Adaptation and validation of a community worker administratered stroke symptom questionnaire in a peri-urban Pakistani community

Date of Award


Document Type


Degree Name

Master of Science in Epidemiology & Biostatistics (MSc Epidemiology & Biostats)


Community Health Sciences


Stroke is the second leading cause of mortality and the leading cause of disability in the world today. The incidence of stroke is rising in low and middle income countries like Pakistan. However, there is little epidemiologic data from these countries to inform policy decisions. Stroke is one of the few conditions that are amenable to surveillance, but for this, validated tools that can be administered by community workers are needed. Our aim was to translate and adapt one such stroke symptom questionnaire, train community health workers (CHWs) in its administration and validate this package against assessment by two neurologists. Methods: The study was carried out at Ibrahim Hyderi, a pen-urban slum of Karachi. Questionnaire for Verifying Stroke free status (QVSFS) which covers most of the common stroke symptoms was adapted and translated into Urdu. Two community workers received training by a neurologist which entailed teaching regarding stroke pathophysiology, symptomatology and detection. Using purposive sampling, the community workers selected 322 community dwelling subjects. Each worker collected data on these subjects. Two neurologists validated findings by assessing each subject. Sensitivity, specificity, positive and negative predictive values and Cohen's kappa was determined for the CHW administered questionnaire tested against assessment by two neurologists. SPSS version 19.0 was used to analyze data. Results: Mean age of the participants was 56.5 years and 71% were women. Most were of Sindhi origin (82.6%) and employed in the fishing industry (60.9%). The overall sensitivity of the questionnaire for picking up stroke was 77% (95% CI: 64.1%46.9%) and the specificity was 85.8% (95% CI: 83.5%-87.5%). The chance corrected agreement using the Cohen's Kappa statistic was 0.51 (95% Cl: 0.38-0.60). Kappa ranged from 0.37 to 0.58 for each of the seven symptom questions. Hemi anesthesia (72.9%) followed by hemiplegia (64.6%) was the most sensitive symptom. An important finding was the different performances of the two CHWs following the first training session, with one being significantly more sensitive (62.5% vs. 89.6%) and the other more specific (97.1 vs. 83.9%) in identifying individuals with stroke. Following a second training session, data was collected on 10% subjects (n=34) again by each CHW while they were still blinded to the final assessment. There was an improvement seen in performance of both workers which became more uniform and agreement with neurologist improved from moderate to substantial. The feasibility of the process of using CHW for stroke surveillance was high, with good community acceptance (<10% refusal to participate), average time for questionnaire administration being 7 minutes, and most subjects (87.8%) finding no difficulty in understanding questions. Conclusion: We found reasonable sensitivity and specificity and moderate agreement between CHW administered stroke symptom questionnaire and assessment by neurologist. However, the questionnaire with the currently used level of training is operator dependent. Repeat training with feedback is needed before this package of stroke symptom questionnaire and trained CHW can be considered for large scale population screening.

This document is available in the relevant AKU library