Development and validation of scales measuring individual, nursing unit, hospital, and community factors related to fertility intentions of female Japanese hospital nurses

Abstract

Background

Scales that comprehensively measure individual, workplace, and community factors related to fertility intentions of women are limited. This study examined the psychometric properties of scales measuring individual (nurse, husband/partner, and child), nursing unit, hospital, and community factors related to fertility intentions of female Japanese hospital nurses. We focused on these nurses because nursing is a gendered profession, and over 90% of nurses in Japan are women. Furthermore, Japanese hospital nurses differ from other female workers in several ways (e.g., professionality, working shifts, financial remuneration).

Methods

This was a methodological study. Data were collected using a cross-sectional, self-administered questionnaire survey. Data from 898 Japanese female hospital nurses (age range, 20–49 years) from 50 hospitals were analyzed. The initial scales were developed through semi-structured interviews, a literature review, an expert review, and a pilot study. Item analysis and exploratory and confirmatory factor analyses were performed, Cronbach’s α was calculated, and known group validity was examined.

Results

The following individual factor scales were obtained: a 12-item, four-domain nurse factor (Economic Needs, Timing of Life Events, Nursing Aspirations, and Career Development) scale, a six-item, two-domain husband/partner factor (Share of Housework and Childcare and Relationship with Husband/Partner) scale, and a three-item, one-domain child factor scale. Likewise, we identified a 13-item, four-domain nursing unit factor (Guilty Feelings toward Colleagues, Unit Nurse Manager’s Management, Workability, and Collegiality) scale, a nine-item, three-domain hospital factor (Access to Legal Rights, Support for Mothers, and Comfort in Hospitals) scale, and a six-item, two-domain community factor (Governmental Family Support and Culture of Working Women) scale. Except for the child factor scale, the comparative fit index was > 0.950, and the root mean square error of approximation was < 0.070. Cronbach’s α ranged from 0.590 (community factor scale) to 0.807 (child factor scale). The scores on the nurse, husband/partner, and hospital factor scales were significantly higher for nurses with high than with low fertility intentions.

Conclusions

These results support the reliability and validity of the nurse, husband/partner, and hospital factor scales. The child, nursing unit, and community factor scales can be further improved.

Background

The percentage of the productive population in Japan is decreasing as a result of declining fertility and increasing life expectancy. As a measure to increase the percentage of the productive population, several policies promoting labor participation by foreigners [1], older adults, and specifically, women [2] have been implemented, as well as policies addressing declining fertility [3].

In Japan, the female labor force participation rate, classified by age group, has been represented by an M-shaped curve that falls in the child-rearing 30s age group. Through policies for balancing work–family responsibilities, the female labor force participation rate in the child-rearing age has improved [4]. However, no such improvements have been seen in the total fertility rate (TFR) [3].

Brzozowska and Mynarska [5] explained that two stages of this process need to be investigated for fertility studies: (a) formation of an intention to have a child, and (b) realization of this intention. The intention to have a child is referred to as fertility intention, the various definitions of which focus on desires, attitudes, or behaviors toward having a child [6].

The choice of whether to have or not to have fertility intentions is an essential part of reproductive rights [7]. Women are more directly affected by fertility than are men, and women should not be coerced into having fertility intentions. For this reason, it is desirable to create an environment in which working women voluntarily have fertility intentions and can combine family and work life.

Bronfenbrenner [8] explained that the nested structure of the environment surrounding a person (e.g., workplace, community) affects that person’s development, which involves a change in roles, such as having a child. Nevertheless, research on women’s fertility intentions has mostly focused on individual as opposed to workplace or community factors. For instance, existing studies have identified numerous factors affecting fertility intention of women, including age, religion, multiple partners, educational status [9], ethnic status, number of siblings, amount of income, perceived acceptable costs of childcare services [10], childbirth readiness, childbirth-related fear [11], partners’ fertility preferences [12], and the division of household chores [13].

Of the limited number of studies on workplace factors, Begall and Mills [14] reported that women are more likely to intend to have a second child when they perceive higher levels of work control. Begall and Mills also found that women with higher levels of job strain had stronger intentions to have a second child if childcare were widely available. In contrast, these women had weaker intentions to have a second child if childcare was not widely available.

Regarding community factors, the community education level has been shown to be associated with fertility intentions [9]. In Japan, at the prefecture level, the share of child welfare expenditures in prefectural finances, the rate of births in which the duration of marriage is shorter than the duration of pregnancy (so-called “shotgun marriages”), the male unemployment rate, and employment rate of women in their 30s were positively related to the TFR [15].

Because research that comprehensively examines the relationships of individual, workplace, and community factors with women’s fertility intentions is limited, we conducted a three-phase study of the fertility intentions of female Japanese hospital nurses. We focused on female Japanese hospital nurses because nursing is a gendered profession. Nursing is often stereotypically viewed as a profession for women [16]. Globally, 65–86% of nurses are women, varying by region [17]. In Japan, over 90% of nurses were women in 2022 [18]. Furthermore, Japanese hospital nurses differ from other female workers in the following ways: they are licensed professionals required to work in rotating shifts, staffed to satisfy a specific nurse-to-patient ratio as dictated by law and the medical reimbursement system; and they are engaged in highly stressful work that affects human lives, has been in shortage, and pays average wages. Economic status is one factor influencing fertility intentions [19], and the mean annual compensation for Japanese nurses in 2019 was US$40,700, which is 1.1 times the mean compensation in Japan [20]. In the first phase of a three-phase study, semi-structured interviews were conducted, and individual, nursing unit, hospital, and community factors related to the fertility intentions of female Japanese hospital nurses were described. In the second phase, four scales measuring the extent to which these factors related to fertility intentions were developed, the variance in fertility intentions among nursing units and hospitals in different communities was examined, and the relationship between individual and nursing unit factors with the fertility intentions of female hospital nurses in Japan was examined [21]. In the third phase, the structures and psychometric properties of the modified scales were examined, and a hierarchical model analysis was conducted to explain the extent to which individual, nursing unit, hospital, and community factors were related to fertility intentions of female hospital nurses in Japan. The structures and psychometric properties of these modified scales are reported.

Methods

Research design

This study was designed as a methodological study using a cross-sectional, nationwide, questionnaire survey.

Participants

The target number of participants was 2,000. DeVellis [22] maintained that a sample of 300 participants was sufficient for scale development. Nevertheless, for the present study, 2,000 participants from 400 nursing units in 80 hospitals for the subsequent hierarchical model analysis were planned. Kreft and Leeuw [23] recommended that at least five observations should be secured for each group at each level for the hierarchical model analysis. Thus, 2,000 participants were recruited from 400 nursing units in 80 hospitals at various prefectures in Japan (80 hospitals × 5 nursing units × 5 nurses = 2000 nurses).

The inclusion criteria were female nurses aged 20–49 years who were working at a hospital in Japan. Female nurses consisted of licensed practical nurses (LPNs), registered nurses (RNs), midwives (MWs), and public health nurses (PHNs). LPNs, MWs, and PHNs were included as participants, as well as RNs, because LPNs can perform the same duties as RNs under the direction of physicians, and MWs and PHNs can perform the same duties as RNs even if they do not have RN licenses under the Act on Public Health Nurses, Midwives, and Nurses of Japan. The age range of the participants was set at 20–49 years, because 99% of babies born in Japan are to mothers within this range [24]. The exclusion criteria were (a) nurses who could not receive the questionnaires at a hospital because of maternity, childcare, or other reasons of absence, and (b) nurses who had worked at their current nursing unit for less than 1 month.

Data collection

Settings

Using stratified random sampling, a total of 800 hospitals with at least 200 beds were invited to participate in the study, with an estimated participation rate of 10%. The strata were created based on the number of beds, hospital owners (e.g., national, prefectural, private), and area (urban vs. rural) from a list of all Japanese hospitals [25] and information on their homepages. Hospitals with at least 200 beds were invited so that five nursing units could be recruited from each hospital. Because the reasons for the decreasing TFR differ between urban and rural areas [26], we divided the hospitals into those in the Tokyo, Nagoya, or Osaka urban conurbation (urban areas) and other areas (rural areas). Of the 800 hospitals, 177 replied to our invitation, of which 53 agreed to participate. Three of these hospitals did not have five nursing units and were therefore excluded. Consequently, a total of 50 hospitals (24 in urban and 26 in rural areas) participated in the study.

Instruments and measures

Scale development comprised the following eight steps: (a) determining what to measure, (b) generating an item pool, (c) determining the format of the scale, (d) having the initial item pool reviewed by experts, (e) considering the inclusion of validation items, (f) administering the items to a development sample, (g) evaluating the items, and (h) optimizing the scale length [22, 27]. We then developed four, six-point, Likert-type scales, adopting (a) and (b) in the first phase and (d) to (h) in the second phase.

First phase: determining what to measure and generating an item pool

We defined Japanese hospital nurses with a positive fertility intention as those who desired to be pregnant and give birth, and were willing to combine their new family role with their current work role. We referred to the ecological model of Bronfenbrenner [8] and a review of the literature on fertility, conducted semi-structured interviews, and conceptualized four factor loci related to fertility intentions of Japanese hospital nurses who were currently working in nursing units or hospitals: individual, nursing unit, hospital, and community factors.

Through the semi-structured interviews with 28 female Japanese hospital nurses in low- and high-TFR prefectures and directed content analysis of the interview transcripts, we identified 71 individual items across six categories (Husband/Partner, Child, Notions about Work–Family Balance, Life Plan, Health Status, and Support from Others), 94 nursing unit items across three categories (Unit Nurse Manager’s Management, Work Environment, and Relationship with Colleagues), 61 hospital items across three categories (Access to Legal Rights, Support System for Mothers, and Retention of Nurses), and 16 community factors across three categories (Family Support Policies, Regional Wage, Women’s Social Position). We added several items based on the literature review and removed and merged other items based on the item contents. At this point, following DeVellis [27], we did not delete redundant items in terms of variability or interest.

Second phase: from determining the scale format to optimizing the scale length

We decided that the scales should be formatted as 6-point, Likert-type scales [27], (from 1 = strongly disagree to 6 = strongly agree) and drafted four. The higher the scores, the better the factors were for fertility intention. Thus, the item score was reversed depending on the item contents. Afterward, a professor in psychology who had been teaching a graduate-level scale development class and four hospital nurse managers reviewed the drafts and provided item-specific advice for relevancy, clarity, and conciseness [27]. Nurse managers were selected because they often possess confidential information related to the fertility of staff nurses. These reviewers also gave advice regarding items not included in the draft [27].

As a result, further items were added, and four scales with 91 individual, 95 nursing unit, 65 hospital, and 16 community items were identified for the pilot study. No items examining social desirability were added to the four scales [27].

Subsequently, we administered the questionnaires, which consisted of the four scales, the items asking about fertility intention, and demographics, to 247 female Japanese hospital nurses aged 20–49 years at two hospitals in Japan. We evaluated the items using item analysis, exploratory factor analysis (EFA), and confirmatory factor analysis (CFA), and reduced the number of items for the third phase.

Because the number of participants who had husbands/partners or children was small, 12 items about husbands/partners and six about children in the individual factors were analyzed separately from nurse factors and remained unchanged from the second phase to the third phase. Similarly, items related to childcare facilities in the hospital factors were excluded from the factor analysis and remained unchanged because the hospitals that participated in the second phase did not have childcare facilities.

Individual factors

Individual factors included nurse, husband/partner, and child factors. For the nurse factors, a 26-item, four-domain (Preconceptions about Working Mothers, Economic Needs, Optimism about Motherhood, and Nursing Aspirations) scale was identified in the second phase and contained in the questionnaire. Interestingly, the item “Women, not men, should do the housework and childcare” was deleted because too many participants endorsed “strongly disagree”, which resulted in a ceiling effect for the reversed score in the second phase. Similarly, for the husband/partner and child factors, the questionnaire contained 12 and six items, respectively. (An additional file provides more detail on the individual, nursing unit, hospital, and community items in the questionnaires used in the third phase [see Additional file 1]).

Nursing unit factors

A 53-item, four-domain scale was included in the questionnaire. The four domains also extracted in the second phase were Unit Nurse Manager’s Management, Tacit Norms, Workload, and Collegiality.

Hospital factors

The questionnaire consisted of a 24-item, three-domain (Support Systems for Mothers, Mother-Friendly Culture, and Retention of Nurses) scale and eight items about childcare facilities.

Community factors

A 14-item, two-domain (Family Support Policies and Social Status of Women) scale was included in the questionnaire.

Fertility intentions

The degree to which nurses desired to be pregnant and give birth while working in their current nursing unit or hospital was measured using two six-point Likert-type items (from 1 = strongly disagree to 6 = strongly agree). Scores of 1–3 and 4–6 were considered to indicate low and high fertility intentions, respectively.

Demographics

The participants were asked about their age, marital status, number of children, basic nursing education, license, length of experience, employment type, and position, among other variables.

Procedures

Each questionnaire was anonymous, but linked with a nursing unit and hospital with an ID number for the subsequent hierarchical model analysis. The contact persons at the hospitals distributed the questionnaires to 10 nurses in each of the five nursing units between April and August of 2019, with an estimated participation rate of 50%. The nurses completed the questionnaire after reading the attached invitation letters, judging that they met the selection criteria, and agreeing to participate in the study. Then, they mailed the completed questionnaires directly to the researchers within 3 weeks of receipt.

Analysis

We examined the distribution, frequency of endorsement, mean, standard deviation (SD), inter-item correlation (IIC), and corrected item-total correlation (ITC) of each item. We then discussed whether to remove an item if it showed (a) a floor or ceiling effect (> 15% of the participants selecting “1” or “6”, respectively) [28], (b) a high IIC (> 0.70) [29], or (c) a low ITC (< 0.20) [30]. Then, we randomly split the sample into two subsamples: A (n = 449) and B (n = 449). Factorability was examined using the Kaiser–Meyer–Olkin test of sampling adequacy (KMO measure) and Bartlett’s test of sphericity (Bartlett’s test). Subsequently, we performed EFA with the maximum likelihood method and promax rotation on subsample A using listwise deletion. The criterion for the factor loading was 0.40. We cross-checked the results of the EFA by CFA on subsample B using full information maximum likelihood estimation. Cronbach’s α and the final ITC for the scale were computed. Welch’s t-test, a robust test for unequal variances, was conducted to examine the known group validity. The scale scores were expected to be higher for nurses with high compared with low fertility intentions. All statistical analyses were conducted using IBM SPSS Statistics 28 and Amos 28 (IBM, Armonk, NY, USA).

Ethical considerations

This study was approved by the Institutional Review Board of the Japanese Red Cross College of Nursing (No. 2018-89). The study aims, significance, and methods (selection criteria of participants, voluntary participation, merits, demerits, and risks of participating in the study, anonymity, and treatment of the data) were explained in the invitation letter attached to the questionnaire. It was also explained to the participants that informed consent was obtained by submitting the completed questionnaires with the invitation letter. Participation in the study was voluntary, and submitting a completed questionnaire was considered to indicate consent.

Results

Participants

Of the 2,500 participants, 1,082 (43.28%) returned a completed questionnaire. A total of 184 responses were excluded from the analysis because they were from (a) male nurses (n = 13) or had an incomplete answer to the question about sex (n = 1), (b) nurses aged > 49 years (n = 10) or had an incomplete answer to the question about age (n = 5), and (c) nurses who had worked at their current nursing unit for less than 1 month (n = 5) or who had an incomplete answer to the question about the length of experience at their current nursing unit (n = 151). Of the 184 responses, one response was excluded because of both (a) and (b). Therefore, a total of 898 (35.92%) responses from 229 nursing units in 50 hospitals were included in the analysis. The response rates for nursing units and hospitals ranged from 0 to 100% and 18–86%, respectively.

The characteristics of the participants are shown in Table 1. The mean age of the participants was 34.26 (SD = 7.45) years. Overall, 377 nurses (41.98%) were single, and 467 (52.00%) were married. About half (n = 454, 50.56%) of the participants did not have any children. Two-thirds of the participants (n = 559, 62.25%) were graduates of a 3-year diploma program to become RNs. All but 10 nurses had an RN license. Regarding these 10 nurses, five had only an LPN license, and the other five had only an MW license. The mean lengths of experience in nursing, at their current hospital, and at their current nursing unit were 11.95 (SD = 7.32), 9.28 (SD = 6.90), and 3.49 (SD = 3.54) years, respectively. The majority of the nurses were permanent workers with full-time work hours (89.20%) and in staff positions (87.75%).

Table 1 Participants’ characteristics

Nurse factor scale

We removed nine (Q21-01, Q21-04, Q21-05, Q21-09, Q21-12, Q21-16, Q21-17, Q21-20, and Q21-21) of the 26 items because of floor or ceiling effects. Two item pairs (Q21-24 and Q21-25, and Q21-25 and Q21-26) had a high IIC, but we retained them because they addressed distinct aspects of income. We excluded one (Q21-18) of two items (Q21-18 and Q21-14) with a low ITC. The content of Q21-14 was love for nursing, which was a repeatedly appearing theme from the interviewees in the first phase. Therefore, we did not remove this item. Through the EFA with the remaining 16 items, we removed three items (Q21-08, Q21-19 and Q21-23) with low factor loadings other than Q21-07. We retained Q21-07 after examining its content. One item (Q21-06) with factor loadings ≥ 0.25 on two domains was excluded. Subsequently, we extracted the 12-item, four-domain structure (Economic Needs, Timing of Life Events, Nursing Aspirations, and Career Development) from the EFA and evaluated the structure using CFA. The results showed the following: χ² = 114.212, df = 48, p≤0.001, comparative fit index (CFI) = 0.956, root mean square error of approximation (RMSEA) = 0.055, and Akaike information criterion (AIC) = 198.212. Cronbach’s α for the entire scale was 0.681.

The scores for Economic Needs were not significantly different between the nurses with low and high fertility intentions while working in the current nursing unit or hospital. Similarly, the scores for Timing of Life Events did not differ between nurses with high and low fertility intentions while working in the current nursing unit. However, these scores were significantly higher for nurses with high than for those with low fertility intentions while working in the current hospital. The scores for the other subscales and the whole scale were significantly higher for nurses with high than for those with low fertility intentions while working in the current nursing unit and hospital (Table 2).

Table 2 Nurse factors

Husband/partner factor scale

Of the 12 items, five (Q21-27, Q21-30, Q21-34. Q21-37, and Q21-38) were removed because of ceiling effects. We did not exclude any items due to a high IIC, but excluded one (Q21-35) of two items (Q21-32 and Q21-35) with a low ITC after examining the contents. A six-item, two-domain structure (Share of Housework and Childcare and Relationship with Husband/Partner) was extracted through the EFA. The results of the CFA showed the following: χ² = 7.657, df = 8, p=0.468, CFI = 1.000, RMSEA = 0.000, and AIC = 45.657. Cronbach’s α for the entire scale was 0.679.

The score of the Share of Housework and Childcare subscale did not differ significantly between nurses with high and low fertility intentions while working in the current nursing unit and hospital. Nevertheless, the scores of the Relationship with Husband/Partner subscale and the whole scale were significantly higher for nurses with high than for those with low fertility intentions while working in the current nursing unit and hospital (Table 3).

Table 3 Husband/partner factors

Child factor scale

From six items, we deleted two (Q21-39 and Q21-44) with floor and ceiling effects. No item was removed due to a high IIC or a low ITC. Next, we performed EFA with the remaining four items. One item (Q21-42) was removed because of a low factor loading. The resultant three-item, one-domain structure was tested using EFA with principal axis factoring for subsamples A and B because neither the maximum likelihood method nor CFA produced a solution for this saturated model. Cronbach’s α for the entire scale was 0.807. The scores for the whole scale did not differ significantly between nurses with low and high fertility intentions while working in the current nursing unit and hospital (Table 4).

Table 4 Child factors

Nursing unit factor scale

Of the 53 items, we removed 12 (Q22-10, Q22-12, Q22-14, Q22-15, Q22-16, Q22-18, Q22-30, Q22-31, Q22-32, Q22-46, Q22-47, and Q22-51) because of floor or ceiling effects. Of these 12 items, four (Q22-14, Q22-15, Q22-16, and Q22-18) were related to demanding workloads and insufficient staffing (17.149–31.292%). Afterwards, we deleted another 12 (Q22-01, Q22-03, Q22-05, Q22-06, Q22-07, Q22-08, Q22-09, Q22-11, Q22-35, Q22-36, Q22-42, and Q22-44) and five (Q22-17, Q22-21, Q22-29, Q22-50, and Q22-52) items because of a high IIC and a low ITC, respectively. A four-domain structure was tested with the remaining 24 items using EFA. Nine items (Q22-19, Q22-24, Q22-26, Q22-27, Q22-33, Q22-39, Q22-48, Q22-49, and Q22-53) with low factor loadings and two items (Q22-28 and Q22-40) with factor loadings ≥ 0.25 on two domains were excluded. Then, a 13-item, four-domain structure (Guilty Feelings toward Colleagues, Unit Nurse Manager’s Management, Workability, and Collegiality) was extracted from the EFA. The results of the CFA showed the following: χ² = 121.365, df = 59, p<0.001, CFI = 0.966, RMSEA = 0.049, and AIC = 211.365. Cronbach’s α for the entire scale was 0.802.

The scores for Guilty Feelings toward Colleagues was significantly lower for nurses with high than for those with low fertility intentions while working in the current nursing unit and hospital. The scores for Unit Nurse Manager’s Management and Workability were significantly higher for nurses with high than for those with low fertility intentions while working in the current nursing unit and hospital. The scores for Collegiality and the whole scale were higher for nurses with high than for those with low fertility intentions while working in the current nursing unit and hospital; however, these differences were not significant (Table 5).

Table 5 Nursing unit factors

Hospital factor scale

Eight items (from Q23-08 to Q23-15) related to the utility of childcare facilities in hospitals were deleted from the 32 items because approximately one-fifth of the participating hospitals did not have childcare facilities. Likewise, 13 items (Q23-01, Q23-03, Q23-07, Q23-16, Q23-17, Q23-18, Q23-19, Q23-20, Q23-23, Q23-25, Q23-28, Q23-31, and Q23-32) were removed because of floor or ceiling effects. In addition, one (Q23-06) of two (Q23-05 and Q23-06) highly correlated items was removed because of redundant content. Q23-04 and Q23-05 were highly correlated (r = .773); however, we retained them because their contents tapped into distinct aspects of labor laws. Subsequently, another item (Q23-27) was eliminated because of a low ITC. After the EFA was conducted with the remaining nine items, we identified a nine-item, three-domain structure (Access to Legal Rights, Support for Mothers, and Comfort in Hospitals). The factor loading for Q23-29 and Q23-30 was < 0.40, but we kept this item because of the stability of the domain and item content. The CFA showed the following: χ² = 30.388, df = 24, p=0.172, CFI = 0.992, RMSEA = 0.024, and AIC = 90.388. Cronbach’s α for the nine items was 0.694.

The scores for Access to Legal Rights were higher for nurses with high than for those with low fertility intentions while working at the current unit or hospital, but the results of Welch’s t-tests were not significant. The scores for Support for Mothers did not differ significantly between the nurses with high and low fertility intentions while working in the current nursing unit. In contrast, the same scores were significantly higher for nurses with high than for those with low fertility intentions while working in the current hospital. The scores for Comfort in Hospitals and the whole scale were significantly higher for nurses with high than for those with low fertility intentions while working at the current unit or hospital (Table 6).

Table 6 Hospital factors

Community factor scale

For five items asking about the utility of the childcare facilities in the community, the responses of “9” (unavailability of childcare facilities) were changed to “1” (strongly disagree). Approximately 90% of those who answered “9” on these five items were participants without children. These five items showed floor effects and were therefore removed. No items showed a high IIC or a low ITC. Therefore, EFA was performed on the remaining nine items. Two items (Q24-11 and Q24-12) with factor loadings < 0.40 were removed. Subsequently, a seven-item, two-domain structure (Governmental Family Support and Culture of Working Women) was evaluated using CFA. The results showed the following: χ² = 37.984, df = 13, p<0.001, CFI = 0.916, RMSEA = 0.065, and AIC = 81.984. Examining the seven items, we removed Q24-08 because it could be interpreted as a question about knowledge of local salary scales. The results of the CFA of the six-item, two-domain structure showed the following: χ² = 18.455, df = 8, p = 0.018, CFI = 0.954, RMSEA = 0.054, and AIC = 56.455. Cronbach’s α for the entire scale was 0.590.

The scores for Governmental Family Support and Culture of Working Women did not differ between nurses with high and low fertility intentions while working at the current unit or hospital. Likewise, the scores for the whole scale did not differ between the two groups while working at the current unit or hospital (Table 7).

Table 7 Community factors

Discussion

Participants

In this study, we recruited female Japanese hospital nurses aged 20–49 years using stratified random sampling. The responses of 898 nurses from 229 nursing units in 50 hospitals were analyzed. DeVellis [22] maintained that a sample of 300 participants was sufficient for scale development, and the present study satisfied this criterion.

Development process, content validity, and structural validity

The development process was appropriate, and content and structural validity were generally supported for the scales. We developed the present scales based on the steps explained by DeVellis [22, 27]. The scale structures, except for that of the child scale, were cross-checked with EFA and CFA in the third phase. The factor analyses were considered proper because the KMO measures were > 0.50 [31] and Bartlett’s test showed a significant difference. Overall, the fit indices of CFA showed acceptable model fit for the five scales. The results of the χ² test were significant for the nurse, nursing unit, and community factor scales; however, significant results had been expected for the χ² tests of the large sample. The CFI was > 0.950 and satisfied the criterion proposed by Kline [32] for the five scales. Likewise, the RMSEA ranged from 0.000 (husband/partner factor scale) to 0.055 (nurse factor scale) and satisfied the criterion of < 0.070 recommended by Steiger [33]. Therefore, we interpreted the contents and structures of the scales as valid, except for the child factor scale.

Interpretability of domains

Nurse factor

Economic Needs, Timing of Life Events, Nursing Aspirations, and Career Development were identified in the third phase. The items in Economic Needs concerned reduced income due to not working night shifts or overtime or using the childcare short-hour system. In 2019, the average ratio of family benefits to GDP was 1.95% in Japan, compared with 2.29% among Organisation for Economic Co-operation and Development (OECD) countries [34]. Since governmental family support is not dependable, economic status is a factor that affects fertility intentions of Japanese people [19]. Although the male breadwinner model is prevalent in Japan [35], some female nurses may have played the role of breadwinner because they have a decent income. Therefore, Economic Needs was extracted as a factor related to fertility intentions.

Timing of Life Events consisted of three items related to the optimal timing for marriage, pregnancy, and childbirth in the workplace. Three unique Japanese values regarding fertility included the prevention of risk in future life planning, the avoidance of criticism from others, and pressure for childcare that children should not have a difficult time [36]. The “3-year-old myth”, which is a tacit norm that mothers should commit to rearing children while they are small, exists in Japan. Ohinata [37] explained that women who want to continue their careers become reluctant to have children as a result of internalizing the “3-year-old myth”. Timing of Life Events suggested that female hospital nurses do not consider marriage, pregnancy, and childbirth “possible when desired”; rather, they may consider these life events “possible when conditions are met” from the viewpoints of life planning, evaluation from others, and childcare.

The items in Nursing Aspirations related to the participants’ affective commitment to the nursing profession. Kleier et al. [38] reported that attachment to a profession, that is, affective commitment, was negatively related to intent to leave. As a result, Nursing Aspirations was extracted as a factor domain related to fertility intention without quitting the current nursing unit or hospital.

Career development consisted of three items related to the dilemma between the desire for motherhood and the expectations of the hospital management team for careers. Gauci et al. [39] maintained that the career advancement of female nurses is slower than that of male nurses because of interruptions due to pregnancy, childbirth, and childcare. In other words, a motherhood penalty exists for female nurses. Therefore, Career Development was extracted as a factor domain related to fertility intentions of the hospital nurses who wanted to continue their careers.

Husband/partner factor

We extracted two domains, Share of Housework and Childcare and Relationship with Husband/Partner. The items in Share of Housework and Childcare concerned the division of housework and childcare with a husband/partner and the ability of the husband/partner to perform housework and childcare. In other words, these items indicated whether female Japanese hospital nurses could delegate housework and childcare to husbands or partners without concern for their performance. In Japan, women spend 15.3% of their time during a 24-hour period on unpaid work for domestic activities such as cleaning, washing, and repair work, whereas men spend only 4.7% [40]. Likewise, women with one and two children spend 11.7% and 16.6% of their time providing care, respectively. In contrast, men spend just 2.5% and 4.1% [40]. Riederer et al. [13] reported that short-term fertility intentions of women in a couple were affected by the division of housework. This effect was influenced by women’s satisfaction with the division, relationship with husband/partner, and parity. Female Japanese hospital nurses are considered to have egalitarian views, as the item “Women, not men, should do the housework and childcare” (the reversed scoring item) was deleted because of a ceiling effect in the second phase. As a result, the division of housework and childcare with husband/partners was obtained as a factor domain related to the fertility intentions of female Japanese hospital nurses.

The items in Relationship with Husband/Partner involved whether the nurses and their husband/partner had a good relationship and frequently discussed work–family balance, and whether the husband/partner desired children. Berninger et al. [41] showed that, among German women, partnership quality was associated with the intention to have a first child. Besides, Araban et al. [42] reported that, among Iranian women, marital satisfaction scores were significantly higher for those who desired to bear children than for those who did not. Thus, in this study, Relationship with Husband/Partner was extracted as a domain in the husband/partner factor related to fertility intentions.

Child factor

The items in the child factor were related to the physical and mental health of the children and their willingness to receive nonparental group care. These items indicated whether mothers could work while absent from their children.

Under Japanese law, employees can claim sick/injured child leave up to 5 days per year for each child. However, some hospital nurses may have difficulty making such claims because many hospital nurses work rotating shifts, and the schedules and staffing of nursing units are determined at least one month beforehand. Furthermore, staffing is set to satisfy the specific nurse-to-patient ratio dictated by law and the medical reimbursement system. This nurse-to-patient ratio is interpreted as an upper rather than a lower limit for staffing in most nursing units. Therefore, no staffing buffer is set, and if a nurse takes sick/injured child leave, her colleagues must return from their days off to cover the shift. If nurses in motherhood continue to take leave, they will also continue to owe their colleagues, which makes it difficult for them to stay at work. Thus, the health of the child is important for the continuation of their current work. If nurses do not think they can continue working after having a child, they will not desire another child while working in the current nursing unit or hospital.

In addition, the availability of nonparental group care is important for mothers who work. Petts et al. [43] reported that the loss of full-time childcare was associated with an increased risk of unemployment for mothers of young children in the US during the COVID-19 pandemic. Both the transition from home to group care and the separation from their mother are stressful for young children [44]. Even when nonparental group care is available, mothers find it difficult to go to work if their children refuse to receive group care because of stress. In such a situation, they will not desire another child while working in the current nursing unit or hospital. The items in the child factor suggest that the situation of the existing child relates to the fertility intentions for the next child in Japanese hospital nurses.

Nursing unit factor

Four domains, Guilty Feelings toward Colleagues, Unit Nurse Manager’s Management, Workability, and Collegiality, were extracted for this factor. The items in Guilty Feelings toward Colleagues showed the attitudes that nurses who are pregnant, give birth, or provide childcare are a second-rate labor force in nursing units. This treatment occurs as a result of the internalization of the tacit norm that “all workers should work like healthy male workers who will never be pregnant, give birth, or provide childcare”.

As mentioned above, the male breadwinner model is prevalent in Japan, and both male and female workers who work like healthy male workers are considered a first-rate labor force, whereas female workers in motherhood are considered a second-rate labor force. For this reason, female workers in motherhood are typically assigned to low-paid, irregular work with no job security and treated as a disposable labor force, whereas male workers tend to secure breadwinner work [35]. Hierarchical relations of power between nurses in or out of motherhood may be produced because of the notion that female workers in motherhood are a second-rate labor force.

Besides, as described previously, the nurse-to-patient ratio is restricted by law and the medical reimbursement system for hospitals. Thus, systems have been established in which the nurses’ colleagues have to compensate through additional shifts or hours if nurses in motherhood cannot provide the expected duties. Nurses internalize the tacit norm and consider that nurses who do not work rotating shifts or overtime because of motherhood owe their colleagues, and perceive them as a second-rate labor force. As a result, nurses regard that nurses in motherhood feel and express remorse toward their colleagues in the first-rate labor force.

The items in Unit Nurse Manager’s Management were associated with managerial support from unit nurse managers regarding pregnancy, childbirth, and childcare. Ooshige et al. [45] described that support from nurse managers and colleagues is required for the continuation of pregnancy among nurses. Unit nurse managers are usually responsible for work assignments, staffing, and scheduling. As a result, their management directly affects workability for nurses in motherhood.

The three items in Workability related to working hours and nursing care delivery systems. Chung [46] maintained that flexible work schedules expedite the integration of work and family roles; however, access to a flexible work schedule was limited in female-dominated workplaces. At Japanese hospitals, nurses’ work schedules are set by organizations with little possibility of change. Nursing care delivery systems influence the continuation of nursing care specifically when nurses in motherhood work irregular shifts, such as shorter childcare hours. If the continuity of nursing care is breached, and quality care is not secured, nurses in motherhood are considered persona non grata. No nurse wants to be persona non grata.

Abe et al. [21] reported that the average perception of workloads among hospital nurses in the nursing unit was related to fertility intentions. However, too many nurses considered that workloads were high, and four items related to workloads were deleted because of floor effects in the third phase. Thus, workload per se was not identified as a domain.

Collegiality was about the attitudes of their colleagues toward motherhood, whereas Guilty Feelings toward Colleagues was about the attitudes of nurses in motherhood. Colleagues’ attitudes are affected by whether they perceive nurses in motherhood as reliable workers. Career breaks of female nurses due to motherhood are regarded to undermine their skills and competency [47]. In addition, colleagues considered that nurses in motherhood were not committed to nursing [48]. In contrast, Ooshige et al. [45] explained that pregnant nurses reported development as nurses through motherhood. Likewise, some nurses may perceive nurses in motherhood as contributing toward quality care. Colleagues may evaluate nurses with children as skilled dependable workers who commit to nursing or as unskilled undependable workers who do not, and this evaluation may affect fertility intentions of nurses while working in the current nursing unit or hospital.

Guilty Feelings toward Colleagues, Unit Nurse Manager’s Management, Workability, and Collegiality were identified as nursing unit factors. All of these domains concerned whether the work environment of the nursing units was mother-friendly.

Hospital factors

Three domains, Access to Legal Rights, Support for Mothers, and Comfort in Hospitals were identified. Japanese laws prohibit discrimination based on pregnancy and childcare. Besides, some laws regarding maternity and childcare leave, time off for sick or injured children, shorter working hours, and the limitation of night work exist. Nevertheless, maternity harassment that includes denied access to legal rights frequently occurs in Japan [49]. Human resources management, such as employment, dismissal, financial remuneration, and working hours or shift patterns, is usually processed at the organizational level in Japanese hospitals. For that reason, denied access to legal rights is likely to occur at the organizational level.

The three items in Support for Mothers were related to support to continue working for nurses in motherhood at the organizational level. For women, decreased perceptions of career encouragement at work are indirectly related to increased turnover intention and decreased career motivation throughout pregnancy [50]. Likewise, Thompson et al. [51] showed that a supportive work–family culture was positively associated with affective commitment and negatively associated with intention to leave the organization.

The items in Comfort in Hospitals were related to wages, aggressive culture, and turnover. In this study, fertility intention was that while working in the current nursing unit or hospital. Thus, it included the desire to continue working in the current nursing unit or hospital, as well as the desire to have a child. Financial remuneration and the organizational work environment, such as violence at work, influence the retention of hospital nurses [52].

Consequently, three domains, Access to Legal Rights, Support for Mothers, and Comfort in Hospitals, were obtained for the hospital factors. These domains were associated with whether nurses in motherhood could work comfortably without facing discriminatory treatment at the organizational level.

Community factors

Two domains, Governmental Family Support and Culture of Working Women, were extracted for the community factor. The items in Governmental Family Support were about family support and after-school care from local governments. Policies that reduce the conflict between work and motherhood have been reported to be effective for the first birth at a younger age [53]. In Japan, local government support generally includes monetary support and direct childcare support such as group care, networking among parents, and short-term stays. The share of child welfare expenditures in prefectural finances is positively related to the TFR [15].

The four items in Culture of Working Women indicated whether working mothers were taken for granted in their communities. In general, strong normative expectations of two-parent families characterized by traditional gender roles are prevalent in Japan [54]. At the same time, the shotgun marriage rate, male unemployment rate, and employment rate of women in their 30s were positively related to the TFR [15]. Thus, fertility intentions may be high in communities with nonnormative expectations of nontraditional gender roles.

The two domains in the community factors were Governmental Family Support and Culture of Working Women. These domains were associated with whether nurses in motherhood can rear children in the community without difficulty.

Scale reliability

The internal consistency and reliability of the scales were acceptable, except for the community factor scale. Cronbach’s α for the nursing unit (0.802) and child (0.807) factor scales ranged between 0.80 and 0.90 (“very good”) [29]. Those of the nurse (0.681), husband/partner (0.679), and hospital (0.694) factor scales ranged between 0.65 and 0.70 (“minimally acceptable”). Nevertheless, Cronbach’s α for community factor was < 0.60 (0.590) and “unacceptable”. Cronbach’s α is increased when well-correlated items with other items in a scale are added [55]. For the community factor scale, the five items related to childcare facilities were correlated with each other. Nonetheless, all five items were deleted because the conversion of the responses from “9” to “1” caused a floor effect. Most of the participants who responded “9” did not have a child. Accordingly, it was suspected that these participants casually responded “9” in lieu of “I don’t know”. Q24-07, the item related to after-school care, did not have an alternative for “9” and thus did not cause a floor effect and was retained. Therefore, Cronbach’s α might have increased if an alternative for “9” had not been set and at least one item related to childcare facilities had been retained.

Construct validity

Construct validity was tested using known-groups validity [56]. For the nurse, husband/partner, and hospital factor scales, the total scores were significantly higher for nurses with high than for those with low fertility intentions, as was expected.

For the child factor scale, the scale scores were not significantly higher for nurses with high than for those with low fertility intentions. In Japan, the average number of actual children per couple is 1.90, and the ideal average number of children is 2.25 [19]. Therefore, nurses with one or more children may not intend to have another child, regardless of the situation of their child. The relationship between the child’s status and fertility intentions needs to be examined more closely.

Equally, the scores for the nursing unit factor were not significantly higher for nurses with high than for those with low fertility intentions. Particularly, the score of Guilty Feelings toward Colleagues was significantly lower for nurses with high than for those with low fertility intentions. Consequently, the score for Guilty Feelings toward Colleagues pulled down the total score. There was a possibility that participants may have interpreted the items within Guilty Feelings toward Colleagues as questions relating to individual perceptions, rather than an assessment of the work environment. Alternatively, nurses with high fertility intentions might evaluate the work environment more critically. In other words, these items were not the antecedents, but the consequences of fertility intentions. Because the items within Guilty Feelings toward Colleagues were reversely scored, withholding the reverse-scoring may solve this problem. Changing the item wordings, removing the domain, or withholding reverse-scoring may improve the scale.

The scores for the community factor scale were not significantly higher for nurses with high than for those with low fertility intentions. The relationships between the community factor and fertility intentions may be influenced by the number of children. Nurses without a child might not have an interest in community childcare resources, as indicated by the responses about the utility of the childcare facilities in the community.

In summary, construct validity was supported for the nurse, husband/partner, and hospital factor scales. However, the child, nursing unit, and community factor scales can be improved further.

Limitations

This study has some limitations. First, the assumptions of independently and identically distributed random variables may be violated because the participants were clustered within nursing units and hospitals. Nevertheless, the recruitment methods used for this study were the best and most feasible way to reflect the population because no nationwide roster of hospital nurses exists in Japan. Second, neither test–retest reliability nor cross-cultural validity were assessed in this study. Several additional studies may be needed to enhance reliability and validity.

Conclusion

In the present study, we developed scales to measure individual (nurse, husband/partner, and child), nursing unit, hospital, and community factors related to fertility intentions of female Japanese hospital nurses. The results confirmed the reliability and validity of the nurse, husband/partner, and hospital scales. However, the child, nursing unit, and community factor scales may be improved further in future research.