Elevated psychological distress and suicidality at entry

The high rate of suicidal ideation at clinical entry – reported by 16.6% of participants – suggests that many women accessing menopause services are doing so in a state of significant psychological distress. This is consistent with broader literature indicating that midlife women, particularly during perimenopause, experience a spike in suicidal thoughts, often under-recognised in healthcare settings.^{Reference Martin-Key, Crowley, Hudaib, Newson and Kulkarni3,Reference Mallon, Broman and Hetta24} Although it is possible that elevated rates in the present sample partly reflect the nature of the clinic – a specialist service to which individuals may self-refer during acute distress – this does not negate the relevance of the findings. Rather, it highlights the unmet needs and urgency that drive help-seeking among this population, many of whom report long-standing difficulties before referral. The gap between distress prevalence and clinical response may stem from the continued medicalisation or minimisation of menopausal symptoms,^{Reference Barber and Charles12} coupled with diagnostic tools that are poorly attuned to menopause-specific experiences.

Although national suicide statistics reflect confirmed deaths rather than ideation,¹⁸ this study supports the argument that suicidal thinking is more prevalent among menopausal women than previously assumed, and that the menopause itself may represent a high-risk psychological state, not simply a hormonal shift. This aligns with neurobiological theories framing perimenopause as a time of neurochemical sensitivity, particularly in systems involved in mood regulation.^{Reference Soares and Zitek4,Reference Brinton, Yao, Yin, Mack and Cadenas25}

Mental health improvements and hormone regimen differences

Mental health outcomes improved across the study period, but not uniformly. At baseline, PHQ-9 scores across treatment groups ranged from 9.5 to 12.3, indicating that many participants were experiencing moderate depressive symptoms at the point of entry. MENO-D scores similarly suggested clinically meaningful distress. These elevated scores, alongside the 16.6% of participants reporting active suicidal ideation at baseline, underscore the importance of systematic screening for psychological distress and suicidality within menopause care settings. Participants receiving combined oestrogen, progesterone and testosterone experienced the most substantial and consistent reductions in depressive symptoms and psychological distress. These findings support prior work highlighting oestrogen’s stabilising effects on serotonergic and dopaminergic systems,^{Reference Gillies and McArthur26} and further suggest that testosterone may augment mood, motivation and energy: factors critical to psychological recovery.^{Reference Scott and Newson14,Reference Montgomery, Bancroft, Fox and Cowan27} Although oestrogen alone has been shown to improve mild to moderate depression in perimenopausal women,^{Reference Zweifel and O’Brien16,Reference Onalan, Onalan, Ustun, Guner and Kilic28} the present study adds to growing evidence that multi-hormonal regimens may be more effective, especially when symptoms are severe or complex.^{Reference Herson and Kulkarni8,Reference Studd and Panay15} However, given the exploratory nature of these subgroup analyses and the small sample sizes in several HRT regimens, these findings should be interpreted with caution. The strong effect sizes observed across groups were generally comparable, suggesting that different hormone combinations may offer broadly similar levels of psychological benefit rather than clear superiority of one regimen over another. The lack of a non-treatment control group further limits causal attribution and prevents firm conclusions about whether improvements were driven solely by HRT or by non-specific therapeutic effects.

Although improvements were most pronounced in certain HRT regimens, it is also possible that mental health gains reflected a broader combination of factors. Participants may have benefited from holistic menopause care, including psychoeducation, improved sleep, symptom validation or concurrent support for anxiety, relationship or occupational stressors.^{Reference Souza29} Lifestyle changes, antidepressant use or progression through the menopausal transition itself may also have contributed to psychological stabilisation.^{Reference Dugan, Norman, Brown, MacMillan, Van Horn and Powell30–Reference Soares, Arsenio, Joffe, Bankier, Cassano and Petrillo32} Given the lack of a non-treatment control group, improvements observed across the study period may reflect both treatment-specific and non-specific effects, and should be interpreted within this broader clinical context.

Conversely, the lack of improvement in the progesterone-only group reflects previous literature highlighting the limited antidepressant properties of progesterone in isolation, as some studies suggest that progesterone may even worsen mood symptoms in susceptible individuals.^{Reference Zweifel and O’Brien16,Reference Dolitsky, Kaufman and Rohan33} However, this result should be interpreted cautiously because of the very small sample size in this group (n = 9). There is also evidence pointing to potential anxiolytic or stabilising effects in certain contexts,^{Reference Toriizuka, Yoneda and Kimura17,Reference de Lignières and Vincens34} suggesting that progesterone’s psychological impact may vary by symptom type, delivery method or individual sensitivity. Future studies should explore these nuances in more detail.

Timing of treatment and perimenopausal vulnerability

Around half of the sample was perimenopausal at baseline, which aligns with literature identifying this phase as a window of increased risk for mood disturbance and suicidality.^{Reference Maki, Kornstein, Joffe, Bromberger, Freeman and Athappilly2} Hormonal fluctuations during perimenopause may induce mood symptoms that are acute, episodic and reactive; distinct from the more stable presentation seen in major depressive disorder.^{Reference Kulkarni, Gavrilidis, Worsley, Van Rheenen and Hayes21,Reference Gibbs, Kulkarni and Aitchison35} These symptoms may also be more responsive to hormonal intervention, particularly when treatment is initiated early. The ‘window of opportunity’ hypothesis^{Reference Soares and Frey36,Reference Lumsden37} suggests that initiating HRT near the onset of menopause yields greater benefit across both somatic and psychological domains. Although baseline HRT status did not significantly moderate treatment effects in this study, improvements observed among those initiating treatment during the study period tentatively support the importance of treatment timing. However, conclusions are limited by the lack of detailed data on HRT duration before baseline. Similar patterns have been observed in bone density and cardiovascular outcomes, where earlier intervention appears to enhance biological responsiveness.^{Reference Lees, Kinsella and Liu-Ambrose38} However high levels of missingness meant that extended timepoints (e.g. time point 3) were excluded from analysis, as their inclusion would have introduced analytic instability. This decision aligns with established recommendations regarding missing data thresholds.^{Reference Jakobsen, Gluud, Wetterslev and Winkel22,Reference Middleton, Nguyen, Moreno-Betancur, Carlin and Lee23}

Suicidality as a distinct symptom dimension

Suicidal ideation declined by over 90% among those who reported it at baseline, with a large effect size, but this improvement was not driven by specific hormone regimens. Instead, the time × baseline suicidality interaction suggests that participants presenting in acute distress improved substantially regardless of treatment type. Although this could partly reflect regression to the mean, the magnitude and consistency of improvement – paired with reductions in depression and self-worth measures – indicates meaningful clinical change for this high-risk subgroup. Notably, suicidality showed only weak associations with total PHQ-9 and MENO-D scores, supporting research that it can occur independently of general depression.^{Reference van Ballegooijen, Riper, Cuijpers, van Oppen, Smit and Donker13,Reference van Bentum, Sijbrandij, Huibers, Holmes, Blackwell and Arntz20} This challenges the tendency to treat suicidality as just another symptom of depression and underscores the importance of assessing it directly, particularly in perimenopausal populations, where emotional reactivity and cognitive shifts may heighten risk outside of classical depressive profiles.

The MENO-D item on self-worth, used as a proxy measure, also improved over time, but was not strongly correlated with PHQ-9 suicidality scores. This further supports the need for menopause-sensitive tools that capture suicidality as a specific and nuanced construct rather than relying on indirect indicators.

Value of menopause-specific measures

Although PHQ-9 and MENO-D total scores were highly correlated, the MENO-D was developed specifically to capture menopause-related psychological features that conventional depression scales may overlook, such as irritability, paranoia and libido changes.^{Reference Kulkarni, Gavrilidis, Worsley, Van Rheenen and Hayes21} These symptom dimensions may represent a distinct profile of perimenopausal depression,^{Reference Parry39,Reference Worsley, Bell, Gartoulla and Davis40} which often manifests more as mood instability than persistent low mood. Although the two scales showed convergence in this sample, the value of the MENO-D lies in its potential to detect cases that would otherwise go unflagged by generic tools. The high correlation observed in this sample may reflect the severity and breadth of symptoms reported by participants, with distress manifesting across both general and menopause-specific domains. This convergence suggests that, in highly symptomatic populations, general depression and menopause-specific mood symptoms may become more tightly intertwined.

Given the unique hormonal and psychological dynamics of perimenopause, further validation of menopause-specific scales is needed. Combining these with general tools may offer a more complete clinical picture, helping to target interventions more effectively.

Lifestyle factors and treatment response

Higher BMI was significantly associated with worse scores on depression, distress and self-worth, consistent with literature linking elevated BMI to lower mood and self-esteem in women.^{Reference Lumsden37,Reference Lees, Kinsella and Liu-Ambrose38} However, BMI did not significantly moderate the effects of HRT, suggesting that hormone therapy may still be effective across weight categories. Still, there is evidence that higher BMI may alter hormone metabolism or influence pharmacodynamics,^{Reference Kristensen, Christensen and Bliddal41,Reference Modugno, Kip, Ness and Rebbeck42} which warrants exploration in future dosing studies.

Smoking showed only weak associations with depression and no impact on treatment response. Although smoking has been linked to serotonergic dysregulation and increased depression risk,^{Reference Fluharty, Taylor, Grabski and Munafò43,Reference Malone, Waternaux, Haas and Mann44} the small proportion of smokers in this sample (around 4%) may have limited statistical power.

Time between assessments did not moderate outcomes, but was controlled for in all models, to account for variability in treatment exposure duration. This ensured that symptom changes were not confounded by differences in follow-up length.

Although this study was based in a UK specialist clinic, the findings have broader relevance for global mental health systems. In many regions, menopause-related psychological distress and suicidality remain under-recognised, and opportunities for early intervention are often missed. Integrating basic mental health screening into menopause services could offer a scalable strategy for reducing distress and suicide risk across diverse care settings.

Limitations

Several limitations should be considered when interpreting these findings. First, the study was observational and based on routinely collected clinical data, which limits the ability to infer causality. Without a control group, improvements may partially reflect non-specific factors such as regression to the mean, placebo effects or increased clinical engagement rather than direct effects of HRT. Additionally, participants were recruited from a private specialist menopause clinic, which may limit generalisability to broader populations. Individuals accessing private care may differ in terms of symptom severity, socioeconomic status, health literacy or prior treatment access compared with those using the National Health Service. Replication in more diverse clinical settings is needed to strengthen the applicability of these findings.

Second, HRT dosage and route of administration (e.g. oral versus transdermal) were not captured. This is a notable limitation given that delivery method can influence both hormone levels and mood-related outcomes.^{Reference Patel45} Additionally, duration of prior HRT use was not consistently documented in the clinical data-set and could not be reliably examined in relation to outcomes. Future studies should systematically assess and compare dosage and formulation variables, and explore the role of HRT treatment duration in more depth, given its potential impact on clinical response.

Third, some HRT subgroups – particularly the progesterone-only group – had small sample sizes, which reduced statistical power and limited the generalisability of subgroup comparisons (see ‘HRT classification’). Additional research with more balanced group sizes is needed to confirm the differential effects observed here.

Fourth, although additional time points were available, data for time point 3 were missing for 59% of participants. Including time point 3 would have introduced analytic instability and reduced the robustness of conclusions. Scholars argue that analyses with >40% missing data should not be considered confirmatory,^{Reference Jakobsen, Gluud, Wetterslev and Winkel22} and studies have shown that multiple imputation techniques perform poorly beyond 50% missingness, particularly when model assumptions may not hold.^{Reference Middleton, Nguyen, Moreno-Betancur, Carlin and Lee23} As such, we limited analyses to the most complete time points (time points 1 and 2).

Fifth, suicidality was assessed with a single item from each scale. Although PHQ-9 item 9 is widely used in clinical practice, it may not fully capture the nuance and range of suicidal thinking. The use of MENO-D item 4 as a proxy for suicidality added value, but was not designed for this purpose, and should be interpreted cautiously.

Sixth, although lifestyle variables such as alcohol use were collected, inconsistent reporting meant this factor could not be reliably included as a covariate. Given alcohol’s established links to depression and suicidality,^{Reference Patel45,Reference Wasserman46} its omission is a limitation. Future research should incorporate standardised, high-quality measures of substance use, sleep, physical activity and psychosocial stressors. Additionally, future research could also usefully explore whether changes in other covariates – if collected thoroughly – such as specific symptom domains (e.g. vasomotor symptoms), mediate the relationship between HRT and mental health outcomes. However, caution is warranted because of the potential multicollinearity between overlapping symptom measures within tools like the MENO-D.

Seventh, item-level outcomes (e.g., PHQ-9 item 9, MENO-D item 4) were ordinal, but analysed as continuous variables in ANOVA, which assumes interval-level data. Although this is common in psychological research and enables parametric modelling with covariates, it may obscure categorical nuances. Findings should be interpreted with caution, and future studies may benefit from categorical or non-parametric approaches.

Finally, both the PHQ-9 and MENO-D, although validated for symptom screening, were not specifically designed to track treatment response. This limits precision when interpreting change scores over time. Broader use of clinician-rated outcomes or structured diagnostic interviews may strengthen future evaluations.

Conclusions

This study provides preliminary evidence that combined HRT – including regimens such as combined oestrogen and progesterone, and combined oestrogen, progesterone and testosterone – may be associated with significant improvements in depressive symptoms and menopause-specific psychological distress among menopausal women. Although suicidality declined over time among those at risk, this effect was not moderated by HRT type, suggesting that broader clinical or psychosocial factors may be at play.

The findings reinforce the importance of assessing suicidality as a distinct outcome during menopause, rather than subsuming it within general depression scores. Standard tools may not fully capture the complexity of mood instability and suicidal ideation during this life stage, highlighting the potential value of menopause-specific instruments like the MENO-D.

Although the observational design limits causal conclusions, the large clinical sample and consistency across measures suggest meaningful psychological improvements associated with multi-hormonal treatment. These results support a more individualised approach to menopause care: one that considers hormonal profile, mental health history and timing of intervention.

Future research should explore the impact of HRT dosage, delivery method and treatment timing by using longitudinal and randomised designs. Direct comparisons of oestrogen-only and oestrogen and testosterone regimens are also warranted, particularly in post-surgical or premature menopause populations, to determine whether testosterone adds incremental benefit beyond oestrogen alone. Further development and validation of menopause-sensitive screening tools are also needed to ensure that suicidality and mood symptoms are accurately identified and addressed. Integrating mental health screening into routine menopause care could improve early detection, guide targeted treatment and ultimately reduce psychological burden during this vulnerable life stage.