Psychometric validation of the Internalised Stigma Scale for Gestational Diabetes Mellitus (ISS-GDM): a cross-sectional study

Abstract

Objectives

To develop and validate the Internalised Stigma Scale for Gestational Diabetes Mellitus (ISS-GDM), a questionnaire measuring self-reported internalised stigma among women with prior gestational diabetes mellitus (GDM). We hypothesised that internalised GDM stigma could be reliably and validly assessed through a short psychometric instrument.

Design

Cross-sectional validation study.

Setting

Follow-up data from the Danish, multicentre Face-it trial for women with prior GDM and their families.

Participants

In total, 248 women completed the ISS-GDM approximately 1 year after their GDM affected pregnancy.

Primary and secondary outcome measures

The primary outcome was psychometric properties of the ISS-GDM, assessed using Cronbach’s alpha, confirmatory factor analysis (CFA) and Rasch analysis (RA). Secondary outcomes included identification of item anomalies (local response dependence, differential item functioning).

Results

A large proportion of respondents endorsed statements reflecting self-disappointment, self-blame and an altered self-perception. Less endorsed statements included feeling inferior to other mothers or guilt towards family members due to GDM. The ISS-GDM demonstrated satisfactory psychometric properties. CFA indicated that item 2 assessing self-perceived capabilities as a mother did not load onto the main factor, while CFA and RA identified local response dependence and differential item functioning by body mass index. After adjustments, a two-factor solution supported calculating a sum score of items 1 and 3–11, with item 2 retained as a stand-alone indicator of perceived parenting capabilities. The 10-item scale demonstrated acceptable reliability (Cronbach’s alpha=0.78).

Conclusions

The ISS-GDM is a reliable and valid tool for assessing internalised stigma among women with prior GDM. Our findings further suggest that a substantial proportion of women with prior GDM experience self-blame and an altered self-perception due to their diagnosis. The ISS-GDM scale enables research into its prevalence, severity and consequences.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study is the first to develop and validate a questionnaire designed to quantify gestational diabetes mellitus (GDM)-specific internalised stigma, addressing a significant gap in the field.

• Development was grounded in stigma theory and international literature, and the Internalised Stigma Scale for Gestational Diabetes Mellitus (ISS-GDM) underwent thorough psychometric testing (confirmatory factor analysis and Rasch analysis), demonstrating validity and reliability.

• The item pool was kept small to minimise respondent burden; test-retest reliability was not assessed for the same reason.

• As validation was conducted among Danish women with prior GDM, additional research is required to assess the scale’s relevance during pregnancy and across cultures.

Introduction

Gestational diabetes mellitus (GDM), defined as glucose intolerance with onset or first recognition during pregnancy, annually affects around 17 million deliveries worldwide.^{1 2} GDM increases the risk of adverse pregnancy outcomes, such as pre-eclampsia, foetal overgrowth and delivery complications.³ Although usually transient, GDM is also known to increase the long-term risk of developing type 2 diabetes, cardiovascular diseases and obesity for both the woman affected by GDM and her offspring.4,6 Women with GDM are also at an increased risk of adverse mental health outcomes, such as postpartum depression.⁷

Furthermore, there is a growing body of research suggesting that GDM is also associated with both enacted and internalised stigma, similar to type 1 and type 2 diabetes.^{8 9} In our study, stigma is defined according to Link and Phelan¹⁰ as a coexistence of four interrelated components in an asymmetrical power dynamic, namely: (1) providing a group of people with a common label, (2) associating the label with negative stereotypes, (3) perceiving the labelled group as separate from the norm, thus resulting in (4) status loss and discrimination. Internalised stigma, or self-stigma, is defined as awareness, agreement and application of the negative stereotypes associated with one’s label and subsequent psychological self-harm.¹¹

Stigma related to GDM is an emerging area of research. A recent scoping review identified evidence of GDM stigma across 19 countries, spanning five continents, highlighting both interpersonal and internalised experiences.⁸ Qualitative studies from across the globe have shown how women may encounter prejudicial remarks from healthcare professionals and relatives, while others internalise stigma, leading to feelings of guilt, shame and threats to their sense of identity.^{9 12 13} Such experiences have been linked to non-disclosure of the diagnosis, reluctance to engage in screening or treatment and poorer psychological well-being.^{9 12 13} In a Danish study, experiences of prejudice from healthcare personnel or the public appeared to be less pronounced; yet, the findings indicated that GDM posed a threat to the women’s self-perceived identity and challenged notions of what it entails to be a good parent.¹³ The identified underlying victim-blaming narratives suggest that women diagnosed with GDM may internalise the stigma associated with GDM.¹³

Building on current evidence, we hypothesise that internalised stigma may affect the well-being and health of women with GDM. While recent evidence has documented GDM stigma, particularly internalised stigma, and its implications across diverse settings, a major limitation remains the absence of a validated instrument to quantify this experience. A recent global consensus to end diabetes stigma explicitly calls for future research to address GDM stigma and valid measures to document the occurrence and harms of GDM stigma and evaluate stigma-reducing interventions.¹⁴ Responding to this call, this study aimed to describe the development and validation of a novel questionnaire measuring self-reported internalised GDM stigma in women with prior GDM in Denmark: The Internalised Stigma Scale for Gestational Diabetes Mellitus (ISS-GDM).

Research design and methods

Study design

This is a cross-sectional validation study of the psychometric properties of the ISS-GDM questionnaire nested in the Face-it randomised controlled trial (ClinicalTrials.gov: NCT03997773). The Face-it trial evaluated a health promotion intervention aiming to reduce the risk of developing type 2 diabetes and increase the quality of life among women with a prior GDM-affected pregnancy and their families.¹⁵ The Face-it study design and results have been reported in detail elsewhere.^{15 16}

Participants

The participants for this study were women with prior GDM participating in the Face-it trial who responded to the follow-up questionnaire administered 12 months after delivery. GDM was defined according to the Danish diagnostic criteria of a 2-hour 75 g oral glucose tolerance test measurement of ≥9.0 mmol/L following a risk-factor-based screening process.¹⁷ In total, 277 women were randomised (2:1) to receive the Face-it intervention or usual care, of which 248 women answered the ISS-GDM questionnaire 12 months after giving birth. At the time of data collection, 18 women were pregnant again. The recruitment strategy, inclusion and exclusion criteria for participation and baseline characteristics of the participants 10–14 weeks postpartum have been described elsewhere.¹⁸

Patient and public involvement

Women with prior GDM were involved in the development phase of this study. Two women with prior GDM participated in cognitive debriefing interviews (as described below) to provide feedback on the wording, clarity and acceptability of questionnaire items and response categories. Their inputs were instrumental in ensuring the face and content validity of the questionnaire. Women with GDM were not involved in setting the research questions, study design, recruitment or conduct of the study, nor were they asked to advise on interpretation or dissemination of the results. Participants were not compensated for their involvement.

Data

The data for this study are based on obstetric data collected from the participants’ medical birth records, self-reported questionnaire data and anthropometric measurements collected 10–14 weeks postpartum (baseline) and 12 months after delivery (follow-up).¹⁵ Data collection commenced in 2019 and was completed in 2023.¹⁵ The data were managed using REDCap electronic data tools hosted by the Capital Region of Denmark.^{19 20}

Internalised GDM-specific stigma

Internalised GDM-specific stigma was measured using the ISS-GDM. The ISS-GDM was developed for this study in 2019 by combining stigma theory,¹¹ items from existing questionnaires measuring internalised stigma related to mental health and type 2 diabetes^{21 22} and findings from qualitative research investigating the lived experiences of women with GDM.23,26 The scale was developed to assess internalised GDM stigma among women with prior GDM, rather than during pregnancy, as it was designed for participants in the Face-it trial, which intervened after childbirth.

An initial pool of items was generated by adapting relevant items from existing questionnaires and creating new items to reflect themes specific to GDM identified in prior qualitative studies. Items addressing similar phenomena were combined to cover relevant aspects of internalised stigma while avoiding redundant items and limiting response fatigue. This process resulted in a draft pool of 11 items.

Before data collection, the content validity of the scale was assessed by conferring with four experts with backgrounds in scale development, stigma and GDM.²⁷ Feedback focused on semantics, clarity and item ordering, and minor modifications were made accordingly. Face validity was assessed through cognitive debriefing interviews with two women with prior GDM.²⁷ Two interviews were conducted to ensure feasibility while adhering to the Face-it timeline for data collection. The women reported that the items were clear and relevant and suggested minor wording changes, which were implemented. They also recommended adding a neutral response option for items 1–10; ‘I neither agree nor disagree’”.

The final scale comprised 11 items: 10 items scored on a 5-point Likert scale ranging from ‘Strongly disagree’ (0 points) to ‘Strongly agree’ (4 points), with items 2 and 10 being reverse-scored due to positive wording (online supplemental appendix A), and one disclosure item (item 11) with three response options: ‘Yes, to people in general’ (0 points), ‘Yes, but only to a few’ (3 points) and ‘No’ (4 points). The ISS-GDM questionnaire was developed and administered in Danish (see online supplemental appendix A). For publication, the questionnaire was translated into English; however, it did not undergo back-translation (table 1).

Table 1. Characteristics of respondents.

Demographic and clinical characteristics (n=248*)	Median (Q1;Q3) or n (%)
Age (years)†	33.4 (30.0;37.0)
Place of birth‡
Born in Denmark	195 (78.9)
Born outside of Denmark	52 (21.1)
Educational status‡
Low (0–10 years)	44 (17.8)
Medium (11–14 years)	121 (49.0)
High (>15 years)	82 (33.2)
BMI pre-pregnancy (kg/m2)§	26.5 (23.0;30.4)
BMI 12 months postpartum (kg/m2) †	26.9 (23.0;31.5)
Family history of diabetes (T2DM, T1DM or GDM)‡ (yes)	169 (68.4)
Number of GDM affected pregnancies‡
1	216 (87.1)
≥2	32 (12.9)
Insulin during most recent GDM affected pregnancy§ (yes)	54 (21.8)
Pregnant 12 months post-partum† (yes)	18 (13.7)
Randomisation status‡ (intervention)	164 (66.1)
Mental health 12 months after giving birth
Anxiety symptoms: GAD-7† (≥ 10)	30 (12.3)
Well-being: WHO-5† (≤50)	82 (33.7)
Stress: PSS†
Low perceived stress (<14)	110 (44.9)
Moderate perceived stress (14-26)	128 (52.2)
High perceived stress (>26)	7 (2.9)

^*Frequencies do not always add to total n=248 due to missing data on some items (place of birth: 1, educational status: 1, PSS: 3).

^†Data collected 12 months after delivery

^‡Data collected at 10–14 weeks postpartum

^§Data collected from medical birth record.

GAD-7, generalised anxiety disorder; GDM, gestational diabetes mellitus; PSS, perceived stress scale; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus.

Exogenous variables

Included exogenous variables encompassed data from the women’s medical birth records (pre-pregnancy body mass index (BMI) (kg/m²), insulin use during pregnancy); trial randomisation status; self-reported medical history and sociodemographic data collected at baseline (number of pregnancies with GDM, family history of diabetes, the woman’s date and place of birth, educational status); BMI (kg/m²) collected at 12-month follow-up; and self-reported measures of mental health, also collected at follow-up (perceived stress scale (PSS),²⁸ generalised anxiety disorder (GAD-7)²⁹ and well-being (WHO-5)³⁰ (table 1). The mental health measures were scored according to published guidelines. The continuous sum scores were used to assess the criterion validity of ISS-GDM. The categorical scoring was used to present the background characteristics of respondents in table 1: GAD-7 (<10: low degree of anxiety, ≥10: high degree of anxiety), WHO-5 (>50: high degree of well-being, ≤ 50: low degree of well-being) and PSS (<14: low perceived stress, 14–26: moderate perceived stress, >26: high perceived stress).

Data analysis

All analyses were performed using R statistical software (V.4.3.0), specifically the packages eRm, iarm and lordiff.31,33

Descriptive statistics were performed on all variables, assessing data quality by the number of missing values and data distribution. Missing data were handled using pairwise deletion, including only cases with complete data for the explored variables. The floor-and-ceiling effect was assessed for each item in the ISS-GDM questionnaire and considered if more than 15% of respondents scored the lowest or highest possible score.³⁴

Psychometric validation methods

The construct and content validity of the ISS-GDM questionnaire were evaluated using confirmatory factor analysis (CFA) and Rasch analysis (RA) for items with ordered response categories.³⁵ CFA aims to test the assumption that all items measure the same latent trait. RA seeks to test whether it is feasible to report all 11 items in the ISS-GDM questionnaire as a sum score without losing or distorting information.³⁵

The underlying assumptions tested in CFA and RA are (1) unidimensionality, (2) monotonicity, (3) homogeneity, (4) local independence, (5) consistency in item correlations and (6) absence of differential item functioning (DIF). Beyond these, the reliability of the questionnaire should also be evaluated.

Unidimensionality was assessed by evaluating the fit of data to a unidimensional CFA model, hypothesising that all items in the questionnaire measure the same latent variable, that is, internalised GDM stigma. Fit statistics for evaluating the model fit were χ² (p≤0.01 considered to indicate multidimensionality), the Comparative Fit Index (CFI), the Tucker Lewis Index (TLI)—which should both ideally be >0.95³⁶ – and the Root mean square of approximation (RMSEA) which should be <0.06.³⁷ Additionally, modification indices (MIs) were also assessed, providing an initial investigation into local response dependence (LRD).³⁸ Consistency was assessed by examining the inter-item correlation matrix, ensuring positive correlations between the items in the questionnaire.

Monotonicity was investigated by assessing the item fit statistics resulting from RA, specifically the infit and outfit statistics. The expected value of these is one, with values close to zero or much greater than one, regarded as evidence of an overfit or underfit of data.³⁵ Homogeneity was investigated by assessing the item response curves and evaluating whether the response categories were ordered for each item.³⁵ LRD was evaluated by looking at the correlation of the residuals and the fit statistic Q3. When the difference between the largest Q3 value and the average residual correlation was larger than 0.20, this was considered evidence of LRD.

DIF was tested according to the following six variables: randomisation status (intervention/usual care group), place of birth (Denmark/Outside of Denmark), insulin use during pregnancy (Yes/No), family history of diabetes (Yes/No), pre-pregnancy BMI (<27 kg/m²/≥27 kg/m²) and BMI 12 months after delivery (<30 kg/m²/≥30 kg/m²). The variables were selected based on previous research indicating that these factors may affect how women with GDM perceive their diagnosis and are affected by GDM stigma.^{8 13} Additionally, pre-pregnancy BMI of ≥27 kg/m² and family history of diabetes were included in the risk-based screening for GDM in Denmark at the time of recruitment¹⁷ and hypothesised to potentially affect the women’s perceptions of causality and thereby possibly enhance feelings of guilt or self-blame. DIF was assessed by running individual logistic ordinal regression analyses for the exogenous variables with significant chi-square values (p≤0.01) considered evidence of DIF.³¹ The magnitude of DIF was evaluated using stratified conditional item characteristic curves.³⁹

Criterion validity was assessed by analysing concurrent validity. We expected a small to moderate negative correlation between ISS-GDM and well-being (WHO-5) and a positive correlation of similar magnitude between ISS-GDM and stress (PSS) and anxiety (GAD-7),⁴⁰ as internalised stigma is conceptually related yet distinct from these constructs.¹¹ The scales were all kept as continuous variables for these analyses.

A Cronbach’s alpha analysis was also conducted, providing a measure of the questionnaire’s reliability. A value between 0.7 and 0.9 was considered acceptable.

The ISS-GDM questionnaire was validated through an iterative process, where all identified anomalies were collectively assessed considering the theory and evidence that originally informed the development of the questionnaire. First, a CFA for ordinal items and an RA (a partial credit model; PCM) were fitted to the observed data. Results were assessed considering the items’ content and the questionnaire’s underlying theoretical framework. Based on these assessments, item reduction and adjustments to the questionnaire scoring were considered in a stepwise manner. This entailed assessing the goodness of fit after each adjustment to the questionnaire. This process was carried out until the goodness of fit and content validity were considered satisfactory.

After assessing the underlying assumptions, item and person location parameters were estimated using RA. The item and person locations shed light on the range of the latent trait that the questionnaire is capable of measuring and the range of the location of study participants. Item and respondent locations along the latent trait were illustrated on a person-item map.³⁵ Comparing the two location parameters indicated whether the questionnaire targeted the appropriate study population.

Data and resource availability

The dataset analysed during the current study is not publicly available due to General Data Protection Regulations but is available from last author KKN on reasonable request.

Results

In total, 248 of the 277 women randomised in the Face-it trial responded to the ISS-GDM at 12 months follow-up. Attrition rates and analyses in the Face-it trial have been reported elsewhere.¹⁶ The background characteristics of the participants in this study are presented in table 1.

Overall, the responses to the ISS-GDM reflected a relatively low endorsement of items. The mean score for all items, except items 5 and 10, was below 1.5, on a scale from 0 to 4, indicating relatively low levels of internalised stigma. Although all response categories were used for all items, a floor effect was identified across the scale, suggesting poor population targeting (table 2). Nevertheless, a notable proportion of respondents agreed or strongly agreed with statements reflecting self-disappointment (item 4: 18.5%), self-blame for developing GDM (item 8: 16%) or potential adverse health outcomes in their children (item 5: 31%) and having an altered self-perception following the GDM diagnosis (item 10: 35%). Less endorsed statements reflected feeling inferior to other mothers (item 7: 4%) or guilt toward child’s family members (item 6: 4%).

Table 2. Response patterns and factor loadings for the 11-item ISS-GDM and 10-item ISS-GDM.

Item nr and wording	Response options, n (%)*					Mean (SD)	Factor loadings, z-value (P)
	Strongly disagree	Disagree	Neither agree nor disagree	Agree	Strongly agree		11-item ISS-GDM (with item 2)	10-item ISS-GDM (without item 2)
1. I am embarrassed or ashamed that I have had GDM	107 (43.1)	51 (20.5)	53 (21.4)	29 (11.7)	8 (3.2)	1.11 (±1.18)
2. I am as good a mother as mothers who have not had GDM†	16 (6.5)	3 (1.2)	4 (1.6)	35 (14.1)	190 (76.7)	0.47 (±1.06)	0.96 (0.34)	–
3. Stereotypes about women with GDM apply to me	102 (41.1)	43 (17.3)	73 (29.4)	24 (9.7)	6 (2.4)	1.15 (±1.14)	7.35 (<0.01)	6.82 (<0.01)
4. I am disappointed in myself for having had GDM	106 (42.7)	54 (21.8)	42 (16.9)	32 (12.9)	14 (5.6)	1.17 (±1.26)	10.82 (<0.01)	11.40 (<0.01)
5. Due to my GDM I feel like it is my fault if my child develops weight or health issues later in life	54 (21.8)	53 (21.4)	65 (26.2)	52 (21.0)	24 (9.7)	1.75 (±1.28)	11.16 (<0.01)	8.80 (<0.01)
6. Due to my GDM I feel guilt towards my child’s close family members (eg, father or grandparents)	156 (62.9)	53 (21.4)	23 (9.3)	12 (4.8)	4 (1.6)	0.61 (±0.95)	6.52 (<0.01)	7.93 (<0.01)
7. I feel inferior to other mothers, who have not had GDM	178 (71.8)	48 (19.4)	12 (4.8)	8 (3.2)	2 (0.8)	0.42 (±0.80)	6.23 (<0.01)	7.71 (<0.01)
8. It is my own fault that I have had GDM	91 (36.7)	47 (19.0)	70 (28.2)	32 (12.9)	8 (3.2)	1.27 (±1.18)	9.06 (<0.01)	9.33 (< 0.01)
9. I feel guilty about prioritising spending time on my own health, instead of for example spending time with my child	77 (31.0)	61 (24.6)	55 (22.2)	39 (15.7)	16 (6.5)	1.42 (±1.25)	5.60 (<0.01)	5.43 (<0.01)
10. The GDM diagnosis has not affected how I see myself†	16 (6.5)	71 (28.6)	43 (17.3)	55 (22.2)	63 (25.4)	1.69 (±1.30)	8.39 (<0.01)	6.89 (<0.01)
11. Have you told family and/or friends about your GDM diagnosis?	Yes, to people in general		Yes, but only to a few		No	0.52 (±1.15)	4.59 (<0.01)	4.96 (<0.01)
	206 (83.1)		40 (16.1)		2 (0.8)

^*Within rows, percentages do not always total 100 due to rounding.

^†Reversed scoring (see online supplemental appendix A).

GDM, gestational diabetes mellitus; ISS-GDM, Internalised Stigma Scale for Gestational Diabetes Mellitus.

The initial CFA (χ²=93.50, df=44, p<0.001) revealed a poor fit. Item 2 did not load onto the latent trait (estimate 0.08, p=0.34), indicating multidimensionality in data. Additionally, the MI suggested LRD for two item-pairs (6 and 7 (MI=13.4); 1 and 4 (MI=13.2)). Anomalies were also identified by assessing the infit and outfit statistics, as analyses revealed that items 2 and 9 seemed to be underfitted to data, with outfit statistics at 4.3 and 1.6 and infit statistics at 2.0 and 1.4, respectively. The Q3 statistics confirmed the LRD indicated by the MI. Finally, DIF was identified for items 3 and 8, which were found to systematically differ relative to the respondents’ pre-pregnancy BMI (<27 kg/m²/≥ 27 kg/m²) and BMI 12 months after giving birth (<30 kg/m²/≥ 30 kg/m²).

Based on these initial analyses, it was decided to omit item 2 from the sum score yet still include it in the questionnaire as a stand-alone item, to allow for further validation across populations and contexts. When testing the fit of data without item 2 to the PCM, anomalies pertained in terms of LRD (analyses not shown). However, after adjusting for this, the tests of fit were considered satisfactory.

All remaining items in the scale loaded onto the same latent trait (χ²=54.056, df=33, p=0.012), the CFI was 0.97, the TLI was 0.96 and the RMSEA was 0.05. Inter-item correlations were all positive, and Cronbach’s alpha was acceptable, with a value of 0.78. Item response curves showed satisfactory homogeneity in data, with the expected ordering of the responses from 0 to 4 points on the Likert scale situated from left to right according to the latent trait. LRD was no longer present in the data as the difference between the largest Q3 value and the average residual correlation was 0.17.

Although the CFA and PCM fit was considered satisfactory (conditional log-likelihood=−1695.62, iterations=129, parameters=37), item 9 still showed infit and outfit statistics of 1.49 and 1.37, respectively. DIF was still identified for items 3 and 8 according to pre-pregnancy BMI (<27 kg/m²/≥ 27 kg/m²) and BMI 12 months after giving birth (<30 kg/m²/≥ 30 kg/m²). Figure 1 illustrates that, on average, respondents with pre-pregnancy BMI ≥27 kg/m² scored higher on items 3 (by 0.8 points) and 8 (by 0.7 points). Likewise, respondents with BMI ≥30 kg/m² 12 months after giving birth scored higher on items 3 (by 1.1 points) and 8 (by 0.7 points) (figure 1). No DIF was identified for the remaining exogenous variables.

Figure 1. Graphical illustrations of the impact of DIF according to pre-pregnancy BMI and BMI 12 months after giving birth in items 3 and 8. Graphs A and B illustrate the respondents' scoring on items 3 and 8, stratified by pre-pregnancy BMI =>27 and <27. Graphs C and D illustrate the respondents’ scoring on items 3 and 8, stratified by BMI 12 months after giving birth =>30 and <30.

Concurrent validity was found with a negative association between ISS-GDM and WHO-5 (r=−0.14, 95% CI −0.26 to −0.02]) and a positive correlation between ISS-GDM and PSS (r=0.31, 95% CI 0.19 to 0.42) and GAD-7 (r=0.17, 95% CI 0.05 to 0.29). Similarly, concurrent validity was indicated with a negative correlation between item 2 and WHO-5 (r=−0.11, 95% CI−0.23 to 0.01) and a positive correlation between item 2 and PSS (r=0.17, 95% CI 0.05 to 0.29) and GAD-7 (r=0.045, 95% CI−0.08 to 0.17).

Finally, figure 2 illustrates investigations of item difficulty, item thresholds and person parameters. The item difficulty locations ranged from item 5 (location=−0.41) to item 11 (location=2.01), with thresholds ranging from a minimum of −1.49 to a maximum of 2.92. The person parameter analyses showed that the respondents ranged from a minimum of −3.33 to a maximum of 1.17, with a median of −0.65 and a mean of −0.78.

Figure 2. Person-Item Map illustrating item difficulty (under the horizontal line) and person parameter (over the horizontal line) along the latent dimension, i.e., internalised GDM-specific stigma, measured using the ISS-GDM questionnaire.

Discussion

The novel ISS-GDM questionnaire fills a critical research gap as the first validated questionnaire measuring GDM-specific internalised stigma. The CFA supported reporting items 1 and 3–11 as a sum score, and a satisfactory fit to the PCM further confirmed the scale’s validity. Item-level patterns revealed that feelings of guilt, self-blame and altered self-perception were among the most frequently endorsed experiences. In contrast, perceived inferiority and guilt toward the child’s family members were less common. Collectively, the findings suggest that despite low overall stigma levels, a notable proportion of women remained affected by their GDM experience 12 months after birth, aligning with existing qualitative studies.⁸ Since the ISS-GDM was developed in 2019, additional research on GDM stigma has emerged,¹² reflecting growing international attention to the topic, including its measurement. The finding that the ISS-GDM is a valid and reliable questionnaire for quantifying self-reported internalised GDM-specific stigma at a group or population level among Danish women with prior GDM therefore marks an important development in the field of GDM stigma research.

The validation of the ISS-GDM identified a few anomalies. First, the CFA suggested that item 2 did not measure the same underlying construct as the remaining items. Item 2 addressed respondent’s perception of their abilities as mothers, which may have reduced salience 12 months after birth when the respondents have gained more concrete experience with motherhood. While the theme was identified in existing qualitative studies,⁸ the underlying theme of motherhood could also be perceived as distinct from the respondents’ perception of having had GDM, as the diagnosis might be perceived as independent of their capabilities as mothers. Similarly, while item 9 appeared underfitted to the PCM, it was considered important to retain, as it reflects broader maternal norms and expectations that may intersect with GDM-related experiences of guilt and self-blame, thus contributing to internalised GDM stigma. We encourage future research to assess the fit statistics of item 2 and item 9 in study populations across cultures and settings.

The CFA and RA indicated LRD between item-pairs 1 and 4, and 6 and 7. The overlap between feeling ‘embarrassed or ashamed’ (item 1) and ‘disappointed in myself’ (item 4) may reflect semantic nuances in the Danish wording, which conveys slightly different meanings than in English. The LRD between items 6 and 7 likely reflects feelings directed toward others, whereas the remaining items concern self-directed feelings. Although LRD might reduce the scale reliability, both pairs were retained to preserve content validity and capture the nuances covered by each item.

Investigations of DIF revealed that responses to items 3 and 8 differed by pre-pregnancy BMI and BMI 12 months after birth. Respondents with a pre-pregnancy BMI≥27 tended to report systematically higher values than those who had a pre-pregnancy BMI <27 (figure 1). The wording of items 3 and 8 indicates a potential overlap between internalised weight stigma and internalised GDM-specific stigma, as common perceptions are that weight is the leading risk factor for GDM and within personal control.¹³ Future research using the ISS-GDM should account for this DIF to ensure comparability across BMI groups.

The item parameters in the PCM indicated that the ISS-GDM can measure high levels of internalised GDM-specific stigma, whereas the person parameters revealed poor targeting, as the study population seemed to have relatively low levels of internalised GDM-specific stigma (figure 2), supported by the floor effect across items (table 2). This was expected, as participants in the Face-it trial were generally more resourceful than the background population, with higher rates of primiparity and singleton deliveries and lower rates of preterm delivery and pre-pregnancy obesity.¹⁸ Attrition analyses further showed that women lost to follow-up had higher BMI, blood glucose and tobacco use and were less likely to be employed,¹⁶ factors that could potentially impact internalised stigma. As the ISS-GDM was administered 12 months after birth, stigma levels during pregnancy or at diagnosis may have been higher at this time. Meanwhile, participation in a trial with a focus on GDM and post-partum well-being may also have increased awareness of stigma. Despite these contextual factors, the ISS-GDM was found to be valid and reliable for assessing high levels of internalised GDM stigma.

Although weak, concurrent validity was observed between the ISS-GDM and measures of perceived stress (PSS), anxiety (GAD-7) and well-being (WHO-5). The weaker correlations compared with people with manifest diabetes suggest that internalised GDM stigma 12 months after birth may be influenced by the transient nature of GDM, a hypothesis that merits further investigation. Future studies could advantageously assess concurrent validity with conceptually related constructs, such as weight bias internalisation, to better understand how internalised GDM stigma aligns with and differs from other forms of diabetes- or weight-related stigma.

A core strength of this study was that it was the first to develop and validate a questionnaire quantifying GDM-specific internalised stigma. The scale was developed by drawing on stigma theory and existing international research on stigma and GDM, and content and face validity were assessed before data collection. The study also employed high-quality data, as data collection followed standard operating procedures for the Face-it trial. Another strength of this study was that the ISS-GDM scale was validated through thorough psychometric testing, assessing the validity and reliability of the questionnaire conducting both CFA and RA.

However, this study is not without limitations. As the ISS-GDM was included in an already lengthy questionnaire assessing the effect of the Face-it trial, test-retest reliability was not assessed to avoid burdening the respondents unnecessarily. For the same reason, the item pool was deliberately kept small, comprising only items considered most relevant by the authors, experts and women with prior GDM as well as items supported by qualitative findings, to capture key aspects of internalised GDM stigma.¹³ Due to limited time and resources available during the development of the ISS-GDM, the questionnaire was also only cognitively de-briefed with two women with prior GDM. Additionally, the scale’s sensitivity to change and acceptability were not assessed and we encourage future studies to include these considerations. Nevertheless, the ISS-GDM was found to be an essential first step towards quantifying internalised GDM stigma.

For future use of the ISS-GDM, we encourage including the full 11-item questionnaire. This is to ensure thorough translation, adaptation and validation of the ISS-GDM in diverse populations and to investigate whether the identified anomalies in this study are replicated in other countries, cultures and settings. The outcome of items 1 and 3–11 can be presented as a sum score ranging from 0 to 100, and item 2 as a stand-alone item also scored from 0 to 100, with higher scores indicating higher levels of internalised GDM stigma (see online supplemental appendix A for scoring guide).

Important remaining knowledge gaps are the extent of experienced stigma due to GDM as well as the level of internalised GDM stigma during pregnancy; such efforts are ongoing.¹² We encourage future research to investigate the potential risk factors and implications of having a high degree of internalised GDM stigma, both on the women’s mental, social and physical health, as well as their healthcare-seeking behaviour.

Our findings indicate that the ISS-GDM is a reliable and valid questionnaire measuring internalised GDM-specific stigma among Danish women with prior GDM. The questionnaire can provide novel insights into the extent and nature of internalised GDM-specific stigma in this population and may serve as a foundation for future studies evaluating stigma.

Data availability statement

Data are available upon reasonable request.