BACKGROUND: Identifying predictors of poor postoperative outcomes is crucial for planning personalized pain treatments. The aim of this study was to examine pain outcomes using cluster analysis in N=2,678 patients from the PAIN-OUT registry at first postoperative day.
METHODS: Indicator variables of the clustering analysis assessed multiple domains, such as clinical and surgical conditions, analgesic-anesthetic variables, desire for more pain treatment and outcome variables of the International Pain Outcome Questionnaire (IPO) summarized as factor scores.
RESULTS: Two-step cluster identified the three-cluster solution as the optimal. Two empirical groups (C1 and C2) included patients with good postoperative outcomes discriminated by peripheral nerve block use, while the other cluster (C3) grouped patients with the worst outcomes, where all patients desired more pain treatment. C3 comprised about 20% of the participants, mostly lower limb, abdominal and spine procedures. The best predictors of belonging to C3 included younger age, being male, preoperative opioid use, bone and fracture reduction procedures, institution, number of comorbidities and morphine equivalents in the recovery room.
CONCLUSIONS: IPO factor scores can be used to select pain outcomes phenotypes in large clinical databases. Most of the predictors were present before the recovery period so perioperative planning should focus in the preoperative and intraoperative periods.