Using Databases to Prevent Inpatient Morbidity

Abstract

Using Databases to Prevent Inpatient Morbidity Dear Ed i tor:In creas ingly, health care data are stored in eas ily ac ces si ble elec tronic for mats. These data pro vide an at trac tive re search op tion for pre dict ing both ben e fi cial and ad verse treat ment out comes. We recently used such data to in ves ti gate the pre dic tion and pre ven tion of de lir ium. There is an ap prox i mate 2% in ci dence of de lir ium among psy chi at ric in pa tients (1), and risk-fac tor pro files for de lir ium of ten im pli cate med i ca tions with anticholinergic ef fects. Mod i fi ca tion of anticholinergic ex po sures may there fore be an ef fec tive pre ven tion for de lir ium.We ex plored the fea si bil ity of us ing an anticholinergic equiv a lency-based formula to pre dict de lir ium in adult psy chiat ric in pa tients. We hoped that the for mula could ul ti mately be in cor porated into a phar macy da ta base, thereby iden ti fy ing pa tients at in creased risk of de lir ium. Cau tion ary "flags" could then en cour age phy si cians to re eval u ate med i ca tion re gimes.Our pro ject in volved a sec ond ary anal ysis of data col lected in a re cent case-control study (2). This study in cluded 31 cases of new-on set de lir ium and 123 matched con trol sub jects. Using published data, we de vel oped a for mula to es ti mate anticholinergic ex po sure lev els for var i ous med i ca tions. We cal cu lated odds ra tios (ORs) to de ter mine whether these es ti mates pre dicted de lir ium risk. Sur pris ingly, only sub jects who received the high est level of anticholinergic unit ex po sure (AUE) from antipsychotic med i ca tions were more likely to de velop de lir ium (OR 15.42, sig nif i cance level 0.001). Lower lev els of AUE from antipsychotics were sta tis ti cally sig nif i cant at a P value ≤ 0.05 but did not strongly pre dict de lirium. There were no con sis tent cor re lations be tween es ti mates of anticholinergic ex po sure for an ti de pressant or antiparkinsonian med i ca tions and the in ci dence of de lir ium. We concluded that a pre dic tive al go rithm based on anticholinergic ex po sures cal cu lated from clin i cal data could not be used as a pre ven tive in ter ven tion.Pro ject data sources in cluded var i ous pub lished es ti mates of each med i cation's av er age rel a tive anticholinergic ef fect (ARAE) and hos pi tal chart data de scrib ing med i ca tions taken and their dos ages. Er ror sources in cluded anticholinergicity es ti mates and doc umen ta tion er rors in clin i cal re cords. Predic tive es ti mates de rived from er ror-prone data sources are vul ner a ble to misclassification bias (3). Anticholinergic ex po sure cat e go ri za tion en tails 3 com pli ca tions: er rors in re cording of type, dos age, and fre quency of med i ca tions; pos si ble pa tient non compli ance with med i ca tion; and crude es tima tions of each med i ca tion's ARAE from di verse lit er a ture sources. These es ti ma tions do not in cor po rate pharmacokinetic fac tors, such as rates of ab sorp tion, me ta...

Cerebral oximetry as a biomarker of postoperative delirium in cardiac surgery patients

“…The American Psychiatric Association recommends that delirium assessment in clinical practice is best achieved when medical diagnosis is supplemented with observational assessment tools [2]. The criterion standard in practice is the Confusion Assessment Method (CAM), which has been validated among intensive care patients with the CAM-intensive care unit (CAM-ICU) [3].…”