Validity and reliability of a pediatric patient classification instrument1

OBJECTIVES: to assess the construct validity and reliability of the Pediatric Patient Classification Instrument. METHODS: correlation study developed at a teaching hospital. The classification involved 227 patients, using the pediatric patient classification instrument. The construct validity was assessed through the factor analysis approach and reliability through internal consistency. RESULTS: the Exploratory Factor Analysis identified three constructs with 67.5% of variance explanation and, in the reliability assessment, the following Cronbach's alpha coefficients were found: 0.92 for the instrument as a whole; 0.88 for the Patient domain; 0.81 for the Family domain; 0.44 for the Therapeutic procedures domain. CONCLUSIONS: the instrument evidenced its construct validity and reliability, and these analyses indicate the feasibility of the instrument. The validation of the Pediatric Patient Classification Instrument still represents a challenge, due to its relevance for a closer look at pediatric nursing care and management. Further research should be considered to explore its dimensionality and content validity.


Introduction
The use of patient classification instrument permits characterization inpatient units, estimating the nursing workload, supporting staff dimensioning, identifying changes in patients' care needs, promoting improvements in team competency and involvement, besides being an objective and practical method to obtain information and statistical data (1)(2)(3) .
In daily practice, it can be observed that patients are classified intuitively through task division, which does not always reflect their care needs. A changed perspective, from the number of tasks that are to be performed to care planning focused on the patients' needs, can expand the possibilities of nursing's health promotion activities and also improve the satisfaction and involvement with the work outcomes. In that sense, it is important to use specific instruments for each clientele.

The Pediatric Patient Classification Instrument
(PPCI) (4) permits classifying pediatric patients in five care categories: Minimal, Intermediary, High dependence, Semi-intensive and Intensive (5) . The factor evaluation instrument consists of 11  and with regard to the equivalence, measured by the agreement between two evaluators' measures when the instrument is applied at the same time (6) .
In the development process of the PPCI, the content validity analysis by experts was performed by means of the Delphi technique and the interrater reliability was verified (4) . As the PPCI is used to support management decisions at pediatric units, its validation process cannot be impervious and demands successive studies to monitor its validity and reliability.
This study intends to assess the construct validity and reliability of the PPCI.

Methods
This correlation study was undertaken at a pediatric unit of a teaching hospital in the interior of the State of São Paulo, which consists of 58 inpatient beds and ten intensive care beds.
Approval for the study was obtained from the distribution of Chi squared and, with a view to the appropriateness of the factor analysis method, the null hypothesis about the identity of the correlation matrix should be rejected, that is, the significance of Bartlett's sphericity test should be inferior to 0.05 (6)(7) .
The construct validity analysis according to the EFA is guaranteed when the total variance explanation represents more than 60% and, according to the Kaiser criterion, factors should be extracted with an Eigenvalue superior to one in order to identify the construct domains (6)(7) .
The commonalities represent the extent of the variance explanation of each indicator based on the factors identified. For the indicator to be representative, its commonality index should be superior to 0.6 (6)(7) . The factor loadings represent the correlation between the indicator and the extracted factor. Thus, coefficients between 0.30 and 0.40 are considered minimal; factor loadings between 0.50 and 0.70 are significant and loadings superior to 0.70 indicate a well-defined structure, which is the target of any factor analysis (6)(7) .
The residues represent the aspects of the variance the indicators do not explain, and a percentage of more than 50% of residues superior to 0.05 is not desirable (6)(7) .
The reliability of the PPCI was assessed by means of the internal consistency with three parameters: itemtotal correlations, inter-item correlations and Cronbach's alpha (α). For the PPCI to be considered reliable, the item-total correlation should be superior to 0.50; the inter-item correlations should be super to 0.30 and Cronbach's alpha superior to 0.70 (6,8) .

Results
The sample characteristics in terms of sex, age and reason for hospitalization are displayed in Table 1.
The sample mostly included patients between one and six years of age, male, predominantly hospitalized due to surgical procedures or respiratory conditions.    For the internal consistency assessment of the PPCI, the inter-item and item-total correlation coefficients are shown in Table 4.

Discussion
The hours for intermediary care, do not reflect the actual care needs of pediatric patients under six years of age (5,9) .
Bartlett's sphericity test indicated that the analyzed data adjust to the EFA and the sample adequacy test, with KMO coefficients that are considered very good for nine indicators, average for Support network and reasonable for Drug therapy, indicating that the EFA results can be generalized and that the variance proportion of the PPCI indicators share a construct (6)(7) .
Based on the EFA, it was verified that the PPCI covers three pediatric nursing care domains: family, patient and therapeutic procedures and, as a factor assessment instrument, its validity does not relate to the number of indicators or situations it covers, but to its concept as a whole, as each indicator of the instrument represents a list of potential care needs (10)(11) .
Each instrument domain does not represent a sum of individual care tasks or procedures, but nursing values based on the notion that the patient needs are multidimensional and depend on the complex objective and subjective interactions (12) .
According to the established criteria, the extraction of three domains represents a care model centered on the child and his/her family, whose care approach presupposes the consideration of the domains that result in the child's health condition: the sick biological body; the child's mental, spiritual and social dimensions; and the family, seen holistically, as responsible for the healthcare shared with the professionals during the hospitalization (10)(11)13) .
The three resulting domains underline the importance of the accomplishment of pediatric nursing interventions inextricably from health promotion, disease prevention, health recovery and rehabilitation, in which it is fundamental to take into account the child and family's singularities with a view to qualified and humanized healthcare (13) .
The presence of more than 50% of residues with coefficients superior to 0.05 and the communality coefficients for the indicators Physiological controls assessment, Cutaneous and mucosal integrity and Eliminations suggested that these indicators could not be considered representative in their respective constructs and, although the exclusion of these indicators may be considered in the literature (6)(7) , developing new studies with interventions in the content of these indicators seems to be more appropriate to improve the clarity of the instrument contents.
The residues represent the aspects of the variance the indicators do not explain (7) . It would be desirable for the residue counts with coefficients superior to 0.05 to be present in less than 50% of the data, which reveals the need for research about the clarity of the instrument contents.  The analyses indicate the feasibility of patient classification through the PPCI, but suggest further research to confirm the three domains identified in the EFA, as well as to review the content validity of the instrument to investigate whether clarity, pertinence or relevance problems caused low factor loadings or the presence of residues superior to 50%.

Conclusion
The construct validity of the PPCI can be proven by the variance explanation superior to 60% in the three domains: Family, Patient and Therapeutic procedures, as well as the factor loadings superior to 0.30 and appropriate coefficients for the other indices that were The validation of the PPCI is a pediatric nursing management resource in attempts to balance the care demand and supply. In addition, the application of the instrument drives clinical nursing assessment towards care delivery that is not only focused on the disease, tasks and therapeutic procedures, but also inspires the assessment of family members and patients, looking at their care needs, and can recover a reference to the range of nursing work.