In this study, Ostrovsky aimed to determine if non-medical workers could use mobile technology and a risk predictive software application could predict the 30-day readmission of clients in the target group. This would be compared with the nurse manager based risk predictive system. The findings were that the about a fifth (22.7%) of the 1064 alerts that a nurse manager responded to were readmitted within 30 days ( Ostrovsky et al, .2016) .
Where elevated risk alert was generated for a patient encounter with the non-medical worker, the patient was 1.12 times more likely to admit that where classification was no elevation. In response to such an alert, if the nurse, using a structured feedback system, also assessed the patient at that encounter as elevated risk, the likelihood of admission increased to 1.25. Interesting if this likelihood dropped to 1.20 the validated unstructured nurse feedback was used to determine elevated risk. The associations between elevated risk score, either by the software or nurse, were positive and relevant as the odds ratio was greater than 1. This finding was statistically significant at the 5% border of the confidence interval was greater than 1 in all cases.
Delegate your assignment to our experts and they will do the rest.
Risk classification sub-categories were high, moderate, mild and baseline based on the predictive score derived from non-medical worker and/or nurse input. 30-day admission rates across the groups as 32%, 20%, 29% and 14% respectively for non- medical workers based input and 33 %, 22%, 20% and 14% for the nurses based ranking. The differences in the admission rate by e risk categories assigned by non-medical workers was statistically significant as reflected ANOVA result as well as between-group t tests yielding p-value < 0.01. The same applied for nurse based ranking except between the groups high (33%) and moderate risk (22%) which statistically insignificant with p-value =0.06.
Limitations of study could be in where the population, study design and statistical methods used. Selection bias existed as the only those under specific Medicare programs were included, with a Caucasian majority of 86% . The use AUC score of 0.56 barely meets the 0.5 mark relevance cut off mark.
References
Ostrovsky, A., O'Connor, L., Marshall, O., Angelo, A., Barrett, K., Majeski, E., ... & Levy, J. (2016). Predicting 30-to 120-Day Readmission Risk among Medicare Fee-for-Service Patients Using Nonmedical Workers and Mobile Technology. Perspectives in health information management , 13 (Winter).
