Methodological considerations in establishing and maintaining longitudinal health workforce studies: Lessons learned from the WiSDOM cohort in South Africa

Figures & data

Figure 1. Key steps in the establishment of the WiSDOM cohort

Figure 2. Stakeholders consulted in the WiSDOM health professional cohort study

Table 1. Summary of WiSDOM cohort study methods, 2017–2020

Method	2017 Baseline	2018 Follow-up	2019 Follow-up	2020 Follow-up
Study population	All final year health professional students (n=571)	Baseline survey respondents (n=511)	Baseline survey respondents, excluding refusals	Baseline survey respondents excluding refusals
Recruitment strategies	Consultative workshop Individual consultations with key stakeholders Customised video, posters and adverts WiSDOM cohort branding	Email to alert cohort members of 2018 survey Circulation of 2 policy briefs	Email to alert cohort members of 2019 survey Circulation of flyer with select 2018 results	Email to alert cohort members of 2020 survey Circulation of newsletter with select 2019 results and reflections from cohort members
Informed consent andvoluntary participation	Yes	Yes	Yes	Yes
Data collection tool	15-20 minute self-administered questionnaire (SAQ) to obtain detailed baseline information Extensive contact details to enable future follow-up	15-20 minute SAQ on changes since baseline survey Verification of contact details	5-10 minute shortened SAQ on changes since 2018 survey Verification of contact details	5-10 minute shortened SAQ on changes since 2019 and experiences during Covid-19 pandemic Verification of contact details
Data collection method	Data collection at Wits Health Sciences Campus in computer laboratory or e-learning room using REDCap Separate data collection sessions for each of 8 professional groups	Web-based survey using REDCap	Web-based survey using REDCap	Web-based survey using REDCap
Optimising response rates	Department of Health talk on community service or internship Motivational talk by young, practising health professional Refreshments Opportunity to complete the SAQ online	Weekly email and short message service (SMS) reminders over a period of 8 weeks Follow-up telephone survey by trained external service provider	Weekly email and SMS reminders over a period of 8 weeks Follow-up telephone survey by team members	Weekly email and SMS reminders over a period of 8 weeks Follow-up telephone survey by trained external service provider
Retention strategies	Feedback meeting in November 2017 Personalised invitations to feedback meeting	Dedicated website Individual birthday message Communicating results Honorarium of R200 voucher (~13 USD)	Dedicated website Individual birthday message Communicating results Honorarium of R200 voucher (~13 USD)	Dedicated website Individual birthday message Communicating results Honorarium of R200 voucher (~13 USD)

REDCap = Research Electronic Data Capture

SAQ=Self-administered questionnaire

Figure 3. Strategies to maintain and retain the WiSDOM cohort

Table 2. Completion rates in the WiSDOM cohort study methods, 2017–2020

	2017: Baseline (n)	Completion Rates						Total Refusals
		Annual				From Baseline
		2017: Baseline response (%)	2017 →2018	2018 →2019	2019 →2020	2017 →2019	2017 →2020
PH	58	95.1%	96.6%	94.6%	98.2%	91.4%	94.8%	1.7%
MD	282	85.7%	66.0%	80.6%	88.6%	53.2%	58.2%	6.7%
DT	17	94.4%	94.1%	100.0%	100.0%	94.1%	94.1%	5.9%
NS	21	95.5%	90.5%	100.0%	100.0%	90.5%	90.5%	0.0%
OT	36	97.3%	94.4%	73.5%	76.5%	69.4%	72.2%	2.8%
PT	46	93.9%	97.8%	88.9%	93.2%	87.0%	89.1%	4.3%
CA	44	93.6%	100.0%	90.9%	100.0%	90.9%	100.0%	0.0%
OH	7	87.5%	100.0%	85.70%	100.0%	85.7%	100.0%	0.0%
TOT	511	89.5%	79.6%	85.7%	91.9%	68.3%	72.8%	4.7%

Table legend: CA = clinical associate; DT = dentist; MD = medical doctor; NS = nurse; OH = oral hygienist; OT = occupational therapist; PH = pharmacist; PT = physiotherapist

Table 3. Logistic regression of medical doctor respondents vs non-respondents

Category	Odds Ratio	[95% CI]	P-Value
Gender	—
Male
Female	1.442	[0.837; 2.484]	0.187
Marital status
Single	—
Married	0.718	[0.285; 1.803]	0.481
Age
Age	0.968	[0.846; 1.106]	0.631
Any children
No	—
Yes	0.559	[0.125; 2.491]	0.446
Born in SA
No	—
Yes	1.375	[0.589; 3.205]	0.461
Area where born
Urban	—
Rural	0.818	[0.307; 2.176]	0.687
Race
White	—
Black African	2.011	[1.023; 3.951]	0.043
Coloured	0.499	[0.089; 2.795]	0.429
Indian	2.593	[1.163; 5.778]	0.020
Profession 1^st choice
No	—
Yes	0.363	[0.146; 0.904]	0.051
Composite SES
Q1 poorest	—
Q2	0.784	[0.294; 2.086]	0.625
Q3	0.817	[0.286; 2.326]	0.705
Q4	0.843	[0.289; 2.452]	0.754
Q5 richest	0.844	[0.270; 2.635]	0.770

Observations 270

Pseudo R² 0.05

Data availability statement

We cannot share data publicly because this is an active cohort study of health professional graduates. The Human Research Ethics Committee (Medical) of the University of the Witwatersrand has imposed restrictions because of the confidential and sensitive nature of the data. https://www.wits.ac.za/ethics/human-research-ethics-committee-medical.