Background and concurrent factors predicting non-adherence to public health preventive measures during the chronic phase of the COVID-19 pandemic

Abstract

Background

To determine factors that predict non-adherence to preventive measures for COVID-19 during the chronic phase of the pandemic.

Methods

A cross-sectional, general population survey was conducted in Israel. Sociodemographic, health-related, behavioral and COVID-19-related characteristics were collected.

Results

Among 2055 participants, non-adherence was associated with male gender, young age, bachelorhood, being employed, lower decrease in income, low physical activity, psychological distress, ADHD symptoms, past risk-taking and anti-social behavior, low pro-sociality, perceived social norms favoring non-adherence, low perceived risk of COVID-19, low perceived efficacy of the preventive measures, and high perceived costs of adherence to the preventive measures.

Conclusion

There appears to be a need for setting out and communicating preventive measures to specifically targeted at-risk populations.

ADHD, adherence, COVID-19, pandemic, preventive measures, public health

Novel Coronavirus 2019 (COVID-19) outbreak has an enormous impact on public health and global economy. In response to the growing pandemic, most states took preventive measures to limit the spread of cases through community transmission of COVID-19. These measures included isolation of infected and suspected patients, use of personal protective equipment (face masks, gloves, etc.), personal hygiene, restrictions on gathering and traveling, social distancing as well as mandatory quarantine and lockdown.¹ Some of the preventive measures depend on how people cooperate, i.e. whether they adhere to the measures or not.²

Despite the potentially harmful consequences for individuals and public health, non-adherence to the preventive measures (non-APM) for the COVID-19 pandemic, mainly at the acute phase, has been reported around the world.^3–7 For designing effective public health policy, it is mandatory to identify the factors that predict non-APM. Recently, several cross-sectional surveys were conducted in an effort to identify predictors of APM during the early phases of the pandemic. For instance, immediately after the detection of the first COVID-19 patient in Hong-Kong, higher levels of adoption of social-distancing measures were associated with being female, living in the geographic regions of Hong Kong that share a border with mainland China, perceiving oneself as having a good understanding of COVID-19 and being more anxious.⁷ In a survey conducted in Israel in April 2020, male gender, not having children, high levels of ADHD symptoms, smoking, past risk-taking behavior and current psychological distress levels, all predicted non-APM. On the other hand, pro-sociality, understanding of the instructions, high perceived risk of COVID-19 and high perceived efficacy of the preventive measures predicted adherence.⁵

As the COVID-19 pandemic continues, preventive measures become a constant part of our lives. The objective of this study was to identify predictors of non-APM at the chronic phase of the pandemic. The literature suggests different conceptualizations of non-APM. Firstly, preventive measures are prescribed by health agencies and therefore can be considered medical instructions and healthy lifestyle. Secondly, non-APM might endanger the non-adherent and his/her vicinity and consequently should be considered as a risk-taking behavior. Finally, preventive measures are often set as laws or regulations, implying that non-APM often mean illicit behavior. Potential predictors of non-APM were chosen for this study based on the literatures regarding the risk factors for non-adherence to medical instructions,⁸ engagement in risk-taking behavior^9,10 and engagement in anti-social behavior (ASB).¹¹ These included four groups of variables: (i) sociodemographic factors: age, gender, marital status, parenthood, ethnicity, religiousness, education, place of living, background migration, pre-outbreak and current level of income, pre-outbreak and current percent of position; (ii) health factors: healthy lifestyle (daily hours of sleep, smoking, physical activity), medical risk factors for COVID-19, subjective health, psychological distress and attention-deficit/hyperactivity disorder (ADHD) symptoms; (iii) behavioral and personality factors: past risk-taking behavior, past ASB and level of pro-sociality; (iv) perceptions regarding the COVID-19 and the preventive measures: the perceived risk of COVID-19, the perceived efficacy of the preventive measures, the perceived norms regarding APM and the perceived cost of APM.

Methods

This study was approved by the ethics committee of the Seymour Fox School of Education at the Hebrew University of Jerusalem. The survey was held from 13 to 23 May 2020, few weeks after the end of the first quarantine. A sample of 2055 online panel respondents (https://www.panel4all.co.il) representing most of the adult Israeli population was recruited.

For the primary outcome measure, non-APM, respondents were asked to rate the extent to which they adhered to each of the 13 preventive measures that were released by the Israeli Ministry of Health at the corresponding period (e.g. social distancing, personal hygiene, facemask). A five-point Likert scale was used: 1= ‘Not at all’, 2 = ‘Somewhat’, 3 = ‘Moderately’, 4 = ‘Strictly’ and 5 = ‘Very strictly’. Individual mean response scores were calculated. Based on the shape of the histogram (see Supplementary materials), a mean score <4 was considered as representing non-adherence.

Measures

The following scales were used to measure the independent variables:

1. Sociodemographic factors: Respondents completed a questionnaire consisting of items regarding age, gender, marital status, number of children, ethnicity, religious affiliation and level of observance, type of education, place of living (country region, and type of community) and background migration. In addition, respondents reported their pre-outbreak level of income (much above average, above average, average, below average, much below average), level of decrease in income since the onset of the coronavirus outbreak (on a 1 = ‘not at all’ to 5 = ‘extreme’ decrease Likert scale) and pre-outbreak and current percent of position.

2. Health factors: Respondents were asked to report their average number of daily hours of sleep, frequency of engaging in intensive physical activity and smoking habits. They also reported whether they are chronically treated or followed up for any of the following reasons (that are considered risk factors for COVID-19)¹²: heart disease, lung disease, liver disease, AIDS, cancer, organ transplantation, diabetes, dialysis, steroid treatment and prophylactic antibiotic treatment. Subjective health was probed by a single-item self-rated health (SRH) Likert scale describing their own health impression, ranging from 1 = ‘poor’ to 10 = ‘excellent’. This scale was found to reflect individuals’ perceptions of their physical health and psychological well-being.¹³ Another scale of subjective wellbeing was used but no analyzed for the current study. An adapted version of the Kessler Screening Scale for Psychological Distress (K6) was used to probe for non-specific psychological distress.¹⁴ For the purpose of the current study, only the first part of the scale was used, in which respondents rated on a five-level Likert scale (1 = ‘All the time’, 5 = ‘None of the time’), the level of six psychiatric common symptoms during the period of the COVID-19 crisis. The questionnaire is sensitive to high levels of mental distress (Kessler et al. 2010) and is used in the annual US National Health Interview Survey.¹⁵ The Hebrew version of the Adult ADHD Self-Report Scale (ASRS-v1.1)^16,17 was used to measure the level of ADHD symptoms. A total score of ASRS was created by averaging the responses to all 18 items. For both the K6 and the ASRS, average scores were dichotomized at the clinical suggested cutoff.^14,16

3. Behavioral and personality factors: The pro-social subscale of the young adult Strengths and Difficulties Questionnaire (SDQ)^18,19 was used for measuring pro-sociality. Respondents rated the extent to which a series of six attributes described them during a 6-month reference period on a three-level response scale (0 = ‘not true’, 1 = ‘somewhat true’ or 2 = ‘certainly true’). A short form of the Adult Risk-Taking Inventory (ARTI)^20,21 was used to measure past engagement in risky behavior. The short form consists of 14 items probing for the frequency of engagement in relatively frequent activities (e.g. sunbathing without sunscreen, smoking marijuana) with respect to their frequency during the preceding year on a rating scale, ranging from 1 = ‘Not at all’ to 7 = ‘On a daily basis’. Previous work has shown that the ARTI has good reliability and validity. Past ASB was assessed using 15-item 4-point frequency scale, ranging from 1 = ‘Not at all’ to 4 = ‘More than 5 times’, adapted from Cho et al.²² For each of the three scales, the average continuous scores were converted to categorical scores by grouping values into four groups with quartiles as cutoff points.

4. COVID-19-related perception factors: Perceptions regarding the COVID-19 and the preventive measures were assessed using several five-point Likert scales: Perceived risk of COVID-19 was assessed by a nine-item self-report questionnaire that was designed for this study based on the risk perception literature.²³ For example, ‘How likely are you to get COVID-19?’. In this sample, the scale had good internal consistency (Cronbach’s α = 0.80). Perceived efficacy of the preventive measures was measured by a self-report questionnaire designed for this study. The scale consists of five items and was composed for measuring participants’ perceived efficacy of the instructions. For example, ‘To what extent do you think that adhering to the preventive measures will reduce the chances that you or your loved ones will get COVID-19?’. In this sample, the scale had good internal consistency (Cronbach’s α = 0.83). Another scale composed for this study, consisted of seven items probing for the perceived costs of APM, including the perceived cost of APM on different domains of wellbeing (e.g. financial, social, spiritual). For example, ‘To what extent adhering to the preventive measures will impair your interpersonal relationships?’. In this sample, the scale had good internal consistency (Cronbach’s α = 0.84). Perceived norms regarding APM were measured by four questions regarding the descriptive (i.e. the prevalence of non-APM) and the injunctive (i.e. the tolerance toward non-APM) norms of the family/friends and the community/workplace they are embedded in.

Analytic approach

First, unadjusted logistic regression analyses were used to calculate the associations between each of the independent variables and the primary outcome. Next, four adjusted models were examined using backward stepwise logistic regressions with a probability of 0.05 for entry and 0.1 for removal. Multicollinearity was examined through Spearman’s rank correlation analysis. For the first model, only the sociodemographic variables were included. For the second model, health-related variables were entered in a second block. Similarly, for the third and fourth models, the second block consisted of the behavioral and personality variables and the COVID-19-related perceptions variables, respectively. P-values were not corrected, a P-value <0.05 was considered statistically significant.

Results

Sample characteristics

Table 1 summarizes the sociodemographic, health-related, behavioral and COVID-19 perceptions related variables.

Table 1

Sociodemographic, health-related, behavioral and personality, and COVID-19-related perceptions characteristics of the sample

Characteristic	N (%)
Adherence to preventive measures
Adherent	1273 (61.9)
Non-adherent	782 (38.1)
Sociodemographic
Gender
Women	1128 (55.4)
Men	909 (44.6)
Age
18–24	309 (16.5)
25–34	529 (28.3)
35–44	377 (20.1)
45–54	263 (14.1)
55–64	266 (14.2)
≥65	127 (6.8)
Ethnicity
Jewish	1834 (89.2)
Non-Jewish	210 (10.2)
Religiousness
Non-religious	1501 (73.7)
Religious	535 (26.3)
Marital status
Single	583 (28.6)
Married or in a relationship Divorced or widowed	1302 (92.3), 157 (7.7)
Having children
Yes	1253 (68.3)
No	582 (31.7)
Higher education
Yes	1054 (51.7)
No	983 (48.3)
Region/district
North	351 (17.1)
Haifa	316 (15.4)
Center	591 (28.8)
Tel Aviv	205 (10.0)
Jerusalem	254 (12.4)
West Bank	60 (2.9)
South	277 (13.5)
Place of living
Urban	1637 (83.4)
Rural	325 (16.6)
Country of birth
Israel	1794 (87.3)
Not Israel	260 (12.7)
Pre-outbreak level of income
Much above average	78 (3.8)
Below average	292 (14.3)
Average	611 (30.0)
Above average	498 (24.4)
Much below average	561 (27.5)
Pre-outbreak percent of position
0%	326 (16.2)
25%	145 (7.2)
50%	207 (10.3)
75%	236 (11.7)
100%	1101 (54.6)
Current percent of position
0%	762 (38.3)
25%	157 (7.9)
50%	168 (8.5)
75%	196 (9.9)
100%	705 (35.5)
Decrease in income (1–5 scale)	Mean = 2.49, SD = 1.51
Health
Daily hours of sleep	Mean = 8.79, SD = 1.37
Smoking
No	1820 (89.0)
Yes	225 (11.0)
Regular physical activity (number of days a week of 30 min vigorous-intensity activity)
0	515 (26.3)
1–2	757 (38.6)
3 or more	688 (35.1)
Risk factors for COVID-19
No	1755 (87.2)
Yes	258 (12.8)
SRH (1–10 scale)	Mean = 8.02
ADHD screener	SD = 1.92
Below cutoff	1729 (84.1)
Above cutoff	326 (15.9)
Psychological distress (K6)
Below cutoff	1513 (75.2)
Above cutoff	500 (24.8)
Behavior and Personality
Risk-taking behavior (ARTI)
1st quartile	517 (25.2)
2nd quartile	446 (21.7)
3rd quartile	521 (25.4)
4th quartile	571 (27.8)
ASQ
1st quartile	476 (25.0)
2nd quartile	544 (28.6)
3rd quartile	395 (20.7)
4th quartile	490 (25.7)
Pro-sociality (SDQ)
1st quartile	527 (26.1)
2nd quartile	337 (16.7)
3rd quartile	461 (22.8)
4th quartile	696 (34.4)
COVID-19-related perceptions
Descriptive and injunctive norms
1st quartile	523 (26.0)
2nd quartile	498 (24.7)
3rd quartile	540 (26.8)
4th quartile	453 (22.5)
Perceived risk of COVID-19
1st quartile	598 (31.3)
2nd quartile	433 (22.7)
3rd quartile	448 (23.4)
4th quartile	432 (22.6)
Perceived efficacy of the preventive measures
1st quartile	470 (23.4)
2nd quartile	471 (23.4)
3rd quartile	568 (28.2)
4th quartile	502 (25.0)
Perceived cost of adherence to preventive measures
1st quartile	574 (28.5)
2nd quartile	450 (22.4)
3rd quartile	542 (26.9)
4th quartile	447 (22.2)

Rate of APM

A significant minority of the participants (38.1%) reported a mean level <4 and was consequently defined as non-adherent.

Predictors of non-APM

Spearman’s correlation coefficients were examined for all independent variables. All correlations were <0.7. The correlations among age, marital status and having children were in the 0.53–0.69 range.

Table 2 presents the unadjusted and adjusted regression analysis results for non-APM. The following variables were found to predict non-APM after adjustment for sociodemographic variables.

Table 2

Non-adherence to the public health preventive measures of the Ministry of Health for the COVID-19 pandemic by a range of sociodemographic, health-related, behavioral and personality, and COVID-19-related perceptions characteristics

	N (%)	Unadjusted OR	Adjusted OR
Gender
Women	382 (33.9)	Ref	Ref
Men	391 (43.0)	1.47 (1.23–1.77)  *	1.40 (1.12–1.75)
Age
18–24	142 (46.0)	3.65 (2.22–6.00)	2.80 (1.47–5.33)
25–34	232 (43.9)	3.35 (2.08–5.40)	3.21 (1.80–5.72)
35–44	142 (37.7)	2.59 (1.59–4.24)	2.85 (1.59–5.11)
45–54	79 (30.0)	1.84 (1.10–3.09)	1.74 (0.94–3.19)
55–64	86 (32.3)	2.05 (1.23–3.43)	2.25 (1.24–4.07)
≥65	24 (18.9)	Ref	Ref
Ethnicity
Jewish	702 (38.3)	Ref
Non-Jewish	75 (35.7)	0.90 (0.67–1.21)
Religiousness
Non-religious	562 (37.4)	Ref
Religious	213 (39.8)	1.11 (0.90–1.35)
Marital status
Single	278 (47.7)	2.13 (1.46–3.11)	1.90 (1.11–3.25)
Married or in a relationship	454 (34.9)	1.25 (0.87–1.80)	1.22 (0.77–1.94)
Divorced or widowed	47 (29.9)	Ref	Ref
Having children
Yes	417 (33.3)	Ref
No	270 (46.4)	1.74 (1.42–2.12)
Higher education
Yes	386 (36.6)	Ref
No	390 (39.7)	1.14 (0.95–1.36)
Region/district
North	143 (40.7)	Ref
Haifa	116 (36.7)	0.84 (0.62–1.15)
Center	215 (36.4)	0.83 (0.63–1.09)
Tel Aviv	72 (35.1)	0.79 (0.55–1.13)
Jerusalem	110 (43.3)	1.11 (0.80–1.54)
West Bank	18 (30.0)	0.62 (0.35–1.13)
South	108 (39.0)	0.93 (0.67–1.28)
Place of living
Urban	623 (38.1)	Ref
Rural	123 (37.8)	0.94 (0.78–1.27)
Country of birth
Israel	705 (39.3)	1.54 (1.16–2.04)
Not Israel	77 (29.6)	Ref
Pre-outbreak level of income
Much below average	219 (39.0)	Ref
Below average	182 (36.5)	0.90 (0.70–1.15)
Average	227 (37.2)	0.92 (0.73–1.17)
Above average	118 (40.4)	1.06 (0.79–1.41)
Much above average	29 (37.2)	0.92 (0.57–1.51)
Pre-outbreak position percent
0%	112 (34.4)	Ref
25%	55 (37.9)	1.17 (0.78–1.75)
50%	83 (40.1)	1.28 (0.89–1.83)
75%	93 (39.4)	1.24 (0.88–1.76)
100%	426 (38.7)	1.21 (0.93–1.56)
Current percent of position
0%	254 (33.3)	Ref	Ref
25%	76 (48.4)	1.88 (1.33–2.66)	2.14 (1.41–3.26)
50%	72 (42.9)	1.50 (1.07–2.11)	1.82 (1.18–2.79)
75%	71 (36.2)	1.14 (0.82–1.58)	1.09 (0.72–1.65)
100%	281 (39.9)	1.33 (1.07–1.64)	1.21 (0.90–1.62)
Decrease in income (1–5 scale)	2028	0.95 (0.89–1.01)	0.91 (0.84–0.99)
Daily hours of sleep	2010	0.92 (0.86–0.99)
Smoking
No	667 (36.6)	Ref
Yes	105 (46.7)	1.51 (1.15–2.00)
Regular physical activity (number of days a week of 30 min vigorous-intensity activity)
0	219 (42.5)	Ref	Ref
1–2	282 (37.3)	0.80 (0.64–1.01)	0.71 (0.53–0.94)
3 or more	237 (34.4)	0.71 (0.56–0.90)	0.68 (0.50–0.92)
Risk factors for COVID-19
No	667 (38.0)	Ref
Yes	92 (35.7)	0.90 (0.69–1.19)
SRH (1–10 scale)	2052	0.98 (0.94–1.03)
ADHD screener
Below cutoff	619 (35.8)	Ref	Ref
Above cutoff	163 (50.0)	1.79 (1.42–2.28)	1.46 (1.04–2.05)
Psychological distress (K6)
Below cutoff	534 (35.5)	Ref	Ref
Above cutoff	225 (45.0)	1.50 (1.22–1.84)	1.41 (1.04–1.91)
Risk-taking behavior (ARTI)
1st quartile	138 (26.7)	Ref	Ref
2nd quartile	132 (29.6)	1.16 (0.87–1.53)	0.92 (0.63–1.32)
3rd quartile	202 (38.8)	1.74 (1.34–2.26)	0.97 (0.67–1.40)
4th quartile	310 (54.3)	3.26 (2.53–4.21)	1.70 (1.16–2.50)
ASQ
1st quartile	101 (21.2)	Ref	Ref
2nd quartile	177 (32.5)	1.79 (1.35–2.38)	1.55 (1.09–2.21)
3rd quartile	153 (38.7)	2.35 (1.74–3.17)	2.11 (1.44–3.09)
4th quartile	285 (58.2)	5.16 (3.89–6.86)	3.71 (2.53–5.45)
Pro-sociality (SDQ)
1st quartile	258 (49.0)	Ref	Ref
2nd quartile	122 (36.2)	0.59 (0.45–0.78)	0.74 (0.51–1.07)
3rd quartile	173 (37.5)	0.63 (0.49–0.81)	0.72 (0.51–1.01)
4th quartile	215 (30.9)	0.47 (0.37–0.59)	0.61 (0.45–0.84)
Perceived non-adherence norms
1st quartile	104 (19.9)	Ref	Ref
2nd quartile	158 (31.7)	1.87 (1.41–2.49)	1.13 (0.77–1.64)
3rd quartile	239 (44.3)	3.20 (2.43–4.21)	1.77 (1.22–2.56)
4th quartile	264 (58.3)	5.63 (4.23–7.48)	3.02 (2.02–4.52)
Perceived risk of COVID-19
1st quartile	317 (53.0)	Ref	Ref
2nd quartile	191 (44.1)	0.70 (0.55–0.90)	0.74 (0.53–1.04)
3rd quartile	146 (32.6)	0.43 (0.33–0.55)	0.49 (0.35–0.70)
4th quartile	90 (20.8)	0.23 (0.18–0.31)	0.31 (0.21–0.47)
Perceived efficacy of the preventive measures
1st quartile	295 (62.8)	Ref	Ref
2nd quartile	208 (44.2)	0.47 (0.36–0.61)	0.75 (0.53–1.06)
3rd quartile	169 (29.8)	0.25 (0.19–0.33)	0.48 (0.34–0.70)
4th quartile	91 (18.1)	0.13 (0.10–0.18)	0.28 (0.18–0.42)
Perceived cost of adherence to preventive measures
1st quartile	181 (31.5)	Ref	Ref
2nd quartile	176 (39.1)	1.40 (1.08–1.81)	1.32 (0.93–1.89)
3rd quartile	225 (41.5)	1.54 (1.21–1.97)	1.35 (0.95–1.92)
4th quartile	182 (40.7)	1.49 (1.15–1.93)	1.98 (1.35–2.89)

^*Note. Adjust OR represents the results of a 2-block backward stepwise logistic regressions (probability of 0.05 for entry and 0.1 for removal) with the sociodemographic variables in the first block and one of the other group of variables in the second block. Bold values represent confidence intervals that do not contain zero.

Sociodemographic variables: The following variables predicted non-APM on adjusted analyses: male gender, younger age (≤64) and being single. Being employed predicted more non-APM; conversely, decrease in income predicted more adherence.

Health-related variables: Less regular physical activity and higher levels of ADHD symptoms and of psychological distress predicted non-APM.

Behavioral and personality factors: High levels of past risk-taking behavior and ASB as well as low levels of pro-sociality, predicted non-APM.

COVID-19 perception variables: Non-APM was predicted by higher perceived non-adherence norms, lower perceived risk of COVID-19, lower perceived efficacy of the preventive measures, as well as higher perceived costs of adherence.

Discussion

Main finding of this study

This study aimed to identify risk factors for non-APM during the chronic phase of the COVID-19 outbreak. Several factors were found to predict non-adherence. Sociodemographic factors included male gender, young age, bachelorhood, being employed and smaller decrease in income. Health-related factors included low physical activity, psychological distress and ADHD symptoms. Behavioral and personality factors included history of risk-taking and ASB, and low pro-sociality. Finally, COVID-19 perception factors included perceived social norms favoring non-adherence, low perceived risk of COVID-19, lower perceived efficacy of the preventive measures, and higher perceived costs of adherence to the preventive measures. Notably, the greatest predictors in terms of OR were lower age, past ASB, low perceived risk of COVID-19, the efficacy of the preventive measures and the norms of adhering to the preventive measures.

What is already known on this topic

The variables that predicted non-APM at the chronic phase of the outbreak in Israel were similar to those that predicted non-APM during the first wave in Israel,⁵ suggesting that similar motivations drive the decision whether to adhere to preventive measures or not. Many of the non-APM predictors that were found in this study have also been reported in studies conducted in other states. For instance, male gender and young age were linked to non-adherence in the USA, Somalia, Saudi Arabia and Hong Kong during the COVID-19 outbreak.^3,4,7,24,25 The negative correlation between employment and adherence in the current study parallels the findings of Porten et al. during the SARS outbreak in Germany.²⁶ Several studies highlighted the association between adherence and perceptions about the infection and the preventive measures in a variety of states during the current pandemic.^27–29 The negative correlation between adherence and the perceived costs of adherence resembles the findings of DiGiovanni et al.³⁰ reporting that perceived economic costs of the quarantine in Canada during the 2004 SARS outbreak were related to non-adherence. The role of social norms has been demonstrated in a study concerning quarantine in Senegal during Ebola outbreak³¹ and in Australia during H1N1 outbreak.³² The negative correlation between adherence and past risk-taking behavior and unhealthy lifestyle is in line with a study reporting that among young adults with hazardous drinking adherence to public policies is suboptimal.³³

What this study adds

Our study adds new predictors of non-adherence including ADHD symptoms, general risk-taking behavior, previous engagement in crime as well as low pro-sociality, which contributes to better prediction of non-APM.

Many of the above listed factors have been shown to predict non-adherence to medical treatment,⁸ risk-taking behavior^9,10 and ASB.¹¹ Accordingly, adherence to preventive measures may be analyzed in all the corresponding theoretical frameworks.

Notably, having medical risk factors for COVID-19 (i.e. background diseases) did not predict higher adherence to preventive measures. A similar independency between objective risk and adherence was found in a study reporting no effect of the total probable cases of SARS on likelihood of adherence.³⁴

The current findings of observable risk factors for non-APM suggest that the nature and the communication of the preventive measures should be targeted for different people. Policymakers may develop specific plans for populations at risk of non-adherence, focusing on messaging, fostering and enforcing preventive measures, as well as on increased monitoring of infection rate. Further research is warranted for identifying other risk factors for non-APM across longer periods and changing contexts and for examining the efficacy of public health policy in promoting APM.

Limitations of this study

In deriving implication for public health, it is important to differentiate between predictors that preceded the COVID-19 outbreak and therefore can be considered risk factors for non-APM, and other variables that coincided with the outbreak and hence their causal relations with non-APM cannot be determined based on a cross-sectional study. The latter include the economic consequences of COVID-19, as well as the perceptions regarding the pandemic and the preventive measures. Nevertheless, these coinciding predictors may still be used for targeting populations at-risk for non-APM. The focus on the Israeli public and the timing of the survey in a relative early period of the pandemic might limit the generalization of the findings to other places and times.

Funding

This research was supported by the Ministry of Science, Technology and Space, Israel.

Authors’ contribution

YP, IB, RS, and HD designed the study. IB, HD, RS, and OGS composed and organized the scales and input the data. YP and RS analyzed the data. YP and RS wrote the first version of the paper and all other authors reviewed and revised the paper.

Conflict of interest

On behalf of all authors, the corresponding author states that there is no conflict of interest.

Yehuda Pollak, Professor

Rachel Shoham, Senior Lecturer

Haym Dayan, PhD Student

Ortal Gabrieli-Seri, PhD Student

Itai Berger, Professor