Background & Goal
Preoperative risk & postoperative outcomes among the extreme elderly are both extensively
debated issues. We aim to determine which conditions predict poor outcomes in this population.
Materials
Prospective observational study. We recruited all patients of 85 years and above admitted for any
scheduled surgical procedure (excluding daycases)
between 20112015.
Apart from demographic data, we recorded preoperative comobidities, functional reserve, nutrition
and congnitive status (mininutritional
assessment, minimental
state examination), polypharmacy
(>7 drugs), dependency (Katz index) and frialty with both Fried´s phenotype (FP) and gait speed, a
surrogate for frailty. The complexity of the procedure (COP) was graded from 13
according to the
simplified J. Hopkins´ criteria.
We studied 4 indicators of poor outcome: the primary endpoint was 30day
mortality. We also
explored morbidity, prolonged hospital stay (PHS) and escalation of care in living conditions (ECLC).
After a bivariate
analysis with X2, Mantel & Hansel´s and Fischer´s test, we used logistic regression
(LR) to develop a model for each outcome variable.
Results and Discussion
127 patients were included: median age 87 (8596).
Overall 30day
mortality was 8%; LR suggests 3
statistically significant predictors: malnutrition (OR/95%CI: 17.6/3110),
COP 3 (25.8/3226),
history
of osteoporosis/pathological fractures (21.8/3176).
Predictors for morbidity included ischemic cardiopathy (3.9/111),
COP 3 (3.6/29),
and a nonfrail
FP (0,3/0.10.8).
The model for PHS also comprised nonfrail
FP (0.3/0.10.8)
along with COP 3 (3.4/19)
and
diabetes (3.3/18).
Finally, the variables independently related to ECLC were: slow gait (2.5/16),
COP 3 (3.2/17)
and
hypertension (3/19).
Interestingly, large surgery (COP 3) was a factor present in all 4 models, and carried a high
predictive weight. Additionally, every model had at least one geriatric variable such as malnutrition &
frialty. The latter, or its surrogate indicated poor outcomes, whereas nonfrail
was a protective factor
(OR< 1). Limitations: we had overestimated the mortality rate while calculating the sample size,
therefore, the precision of the model for mortality is lower, with broad 95%CIs.
Conclusions
The COP and geriatric variables tend to be overlooked in regular preoperative assessements, yet
they are particularly relevant in the extreme elderly. Along with comorbidities these variables should
be studied to help decision making.

Preoperative geriatric conditions as predictors of surgical outcomes in fourth-age patients