Net response of the AugUR COVID-19 spring 2020 survey was very high
Among the 2314 individuals to whom the questionnaire of the AugUR COVID-19 spring 2020 survey was sent out, 2088 individuals were contactable (i.e. alive and received questionnaire) and the filled-out questionnaire was returned by 1850 individuals (“AugUR COVID-19 survey participants”, questionnaire completion May 13th to Aug 26th, 2020). This resulted in a net response of 89% (Fig. 1), lower among women than men (87% versus 90%) and among the very old (80+) versus 73-79 years (91% versus 87%). The 1850 participants included 48% men, survey age ranged from 73 to 98 years (birth years 1922 – 1947), 5% were smoker and mean BMI was 27.6 kg/m2 (Table 1). Few women, but 57% of men had ≥13 years of education. One main reason for non-response was illness (39% too ill among 110 in non-responder phone follow-up). When comparing the lost-to-follow-up (350 non-responders, 114 died since prior visit) with the 1850 participants based on information from the prior study center visit, we found fewer men, more smoker, lower QOL, and less physical activity (Table 1).
Majority of participants was at increased risk for severe COVID-19 by medical conditions
For individuals at any age, the CDC classifies medical conditions with strong evidence for increased risk for severe COVID-19 [5], most of which are common in the elderly (cancer, chronic kidney disease, chronic bronchitis, obesity, serious heart conditions, or type 2 diabetes). Based on the information assessed at the prior study center visit (mean time between survey and prior visit = 1.8 years; < 1 year: n = 524, 28%; 1-3 years: n = 1029, 56%; > 3 years: n = 297, 16%), we derived frequencies of these medical conditions (Fig. 2A, Supplementary Table 1). We found 74% of our 1850 participants with at least one of these conditions and thus at increased risk for severe COVID-19 (Fig. 2B). This risk group was larger among men than women (76% versus 72%), mostly due to more men with serious heart conditions, and the risk group increased by 10-year age-group (71, 79, 95% for those aged 70-79, 80-89, 90+, respectively; Fig. 2B). When extending to CDC conditions listed as possible risk factors for severe COVID-19 [5] (asthma, hypertension, cerebrovascular disease, current/former smoking), the risk group increased to 94% among the 1850 participants (Supplementary Table 1). In summary, the majority of participants was at increased risk for severe COVID-19 beyond old age by pre-existing medical conditions.
Frequency of participants at increased risk for severe COVID-19 beyond old age. Shown are frequencies of individuals with medical conditions assessed at the prior study center visit (among the 1850 participants of this survey): A having a medical condition listed to increase or possibly increase risk for severe COVID-19 [5], B having ≥1 condition listed to increase risk (cancer, chronic kidney disease, chronic bronchitis, obesity, serious heart conditions, type 2 diabetes) [5] by 10-year age-groups and sex (blue and orange), men&women combined by age-group and all combined (gray)
Only four participants reported infection and all had mild consequences
We asked whether participants had undergone testing for SARS-CoV-2 infection and whether any test result had been positive. Among the 1850 participants, 52 reported a test (test dates March 21th – June 15th, 2020; reasons for testing: contact to infected, symptoms, returning from risk areas, other, n = 5, 15, 0, or 19, respectively). Four were tested positive (8% of 52, 0.2% of 1850): their age ranged from 76 to 95 years, three men, all non-smoking, three at increased risk for severe COVID-19 due to medical conditions. Two reported to live alone, two with partners; the partners were also tested, but not infected. Their QOL ranged from 50 to 80 (IQR of all at survey 50-80, at prior visit 60-85 on a scale 0, worst, to 100, best).
All participants, irrespective of infection or previous SARS-CoV-2 testing, were asked about experienced symptoms since Feb 1st. 2020. Of the 1850 participants, 23% reported at least one symptom considered COVID-19 related (cough, shortness of breath, respiratory problems, fever/chills, or loss of taste/smell [14], Supplementary Table 2). A loss of taste or smell, considered specific to SARS-CoV-2 infection, was reported by 2% (none of the four infected). Two of the four infected reported any symptom (cough, difficulty breathing, pain in extremities, diarrhea, headache, rhinitis), but none of the four reported bronchitis or pneumonia.
When linking these observations to the infection occurrence among the 46,461 inhabitants aged 70+ in the study capture area (infections mostly March – June 2020, Fig. 3A), we found the proportion of positive tested (0.3%; n = 109) and the 4.3 expected individuals with infection among the 1850 participants to fit well to our observation. Given the 16 individuals aged 70+ in the study area who died with COVID-19 (0.03% of the 46,461), the expected number of 0.6 deaths among the 2314 eligible individuals indicated little to no bias from COVID-19 related death. Of note, those aged 70+ comprised 13% of study region inhabitants, 8% of those tested positive, and 64% of those with COVID-19 related death (Fig. 3B).
The SARS-Cov-2 epidemic situation in the study capture area until August 2020 for all inhabitants and those aged 70+. We derived the numbers of SARS-CoV-2 infections and COVID-19 related deaths in the study area (city and county of Regensburg) from the Bavarian Food and Health Safety Authority (Landesamt für Gesundheit und Lebensmittelsicherheit) for the survey observation period (Feb 1st to Aug 26th, 2020). Shown are (A) number of newly reported infected per day, (B) cumulative number of deaths. Those aged 70+ comprise 13% of study region residents, 8% of infected, and 64% of COVID-19 related deaths
Household and some aspects of outside contacts during first wave lockdown
We were interested in how the number of infected participants, which were very few, related to participants’ isolation during the first wave lockdown (March 2020 to June 17th, 2020), when old aged individuals were advised to avoid public transportation and doing errands themselves. Larger households, particularly when including younger household members, were reported to increase risk of infection [15]. Of the 1850 participants, 36% reported to live alone (more women than men), 62% lived with at least one more person in a private household and 1% in a senior residence (Table 2). At the time of questionnaire completion (May 13th to Aug 26th, 2020), 92% reported at least one of the following: 81% of participants reported to do their own errands, 26% to use public transportation, 18% had a help come to their home, 3% lived with a younger generation person in the household, and 1% had contact with an infected person (Table 2). Since the lockdown ended June 16th, 2020, we conducted sensitivity analyses restricting to the 1734 participants with immediate response (i.e. questionnaire return until June 12th, 2020). This yielded the same results (Supplementary Table 3). Overall, most participants sustained at least some type of outside contacts.
Participants reduced outside contacts and refrained from medical appointments
We asked participants whether they had changed their behavior with regard to public transport, obtaining food, or healthcare seeking at the time of questionnaire completion compared to before the pandemic (as of Feb 1st, 2020). A substantial proportion reported less use of public transport, less errands on their own, and increased food delivery (33, 42, 29%, respectively, Table 3), which was more pronounced among women than men; rather few reported the opposite change.
Almost a third (29%) refrained from medical consultations, more women than men, but no difference when comparing the 80+ to the 73- to 79-year-old (Table 3). When restricting to the 1734 immediate responders, we found the same (Supplementary Table 4). Decreased healthcare-seeking behavior can be potentially problematic. We were thus interested whether we could identify susceptible subgroups. When analyzing the association of socio-demographic factors (age, sex, education, living alone) and pre-existing medical conditions with the odds of having refrained from medical consultation by logistic regression, this indicated higher susceptibility in women (OR = 1/0.7 = 1.4 across models, P < 0.003, Supplementary Table 5A). Notably, we do not know how many medical appointments were canceled by physicians or hospitals.
Participants reported a change towards less physical activity, but not for increased smoking or alcohol consumption
We asked participants whether they had changed their lifestyle with regard to sedentary or addictive behavior (physical activity, TV consumption, smoking, and alcohol consumption) at the time of questionnaire completion compared to before Feb 1st, 2020 (same, less now, more now). A quarter (26%) reported that they were less physically active versus 2% more active and 14% with more TV consumption versus 2% less, both more pronounced in women (Table 3). There was no trend towards more smoking or more alcohol consumption (7% more smoking vs. 11% less, 2% with more alcohol consumption vs. 2% with less; Table 3). The majority of participants, 60% (52% among women, 68% among men), did not report any change in these lifestyle factors. Sensitivity analyses restricting to the 1734 immediate responders yielded similar results (Supplementary Table 4). We were interested to identify subgroups particularly susceptible to change. When modelling the association of socio-demographic factors and pre-existing medical conditions with the odds of perceived decreased physical activity or increased TV consumption by logistic regression, we found significantly increased odds for increased TV consumption among the old aged versus the very old (73-79 years vs. 80+), among women, higher educated, and those living alone (P < 0.01, Supplementary Table 5A). For decreased physical activity, higher odds were found for women (OR = 1/1.7 across models, P < 0.001) and a tendency for the higher educated (Supplementary Table 5A).
Lifestyle changes quantified from during and pre-lockdown information showed a similar pattern as perceived changes
While the report that one’s own lifestyle was perceived as having changed is a noticeable parameter, we also derived the “quantified” change by comparing the current (i.e. during lockdown) with the pre-pandemic report of the respective factor. First, we aimed to understand dependencies of during-lockdown reports of lifestyle factors (physical activity category, cigarettes smoked daily, alcoholic drinks consumed daily) on socio-demographic factors (age, sex, education, living alone) and pre-existing medical conditions. We found all evaluated factors, but not “living alone”, to be associated with at least one lifestyle factor (Supplementary Table 5B).
Second, we quantified the difference between during- and pre-lockdown reports for each of the 1850 participants (change in physical activity category, difference in number of cigarettes, difference in number of drinks) and analyzed also the sample restricted to the 524 individuals with prior visit < 1 year before lockdown (March 2019- March 2020). We found 24% with decreased versus 5% with increased physical activity (19% versus 8% among the 524), almost no difference in cigarettes smoked and no difference in drinks consumed among the 1850 participants and in the 524 sub-set (Table 4).
Third, we modelled the association of socio-demographic factors and pre-existing medical conditions with quantified change restricting to the 524 sub-set (for physical activity and drinking, not for smoking due to only 14 smokers). We found no significant effect from any of the included covariates, except a reduced number of alcoholic drinks by increased age (P < 0.001) and a tendency of reduced physical activity for women (Supplementary Table 5C).
Fourth, when comparing the pattern of quantified change by the categories of perceived change (same, less now, more now), we found a consistent pattern in the 1850 (Fig. 4A-C right column), also when restricting to the 524 individuals with prior visit < 1 year before lockdown (Fig. 4A-C left column). Overall, the evaluation of quantified change supported the findings for perceived change.
Comparing quantified differences in lifestyle factors and QOL with perceived changes. We derived categories of perceived changes in lifestyle and QOL reported during lockdown (same, less/better now, more/worse now) with the quantified change of the report during lockdown compared to the report pre-lockdown. By category of perceived change, we show the distribution of the quantified change for all participants (prior visit April 2016 – March 2020, n = 1850, mean time before lockdown = 1.76 years, SD = 0.93; left column) and restricted to those with prior visit < 1 year before lockdown (March 2019 – March 2020, n = 524; right column) where information on both perceived and quantified changes was availble. Shown are (A) difference in number of cigarettes smoked daily (among current smokers at survey or prior visit, 43 smokers in left column, 13 smokers in right column), (B) difference in number of alcoholic drinks consumed daily (among alcohol consumers at survey or prior visit, n = 1357 or 385, respectively), (C) difference in QOL score (n = 1657 or 462, respectively)
A large proportion of participants perceived a worse QOL
The situation during the lockdown in March 2020 was exceptional and potentially hard on the QOL, particularly for this old age group. We thus asked participants whether they perceived a change in QOL (same, better now, worse now) compared to Feb 1st, 2020. For the 1850 participants, the majority reported no change (61%), but a substantial proportion perceived a change towards worse QOL (38% worse, 0.3% better; Table 3), more pronounced among women than men (41 and 36%, respectively). When modelling socio-demographic factors for association with a perceived worsening of QOL, we found women (P < 0.05, all models) and the higher educated (P < 0.001, all models) to be more susceptible (Supplementary Table 5A).
While the report of a perceived worsening of QOL is a noteworthy feeling of the participant, we also derived the quantified change by comparing QOL scores reported during lockdown with reports pre-lockdown within participants. We found a median of zero difference in the 1850 participants and in the 524 sub-set with prior visit < 1 year before lockdown with small variation (IQR (− 15) to (+ 10) and (− 10) to (+ 10), respectively; Table 4). When modelling socio-demographic factors for association with quantified difference in QOL restricting the 524, we found no association (Supplementary Table 5C). When comparing the quantified change in QOL by categories of perceived change, we found some consistent pattern of quantified change in QOL with perceived change, which, however, almost disappeared when restricting to the 524 participants with prior visit < 1 year before lockdown (Fig. 4C). Notably, QOL reported at survey was highly associated with pre-existing medical conditions independent of socio-demographic factors: mean QOL among 80-year-old women was 69 points, without and with adjustment for education and living alone (model I, model II, Supplementary Table 5B), but increased to 73 points among 80-year-old women without medical conditions; medical conditions decreased QOL significantly by 5 points (model III, P < 0.001). This suggests that the QOL, despite being a rough scale, captured aspects of overall health.
Overall, the large majority of participants reported no change in QOL, but 38% perceived a worse QOL, more women and higher educated, but the reported QOL score during lockdown was surprisingly similar to the QOL score reported pre-lockdown.
link
