Study design
A cross-sectional study design was conducted using a pre-tested structured questionnaire to examine the relationship between mental health and health work functioning for health workers in Uganda during the COVID-19 pandemic.
Study setting
This study was conducted in the various health facilities and institutions supporting the COVID-19 task force at the national and sub-national levels. The health facilities included private and public facilities providing health care services during the pandemic and were directly (case management) or indirectly (other complimentary services during the response) involved in the COVID-19 response. The COVID-19 task force included multi-sectorial and multi-disciplinary organs of relevant government and non-governmental organizations working through its subcommittees at the national and the district level.
Study population
The study participants were health workers involved in the routine health care services or involved in at least one of the six subcommittees of the national or districts task forces. The subcommittees included surveillance; coordination (resource allocation and budgeting); risk communication and social mobilization; case management (clinical management), infection prevention and control; logistics/supply chain management and laboratory. The category of health workers included General and specialized medical practitioners, Pharmacists, Allied Health professionals, Nurses and health management personnel supporting the national and subnational task forces of the COVID-19 in Uganda. The workers came from the private, public sector and, some were implementing partners supporting the national and subnational response.
Sampling procedure and sample size
Multi-stage cluster sampling was used to estimate 1530 participants for the study. The researchers formed three clusters from the population of health workers providing healthcare services and those in the COVID-19 task forces in Uganda. The district local governments (DHO) cluster consisted of health workers in the 134 districts countrywide, referral hospitals cluster consisting of health care workers in 15 referral hospitals and the National Task Force cluster consisting of multiple stakeholders headed by the Ministry of Health. Ten health workers were selected from each subnational cluster (i.e. DHO and referral hostile cluster) and, forty health workers came from members of the National Task Force at the MOH.
The researcher assumed that in the district cluster, one person would be randomly sampled from each of the six pillars (subcommittees) of the district task force, and the other four persons would come from the health workforce in the health facilities within the district. At the referral hospital cluster, two people came from the COVID-19 treatment center and, eight were health care workers providing routine health care services within the hospital. At the national level cluster, at least six persons were randomly sampled from each of the six pillars (committee) of the national task force for the COVID-19 response in the Ministry of Health. The other four were health care workers who were not part of the NTF but had played a critical role in the COVID-19 response like the scientific committee and the Information Communication and Technology (ICT) team of the MOH.
Data collection instruments
The mental health status of health care workers were measured using the Depression Anxiety Stress Scale 21 (DASS21) questionnaire adopted from the Psychology Foundation of Australia available at http://www2.psy.unsw.edu.au/dass/ (PFA, 2018)[23]. The tool was designed to assess the severity of the core symptoms of depression, anxiety, stress; and to give a possible indication of a clinical disturbance. DASS21 includes three scales, each containing seven questions. The participants were required to use 4-point severity or frequency scales to rate the extent to which they experienced each state over the past week. The measure for depression, anxiety and stress were calculated by totaling up the scores for the relevant items [23]. A detailed description of the question is in the appendix four.
The measure of health work functioning was subjective using a questionnaire adapted from the Nurses Work Functioning Questionnaire (NWFQ) by Williams (2017) used under the license code CC BY-SA 3.0 (Attribution-ShareAlike 3.0 Unported) available at http://creativecommons.org/licenses/by-sa/3.0/. The self-reported questionnaire was job-specific and related better to the work context compared to other tools. The tool was originally designed to be completed by nurses and allied health professionals. In this study, the questionnaire was adapted to measure work function impairment for other categories of professional healthcare workers who might be experiencing depression, anxiety or stress symptoms.
Response formats had three-category scales. The content of the response scales varies between likert-type scales (from 0, strongly disagree, to 4, strongly agree), relative frequency categories (from 0, almost never, to 4 almost always). The sum scores of the subscales ranged from 0 to 100 and were calculated as follows: (sum of item scores100) / (number of items of the subscale maximum item score). The sum score of the total NWFQ score was calculated based on the sum of all 18 items according to the same principle. A higher total score indicates more impairment in work functioning.
Data collection
An online survey questionnaire was used to collect the quantitative data. I administered the tool to healthcare workers in public and private health facilities as well as members of the National and subnational task force for the COVID-19 response in Uganda. The tool was shared through email and social media groups (Whatsapp® and Telegram®) to the representatives of the districts (Districts health officers and surveillance focal persons), representatives of the referral level (Hospitals Directors and Pharmacists) and chairpersons of the difference subcommittees of the National taskforce. The tool was disseminated by sharing with the different professional organizations including, Pharmacists, Nurses and Midwives. Further dissemination involved social media platforms including, Expanded Programme for Immunization (EPI) /Integrated Diseases Surveillance and Response (IDSR) (Whatsapp®), Medicine Management Supervisors platform (using email and Whatsapp®), The Districts Health Officers (DHOs) platform (using email and Whatsapp®), the National districts taskforce for COVID-19 response (using email and Whatsapp®) and the various Hospital Whatsapp® and Telegram® groups at the referral and districts hospitals across the country and to other individual health workers using different media platforms including Instagram® and Facebook®.
Upon opening the tool, participants were instructed to read and consent to participate in the study. Once a participant has consented, the tool opens and, the participant could proceed to answer the questions and later submit. The data collection was from June to July 2021 included socio-demographic, existing concerns of healthcare workers on mental and psychosocial problems, and factors relating to work function impairment.
The qualitative data was collected using key informant interviews and focus group discussions. The four key informants included a psychiatrist, the chairperson of the risk communication and psychosocial support, one of the Directors of a referral hospital and a District Health Officer. The interview explored the effect of the change in the course of the pandemic on the mental health of health workers and the relation of work function impairment. The key informants for the interview were purposively selected based on their knowledge, experience and expertise in the COVID-19 pandemic. The researchers carried out two focus group discussions in two referral hospitals in the Northern region and the other in the Southwestern region. The participants included health workers in the response including, the Surveillance officer, General and specialized medical practitioners, logisticians (Pharmacists), community health workers, nurses and allied health professionals working during the response. The information from the key informant interviews and focus group discussions were recorded, coded and summarized into themes. The results were used to supplement the quantitative findings from the survey and gather a deeper understanding of the pandemic and mental health.
Inclusion criteria
The study included health care workers involved in routine healthcare services and those supporting at least one of the six pillars of the task forces at the national or district level. They must have given written consent before participating in the study. The health workers came from the fifteen referral hospitals including, Lira, Gulu, Arua, Soroti, Mbale, Jinja, Masaka, Mbarara, Bombo, Entebbe, Moroto, Hioma, Fort portal, Naguru and Mulago. Health workers came from the district local governments and, others were members of the NTF responding to the COVID-19 pandemic.
Exclusion criteria
Interns, non-medical staff were not involved directly in the COVID-19 response and, health workers who did not consent were excluded from the studies.
Data analysis
Univariate analysis
The data were subjected to a descriptive and inferential analysis using the Statistical Package of STATA® version 16 (StataCorp LLC, Texas USA). Descriptive statistics, including mean, percentages, frequencies, and charts, were used to provide a descriptive account of the study findings. Meaningful patterns were identified from the summarized data to give further insight into the multivariable analysis.
Multivariable analysis
Mental health status had three dependent variables: depression, stress, anxiety. In this study, I used ordered logistics regression to estimate the maximum likelihood to evaluate the probability of each variable of mental health in relation to the health work function impairment. The four outcomes of each dependent variable included normal, mild, moderate and severe or extremely severe mental health state. The base outcome for comparing each dependent variable being “normal” for each dependent variable. The results were reported as more or less likely in relation to the base outcome of each dependent variable while we hold other variables constant.
I estimated three ologit models using each of the six explanatory variables for health work functioning in relating to each of the dependent variable of mental health status of the participants. The results were reported in terms of odds ratio. Eq. 1 below presents the general specification of the cross-sectional models for the study. The study assumed that the accuracy of each dependent variable of mental health status m was a linear function (Mental health f*(m, i)) of the six explanatory variable. The six explanatory variables included cognitive aspects of task execution and general incidents (cognitive), ability to cause incident at work (incident), avoidance behaviors (avoidance), conflicts and irritation by colleagues (conflict), impaired contact with patients and their family (contact) and lack of energy and motivation (motivation). These independent variables were arranged as continuous variables in the analysis for the observations i. I assumed that each independent variable had a set of values for each outcome of every dependent variable of mental health m and that the maximum likelihood of each outcome m is independent of the other outcomes in the variable and the outcome of the other dependent variables of mental health.
βm is the regression coefficient associated with each explanatory variable and the mth outcome of mental health. I also included a vector of control variables X and unobserved variables (error term) εi for all the observations i.
From empirical studies, I identified a range of control variables that was included in vector X. I included socio-demographic characteristics of health workers such as age, gender, marital status, the highest level of education attained, their profession, religion and income. I also included a variable to understand the role of health workers during the response and one to determine whether they participated in any other outbreak before COVID-19. The variables allowed the researchers to control for differences in the observable characteristics of health workers.
The researcher estimated the relationship between mental health and work impairment using Eq. 1 above and including all the control variables mentioned above in an upward process. I used the relative risk ratio to predict the interpretation of results. The results were interpreted as the more or less likely relative risk of the outcome occurring for every unit change in an explanatory variable, holding other factors constant. The results for the model were presented in a table. The level of significance for the variables was determined using a p-value of < 0.05.