Data collection
The analysis was based on a population-representative survey of the German resident population aged 18-69 years. The survey was conducted online by the certified survey company (ISO 26362) Respondi AG, the survey period was from the 17 to the 28 February 2021. The survey was conducted at a time when restrictions on public life were imposed on the population in the form of a second lockdown (starting in December 2020).
The sample was based on a multiple stratified quota selection according to the nested characteristics of gender, age, and federal state/group of federal states. The specifications of the sample were based on the population update of the Federal Statistical Office in Germany. Respondents were recruited from a pool of individuals who had indicated their general willingness to participate in anonymous surveys. Due to minor deviations of the quota characteristics from the specifications, the data were weighted by adjusting the structure of the sample with regard to gender, age, and federal state, in order to align the sample with the data of the Federal Statistical Office of Germany. The sample forms a reduced image of the basic population and is thus representative of the German population in the age group studied (18-69 years).
Survey instruments
Those suffering from depression were identified by self-description: “Have you already come into contact with depression?” The relevant answering options for the selection of the subsample were: “Yes, I have already been diagnosed with depression” and “Yes, I think that I have already had depression myself, but I didn’t undergo a formal diagnostic procedure." The effects of the COVID-19 pandemic on depression were asked as follows: "In the past 6 months, have you experienced any worsening of your health with respect to your depression as a result of the Corona-associated measures?” (In Germany, the term “Corona” was used in all political and media communications to refer to COVID-19). The possible answers were “My depression has gotten worse (worsening of the depressive symptoms)”, “I relapsed into a depressive episode”, “I had suicidal thoughts or impulses”, “I attempted suicide”, “There were other deteriorations regarding my depression ”. As an exclusion of the above-mentioned items, there was the item: “No, there was no worsening of my depression illness”. The answer fields were selected by ticking (yes / no), with multiple responses to the first five items being possible. Restriction of treatment for respondents with depression were asked as follows: "Has the situation due to the Corona crisis had any effects on the care (treatments, therapies, offers of help) for your depression in the past 6 months? (You can tick several options)”. The question included 9 items, which are listed in Table 2. Effects of home isolation on respondents with depression were asked as follows: “You said that in the past 6 months you have spent most of the day in your home environment and only leave it in urgent cases. What effects have you observed about yourself?” Scale ranged from 1 - strongly agree, 2 - agree, 3 - agree less, 4 - disagree. The question included 13 items, which are listed in Table 3.
Sample
The survey comprised N = 5,135 people (n = 2,547 female, 49.6%). Of these, n = 1,994 respondents (38.8%) had already had experience with depression themselves, including n =1,038 people (20.2%) self-reported to have been diagnosed with depression (n = 622 female, 59.9%).
Statistical analysis
A CHAID analysis (Chi-square Automatic Interaction Detectors) was carried out to identify particularly affected risk groups. [5] While it is often used as a data mining technique in various scientific disciplines due to its exploratory nature, CHAID has so far rarely been used in connection with research into depressive disorders. In a CHAID analysis, an algorithm, based on a dependent variable, checks for a large number of independent variables with the help of chi-square which tests how much the actual cell distribution deviates from the expected cell distribution. The variable with the highest chi-square value leads to the first branching in a tree diagram with two or more segments (nodes). Then, the next independent variable with the highest chi-square value is searched for each node. The process is continued until no more significant chi-square tests occur or a previously defined minimum size for the nodes has been reached. Depending on the data level, the algorithm automatically summarizes the values of the variables if this leads to a higher significant chi-square value. With a CHAID analysis, all target group segments within the surveyed population can be identified in which a characteristic occurs particularly frequently. In addition, the case numbers of the nodes are shown, which are an important indicator of the relevance of the respective segment.
In the present study, a CHAID analysis was carried out for all respondents with a depressive illness (diagnosed according to the respondents or based on self-description) who reported negative consequences of the public health measures associated with the COVID-19 pandemic on their depressive illness. The following independent variables were included in the analysis: gender, age group (18-29 years, 30-39 years, 40-49 years, 50-59 years, 60-69 years), regional allocation of the place of residence (East Germany, West Germany), type of place of residence (in or in the immediate vicinity of a large city, in or in the immediate vicinity of a small town, in rural areas), marital status (married, registered civil partnership, single with a partner in the household, single without a partner in the household, divorced / widowed with partner in the household, divorced / widowed without a partner in the household), school education (elementary school, secondary school, high school diploma, yet no general school leaving certificate), number of previous depressive episodes (one depressive episode, 2-3 depressive episode, more than 3 depressive episodes, chronically depressed), diagnosed depression (Yes, I have already been diagnosed once with depression, I think that I have already had depression myself, but no diagnosis has been made), household size (living alone, 2, 3, 4, 5 and more people), children under 14 years in the household (yes, no). The number of nodes was not limited, the minimum number of cases per branch was specified with N=50. The data was analyzed with IBM SPSS 25.