Personalizing Traditional Chinese Medicine: A Longitudinal Study on How Constitution-Based Configurations Relate to Quality of Life During Comprehensive Intervention

Abstract

Objective

Traditional Chinese Medicine (TCM) constitution theory underpins personalized health management, but most studies examine constitution types in isolation, whereas overlapping constitutions are the rule rather than the exception in practice. This study examined how combinations of TCM constitution subscale scores were related to health-related quality of life (QoL) over six months of constitution-based comprehensive intervention.

Methods

In a single-arm prospective study, 509 adults aged 19–84 years completed a 6-month constitution-based self-management programme. Constitution was assessed with the 60-item Constitution in Chinese Medicine Questionnaire (CCMQ-60), which yields a Gentleness (Type A) score and scores on eight biased types (Qi-Deficiency, Yang-Deficiency, Yin-Deficiency, Phlegm-Dampness, Damp-Heat, Blood-Stasis, Qi-Depression, Special-Diathesis). QoL was measured with the 36-Item Short Form Health Survey (SF-36) at baseline, 3 months, and 6 months. Dynamic (panel) and “bottom-up” fuzzy-set qualitative comparative analysis (fsQCA) were used to identify constitution configurations associated with high QoL at each time point.

Results

The mean SF-36 total score rose from 72.7 ± 16.1 at baseline to 75.1 ± 14.6 at 3 months and 75.6 ± 15.5 at 6 months (mean change 2.9 points, 95% CI 1.7 to 4.1; standardized change 0.21). No single constitution subscale was a necessary condition for high QoL (pooled consistency < 0.90). Dynamic fsQCA identified 5 overlapping configurations (consistency 0.946–0.953; solution coverage 0.540), each characterized by a high Gentleness score combined with low scores on most biased constitutions; unique coverage of individual configurations was small (0.006–0.049). Bottom-up analysis at a PRI threshold of 0.75 retained low Yang-Deficiency or low Blood-Stasis together with low Qi-Depression at baseline and 3 months (consistency 0.867–0.873), whereas at 6 months the retained pairs were low Qi-Deficiency with low Yang-Deficiency or low Blood-Stasis, and low Yang-Deficiency with low Phlegm-Dampness (consistency 0.856–0.866). Consistency of the configurations exceeded 0.90 in every age and sex stratum.

Conclusion

During the programme, high QoL was associated with a high Gentleness score combined with low scores on several biased constitutions rather than with any single constitution type, and the pairs of biased constitutions that distinguished high QoL at 6 months differed from those at baseline and 3 months. Because the study had no control group, these configurational patterns are descriptive and hypothesis-generating; they point to combinations of constitutions that could be prioritized and tested in controlled evaluations of constitution-based personalized health management.

Background

Traditional Chinese Medicine (TCM) constitution refers to the inherent, relatively stable characteristics of the human body formed through genetic endowment and acquired factors during the life process. These characteristics encompass structural morphology, physiological functions, and psychological states.^1,2 The most widely used classification system is the nine types of constitution:^3–5 Gentleness (Type A), Qi-Deficiency (Type B), Yang-Deficiency (Type C), Yin-Deficiency (Type D), Phlegm-Dampness (Type E), Damp-Heat (Type F), Blood-Stasis (Type G), Qi-Depression (Type H), and Special-Diathesis (Type I). Gentleness describes a balanced state of yin and yang with good adaptability, and the other eight types describe deviations from that balance; for this reason they are conventionally termed “biased” constitutions.^2,6 TCM constitution identification involves extracting constitution characteristics suitable for populations from the complex and variable individual constitution phenomena.⁷ Its multidimensional nature is reflected in how physiological and psychological states develop differently under the combined influence of genetic and environmental factors, similar to health-related quality of life.⁸ This provides an effective evaluation system for health assessment and promotion. Through TCM constitution identification and evaluation, individuals with similar physiological and psychological characteristics can be categorized, facilitating an overall understanding of the health status of the population.⁹ Previous studies on the relationship between TCM constitution and quality of life have primarily focused on comparing the quality of life among different TCM constitution groups and examining the simple correlations or independent effects of each TCM constitution factor on quality of life. However, they often fail to consider the interaction between various dimensions of TCM constitution and their influence on quality of life. Statistical analyses have mainly used methods such as t-tests, Spearman correlation analysis, multiple stepwise regression, and logistic regression.^10–14 These methods fall short of providing a comprehensive assessment of the relationship between multidimensional TCM constitution and quality of life. Some studies suggest that enhancing Gentleness constitution scores while reducing biased constitution scores benefits health,¹⁵ yet they do not address how changes across constitutional dimensions collectively affect QoL, suggesting that the depth and precision of such research require further advancement.

There are both theoretical and empirical reasons to expect that constitutions matter in combination rather than one at a time. In constitution theory the nine types are not mutually exclusive categories but dimensions of one body: a person is assigned a dominant type, yet in population surveys most adults score above the threshold on two or more biased subscales, a situation described in the literature as compound or mixed constitution.^2,6,8 The types are also causally linked in TCM reasoning; qi deficiency is held to give rise to blood stasis and phlegm-dampness, and yang deficiency to dampness, so that deficiency-type and stagnation-type deviations tend to travel together and reinforce one another. Regression models that enter subscale scores as additive terms treat these dimensions as independent and cannot tell us which combinations are compatible with good health and which are not. Whether the same combinations remain relevant as an intervention proceeds is a further open question: constitution is defined as relatively stable, but converted scores do shift over months of lifestyle regulation, and if deficiency-type constitutions respond first, the biased constitutions that still distinguish people with high QoL may change over the course of a programme. We did not pre-specify formal temporal hypotheses; our general expectation from this reasoning was that a high Gentleness score would feature in every pattern associated with high QoL, and that the accompanying biased constitutions would differ between the start and the end of the intervention.

Two points of clinical context frame the present analysis. First, constitution-based intervention as practised in Chinese health-management centres is a package of individualized advice on emotion, diet, daily living, exercise and acupoint self-care, delivered by TCM practitioners and health managers and then carried out by the participant; there is no established standard of care in this setting against which it is routinely compared, and programmes are usually organized in cycles of three to six months because constitution scores are re-assessed at those intervals.^16,17 Second, for the SF-36 a change of about half a standard deviation is generally regarded as the smallest difference that is noticeable to patients;¹⁸ we use this benchmark when describing the size of the changes observed.

Qualitative Comparative Analysis (QCA), introduced and developed by American sociologist Charles Ragin in 1987, is a research methodology grounded in configurational thinking. It is particularly well-suited for investigating issues of equifinality, where multiple pathways lead to the same outcome.¹⁹ A “configuration” refers to a specific combination of conditions that together are associated with a particular outcome, essentially examining how conditions interact to yield certain results. By identifying multiple configurations that are sufficient for the same outcome, QCA provides a framework for understanding the complex relationships between configurations of conditions and the outcome of interest.

Traditional QCA, constrained by its theoretical foundations and tools, is primarily limited to cross-sectional data analysis, making it difficult to explore the longitudinal patterns of configurations over time. In contrast, dynamic QCA overcomes the limitations between panel data and traditional QCA by measuring across three dimensions: between groups, within groups, and pooled data. It employs a consistency-adjusted distance to capture the variations in consistency across time and case dimensions. In this study, we use the scores from nine subscales of TCM constitution as the condition variables, and the overall quality of life score as the outcome variable. Adopting the dynamic QCA method, based on the theory and methodology proposed by García-Castro and Ariño,²⁰ we analyze the configurational relationship between TCM constitution and high quality of life before the intervention, after 3 months, and after 6 months of intervention. Previous research has pointed out that traditional QCA’s “top-down” approach, relying on truth tables and Boolean minimization, can generate configurations that are either insufficient to explain the outcome or include unnecessary configurations that can be eliminated.²¹ Haien Ding proposed and validated a “bottom-up” QCA method,²² which generates all minimally sufficient conditions leading to the outcome for analysis. Thus, we apply this method to further analyze the data, aiming to identify the core configurations of TCM constitution that are associated with high quality of life, thereby deepening our understanding of their relationship.

Accordingly, the study addressed three questions. (1) Is a high (or low) score on any single constitution subscale a necessary condition for high QoL? (2) Which combinations of subscale scores are sufficient for high QoL, and do they share a common core? (3) Do the combinations identified before the intervention persist at 3 and 6 months, or are they replaced by others? The study was a single-arm observational analysis of an intervention programme; it was not designed to estimate the effect of the intervention on QoL, and the findings are framed as associations.

Data Sources and Measurement

Data Sources

The data for this study was sourced from a TCM constitution intervention project. Discriminant analysis was used to determine the TCM constitution of the study participants. Based on the results of the determination of the TCM constitution, a comprehensive intervention programme was formulated, including five aspects of emotion, diet, daily living, exercise and meridians. The general principles of the intervention programme include: (1) maintain emotional stability and happy mood; (2) eat a balanced and reasonable diet; (3) live a regular life and get enough sleep; (4) adhere to the right amount of exercise; (5) pay attention to the health of acupoints in daily life. The specific intervention programme for each body type is based on the different characteristics of the 9 types of TCM constitutions. For example, participants whose dominant type was Qi-Deficiency were advised to favour warm, easily digested and qi-tonifying foods, to avoid overexertion and to practise low-intensity, high-frequency exercise such as walking or Tai Chi, and to massage Zusanli (ST36), Guanyuan (CV4) and Qihai (CV6); those with a Phlegm-Dampness type were advised to eat light and to avoid greasy, sweet and cold foods, to undertake gradual low- to moderate-intensity whole-body exercise, and to massage Zusanli (ST36) and Fenglong (ST40). The complete constitution-specific plans are given in Supplement 1, and the study protocols have been described in earlier reports.^16,17 The project was conducted between February 2013 and January 2014 at the Center for Disease Prevention at Zhuhai Hospital of Guangdong Provincial Hospital of Traditional Chinese Medicine, a hospital-based health-management department that provides constitution identification and health examination to the general public. Adults attending the centre during the recruitment period were informed of the project by centre staff and invited to take part; recruitment was by convenience among those who met the inclusion criteria and gave consent, and 595 participants were enrolled.

Inclusion criteria were as follows: (1) participants were 18 years or older and possessed a certain level of literacy, enabling them to independently complete a questionnaire survey; (2) participants provided signed informed consent and voluntarily agreed to take part in the comprehensive TCM constitution intervention study. Exclusion criteria included: (1) individuals with mental illness or behavioral disorders; (2) those with serious diseases or other conditions that prevented them from completing the survey or intervention. The criteria were deliberately broad because the programme was aimed at the general adult population rather than at a disease group; participants were not required to have, or to be free of, any particular condition, and self-reported medical history was recorded at baseline for description.

Based on the participants’ TCM constitution type identified before the intervention, researchers implemented a targeted comprehensive TCM constitution intervention tailored to the dominant TCM constitution type for a duration of six months. The intervention was delivered by TCM physicians and health-management staff of the centre who had received unified training on the protocol. At the baseline visit each participant was given face-to-face feedback on his or her constitution result together with the individualized plan covering the five domains and the printed manual for the relevant type; the plan was then carried out by the participant in daily life. Participants were followed up by telephone and asked to record implementation in a monitoring manual, which was checked at the 3- and 6-month visits, when the questionnaires were re-administered and the plan was reviewed. Adherence to the programme, and its variation across age groups, has been analysed and reported separately¹⁷ and is not re-analysed here. Assessments were conducted at three time points: baseline (before intervention), three months into the intervention, and at the six-month mark following intervention.

A standardized protocol was developed prior to the study, and all researchers received training to ensure consistency in data collection. During the study, researchers conducted surveys following a uniform set of standards, while monitoring participants through various methods, including monitoring manuals and phone check-ins. Quality control measures, such as assessing the completeness of questionnaire responses and performing logical consistency checks, were employed to verify the accuracy of the survey data and intervention outcomes. An EpiData 3.1 database was established, and data entry was performed using a double-entry, cross-verification method by two independent data collectors. A total of 509 participants completed the six-month comprehensive TCM constitution intervention (Figure 1), and this analysis is based on these 509 participants. Of the 595 enrolled, 58 did not attend the 3-month assessment and a further 28 did not attend the 6-month assessment (86 in total, 14.5%); none withdrew consent; 51 could not be contacted and 35 were lost for other reasons (Figure 1). Because dynamic QCA requires a complete panel, participants with a missing wave were not imputed. Baseline characteristics of completers and non-completers are compared in Data Description, Descriptive Analyses and Calibration and Supplementary Table S1. The study was approved by the Ethics Committee of Beijing University of Chinese Medicine (Approval No. 2012BZHYLL0301).

Figure 1.

Flowchart of the research process.

Measurement Tools

Constitution in Chinese Medicine Questionnaire - 60 Items

Participants underwent TCM constitution assessments using the Constitution in Chinese Medicine Questionnaire with 60 items (Supplement 2 and Supplement 3), at baseline and at the 3- and 6-month visits. The CCMQ-60 was developed by our group and has shown acceptable internal consistency and test–retest reliability as well as construct validity in Chinese adults;^3,23 it forms the basis of the national industry standard for constitution classification²⁴ and has since been validated in other populations, including older adults.¹⁵ We did not re-estimate its psychometric properties in the present sample. This scale comprises nine subscales: gentleness constitution (A), qi-deficiency constitution (B), yang-deficiency constitution (C), yin-deficiency constitution (D), phlegm-dampness constitution (E), dampness-heat constitution (F), blood-stasis constitution (G), qi-depression constitution (H), and special-diathesis constitution (I). Except for the gentleness constitution, the other eight types are categorized as biased constitutions, because each describes a departure from the balanced state that the gentleness subscale measures (see Background). Each item on the scale is rated using a five-point Likert scale, where respondents choose the most appropriate answer from the following options: “1 – none (never),” “2 – rare (a little),” “3 – sometimes (somewhat),” “4 – often (quite),” and “5 – always (very).” The scores for each of the nine subscales are calculated by summing the item scores to obtain a raw score, which is then converted to a standardized score using the formula: Converted score = (Actual score − Minimum possible score) / (Maximum possible score − Minimum possible score) × 100. The resulting score ranges from 0 to 100, with higher scores indicating a stronger tendency toward a particular TCM constitution type. The evaluation criteria are as follows: if the participant’s converted score on the gentleness constitution subscale is ≥ 60 and all eight biased constitution subscales are < 30, the participant is classified as having a gentleness constitution. For the eight biased constitutions, a converted score of ≥ 30 on any subscale indicates that the participant is classified as having that specific TCM constitution.²³ These cut-offs (≥ 60 for gentleness; < 30 and ≥ 30 for the biased subscales) are those specified in the national standard for constitution classification²⁴ and are the criteria used in the great majority of published CCMQ studies;²⁵ they were derived in the original scale-development work from the score distributions of large general-population samples and expert consensus, and we applied them unchanged. If two or more biased constitution scores exceed 30 points, discriminant analysis is used to determine the dominant constitution type, which was used to assign the intervention plan. The QCA analyses reported below do not use the categorical type; they use the nine converted subscale scores directly, so that every participant contributes information on all nine dimensions.

Chinese Version of the Short Form Health Survey (SF-36)

Participants’ health-related quality of life (QOL) was assessed using the Chinese version of the SF-36 at the same three visits. The QOL assessment covers various dimensions related to physical, psychological, emotional, and social functioning, aligning with the principles of TCM constitution. The SF-36 is a widely used self reported measure of quality of life, consisting of 36 items across eight dimensions: physical functioning, role physical, bodily pain, general health, vitality, social functioning, role emotional, and mental health.²⁶ These dimensions are grouped into two major domains: physical and mental health; the Chinese version used here was translated and validated in a mainland Chinese general population.²⁷ Each dimension, domain, and the overall score can be converted into a 0 to 100 scale, with higher scores indicating better quality of life. In this study, the overall SF-36 score was used as the primary indicator to evaluate participants’ health-related QOL. We chose the total score rather than the two component summaries because QCA requires a single outcome set, because constitution theory treats physical and mental functioning as one integrated state, and because in this population the physical and mental component scores were highly correlated and changed in parallel (Data Description, Descriptive Analyses and Calibration); the component summaries are reported descriptively alongside the total score. As a benchmark for interpreting change, half a standard deviation of the baseline total score¹⁸ corresponded to about 8 points in this sample.

Data Analysis and Results

Data Description, Descriptive Analyses and Calibration

The measurement data were approximately normally distributed and are expressed as mean ± standard deviation (x ± s), while qualitative data are presented as frequencies and percentages. Changes in scores between time points were summarized as mean differences with 95% confidence intervals from paired t-tests, together with the standardized change (mean difference divided by the standard deviation of the difference). Completers and non-completers were compared on baseline characteristics with independent-samples t-tests and chi-square tests. Associations between age and baseline scores were examined with Spearman correlation. These descriptive analyses were carried out in R version 4.2.1. Among the 509 participants, ages ranged from 19 to 84 years (mean 47.7 ± 15.0; 240 participants were under 45, 141 were 45–59 and 128 were 60 or older), with 158 males (31.04%) and 351 females (68.96%). A little over 40% reported at least one physician-diagnosed chronic condition at baseline, most commonly hypertension (15.3%), hyperlipidaemia (15.3%), fatty liver (12.2%) and diabetes (10.6%); the remainder described themselves as generally healthy. According to the single standard of TCM constitution identification, 148 participants (29.08%) were identified as having a gentleness constitution, while the distribution of the gentleness and other eight biased constitutions is detailed in Table 1. Mixed constitution was common: at baseline 92.3% of participants scored ≥ 30 on at least one biased subscale and 75.8% on two or more, which is the situation that motivated a configurational rather than a single-type analysis.

Table 1.

The Distribution of TCM Constitution Types Among 509 Research Subjects

TCM Constitution Type	Frequency (n)	Percentage (%)
Gentleness constitution (A)	148	29.09
Qi-deficiency constitution (B)	53	10.41
Yang-deficiency constitution (C)	60	11.79
Yin-deficiency constitution (D)	19	3.73
Phlegm-dampness constitution (E)	59	11.59
Dampness-heat constitution (F)	70	13.75
Blood-stasis constitution (G)	47	9.23
Qi-depression constitution (H)	28	5.50
Special diathesis constitution (I)	25	4.91

Notes: Due to rounding percentages to two decimal places, the cumulative value does not equal 100%. Therefore, the proportion of individuals with a gentleness constitution was adjusted from 29.08% to 29.09%.

The 86 participants who did not complete the three assessments did not differ from the 509 completers in age (46.6 ± 14.6 vs 47.7 ± 15.0 years, p = 0.54), sex (72.1% vs 69.0% female, p = 0.65), baseline SF-36 total score (71.1 ± 17.1 vs 72.7 ± 16.1, p = 0.39), any of the nine baseline constitution scores (all p > 0.3) or the distribution of dominant constitution type (p = 0.65) (Supplementary Table S1). Attrition therefore does not appear to have been selective on the variables analysed, although selection on unmeasured factors cannot be excluded.

Among completers, the mean SF-36 total score increased from 72.7 ± 16.1 at baseline to 75.1 ± 14.6 at 3 months and 75.6 ± 15.5 at 6 months (Table 2 and Figure 2A). The change from baseline to 6 months was 2.9 points (95% CI 1.7 to 4.1; standardized change 0.21); almost all of it occurred in the first three months, and the change between 3 and 6 months was negligible (0.5 points, p = 0.40). Physical and mental component summaries moved in parallel (changes of 3.3 and 2.5 points, respectively). At the individual level 31.2% of participants improved by 8 points or more (about half a baseline standard deviation) and 16.9% declined by that amount, so the average change conceals considerable heterogeneity. Over the same period the gentleness score rose by 4.1 points and all eight biased scores fell, the largest declines being in blood-stasis (−5.5), qi-deficiency (−5.0) and dampness-heat (−4.5) (Table 2 and Figure 2B). These uncontrolled before–after changes are reported to describe the sample; they are not estimates of an intervention effect.

Table 2.

SF-36 Scores and TCM Constitution Subscale Scores at Baseline, 3 Months and 6 Months (n = 509)

Variable	Baseline	3 Months	6 Months	Change, Baseline to 6 Months (95% CI)	Standardized Change	p
Health-related quality of life (SF-36)
SF−36 total score	72.72 ± 16.11	75.14 ± 14.55	75.62 ± 15.48	2.90 (1.68 to 4.12)	0.21	<0.001
Physical component summary	72.03 ± 16.54	74.37 ± 15.39	75.34 ± 16.47	3.31 (2.07 to 4.54)	0.23	<0.001
Mental component summary	73.41 ± 19.23	75.91 ± 17.12	75.90 ± 17.50	2.49 (0.97 to 4.01)	0.14	0.001
TCM constitution (CCMQ-60 converted score)
Gentleness (A)	59.30 ± 15.46	61.89 ± 14.04	63.37 ± 13.65	4.08 (3.12 to 5.03)	0.37	<0.001
Qi-deficiency (B)	34.86 ± 16.17	31.55 ± 14.80	29.88 ± 14.05	−4.98 (−6.00 to −3.96)	−0.43	<0.001
Yang-deficiency (C)	34.45 ± 22.54	33.04 ± 20.85	30.47 ± 19.09	−3.98 (−5.01 to −2.95)	−0.34	<0.001
Yin-deficiency (D)	28.32 ± 15.35	27.10 ± 13.60	25.21 ± 13.41	−3.11 (−4.07 to −2.16)	−0.28	<0.001
Phlegm-dampness (E)	29.47 ± 16.20	28.02 ± 15.27	26.57 ± 14.75	−2.91 (−3.84 to −1.98)	−0.27	<0.001
Dampness-heat (F)	29.08 ± 17.44	26.13 ± 16.05	24.53 ± 15.09	−4.54 (−5.57 to −3.51)	−0.38	<0.001
Blood-stasis (G)	34.33 ± 14.40	31.78 ± 13.94	28.81 ± 13.04	−5.52 (−6.36 to −4.68)	−0.57	<0.001
Qi-depression (H)	28.75 ± 17.30	26.05 ± 15.33	24.61 ± 14.46	−4.14 (−5.22 to −3.06)	−0.33	<0.001
Special diathesis (I)	20.87 ± 15.67	19.79 ± 14.25	18.36 ± 13.55	−2.51 (−3.45 to −1.56)	−0.23	<0.001

Notes: Values are mean ± standard deviation unless stated otherwise. Change is the mean within-person difference between 6 months and baseline with its 95% confidence interval from a paired t-test; the standardized change is the mean difference divided by the standard deviation of the differences. Constitution scores are converted scores (0–100); a higher gentleness score and lower biased-constitution scores indicate a more balanced constitution. These are uncontrolled before–after changes and do not estimate the effect of the intervention. Abbreviations: CCMQ-60, Constitution in Chinese Medicine Questionnaire (60 items); CI, confidence interval; SF-36, 36-Item Short Form Health Survey; TCM, Traditional Chinese Medicine.

Figure 2.

Quality of life, constitution scores and minimally sufficient constitution configurations over the 6-month intervention (n = 509). (A) Mean SF-36 total score and physical and mental component summary scores, with 95% confidence intervals, at baseline, 3 months and 6 months. (B) Mean converted scores (0–100) of the nine CCMQ-60 subscales at each time point. (C) Two-condition configurations retained by bottom-up QCA at a PRI threshold of 0.75 at each time point; cell shading indicates consistency and each cell gives consistency and raw coverage; grey cells indicate that the configuration was not retained at this threshold at that time point (full results at PRI 0.65, 0.70 and 0.75 are in Supplementary Table S3). “~” denotes a low score (non-membership in the high-score set) and “*” denotes logical AND. (A and B) describe uncontrolled before–after change and are not estimates of an intervention effect.

Abbreviations: CCMQ-60, Constitution in Chinese Medicine Questionnaire (60 items); PRI, proportional reduction in inconsistency; QCA, qualitative comparative analysis; QoL, quality of life; SF-36, 36-Item Short Form Health Survey.

Age was related to the constitution profile at baseline. Older participants had somewhat higher gentleness scores (Spearman ρ = 0.17) and lower scores on qi-deficiency (ρ = −0.23), yin-deficiency (ρ = −0.24), dampness-heat (ρ = −0.37) and qi-depression (ρ = −0.32) (all p < 0.01), while yang-deficiency was only weakly related to age (ρ = −0.12) and blood-stasis and special diathesis were not. Age was not appreciably correlated with the baseline SF-36 total score (ρ = 0.08, p = 0.06), but the change in QoL over six months differed by age group (mean change 5.1 points under 45 years, 0.2 points at 45–59 and 1.7 points at 60 or older; p = 0.002). Age was therefore treated as a potential source of heterogeneity, and the sufficiency of the configurations identified below was re-examined within age and sex strata (Sufficiency Analysis of Condition Configurations).

Before conducting necessity and configuration analyses of individual conditions in dynamic QCA, it is essential to calibrate the variable data according to the analytical requirements. This involves adjusting the values of all variables to a scale between 0 and 1 using the calibration function. Taking the total score of quality of life as an example, a higher quality of life score corresponds to a calibrated value closer to 1, while a lower score results in a value nearer to 0. Calibration should be based on three “anchor points” for each variable. Given the characteristics of the variable data, a direct calibration method is employed, setting the 95th percentile, 50th percentile, and 5th percentile as calibration anchors, which represent fully in, the crossover point, and fully out, respectively. The percentiles were computed once from the pooled data of all three time points (509 participants × 3 waves = 1527 case-observations), so that a single set of anchors, and hence the same definition of “high quality of life” and of a “high” score on each constitution subscale, applied at baseline, 3 months and 6 months. Anchors were taken from the data rather than from external norms because no consensus thresholds exist for a “high” SF-36 total score or for “high” membership on a constitution subscale in this population; the 5th and 95th percentiles were preferred to more extreme values so that full membership and full non-membership were each anchored on a reasonable number of observations rather than on a few outliers. The direct method fits a logistic function separately above and below the crossover, so the transformation does not assume that the distribution is symmetric about the median. For instance, if the total quality of life score reaches or exceeds the 95th percentile across all cases, it can be classified as “high”, thereby defaulting to a calibrated value of 1. Conversely, scores at or below the 5th percentile are deemed “low”, resulting in a default value of 0. The 50th percentile, being centrally positioned, does not provide a definitive indication of whether the quality of life is “high” or “low”. Following the calibration of variable values based on the established anchor points, to prevent the “exclusion” of sample cases in subsequent analyses, the membership value of 0.5 is adjusted to 0.501.²⁸ This adjustment is needed because a case with membership of exactly 0.5 belongs to neither the set nor its complement and would be dropped when the truth table is constructed; because converted constitution scores take a limited number of discrete values, a number of observations fell exactly on the median of a subscale. The adjustment moves the affected values by 0.001 and, as a check, recomputing the parameters of fit of the final configurations with the adjustment removed changed consistency and coverage by less than 0.001. The data and calibration results are presented in Table 3.

Table 3.

Data Situation and Variable Calibration (Score)

Variable	Calibration Anchors			Description
	Fully in (95th Percentile)	Crossover Point (50th Percentile)	Fully Out (5th Percentile)	Mean	SD	Max	Min
Outcome variable (quality of life)
Total quality of life	93.75	78.00	44.94	74.49	15.43	100.00	12.63
Condition variables (TCM constitution)
Gentleness constitution (A)	84.38	62.50	37.50	61.52	14.49	100.00	9.38
Qi-deficiency constitution (B)	59.38	31.25	9.38	32.10	15.17	87.50	0.00
Yang-deficiency constitution (C)	71.43	28.60	3.57	32.66	20.93	96.40	0.00
Yin-deficiency constitution (D)	52.17	25.00	6.25	26.88	14.20	78.10	0.00
Phlegm-dampness constitution (E)	56.25	28.10	6.25	28.02	15.46	90.60	0.00
Dampness-heat constitution (F)	54.20	25.00	4.17	26.58	16.32	87.50	0.00
Blood-stasis constitution (G)	53.60	32.14	7.14	31.64	13.98	82.10	0.00
Qi-depression constitution (H)	53.57	25.00	3.57	26.47	15.82	89.30	0.00
Special diathesis constitution (I)	46.43	17.86	0.00	19.67	14.54	75.00	0.00

Notes: Anchors are percentiles of the pooled data from the three time points (1527 case-observations) and were applied unchanged at baseline, 3 months and 6 months. Mean, SD, Max and Min also refer to the pooled data. Abbreviations: SD, standard deviation; TCM, Traditional Chinese Medicine.

Analysis of Necessity Based on Dynamic QCA for Single Conditions

The criterion for assessing necessary conditions is based on the consistency level, which reflects the accuracy of the condition variable in leading to the outcome variable. If the consistency value exceeds 0.9, the condition can be considered a necessary condition for the outcome variable.²⁹ Panel data QCA analysis yields results for pooled consistency, pooled coverage, BECONS adjusted distance, and WICONS adjusted distance. Currently, the consistency distances calculated by the panel QCA package in R are based on Euclidean distances, and further adjustment is needed following the correction formulas provided by Garcia-Castro²⁰ to ensure the distance standards are applicable to datasets of any sample size. Pooled consistency and pooled coverage are used to evaluate the overall consistency and coverage of conditions across time and individuals. Consistency distance is used to assess whether significant temporal or individual effects are present. If the BECONS distance or WICONS distance is too large, it indicates that significant temporal or individual (or cluster) effects are influencing the subset relationship between conditions and outcomes, which necessitates further investigation into the necessity of the conditions.^30,31 When the adjusted distance is less than 0.2, the accuracy of the pooled consistency is considered high and can be used as a basis for judgment.²⁰ The results of the current study, shown in Table 4, indicate that the between group adjusted distances for all nine condition variables are less than 0.2, and the pooled consistency values are all below 0.9, suggesting that the strength of a single TCM constitution is not a necessary condition for the outcome variable. However, the within group adjusted distances exceeding 0.2 imply that the relationship between TCM constitution and quality of life is strongly influenced by individual differences, warranting further exploration into the relationship between condition configurations and outcomes. The data calibration and necessity analysis were conducted using the SetMethods package in R version 4.2.1.

Table 4.

Necessary Condition Analysis

Condition Variable	High Quality of Life
	Pooled Consistency	Pooled Coverage	BECONS Adjusted Distance	WICONS Adjusted Distance
High gentleness constitution (A)	0.785	0.820	0.026	0.249
Non high gentleness constitution (~A)	0.523	0.542	0.034	0.453
High qi-deficiency constitution (B)	0.517	0.554	0.034	0.475
Non high qi-deficiency constitution (~B)	0.791	0.800	0.031	0.249
High yang-deficiency constitution (C)	0.534	0.560	0.023	0.475
Non high yang-deficiency constitution (~C)	0.755	0.779	0.013	0.294
High yin-deficiency constitution (D)	0.553	0.577	0.039	0.453
Non high yin-deficiency constitution (~D)	0.740	0.768	0.008	0.294
High phlegm-dampness constitution (E)	0.515	0.575	0.028	0.475
Non high phlegm-dampness constitution (~E)	0.782	0.763	0.015	0.272
High dampness-heat constitution (F)	0.547	0.582	0.067	0.453
Non high dampness-heat constitution (~F)	0.733	0.746	0.041	0.317
High blood-stasis constitution (G)	0.561	0.583	0.010	0.407
Non high blood-stasis constitution (~G)	0.739	0.771	0.005	0.272
High qi-depression constitution (H)	0.524	0.552	0.026	0.475
Non high qi-depression constitution (~H)	0.775	0.797	0.026	0.249
High special diathesis constitution (I)	0.541	0.592	0.067	0.520
Non high special diathesis constitution (~I)	0.744	0.737	0.036	0.385

Notes: “~” denotes the set of non-high scores; for example, ~A indicates non-high gentleness constitution scores, and similarly for the other conditions. Abbreviations: BECONS, between-group consistency; WICONS, within-group consistency.

Sufficiency Analysis of Condition Configurations

Configurational analysis is at the core of the QCA method, aiming to explore how different combinations of antecedent conditions relate to the outcome. The primary criterion for this assessment is the level of consistency regarding sufficiency, where Schneider and Wagemann suggest a threshold of no less than 0.75 for consistency.³² To conduct a configurational analysis, a truth table must be constructed. Drawing on previous research, we set the minimum number of cases required for a configuration leading to high quality of life at 10, meaning that a configuration is considered meaningful only if it appears in at least 10 cases. Furthermore, for the configuration to meet the analytical requirements, its consistency (understood as accuracy) must exceed 0.8, and the PRI (Proportional Reduction in Inconsistency, which measures the degree to which a configuration is a subset of the outcome rather than of its negation) must be at least 0.6.³³ It should be noted that consistency, PRI and coverage are descriptive parameters of set relations, not test statistics; QCA does not produce p-values, and the thresholds are conventions for deciding which set relations are strong enough to be reported.^32,34 Accordingly, no multiplicity correction is applied. Instead, robustness is examined by varying the thresholds and by checking whether the same configurations recur across time points and, in this study, across age and sex strata.

Given the advantages of panel data analysis and the emphasis on the precision of configurations as “minimally sufficient conditions,” we utilized dynamic QCA and “bottom-up” QCA with nine TCM constitutions, replacing traditional QCA techniques (such as fsQCA). The results obtained provide complementary insights and validation. (1) Dynamic QCA Method: Using the SetMethods package in R version 4.2.1, the configuration consistency threshold was set at 0.8, and the PRI value at 0.7 to construct a temporal data truth table This allowed us to identify the configurations of physical constitutions associated with high quality of life across three time periods, before and after the intervention (Table 5). The consistency values for all five configurations were above 0.9, indicating a high level of set-theoretic consistency. A condition common to all configurations associated with high quality of life was a high score in the gentleness constitution, while the scores for the other eight biased constitutions were relatively low to varying degrees. The five configurations differ from one another only in which one or two biased constitutions are left unrestricted (dampness-heat and qi-depression in configuration a; phlegm-dampness and dampness-heat in b; blood-stasis in c; yin-deficiency and special diathesis in d; qi-deficiency in e). They therefore overlap heavily: the raw coverage of each is 0.40–0.44, but the unique coverage is 0.006–0.049, and together they cover 54% of the membership in the high-QoL set. Each configuration was held (membership > 0.5) by 142–187 of the 1527 case-observations and the solution as a whole by 309. Between-group and within-group adjusted distances were all below 0.01, indicating that these pooled parameters were stable across the three time points and across individuals. (2) “Bottom-Up” QCA Methods: This method is derived from Coincidence Analysis (CNA) and the logical comparison approach of Qualitative Comparative Analysis (QCA),^22,35 and was implemented in this study using the QCA package in R version 4.2.1. Unlike the traditional top-down approach to QCA, the “Bottom-Up” QCA Methods begins by examining the sufficiency of individual conditions and then incrementally combines conditions until all minimally sufficient configurations are identified. This approach better aligns with the analytical objectives of QCA, as it generates solutions that meet the criteria of sufficiency, necessity, and completeness. Compared to the top-down logic of traditional QCA, “Bottom-Up” QCA Methods offers several advantages: ① Avoidance of truth tables: It does not rely on binary truth tables, which helps avoid ambiguity in fuzzy set data; ② Avoidance of counterfactual assumptions: The method is entirely based on empirical data, removing the need for researchers to make assumptions during the analysis, thereby mitigating the risk of erroneous assumptions; ③ Reduction of complexity: More relevant conditions can be included without increasing the complexity of the solution; ④ Addressing omitted variable issues: Adding new conditions does not affect the accuracy of previously identified conditions.

Table 5.

Results of the Dynamic QCA Configuration Analysis

Condition Variable	Configuration
	a	b	c	d	e
Gentleness constitution (A)	●	●	●	●	●
Qi-deficiency constitution (B)	⊗	⊗	⊗	⊗
Yang-deficiency constitution (C)	⊗	⊗	⊗	⊗	⊗
Yin-deficiency constitution (D)	⊗	⊗	⊗		⊗
Phlegm-dampness constitution (E)	⊗		⊗	⊗	⊗
Dampness-heat constitution (F)			⊗	⊗	⊗
Blood-stasis constitution (G)	⊗	⊗		⊗	⊗
Qi-depression constitution (H)		⊗	⊗	⊗	⊗
Special diathesis constitution (I)	⊗	⊗	⊗		⊗
Consistency	0.948	0.947	0.948	0.946	0.953
PRI	0.890	0.889	0.892	0.888	0.898
Coverage	0.430	0.433	0.431	0.444	0.401
Unique coverage	0.016	0.019	0.036	0.049	0.006
Cases with membership > 0.5 (of 1527)	176	173	177	187	142
BECONS adjusted distance	0.003	0.008	0.005	0.003	0.003
WICONS adjusted distance	<0.001	<0.001	<0.001	<0.001	<0.001
Overall PRI	0.858
Overall solution consistency	0.927
Overall solution coverage	0.540
Cases covered by the overall solution (membership > 0.5)	309

Notes: Large filled circles (●) indicate the presence of a core condition (high score), and large crossed circles (⊗) indicate the absence of a core condition (low score). All marked conditions are core conditions; no peripheral conditions are displayed. Blank cells indicate that the condition is unrestricted and may be either high or low. Gentleness constitution is present in all five configurations and is referred to in the text as the common (core) condition; the biased constitutions enter in differing combinations and are referred to as peripheral conditions. This usage is descriptive and does not rest on a comparison of parsimonious and intermediate solutions. Cases are case-observations (509 participants × 3 time points) whose fuzzy-set membership in the configuration exceeds 0.5. Abbreviations: BECONS, between-group consistency; PRI, proportional reduction in inconsistency; QCA, qualitative comparative analysis; WICONS, within-group consistency.

Using the superSubset function in the QCA package (which allows researchers to identify all minimally sufficient conditions in sufficiency analysis and automatically remove redundant conditions), we performed a bottom-up analysis at three different time points: before the intervention, three months post-intervention, and six months post-intervention. In this analysis, the scores of nine TCM constitution types were used as conditions, and the total quality of life score as the outcome. In order to obtain relatively comprehensive minimum sufficient condition configuration information, PRI values were set to 0.65, 0.70 and 0.75 respectively as a graded robustness check: the lowest threshold gives the fullest list of minimally sufficient conditions, and raising it progressively retains only those whose membership is most clearly a subset of high QoL rather than of its complement. The complete results at all three thresholds are given in Supplementary Table S3; the main text reports the strictest threshold (PRI ≥ 0.75) (Table 6 and Figure 2C).

Table 6.

Minimally Sufficient Constitution Configurations for High Quality of Life Identified by Bottom-up QCA at a PRI Threshold of 0.75, by Time Point

Time Point	Configuration	Consistency	PRI	Raw Coverage	Unique Coverage
Before intervention	~C*~H (low yang-deficiency and low qi-depression)	0.872	0.783	0.645	0.021
	~G*~H (low blood-stasis and low qi-depression)	0.873	0.779	0.622	0.023
	~H*~I (low qi-depression and low special diathesis)	0.861	0.756	0.614	0.018
	Overall solution	0.830	0.724	0.729	—
3 months after intervention	~C*~E (low yang-deficiency and low phlegm-dampness)	0.872	0.764	0.644	0.053
	~C*~H (low yang-deficiency and low qi-depression)	0.867	0.762	0.667	0.023
	~G*~H (low blood-stasis and low qi-depression)	0.867	0.756	0.631	0.060
	Overall solution	0.825	0.705	0.780	—
6 months after intervention	~B*~C (low qi-deficiency and low yang-deficiency)	0.856	0.751	0.672	0.023
	~B*~G (low qi-deficiency and low blood-stasis)	0.860	0.751	0.641	0.061
	~C*~E (low yang-deficiency and low phlegm-dampness)	0.866	0.765	0.633	0.041
	Overall solution	0.817	0.702	0.774	—

Notes: Configurations are minimally sufficient combinations of two conditions identified with the superSubset function; single conditions and pairs retained at the less stringent thresholds of PRI ≥ 0.65 and ≥ 0.70 are listed in Supplementary Table S3. “~” denotes a low score (non-membership in the set of high scores) on the subscale indicated by the letter (B, qi-deficiency; C, yang-deficiency; E, phlegm-dampness; G, blood-stasis; H, qi-depression; I, special diathesis); “*” denotes logical AND. Raw coverage is the proportion of membership in the high-QoL set covered by the configuration; unique coverage is the proportion covered by that configuration alone. Abbreviations: PRI, proportional reduction in inconsistency; QCA, qualitative comparative analysis; QoL, quality of life.

When the PRI was set at 0.65, the results provided a relatively comprehensive overview of the minimally sufficient configurations across the three stages that were associated with high quality of life (Supplementary Table S3). High gentleness scores (A) and low qi-deficiency scores (~B) were each sufficient on their own at baseline and 3 months, and A and low qi-depression scores (~H) at 6 months, with consistency of 0.80–0.83. In addition, various pairs of low biased constitution scores were also identified as minimally sufficient conditions for high quality of life, which aligns with the results of the dynamic QCA but with more detailed information.

At a PRI value of 0.70, while high gentleness constitution scores remained a sufficient condition, some differences were observed in the configurations of low biased constitution scores compared to the PRI 0.65 analysis. Before the intervention and three months post-intervention, configurations such as low yang-deficiency constitution scores (~C), low blood-stasis constitution scores (~G), and low qi-depression constitution scores (~H) frequently appeared as components of pairs associated with high quality of life. After six months, configurations including low qi-deficiency constitution scores (~B), low yang-deficiency constitution scores (~C), low phlegm-dampness constitution scores (~E), and low qi-depression constitution (~H) appeared more frequently, reflecting the shifting relationship between constitution configurations and high quality of life throughout the intervention process.

When the PRI was set at 0.75, a stricter and more precise filtering of the minimally sufficient conditions for high quality of life was achieved (Table 6 and Figure 2C). Before the intervention and three months post-intervention, constitution configurations associated with high quality of life consistently included low yang-deficiency constitution scores and either low qi-depression constitution scores (~C*~H) or low blood-stasis constitution scores combined with low qi-depression constitution scores (~G*~H), with consistency of 0.867–0.873 and raw coverage of 0.62–0.67. Additionally, before the intervention, configurations involving low qi-depression constitution scores and low special-diathesis constitution scores (~H*~I) were also identified, while three months post-intervention, low yang-deficiency constitution scores combined with low phlegm-dampness constitution scores (~C*~E) emerged. The same constitution configurations that were associated with high quality of life 6 months after the intervention as 3 months earlier had low yang-deficiency constitution scores and low phlegm-dampness constitution scores (~C*~E), and differed in that there were also low qi-deficiency constitution scores and low yang-deficiency constitution scores (~B*~C) or low qi-deficiency constitution scores and low blood-stasis constitution scores (~B*~G) (consistency 0.856–0.866). We emphasize that the retention or dropping of a pair between time points reflects its PRI crossing the 0.75 threshold; the pairs dropped at 6 months (~C*~H, ~G*~H) remained sufficient at PRI ≥ 0.70 (Supplementary Table S3), so the change is a shift in the relative ordering of configurations rather than the disappearance of a relationship, and no statistical test of the difference between time points was performed.

As a check on whether the configurational results were driven by the composition of the sample, the five dynamic-QCA configurations were re-evaluated within age strata (< 45, 45–59 and ≥ 60 years) and by sex using the pooled calibration (Supplementary Table S2). Consistency remained above 0.90 in every stratum (0.901–0.973 for the individual configurations; 0.879–0.953 for the overall solution), and the same high-gentleness/low-biased pattern held throughout. Coverage was higher in participants aged 60 or older (solution coverage 0.617 vs 0.505–0.519 in the younger strata) and in men (0.587 vs 0.509 in women), meaning that the identified configurations account for a larger share of high-QoL cases in those groups, but no stratum produced a qualitatively different pattern. This does not adjust for age or sex in the way a regression model would, and other potential confounders (comorbidity, medication, socioeconomic position, lifestyle) were not examined; the limitation is discussed in Discussion.

Discussion

This study, by analyzing the relationship between TCM constitutions and quality of life, describes the association between TCM constitution configurations and quality of life across different stages of a 6-month constitution-based intervention (Figure 2; summarized schematically in Figure 3). Overall, higher scores in the gentleness constitution, along with lower scores in qi-deficiency constitution and yang-deficiency constitution, were associated with better quality of life. This observation is in line with the findings of Sun et al,³⁶ who reported that regulating TCM constitutions such as qi-deficiency constitution and yang-deficiency constitution through TCM interventions was accompanied by improved cognitive function and mental health in elderly patients, especially those with long-standing qi-deficiency constitution and yang-deficiency constitution. Qi deficiency and Yang deficiency belong to the same category of deficiency. People with qi deficiency are often characterised by easy fatigue and shortness of breath, while people with yang deficiency are often characterised by fear of cold, coldness and lack of warmth in the hands and feet.¹² Both qi and yang deficiency are held in TCM theory to cause low energy, low immunity and poor digestion, leading to a decline in quality of life. If so, measures directed at qi and yang deficiency would be a natural place to start when the aim is to improve quality of life, a proposition that the present data are consistent with but cannot establish.²⁵ Additionally, the reduction in blood-stasis constitution and phlegm-dampness constitution scores during the intervention is compatible with the TCM principle of treating both the root cause and symptoms. In traditional Chinese medicine, Blood-stasis and Phlegm-dampness are often closely linked to chronic symptoms such as pain and fatigue; on this reasoning, regulating these constitutions would be expected to alleviate patients’ symptoms and improve their quality of life. Liu et al,³⁷ in their study on patients with postherpetic neuralgia, found that regulating phlegm-dampness constitution and blood-stasis constitution was associated with reduced pain and improved quality of life, which is consistent with the direction of our findings.

Figure 3.

Schematic summary of the configurational findings. The upper part depicts the two analytic approaches applied to the 509 participants at baseline (circle), 3 months (triangle) and 6 months (star): the five dynamic-QCA configurations (left; each combines a high gentleness score, ↑, with low scores, ↓, on the biased constitutions listed) and the minimally sufficient two-condition configurations retained by bottom-up QCA at PRI ≥ 0.75 at each time point (right). The centre shows the nine constitution types (A–I). The lower part restates the two sets of results: high gentleness with broadly low biased constitutions from dynamic QCA, and the time-specific pairs from bottom-up QCA. The figure is an illustrative summary of Table 5 and Table 6; it does not represent statistical tests of change between time points. Abbreviations as in Figure 2.

Two features of the configurational results deserve comment. The first is that a high gentleness score appears in every configuration. This is partly a matter of measurement: the gentleness subscale asks about energy, sleep, appetite, adaptability to weather and mood, content that overlaps with the vitality, general health and mental health domains of the SF-36, and by construction its score moves inversely with the biased subscales. It is also what constitution theory would predict, since gentleness is the balanced state that the biased types deviate from, so that high QoL without a reasonably high gentleness score would be surprising. The second feature is that the five dynamic-QCA configurations overlap heavily, with unique coverage of at most 0.049. We do not read this as five distinct pathways to high QoL in the sense of strong equifinality. Rather, the data describe one broad pattern (high gentleness with low scores on most biased constitutions) within which one or two biased constitutions can remain elevated without the person leaving the high-QoL set, and the configurations differ in which those are. The more informative findings are therefore the bottom-up results, which identify the specific pairs of biased constitutions whose joint absence is, on its own, sufficient for high QoL at each time point.

In terms of the course of the intervention, low qi-depression constitution scores combined with low yang-deficiency constitution, low blood-stasis constitution scores, and low special-diathesis constitution scores, respectively, were the minimum sufficient conditions for high quality of life before the intervention began. As the TCM constitution interventions progressed, these configurations shifted to low qi-deficiency constitution scores and low yang-deficiency constitution scores or low blood-stasis constitution scores, as well as low yang-deficiency constitution scores combined with low phlegm-dampness scores. This parallels the findings of Zhang et al,³⁸ who reported improvements in respiratory symptoms through phlegm-dampness elimination using TCM treatments. One reading of the shift, consistent with the expectation set out in Background, is that qi-depression, which is largely an emotional and stress-related dimension, was the biased constitution that most clearly separated high from low QoL in an unselected sample, and that once participants had spent several months attending to emotion, sleep and daily routine it no longer did so, leaving the deficiency-type constitutions (qi- and yang-deficiency) and their stagnation-type sequelae (blood-stasis, phlegm-dampness) as the discriminating features. TCM theory describes these mechanisms in terms of promoting qi,^39,40 supporting yang and resolving blood stasis. This study preliminarily outlines such a sequence, but the shift was not subjected to a statistical test, the retained pairs at successive time points overlap substantially, and the sequence should be regarded as a hypothesis for prospective testing rather than an established progression.

The age pattern observed here is also worth noting. Older participants had, on average, higher gentleness and lower deficiency- and heat-type scores at baseline, which is consistent with earlier large-sample descriptions of constitution across the lifespan⁸ and may reflect both survival and lifestyle differences; younger participants, who started from a lower QoL and a more biased profile, showed the largest change over six months. Despite these differences the configurations held with high consistency in all age strata and in both sexes, and coverage was, if anything, higher in the older group. Age therefore modifies where people sit within the configurational space more than it changes the configurations themselves, at least on the evidence available here.

The findings also illustrate some of the practical features of constitution-based intervention. Firstly, TCM constitution intervention is able to create personalised intervention plans based on the diverse TCM constitutions of participants, in this study through a written plan and manual covering emotion, diet, daily living, exercise and acupoint self-care that participants implemented themselves. Secondly, the intervention is comprehensive rather than single-component, which is in keeping with the TCM emphasis on balancing of yin and yang and on “treating the disease before it occurs”.^41,42 Whether these features translate into an effect on quality of life beyond what would be seen with time, attention or general health advice is a question for controlled studies, and this study does not answer it.

Methodologically, the application of dynamic QCA method offers distinct advantages in describing the complex relationship between TCM constitutions and quality of life. Traditional statistical methods, such as regression analysis, often struggle to capture the intricate interactions between multiple constitutions, especially in terms of temporal changes.⁴³ In contrast, dynamic QCA method, through the analysis of panel data, allows the configurations associated with high QoL to be compared across time points and tests, through the adjusted between- and within-group distances, whether pooled parameters are stable over time and across individuals. This method not only allows for the analysis of TCM constitution states at single time points but also tracks the dynamic changes in TCM constitutions throughout the intervention. Additionally, the “Bottom-up” QCA method, by incrementally identifying minimally sufficient conditions, avoids the counterfactual assumptions and Boolean minimization of the truth-table approach and returns every minimally sufficient combination rather than a single solution, providing a more comprehensive picture of constitution configurations. We regard the present study as much a demonstration of this analytic approach in constitution research as a substantive clinical finding.

The study has substantial limitations, and they bear directly on how the results should be read. The most important is the absence of a control or comparison group. The 509 participants all received the intervention, so the improvement in QoL and the fall in biased constitution scores between baseline and 6 months cannot be attributed to the intervention; regression to the mean, seasonal variation (the study ran from February to the following January), repeated measurement, and the natural course of participants’ health could each have contributed, and the size of the average change (about 0.2 standard deviations) is below the half-standard-deviation benchmark usually taken as noticeable to patients.¹⁸ What the study can describe is which combinations of constitution scores accompanied high QoL at each time point; it cannot say that changing those constitutions would change QoL. We have framed the findings accordingly, and the clinical statements in earlier drafts have been recast as hypotheses. Second, the QCA models contain only the nine constitution scores. Age, sex, comorbidity, medication use, education, socioeconomic position and lifestyle were not included as conditions, both because QCA is limited in the number of conditions it can handle with this number of cases and because the aim was to characterize constitution configurations rather than to build a predictive model. The stratified analysis shows that the configurations were stable across age and sex, but confounding by the other factors, several of which were not measured in enough detail, remains possible. Third, 14.5% of those enrolled did not complete all three assessments. Completers and non-completers were similar on the baseline variables we could compare, and complete-case analysis was dictated by the requirements of panel QCA, but differential loss on unmeasured characteristics cannot be ruled out. Fourth, the calibration anchors were derived from the data. This is the usual practice when external norms are lacking, and a single pooled set of anchors was used across the three waves, but “high QoL” is defined relative to this sample and the configurations may look different in a population with a different QoL distribution; using the same data to calibrate and to analyse also means the parameters of fit are best regarded as descriptive.

Several further considerations apply. The sample of 509 participants represents a substantial cohort for a 6-month intervention study, though recruitment was by convenience from a single hospital-based centre in Zhuhai, Guangdong Province, and about 70% of participants were women, so the findings should not be generalized to other regions or settings without replication.⁴⁴ The 6-month observation captured temporal changes through dynamic QCA, but long-term sustainability was not examined. Whether beneficial constitution configurations last beyond this period or need periodic adjustments is unknown. The intervention itself was a multi-component self-management package, and adherence, which varied by age group,¹⁷ was not modelled here; we cannot say which components, if any, were responsible for the changes in constitution scores.

Self reported questionnaires, validated and widely used for assessing TCM constitutions and quality of life, carry inherent subjectivity influenced by emotional states and cognitive factors, and because constitution and QoL were both self-reported at the same visit, part of their association may reflect shared response tendencies. Incorporating objective physiological indicators such as immune system function and blood biochemical markers could complement these findings while elucidating the biological mechanisms underlying the constitution–QoL relationship. We also used the SF-36 total score as a single outcome; analyses of the component summaries or of individual domains might identify different configurations, particularly for mental health. From a methodological perspective, analyzing constitution configurations associated with persistently low quality of life was not undertaken here and would provide complementary insights; QCA treats the outcome and its negation asymmetrically, so the configurations for low QoL cannot be inferred from those reported.

Conclusion

In summary, in adults taking part in a 6-month constitution-based comprehensive intervention, combinations of TCM constitutions, rather than single constitution types, were associated with high quality of life. Every configuration combined a high gentleness score with low scores on several biased constitutions, and the configurations overlapped to a large degree. At baseline and 3 months, low Yang-Deficiency with low Qi-Depression (low C + low H) and low Blood-Stasis with low Qi-Depression (low G + low H) were the most consistent two-condition configurations linked to high quality of life; at 6 months the key configurations were low Qi-Deficiency with low Yang-Deficiency or Blood-Stasis (low B + low C / low B + low G), and low Yang-Deficiency with low Phlegm-Dampness (low C + low E). This pattern is consistent with the idea that deficiency-type constitutions and their stagnation-type sequelae become the discriminating features once emotional and daily-routine deviations have been addressed, but the design was uncontrolled and the pattern was not tested statistically. The results are therefore best treated as hypothesis-generating: they identify specific combinations of constitutions that could be prioritized as targets, and as outcomes, in controlled trials of constitution-based personalized health management, and they show that dynamic and bottom-up QCA are workable tools for that purpose.

Funding Statement

This project was supported by Open Bidding for Selecting the Best Candidates Project of Beijing University of Chinese Medicine of JIE BANG GUA SHUAI Project (2022-JYB-JBZR-036). The funder played no role in the study’s design, execution, data collection, management, analysis, interpretation, manuscript preparation, review, or approval, nor in the decision to submit the manuscript for publication.

TCM Constitution Types

Type A: Gentleness Constitution; Type B: Qi-Deficiency Constitution; Type C: Yang-Deficiency Constitution; Type D: Yin-Deficiency Constitution; Type E: Phlegm-Dampness Constitution; Type F: Damp-Heat Constitution; Type G: Blood-Stasis Constitution; Type H: Qi-Depression Constitution; Type I: Special-Diathesis Constitution.

Clinical Trial Number

This study is a self-care lifestyle intervention undertaken by the study participants. It does not involve invasive procedures or interventions by physicians or nursing staff; therefore, a Clinical Trial Number is not applicable.

Declaration of Generative AI and AI-Assisted Technologies in the Manuscript Preparation Process

The authors did not use generative AI or AI-assisted technologies in the writing process.

Abbreviations

TCM, Traditional Chinese Medicine; QoL, Quality of Life; CCMQ-60, Constitution in Chinese Medicine Questionnaire (60 items); SF-36, 36-Item Short Form Health Survey; QCA, Qualitative Comparative Analysis; fsQCA, Fuzzy-set Qualitative Comparative Analysis; PRI, Proportional Reduction in Inconsistency; CAN, Coincidence Analysis; BECONS, Between-group Consistency; WICONS, Within-group Consistency; CI, confidence interval; SD, standard deviation.

Data Sharing Statement

All data are derived from the self-established databases. The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Ethics Approval

This study was approved by the Medical and Experimental Animal Ethics Committee of Beijing University of Chinese Medicine (Approval Number: 2012BZHYLL0301). The procedures used in this study adhere to the tenets of the Declaration of Helsinki.

Consent to Participate

Informed consent was obtained from all individual participants included in the study.

Consent to Publish

All individual participants signed informed consent regarding publishing their data.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.