The most recognized trends of transmission of COVID-19 are that based on publicly reported data. They are updated and disseminated by provinces and territories via public media, available from various internet sources, such as Canada COVID-19 Situation Dashboard produced by ArcGIS, https://www.covid-19canada.com/, or https://www.worldometers.info/coronavirus/country/canada/. These trends are typically presented by date of report. They are strongly affected by confounding factors such as testing patterns, reporting patterns, computer glitches, weekdays vs weekends, etc. A different trend presentation is the “Epidemic curve” based on date of illness onset*, such as that published by Government of Canada on Canada.ca. This is more relevant to the disease transmission. However, reporting delays make the numbers of cases with date of onset in recent days tend to be more incomplete compared to cases with dates of onset quite a long time ago. This causes a time-bias with an artificial decline for recent cases. It is more informative to use statistical models to adjust for this reporting delay in order to illustrate the trend in the recent past, as a form of now-casting. Reporting delay adjustments are performed by using survival analysis techniques for right-truncated data to calculate adjust weights. The mot relevant epidemic trends should be presented by dates at transmission, but they are not directly observable from data. They are revealed through statistical models that take the incubation time distribution into account. If we call the reporting delay adjustment “now-casting”, then the estimated trends by dates of transmission is “back-casting”. The algorithm is applied to the reporting delay adjusted trend by date of onset. We demonstrate our results based on two-step analysis of reported data in five selected provinces with Step 1: now-casting trends by date of onset through reporting delay analysis, and Step 2: back-casting trends by date of transmission. For each province, trend representations are plotted on the same chart according to three different event markers: by time at transmission, by time of onset and by time of reporting. The trend by date of transmission and the trend by date of onset share close resemblance, separated by approximately 5 days of the average incubation periods. The trend by date of transmission and the trend by date of report are far apart by 10 days or more. They still have some resemblance. The trend by date of report, as the most visible trend that the public sees, not only reflects the past transmission taking place 10 to 15 days prior, but also influenced by other confounding factors. The importance messages of these results are: (i) recognition of different transmission patterns and timing in different provinces; (ii) recognition of the time-delay between what we see based on reported data and what might have happened in the past; (iii) challenges to mathematical modelling in general while discussing fitting models to data; (iv) future directions for both modelling and for improvement of data collection.