Psychometric Properties of The Sub-Health Measurement Scale V1.0 for Assessing Suboptimal Health Status of Midwives: A Multi-Centre Cross-Sectional Study


 Purpose Midwives with high work stress are prone to suboptimal health status (SHS). However, there is a lack of unified measurement standard for SHS. Sub-Health Measurement Scale V1.0 (SHMS V1.0) is a multidimensional inventory that has been proved to have adequate psychometric properties in other populations. Therefore, this study aimed to evaluate validity and reliability of SHMS V1.0 in midwives.Methods In this cross-sectional study, 842 midwives from 46 midwifery institutions completed the SHMS V1.0. Internal consistency and test-retest reliability of SHMS V1.0 were assessed with Cronbach's alpha and Pearson correlation coefficient. Project analysis and confirmatory factor analysis (CFA) were performed to assess construct validity. Concurrent validity of SHMS V1.0 was evaluated by Pearson correlation analysis with the 14-item Fatigue Scale (FS-14).Results The total Cronbach's alpha and test-retest reliability coefficient of SHMS V1.0 was 0.908 and 0.804, respectively. The correlation coefficients between each item and its corresponding dimension, and those between each dimension and its corresponding subscale were all greater than 0.4. The CFA supported the structure of SHMS V1.0 with strong factor loadings and excellent fit indexes in the first-order and modified second-order factor model. The concurrent validity for SHMS V1.0 showed large correlation with FS-14, but social subscale showed medium correlations with total fatigue, physical fatigue and mental fatigue. Conclusions The SHMS V1.0 is a valid and reliable instrument for measuring SHS for midwives. Given the severity and harmfulness of SHS in midwives, SHMS V1.0 is recommended as an important tool for clinical practice and research.


Introduction
As an integral part of healthcare professionals, midwives play an important role in ensuring and promoting the health of pregnant women and newborns [11]. Midwives are especially prone to SHS due to frequent rotating shift, shortage of human resources and high work stress [12][13][14]. Midwives have reported moderate to severe levels of exhaustion on 22-50% of all shifts and rest days [15], and there were 59.3% of midwives with SHS according to the TCM guidelines for SHS [16]. The United Nations Population Fund and the World Health Organization (WHO) have pointed out that there was a large shortage of midwives in the world [13]. Furthermore, the number of older mothers and obstetric complications has increased without a proportional increase in midwives since the opening of Chinese two-child policy [17], which makes midwives bear more work load and suffer from sleep disorders, fatigue, anxiety and burnout, and then accounts for lower job satisfaction, higher turnover and degradation in work quality [18][19][20]. Therefore, it is important to detect SHS among midwives for preventing it.
However, there is a lack of uni ed measurement standard for SHS. The diagnosis of SHS can be divided into subjective and objective measures, and the objective measurements for SHS such as blood pressure, pulse, lipid and plasma glycoprotein are still being explored [21,22]. Subjective measurement is the main clinical diagnostics for SHS after excluding speci c disease according to comprehensive physical examination [1]. SHS questionnaires have been developed and are widely used as diagnostic instruments of SHS in China, including Sub-Health Measurement Scale V1.0 (SHMS V1.0) [23], 36-item Short-Form Health Survey Questionnaire (SF-36) [24], and Suboptimal Health Status Questionnaire-25 (SHSQ-25) [25]. In particular, SHMS V1.0 is a multidimensional inventory that reports physiological, psychological and social symptoms [23], which is highly consistent with the WHO de nition of health [26]. And SHMS V1.0 has been proved to have adequate psychometric properties in nurses (with Cronbach's alpha and split half reliability coe cient of 0.924, 0.863) [27], urban residents (with Cronbach's alpha and split half reliability coe cient of 0.928, 0.847) [28], and college students (with good construct validity) [29], which all indicate that it is a reliable and valid measurement scale for SHS. However, the reliability and validity of SHMS V1.0 in midwives are not known. Therefore, the aim of this study was to investigate the psychometric properties of SHMS V1.0 in midwives.

Study design and participants
A multi-centre cross-sectional study was conducted among midwives through strati ed cluster random sampling in Fujian province of China from March to June 2019. The sample size was calculated based on the item numbers (35 items×10=350) [30]. Thus, the minimum sample size was 420 after considering a 20% non-response rate. The inclusion criteria for midwives were as follows: (1) obtained professional quali cation certi cates and work as independent position; (2) volunteer to participate in this study. The exclusion criteria were as follows: (1) took maternity leave or left their hospital for training for more than one month; (2) diagnosed as acute or chronic diseases. According to the annual report number of midwives in the province, 186 midwifery institutions with at least 5 midwives were selected in the study.
Then total 900 participants eligible for inclusion were recruited from 37 secondary midwifery institutions and 9 tertiary midwifery institutions, which were randomly selected according to the proportion of secondary and tertiary midwifery institutions in the 186 midwifery institutions by computer-generated random numbers.
The study was approved by the ethical committee of the main researcher's hospital. Before the survey, oral informed consent was obtained from all nursing departments of the 46 hospitals and all participants.

Measures
Online questionnaires were used to collect the data in the study including socio-demographic questionnaire, SHMS V1.0 and the 14-item Fatigue Scale . The socio-demographic questionnaire was designed by the researchers, including hospital nature, age, education level, years of midwifery experience, professional rank, professional position, marital status, and monthly income.
The FS-14 developed by Trudie Chalder et al [31] was used to measure the severity of participants' fatigue over the last month. FS-14 comprises 14 items with two dimensions: physical (items 1-8) and mental (items 9-14) fatigue. The scoring pattern for each item is a four-point Likert scale (0=less than usual, 1=no more than usual, 2=more than usual, 3=much more than usual). The total fatigue score obtained by adding up all items ranges from 0 to 42, with higher scores indicating more severe fatigue. The Chinese version of the FS-14 has been shown acceptable psychometric properties [32].

Data collection
The questionnaires survey was conducted through an online platform named "WenJuanXing"( ) which was used to make the survey link. Before the survey, the nurses in charge of labor wards in the 46 midwifery institutions were trained and informed of the purpose of the research, the requirements of choosing participants, the lling method and precautions of the questionnaire by phone. The main researcher sent the survey invitation and questionnaire link to those nurses in charge through Wechat which is an online communication platform. Then 46 nurses in charge selected the subjects according to the inclusion and exclusion criteria, explained the purpose of this study and the lling method of questionnaire, and sent the survey link to the eligible participants. All midwives who agreed to participate in the survey clicked on the survey link, completed and submitted the questionnaire. According to the data integrity of the "WenJuanXing" platform, this process was promoted by weekly reminders and data collection was completed after three reminders. Besides, 70 midwives were randomly selected and received the same tests again to investigate the test-retest reliability of SHMS V1.0 two weeks after the initial test.

Data analysis
All data were analyzed using the IBM SPSS 25.0 and IBM AMOS 24.0, which was used to conduct con rmatory factor analysis (CFA). Frequency, percentage, means and standard deviations (SD) were used to describe socio-demographic characteristics of midwives. For the reliability of SHMS V1.0, the internal consistency and test-retest reliability were performed by Cronbach's alpha coe cient and Pearson correlation coe cient, with values of at least 0.70 indicating acceptable reliability [33]. For the validity of SHMS V1.0, project analysis (item-total correlation) was used to test the homogeneity of each dimension and subscale, with the correlation coe cient greater than 0.40 considered as acceptable homogeneity [30]; the concurrent validity was evaluated by Pearson correlation analysis with FS-14, with the effect size as "Low" (the value of correlation coe cient varied around 0.1), "Medium" (the value of correlation coe cient varied around 0.30) and "Large" (the value of correlation coe cient varied more than 0.50) [34]; and the construct validity was performed by CFA, with the factor loadings being at least 0.40 indicating a su cient correlation [35]. Good t indexes for the models were indicated by the ratio between χ 2 and the degrees of freedom of the model (χ 2 /df) less than 2, Root Mean Square Error of Approximation (RMSEA) less than 0.08, Incremental Fit Index (IFI), Tucker-Lewis index (TLI) or Comparative Fit Index (CFI) greater than 0.90, Parsimony Goodness-of-Fit Index (PGFI) greater than 0.50 [30,36]. All tests were two-sided and p-values less than 0.05 were regarded as statistically signi cant.

Results
A total of 900 midwives were eligible and invited to participate in the study, of which 842 completed all questionnaires, with 25 without submitting questionnaire and 33 unaccomplished questionnaires.

Participant characteristics
The participant socio-characteristics are shown in Table 1. Those midwives were from general (65.80%), specialized (28.15%) and private (6.06%) hospital. The mean age of midwives was 31.41 years (SD = 6.42). Most midwives (54.16%) had a junior college degree. Approximately half of midwives had less than ve years of midwifery (45.25%) and senior professional rank (49.29%) with 58 nurses in charge (6.89%). Most midwives were married (71.85%) and 61.40% of midwives had monthly income ranging from 3000 to 5999RMB.

Reliability
In terms of internal consistency of SHMS V1.0, the total Cronbach's alpha coe cient of the scale was 0.908. The Cronbach's alpha coe cients for physiological, psychological and social sub-health subscale were 0.714, 0.893 and 0.846, respectively. In terms of test-retest reliability of SHMS V1.0, the Pearson correlation coe cient was 0.804. The test-retest reliability for physiological, psychological and social sub-health subscale were 0.715 (p < 0.001), 0.861 (p < 0.001) and 0.503 (p = 0.010), respectively.

Project analysis
The results of project analysis are shown in Table 2 and Table 3. The correlation coe cients between each item and its associated dimension score ranged from 0.619 to 0.890, higher than those between the item and other dimensions, which were all statistically signi cant (p < 0.01). The correlation coe cients between each dimension and its associated subscale score ranged from 0.614 to 0.911, higher than those between the dimension and other subscales, which were all statistically signi cant (p < 0.01). The dimension-total correlations coe cients varied between 0.615 and 0.800, which were statistically signi cant (p < 0.01).

Con rmatory Factor Analysis
The results of CFA including the rst-order factor model, the second-order factor model and the modi ed second-order model are showed in Figure 1, Figure 2 and Figure 3. The t indexes for CFA are showed in Table 4. For the rst-order factor model, the loadings of each item in the corresponding dimension were greater than 0.4. The t indexes revealed the following results: χ 2 /df was less than 2; RMSEA was less than 0.08; IFI, TLI and CFI were greater than 0.90; PGFI was greater than 0.50. For the second-order factor model, the loadings of each item in the corresponding dimension were greater than 0.4 and the loadings of each dimension in corresponding subscale were greater than 0.6, but the model demonstrated no optimal t indexes with χ 2 /df greater than 2 and TLI less than 0.90. Based on modi cation indexes, two covariance factors between the errors for item 30 to 31, and item 20 to 21 were added. The modi ed second-order factor model showed good model t indexes with χ2/df less than 2, RMSEA less than 0.08, IFI, TLI and CFI greater than 0.90, PGFI greater than 0.50.

Concurrent validity
Correlation analysis between SHMS V1.0 and FS-14 was shown in Table 5. Large negative correlations were observed between the scores of SHMS V1.0 and FS-14 including total fatigue (r = -0.609, p < 0.001), physical fatigue (r = -0.594, p < 0.001), mental fatigue (r = -0.553, p < 0.001). Physiological sub-health had large negative correlations with total fatigue and physical fatigue, and medium correlation with mental fatigue. Psychological sub-health had large negative correlations with total fatigue, physical fatigue and mental fatigue, while social sub-health showed medium correlations with FS-14.

Discussion
Page 7/23 SHS has become a global public health challenge, but there is no uni ed diagnostic standard for it. Therefore, research on the evaluation method of SHS has attracted the attention of researchers. Many scales for SHS measurement have been developed in China, but some of them are either limited to traditional Chinese medical symptoms, or only focus on physiological indicators, which cannot re ect the connotation of SHS comprehensively. SHMS V1.0 was designed according to the de nition of health from WHO, who de ned health as a state of complete physiological, mental and social well-being and not merely the absence of disease or in rmity [26]. All items of SHMS V1.0 have been selected through Delphi expert consultation and eld investigation [23], which would ensure the authority and practicability of the selected indexes. It can be concluded that SHMS V1.0 has good content validity. Thus, the purpose of this study was to evaluate the reliability and validity of SHMS V1.0 in midwives, in order to con rm whether SHMS V1.0 can accurately and reliably re ect the SHS of midwives for further research.
Reliability refers to the stability and consistency of a tool when it is used to investigate the same object. The present study showed that Cronbach's alpha coe cient for SHMS V1.0 was greater than 0.90, which was considered as excellent internal consistency [34]. The Cronbach's alpha coe cient for three subscale of SHMS V1.0 were all greater than 0.70, showing acceptable internal consistency. The test-retest coe cient for SHMS V1.0, physiological subscale and psychological subscale were all greater than 0.7, indicating acceptable reliability, but that for social subscale was lower than 0.7, maybe duo to small sample size for test-retest reliability. Overall, the overall reliability of SHMS V1.0 is good and it can evaluate SHS of midwives stably.
Validity refers to the matching degree between scale structure and measurement results, so as to evaluate the accuracy of the scale. In the study, the results of project analysis showed that the correlation coe cients between each item of SHMS V1.0 and its corresponding dimension, as well as those between each dimension and its corresponding subscale were all greater than 0.6, indicating high homogeneity for each dimension and subscale; CFA showed that factor loadings in the rst-order factor model and the second-order factor model were greater than 0.4, which indicated that all items and dimensions of SHMS V1.0 were designed reasonably. Furthermore, all t indexes for CFA in the rst-order and modi ed secondorder factor model reached ideal values, showing good tting effect of the models. In the modi ed second-order factor model, there was error correlation between item 20 and 21, consistent with the study of Cordova [37], which reported that spirit were associated with mood (Item 20: Do you feel nervous? Item 21: Do you feel bad or depressed?). There was also error correlation between item 30 and 31 in the modi ed model. Interpersonal relationship may lead to bene ts in subjective well-being and performance [38]. Therefore, there are interacting effects between satisfaction with interpersonal relationships and performance in life, work or study (Item 30: Are you satis ed with your interpersonal relationship in society? Item 31: Are you satis ed with your performance in life, work and study?). Overall, the results illustrated that SHMS V1.0 had good construct validity among midwives.
Concurrent validity is to evaluate the effectiveness of the new tool by testing the correlation between new tool and reference standard. The present study took the FS-14 as a standard, which was used to measure physical and mental fatigue similar to SHS. Pearson correlation analysis showed strongly negative correlations between SHMS V1.0 and FS-14, but SS showed smaller correlations with FS-14 including total fatigue, physical fatigue and mental fatigue. Therefore, it can infer that SHMS V1.0 can more comprehensively re ect SHS of midwives than FS-14, especially for social health symptoms. Regarding PS, bad posture (bending and twisting) and standing for long periods during midwifery process make midwives physically tired [39]. Lack of sleep and biorhythm disorder which may be due to frequent rotating shifts work can also aggravate PS [20]. For MS, it takes much time, energy and emotion for midwives to deal with women's concerns and anxieties during the intrapartum period [40]. At the same time, high tension because of concerning the health of mothers and newborns, worrying about medical disputes, and facing further rescue in midwifery workplaces can easily lead to MS [41]. In regard to SS, Chinese midwives are subordinate to the nursing profession and there is no independent policy and professional system for midwives, which make them lack social recognition. In addition, most midwives are women who have more family responsibility than men, which weaken social support and social adaptation of professional women leading to SS [42]. Therefore, SHMS V1.0 is suitable for measuring the SHS of midwives

Strengths and limitations
The results of this study contribute to the limited research on the measurement of midwives' SHS. And it was the rst and a multi-hospital study, with enough samples to evaluate the psychometric properties of SHMS V1.0 in midwives, which provides an inspiration for measuring SHS of midwives. Meanwhile, there are several limitations to this study. Firstly, the study was only conducted in one province of China, which might limit the generalizability to all Chinese midwives. Second, the races or characteristics of the sample usually affect the reliability and validity of questionnaires in general, so the universal applicability of SHMS V1.0 should be carefully considered.

Conclusion
The internal consistency coe cient and test-retest reliability coe cient showed SHMS V1.0 had acceptable reliability. Project analysis and CFA showed that SHMS V1.0 had good construct validity. The results of concurrent validity analysis indicated that SHMS V1.0 was an available tool to measure SHS of midwives including fatigue. In conclusion, SHMS V1.0 with excellent psychometric properties can accurately and reliably evaluate the SHS of midwives.

Declarations Author contributions
All authors contributed to the study conception and design. Material preparation and data collection were performed by Xiao-Qian Chen, Qing-Xiang Zheng and Yu-Qing Pan, data analysis was performed by Xin-Xin Huang and Gui-Hua Liu. The rst draft of the manuscript was written by Xiao-Qian Chen, Xiu-Min Jiang and Qing-Xiang Zheng commented on previous versions of the manuscript. All authors read and approved the nal manuscript.

Con ict of Interest statement
No con ict of interest has been declared by the authors.

Acknowledgments
We would be grateful to all of the 46 hospitals assisting us in recruiting participants for this study. We acknowledge 46 head nurses for assistance in collecting data. We thank all midwives who participated in the study.

Consent to participate
Informed consent was obtained from all individual participants included in the study.       The second-order factor analysis for SHMS V1.0 The modi ed second-order factor analysis for SHMS V1.0

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.