The aim of this study is to translate and validate the APAIS for the Portuguese oncologic population. The use of a quick and easy-to-administer instrument to evaluate anxiety is of foremost importance in the perioperative setting as anxiety leads to a more challenging anesthetic management and a worse patient experience. Our study revealed an anxiety prevalence of 61%, higher than other published oncological reports.
Education level of our sample was in line with the average elder Portuguese population (52% of the people aged 65 and over only attained primary school education, PORDATA 2018). Thus, in order to obtain a better comprehension, warranting an appropriate phrasing was crucial. Following WHO guidelines, after cross-validation of the direct translation, the final questionnaire resulted from additional cognitive interviewing.
Contrary to previous validations, all questionnaires were applied after the patient’s admission on the day before surgery, instead of immediately before the surgery. We believe this provided a less stressful environment for the patients.
Results showed high-quality psychometric properties. Scale reliability revealed a Cronbach’s alpha of 0.81, consistent with the one obtained on the original scale and further validations. In contrast, confirmatory factor analysis revealed a best fit with a three-dimensional model (anxiety about anesthesia, anxiety about surgery, desire for information), differently from the original scale. This model has already been described for the French scale validation (19). Differences might be explained by cultural and educational reasons, as well as the life-changing nature of the oncological disease and its surgical treatment. Thus inter-item correlation was moderate to high on both anxiety dimensions. A weaker correlation, however, was observed in the desire for information dimension. A paternalistic doctor-patient relationship, based on trustworthiness, may be a reason for the mismatch between reported anxiety and need for information.
Spearman’s correlation established the capacity for the Portuguese version of the APAIS to explore anxiety states. The STAI-Y1 convergent validity evaluation correlated slightly better to the APAIS than HADS-D did (Spearman’s rho 0.580 vs 0.455). Albeit the presence of convergent correlation in both scales, these results were not as disparate as anticipated. As shown in the HADS-D final scores, 18% of the studied population also suffered from depression. Although clinically different, anxiety and depression are both frequent comorbidities in oncological patients, and sometimes coexist, which may justify our observations. Indeed, results from correlation testing between APAIS and HADS-D suggested that preoperative anxiety is usually present in cancer patients suffering from depression (Fig. 1.b), an association that should be tested in future works.
For this specific population, two cutoffs determined by the ROC curves could be used. For a higher sensitivity (0.86) a cutoff of 11 would be recommended, similar to the original version. However, taking all the results into account, a cutoff of 12 is suggested in order to gain specificity (specificity 0.79),
That said, this study presents some limitations. First, the studied population is restricted to a single hospital and it only includes cancer surgery patients. As previously mentioned, cancer patients are known to have higher baseline anxiety and depression. Furthermore, many of these patients undergo multiple diagnostic and therapeutic procedures during the course of the disease, somehow modulating their perioperative-related anxiety. Secondly, several patients were not included due to low literacy, meaning that further works should be developed in order to evaluate the applicability of a verbal assessment in this group. One factor contributing to the difficulty of answering the questionnaires was their length, particularly the ones used for comparative measures, not APAIS itself. Also, convenience sampling was applied in this study. Last, and unintentionally, no ASA IV or V were sampled from the population.