What are the predominant factors predicting trend of post-traumatic growth for patients after bone tumour surgery? A longitudinal observational study

Abstract

Aim

To investigate the predominant factors predicting trend of post-traumatic growth for patients after bone tumour surgery.

Methods

We conducted a longitudinal observational study with convenience sampling. Follow-up examinations were conducted every two weeks over a total follow-up period of one month. Post-traumatic growth (PTG) levels were measured using the Chinese version of the Post-Traumatic Growth Inventory (PTGI-C). Latent Category Growth Model (LCGM) was applied to explore the latent trajectory categories of PTG. Univariate and multivariate analyses were conducted to explore the predictors and influencing factors for patients with different trajectories.

Results

A total of 153 patients were included in the study, and 141 patients completed the follow-up. The means of PTGI-C at different monitoring time were 41.24 ± 21.94, 44.56 ± 23.33, and 40.69 ± 22.63, respectively. Repeated measures analysis of variance with univariate revealed statistically significant differences in PTGI-C across different time (F = 3.372, P = 0.02). Patients were categorized into two latent subgroups: an increasing trend group (intercept = 44.276, slope = 5.152, P < 0.001) and a decreasing trend group (intercept = 35.164, slope = − 4.718, P < 0.001) based on two-category LCGM model. Low levels of post traumatic stress (vs. high levels) were associated with increased odds of decreasing PTG trajectory (OR 2.67, P = 0.015), indicating a risk factor. Unmarried status (vs. married) and giant cell tumour (OR = 0.223, P = 0.047), fibrogenic tumours (OR = 0.127, P = 0.007), cartilaginous origin tumour (OR = 0.194, P = 0.034), and other tumours (OR = 0.169, P = 0.023, vs. osteosarcoma) were protective factors.

Conclusion

The classification of PTG change trajectory suggested different positive psychological processes in patients with bone tumours. Identifying the influencing factors of patients in the PTG decline trend group can guide nurses to provide effective intervention in time.

Clinical trial number

ChiCTR2300074170 (Registration Date: 1st August 2023).

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s40359-026-05047-7.

Background

Bone tumours, which encompass diverse mesenchymal tumours originating from bone, cartilage, and connective tissues, represent a highly heterogeneous group ranging from indolent to aggressive and metastatic [1, 2]. These tumours were relatively rare, accounting for 2% of all human neoplasms. According to the latest Cancer Statistics report, there were an estimated 3770 new cases of bone and joint cancers in the United States, resulting in 2,190 deaths [3]. The overall age-standardized rate for malignant neoplasm of bone and articular cartilage is 1.82/100,000 in China [4]. Notably, advancements in treatment have placed the management of this disease on the brink of progress, leading to improved survival outcomes for patients with bone tumours [5–6]. The 5-year overall survival rate for individuals with malignant bone tumours has significantly increased, with current rates ranging from 65% to 70%, a substantial improvement from the previous rate of 20% [5, 7].

Surgery was a crucial component of treatment for nearly all bone tumour patients, involving limb salvage and reconstruction techniques utilizing modular tumour endoprostheses [8–11]. In cases where the disease progresses following primary treatment, the panel recommends considering radiotherapy and/or primary site surgery to address local control or provide palliative care [11]. However, surgery itself was a traumatic procedure, characterized by a sizable incision and associated risks of complications (approximately 15.4%~27.27%), such as periprosthetic joint infection, periprosthetic bone fracture, and wound infection [8, 12, 13]. Additionally, patients underwent long-term rehabilitation exercises following surgery [13]. Some of them would experience the impaired body image, perceived control, and self-esteem problems after surgery [14–17]. In particular, patients with malignant bone tumours such as osteosarcoma often experience severe physical and mental burden when facing the dramatic changes in body image caused by amputation [14, 15] They often adopt avoidant coping strategies and tend to be socially isolated [14].

Malignant bone tumour patients often underwent chemotherapy/radiotherapy following surgery, which frequently led to adverse reactions like nausea, vomiting, and bone marrow suppression [18–21]. These side effects, especially from chemotherapy significantly impacted patient comfort and gave rise to a range of psychological problems, thereby diminishing quality of life and impeding the recovery process [19–21]. A great number of patients with cancer were prone to experiencing psychological disturbances with high prevalence rates during the treatment. Fauske et al. conducted a longitudinal study of 18 survivors with bone tumours found that only three participants felt they returned to a life similar to that before their illness, while 15 believed that their lives and bodies had undergone significant changes, with five falling into both psychological and physiological distress, experiencing decreased physical comfort and increased pain [15]. Luzzati et al. investigated the psychological status of 149 postoperative spinal tumour patients, and found that 49 participants had depressive symptoms (constituting 32.9%), among which 24 (constituting about 49.0%) were mildly depressed, six (constituting about 12.5%) were moderately depressed, and 18 (constituting about 37.5%) were severely depressed [16]. Negative emotions can affect the patient’s recovery process, and there was a possibility of inducing certain adverse reactions or complications which would impair quality of life [22, 23].

After experiencing a series of medical interventions such as diagnosis and treatment, patients not only experience negative psychological states, including the mental health problems but also positive changes or growth, such as mental wellbeing [24]. Positive psychology was a science that focused on human virtues, strengths, and other positive qualities, and studies positive emotional experiences, cognitive processes, personality traits, creativity, talent development, and other related topics [25]. Post-traumatic growth (PTG) was a part of positive psychology. It referred to the collection of psychological positive changes experienced by individuals after struggling with traumatic events or situations, first proposed by American scholars Tedeschi and Calhoun in 1996 [24]. The factors related to PTG mainly focused on demographic characteristics, personality traits, coping strategies, and others [26]. At the same time, research has also found that individual positive psychology, such as PTG, were supposed to improve their post traumatic status and promote positive changes in both physical and mental health [27]. The population of cancer patients who experienced PTG had a better prognosis, with significantly higher quality of life and disease recovery compared to those who did not experience PTG [28, 29].

At present, the research on the psychological status of patients with bone tumour mainly focuses on the incidence of mental health problems, such as anxiety and depression, and the influencing factors of each negative psychological state [16]. At the same time, the relationship between psychological status and rehabilitation was also reported, but it did not focus on the psychological status and positive changes in the short term after surgery [23]. Based on the functional-descriptive model of post-traumatic growth (PTG) proposed by Tedeschi and Calhoun [30], PTG is the positive psychological change that emerges from an individual’s struggle with highly challenging life events, such as bone tumour diagnosis and surgery. PTG is not merely the absence of distress but arises through cognitive processing, particularly deliberate rumination, that reconstructs shattered cognitive schemas [30]. In this study, we operationalised this framework by examining influencing factors of PTG through four dimensions: patients’ demographic characteristics, the impact of surgery on the individual, individual coping traits, and somatic symptoms (pain), thereby exploring how these factors collectively shape PTG following bone tumour surgery. By applying latent class growth modelling (LCGM) to analyze short-term follow-up data, this study differs from existing research by identifying distinct PTG trajectories and their predictors following bone tumour surgery. These findings will inform the development of targeted, timely psychosocial interventions to improve patients’ mental wellbeing.

Materials and methods

Study design and Settings

This longitudinal observational study was conducted between December 2019 and December 2020 at a tertiary care hospital. Participants were recruited from patients attending the orthopaedic oncology outpatient clinic. A convenience sampling method was employed. Ethical considerations were upheld throughout the study, adhering to the principles outlined in the Declaration of Helsinki. The study received approval from the West China Hospital Ethics Committee (approval number: 2019-Review-1091) and all participants provided informed consent by signing consent forms. This study has been registered on https://www.clinicaltrials.gov/. with the number of ChiCTR2300074170 (Date: 1st August 2023).

Participants

Eligible participants had a histopathologically confirmed primary bone tumour (benign or malignant) located in the appendicular skeleton. This included the upper limb (hand, radius, ulna, humerus) and lower limb (foot, tibia, fibula, femur). For long bones, tumours could originate from the epiphysis, metaphysis or diaphysis; for short bones (e.g., hand or foot), tumours from any anatomical level were included. All participants underwent limb-salvage en bloc tumour resection (amputation cases were excluded).

Inclusion and exclusion criteria

Inclusion criteria were: (1) age ≥ 18 years; (2) clear consciousness, defined as the ability to provide accurate responses to verbal questioning and to complete self-administered questionnaires; (3) primary bone tumour confirmed by histopathology; (4) tumour location as specified above; (5) scheduled limb-salvage surgery. Exclusion criteria were: (1) mental illness or cognitive impairment hindering cooperation with the research; (2) concurrent heart, liver or renal insufficiency, or other malignant diseases (e.g., other metastatic cancer, leukaemia); (3) previous bone tumour surgery on the same limb; (4) pathological fracture at presentation requiring emergency surgery. During follow-up, participants were excluded if they underwent re-operation due to disease deterioration, experienced other major stressful life events, withdrew voluntarily or had more than 20% missing data.

Sample size calculation

The main outcome measure in this study was a continuous variable, for which the sample size was calculated using the following formula.

Z_α/2 was set at 1.96, and Z_β was set at 1.28. Based on preliminary pilot testing, the estimated value for σ was determined to be 10, while Δ was set at 3. The sample size was then computed using PASS15.0 software, resulting in a required sample size of 117 participants. Accounting for an anticipated attrition rate of 20%, the sample was expanded to 146 participants.

Follow-up and data collection

Follow-up examinations, including telephone or WeChat interviews, as well as physical examinations at the Outpatients Department, were conducted every two weeks. Data were collected at four time points: three days before surgery (T1), two days after surgery (T2), 15 days after surgery (T3), and 30 days after surgery (T4). The total follow-up period lasted one month.

Measurement

Patients were asked to fill out several questionnaires. The questionnaires included five parts, which recorded: (1) General information questionnaire; (2) Chinese version of the post-traumatic growth Inventory (PTGI-C); (3) The Chinese translation of the Impact of Event Scale-Revised (CIES-R); (4) The Chinese Trait Coping Style Questionnaire (TCSQ); and (5) Number Rating Scale (NRS).

General information questionnaire

The general information questionnaire was designed by the researchers which was mainly used to collect demographic information and disease-related information on the participants. The demographic information includes age, gender, Body Mass Index (BMI), education, occupation, economic condition, marital status and religious beliefs. Disease-related information covers tumour type, tumour size, tumour location, type of surgery, length of stay, etc. We collected this data when patients were hospitalized (T1).

Chinese version of the post-traumatic growth inventory (PTGI-C)

PTGI-C was used to assess the positive outcomes reported by people who have experienced a traumatic event. The original scale was developed by foreign scholars Tedeschi and Calhoun. It was introduced into China in 2008 by Chinese scholars Wang Ji et al. The content of the scale was translated and adapted [31], and has been validated in patients with bone tumours, with a Cronbach’s alpha coefficient of 0.91 and a goodness-of-fit index of 0.90 [32]. It contained 21 entries divided into five dimensions: relevance to others, new possibilities, personal strength, spiritual change and appreciation of life [31]. The scale was based on a 6-point Likert scale, with each item ranging from “not at all” to “very much” corresponding to a score of 0 to 5, and a total score of 0 to 100, with higher scores indicating higher levels of PTG [31]. We collected the PTGI-C in 4 time (T1-T4).

The Chinese translation of the impact of event scale-revised (CIES-R)

CIES-R was a widely used self-assessment scale that measures the degree to which an individual was psychologically affected after experiencing a traumatic event [33]. The original scale was developed by Horowitz et al. and it was revised by Weiss and Marmar. We used the revised version was translated in Chinese by Guo Suran et al. with the internal consistency and half-score reliability of the scale was > 0.8, and the validity scale correlation validity was 0.55 [34]. It consisted of 3 sub-dimensions, namely intrusion, avoidance and high arousal, and contains 22 entries [34]. Scoring was based on a 5-point Likert scale with a total score of 0 to 88. A score of 0 to 8 indicates no clinical symptoms of post traumatic stress ( PTS), 9 to 25 the presence of mild symptoms, 26 to 43 moderate symptoms, and ≥ 44 severe symptoms [35]. A score > 26 indicates significant PTS symptoms, and a score > 35 can be used as a criterion for diagnosing PTSD [34]. Patients’ CIES-R scores were collected 3 days before surgery and 2 days after surgery (T1 and T2). In our study, patients were divided into a low (languishing) and high (flourishing) PTS group based on their CIES-R score two days after surgery using a cut-off score of 26 for the analysis of influencing factors. It indicated the influence of surgery mainly when patients just experienced the surgery.

The Chinese trait coping style questionnaire (TCSQ)

TCSQ was a questionnaire developed by Chinese scholars to assess the coping style of individuals which was used in this study to assess the type of coping style adopted by patients in response to the surgical event. It was one of the influencing factors of PTG to be explored. When confronted with a stressful event that may act as a stressor (in this study, surgery was the stressful event), individuals tended to adopt different cognitive, emotional or behavioural response strategies, which are defined as coping styles [36]. Researchers have categorized these coping styles into positive and evasive type [36]. Both positive and evasive coping are relatively stable coping strategies that existed in individuals and were related to personality traits [37]. The Cronbach’s α for the positive and evasive coping dimensions were 0.70 and 0.69, respectively, with a calibration correlation validity of 0.60 [38]. They suggested that the instrument had good reliability and validity. We collected this data when patients were hospitalized (T1).

Number rating scale (NRS)

This scale was the most widely used unidimensional scale. The scale consisted of 11 points from 0 to 10, representing a score of 0–10. 0 meant no pain at all while10 meant severe pain. The score increased from small to large to indicate progressively increasing pain. Patients chose the value that best matched the pain they felt at that moment, based on how they experienced it. Pain was one of the influencing factors of PTG explored in this study which we collected in both 4 times (T1-T4).

Statistics

Data were analysed using SPSS software (version 26.0; SPSS Inc., Chicago, IL, USA) and MPLUS 8.4. The statistical description of the measures was done by means ± standard deviation, median and interquartile spacing. Statistical analysis was done by t-test, analysis of covariance, rank sum test and other tests. For categorical data, statistical descriptions such as composition ratios and rates were used, and statistical analyses were performed using chi-square tests and fisher’s exact probability method. We used latent class trajectory models to identify trajectories of PTG over time. This was a specialized form of finite mixture modelling and was designed to identify latent classes of individuals following similar progressions of a determinant over time or with age. Due to the small sample size, the χ² test was used to identify possible predictors, and differences were considered statistically significant at P ≤ 0.10 to avoid omission, with the remaining statistical test level α = 0.05. One of the screening methods for logistic regression variables was to first screen variables from a large number of potential independent variables (based on previous studies and our own clinical experience) by designing a more relaxed Type I error rate through univariate statistical analysis [39].The fit indicators for the Latent Class Growth Model (LCGM) included the Akaike Information Criterion (AIC), Bayesian information criterion (BIC), sample-corrected Bayesian information criterion (aBIC), entropy, LoMendell likelihood ratio test (LMR)likelihood ratio test (LRT) and Bootstrap-based likelihood ratio test (BLRT). The best model was determined by combining practical implications and statistical indicators. Finally, logistic regression was used to analyze the factors influencing PTG in patients with different latent trajectories.

Results

According to the inclusion and exclusion criteria of this study, a total of 153 study subjects who met the criteria were included. During the follow-up period of 1 month after surgery, a total of 12 patients were lost to follow-up, with a loss rate of 7.8%. The reasons for loss of follow-up were as follows: change of condition within 1 month after surgery, including serious post-operative complications (2 cases) and poor wound healing (1 case). Patients could not be contacted after discharge and did not attend regular outpatient check-ups (4 cases). Refusal to accept follow-up surveys for personal reasons (4 cases). And loss of contact due to sudden natural disasters in the area (1 case). All of them were excluded from the study, and a total of 141 patients were included.

General information of participants

A summary of the demographic and clinical characteristics of the 141 participants is presented in Table 1. In brief, the sex distribution was balanced (50.4% male, 49.6% female), with a mean age of 33.48 ± 12.47 years. The majority of patients were married (58.9%), and the largest educational group held a bachelor’s degree or above (27.7%). The lower limb was the most common tumour site (67.4%), and giant cell tumour of bone (23.4%) and osteosarcoma (20.6%) were the predominant tumour types. Benign bone tumours accounted for 58.9% of cases. Surgical procedures included tumour resection with bone reconstruction (33.3%) or tumour arthroplasty (34.0%). The mean length of hospital stay was 7.67 ± 2.34 days. The mean PTGI-C scores at three time points were 41.24 ± 21.94, 44.56 ± 23.33, and 40.69 ± 22.63, respectively, with a statistically significant difference across time points (repeated-measures ANOVA: F = 3.372, P = 0.02).

Table 1.

General information of bone tumour patients with surgery

Trait	Category	Number	Composition Ratio (%) /`x±s / M(QR1)
Gender	Male	71	50.4
	Female	70	49.6
Age（y）		141	33.48±12.47
BMI（Kg/m2）		141	22.33±3.29
Marital	Married	83	58.9
	Unmarried	58	41.1
Education	Junior high school	31	22.0
	High school and secondary school	37	26.2
	Junior colleges	34	24.1
	Bachelor degree or above	39	27.7
Religion	YES	7	5.0
	NO	134	95.0
Income Monthly	＜2000	30	21.3
	2000~5000	47	33.3
	5001~10000	44	31.2
	＞10000	20	14.2
Residence	City	82	58.2
	County	28	19.9
	Countryside	31	22.0
Site of the tumour	Upper limbs	40	28.4
	Lower limbs	95	67.4
	Other	6	4.3
Type	Giant cell tumour	33	23.4
	Osteosarcoma	29	20.6
Fibrogenic tumours		16	11.3
Other mesenchymal tumours		26	18.4
Cartilage-origin tumour		25	17.7
	Other	12	8.5
Histology	Benign	83	58.9
	Malignant	58	48.1
Surgery type
Tumour segment resection		35	24.8
Tumour segment resection and elongated bone reconstruction		47	33.3
Tumour segment resection plus joint replacement		48	34.0
	Other	11	7.8
Implants	YES	93	66.0
	NO	48	34.0
History of surgery	YES	87	61.7
	NO	54	38.3
Time since diagnosis (month)		142	5 [2, 12]
Chemotherapy	YES	42	29.8
	NO	99	70.2
Sleep quality
	Very good	5	3.5
	Good	50	35.5
	General	54	38.3
	poor	32	22.7
Sleep duration/per day (hours)
	＜5h	31	22.0
	5~6h	32	22.7
	6~7h	44	31.2
	＞7h	34	24.1
Hospital stay（days）		141	7.67±2.34
PTGI-C2	T1	141	41.24±21.94
	T2	141	44.56±23.33
	T3	141	40.69± 22.63
	T4	141	46.62±22.99
			F=3.372, P=0.025

¹Indicates Interquartile range. ²PTGI-C Chinese version of the Post-Traumatic Growth Inventory

Model fitting and selection of PTG trajectories

PTGI-C scores of patients at 4 time points were taken as the observation index with setting the free estimation of time parameters, and 1–5 categories were extracted successively for model fitting analysis. The results show that only 2 types of models have significant LMR, LRT and BLRT values. At the same time, with the increase of the number of classifications, AIC, and aBIC values increased first and then decreased. The BIC increases and Entropy reaches its maximum value (Table 2). Above all, this study applied the 2 LCGM model of categories and combining the characteristics of the trajectory of change in PTGI-C, PTGI-C were divided into 2 subgroups. The results showed that category group 1 had a higher PTGI-C score at the admission (intercept = 44.276) and mean of the PTGI-C with an overall increasing trend (slope = 5.152 P < 0.001), indicating that the PTG were maintained at a high level. Therefore, category group 1 was named the “increasing trajectory group”. The PTGI-C score at the admission was lower than that of category 1 (intercept = 35.164) and the entire line was below the average, with an overall decreasing trend (slope = − 4.718, P < 0.001), so category 2 was named the “decreasing trajectory group”. The potential categories of the trajectory of the change in the PTG of the two groups are shown in Table 3; Fig. 1. The scores of PTGI-C for different periods in each group are shown in Table 4, including 91 cases (64.54%) in the “increasing trajectory group”, and 60 cases (35.46%) in the “decreasing trajectory group”.

Table 2.

Fitting Results of LCGM

Number of Classes	AIC	BIC	saBIC	Entropy	LMR	LRT	BLRT
1	4848.167	4874.706	4846.230	-		-	-
2	4840.262	4875.647	4837.679	0.717	0.0399*	0.0389*	0.0150*
3	4841.256	4885.487	4838.028	0.743	0.6690	0.6849	0.4350
4	4836.849	4889.927	4832.975	0.754	0.3986	0.4119	0.2500
5	4834.656	4896.580	4830.136	0.797	0.2341	0.2624	0.0609

Abbreviations: AIC Akaike information criterion, BIC Bayesian information criterion, saBIC sample size adjusted BIC, LMR LoMendell likelihood ratio test, LRT Likelihood Ratio Test, BLRT Bootstrap likelihood ratio test

* indicates P＜0.05

Table 3.

Intercept and Slope of Each Class

Class	Intercept	P	Slope	P
Class 1(Increasing trajectory group)	44.276	P<0.001	5.152	P<0.001
Class 2(Decreasing trajectory group)	35.164	P<0.001	-4.718	0.003

Fig. 1.

Potential categories of PTGI-C trajectory

Table 4.

Univariate Analysis of Patient-Reported Outcome Trajectories

Projects	Category	Increasing trajectory Group (n = 91)	Decreasing trajectory Group (n = 50)	χ2 /t	P
Gender (%)	Male	44 (48.4)	27(54.0)	0.41	0.521
	Female	47 (51.6)	23(46.0)
Age(year)	<27	42(46.2)	13(26.0)	6.27	0.042*
	28–39	26(28.6)	23(46.0)
	>40	23(25.2)	14(28.0)
BMI(Kg/m2)		22.16 ± 3.38	22.65 ± 3.13	-0.86	0.390
Marital status (%)	Married	47(51.6)	36(72.0)	5.52	0.019*
	Unmarried	44 (48.4)	14(28.0)
Education (%)	Junior high school	21(23.1)	10 (20.0)	5.17	0.160
	High school and secondary school	22(24.2)	15(30.0)
	Junior colleges	18(19.8)	16(32.0)
	Bachelor’s degree or above	30(32.9)	9(18.0)
Religion (%)	YES	5(5.5)	2(4.0)	0.15	0.196
	NO	86(94.5)	48(96.0)
Income Monthly (%)	<2000	17(18.7)	13(26.0)	1.11	0.775
	2000 ~ 5000	32(35.1)	15(30.0)
	5001 ~ 10,000	29(31.9)	15(30.0)
	>10,000	13(14.3)	7(14.0)
Residence (%)	City	50(54.9)	32(64.0)	1.15	0.563
	County	19(20.9)	9(18.0)
	Countryside	22(24.2)	9(18.0)
Site of the tumour (%)	Upper limbs	26(28.6)	14(28.0)	2.69	0.261
	Lower limbs	63(69.2)	32(64.0)
	Other	2(2.2)	4(8.0)
Type (%)	Giant cell tumour	25(27.5)	8(16.0)	10.28	0.064
	Osteosarcoma	20(22.0)	9(18.0)
	Fibrogenic tumour	7(7.7)	9(18.0)
	Other mesenchymal tumour	17(18.7)	9(18.0)
	Cartilage-origin tumour	18(19.8)	7(14.0)
	Other	4(4.3)	8(16.0)
Histology (%)	Benign	57(62.6)	26(52.0)	1.51	0.219
	Malignant	34(37.4)	24(48.0)
Surgery type (%)	Tumour segment resection	18(19.80	17(34.0)	3.58	0.310
	Tumour segment resection and elongated bone reconstruction	32(35.2)	15(30.0)
	Tumour segment resection plus joint replacement	33(36.3)	15(30.0)
	Other	8(8.8)	3(6.0)
Implants (%)	YES	62(68.1)	31(32.0)	0.54	0.462
	NO	29(31.9)	19(38.0)
History of surgery (%)	YES	58(63.7)	29(58.0)	0.45	0.503
	NO	33(36.3)	21(42.0)
Time since diagnosis (month)		6 [2,12]	5[2.10]	2048.00b	0.325
Chemotherapy (%)	YES	23(25.7)	19(38.0)	2.50	0.114
	NO	68(74.7)	31(61.0)
Sleep quality (%)	Very good	4(4.4)	1(2.0)	1.331	0.752
	Good	31(34.1)	19(38.0)
	General	37(40.7)	17(34.0)
	poor	19(20.9)	13(26.0)
Sleep duration/per day (hours)	<5 h	18(19.8)	13(26.0)	2.59	0.460
	5 ~ 6 h	22(24.2)	10(20.0)
	6 ~ 7 h	26(28.6)	18(36.0)
	>7 h	25(27.4)	9(18.0)
Hospital stay(days)		7.67 ± 2.41	7.68 ± 2.22	0.02	0.981
Coping style	Positive	24(26.4)	6(12.0)	3.98	0.046*
	Evasive	67(73.6)	44(88.0)
CIES-R3	T1	14 [7,24]	10 [3,18]	1825.502	0.053
	T2	27 [9,32]	12.5 [5, 28]	1674.50b	0.010*
PTS4	High level	48(52.7)	16(32.0)	5.60	0.018*
	Low level	43(47.3)	34(68.0)
Pain level	T1	1.91 ± 1.82	2.24 ± 2.28
	T2	4.20 ± 2.07	4.64 ± 2.06
	T3	2.98 ± 1.99	3.08 ± 1.88
	T4	2.35 ± 1.55	2.22 ± 1.43	244.946	<0.001*
PTGI-C5	T1	45.42 ± 20.93	33.21 ± 21.80
	T2	50.15 ± 20.26	34.41 ± 25.26
	T3	51.73 ± 18.67	19.98 ± 12.69
	T4	60.34 ± 13.31	19.78 ± 11.05	100.416	<0.001*

* Indicates P<0.05. ¹ Indicates Fisher’s exact probability method. ² Indicates Mann-Whitney U test method. ³CIES-R The Chinese translation of the Impact of Event Scale-Revised. ⁴ PTS Post Traumatic Stress. The Cut-off of CIES-R was 26. CIES-R ≥ 26 indicates higher PTS while CIES-R < 26 indicates lower status of PTS. ⁵PTGI-C: Chinese version of the Post-Traumatic Growth Inventory. ⁶ Indicates results of repeated measures analysis of variance

Univariate analysis and logistic regression analysis of factors influencing the trend of PTG

The results showed that the two trajectory categories of age, marital status, coping style, CIES-R on post-operative 2 days and Post Traumatic Stress (PTS) were statistically significant (P < 0.05) when comparing PTGI-C scores, as shown in Table 4. The variables with statistical significance in the univariate analysis (P<0.10) were used as independent variables, and the two groups of trajectory categories were used as dependent variables for the Binomial Logistic Regression Analysis. The forward stepwise regression method based on maximum likelihood estimation was used for the analysis. The influencing factors for the decreasing trajectory group relative to the increasing trajectory group were marital status, type of tumour, and PTS (P<0.05). The results showed that unmarried (Ref =Married, OR = 0.381, P = 0.019), low level of PTS (Ref =High level, OR = 2.666, P = 0.015), Giant cell tumour (Ref = Osteosarcoma, OR = 0.223, P = 0.047), Fibrogenic tumours (Ref = Osteosarcoma, OR = 0.127, P = 0.007), cartilaginous origin tumour (Ref = Osteosarcoma, OR = 0.194, P = 0.034), and other tumours (Ref = Osteosarcoma, OR = 0.169, P = 0.023) were predictive factors for PTG trajectory. The fit of this Binary Logistic regression model was P = 0.947(P>0.05) in the Hosmer Lemeshow test. The Omnibus test showed that the model was significant in general (P = 0.002<0.05), and the model predictive power of Binomial Logistic Regression fitting equation was 66.0%. The result was shown in Table 5.

Table 5.

Logistic Regression Analysis of Patient Reported Outcome Trajectories

Variables	B	SE	Waldχ2	p	OR (95%CI)
Marital status
Unmarried (Ref =Married)	-0.965	0.412	5.481	0.019	0.381[0.170–0.855]
Type
Giant cell tumour (Ref = Osteosarcoma)	-1.503	0.756	3.948	0.047	0.223[0.051, 0.980]
Fibrogenic tumour (Ref = Osteosarcoma)	-2.063	0.768	7.210	0.007	0.127[0.028,0.573]
Other mesenchymal tumour (Ref = Osteosarcoma)	-0.684	0.820	0.696	0.404	0.505[0.101,2.517]
Cartilage-origin tumour (Ref = Osteosarcoma)	-1.640	0.774	4.485	0.034	0.194[0.043,0.885]
Other (Ref = Osteosarcoma)	-1.778	0.780	5.193	0.023	0.169[0.037,0.780]
PTS1
Low level (Ref = High level)	0.981	0.402	5.953	0.015	2.666[1.213, 5.860]

Omnibus Tests of Model Coefficients showed P = 0.002<0.05. Hosmer and Lemeshow showed P = 0.947>0.05. Model predictive power was 66.0%. ¹ Indicates Post Traumatic Stress

Discussion

To our knowledge, this study was the first one which focused on the trajectories of PTG among bone tumour patients at the early time after surgery. It was also the first study to explore the categories and predictive factors of PTG trajectories when patients finished the surgery. We found that the patients could be divided into two group basing on the PTG trajectories including increasing trend group and decreasing trend group. Interestingly, age, marital status, coping style, CIES-R on post-operative 2 days, level of PTS, and categories of tumours would predict the tendency of PTG basing on univariate analysis and multivariate analysis.

We selected four time points for repeated measurements of PTG based one patients’ treatment process. In this research, patients were usually admitted to hospital 3 days before surgery when general information could be collected clearly. The time chosen for follow-up was based on the patient’s post-operative recovery process. They had their wound stitches removed in the outpatient clinic on 15 days after surgery, at which point the wound healed and HCWs were available to guide the patient through the full recovery. In addition, their first post-operative review in the outpatient clinic on 30 days after surgery, which was the time of high incidence of various post-operative complications in bone tumours and patients should be emphasized [13]. Patients with malignant bone tumours who also required a combination of neoadjuvant chemotherapy typically begin their first postoperative chemotherapy session 30 days later which may result in new traumatic events [40, 41].

Additionally, interests in the experiences of bone tumour patients who just went through the diagnosis and therapy induced us emphasize on the PTG level during 1 month after surgery. A traumatic event may challenge a set of deeply held assumptions about the world and the self [42]. Based on the theoretical model of post-traumatic growth, it can be inferred that individuals need to reconstruct all aspects such as schema and core values to produce PTG in the early stage of traumatic events,, which requires a certain reaction time [42, 43]. But the length of this time was still inconclusive [43]. Previous research reported the long-term PTG level of patients who had been diagnosed or treated of cancers for years or even decades [24, 25], but none of them focused on the changes of PTG in patients with tumours undergoing therapeutic surgery for the first time. Latest studies have shown that PTG occurring in the short term after traumatic events was often more meaningful, which would affect the patient’s long-term PTG level and quality of life [44, 45]. Hence, our research aimed to explore the change of PTG in the early stage of treatment among bone tumour patients. If risk factors of PTG reduction trajectory could be identified in the early stage of tumour-related traumatic events, it would be effective to the improvement or maintenance of PTG levels later for patients [46].

We found that patients undergoing bone tumour surgery, including those with benign and malignant bone tumours, had two groups of latent classes of PTG trajectories.The outcomes were different with a longitudinal study of patients with head and neck cancer which found positive psychology increased and then decreased early in diagnosis and treatment [47]. Fauske et al. conducted the qualitative study and showed that patients with bone sarcoma were more likely to have emotional problems [15]. It was also reported that different categories of PTG change trajectories existed in different tumour population [48]. However, these studies did not explore the trajectory classes and influencing factors of PTG in patients with bone tumours. Through the establishment of latent category trajectory, we were able to more accurately identify the PTG ascending group and decline group. On this basis, precise intervention for high-risk bone tumours with decreased PTG could be targeted.

We also identify the unmarried status was a protective factor for the decrease of PTG which was a novel finding. Differently, relevant study have found that being married was a protective factor for low PTG levels in cancer patients [49]. The study inferred that married people received more social support than unmarried people which was a mediator of marital status and PTG in Korea [49]. However, this may be related to the gender of the sample included in. Previous studies tended to analyse PTG levels only in men or women [49, 50], but our study included both male and female patients for analysis. Moreover, the family support of unmarried patients often came from all family members, while the family support of married patients was mainly from spouse [51]. If the level of marital adjustment was too low, it was likely to affect the early PTG level [52]. However, psychological support should be provided to patients regardless of marital status, for example, through family health education to enhance social support and caregiving skills.

Secondly, the active coping strategies could help patients build up high level of PTG. Our findings were similar with other research findings. A cross-sectional study of Asian cancer patients also showed that positive coping style was beneficial for personal growth [53]. Gori and his colleges also confirmed the relationship between coping style and PTG level [54]. However, the predictive and practical effects of positive coping styles on the trajectory of PTG change cannot be well interpreted in aforementioned studies. Although many different components were thought to interact in intricate ways to cause post traumatic development, cognitive processes seem to play a specific role in these interactions [55]. Tumour, on the other hand, was an intensely signifying event that destroys an individual’s sense of self and somatopsychic link, necessitating an introspective concentration, developing awareness, and meaning-making of one’s own experience to permit complete elaboration [56]. Considering prior research and the present study, we might infer that active coping strategies were able to help bone tumour patients actively complete cognitive processing of traumatic events and reconstruct the world assumption through deliberate rumination [55]. Therefore, identifying bone tumour surgery patients who had evasive coping strategies and teaching them to cope with traumatic events may be beneficial for medical staff to improve PTG.

Thirdly, there were also significant differences in PTG levels among patients with different types of bone tumours. Compared with osteosarcoma, other types of bone tumours were protective factors for PTG reduction. Osteosarcoma, which ranks first among malignant bone-related cancers in adolescents, had complex heterogeneity and abnormally generated immature osteoid matrix [57]. About 35%-45% of patients with osteosarcoma have chemoresistance and poor response to treatment, and these patients are prone to distant metastasis [58]. The 5-year survival rate of patients with distant metastasis was less than 20% [58]. Hence, once diagnosed as osteosarcoma, it must bring a shock to patients and their families. In the early stages of treatment, it was often difficult to maintain a high level of PTG. However, challenges for patients with bone tumours, including diagnosis of osteosarcoma, treatment persistence, prognosis, altered body image, and changes in functional autonomy or role, may recur throughout the disease trajectory, particularly in those with malignant disease [59]. This suggests that it is necessary for healthcare workers to provide empathy and support to patients with bone tumours, irrespective of whether the tumour is osteosarcoma.

Finally, this study validated the positive correlation between PTS and PTG in patients with bone tumours. PTS was one of the most frequently reported negative psychological phenomena in cancer patients. It mainly included post traumatic stress disorder (PTSD) and post traumatic stress symptom (PTSS) [60]. They were clinical diseases and Sub-syndromic symptoms, respectively [44]. Previous research evaluating the association between PTS and PTG found varying results in terms of the degree, direction, and linearity of this relationship [44]. Some research [61, 62] claimed a positive association between PTS and PTG, while another [63], claimed a negative relationship. Our findings were in line with the research results of Morrill [62], Groarke [32], Chen and other scholars [64]. Some researcher believed that the positive correlation between the two was due to the fact the more threatening an event was, the more it would promote the alternate occurrence of growth and stress response. Then they would lead to a stronger relationship between the two structures [44, 65], resulting in the result of both growths. Although a higher level of psychological impact may cause a higher level of PTG, patients need to bear a certain degree of psychological pain [62]. However, a high PTS rate was maladaptive and can hinder the progression of PTG [63]. An individual with an excessive level of PTS could not have the capacity to reconstruct cognition, such as core beliefs, which may form the basis of PTG [65]. As HCWs when evaluating the psychological status of patients, we should include both positive and negative psychological states in the evaluation, which was conducive to accurately evaluate all the psychological states of patients with bone tumours, so as to provide more targeted psychological interventions to them and alleviate some of the pain on the premise of creating conditions conducive to PTG.

Limitations

This study has several limitations. First, the single-centre design and short follow-up limit the generalisability of the findings. Bone tumours are rare, and the sample was limited. Larger, multi-centre studies with extended follow-up are required to confirm the trajectory of PTG and its determinants. Second, the use of self-report questionnaires may introduce response bias, and the timing of measurements may not fully capture the dynamic course of PTG. Third, although the inclusion of diverse tumour types may have reduced selection bias, it could also have lowered internal validity. Univariate analysis showed no difference between benign and malignant tumours, but more complex statistical modelling should be considered in future work. Fourth, we acknowledge that effect sizes were not reported alongside p-values, which limits the interpretability of the magnitude and clinical significance of our findings. Future studies should routinely include effect size estimates, such as Cohen’s d or partial η², to facilitate a more comprehensive understanding of the results. Finally, other potential confounders, such as personality traits, additional life events, and perception of stress, were not examined and need further investigation.

Conclusion

Our study was a 1-month longitudinal study of patients undergoing bone tumour surgery. Based on the concept of individual heterogeneity, the LCGM was used to confirm the trend of PTG level in patients with bone tumour surgery into an increased group and a decreased group. Marital status, coping style, tumour type and level of PTS were the influencing factors of PTG decreasing trend group through univariate analysis and logistic regression. These findings provide preliminary empirical support for positive psychology frameworks in orthopaedic oncology populations and may offer a basis for developing targeted psychosocial interventions in the future.

Authors’ contributions

YL designed the research, collected and analysed the data of bone tumour patients, and drafted this manuscript. XFZ collected and analysed the data of bone tumour patients. PFL modified the research design and assisted investigating of data among bone tumour patients and revised the manuscript. YTW and JW revised the statistic model of analysing trend of post traumatic growth among bone tumour surgery patients. CQT and HD defined the criteria for participants of the study and supported the idea of methodology of this research. JLC and NN came up with the idea of the research, supervised the process of research and revised the manuscript.

Funding

Not applicable.

Data availability

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

Declarations

Ethics approval and consent to participate

λ Ethical considerations were upheld throughout the study, adhering to the principles outlined in the Declaration of Helsinki. The study received approval from the West China Hospital Ethics Committee (approval number: 2019-Review-1091) and all participants provided informed consent by signing consent forms.

Consent for publication

λ Authors, institutions and participants have consent of publication. All authors have reviewed the manuscript before submission.

Competing interests

The authors declare no competing interests.