A novel plot for the early alert of epidemic growth using regional targets: the doubling plot

Abstract Background During the pandemic, restrictions set by the Italian Government were primarily based on the regional level of key parameters including hospitalization and incidence rates. We aimed to build a specific plot to monitor trends and trigger early alerts, with daily updates publicly available on a National Portal. Methods A multidisciplinary team conceived and implemented a new composite plot, developing ad hoc R scripts on top of a specialised database, built in collaboration with the Ministry of Health. We calculated the doubling time Td as log(2)/log(1+r/100), where r is the daily change of target parameters and Td ranges between (0,+-infinity), and not determined for constant or missing values. We calculated Td daily, as either doubling (growth) or halving (decrease) time. To visualize trends, we assembled two different types of graphs: a bivariate plot showing the path of each point (Td, target parameter) over time, and a line plot of Td over time. The Y axis was inverted for doubling times, as lower Td indicate higher alert in this case. The two graphs were arranged in lines, using cutoffs for excessive high values for doubling times and low values for halving times. A third line was included to display trends of the target parameter over time. Results The plot was successfully realized and published on the Portal for all regions in February 2021 (https://www.agenas.gov.it/covid19/web/index.php?r=english%2Fdoubling&q=ITA&t=0). Since July 2021, we used the doubling plot to monitor the three main parameters adopted to set restrictions for Covid-19: a) occupancy rates in intensive care; b) occupancy rates in medical wards; c) weekly incidence rates. The plot highlighted growth trends and early alerts, particularly in the initial phases of growth. Conclusions The doubling plot can provide useful information to trigger early responses for pandemic control in decentralised governance. R code is available open source from AGENAS for free use. Key messages • The doubling plot was conceived and implemented on a National Portal to trigger early alerts of Covid-19 progression in Italian Regions and Autonomous Provinces. • The plot could be rapidly adapted to legislative parameters and can be useful in different situations to monitor epidemic growth and support public health policies.


Background:
Purchasing power parities (PPPs) are indicators of price level differences for all goods and services across countries. Their calculation follows the É KS method (https://ec.europa.eu/ eurostat/cache/metadata/en/prc_ppp_esms.htm). Prices of pharmaceuticals are collected by national statistical offices using different methods, from asking prices in pharmacies to retrieving scanner data. The sample is limited to 150 topselling medicines, which are not available in all countries. Consequently, the PLIs (price level indices) and PPPs derived are sometimes based on only 50 pharmaceuticals. In a cooperation between Eurostat and EURIPID the PPP calculation was alternatively done with the EURIPID database for 28 countries.

Methods:
The study compared the PLIs/PPPs derived from the E20-2 ''Furniture and health'' survey (aka CGS) with those from EURIPID for 2018-2020. The main challenge was to identify comparable products from the 224,448 products in EURIPID. For this we grouped those that shared the same 1) ATC, 2) active substance(s) & strength(s), 3) pack size group and 4) dosage form group (e.g., oromucosal) resulting in 157,186 distinctive products compared to 1,928 included in CGS. We used the Gross Retail price (GRP) as defined in the Eurostat PPP manual.

Results:
The ranking of PPPs was similar in both approaches, with Switzerland and Iceland in the lead and Poland and Hungary in the end. Only for some countries, e.g. the Netherlands deviations were identified. The Pearson correlation between the 2020 PLI for the Euripid subsample using the asterisk method with all products marked as representative and the CGS results was 0,946.

Conclusions:
Results from EURIPID show the same trend as the CGS. It is possible to replace the national data collection by a central source. This would reduce the data collection burden on the statistical offices and allows a closer monitoring of the evolution of pharmaceutical prices (bi-annual PPP publication instead of current 3-year interval) Key messages: To monitor affordability of medicines for all citizens it is important to compare prices on a regular basis with a simple tool as e.g. some products prices differed by 1000-times across countries. The EURIPID database (www.euripid.eu) allows detailed analysis of pharmaceutical prices in Europe and is available for free to non-commerical researchers.
Abstract citation ID: ckac129.344 A novel plot for the early alert of epidemic growth using regional targets: the doubling plot

Background:
During the pandemic, restrictions set by the Italian Government were primarily based on the regional level of key parameters including hospitalization and incidence rates. We aimed to build a specific plot to monitor trends and trigger early alerts, with daily updates publicly available on a National Portal.

Methods:
A multidisciplinary team conceived and implemented a new composite plot, developing ad hoc R scripts on top of a specialised database, built in collaboration with the Ministry of Health. We calculated the doubling time Td as log(2)/log(1+r/ 100), where r is the daily change of target parameters and Td ranges between (0,+-infinity), and not determined for constant or missing values. We calculated Td daily, as either doubling (growth) or halving (decrease) time. To visualize trends, we assembled two different types of graphs: a bivariate plot showing the path of each point (Td, target parameter) over time, and a line plot of Td over time. The Y axis was inverted for doubling times, as lower Td indicate higher alert in this case. The two graphs were arranged in lines, using cutoffs for excessive high values for doubling times and low values for halving times. A third line was included to display trends of the target parameter over time.

Results:
The plot was successfully realized and published on the Portal for all regions in February 2021 (https://www.agenas.gov.it/ covid19/web/ index.php?r = english%2Fdoubling&q = ITA&t = 0). Since July 2021, we used the doubling plot to monitor the three main parameters adopted to set restrictions for Covid-19: a) occupancy rates in intensive care; b) occupancy rates in medical wards; c) weekly incidence rates. The plot highlighted growth trends and early alerts, particularly in the initial phases of growth.
Notification and contact tracing systems of COVID-19 hold a vast amount of information on transmission chains of the virus. It can be hard to gain an understanding of these due to the number of individuals involved. Network analysis can be used as a method of visualising these systems, gaining understanding of transmission chains and as a potential tool for monitoring outbreaks. It can link cases together and show how far reaching initial infections were. Here, a system developed in the programming language R links the Irish infectious disease notification system and contact tracing system together and creates a network representation of the result. The system finds any cases or close contacts linked in any manner to the known cases. The result is a network from the earliest found case to the latest found case or contact related to the outbreak. A large outbreak of COVID-19 occurred in the student population in the West of Ireland in February 2021 with 449 cases linked to it by the Department of Public Health at the time. Using the system, 192 further positive cases were found to be linked to the outbreak. A total of 1,431 individuals were linked in some manner to it with 68% in the 19-24 age group and less than 1% in the 65+ age group. This takes a matter of seconds to run and highlights clusters within the outbreak, the largest of which had 96 cases and 121 not detected close contacts. Visualising the transmission chains also showed that there were no other large clusters outside of the outbreak at the time. A system such as this can link cases to outbreaks not previously linked, dramatically reduce the time taken to link cases together and visualise transmission chains to gain a deeper understanding of what is happening. This automated system frees up resources to allow for deeper investigation into cases and situations of concern. It also has the potential to link outbreaks together and spot previously unnoticed situations of concern. Key messages: Network analysis is beneficial for the monitoring of the spread of an infectious disease like COVID-19.
Here, 192 cases were found to be linked to an outbreak that was thought to have 449 cases in it.

Methods:
We measured HIL using the HILM-NL questionnaire, a validated and translated version of the HILM. In February 2020, the HILM-NL was sent to 1,500 members of the Nivel Dutch Health Care Consumer Panel. The response rate was 54% (n = 806). Higher HILM-NL scores imply a higher selfassessed ability in choosing and using health insurance.

Results:
There is a wide variation in HIL among citizens in the Netherlands. The average total HILM-NL score is 55.14 (ranging from 21-84). Lower-educated citizens (p<.04) and citizens with lower income (p<.01) are relatively more likely to have lower HIL, than, respectively, higher-educated citizens and citizens with higher income.

Conclusions:
Citizens who completed less education or earn a lower income are relatively more likely to have difficulty choosing and using a health insurance policy. It is important to support these vulnerable groups, so that Dutch citizens in general will be better able to choose a policy that fits their needs and preferences. This should ensure that citizens are less likely have to deal with inadequate coverage and unexpected costs.

Key messages:
There is a wide variation in HIL among citizens in the Netherlands. Citizens who completed less education or earn a lower income are relatively more likely to have difficulty choosing and using a health insurance policy.