Validity and reliability of a targeted nutrition-risk screening tool for mental health services.

BACKGROUND

Existing nutrition-risk screening tools focus primarily on undernutrition and do not capture the broader risks experienced in mental health settings.

OBJECTIVES

This study aimed to evaluate the measurement properties of a targeted nutrition-risk screening tool developed for mental health and alcohol and other drug (AOD) services.

METHODS

A multicenter validation study was conducted in 12 public and private mental health and AOD services in Australia. People receiving care in inpatient, community, forensic, and residential mental health settings and 1 AOD setting were recruited through convenience sampling. The NutriMental Screener was administered at 3 time points to assess interrater reliability (2 healthcare professionals within 24 h) and test-retest reliability (repeat administration ∼1 wk later). Concurrent criterion validity was examined through comparison with a dietitian-completed nutrition assessment incorporating established measures of constipation, malnutrition risk, eating disorder symptomatology, and food insecurity. Reliability was evaluated using Cohen's κ coefficients, and sensitivity and specificity analyses were conducted for domain-specific validity. Feasibility was assessed using completion time and referral outcomes. Findings were reviewed through a structured expert consultation process to refine tool instructions, scoring response layers, and referral processes and frameworks.

RESULTS

A total of 207 participants completed the initial screening, with 198 completing all assessments. Test-retest and interrater reliability demonstrated moderate to almost perfect agreement across items (κ = 0.421-0.939, all P < 0.001). Domain-specific concurrent validity analyses showed clinically acceptable sensitivity-specificity trade-offs across constructs. Median completion time was ∼5 min. Screening generated a dietitian referral in 40.1% of cases and prompted clinical team discussion in a further 34.8% of cases. Expert-informed refinement resulted in minor modifications to wording and referral framework, producing a clinically ready tool.

CONCLUSIONS

The NutriMental Screener demonstrated acceptable reliability, clinically meaningful validity, and high feasibility. This tool provides an alternative to undernutrition-focused screening tools currently employed in mental health settings and preliminary evidence for use in AOD services.