**Background:** Nursing management requires tools that accurately measure patient care demand to support staff dimensioning and resource allocation. While the Pediatric Patient Classification Instrument (PPCI) included family support indicators, no equivalent instrument existed for adult patients despite the increasing presence of family companions during adult hospitalizations. This study aimed to adapt and validate an adult patient classification instrument that emphasizes the family support network.
**Methods:** This methodological study was conducted in three phases between November 2016 and November 2018 at a general, public teaching hospital in São Paulo, Brazil. Phase 1 involved adapting the PPCI for adult patients through literature review and author authorization. Phase 2 comprised content validation by seven experts (two professors/researchers in instrument validation, three nursing management nurses, and two patient care nurses; experience range 7–21 years). Experts rated each indicator on a four-point Likert scale for relevance and clarity; the Content Validity Index (CVI) was calculated with a minimum threshold of 0.8. Agreement with the response scale organization was assessed using a dichotomous scale (>70% considered valid). Phase 3 evaluated measurement properties by applying the instrument to all patients admitted to a general adult ward over one month. Nurses who performed classifications received prior training. The minimum sample size was 110 patients (10 per item). Confirmatory factor analysis using partial least squares structural equation modeling assessed convergent validity (AVE >0.5), factor loadings (>0.5), discriminant validity (Fornell-Larcker criterion and cross-loadings), and internal consistency (composite reliability ≥0.6, Cronbach's alpha).
**Key Results:** In Phase 1, the APCI preserved the original 11 indicators across three domains: Family (2 indicators), Patient (6 indicators), and Therapeutic Procedures (3 indicators). The response scale ranged from 1 to 4 points per indicator, with total scores classifying patients as minimal (11–17 points), intermediate (18–23), high dependency (24–30), semi-intensive (31–36), or intensive (37–44). In Phase 2, all indicators achieved CVI values between 0.85 and 1.00 in a single round. Agreement with the graded response scale ranged from 71.4% to 100%. In Phase 3, 902 patients were classified; 121 were excluded due to missing data, yielding 781 valid classifications. Among these, 39 (5%) were minimal care, 133 (17%) intermediate, 375 (48%) high dependency, 187 (24%) semi-intensive, and 47 (6%) intensive. Confirmatory factor analysis showed AVE values of 0.94 (Family), 0.57 (Patient), and 0.52 (Procedures). Composite reliability was 0.97, 0.89, and 0.76 respectively. Cronbach's alpha was 0.94 (Family), 0.84 (Patient), and 0.52 (Procedures). All factor loadings exceeded 0.5 within their respective domains. The square root of AVE for each domain (Family: 0.97, Patient: 0.76, Procedures: 0.72) exceeded inter-domain correlations, satisfying the Fornell-Larcker criterion. Cross-loading analysis confirmed each indicator loaded highest on its intended domain.
**Clinical Implications:** The APCI is the first validated adult patient classification instrument that incorporates the family support network as a dimension of nursing care demand. This addresses the reality that family members are increasingly present during adult hospitalizations, as supported by the Elderly Statute, the SUS humanization policy, and regulatory requirements for companion coverage. The instrument can guide short-term decisions about staffing and care prioritization, as well as medium- and long-term planning for discharge and post-discharge support. By including family indicators, the APCI aligns with the WHO Global Action Plan for Patient Safety 2021–2030, which emphasizes engaging patients and families in safe care. The authors note that Cronbach's alpha for the Therapeutic Procedures domain (0.52) fell below the 0.6 threshold, but they considered composite reliability (0.76) more appropriate for PLS analysis. Study limitations include the exclusion of indicators for contact precautions and discharge preparation, which were suggested by experts but not incorporated to maintain fidelity to the original instrument structure.