Transit Time and Recovery in Prehospital Emergency Care: Meta-analysis

Background: People living in rural areas usually suffer comparatively disadvantaged emergency health care than those living in urban areas, reasons including long transit time due to geographic factors. Implementation of technology is suggested to improve the access to medical care for the rural population, such as remotely-supported ultrasound and satellite/cellular communications. Methods: Screening of eligible studies were conducted based on inclusion an exclusion criteria. A comprehensive search was conducted by using following database: EMBASE, Medline, Cochrane library and Scopus. Quality assessment tool for observational cohort and cross-sectional study is used for assessing the risk of bias. Transit time was mostly classified into two-stage time symptom onset-balloon time and door-balloon time. In symptom onset-balloon time, we divided into short time group and long time group in our study. The time group were defined based on the median or mean transit time among patients and we also used same way to set up time groups among door-balloon time. The collected data were used for quantitative analysis, they were inputted into Review Manager Software (v5.3) to produce summary results. Results: Ten studies representing 71099 patients were included in the meta-analysis. All studies were retrospective and prospective observational studies and RCTs in which patients experienced ST-elevation myocardial infarction (STEMI) and were treated with percutaneous coronary intervention (PCI). Combined in meta-analysis the odds ratio for mortality in onset-balloon and door-balloon time was 0.82 (CI 0.70, 0.96) and 0.62 (CI 0.53, 0.74) respectively. The forest plot of both onset-balloon time and door-balloon time showed a moderate heterogeneity, with I2 = 31%, P =0.18 and I2 = 67%, P <0.05 respectively. Conclusion: Results from Meta-analysis report less mortality in shorter transit time than

3 that in longer transit time. The demographic characteristics and the sample size of the difference between the long time and short time groups may be the reasons for heterogeneity. The result of the review may potentially support the application of remotely-supported ultrasound and satellite/cellular in prehospital emergency for better health outcome in long transit time.

Background
People in different geographic areas experiences discrepant in health services, which includes the density of paramedics, distances to treatment centers, and type of transportation to hospitals [1]. According to Horne et al 2000, medical knowledge acquisition, demographic, symptom relevant, and clinical factors are likely to affect the prehospital transit time [2]. In many countries, it is common that people living in rural areas suffer comparatively disadvantaged emergency health care than those living in urban areas, most have experienced long transit time due to geographic factors [3].
Delays in receiving and reaching healthcare may lead to serious health issues [4]. But implementation of technology is suggested to provide a fair medical chance for the rural population, such as remotely-supported ultrasound and satellite/cellular communications.
These implementations involve paramedics that are remotely-supported and guided by hospital-based experts via telecom-systems, which may lead to early diagnosis and provide pre-hospital treatment as quickly as possible before patients arriving at the hospital [5].
As for many time critical diseases, it is necessary to obtain treatment as quickly as possible. In this context, acute myocardial infarction (AMI) serves as an example because it is a widespread disease all over the world, and it is one of the most serious type of coronary heart disease with a high mortality rate [4]. Minimizing the time of reperfusion therapy of AMI patients may be an effective method to control the death rate [6]. 4 However, one of the main reasons influencing on therapeutic time may be distance. As so far, AMI mortality in rural areas is higher than that in urban areas because patients living in remote areas cannot receive suitable medical treatment in time [7]. Some individual factors also influence the timeliness of reception of AMI therapy, such as medically underserved setting and absence of medical knowledge [8]. Several studies have been conducted to identify the possibility of reducing mortality through reducing prehospital transit time. But a systematic review of the relationship between transit time and the mortality of acute myocardial infarction has not been conducted.

Aim
The aim of this study is to assess the relevant transit time and recovery of acute myocardial infarction in prehospital emergency care.

Literature search strategy
We have systematically searched the following databases: EMBASE, Medline, Cochrane library and Scopus for all type of observational studies, qualitative studies and randomized controlled trials (RCTs). A combination of MeSH and keywords were involved in search strategy, including "transit time", "acute myocardial infarction patients", "emergency care", "remote". More details of the search processes were listed in Appendix A. All included studies were managed by Refworks. Thirty-two studies were identified via screening article titles and abstracts for eligibility. Full-text articles were obtained to examine eligibility for data extraction. Any limitations in the studies were discussed among the three authors (FX, PW, WSFC). One author (FX) scanned all records first and then discussed any disagreements with other two reviewers (PW, WSFC).

Eligibility criteria 5
Eligible studies were screened based on inclusion and exclusion criteria ( Table 1). The studies were considered as eligible, if a) the study types were either an observational study, randomized controlled trial or a qualitative study. b) studies which report adult participants and acute myocardial infarction patients. c) trials were conducted in emergency department, hospital, prehospital setting and clinical setting. d) the outcomes were focused on measurable mortality. e) the article is from 1990 to present day and the article language is in English.

Assessment of risk bias
The nine selected studies were evaluated by a single researcher (FX) using Quality assessment tool for observational cohort and cross-sectional study [9]. One study was assessed by ROBINS-I (Risk Of Bias In Non-randomized Studies -of Interventions) [10]. The primary researcher (FX)then discussed the disagreements with another two reviewers (PW, WSFC). The risk of bias for observational studies was assessed by the following aspects: 1) Research question: Did the authors demonstrate their aim in research? Whether readers can get an idea of clear research goal easily. 2) Study population: Was selection of exposed and non-exposed group from the same population? Did the authors make inclusion and exclusion criteria based on demographics, medical history, and time period? Did the authors select participants according to inclusion criteria? If there are fewer than 50% of eligible people participated in the study, it should be considered an increased risk of bias. 3) Did the authors explain why select the number of participants to analyze? Whether they have recorded or discussed the statistical power of the study. 4) Whether exposure(s) of interest measured before the outcome(s) being measured, the step can determine whether or not an exposure causes an outcome. 5) Did the study allow adequate time to observe an outcome except for cross-sectional analyses? 6) For some exposures that are defined as a range, were different categories of exposure evaluated? 6 Blind is unnecessary in some study types. 7) Were the exposures clearly defined and accurately measured? Were the measurement tools reliable? Were the results valid and measured objectively? 8) Repeated exposure assessment: were exposures are measured many times? The step can increase the accuracy of the results. 9) Can we be confident in the assessment of exposure? Was the outcome clearly defined and accurately measured? Several studies measured data from multiple time periods, therefore, these studies were pooled into two groups on the basis of median or mean. The population with less transit time than the median or mean is regarded as short time group, otherwise, it is the long time group.

Data extraction
A data extraction form based on the Cochrane handbook for systematic review of intervention [11] was used by FX to collect information from included studies. The form contained: study ID (last name of the first author and publication date), study country, type of the study, transit time (min), the number of study population, number analyzed, mortality, age. We have reviewed articles in the publicly available journal, so ethics approval was not required.

Meta-analysis
The collected data were used for quantitative analysis, they were inputted into Review Manager Software (v5.3) to produce summary results through comparing a dichotomous outcome. A forest plot was implemented through pooling data and comparing number of deaths in each group among individual studies. We used random effects models to calculate summary odds ratio and confidence interval of the summary estimate.

Search results
The detailed process of article screening for the systematic review is shown in Fig. 1. A total number of 1020 publication were obtained. The remaining records were 664 after removing duplicates. Then 32 references were assessed after screening the title and 8 abstract. Twenty-two records were excluded with the following reasons: eight studies lacked transit time data; nine did not report mortality; three did not represent acute myocardial infarction and one trial did not report patients in an emergency care setting.
Another four articles were excluded because of the content did not show specific figures on mortality and one recorded insufficient mortality. Overall, ten studies were included for meta-analysis.  Table 2  Overall, data from six studies [14,16,17,18,26,27] were divided into short/long time groups directly in original articles. Moreover, another four studies [15,19,20,21] with multiple time periods were classified into short/long time group via pooling data for metaanalysis based on the median or mean time.

Symptom onset-balloon time
Relative odds ratio and corresponding 95% confidence interval are showed in Fig. 4. In eight studies with 39577 patients, six trials favour short time group and two studies favour long time group. Random effects meta-analysis of the point estimate was 0.82 (CI 0.70, 0.96). Heterogeneity between study results was evaluated via examination of the forest plots and quantified by using I 2 statistic [22]. Heterogeneity in symptom onsetballoon time was moderate among studies (I 2 = 31%, P = 0.18). Population in short time group were non-shock patients, whereas AMI patients in long time group suffered shock. It is possible that shock patients may experience longer emergency care time on the spot [23]. The bias introduced by follow-up was not considered as an assessment of mortality was conducted within a short-term (30-day).

Heterogeneity
Both of the estimates of I 2 assessed in the meta-analysis would be found to be moderate. I 2 = 31% is for symptom onset-balloon time; I 2 = 67% is for door-balloon time. The value of I 2 was calculated from both the scope of the study and the between study variance [24].
The study including different clinical characteristics, research methods and statistical methods could contribute to the value of I 2 . Thus, the summary estimation cross the average population have various features, this should be carefully used and discussed.
Deficiencies often existed in studies. The studies involved in this review include different demographic characteristics. Participants in two studies [15,16] were from Asia, one study was selected from the Middle East [14]. Four studies [17,19,20,21] were conducted in the United States and three studies [18,26,27] were from Europe. It is possible that different races have different prognosis for the same disease, which can introduce heterogeneity. Additionally, screening trials include retrospective and prospective observational studies, recall bias is less likely to be avoided in retrospective studies and survivor bias could affect clinical outcomes [16]. It is difficult that to assess the time of onset of symptom sometimes so that it would result in errors in door-balloon time [17]. Different trials that were grouped with different standards and the results could be influenced by variables. For example, in Brodie 2003, patients with shock experienced longer time-to-reperfusion and mortality are also different from patients without shock. A few studies [15,18] showed the opposite result among mortality with other studies, which was caused possibly by insufficient sample size.
In addition to characteristics mentioned above, a number of others vary in different studies, which can lead to heterogeneity. These situations include differences in inclusion and exclusion criteria, the instruments used and the medical levels of paramedical personnel [25].

Limitation
Not all studies had reported median or mean of symptom onset-balloon time and doorballoon time and corresponding mortality directly. Some studies had only recorded the approximate range of the transfer time and the mortality in the time range. For comparison in meta-analysis, we pooled the data of periods of time in these studies according to the median or mean transit time that was mentioned in the context. The time for patients to transfer to hospital that was less than the median/mean time was regarded as "short time group", otherwise, the group was judged as "long time group". This kind of grouping may have an impact on the analysis result. Besides, participation in these studies from observational study was not randomised. As a result, clinical characteristics of patients would introduce confounders into differences of mortality of different periods of time so that confounder bias caused by underlying disease is likely to be ineluctable.
We limited the search language to English, which would contribute to linguistic bias.
Meanwhile, limit number of studies may not be representative.

Comparison with previous studies
This review showed patients with STEMI are more likely to survival when experiencing shorter transit time, in spite of other variables. Needleman et al 2011 reported that the availability of skilled staff, staffing levels and the number of paramedics are potential factors to contribute to mortality [28]. Kulkarni et al 2013 showed a similar outcome. 30day mortality rate of patients with acute myocardial infarction in the low density of 13 cardiovascular disease experts was higher than those in high-density areas, which suggested that the outcome of patients with AMI might be influenced by the availability of cardiology specialists in regional care systems [29]. Other factors, including patients' own condition (sleep deprivation and fatigue [30], genetic factor) and distance from home to hospital [31], can result in increased risk of death for AMI.
As for the factors affecting the transit time, availability of cardiologists and cardiac catheterization laboratory staff are considered to be associated with that [32]. The increase in the time interval between getting the electrocardiogram (EGG) and arriving at the catheterization laboratory would lead to the increase in almost all door-balloon time [32]. Delay to make decision to seek care and delay to receive care may be the reason for the increase in the overall transit time [2,4].

Conclusion
The meta-analysis for included studies report less mortality in shorter transit time than that in longer transit time. The demographic characteristics and the sample size difference between the long time and short time groups may be the reasons for heterogeneity. Multiple factors, such as availability of skilled staff, the density of paramedics and the time between electrocardiogram and catheterization laboratory could lead to differences in mortality and time intervals. Furthermore, data in a few studies were pooled roughly according to the time of median or mean in studies, which may have an impact on the results of the analysis. The result of the review may potentially support the application of remotely-supported ultrasound and satellite/cellular in prehospital emergency for better health outcome in long transit time.

Ethics approval and consent to participate
Not applicable

Consent for publication
Not applicable

Availability of data and material
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing Interests
Authors have no conflict of interest to declare.

Funding
No funding was obtained for this study.

Authors' contribution
Any limitations in the studies were discussed among the three authors (XF, PW, WSFC). XF scanned all records first and then discussed any disagreements with other two reviewers (PW, WSFC). The nine selected studies were evaluated by XF and then discussed the disagreements with another two reviewers (PW, WSFC).

Acknowledgements
We would like to thank Mr. Robert Polson, University of Highlands and Islands for his assistance in literature search strategy.