Cost of Hospitalization in Children Who Develop Perioperative Respiratory Event

Overview

Mortality related with cardiac arrest in anesthetized children has diminished over several decades from 2.9 per 10000 anesthesias in 1961 to 0.21 per 10000 anesthesias in 2007.(1) Even though the mortality rate related to pediatric anesthesia is much lower than before, respiratory complications related with peri-operative cardiac arrest are as high as 27% according to the Pediatric Peri-operative Cardiac Arrest (POCA) Registry.(2).Therefore, peri-operative respiratory event (PRE) in pediatric anesthesia such as laryngospasm, stridor, bronchospasm, desaturation and reintubation are crucial. Stridor and reintubation occur after the children are extubated, mostly in the PACU period. Laryngospasm, stridor, bronchospasm and wheezing can lead to desaturation and the need for reintubation. Those PRE, especially peri-operative desaturation, can prolong PACU stay especially if PRE develops in the PACU.(8,9) PRE occurring during the intraoperative period can also prolong PACU stay if children are observed at PACU and not transferred directly to the intensive care unit (ICU). Some children require oxygen therapy in the PACU and continue at the ward. Some need endotracheal tube intubation with spontaneous breathing or are placed on mechanical ventilator. Thai AIMS (10) reported that desaturation at PACU was associated with re-intubation, prolonged mechanical ventilation and unplanned ICU admission. Oxygen supplement need, prolonged mechanical ventilation or unplanned ICU admission can produce extra days of hospitalization or increase the cost of hospitalization from extra-cost payment eg; oxygen therapy, mechanical ventilator, cost of ICU stayed. Furthermore, the short-term sequelae regarding cost of hospitalization of children who develop PRE has never been evaluated or compared with the cost in children who do not develop PRE. Higher cost of hospitalization after occurrence of PRE in anesthetized children will have an impact to the hospital policy maker. Minimizing PRE can save on the cost of hospital care to the public hospital or other health sectors. Therefore, we would like to compare days of hospitalization and cost differences of hospitalization between children who develop PRE and children who do not develop PRE at a tertiary care hospital in southern Thailand

1. Study design A prospective observational study 2. Study setting Children aged \< 15 years who received general anesthesia in Songklanagarind Hospital between November 2012 and December 2013 3. Study samples Inclusion criteria Children aged \< 15 years who received general anesthesia and developed PRE or who had no PRE in Songklanagarind Hospital between November 2012 and December 2013. Exclusion criteria 1. ASA classification 4 or 5 2. Preoperative SpO2 \< 95% at room air 3. Required preoperative endotracheal tube intubation (ETT) or mechanical ventilation 4. Had congenital cyanotic heart disease 4\. Study size Sample size calculation The sample size calculation is based on a comparison between groups of the outcomes of days and cost of hospitalization between PRE and non-PRE group 5. Sampling technique Purposive identification of children aged \< 15 years old who receive general anesthesia and develop Peri-operative respiratory event (PRE) between November 2012 and December 2013.was selected for PRE or exposed group. Peri-operative respiratory event (PRE) is defined as any respiratory event eg; laryngospasm, stridor, wheezing, bronchospasm, reintubation, with or without desaturation, which occurs during the intraoperative or PACU period. For non-PRE or non-exposed group, the matching with exposed group on the same demographic data will be performed as below. * Matching on the same year of anesthesia * Matching on the same type of surgery * Matching on the same type of anesthesia After matching the same type of surgery and anesthesia with exposed group, non-exposed group will be randomly selected by simple random sampling technique if there are more than 4 suitable non-exposed subjects available. The ratio of exposed group and non-exposed group will be 1 to 4. 6 Outcome variables The outcome variables are divided into days of hospitalization and cost of hospitalization. Days of hospitalization Days of hospitalization is the days that the children stay in the hospital until discharge home Days of hospitalization will be compared between PRE group and non-PRE group Cost of hospitalization Cost of hospitalization is the total cost of hospitalization including direct and indirect costs (25,26) * Direct cost are the hospital charge multiplied by cost-to-charge ratio (CCR) * Hospital charges are all the medical cost including medication, procedure related with anesthesia or surgery, material cost related with procedure, extra-cost from complication postoperatively: ICU stay, cost of ventilator or oxygen therapy, accommodation and meals charge * Indirect costs are the cost including transportation (T), productivity loss eg; cost of work leave (L)/day Therefore, cost of hospitalization * (hospital charge\* CCR) + T + (L\* days of hospitalization) According to the payer perspective, the CCR is equal to 1 Cost of hospitalization will be compared between PRE group and non-PRE group. The total cost of hospitalization from each group is a transfer payment not a real payment because 80% of the hospital charge is subsidized with the insurance coverage by government eg; universal coverage (UC), social security, government corporation or Comptroller General Department, which cannot reflect the societal cost. Moreover, we do not consider the cost of hospitalization following the disease related groups (DRG) payment in this project. 7\. Independent variables Patient related factors 1. Age 2. Sex 3. Weight 4. Height 5. ASA classification 6. History of upper respiratory tract infection 7. History of parental smoking 8. History of snoring: yes vs no 9. Postoperative complications - Related to surgery \- Related to anesthesia * Related to anesthesia and surgery Anesthetic related factors 1\. Type of surgery 2. Choice of general anesthesia 3. Technique of general anesthesia 4. Induction agent 5. Intubation agent 6. Inhalation agent used 7. Gas mixed with oxygen used 8. Type of narcotic 9. Duration of anesthesia 10. Period of PACU stayed 11. Developed any respiratory events : Yes vs no 12. Peri-operative respiratory event 13. The period of respiratory events developed : intraoperative vs PACU 14. The lowest oxygen saturation of respiratory events developed 15. Postoperative on oxygen therapy 16. Postoperative on mechanical ventilation 8\. Data processing and analysis Descriptive component Analysis will be performed with the R program version 2.14.1. Descriptive statistics will be computed for all variables and include frequency, proportion, mean + SD and median (range). Predictor variables will be categorical data either originally or if continuous, categorized by selected suitable cut points. Analytic component The days of hospitalization will be compared between PRE and non-PRE group using simple linear regression model. The other explanatory variables that may be related to prolonged hospitalization will be analyzed using multiple linear regression models if showing some evidence of differing across outcome categories in univariate analysis (p ≤ 0.2). The magnitudes and precision of associations will be indicated by adjusted coefficient and their 95% confidence intervals. Total cost of hospitalization will be compared between PRE and non-PRE group using simple linear regression model. Cost of hospitalization will be calculated both using CCR 0.4 (hospital perspective) and 1 (payer perspective). The other explanatory variables that may be related to higher cost of hospitalization will be analyzed using multiple linear regression models if showing some evidence of differing across outcome categories in univariate analysis (p ≤ 0.2). The magnitudes and precision of associations will be indicated by adjusted coefficient and their 95% confidence intervals. For both days and cost of hospitalization, if the distribution is not suitable for linear regression or the assumption of linear regression is not met, the other alternative analysis methods may be preferred eg; logarithmic transform, quantile regression or Cox regression. In multivariate modeling techniques, association with outcome will be considered significant if the likelihood ratio p-values are ≤ 0.05.