CARDIOVASCULAR JOURNAL OF AFRICA • Volume 33, No 5, September/October 2022 262 AFRICA or lower (cold spots) incidence than expected, based on the average incidence in all clusters, but also surrounded by outlier clusters. In all analyses, we present clusters with 90, 95 and 99% confidence interval. Finally, proximity analysis was used to determine the projected driving time from each OHCA location to the closest PCI-capable facility. ArcGIS calculates drive times using predicted typical traffic trends. Typical traffic trends for each road are determined within ArcGIS by averaging a week’s historic real-time driving speeds in five-minute intervals. Each hexagon was then assigned the resulting driving time for visualisation. All geospatial analyses were performed using ArcGIS Pro 2.6.3 (Esri, California, United States) and ArcGIS Online (Esri, California, United States). Results For the study period, a total of 929 patients with OHCA received an EMS response in the city of Cape Town metropole, corresponding to an annual prevalence of 23.2 per 100 000 persons. Detailed clinical data are reported elsewhere.24 The median [interquartile range (IQR)] age of the sample was 63 (26) years and most OHCAs occurred in private compared to public locations (n = 99; 10.7% vs n = 740; 79.7%). Only 1.3% (n = 13) of patients had ROSC and 97.8% (n = 909) of patients were declared dead on the scene.24 Figs 1–3 outline the temporal distribution of OHCAaccording to month of the year (Fig. 1), day of the week (Fig. 2) and time of day (Fig. 3). The distribution of OHCA incidence was not explained by month of the year [χ2 (11) = 17.967, p = 0.08] or day of the week [χ2 (6) = 4.034, p = 0.67]. However, a statistically significant variation in OHCA incidence was explained by time of day [χ2 (23) = 121.446, p < 0.01], with 30% (n = 279) of all OHCAs occurring in a five-hour period from 05:00 to 09:59 in the morning, and again in the early afternoon (Fig. 3). Fig. 4 provides the output maps following geospatial analyses. Map 1 shows the cluster and outlier analyses according to aggregation by census tract sub-place. While the results achieved were reasonable, there was also an unexpected number of outliers occurring in the larger, low-population sub-places at the edge of the metropole. This led to the use of a second aggregation method. Map 2 shows the cluster and outlier analyses according to aggregation by hexagonal tessellation. Map 3 shows the output from the hotspot analysis. The analysis results in a large area of hotspots (99% confidence interval) over the centre of the metropole, Cape Flats and southern suburbs. While data on the actual ambulance location at the time of dispatch were not available for meaningful analysis, the median (IQR) response time was 26:30 (20:44) minutes. Map 4 shows the driving time, per hexagon, to the closest PCI facility. Through proximity analysis, we calculated the median (IQR) driving time from the incident to the closest PCI-capable facility to be 10:22 (08:05) minutes. Discussion This study aimed to describe the temporal and geographic distribution of OHCA in the city of Cape Town metropole, in the Western Cape province of South Africa. A secondary aim was to describe OHCA locations as they relate geographically to PCI-capable facilities. A peak in OHCA was seen during the early to mid-morning and early afternoon, but day of the week and month of the year did not explain variation. Hotspots of OHCA incidents were identified around residential areas. Generally, PCI-capable facilities were close to the location of OHCA, but some areas remained underserved. A peak incidence of OHCA between 05:00 and 10:00 could likely be explained by relatives discovering patients who have suffered cardiac arrest in the morning, upon waking, and recognising the need for emergency assistance. Returning home from work to discover a relative is also supported in the peaks observed between 17:00 and 19:00, especially considering that the overwhelming majority of cases occurred at home. While it has not been demonstrated in more recent studies,25-27 previous studies suggest a link between night-time circadian rhythms and OHCA in the early hours of the morning.28 A more recent study observed morning (10:00–11:00) and evening (20:00–21:00) peaks.29 While a modest peak during the mid-morning is observable in our data, suggesting circadian variation, the largest peaks were observed in the early morning, supporting the theory of discovery. A protracted time to discovering a cardiac arrest victim might also be reflected in the way in which callers to an emergency contact centre describe the situation. A recent study from the Western Cape reported that callers often described cardiac arrest Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec Number of OHCAs 120 100 80 60 40 20 0 79 67 61 73 75 84 105 70 80 85 70 80 Fig. 1. Distribution of OHCA according to month of the year in 2018. Mon Tues Wed Thurs Fri Sat Sun Number of OHCAs 160 140 120 100 80 60 40 20 0 122 134 141 143 138 118 133 Fig. 2. Distribution of OHCA according to day of the week in 2018.
RkJQdWJsaXNoZXIy NDIzNzc=