A Series Of Study In ED Overcrowding In China

[Background]Emergency department overcrowding (EDO) is a severe worldwide concern and is associated with significant negative outcomes, including unnecessary deaths. Yet the study of EDO is hindered by lack of widely accepted evaluation system.[Objective]To establish the objective EDO measuring method by observing emergency patient flow and EDO varying pattern as well as evaluating the variables affecting EDO. Aim to establish a practicable and balanced EDO assessment system suitable for domestic operation in Chinese ED setting.[Methods]The studies on emergency department overcrowding were systemically reviewed. The emergency patient flow and the EDO varying pattern during2008and2010in a2000-bed academic teaching hospital were observed prospectively. The rhythms of24h and the holiday effect as well as the policy effect on EDO were observed. The factors that might affect EDO were evaluated by multiple linear regression analysis. The National Emergency Department Overcrowding Scale (NEDOCS) was modified. The database of factors which might affect EDO was founded and the multidimensional ED Overcrowding Scale (MEDOS) items were screened. Consistency check among the Visual Analogue Scale (VAS), NEDOCS and MEDOS was performed.[Results]PARTIThe studies and papers about EDO are steadily increasing in recent years. The number of papers on this issue is8during1974and1988. The number elevated to325during1989and2002. Yet the number has climbed up to1254dramatically (which was79%of all the papers on this issue) during2003and2012. The number of countries which have reported ED overcrowding phenomenon has also been on the increase with the number of48till Feb2012. Among all the papers on this issue,86.5%were single institution studies. There has been no multi-center, randomized control clinical trial on this issue till now. The bottleneck of EDO research is the lack of the widely accepted EDO objective measuring technique.PART ⅡDistinctive patient flow patterns on workdays was observed with the patients volume peak flow in20:00-22:00, low ebb in4:00-6:00, while overcrowding score was2hours lag behind. Emergency patient flow significantly increased at weekend and long holiday with a bimodal pattern effect with a peak flow at10am to12pm and8pm to10pm. ER patient volume is affected by policy factors, but the changes only apply to non-critical patients while the number of critical patients remains consistent. Multivariate regression analysis showed that a period of time (2hours) emergency patient number (B=0.027, P=0.001), emergency bed occupancy rate (B=5.25, P<0.001) correlated with the next period of time ER overcrowding significantly.PART ⅢThis part was composed of four mainly results. The first one was focused on VAS evaluation. It was shown that VAS-p value evaluated by physicians was significantly lower than VAS-n value evaluated by nurses (6.49±1.82vs7.12±1.78, P<0.001). The reliability analysis (Kappa test) showed that Kappa value is as low as0.112,(P<0.001) which meant the consistency between VAS-p and VAS-n is weak. VAS-m (average value of VAS-p and VAS-n) was adopted while comparing different evaluation system.The second one was focused on NEDOCS evaluation. It is shown that the NEDOCS value correlated well with VAS-m value (r=0.714, P<0.001). The paired sample T test showed that NEDOCS significantly higher than VAS (VAS-mx20),155.5±36.4vs136±33.4, t=17.26, P<0.001.It was found that11.4%of the evaluation value of NEDOCS higher than200, the upper limit of the reference range (0-200).The reason of this phenomenon is that the Chinese emergency system is much different than that of the US. Then multiple linear regression was performed with VAS-m as outcome variable, Pbed/Bt, Xn, Wtime and ABI as independent variable. The modified NEDOCS (NEDOCSBJ) model was thus established: NEDOCSBJ=83.563x (Pbed/Bt)+7.201×(Xn)+0.116×Wtime+0.302×ABI+2.835The third one was focused on establishing multidimensional emergency department overcrowding scale (MEDOS) evaluation system through screening the variables data for scale items which include nine objective items and three subjective items. It was shown that the average value of MEDOS was25.4±5.8(n=552,5～39). The correlation analysis showed that MEDOS correlates with VAS-m (r=0.664, P<0.001) and NEDOCS (r=0.939, P<0.001). The Split-half reliability testing showed that Split-half coefficient was0.817(P<0.001) which means the high internal consistency of MEDOS evaluation system.The fourth one was focused on consistency check among VAS-m, NEDOCSBJ and MEDOS by using Bland-Altman Plot. None pairwise substitutability was found between nether VAS-m and NEDOCSBJ, or NEDOCSBJ and MEDOS, or VAS-m and MEDOS. The methods with moderately consistency included VAS-m and NEDOCSBJ, NEDOCSBJ and MEDOS. Yet the former could not be well accepted in clinical practice since the95%confidence interval of Bland-Altman plot is rather low. Intra-class correlation coefficients (ICC) of NEDOCSBJ and MEDOS, as well as Bland-Altman plot (95%limits of agreement:-6.74,30.66), was fairly acceptable, which meant the consistency between the two methods. Difference mean of the two methods was11.95which may be due to systemic bias and could also be acceptable in clinical practice.[Conclusions]The systemic review on EDO studies showed that the standardized measurement of EDO has become the bottleneck of EDO study. It is very important and urgent for ED management and staff to establish an objective and effective EDO evaluation system. It has been proved that the National Emergency Department Overcrowding Study (NEDOCS) was inapplicable in domestic emergency system due to the differences practice pattern between China and US. It seems that modified NEDOCS (NEDOCSBJ) model has potential value for EDO evaluation, yet it is less practical in emergency environment since it is difficulties to obtain the primary data, which need advance patient information track system. MEDOS seems to a promising EDO evaluation method not only because it is intuitive, ready to evaluate, but also because the availability and good proportionality and balance of the data.