Beyond the Symptom Score: What Standard Psychiatric Rating Scales Fail to Capture — A Narrative Review

1. Introduction

Psychiatric outcome measurement has long been anchored to symptom severity scales. The Positive and Negative Syndrome Scale (PANSS), the Hamilton Depression Rating Scale (HAM-D), the Montgomery–Åsberg Depression Rating Scale (MADRS) and their self-report counterparts served the field’s pragmatic needs well: they standardized diagnosis, enabled regulatory demonstrations of drug efficacy and gave clinicians a common language. They have also quietly defined what counts as improvement. Methodologists have argued that excessive reliance on the symptoms that constitute diagnostic criteria “has impoverished clinical assessment,” obscuring demoralization, psychological well-being, mental pain, social adjustment, illness behavior and patient satisfaction (Guidi et al., 2018).

This gap has practical consequences. Measurement-based care—the routine use of patient-reported and clinician-rated measures to guide treatment—demonstrably improves on usual care, yet fewer than one in five practitioners use it (17.9% of psychiatrists, 11.1% of psychologists and 13.9% of master’s-level practitioners), and as few as 5% measure at the empirically supported every-session frequency (Lewis et al., 2019). Where measurement is adopted, the instruments chosen are usually symptom scales, so the outcomes that matter most to patients may remain invisible even in services that do measure.

Previous critiques have typically examined one neglected outcome at a time—functioning, quality of life, or personal recovery. This review brings these literatures together to answer a single question: which clinically and personally meaningful improvements do standard psychiatric rating scales fail to capture, and how should measurement change in response? We map the evidence across ten outcome domains and two cross-cutting methodological limitations (Figure 1), summarize candidate measures for each (Table 1) and propose a pragmatic, tiered outcome set for routine care and research (Table 2).

2. Methods

This is a narrative review designed and reported with reference to the Scale for the Assessment of Narrative Review Articles (SANRA) criteria (Baethge et al., 2019). Targeted searches of PubMed/MEDLINE, Scopus and Google Scholar combined terms for outcome measurement (“rating scale,” “outcome assessment,” “measurement-based care,” “patient-reported outcome”) with domain-specific terms (functioning, cognition, social cognition, negative symptoms, anhedonia, subjective response, quality of life, well-being, personal recovery, self-stigma, mortality, sleep, ecological momentary assessment, digital phenotyping and shared decision-making). Reference lists of key reviews were hand-searched, and seminal conceptual papers were retrieved by citation chaining.

Priority was given to systematic reviews and meta-analyses, large measurement-validation studies, randomized controlled trials, and consensus statements or Commission reports; English-language, peer-reviewed sources were included. A domain was retained when three conditions were met: (a) evidence showed that it is dissociable from, or only weakly correlated with, symptom severity; (b) it carries prognostic value or is prioritized by patients; and (c) it is absent from, or under-represented in, conventional symptom scales. Evidence was synthesized thematically. Consistent with the narrative design, no formal risk-of-bias assessment or quantitative pooling was undertaken.

3. Results

3.1. Structural Limits of Symptom-Focused Scales

Several design features limit what conventional scales can see. First, they are optimized for diagnostic and severity classification rather than for detecting change in domains patients value. Second, they suffer from measurement problems that reduce assay sensitivity—the capacity to detect a true treatment effect. In depression trials, low assay sensitivity has been attributed to inappropriately broad measures, recall bias and poor inter-rater reliability, compounded by difficulty separating trait-like from state-like features of symptoms (Mofsen et al., 2019). Third, clinician administration, long treated as the gold standard, may compound rather than correct response bias; clinicians may magnify rather than neutralize patients’ reporting errors (Mofsen et al., 2019).

The field has begun to acknowledge these limits. Studying symptoms in the context of everyday life is increasingly recognized as a necessary addition to conventional assessment (Myin-Germeys et al., 2018), and the WPA–Lancet Psychiatry Commission on the Future of Psychiatry recommended a dimensional component to diagnosis and the broader uptake of measurement-based care using both clinician-administered and self-report instruments (Bhugra et al., 2017).

3.2. Domain 1: Real-World Functioning and Functional Capacity

Functioning is perhaps the most consequential omission. Although depression trials conventionally report both symptom severity and functional impairment, the relationship between them “remains unexpectedly weak and often bidirectional,” and functional change often lags behind symptomatic change (McKnight and Kashdan, 2009). Symptom remission is therefore not equivalent to functional recovery (Yang et al., 2022).

The divergence is measurable. Quality-of-life measures correlate only weakly to moderately with symptom measures, and symptom and quality-of-life trajectories diverge over time, both naturalistically and in response to treatment; some patients maintain role functioning despite severe symptoms, whereas others report marked functional decrements despite few symptoms (Murray et al., 2017).

Functional outcome also has a layered structure that symptom scales cannot resolve. Cognitive performance predicts performance-based functional skills, and those skills—rather than cognition directly—account for variance in work, community activities and interpersonal function; negative symptoms and depression relate to real-world performance independently of cognition (Bowie and Harvey, 2006). In the large VALERO validation study, correlations between cognition or functional capacity and real-world outcomes ranged from modest to negligible depending on the rating method, and many real-world functioning scales were “not strongly related to indices of functional abilities that are the state of the art outcomes in treatment studies” (P.D. Harvey et al., 2011).

3.3. Domain 2: Cognition and Social Cognition

Cognitive impairment is a core feature of psychotic disorders and is largely invisible to symptom scales. Its correlations with hallucinations and delusions are low, yet cognitive deficits are reliable predictors of functional outcome and disability—more strongly so than psychotic symptoms—partly because impairment limits the benefit patients can derive from rehabilitation (Green, 2016). The apparent size of this relationship depends on how functioning is measured: correlations with performance-based measures such as the UCSD Performance-based Skills Assessment range from r = 0.4 to 0.8, whereas those with interview-based ratings range from 0.1 to 0.3 (McCutcheon et al., 2023; Figure 2B). Cognitive impairment is also linked to poorer treatment adherence, more hospital admissions and longer stays, and—together with negative symptoms—is a major driver of lost productivity in a population in which 80–90% are unemployed (McCutcheon et al., 2023).

Social cognition is similarly neglected. Social-cognitive deficits explain variance in daily and community functioning beyond that explained by basic neurocognition (Shamsi et al., 2011), and social cognition has a multidimensional structure with distinct relationships to functional outcome (Mancuso et al., 2011). Because cognitive and social-cognitive competence are more proximal to functional capacity than to real-world functioning, their relationship with community performance is attenuated by personal factors such as motivation and self-efficacy and by socio-environmental factors such as disability policy (Mancuso et al., 2011).

3.4. Domain 3: Negative Symptoms, Motivation and Anhedonia

Negative symptoms add explanatory power that positive-symptom-weighted scores miss. Global negative symptoms contribute substantial variance in community functioning and employment beyond that explained by neurocognition and functional capacity (Yang et al., 2021), and they can mediate the path from functional capacity to real-world functioning (Couture et al., 2011).

Conventional instruments also misrepresent the construct. Anhedonia has long been defined as an inability to experience pleasure, but reward deficits in depression and schizophrenia are broader than hedonic response (Treadway and Zald, 2013). Widely used anhedonia measures emphasize consummatory pleasure, with little attention to anticipation, drive or effort (Nusslock and Alloy, 2017; Rizvi et al., 2016). Hedonic function is better understood as a sequence—reward association, interest and desire, anticipation, motivation, effort, hedonic response and feedback integration—and parsing these facets is essential to understanding their neurobiology (Rizvi et al., 2016). Impairments in the ability to pursue and learn about reward are documented across depression, schizophrenia and addiction (Thomsen, 2015). A patient whose motivation recovers while hedonic ratings remain unchanged will appear unimproved on most current scales.

3.5. Domain 4: Subjective Experience of Medication and Tolerability

Symptom scales report efficacy; they rarely capture how treatment feels. In a survey of 250 outpatients with schizophrenia-spectrum disorders, 93.6% rated efficacy and 83.6% rated side effects as “very” or “most” important when deciding whether to take prescribed medication, 87.6% valued being able to think more clearly, and weight gain, physical restlessness and somnolence were rated highly important by 61.6%, 60.8% and 58.8%, respectively (Achtyes et al., 2018; Figure 2A).

Subjective response is a legitimate clinical signal. Reviews of patients’ subjective experiences with antipsychotics concluded that clinicians should not dismiss such complaints as unreliable and that researchers should not treat them as non-scientific, because they inform both psychopathology and management (Awad et al., 1995). Treatment satisfaction depends on multiple factors beyond symptom reduction, and adverse effects deserve standing as outcome measures in their own right (Hamer and Haddad, 2007). Because these experiences shape adherence, they shape every downstream outcome.

3.6. Domain 5: Quality of Life and Positive Mental Health

Quality of life and well-being constitute a separate outcome axis. In people with schizophrenia, objective and subjective quality-of-life indices overlap only modestly, and patients’ reports remain essential even when they diverge from objective information, because motivation to engage in treatment rests on the person’s own appraisal of their functioning; unmet needs are a principal driver of quality of life in primary psychosis (Maj et al., 2021).

The conceptual basis is the dual-continua model: the absence of mental disorder is not the presence of positive mental health, and the two are correlated but distinct constructs that should both be assessed (Lukat et al., 2016). Keyes operationalized mental health as 13 dimensions of emotional, psychological and social well-being; “flourishing” requires high levels on at least one of two hedonic dimensions plus six of eleven dimensions of positive functioning, and complete mental health is the absence of mental illness together with flourishing (Keyes, 2002; Slade, 2010). Defining well-being through positive functioning broadens the field and improves prediction of distress and dysfunction (Dodge et al., 2012), and well-being can be measured reliably with instruments such as the Mental Health Continuum–Short Form, the Warwick–Edinburgh Mental Well-being Scale and the Flourishing Scale (Bohlmeijer and Westerhof, 2021).

3.7. Domain 6: Personal Recovery

Standard scales cannot capture recovery as patients define it. The dominant clinical view has defined recovery by symptom remission and level of functioning rather than by the individual’s subjective experience, and it often overlooks substantial disagreement between providers and clients about quality of life, with clients typically reporting better outcomes than their providers (Leonhardt et al., 2017). Personal recovery, by contrast, is a process defined and assessed by the person (Leonhardt et al., 2017), consistent with a recovery perspective that prioritizes agency, context, meaning-making and lived experience (Murray et al., 2017).

The most widely adopted operationalization is the CHIME framework—Connectedness, Hope and optimism, Identity, Meaning in life and Empowerment—derived from a systematic review and narrative synthesis of the personal recovery literature (Leamy et al., 2011). Clinical recovery does not guarantee personal recovery: symptom reduction does not necessarily increase personal recovery, which has been argued to offer a more holistic benchmark for evaluating services (Argentzell et al., 2020).

3.8. Domain 7: Internalized Stigma

Stigma affects outcomes through internalized processes that no symptom item indexes. About one-third of people with severe mental illness report high internalized stigma, which is associated with low self-esteem, empowerment, social support and hope, poorer adherence and lower subjective quality of life (Mashiach-Eizenberg et al., 2013). Internalized stigma is negatively related to hopefulness, self-esteem and social functioning, and it erodes the self-efficacy people need to pursue life goals (Yanos et al., 2008). Self-stigma threatens empowerment, which in turn predicts quality of life (Corrigan et al., 2010), and meta-analytic evidence confirms an association between stigma and subjective quality of life in psychosis (Degnan et al., 2021). A patient whose self-stigma falls may be meaningfully better, yet no conventional severity score will register it.

3.9. Domain 8: Physical Health and Premature Mortality

Conventional psychiatric scales are silent on the physical-health burden that dominates long-term prognosis in severe mental illness. Physical disorders account for approximately 70% of premature deaths in this population, with cardiovascular and metabolic disease the leading contributors (Stubbs et al., 2018). People with severe mental disorders have two- to three-fold higher average mortality and a 10–20-year reduction in life expectancy, with cardiovascular morbidity and mortality 1.5–3 times higher and diabetes prevalence two- to three-fold higher than in the general population (Gronholm et al., 2021; Figure 2D). Most systematic reviews report all-cause mortality 2–3.5 times that of the general population, and the gap has widened over time (de Mooij et al., 2019). The Lancet Psychiatry Commission accordingly identified cardiometabolic disease as a transdiagnostic target requiring reductions in both prevalence and life-course impact (Firth et al., 2019). An antipsychotic that halves PANSS scores while adding 10 kg of weight registers as a success on the primary outcome.

3.10. Domain 9: Sleep and Circadian Function

Sleep disturbance is a transdiagnostic process that symptom scales treat, at best, as a secondary item. Sleep disturbance and psychiatric disorder appear to be mechanistically intertwined, sharing serotonergic and dopaminergic substrates (A.G. Harvey et al., 2011). Insomnia, often assumed to be epiphenomenal to the “primary” disorder, is better understood as an under-recognized contributor to the onset and maintenance of psychiatric disorders (Harvey, 2008). In a highly comorbid sample, insomnia-type sleep disturbance and emotion dysregulation each uniquely predicted anxiety, depression, post-traumatic stress and substance-use symptom severity (Fairholme et al., 2013). In a community-based randomized trial, a transdiagnostic sleep and circadian intervention improved selected measures of functioning, comorbid symptoms and sleep–circadian outcomes—an axis of improvement distinct from the primary symptom scale (Harvey et al., 2021).

3.11. Domain 10: Shared Decision-Making and Person-Centered Process

Standard scales also omit the process outcomes of care. Recovery-oriented care requires greater emphasis on collaboration among professionals, people with mental health conditions and their families (Aoki et al., 2022), and the recovery model locates control with the individual rather than the professional (Slade, 2010). Offering people interventions that fit their preferences increases the likelihood that treatment will enhance personal meaning, satisfaction and quality of life (Schauer et al., 2007), and shared decision-making is best understood as one tool within a broader person-centered recovery-planning process oriented to a life the person has reason to value (Davidson et al., 2017). These are genuine outcomes of care, yet they are invisible to a symptom rating.

3.12. Cross-Cutting Limitation 1: Momentary, Contextual and Passive Dynamics

Retrospective, clinic-based scales cannot capture dynamics. Experience sampling methodology—structured, repeated self-report in daily life—captures variability over time, identifies internal and situational determinants of that variability and enables finer-grained evaluation of treatment effects beyond the clinic (Myin-Germeys et al., 2018). Psychometric challenges of ecological momentary assessment, including whether responses are truly momentary, response standards, selection bias and missingness, are being actively addressed (Stone et al., 2023).

Passively collected data extend this further. Personal sensing—collecting and analyzing data from sensors embedded in daily life—may support monitoring and next-generation mobile interventions (Mohr et al., 2017). Digital phenotyping, the “moment-by-moment quantification of the individual-level human phenotype in situ using data from personal digital devices,” offers fine-grained information potentially more relevant to illness phenotypes than active reports alone (Oudin et al., 2023). Smartphone sensor data have been associated with daily stress and with changes in depression, stress and loneliness (Ben-Zeev et al., 2015), and passive data predicted social anxiety severity (r = 0.702 between predicted and observed scores) with discriminant validity from depression and affect (Jacobson et al., 2020). What counts as a digital biomarker nevertheless remains debated (Montag et al., 2021).

3.13. Cross-Cutting Limitation 2: Discordance Between Raters

Different raters produce different pictures, and standard practice rarely reconciles them. Clinician-rated and self-report measures of depression carry complementary information, and complete assessment requires both (Uher et al., 2012). In schizophrenia, self-assessment accuracy is mood-dependent: in 406 outpatients, those reporting very low depression overestimated their everyday functioning relative to informant reports, whereas greater depression was associated with more accurate self-assessment (Harvey et al., 2017). Depression severity correlated with less overestimation of functioning, a response bias that substantially affects the accuracy of self-reported abilities (Durand et al., 2015). A single rating—clinician or patient—therefore captures only part of whatever improvement is, or is not, occurring.

4. Discussion

4.1. Principal Findings

Across ten domains and two cross-cutting limitations, the evidence converges on five conclusions. First, improvement is multidimensional. Symptoms, functioning, cognition, subjective experience, quality of life, personal recovery, physical health and sleep are related but distinct outcome axes; positive mental health and psychopathology are correlated but independent continua (Lukat et al., 2016), and symptom and quality-of-life trajectories diverge over time (Murray et al., 2017).

Second, the patient’s perspective is a core measure, not a soft supplement. Asked what defines remission, depressed patients ranked the presence of positive mental health, a return to their usual self and a return to usual functioning above the absence of symptoms (Zimmerman et al., 2006), and patient-reported functioning and quality of life add prognostic information on relapse beyond symptom severity (IsHak et al., 2013). The recovery literature is built on personally, not clinically, defined outcomes (Leamy et al., 2011; Leonhardt et al., 2017).

Third, function-relevant constructs require function-relevant measures. Because cognition and negative symptoms predict real-world outcome but correlate weakly with positive symptoms (Green, 2016; Yang et al., 2021), and because many real-world functioning scales relate weakly to state-of-the-art functional-capacity measures (P.D. Harvey et al., 2011), instrument choice determines which improvement is visible (Figure 2B).

Fourth, context and time matter. Momentary and passive methods capture variation, context and dynamics that retrospective, clinic-based scales cannot (Myin-Germeys et al., 2018; Oudin et al., 2023).

Fifth, physical health is an outcome domain in its own right. With all-cause mortality two to three-and-a-half times higher and life expectancy shortened by 10–20 years (de Mooij et al., 2019; Gronholm et al., 2021), a measurement system that ignores cardiometabolic outcomes measures the wrong thing.

4.2. Implications for Practice, Trials and Policy

These findings do not argue for abandoning symptom scales, which remain indispensable for diagnosis, safety monitoring and regulatory comparability. They argue for surrounding them. We propose a tiered minimum outcome set (Table 2). Tier 1, intended for all routine care, including busy outpatient clinics in low- and middle-income countries, pairs a brief symptom measure with a generic disability measure, a quality-of-life or well-being measure, basic cardiometabolic indicators, an insomnia screen and a side-effect checklist—almost all freely available and self-completed in minutes. Tier 2 adds informant-rated functioning, brief cognitive assessment, negative-symptom, recovery and self-stigma measures for specialist and early-intervention services. Tier 3 adds comprehensive cognitive and functional-capacity batteries, ecological momentary assessment or passive sensing, and hard physical-health endpoints for trials.

Standardized outcome sets developed through consensus, such as the International Consortium for Health Outcomes Measurement (ICHOM) standard set for depression and anxiety, already combine symptom burden with functioning and treatment-related outcomes (Obbarius et al., 2017), and our framework is compatible with them. For trials, regulators and sponsors should consider designating at least one functioning, quality-of-life or recovery measure as a key secondary endpoint. For services, measurement-based care is more likely to be adopted when measures answer questions that clinicians and patients actually ask (Lewis et al., 2019). Where clinician time is scarce, self-report and informant measures can extend coverage at low cost, but their divergence from clinician ratings should be interpreted as information rather than error (Harvey et al., 2017; Uher et al., 2012).

4.3. Future Research

Priorities include establishing minimal clinically important differences for functioning, recovery and self-stigma measures; validating these instruments across languages and cultures, particularly in South Asia and other low-resource settings; testing whether broader outcome sets change clinical decisions and patient outcomes; and defining which digital signals qualify as valid, equitable and privacy-preserving outcome measures (Montag et al., 2021; Stone et al., 2023).

4.4. Strengths and Limitations

This review integrates, within a single framework, literatures that are usually examined separately, and it translates the synthesis into a pragmatic measurement proposal. It is, however, narrative rather than systematic: study selection may be subject to bias, and no formal risk-of-bias assessment or meta-analytic pooling was performed. Several domains rest on heterogeneous measures and samples, and much of the functional-outcome literature derives from schizophrenia-spectrum and mood disorders, limiting generalization to the full diagnostic spectrum. Each instrument discussed carries its own psychometric limitations, and the tiered outcome set is a proposal that requires empirical testing. Finally, the emphasis on highly cited work may under-represent recent methodological innovation.

5. Conclusions

Standard psychiatric rating scales answer the question, How severe are the symptoms? They do not answer the questions patients, families and health systems ask: Can this person function, work and live independently? Is cognition improving? Is the treatment tolerable? Is life worth living again? Is the person recovering on their own terms? Are they physically healthy and sleeping well? How do these things fluctuate in daily life? The evidence reviewed here shows that meaningful improvement occurs—and is routinely missed—across functional, cognitive, motivational, experiential, recovery-related, stigma-related, physical and dynamic dimensions. Broadening psychiatric measurement to include these domains is a scientific necessity, not merely a humane preference, because these domains predict the outcomes that matter most and current instruments cannot see them.