Markov switching modeling of covid-19 case counts and associated clinical and epidemiological characteristics in Kermanshah province, western Iran

Background and Objectives: To effectively combat the novel coronavirus (COVID-19), it is crucial to have an accurate understanding of its epidemiological patterns and how it spreads within the community. In this context, a thorough grasp of statistical and epidemiological methods is more important than ever for policymakers and decision-makers. This study utilized a two-state Markov Switching model to analyze daily confirmed COVID-19 case counts, identify periods of outbreak and non-outbreak, and examine the overall clinical and epidemiological characteristics associated with daily case counts during these different periods.
Methods: This time-series study included individuals who presented to hospitals in Kermanshah province with COVID-19-related symptoms from January 31, 2020, to January 5, 2021. Data were extracted from the MCMC system, which is affiliated with the Ministry of Health's emergency services. We used a two-state Markov Switching model to model daily COVID-19 case counts and examine their temporal association with daily aggregate clinical and epidemiological characteristics.
Results: We recorded 9,255 confirmed COVID-19 cases, including 4,912 males and 4,343 females. The mean age was 54.65 years (median: 55.27 years). Among patients, 14.6% had hypertension, 8.8% had diabetes, and 8% had cardiovascular disease. In the outbreak regime (State 2), the daily proportion of confirmed cases with asthma was positively associated with daily case counts, whereas chronic neurological disorders (State 2) and chronic liver disease (State 1) showed negative associations. In the non-outbreak regime (State 1), contact history with a patient with coronavirus was positively associated with daily case counts.
Conclusion: The Markov Switching approach provides a complementary framework for characterizing temporal variation in COVID-19 case counts and identifying outbreak and non-outbreak regimes. Further studies are needed to evaluate its applicability to infectious disease surveillance.