to recruit participants from August to September 2022.

Stage 1 (Institution Sampling): Using non-probability convenience sampling, 6 institutions that agreed to participate were selected from an initial pool of 12 nursing-offering colleges and universities in Southwest China (Chongqing Municipality, Sichuan Province, Guizhou Province, and Yunnan Province). These 6 institutions, comprising 4 medical universities and 2 general colleges, represented diverse nursing education models (4-year undergraduate programs and 3-year junior college programs), aiming to enhance sample diversity. Stage 2 (Student Sampling): Within each participating institution, cluster sampling was used, where all eligible first-year nursing students were invited as potential subjects. The specific inclusion criteria were: formally registered first-year nursing students who were to complete their first academic year by July 2022. Exclusion criteria included students with known mental disorders, those on leave of absence or participating in overseas exchange programs.

A total of 1,395 students responded to the online questionnaire. After excluding 225 questionnaires with incomplete information, 1,170 participants were included in the final analysis. Regarding the explanation of sample representativeness: Due to the use of non-probability sampling for institutional selection in the first phase, the sample has certain limitations in terms of regional distribution and institutional type. The research conclusions are mainly applicable to the regions involved in the survey, or to the groups of new nursing students in institutions with similar conditions to these regions. We have clearly identified this as one of the limitations of this study in the paper, to remind readers to exercise caution when interpreting and generalizing the research conclusions.

Theoretical framework

This study integrates Bronfenbrenner’s bioecological model, LCT, and DCM to construct a multi-level, dynamic analytical framework, aiming to unravel the mechanisms underlying learning adaptation among nursing freshmen during the pandemic. As shown in Fig. 1,

Fig. 1.

Learning adaptation of nursing freshmen during pandemic-induced isolation based on the bioecological model

The microsystem, corresponding to individual-level psychological and behavioral characteristics, is operationalized using the Connor-Davidson Resilience Scale-10 (CD-RISC-10) for resilience and the Mobile Phone Addiction Tendency Scale for mobile phone addiction. The mesosystem, reflecting interactions between family and school environments, is operationalized via the Inventory of Parent and Peer Attachment (IPPA) for parental attachment. The macrosystem, representing the impact of broader sociotechnical contexts on individuals, is operationalized using the Interaction Anxiousness Scale (IAS) for interaction anxiety. Furthermore, Life Course Theory emphasizes that the isolation period during the pandemic constitutes a “critical period” for the academic development of nursing freshmen. In contrast, the Developmental Cascade Model is applied to explain how factors across different systems interact across time and domains, ultimately shaping distinct adaptation trajectories.

Study instruments

Sociodemographic characteristics

The study collected data on the participants’ age, gender, and the type of school they attended (university vs. college). To account for pre-nursing academic backgrounds, we documented students’ high school subject tracks (Physics or History), reflecting China’s “3 + 1 + 2” education reform policy. In which, physics track students focus on STEM subjects (e.g., advanced mathematics, physics), typically qualifying for science/engineering degrees, while history track students emphasize humanities/social sciences (e.g., history, political science), generally pursuing related fields [33]. While all students complete basic coursework in both domains, this tracking system fundamentally shapes their cognitive development and career pathways. Additional variables included whether participants served as class cadres and only-child status (a recognized sociocultural factor in China).

Learning adaption

Learning adaption was measured by the learning adjustment of undergraduate scale, which was developed by Feng et al. [34]. The scale is a 29-item instrument that assesses learning adjustment across five key dimensions: learning motivation, Teaching Models, learning ability, learning attitude, and learning environment. It utilizes a 5-point Likert scale, with 1 indicating “completely disagree” and 5 signifying “completely agree.” Intermediate scores are allocated progressively, and 18 items are reverse-scored to adjust for negatively worded statements. An elevated score in each dimension and in the composite assessment reflects superior levels of learning adaption. The scale has exhibited robust reliability, with an overall Cronbach’s alpha of 0.928, and dimensions showing alpha values between 0.71 and 0.80.

Bioecological framework

Microsystem level

Resilience was measured using the Connor-Davidson Resilience Scale (10-item version), a revision of the original scale by Campbell-Sills in 2007 [35]. This assessment employs a 5-point Likert scale, where “0” corresponds to “never” and “4” denotes “almost always.” The total score on this scale spans from 0 to 40, with higher scores indicating greater resilience. The Cronbach’s alpha coefficient for this study was 0.947, suggesting excellent internal consistency and reliability.

The Mobile Phone Dependency Scale, developed by Xiong et al. [36], comprises 16 items distributed across four dimensions: withdrawal symptoms, salient behavior, social reassurance, and mood change. It applies a 5-point Likert scale, with 1 indicating “very inconsistent” and 5 signifying “very consistent.” The scale’s scores extend from 16 to 80, where an elevated score suggests increased reliance on mobile phones. The Cronbach’s alpha coefficient, calculated at 0.925, reflects a high level of internal consistency and reliability.

Mesosystem level

Parental attachment was assessed using the abbreviated subscale version of Armsden and Greenberg’s Parental and Peer Attachment questionnaire [37], translated by Wang et al. in 2007 [38]. This subscale is composed of 10 items that evaluate three key dimensions of parent-child relationships: communication, trust, and alienation. Responses are measured on a 5-point Likert scale. The aggregate scores for this subscale vary from 10 to 50, with higher scores reflecting a more secure and connected parent-child attachment. The Cronbach’s alpha coefficient for this study was 0.897, demonstrating satisfactory internal consistency and reliability of the measure.

Macrosystem level

Interaction anxiety was assessed using the interaction anxiousness scale (IAS), which is designed to measure the propensity for subjective experiences of social anxiety independent of behavioral manifestations [39]. The IAS comprises 15 self-report items scored on a 5-point Likert scale, ranging from 1 (“not at all like me”) to 5 (“very much like me”). The scale demonstrates excellent internal consistency with a Cronbach’s alpha coefficient exceeding 0.87. Additionally, the eight-week test-retest reliability coefficient is 0.80. Given that our study population consists of first-year students, the item “8. I become anxious during job interviews” was omitted from the scale to ensure relevance to their lived experiences.

Data collection

The survey questionnaire was made accessible through a two-dimensional code or direct link, accompanied by a set of standardized guidelines that outlined the survey’s purpose and necessary precautions. Distribution was facilitated by counselors via WeChat groups. Participants who were interested were invited to engage with the survey; after providing their informed consent, they were granted access to the survey platform. Should a participant not meet the established inclusion and exclusion criteria, their participation was terminated. Participation was strictly voluntary, and participants were empowered to withdraw at any point during the study. All data collected were submitted anonymously, with access restricted to members of our research team for the sole purpose of conducting the study.

Statistical analysis

Latent Profile Analysis (LPA) was conducted using Mplus (version 8.3) to discern distinct profiles of learning adaptation among participants. To address scale inconsistencies across various instruments, data were standardized through z-score transformation. Model parameters were estimated using maximum likelihood within the Latent Class Analysis (LCA) framework, with individuals classified into the group with the highest posterior probability. Model fit was assessed using the Bayesian Information Criterion (BIC), Sample-Size Adjusted BIC (aBIC), and Akaike Information Criterion (AIC). The appropriateness of the latent profile number was ascertained when the average posterior probability was above 0.8, each profile’s proportion exceeded 5%, and the Entropy index was greater than 0.8 [40, 41]. The Bootstrapped Likelihood Ratio Test (BLRT) for K latent profiles was deemed significant with a p-value less than 0.05, suggesting a superior fit for the K-profile LPA model over the (K-1)-profile model.

Categorical data were represented through counts and proportions, with inter-group comparisons conducted using the chi-square test or Fisher’s exact test when appropriate. Continuous data adhering to a normal distribution were detailed using means and standard deviations, and inter-group comparisons were executed using Analysis of Variance (ANOVA). Post-hoc pairwise comparisons utilized the Student-Newman-Keuls (SNK) q test. Covariance analysis was applied to control for confounding variables when evaluating learning adaptation scores across different latent profiles. All statistical analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). A P-value of less than 0.05 was considered indicative of a statistically significant difference.

Results

Descriptive statistics of variables

The descriptive statistics of variables were presented in Table 1. Notably, the Cronbach’s alpha values for all items and scales exceed the threshold of 0.7, signifying adequate internal consistency and reliability.

Table 1.

Descriptive statistics and reliability of variables

Variables	Items	Cronbach α	Scale Mean ± SD	Item Mean ± SD	Max	Min
Learning Adaptation
Learning Motivation	8	0.881	27.01 ± 5.41	3.38 ± 0.68	8	40
Teaching Models	7	0.797	25.61 ± 4.59	3.66 ± 0.66	7	35
Learning Ability	6	0.872	21.14 ± 4.11	3.52 ± 0.68	6	30
Learning Attitude	4	0.709	14.98 ± 2.81	3.75 ± 0.70	4	20
Learning Environment	4	0.763	13.13 ± 3.11	3.28 ± 0.78	4	20
Overall	29	0.938	101.88 ± 16.44	3.51 ± 0.57	41	145
Resilience
Overall	10	0.955	35.33 ± 6.92	3.53 ± 0.69	10	50
Mobile Phone Addiction
Withdrawal Symptoms	6	0.839	17.07 ± 4.52	2.85 ± 0.75	6	30
Salient Behavior	4	0.86	9.74 ± 3.23	2.44 ± 0.81	4	20
Social Reassurance	3	0.83	8.59 ± 2.57	2.86 ± 0.86	3	15
Mood Change	3	0.754	8.08 ± 2.56	2.69 ± 0.85	3	15
Overall	16	0.936	43.48 ± 11.45	2.72 ± 0.72	16	80
Parental Attachment
Communication	3	0.833	9.97 ± 2.57	3.32 ± 0.86	3	15
Trust	3	0.895	10.63 ± 2.55	3.54 ± 0.85	3	15
Alienation	4	0.845	9.79 ± 3.37	2.45 ± 0.84	4	20
Overall	10	0.884	30.40 ± 4.08	3.04 ± 0.41	10	50
Interaction Anxiety
Overall	14	0.769	42.85 ± 6.89	3.06 ± 0.49	14	70

SD, standard deviation

In terms of scores, the aggregate mean score for learning adaptation is 3.51 ± 0.57, signifying a positive level of learning adaptation among the participants. Within this domain, the dimension of learning attitude exhibits the highest mean item score at 3.75 ± 0.70, and the learning environment dimension has the lowest mean item score at 3.28 ± 0.78. resilience scores are aggregated to a mean of 3.53 ± 0.69, reflects a relatively high level of resilience. Parental attachment and interaction anxiety both exhibit mean scores that are indicative of a moderate degree of these constructs, with means of 3.04 ± 0.41 and 3.06 ± 0.49, respectively. Of particular interest, the alienation subscale of parental attachment registers the lowest mean score at 2.45 ± 0.84. in the context of mobile phone addiction tendencies, the overall mean score is 2.72 ± 0.72. the dimension of salient behavior is marked by the lowest mean of 2.44 ± 0.81, while social reassurance shows the highest mean score of 2.69 ± 0.85.

Results of latent profile analysis

The study used the Harman single factor for common method bias. The results showed that 12 common factors with eigenvalues > 1 were precipitated from the EFA and the maximum common factor had an explanatory rate of 31.92% (< 40% critical value). Therefore, the relationship between variables in this study was less affected by common method bias.

In this study, resilience, parental attachment, interaction anxiety, and tendencies towards mobile phone addiction were subjected to standardized transformation using z-scores prior to LPA. The fit indices for five potential LPA models were presented in Table 2. While the AIC, BIC, and aBIC values decreased with an increasing number of Profile es, the fifth profile in the five-profile LPA model contained only eight cases, leading to its exclusion due to insufficient representation.

Table 2.

Fit statistics for the latent profile analysis

Latent Profile	Loglikelihood	AIC	BIC	aBIC	Entropy	BLRT (P)	Cases (%)	Average Proportion (%)
1	-6638.631	13293.263	13333.781	13308.370	—	—	C1 = 1170(100%)	C1 = 1.00
2	-6212.926	12451.851	12517.693	12476.400	0.878	< 0.001	C1 = 206(17.61%)/ C2 = 964(82.39%)	C1 = 0.91/ C2 = 0.98
3	-6108.977	12253.955	12345.121	12287.946	0.901	< 0.001	C1 = 207(17.69%)/ C2 = 903(77.18%)/ C3 = 60(5.13%)	C1 = 0.93/ C2 = 0.97/ C3 = 0.87
4	-6027.839	12101.678	12218.168	12145.112	0.885	< 0.001	C1 = 64(5.47%)/ C2 = 826(70.60%)/ C3 = 214(18.29%)/ C4 = 66(5.64%)	C1 = 0.91/ C2 = 0.96/ C3 = 0.85/ C4 = 0.88
5	-5972.140	12000.281	12142.094	12053.157	0.902	< 0.001	C1 = 113(9.66%)/ C2 = 217(18.55%)/ C3 = 762(65.13%)/ C4 = 70(5.98%)/ C5 = 8(0.68%)	C1 = 0.92/ C2 = 0.87/ C3 = 0.96/ C4 = 0.93/ C5 = 0.99

AIC, Akaike Information Criterion; BIC, Bayesian Information Criterion; BLRT, Bootstrapped Likelihood Ratio Test

Consequently, the four-profile LPA model was selected for this study. This model demonstrated a significant BLRT with a p-value of less than 0.05 and an Entropy index greater than 0.885, indicating a clear and reliable classification. Additionally, the average posterior probability for the four latent profiles was greater than 0.8, further supporting the robustness of the classification. Table 4 details the relationship between the four profiles and the variables of resilience, attachment to parents and peers, interaction anxiety, and tendencies towards mobile phone addiction.

Table 4.

Comparative analysis of bioecological attributes among four profiles

Variables	Profile 1 (n = 64)	Profile 2 (n = 826)	Profile 3 (n = 214)	Profile 4 (n = 66)	F	P
Resilience	29.67 ± 8.67	33.35 ± 5.04a	40.68 ± 5.82ab	48.17 ± 3.56abc	256.648	< 0.001
Mobile Phone Addiction	65.52 ± 7.11	47.20 ± 5.20a	29.20 ± 6.01ab	21.89 ± 6.92abc	1251.308	< 0.001
Parental Attachment	29.06 ± 5.59	29.93 ± 3.39	30.79 ± 3.68a	36.36 ± 6.17abc	61.984	< 0.001
Interaction Anxiety	54.08 ± 6.67	43.87 ± 4.72a	39.95 ± 6.05ab	28.73 ± 6.34abc	294.100	< 0.001

After pairwise comparisons, ^a was significantly different from Profile 1; ^b was significantly different from Profile 2; ^c was significantly different from Profile 3, P < 0.05

Demographic characteristics of participant profiles

The demographic characteristics of the participants were analyzed to identify any significant differences among the four latent profiles, as shown in Table 3. Notably, the chi-square test revealed significant disparities in the distribution of educational type (χ2 = 21.797, p < 0.001) and high school subjects studied (χ2 = 23.012, p < 0.001) across the profiles. This divergence could imply that the educational environment and academic focus during high school might be linked to the psychological profiles that influence learning adaptation.

Table 3.

Demographic characteristics of participants (n = 1170)

Variable	Number of Cases	Proportion (%)	Profile 1 (n = 64)	Profile 2 (n = 826)	Profile 3 (n = 214)	Profile 4 (n = 66)	χ2	P
Age (Years)
16 ~ 17	38	3.25	0(0.00)	26(3.15)	10(4.67)	2(3.03)	6.02	0.738
18	666	56.92	42(65.63)	471(57.02)	119(55.61)	34(51.52)
19	399	34.1	19(29.69)	281(34.02)	74(34.58)	25(37.88)
20	67	5.73	3(4.69)	48(5.81)	11(5.14)	5(7.58)
Gender
Male	171	14.62	7(10.94)	120(14.53)	29(13.55)	15(22.73)	4.373	0.224
Female	999	85.38	57(89.06)	706(85.47)	185(86.45)	51(77.27)
School type
College	246	21.03	12(18.75)	199(24.09)	33(15.42)	2(3.03)	21.797	< 0.001
University	924	78.97	52(81.25)	627(75.91)	181(84.58)	64(96.97)
Class Cadre
Yes	925	79.06	50(78.13)	656(79.42)	171(79.91)	48(72.73)	1.789	0.617
No	245	20.94	14(21.88)	170(20.58)	43(20.09)	18(27.27)
High School Subjects
Physics	262	22.39	14(21.88)	189(22.88)	45(21.03)	14(21.21)	23.012	< 0.001
History	501	42.82	23(35.94)	378(45.76)	85(39.72)	15(22.73)
Others	407	34.79	27(42.19)	259(31.36)	84(39.25)	37(56.06)
Only Child
Yes	932	79.66	48(75.00)	673(81.48)	161(75.23)	50(75.76)	5.748	0.125
No	238	20.34	16(25.00)	153(18.52)	53(24.77)	16(24.24)

Description of bioecological attributes among four profiles

Figure 2; Table 4 illustrates the distribution of standardized scores of bioecological attributes among four profiles. The analysis reveals significant differences (P < 0.001) across all variables: resilience, parental attachment, interaction anxiety, and mobile phone addiction. Notably, a clear gradient in scores is observed among the profiles, indicating a systematic variation in these attributes that may underlie distinct patterns of learning adaptation.

Fig. 2.

Standardized scores of bioecological attributes among four profiles

Association of four profiles with learning adaptation

After adjusting for confounding factors, significant statistical differences (P < 0.001) were observed across all learning adaptation dimensions for the four profiles (Fig. 3). Profile 4 demonstrated the highest scores in every assessed category and overall learning adaptation, followed by Profile 3 and Profile 2, with Profile 1 showing the lowest scores. Details were shown in Table 5. The profiles have been designated with specific labels that reflect their learning adaptation characteristics: ‘Struggling Learners’ (Profile 1, 5.47%), ‘Moderate Engagers’ (Profile 2, 70.60%), ‘Adaptable Strivers’ (Profile 3, 18.29%), and ‘Optimal Adapters’ (Profile 4, 5.64%).

Fig. 3.

Standardized scores of learning adaption among four profiles

Table 5.

Comparative analysis of learning adaptation among four profiles

Learning Adaptation	Mean ± SD							Mean ± SD *
	Profile 1 (n = 64)	Profile 2 (n = 826)	Profile 3 (n = 214)	Profile 4 (n = 66)	F	P		Profile 1 (n = 64)	Profile 2 (n = 826)	Profile 3 (n = 214)	Profile 4 (n = 66)	F	P
Learning Motivation	22.42 ± 6.48	25.59 ± 3.86a	30.92 ± 5.11ab	36.65 ± 4.39abc	221.908	< 0.001		22.73 ± 0.61	25.99 ± 0.31a	31.17 ± 0.40ab	36.56 ± 0.59abc	204.887	< 0.001
Teaching Methods	23.22 ± 4.20	24.34 ± 3.84a	29.31 ± 3.96ab	31.76 ± 3.81abc	157.765	< 0.001		23.05 ± 0.55	24.26 ± 0.28a	29.14 ± 0.36ab	31.50 ± 0.54abc	148.291	< 0.001
Learning Ability	18.56 ± 4.24	20.20 ± 3.39a	23.71 ± 3.83ab	27.14 ± 4.01abc	130.152	< 0.001		19.01 ± 0.51	20.69 ± 0.26a	24.14 ± 0.33ab	27.36 ± 0.49abc	120.252	< 0.001
Learning Attitude	12.94 ± 3.17	14.30 ± 2.43a	17.13 ± 2.22ab	18.53 ± 1.90abc	141.43	< 0.001		12.77 ± 0.34	14.16 ± 0.17a	16.93 ± 0.22ab	18.26 ± 0.33abc	132.731	< 0.001
Environmental Factors	11.11 ± 3.73	12.44 ± 2.49a	15.07 ± 3.05ab	17.50 ± 3.13abc	122.944	< 0.001		10.96 ± 0.39	12.30 ± 0.20a	14.95 ± 0.25ab	17.44 ± 0.38abc	122.31	< 0.001
Overall	88.25 ± 16.93	96.87 ± 11.63a	116.14 ± 13.85ab	131.58 ± 13.01abc	287.587	< 0.001		88.52 ± 1.77	97.41 ± 0.90a	116.33 ± 1.16ab	131.12 ± 1.72abc	269.382	< 0.001

* Adjusted for age, gender, type of school, class cadre, high school subjects, and only child

After pairwise comparisons, ^a was significantly different from profile 1; ^b was significantly different from profile 2; ^c was significantly different from profile 3, P < 0.05

Discussion

Principal results

This study identified four distinct adaptation profiles among nursing freshmen during the pandemic, highlighting the interplay of individual (resilience, phone addiction), familial (parental attachment), and societal (interaction anxiety) factors. These findings align with Bronfenbrenner’s bioecological model, demonstrating how micro-, meso-, and macrosystem factors collectively shape learning adaptation.

Our findings indicate that students’ adaptation scores were at a moderate level, consistent with Oliveira Silva [42]. The positive learning attitude of first - year students, as noted by Hou et al. [43], likely contributed. Higher scores might be due to future - oriented nursing career goals during COVID − 19, though this needs more research. Despite this, first year students still faced learning environment adaptation challenges [44].

This study uncovered significant heterogeneity among factors influencing learning adaptation. Statistical analysis revealed four distinct latent profiles with marked differences in adaptation levels. The heterogeneity among students implies that a one - size - fits - all approach to educational interventions during the pandemic would be ineffective. Our findings align with Shi et al. [45] research, who, in the context of MOOCs (Massive Open Online Courses), identified three behavioral attributes - effort regulation, self - assessment, and learner engagement - crucial for student success. This indicates that, similar to the diverse learning environment of MOOCs, in our study of nursing freshmen during the pandemic, multiple factors are at play in determining learning adaptation. We also found that school type and high - school academic track impact learning adaptation classification. As Ding et al. [46] noted, students from different academic backgrounds vary in adaptation, especially in nursing’s science - heavy first year. Recognizing this heterogeneity, categorizing students into profiles is crucial. It allows for targeted educational strategies, as a one - size - fits - all approach won’t address the diverse needs stemming from differences in individual traits and academic histories.

The ‘Struggling Learners’ subgroup, comprising 5.47% of the participants, demonstrated the lowest levels of learning adaptation. Despite being a relatively small proportion, their situation is of great concern as they are at risk of academic failure or dropping out. The pandemic-induced isolation was a unique backdrop that significantly influenced the learning adaptation of nursing freshmen, and mobile phone addiction was one aspect of this complex picture. Drawing from developmental theories, such as Mandleco and Peery’s [47] resilient system model, their low resilience levels make it arduous for them to overcome the stress of isolation and online learning. Bowlby’s [48] attachment theory suggests that their parental attachment could be either too weak, failing to offer essential emotional support, or overly strong, resulting in over - dependence and self - regulation difficulties in learning. In line with a meta- analysis [49] indicating a positive correlation between social anxiety and mobile phone addiction, this group likely experiences high interaction anxiety and mobile phone addiction, further impeding their learning progress.

In contrast, the ‘Moderate Engagers’, who make up 70.60% of the sample, show a middle - ground adaptation. Their resilience, as per the resilient system model, is at a moderate level, allowing them to cope moderately well with the educational changes. Parental attachment, in accordance with family - related theories, provides sufficient but not outstanding support. Their interaction anxiety and mobile phone use are within a manageable range, enabling them to engage in learning without significant disruptions.

The ‘Adaptable Strivers’, constituting 18.29% of the participants, exhibit good learning adaptation. Their relatively high resilience, as predicted by developmental frameworks, helps them adjust effectively to the new educational environment. McLoyd’s [50] Family Stress Model implies that their healthy parental attachment offers a stable base for learning. With low interaction anxiety, they can actively participate in online learning, and they manage their mobile phone use proficiently, leveraging technology for learning rather than being hindered by it.

Finally, the ‘Optimal Adapters’, representing 5.64% of the group, achieve an outstanding level of learning adaptation. Exceptionally high resilience, ideal parental attachment, minimal interaction anxiety, and negligible mobile phone addiction tendencies are characteristic of this subgroup. Their academic background and school type may also play facilitating roles, further enhancing their overall adaptation.

Our study underscores the complex interplay of these bioecological attributes in shaping student adaptation. Resilience and parental attachment are consistently associated with positive educational outcomes, while interaction anxiety and mobile phone addiction can have a negative impact, though the relationship is complex. For example, while some research [51] shows a negative link between social anxiety, mobile phone addiction, and academic performance, other studies [52] suggest a more nuanced relationship where, in certain cases, increased anxiety and mobile phone addiction might be associated with better performance, highlighting the need for a more in - depth understanding of these factors within each subgroup.

Practical implications for nursing education

Identifying distinct profiles of learning adaptation among nursing freshmen during the pandemic provides guidance for developing targeted nursing curriculum design, optimization of student support, and mental health interventions.

The significant impact of mobile phone addiction and interaction anxiety on learning adaptation highlights the need to integrate technology regulation and communication skill training into the nursing curriculum [53]. Drawing on the principle of “theory-practice integration” in nursing education, two actionable adjustments are proposed: First, embed “digital literacy modules” in fundamental nursing courses [54]. For instance, in the Introduction to Nursing course, add corresponding class hours of the “Balancing Virtual Learning and Digital Well-being” module. This module teaches time management strategies for online resources and self-assessment tools for mobile phone dependency, and it addresses the high prevalence of technology-related learning distractions among nursing freshmen. Such distractions have been linked to impaired cognitive attention and academic performance in post-pandemic educational research [55]. Second, optimize simulation-based clinical training to reduce social anxiety. Research indicates that 42.1% of nursing students experience moderate social anxiety, and this anxiety directly impairs their performance in clinical skill operations and interactions with patients [56]. We recommend revising the Clinical Communication course through the following measures: (1) reducing the size of observation groups to 3–4 students (instead of the conventional 12) to minimize evaluation pressure [57]; (2) incorporating “low-stakes practice sessions” that do not involve video recording or external observers to foster a psychologically safe learning environment [58]; and (3) utilizing role-playing exercises that gradually increase the complexity of social interactions (e.g., progressing from peer practice to simulated patient interactions). These adjustments align with the need to address “disruptions in professional interactions,” which is a key challenge in post-pandemic clinical learning environments identified by qualitative studies [59].

The critical role of parental attachment and resilience in positive adaptation underscores the necessity of establishing a multilayered student support system involving educators, families, and institutional resources [60]. At the educator level, train nursing faculty to serve as “frontline support navigators” through targeted workshops. As identified in recent studies, nursing teachers can effectively detect psychological distress through changes in academic performance and social withdrawal; however, they often lack structured strategies to address issues such as social anxiety and mobile phone addiction [61]. We propose implementing mandatory “Mental Health First Aid” training for all nursing instructors, focusing on nonviolent communication techniques and crisis identification checklists [62]. For instance, an instructor who notices a student’s persistent absence from clinical rotations could initiate a private conversation using phrases like, “I’ve noticed you’ve missed recent sessions—would you like to share any challenges you’re facing?” to create a safe space for disclosure. This aligns with recommendations to strengthen educators’ capacity to address post-pandemic psychological challenges. At the family level, establish a “parental attachment enhancement program” for first-year nursing students. Given that social anxiety mediates the relationship between psychological stress and mobile phone addiction, parental support acts as a critical buffer against such cascading effects [63]. Organize quarterly online seminars for families covering topics such as “Supporting Nursing Students’ Transition to Clinical Training” and “Recognizing Signs of Academic Burnout [64]. " Provide families with a digital resource kit containing articles, short videos, and contact information for the university’s wellness center to facilitate collaborative support.

The heterogeneous nature of the adaptation profiles demonstrates that one-size-fits-all interventions are ineffective, necessitating tailored strategies that target specific risk factors. For students with high interaction anxiety, implement a cognitive-behavioral intervention that addresses the mediating role of loneliness in the relationship between social anxiety and mobile phone addiction [65]. A structured psychological education program—combining psychoeducation on the mechanisms of social anxiety and attention externalization training—has been shown to reduce anxiety scores and improve learning performance among nursing students. We recommend adapting this intervention for first-year students: include interactive exercises (e.g., identifying automatic negative thoughts during patient interactions) and homework assignments (e.g., gradual exposure to small group discussions), and use the Interaction Anxiousness Scale (IAS) to track progress. For students with comorbid low resilience and high mobile phone addiction, develop an integrated “resilience building and digital detox” program. Given that mobile phone addiction negatively predicts safety behaviors in clinical settings [66], the program should combine resilience training e.g., “Developing realistic goals during stressful periods”) with structured digital detox activities (e.g., 2-hour “device-free” blocks during clinical simulations) [67]. Pair students with senior nursing mentors who have completed the program to provide peer support, as peer role models have been shown to enhance intervention adherence [68]. Institutionalize these interventions by embedding them in the services of the university’s wellness center, with clear referral pathways: faculty identify at-risk students through the aforementioned training, refer them to the center, and receive follow-up updates (while maintaining confidentiality).

Strength and limitations

A notable strength of this research is its application of a person-centered methodology, which effectively captures the heterogeneous nature of learning adaptation among nursing students and underscores the necessity of tailored educational interventions.

Notwithstanding these strengths, the study has limitations. The primary measures were based on self-report, which may be subject to social desirability bias; the inclusion of objective data in future work would strengthen the findings. Furthermore, the initial stage of our multi-stage sampling involved non-probability sampling for institutional recruitment. This approach, while practical under pandemic constraints, means our sample may not be fully representative of the broader population of nursing freshmen in terms of geographic distribution and institution types. This potential lack of representativeness, coupled with the small size of some identified profiles and unmeasured variations in curricular implementation across schools, suggests that caution should be exercised in generalizing the results. Future research with more robust, probability-based sampling designs is recommended.

Conclusion

In conclusion, this study offers a nuanced perspective on the learning adaptation of nursing freshmen during pandemic-induced isolation. It highlights the importance of a bioecological understanding of student adaptation and the need for tailored educational interventions. As the nursing profession continues to evolve, the insights provided by this research will be instrumental in shaping a resilient and adaptable future workforce.

Author contributions

Huanhuan Huang, Xudong Tian, Qi Huang, Zhiyu Chen, Xin Yang, Jiao Tang, Wanyu Tang, Yetao Luo contributed to the conceptualization and design of the study. Huanhuan Huang and Zhiyu Chen were responsible for the methodology and formal analysis. Xudong Tian, Zhiyu Chen, and Jiao Tang contributed to the investigation and resources. Xin Yang and Wanyu Tang performed the data curation. Huanhuan Huang and Yetao Luo were in charge of writing the original draft. Jiao Tang and Huanhuan Huang contributed to the review and editing of the manuscript. All authors have read and agreed to the published version of the manuscript.

Funding

This study was funded by Chongqing Municipal Education Commission's l4th Five-Year Key Discipline Support Project (approval number: 20240102; 20240202). However, the funders had no role in the study design, data collection, management, analysis or interpretation, manuscript writing or the decision to submit the report for publication.

Data availability

The data and materials are available from the corresponding author.

Declarations

Ethics approval and consent to participate

This study was approved by the Medical Ethics Committee of The First Affiliated Hospital of Chongqing Medical University (No. 2021-430). It adhered to the principles outlined in the Declaration of Helsinki. After explaining the anonymity and confidentiality of participation, nursing freshmen were informed that they could refuse to participate or withdraw from participation at any time without penalty. All participants provided written informed consent.

Consent to publish

Not applicable.

Competing interests

The authors declare no competing interests.