Spatiotemporal Diffusion Patterns Associated with COVID-19 in the Tel Aviv-Jaffa and Haifa (Israel) Metropolitan Regions

Abstract

Social and cultural diffusion determines how behavioral phenomena spread among communities. The COVID-19 pandemic, which emerged globally at the beginning of 2020, triggered changes in human behavior in various settlements and regions. In this study, we use a spatial approach to examine diffusion patterns during the Omicron wave (December 2021–February 2022). We collected data on daily testing and confirmed cases from the Israeli Ministry of Health (MoH) database, as well as population characteristics from the Israel Central Bureau of Statistics (CBS). These data were normalized per population, classified regionally and analyzed spatially using GIS, to verify the significance of the results. We found a contagious diffusion pattern apparent spatially in the metropolitan regions of Tel Aviv-Jaffa and Haifa (Israel). Accordingly, the undulating pattern of the number of COVID-19 tests and confirmed cases began in the center of the given metropolitan region (populated with high-class settlements) at the beginning of the wave, spread out to the periphery (populated with high-class settlements) toward the mid-wave period, and returned to the center when the wave ended. Additionally, we have seen that these patterns do not accord with the morbidity spread, implying that social characteristics may have been dominant in determining the diffusion pattern.

1. Introduction

Social phenomena tend occasionally to spatially diffuse between geographic features and regions [1]. Spatial diffusion is defined as the spread of a phenomenon over space and time from a primary source outward to (a) remote destination(s).

The diffusion may be either sudden—such as the propagation of damage during an earthquake [2]—or a continuous one—such as the international diffusion of anti-terrorism laws [3]. In the field of human geography, a diffusion of behavioral change affected by processes occurring elsewhere is termed cultural diffusion [4]—that is, a process associated with a social phenomenon that disseminates from one area to another in the course of a specified period [5]. Coined by the French scholar Leo Frobenius in the early 20th century, cultural diffusion was initially defined as “the spread of cultural items—such as ideas, styles, religions, technologies, languages—between individuals, whether within a single culture or from one culture to another over space and time” [6]. Yet, as the discipline of geography has evolved, so does the term, and ever since, cultural diffusion has often been broadly classified into two main types: relocation diffusion and expansion diffusion.

Relocation diffusion occurs when a phenomenon shifts from its source to a new place, ceasing to exist at the original location. For instance, the 19th-century immigration from the Eastern to the Western United States of the Mormon community [7]. Expansion diffusion describes a phenomenon that continues to exist at its original location while spreading to new regions [8]. Expansion diffusion can be divided into two subtypes: hierarchical diffusion and contagious diffusion. Hierarchical diffusion occurs when the main distribution source of a phenomenon is tagged as the leader in a spatial hierarchy, while other spreading sources are hierarchically leveled based on a spatial order. An example is the spread of Rotary clubs throughout the United States in the most populated cities at the beginning of the 20th century [9]. Contagious diffusion, on the other hand, is based on distance decay principles, with dissemination into the neighboring or nearest area [8]—for example, the transition of European hunter-gatherers to farming by acquiring domestic plants, animals and knowledge from neighboring farmers [10].

Cultural phenomena presenting patterns of cultural diffusion can be assessed both qualitatively and quantitatively. Qualitative analysis methods rely on the interpretation of visual sources (e.g., maps and photographs), the characterization of communities, and social examination at the individual level. For example, the spread of the term “public intellectual” in the English press in Canada that reveals a continuous diffusion from the US press [11]. Quantitative methods, on the other hand, rely on statistical and mathematical models to analyze diffusion patterns by correlative models (i.e., variables are correlated), continuous models (the independent variable has continuous values—e.g., time) and discontinuous models (the independent variable is noncontinuous—e.g., place names). An example of spatial quantitative methods employed in the study of a cultural diffusion is an analysis of the patterns of dissemination of village names in neighboring regions in China that shows diffusion patterns where names spread outward from the center through spatial proximity while also appearing in distant locations due to population migration [5].

This article examines whether spatial diffusion patterns associated with the COVID-19 virus are detectable, based on a weekly analysis of three indicators: the number of COVID-19 tests normalized with the population size, the number of confirmed cases normalized with the population size and the ratio between them—namely, the Percent Positivity Index (PPI). The research question is as follows: did the spatial distribution of COVID-19 tests and confirmed cases have a societal influence? In other words, are the ratios of COVID-19 tests and confirmed cases determined solely by epidemiological conditions, or are they also influenced by social factors? We hypothesize that the diffusion of COVID-19 tests and confirmed cases are influenced by social and cultural factors. As test cases, we examine the Tel Aviv-Jaffa and Haifa (Israel) metropolitan regions, for they are socially, culturally and ethnically diverse.

1.1. The COVID-19 Virus as a Spatial Phenomenon

SARS-CoV-2, also referred to as COVID-19—a recently discovered coronavirus strain—was first detected in Wuhan, China, in December 2019 [12]. During February 2020, the virus began to spread globally, triggering significant fear and concern [12]. Commonly, once infection occurred in a given region, the number of confirmed cases and morbidity rate increased sharply, reaching a peak, and then rapidly declined back to normal, demonstrating wave propagation with peaks and lows [13]. To prevent the spread of the virus, restrictions were implemented worldwide, including lockdowns and the isolation of people suspected of being infected [14,15]. To detect the presence of infection, polymerase chain reaction (PCR) tests were used, gaining popularity [16] and becoming the primary method for detecting infection cases [15].

The spread of COVID-19 has been studied from numerus epidemiological perspectives, including the geographical propagation of the virus [17,18,19,20]. Diffusion patterns play an influential epidemiological role in terms of the virus’s propagation, and the conducted studies show they are usually predictable rather than random [21], demonstrating a combination of hierarchical and contagious diffusions patterns [8]. Yet, some studies demonstrate that propagation patterns are subjected to societal influence [22,23,24]. That is, the propagation is influenced by social factors and not just by epidemiological parameters. This is apparent, in particular, when examining social media and crowdsourcing initiatives that inform about the progress of the contagion, implying the potential influence of information transmitted within the community and raised awareness [25].

One of the indicators of the virus’s spread is the rate of confirmed cases. However, it should be considered with caution for several reasons. First, the PCR test indicates whether a person was infected with COVID-19 up to three months prior to the test and does not necessarily indicate an active illness [26]. Second, many COVID-19 carriers developed no symptoms and may have not tested themselves or obeyed the authorities’ call to be tested [27], thus causing the rate to be underestimated. Third, COVID-19 outbreaks are difficult to understand, since there is no clear insight into the effectiveness in predicting the outbreak using traditional epidemiological models in comparison to computational models [28]. Thus, one might discover that obedient communities whose members took a large number of PCR tests may have subsequently also presented a large number of confirmed cases [29]. It does not necessarily indicate they have higher morbidity but rather implies a phenomenon that may be subject to the social behavior of communities. Indeed, studies conducted worldwide show that the number of COVID-19 tests of people from a low socioeconomic status is relatively smaller than their ratio in the general population [30,31,32,33]. It is possible that behavioral emotional motives, which also have an economic impact, such as fear of self-isolation and absence from work, outweigh social motivations such as recognizing public health issues and consequently adhering to preserving the health of family and relatives [30]. Additionally, it has been demonstrated that risk perception and worries influence people’s decision to take excessive preventive measures [34].

1.2. The COVID-19 Virus in Israel

The first case of COVID-19 was confirmed in Israel on 21 February 2020 [35], not long after the first cases in Europe were reported. A month later, the country experienced the impact of the first wave of the pandemic. The government responded with lockdowns, the cancellation of leisure activities, and imposing restrictions on social gatherings. The peak of that wave appears to have been on 31 March 2020, with over 7000 new confirmed cases per day [36,37]. After a gradual decline during April and May 2020, Israel began to lift restrictions: schools reopened at the end of April, and by the end of May, cafes, hotels, and tourist attractions had partially opened again. A second wave of the virus began in June 2020—this time with government regulations on city-specific restrictions, based on local morbidity indicators, namely following the “Traffic Light Model” [38]. In September 2020, a three-week lockdown was imposed again, with the rise of the wave peaking toward the end of that month. The third wave began in December 2020, by the time the entire population began to get vaccinated for the first time. Nevertheless, a third lockdown was imposed in late December 2020 and lifted only in early February 2021, for the wave had peaked in mid-January. In July 2021, a fourth wave hit Israel, driven by the highly contagious Delta variant, which peaked in September 2021 with over 11,000 new daily confirmed cases. Throughout October and November 2021, restrictions were gradually eased [37]. The fifth wave, caused by the rapidly transmitted Omicron variant, was the fastest to rise among the waves that hit Israel. It began at the end of December 2021 and lasted until the end of February 2022, with the number of new daily confirmed cases reaching over 80,000 at its peak [36].

2. Materials and Methods

2.1. The Study Area

We selected the metropolitan regions of Tel Aviv-Jaffa and Haifa in central and Northern Israel, respectively, to examine potential spatial diffusion patterns. The Tel Aviv-Jaffa metropolitan region extends from Netanya in the north to Yavne in the south (Figure 1a), and it encompasses forty-six settlements with a total population of approximately 3.2 million. Of those, about 4% are Arabs, 10% are ultra-Orthodox Jews. Three settlements—Tel Aviv-Jaffa itself, Ramla and Lod—are defined as mixed Jewish and Arab cities (

Table 1. The division of Tel Aviv-Jaffa metropolitan settlements into five consecutive rings starting from the core (the jurisdiction area of Tel-Aviv to ring 4): Ring—the ring ID; Distance—the distance from the metropolitan center (KM); Settlements—the number of included settlements within the given ring; Pop—total population; Arab pop.—the Arab population; Arab per.—the ratio of Arabs out of the total population; Ultra-Orthodox pop.—ultra-Orthodox population; Ultra-Orthodox per.—the ratio of ultra-Orthodox out of the total population.

Ring	Distance	Settlements	Pop	Arab Pop.	Arab Per.	Ultra-Orthodox Pop.	Ultra-Orthodox Per.
Core	-	1	461,352	22,232	4.82	5104	1.11
1	3.5	9	937,332	2855	0.30	201,904	21.55
2	8.5	11	786,290	1076	0.14	33,456	4.26
3	13.5	13	495,943	83,108	16.76	56,848	11.48
4	19.5	12	542,124	28,449	5.25	29,740	5.49
Total		46	3,223,041	137,720	4.27	327,052	10.15

). The metropolitan center, as well as the largest city, is Tel Aviv-Jaffa, with more than 460,000 residents.

The borders of the Haifa metropolitan region are Zichron Yaakov in the south, Nahariya in the north (Figure 1b), and it contains forty-three settlements with nearly one million residents. Approximately 35% of the population is Arab, 6% is ultra-Orthodox Jewish. Two settlements—Haifa and Akko—are defined as mixed Jewish and Arabs cities (

Table 2. The division of Haifa metropolitan settlements into five consecutive rings starting from the core (the jurisdiction area of Haifa to ring 4): Ring—the ring ID; Distance—the distance from the metropolitan center (KM); Settlements—the number of included settlements within the given ring; Pop—total population; Arab pop.—the Arab population; Arab per.—the ratio of Arabs out of the total population; Ultra-Orthodox pop.—ultra-Orthodox population; Ultra-Orthodox per.—the ratio of ultra-Orthodox out of the total population.

Ring	Distance	Settlements	Pop	Arab Pop.	Arab Per.	Ultra-Orthodox Pop.	Ultra-Orthodox Per.
Core	-	1	285,542	34,443	12.06	16,213	5.68
1	4	6	229,384	854	0.37	3528	1.54
2	10.5	11	210,035	137,488	65.46	12,768	6.08
3	16.5	13	135,171	106,575	78.84	257	0.19
4	21.5	12	190,747	83,107	43.57	6684	3.50
Total		43	1,050,879	362,467	34.49	39,450	3.75

). Haifa is the largest city in the metropolitan region, with over 280,000 residents, while Shfar‘am is the largest non-Jewish settlement with 42,000 residents. Table 1 and Table 2 provide additional demographic details for both regions.

2.2. Data

2.2.1. Data Sources and Examined Period

We chose to examine the period of the Omicron wave for this study because of the relatively high number of new daily confirmed cases in Israel, as well as the fact that no lockdowns were in force at the time, thus social mobility was allowed with no restrictions. The beginning, ending and peak dates of the wave were determined by detecting significant alterations in the number of confirmed cases. Accordingly, the omicron wave lasted 12 weeks, starting on 20 December 2021 and reaching an end on 7 March 2022 (Appendix A, Figure A1 and

Table A1. Division of the Omicron wave period into 12 weeks of examination.

Week Number	Dates
1	14 December 2021–20 December 2021
2	21 December 2021–27 December 2022
3	28 December 2021–3 December 2022
4	4 January 2022–10 January 2022
5	11 January 2022–17 January 2022
6	18 January 2022–24 January 2022
7	25 January 2022–31 February 2022
8	1 February 2022–7 February 2022
9	8 February 2022–14 February 2022
10	15 February 2022–21 February 2022
11	22 February 2022–28 February 2022
12	1 March 2022–7 March 2022

). These dates conform with the dates published officially by the Israeli MoH and the World Health Organization (WHO).

For testing our hypothesis throughout the examined period, we used the following datasets:

(1)

The number of daily COVID-19 tests and confirmed cases in each Israeli settlement of more than 2000 residents. The dataset was downloaded from the Israeli MoH dashboard (https://data.gov.il/dataset/covid-19/resource/8a21d39d-91e3-40db-aca1-f73f7ab1df69/download/corona_city_table_collectionver_00216.csv, accessed on 26 December 2022);

(2)

The population characteristics for each of the examined settlements was downloaded from the Central Bureau of Statistics (CBS) website (https://www.cbs.gov.il/he/settlements/Pages/default.aspx?mode=Yeshuv, accessed on 26 December 2022).

We combined these two datasets based on an existing common identifier (the settlement code) and transformed them into a unified dataset containing the number of COVID-19 tests and the number of confirmed cases per day throughout the Omicron wave in each of the inspected settlements (a fragment of the dataset is presented in Appendix A,

Table A2. A fragment of the dataset used in this study. The dataset is a result of joining the datasets downloaded from MoH and CBS. The columns include the following: Date– the starting date of the week; Code—the CBS city code of the settlements; Name—the name of the settlements; Pop—population number; Longitude and Latitude—central coordinates of the settlements in Israel Transverse Mercator (ITM) coordinate reference system; Confirmed—the number of confirmed cases; Tested—the number of conducted tests.

Date	Code	Name	Pop	Longitude	Latitude	Confirmed	Tested
20 December 2021	3000	Jerusalem	966,209	220,183	632,033	69,752	1317
20 December 2021	5000	Tel Aviv—Yafo	461,352	180,489	665,757	45,502	558
20 December 2021	4000	Haifa	285,542	198,947	744,928	24,070	241
20 December 2021	8300	Rishon Leziyyon	254,238	180,285	652,959	30,648	602
20 December 2021	7900	Petah Tiqwa	248,005	189,330	666,948	20,289	230
20 December 2021	70	Ashdod	225,974	167,582	635,675	18,378	140
20 December 2021	7400	Netanya	221,850	186,926	690,287	16,897	129
20 December 2021	9000	Be’er Sheva	211,249	179,975	573,353	19,647	152
20 December 2021	6100	Bene Beraq	204,616	184,664	666,370	5447	77
20 December 2021	6600	Holon	195,696	179,547	657,938	15,042	239

; complete data are available in the Supplementary Materials)

2.2.2. Data Preprocessing

Spatial Classification of the Study Area

The introduction of GIScience (Geographic Information Science) has substantially fostered the ability to analyze spatial and temporal data [38], while suggesting new approaches to promote scientific disciplines [39,40,41]. GIScience offers a variety of tools and functions to analyze spatiotemporal diffusion patterns, such as spatial autocorrelation (Moran’s I), K-nearest neighbors (K-NN) or hotspots analysis [42]. These spatial tools and others have already been used successfully in several studies investigating diffusion. For instance, the analysis of Dengue fever diffusion in Thailand using hotspot analysis [43]; the application of spatial autocorrelation to model the spatial diffusion of the HIV/AIDS epidemic in the United States [44]; the inspection of youth crime in the United States using a combination of spatial autocorrelation and the K-nearest neighbors algorithm [45]; and the worldwide spatial propagation of COVID-19 using different combinations of such tools [46]

We summarized the daily values of COVID-19 tests and confirmed cases into weekly summations in accordance with the 12 weeks under examination. The weekly COVID-19 tests and confirmed cases values were then normalized to 10,000 residents for each of the settlements included within the Tel Aviv-Jaffa and Haifa metropolitan regions. Using the ESRI© ArcGIS Pro 3.1 software, the joint dataset (see the previous Data Section 2.2.1) was converted into a GIS layer registered to the Israel Transverse Mercator (ITM) coordinate reference system [47]. Subsequently, the core of each metropolitan area was defined by the municipal boundaries of the central city. We divided each metropolitan region into concentric rings based on their distance from the metropolitan core (Figure 1a,b). Each of the examined settlements was then associated geographically with one of the rings by the location of the settlement’s centroid. We set slightly different distance criteria, determining the distance from the metropolitan core of the rings for the two metropolitan areas, so that the number of settlements in each of the rings was as equal as possible, ensuring a comparable number of observations in each zone. Accordingly, for the Tel Aviv-Jaffa metropolitan region (Figure 1a), the distance of the buffer for rings 1, 2, 3 and 4 was set to 3.5, 8.5, 13.5 and 19.5 km, respectively (Table 1), while for the Haifa metropolitan region (Figure 1b) it was set to 4, 10.5, 16.5 and 21.5 km for the same rings, respectively (Table 2).

Socioeconomic Classification of the Inspected Population

We have associated the settlements included within each of the rings with an averaged adjacent socioeconomic cluster as defined by the Central Bureau of Statistics (CBS). That way, we were able to examine the COVID-19 tests and confirmed cases propagation also in light of socioeconomic factors. The concentric buffer rings surrounding the cores of the Tel Aviv-Jaffa and Haifa metropolitan regions demonstrate a decaying socioeconomic pattern—that is, the core is ranked with a high socioeconomic cluster (high socioeconomic status), and the rank decreases as the distance from the centroid grows. For instance, the Tel Aviv-Jaffa core is ranked as socioeconomic cluster 8, while rings 3 and 4 rank as clusters 6 and 7, respectively. On the other hand, the Haifa core is ranked with socioeconomic cluster 7, while rings 3 and 4 belong to clusters 4 and 5, respectively (Figure 2). For clarity and simplicity, the decile values of the CBS socioeconomic clusters were grouped and reclassified into three classes: low-class (clusters 1–4), mid-class (clusters 5–7) and high-class (clusters 8–10). This regrouping was considered necessary because (A) the extreme clusters (1 and 10) consist of very few settlements, making them statistically marginal when analyzed on their own. (B) The study only consists of 45 settlements within each metropolitan region; if we were to divide these into ten distinct socioeconomic clusters, there would be very few settlements per cluster, making it difficult to make meaningful comparisons. We ensure a sufficient number of settlements in each socioeconomic tier by consolidating the clusters into three broad socioeconomic groups. (C) creating categories enhances the ability to distinguish between different sociocultural communities in Israeli society. In particular, Arab and ultra-Orthodox populations are predominant in clusters 1–4, whereas more secular and liberal communities are typically found in clusters 8–10. [48] (Attachment: Figure A2).

2.3. Implemented Analyses

2.3.1. COVID-19 Propagation Indicators

As indicators for testing our hypothesis, we used three parameters: (1) COVID-19 tests; (2) confirmed cases (i.e., the number of positive tests); and (3) the ratio between them, namely, the Percent Positivity Index (PPI). These parameters were examined spatially and verified statistically to identify potential propagation patterns and examine whether social characteristics were dominant enough to form a spatio-social diffusion. The weekly values of these indicators, classified spatially into the core and four concentric rings, are presented in Appendix A,

Table A3. The number of tests, the number of confirmed cases and the ratio between them during the 12 weeks of the Omicron wave in settlements belonging to Tel Aviv metropolitan and classified into 5 concentric regions.

Ring	Core			1			2			3			4
	Tested	Verified	Ratio (%)	Tested	Verified	Ratio (%)	Tested	Verified	Ratio (%)	Tested	Verified	Ratio (%)	Tested	Verified	Ratio (%)
week 1	45,502	558	1.23	73,066	826	1.13	77,961	1163	1.49	38,665	424	1.10	43,293	418	0.97
week 2	62,556	2044	3.27	115,350	3074	2.66	128,396	3812	2.97	54,409	1334	2.45	66,823	1613	2.41
week 3	123,505	9089	7.36	205,961	13,163	6.39	217,922	13,114	6.02	117,132	6391	5.46	122,372	6662	5.44
week 4	161,534	18,214	11.28	308,918	36,578	11.84	308,711	31,337	10.15	187,208	19,484	10.41	205,813	20,837	10.12
week 5	150,126	22,477	14.97	298,434	45,694	15.31	319,245	46,243	14.49	208,085	29,990	14.41	215,251	31,259	14.52
week 6	111,495	20,637	18.51	209,974	38,028	18.11	232,688	42,508	18.27	141,844	26,170	18.45	154,334	29,774	19.29
week 7	83,606	17,002	20.34	152,859	31,723	20.75	157,722	33,893	21.49	99,137	21,483	21.67	106,435	24,219	22.75
week 8	58,741	11,556	19.67	102,376	20,703	20.22	107,960	23,315	21.60	62,716	14,158	22.57	70,918	16,030	22.60
week 9	48,270	7154	14.82	74,396	12,150	16.33	77,911	14,081	18.07	41,098	7559	18.39	50,304	9452	18.79
week 10	38,884	4629	11.90	58,460	7676	13.13	59,640	9056	15.18	31,648	4410	13.93	36,477	5238	14.36
week 11	34,931	4018	11.50	49,144	5588	11.37	48,811	6834	14.00	26,465	3096	11.70	28,457	3738	13.14
week 12	30,991	3454	11.15	43,336	4911	11.33	42,078	5565	13.23	23,561	2538	10.77	26,649	3210	12.05

and

Table A4. The number of tests, the number of confirmed cases and the ratio between them during the 12 weeks of the Omicron wave in settlements belonging to Haifa metropolitan (Figure 1) and classified into 5 concentric regions.

Ring		Core			1			2			3			4
	Tested	Verified	Ratio (%)	Tested	Verified	Ratio (%)	Tested	Verified	Ratio (%)	Tested	Verified	Ratio (%)	Tested	Verified	Ratio (%)
week 1	24,070	241	1.00	15,678	80	0.51	12,204	91	0.75	9080	85	0.94	10,797	564	5.22
week 2	27,767	564	2.03	21,438	342	1.60	13,423	265	1.97	8535	223	2.61	13,370	238	1.78
week 3	51,070	2712	5.31	52,322	2713	5.19	27,164	1432	5.27	15,050	773	5.14	28,900	1323	4.58
week 4	82,568	7677	9.30	84,420	7629	9.04	46,170	5066	10.97	25,643	2235	8.72	47,986	4131	8.61
week 5	83,945	12,457	14.84	85,012	11785	13.86	68,447	13,389	19.56	44,668	7932	17.76	65,529	10,106	15.42
week 6	71,378	13,905	19.48	71,699	13327	18.59	54,903	12,877	23.45	40,946	9479	23.15	58,656	12,073	20.58
week 7	52,935	12,963	24.49	51,072	12289	24.06	35,747	9496	26.56	25,082	6928	27.62	38,555	9527	24.71
week 8	37,252	8663	23.26	34,459	8184	23.75	22,197	4863	21.91	15,579	3500	22.47	23,575	5246	22.25
week 9	26,204	4739	18.09	22,371	4581	20.48	13,724	2257	16.45	9907	1709	17.25	16,563	3055	18.44
week 10	19,034	2633	13.83	16,215	2617	16.14	10,147	1117	11.01	7765	887	11.42	11,665	1542	13.22
week 11	15,881	1918	12.08	12,992	1816	13.98	8752	718	8.20	6123	483	7.89	9779	1004	10.27
week 12	14,897	1547	10.38	11,598	1442	12.43	7389	555	7.51	5307	341	6.43	8315	832	10.01

for the Tel Aviv-Jaffa and Haifa metropolitan regions, respectively. The weekly values of these indicators classified into the three socioeconomic groups (low-class, mid-class, high-class) are listed in Appendix A,

Table A5. The number of tests, the number of confirmed cases and the ratio between them during the 12 weeks of the Omicron wave in settlements belonging to Tel Aviv metropolitan (Figure 1) and classified into three socioeconomic group.

Week Number	Low-Class			Mid-Class			High-Class
	Tested	Verified	Ratio (%)	Tested	Verified	Ratio (%)	Tested	Verified	Ratio (%)
week 1	16,952	198	1.17	125,127	1621	1.30	84,861	921	1.09
week 2	19,174	411	2.14	175,305	5041	2.88	197,567	5624	2.85
week 3	39,594	3550	8.97	361,669	21,644	5.98	374,243	22,560	6.03
week 4	101,676	16,247	15.98	534,761	57,643	10.78	498,850	49,343	9.89
week 5	95,154	17,773	18.68	593,869	90,576	15.25	489,686	65,598	13.40
week 6	52,871	9608	18.17	394,536	76,945	19.50	356,479	61,985	17.39
week 7	36,052	6856	19.02	260,151	59,188	22.75	274,274	55,288	20.16
week 8	10,767	1831	17.01	187,021	42,050	22.48	188,767	38,623	20.46
week 9	11,446	1394	12.18	127,928	23,913	18.69	142,025	23,037	16.22
week 10	11,715	863	7.37	93,005	13,994	15.05	114,383	15,274	13.35
week 11	10,809	581	5.38	46,487	5950	12.80	101,418	12,463	12.29
week 12	8997	426	4.73	78511	10150	12.93	82879	9827	11.86

and

Table A6. The number of tests, the number of confirmed cases and the ratio between them during the 12 weeks of the Omicron wave in settlements belonging to Haifa metropolitan (Figure 1) and classified into three socioeconomic groups.

Week Number	Low-Class			Mid-Class			High-Class
	Tested	Verified	Ratio (%)	Tested	Verified	Ratio (%)	Tested	Verified	Ratio (%)
week 1	17,740	132	0.74	49,532	389	0.79	4557	31	0.68
week 2	16,698	356	2.13	61,601	1155	1.87	6234	121	1.94
week 3	25,646	1425	5.56	132,568	6741	5.08	16,292	787	4.83
week 4	55,844	5980	10.71	211,864	19,101	9.02	19,079	1657	8.68
week 5	102,286	20,863	20.40	226,977	32,446	14.29	18,338	2360	12.87
week 6	88,617	21,274	24.01	192,991	37,412	19.39	15,974	2975	18.62
week 7	51,463	13,551	26.33	137,648	34,373	24.97	14,280	3279	22.96
week 8	30,409	5899	19.40	93,021	22,341	24.02	9632	2216	23.01
week 9	19,846	2625	13.23	62,559	12,461	19.92	6364	1255	19.72
week 10	15,215	1148	7.55	44,806	6981	15.58	4805	667	13.88
week 11	13,726	616	4.49	36,073	4815	13.35	3728	508	13.63
week 12	11,482	415	3.61	32,704	3874	11.85	3320	428	12.89

for the Tel Aviv-Jaffa and Haifa metropolitan regions, respectively.

It is important to note that only the spatial and sociocultural dimensions of COVID-19 diffusion patterns are examined in this study. Despite the fact that it is clear that other factors have an impact, this study does not consider other potential influences, such as epidemiological, environmental, physical ones or any others.

2.3.2. Verification Using Statistical Tests

The Chi-squared test was chosen for our study due to the fact that it is aimed at determining if there is a fit between two sets of proportions [49]. By comparing the observed and expected values based on population proportions, one can detect spatial relationships and deviations from expected patterns. In our case, we examined whether the proportions of the observed COVID-19 tests (the conducted tests), classified into the five rings, fit those of the expected tests considering the size of the population in each of the rings. A similar inspection was made to the proportions of the confirmed cases. The resulting χ² score captures the overall deviation between the observed and expected values. The higher the χ² score, the greater the gap between the observed and expected proportions. We conducted this test for each of the 12 weeks, marked as T_n. For every week T_n, we computed a separate χ² value and compared it to a critical value determined by the chosen significance level (α = 0.05) and the degrees of freedom of 4(k − 1); the critical value is approximately 9.49 (rounded to 10) and reflects the threshold beyond which we reject the null hypothesis:

Hypothesis 1 (H1).

-

Null hypothesis (H1₀).

there will be no differentiation between the proportions of the observed and expected values during the 12 examined weeks:

$$(H{\mathbf{1}}_{\mathbf{0}}):{\chi }_{\left(\mathit{Tn}\right)}^2<10$$

Alternative hypothesis (H1₁).

there is a difference between the proportions of the observed and expected values during the 12 examined weeks:

$$(H{\mathbf{1}}_{\mathbf{1}}):{\chi }_{\left(Tn\right)}^2>10$$

The null hypothesis (H1₀) suggests that there will be no differentiation between the proportions of the observed and expected values during the 12 examined weeks, thus indicating a constant ratio between the observed and expected values along the entire period of the wave. The alternative hypothesis (H1₁) assumes there is a difference between the two sets, and these differences may change as the wave progresses in time. Additionally, we examined whether these χ² values remain the same across the 12 weeks, or if they change over time as the wave progresses:

Hypothesis 2 (H2).

-

Null hypothesis (H2₀).

no change in the gap between the observed and expected proportions across the 12 weeks:

$$(\mathbf{H}{\mathbf{2}}_{\mathbf{0}}):{\chi }_{\left(T1\right)}^2={\chi }_{\left(T2\right)}^2={\chi }_{\left(Tn\right)}^2$$

Alternative hypothesis (H2₁).

the deviation between the observed and expected proportions varies over time:

$$(\mathbf{H}{\mathbf{2}}_{\mathbf{1}}):{\chi }_{\left(T1\right)}^2\ne {\chi }_{\left(T2\right)}^2\ne {\chi }_{\left(Tn\right)}^2$$

The second null hypothesis (H2₀) suggests all weekly χ² values are equal, implying no change in the gap between the observed and expected proportions across the 12 weeks. The second alternative hypothesis (H2₁) assumes that not all χ² values are the same, suggesting that the deviation between the observed and expected proportions varies over time.

By comparing the computed χ² for each week to the critical threshold, we can determine whether certain rings have significantly higher or lower observed values than expected. Furthermore, examining the changes in χ² scores over the 12 weeks helps us identify whether the patterns of testing or confirmed cases shift during different phases of the wave, reflecting potential peaks or lows in different rings.

3. Results

3.1. The Distribution of COVID-19 Tests and Confirmed Cases in the Metropolitan Regions of Tel Aviv-Jaffa and Haifa

Table 3. The observed confirmed cases (upper section) and COVID-19 tests (lower section) vs. the expected values based on the population distribution within the Tel Aviv-Jaffa metropolitan rings.

Confirmed Cases
	Observed					Total	Expected
	Core	Ring 1	Ring 2	Ring 3	Ring 4		Core	Ratio	Ring 1	Ratio	Ring 2	Ratio	Ring 3	Ratio	Ring 4	Ratio
Week 1	558	826	1163	424	418	3389	485	115%	986	84%	827	141%	522	81%	569	73%
Week 2	2044	3074	3812	1334	1613	11,877	1698	120%	3456	89%	2898	132%	1829	73%	1995	81%
Week 3	9089	13,163	13,114	6391	6662	48,419	6924	131%	14,090	93%	11,814	111%	7457	86%	8134	82%
Week 4	18,214	36,578	31,337	19,484	20,837	126,450	18,082	101%	36,797	99%	30,854	102%	19,473	100%	21,244	98%
Week 5	22,477	45,694	46,243	29,990	31,259	175,663	25,120	89%	51,118	89%	42,862	108%	27,052	111%	29,511	106%
Week 6	20,637	38,028	42,508	26,170	29,774	157,117	22,468	92%	45,721	83%	38,337	111%	24,196	108%	26,396	113%
Week 7	17,002	31,723	33,893	21,483	24,219	128,320	18,350	93%	37,341	85%	31,310	108%	19,761	109%	21,558	112%
Week 8	11,556	20,703	23,315	14,158	16,030	85,762	12,264	94%	24,957	83%	20,926	111%	13,207	107%	14,408	111%
Week 9	7154	12,150	14,081	75,59	9452	50,396	7207	99%	14,665	83%	12,297	115%	7761	97%	8467	112%
Week 10	4629	7676	9056	4410	5238	31,009	4434	104%	9024	85%	7566	120%	4775	92%	5210	101%
Week 11	4018	5588	6834	3096	3738	23,274	3328	121%	6773	83%	5679	120%	3584	86%	3910	96%
Week 12	3454	4911	5565	2538	3210	19,678	2814	123%	5726	86%	4801	116%	3030	84%	3306	97%
Tests
	Observed					Total	Expected
	Core	Ring 1	Ring 2	Ring 3	Ring 4		Core	Ratio	Ring 1	Ratio	Ring 2	Ratio	Ring 3	Ratio	Ring 4	Ratio
Week 1	45,502	73,066	77,961	38,665	43,293	278,487	39,824	114%	81,040	90%	67,951	115%	42,887	90%	46,786	93%
Week 2	62,556	115,350	128,396	54,409	66,823	427,534	61,137	102%	124,412	93%	10,4318	123%	65,840	83%	71,826	93%
Week 3	123,505	205,961	217,922	117,132	122,372	786,892	112,526	110%	228,986	90%	192,002	113%	121,181	97%	132,198	93%
Week 4	161,534	308,918	308,711	187,208	205,813	1,172,184	167,622	96%	341,106	91%	286,013	108%	180,516	104%	196,927	105%
Week 5	150,126	298,434	319,245	208,085	215,251	1,191,141	170,333	88%	346,622	86%	290,638	110%	183,436	113%	200,112	108%
Week 6	111,495	209,974	232,688	141,844	154,334	850,335	121,598	92%	247,447	85%	207,482	112%	130,952	108%	142,856	108%
Week 7	83,606	152,859	157,722	99,137	106,435	599,759	85,766	97%	174,530	88%	146,341	108%	92,363	107%	100,760	106%
Week 8	58,741	102,376	107,960	62,716	70,918	402,711	57,588	102%	117,189	87%	98,261	110%	62,017	101%	67,655	105%
Week 9	48,270	74,396	77,911	41,098	50,304	291,979	41,753	116%	84,966	88%	71,243	109%	44,965	91%	49,052	103%
Week 10	38,884	58,460	59,640	31,648	36,477	225,109	32,191	121%	65,507	89%	54,927	109%	34,667	91%	37,818	96%
Week 11	34,931	49,144	48,811	26,465	28,457	187,808	268,57	130%	54,652	90%	45,825	107%	28,922	92%	31,552	90%
Week 12	30,991	43,336	42,078	23,561	26,649	166,615	23,826	130%	48,485	89%	40,654	104%	25,659	92%	27,991	95%

and

Table 4. The observed confirmed cases (upper section) and COVID-19 tests (lower section) vs. the expected values based on the population distribution within the Haifa metropolitan rings.

Confirmed Cases
	Observed					Total	Expected
	Core	Ring 1	Ring 2	Ring 3	Ring 4		Core	Ratio	Ring 1	Ratio	Ring 2	Ratio	Ring 3	Ratio	Ring 4	Ratio
Week 1	241	80	91	85	55	552	150	161%	120	67%	110	83%	71	120%	100	55%
Week 2	564	342	265	223	238	1632	444	127%	356	96%	326	81%	211	106%	295	81%
Week 3	2712	2713	1432	773	1323	8953	2435	111%	1952	139%	1791	80%	1155	67%	1620	82%
Week 4	7677	7629	5066	2235	4131	26,738	7273	106%	5829	131%	5348	95%	3449	65%	4840	85%
Week 5	12,457	11,785	13,389	7932	10,106	55,669	15,142	82%	12,136	97%	11,134	120%	7181	110%	10,076	100%
Week 6	13,905	13,327	12,877	9479	12,073	61,661	16,772	83%	13,442	99%	12,332	104%	7954	119%	11,161	108%
Week 7	12,963	12,289	9496	6928	9527	51,203	13,927	93%	11,162	110%	10,241	93%	6605	105%	9268	103%
Week 8	8663	8184	4863	3500	5246	30,456	8284	105%	6639	123%	6091	80%	3929	89%	5513	95%
Week 9	4739	4581	2257	1709	3055	16,341	4445	107%	3562	129%	3268	69%	2108	81%	2958	103%
Week 10	2633	2617	1117	887	1542	8796	2393	110%	1918	136%	1759	64%	1135	78%	1592	97%
Week 11	1918	1816	718	483	1004	5939	1615	119%	1295	140%	1188	60%	766	63%	1075	93%
Week 12	1547	1442	555	341	832	4717	1283	121%	1028	140%	943	59%	608	56%	854	97%
Tests
	Observed					Total	Expected
	Core	Ring 1	Ring 2	Ring 3	Ring 4		Core	Ratio	Ring 1	Ratio	Ring 2	Ratio	Ring 3	Ratio	Ring 4	Ratio
Week 1	24,070	15,678	12,204	9080	10,797	71,829	19,537	123%	15,659	100%	14,366	85%	9266	98%	13,001	83%
Week 2	27,767	21,438	13,423	8535	13,370	84,533	22,993	121%	18,428	116%	16,907	79%	10,905	78%	15,300	87%
Week 3	51,070	52,322	27,164	15,050	28,900	174,506	47,466	108%	38,042	138%	34,901	78%	22,511	67%	31,586	91%
Week 4	82,568	84,420	46,170	25,643	47,986	286,787	78,006	106%	62,520	135%	57,357	80%	36,996	69%	51,908	92%
Week 5	83,945	85,012	68,447	44,668	65,529	347,601	94,547	89%	75,777	112%	69,520	98%	44,841	100%	62,916	104%
Week 6	71,378	71,699	54,903	40,946	58,656	297,582	80,942	88%	64,873	111%	59,516	92%	38,388	107%	53,862	109%
Week 7	52,935	51,072	35,747	25,082	38,555	203,391	55,322	96%	44,339	115%	40,678	88%	26,237	96%	36,814	105%
Week 8	37,252	34,459	22,197	15,579	23,575	133,062	36,193	103%	29,008	119%	26,612	83%	17,165	91%	24,084	98%
Week 9	26,204	22,371	13,724	9907	16,563	88,769	24,145	109%	19,352	116%	17,754	77%	11,451	87%	16,067	103%
Week 10	19,034	16,215	10,147	7765	11,665	64,826	17,633	108%	14,132	115%	12,965	78%	8363	93%	11,734	99%
Week 11	15,881	12,992	8752	6123	9779	53,527	14,559	109%	11,669	111%	10,705	82%	6905	89%	9688	101%
Week 12	14,897	11,598	7389	5307	8315	47,506	12,922	115%	10,356	112%	9501	78%	6128	87%	8599	97%

present the observed COVID-19 tests and confirmed cases (positive tests) vs. the expected values considering the population size of the Tel Aviv-Jaffa and Haifa metropolitan regions, respectively. As can be observed in the core ring, at the beginning of the wave, the number of tests was higher than the expected value, while during the wave’s peak, this number drops below the expected value, only to rise above the expected again as the wave ends. A similar trend is observed in the number of confirmed cases, with this trend being seen in both metropolitan regions. In contrast, rings 3 and 4 (the peripheral rings) display a polarized trend. The number of COVID-19 tests starts below the expected values at the beginning of the wave, rises above the expected at the peak and then returns below the expected as the wave fades. Similar results are observed in the number of confirmed cases, consistent across both metropolitan regions. The trend in rings 1 and 2 is less obvious than in the other rings, but changes along the wave can still be observed in these rings.

3.2. The Distribution of COVID-19 Tests and Confirmed Cases in Israel

In general, high socioeconomic groups are associated with higher odds of perceived access to COVID-19 testing. That is, among communities of a low income, the rate of testing tends to be lower than in communities of a high income [50]. This was apparent in the general population of Israel, where the ratios of COVID-19 tests and confirmed cases in low socioeconomic clusters are smaller than their proportions in the general population. The lower ratios (0.28 and 0.32) are associated with the 1st socioeconomic cluster while the highest values (1.63 and 1.64) are associated with the 10th socioeconomic cluster. This phenomenon is seen also when grouping the ten clusters into three sub-socioeconomic groups. Accordingly, the low-class group tends to be tested less than the high-class group. This is also true when inspecting the number of confirmed cases, which are subjected to the number of tests. In the low-class group, the ratio between the observed and expected confirmed cases was 0.67, while in the high-class group it was 1.35 throughout the entire period (Appendix A,

Table A7. The ratio between the observed and expected values (in relation to population size) for the COVID-19 tests and confirmed cases in Israel, classified into ten socioeconomic clusters throughout the entire Omicron period.

Classification By Socioeconomic Cluster
Socioeconomic Cluster	COVID-19 Tests Ratio (Observed/Expected)	Confirmed Cases Ratio (Observed/Expected)
1	0.28	0.32
2	0.66	0.68
3	0.82	0.7
4	0.95	0.86
5	1.03	1.01
6	1.15	1.16
7	1.15	1.23
8	1.33	1.29
9	1.46	1.49
10	1.63	1.64
Classification by Socioeconomic Group
Low-class	0.69	0.67
Mid-class	1.11	1.13
High-class	1.37	1.35

).

3.3. The Distribution of COVID-19 Tests and PPI in the Tel Aviv-Jaffa and Haifa Metropolitan Regions

The phenomenon by which low socioeconomic groups tend to be tested less can be seen also when examining the data associated with the Tel Aviv-Jaffa and Haifa metropolitan regions (Figure 3). In the Tel Aviv-Jaffa region, the ratio between the observed and expected COVID-19 tests differ significantly between the low-class and high-class groups. The former was tested 40–80% less than expected (a ratio of 0.2–0.6), while the latter was tested 30–90% more than expected (a ratio of 1.3–1.9). The gap between the two ratios remains large throughout the entire period (Figure 3a). In the Haifa region, the ratio between the observed and expected COVID-19 tests varies significantly between the low-class group and the high-class group (Figure 3b). In the first three weeks of the wave, the ratio in the high-class group ranged between 1.18 and 1.47, that is, about 18–74% tests more than expected. The ratio in the low-class group was between 0.79 and 0.47, that is, 21–53% tests less than expected. In weeks 4–5, the gap between these groups was almost equal, where both groups were tested between 2% and 6% less than expected. In week 6–7, the ratios reverted to the initial state, where the high-class group was tested about 30% more than expected, while the low-class group was tested about 20% less than expected.

The testing trend described above is prominent in light of the PPI inspection. One can see that in general, the ratio between the number of tests and the number of confirmed cases presents a typical wave shape. Accordingly, there are low values at the beginning and ending of the wave with peaks in its middle, regardless of the socioeconomic group. That is, if the PPI is an epidemiological propagation indicator, and the number of COVID-19 tests behaves differently, then other, non-epidemiological factors might be dominant in determining propagation. This is more interesting when examining propagation classified into the three socioeconomic groups. As the Omicron wave began, the low-class group was associated with a higher rate of PPI in comparison with the high-class group in both metropolitan regions (Figure 3c,d). However, this tendency was reversed in weeks 5 and 8 in the Tel Aviv-Jaffa and Haifa metropolitan regions, respectively, with the low-class group PPI ratio dropping below the higher-class one for the remainder of the wave.

3.4. The Spatial Propagation of COVID-19 Tests

Because the COVID-19 test distribution behaves differently than that of the PPI, a spatial qualitative examination of the tests was carried out. Figure 4 and Figure 5 present the COVID-19 test distribution (normalized per 10,000) over the entire period for the Tel Aviv-Jaffa and Haifa metropolitan regions, respectively. Notably, there are shifts from the center to the periphery and back along the 12 weeks of the Omicron wave. In Tel Aviv-Jaffa metropolitan (Figure 4), starting at week 1, the core and ring 1 present a high ratio of normalized COVID-19 tests (quintiles Q4–Q5 of the distribution), while the peripheral rings (3–4) are attributed with lower ratios. In week 5, the trend alternates, as Tel Aviv-Jaffa and its immediate vicinity demonstrate lower ratios, while the peripheral settlements (e.g., Yavne, Kfar Sava and Rosh Ha-‘Ain) are attributed with higher ratios. In week 9, the trend alternates again, and higher ratios of normalized tests shift back from the peripheral regions to the metropolitan center. At the end of this wave (week 12), the reversed diffusion is complete, as the central region is attributed with high ratios of normalized tests, similar to the values that prevailed at the beginning of the wave (week 1). A similar phenomenon can be seen in the Haifa metropolitan region (Figure 5). In week 1, the core and ring 1 (henceforth “the center”) peak, while rings 2–4 (henceforth “the periphery”) have lower rates, although a few settlements (e.g., Zikhron Yaakov and Nahariya) do demonstrate higher ratios. As the wave progresses, the trend gradually changes, reversing almost completely in week 5: apart from a few exceptions (e.g., Zikhron Yaa’kov, Yoqnea’m, Naharia), the peak diffuses into rings 2–4, while the center is left with lower values. Week 9 marks the transition of the diffusion back from the peripheral rings to the core and ring 1, as the ratio in the center gradually increases from this week (week 9) onward, reaching in weeks 11–12 the values of week 1.

This tendency is reinforced by the distributions presented in Table 3 and Table 4. Accordingly, the ratio between the observed and expected COVID-19 tests in the Tel Aviv-Jaffa and Haifa metropolitan regions core begins in week 1 with high values of 114% and 123%, respectively. They then drop to 88% and 82% in week 5 and increase back to 130% and 121% during week 12. On the other hand, in ring 4 the trend is the opposite: it starts with 93% and 83% in the first week for the Tel Aviv-Jaffa and Haifa metropolitan regions, respectively; increases to 108% during week 6 for ring 4 in both metropolitan regions; and drops back to 95% and 97% in week 12—that is, at the beginning of the wave the number of actual COVID-19 tests in the cores is greater than expected, while during the mid-wave they decrease to less than expected and rise back toward the end of the wave. The opposite trend is demonstrated in the peripheral settlements represented by ring 4 in both metropolitan regions.

3.5. The Distribution of COVID-19 Tests and PPI Classified into Buffered Rings

Figure 6a,b present the distribution of COVID-19 tests classified into three ring groups (core and ring 1; ring 2; and rings 3–4) for the Tel Aviv-Jaffa and Haifa metropolitan regions, respectively. Accordingly, the ratio between the observed and expected COVID-19 tests at the center (blue line) of these metropolitan regions starts with high values, then decreases and rises back again toward the end of the wave. On the other hand, the ratio in the periphery (orange line) is polarized. It begins with low values at the beginning of the wave, peaks in the middle and then decreases back again toward the end of the wave. This trend is similar to the one described in Figure 3a,b, supporting the fact that that the high-class socioeconomic group is concentrated in the core of the metropolitan regions and the low-class one in the periphery.

The findings associated with the PPI rate in relation to the socioeconomic groups is apparent spatially when examining the distribution in the metropolitan rings. Likewise, the PPI ratio behaves typically of wave propagation, with low values at the beginning and ending of the wave and a peak towards its mid period. In the Tel Aviv metropolitan region, the center initially had the highest PPI ratio until week 5, when it dropped to below the peripheral rings (Figure 6c). In the Haifa metropolitan region, the center was slightly higher until week 4, then it dropped below the peripheral rings until week 8 but then rose back to the highest PPI ratio (Figure 6d). Once again, the behavioral difference between the COVID-19 tests and PPI propagation imply the propagation of the former is not merely epidemiological, and other factors might be dominant as well.

4. Discussion

The COVID-19 pandemic exposed complex interactions between social structures, spatial dynamics and epidemiological monitoring. Our study reveals that the propagation of COVID-19 testing and confirmed cases is not solely an epidemiological phenomenon but a complex process influenced by socioeconomic and spatial factors.

To assess the spread of the virus and the level of morbidity, health authorities used various indices. One of them is the PPI, which refers to the ratio of positive COVID-19 tests (confirmed cases) per total tests conducted [51,52]. In Israel, this ratio was monitored closely from the beginning of the pandemic by the Israeli MoH. The PPI demonstrates that the level of morbidity starts with lower values at the beginning of the wave, increases toward the peak and decreases toward the end of the wave (Figure 3 and Figure 6). This trend behavior is similar for each ring, whether it is in the center or the periphery. In other words, the epidemiological indicator of morbidity behaves similarly in all rings and regions. This trend is different from the results associated with the COVID-19 tests and confirmed cases, stressing that regarding the latter, other, non-viral factors may be dominant.

Unlike the PPI, the propagation of COVID-19 tests behaves differently. The social structure in both metropolitan regions is not unified; high-class socioeconomic groups reside in the center, and the farther away a settlement is from the core of the metropolitan area, the lower its socioeconomic ranking. Considering that low-class communities tend to be tested less than their proportion in the entire population [30,31,32,33], one would expect that the number of COVID-19 tests would remain high at the geographic center and low at the peripheral rings along the entire Omicron wave. However, the examination of the low-class socioeconomic group (which populates the metropolitan periphery) reveals that this group is indeed being tested, with a small rate at the beginning of the wave, but gradually the ratio of tests rises until reaching a peak towards the middle of the wave (weeks 5–7). Then, the ratio of testing decreases back as the wave is progressing to its end. In the high-class socioeconomic group (which populates the metropolitans center), the trend is polarized; it starts with high values, then drops towards mid-wave and rises back again as the wave ends (Figure 3). The conjunction of spatial and social perspectives in terms of COVID-19 tests implies that its propagation might also be influenced by social factors. This insight is stronger in the Haifa metropolitan region than in the Tel Aviv-Jaffa region and forms a pattern of contagious diffusion in which the phenomena of high COVID-19 testing diffuses between neighboring spatial rings (from the core to periphery and back). Social factors can be explained by commuting habits between the center and peripheral regions [53], as demonstrated by the cases of 20th-century pandemics such as cholera [54], Spanish flu [55] and AIDS [56] or different public responses to the viral spread fostered and echoed by the media [25,57]. Either way, the number of COVID-19 tests seems to be a false indicator for the actual virus spread.

This seems to be also the case of the confirmed cases, which during the pandemic were used as a proxy for the virus spread. Not surprisingly, when examining the ratio of the confirmed cases it seems that it follows the trend of the COVID-19 tests, in particular concerning the interplay between the core and periphery of the metropolitan regions (Figure 7). Accordingly, at the beginning of the wave high rates were associated with the core. Then, towards the mid period, the higher rates reversed and were associated with the periphery, and then it flipped back. This is not surprising—when a large number of tests are being conducted, it is reasonable to assume that more people will be revealed as positive to COVID. In other words, if the number of confirmed cases depends on the number of conducted COVID-19 tests, and the latter in turn is subjected to social influence in time and space, then a high ratio of confirmed cases does not necessarily indicate high morbidity but rather reflects the higher responsiveness of the community to getting tested. Considering that many COVID-19 carriers do not develop any symptoms [58], one can conclude that the number of the confirmed cases does not necessarily reflect the morbidity level nor the development of viral hotspots.

In terms of diffusion type, the propagation of COVID-19 tests and confirmed cases demonstrates a contagious diffusion pattern, whereby they diffuse from the center outward and then back inwards (Figure 4, Figure 5 and Figure 6). In the Haifa metropolitan region, a typical contagious diffusion model is well identified. For instance, the diffusion from Abu Sanan to the surrounding settlements (Figure 4), until complete contagion was reached during the peak of the wave (weeks 6–7). In the Tel Aviv-Jaffa metropolitan region, a similar process is detected in the Sharon settlements (Figure 5), where the diffusion began in Kfar Saba and Hod Hasharon toward Ra‘anana, Ramat Hasharon and Herzliya, until it reversed back to the metropolitan center. The pattern of contagious diffusion is typical also in other virus spread scenarios such as in Brazil [59], India [17], Iran [60], Indonesia [61] and Bangladesh [62].

The results presented in this paper are partial, for the Israeli MoH data does not include settlements with less than 2000 people, which may have affected the concluded patterns. Additionally, during pandemics, it is essential to differentiate between the actual viral propagation and social influence [63]. Obviously, other viral factors such as vaccination rates, vaccine types or quarantine and lockdown policies may also have an influence. In addition, numerous other factors, including environmental and climatic variables, could further shape the diffusion patterns. However, addressing these shortcomings is complex and should include additional epidemiological factors and other considerations [64,65], which are beyond the scope of this paper.

5. Conclusions

The ratios of COVID-19 tests and confirmed cases exhibit a spatial and social pattern of contagious diffusion in the Tel Aviv-Jaffa and Haifa metropolitan regions. It is an undulating movement beginning in the center, spreading outward to the edges, and returning to the center as the wave declines. As a result of comparing the PPI to the detected diffusion patterns, one may conclude that the diffusion of COVID-19 testing and confirmed cases in these two regions did not always correspond to morbidity models, suggesting that they were influenced by social factors during the Omicron wave. This is a preliminary inspection of two metropolitan regions in Israel, which will hopefully inspire further examination of this phenomenon in other regions worldwide.