Air Pollution in Kosovo: Short Term Effects on Hospital Visits of Children Due to Respiratory Health Diagnoses

Abstract

The Republic of Kosovo is a small country in the Balkans. The capital city of Pristina hosts most of its population and is situated in a mountain basin with poor air exchange, especially during winter. Domestic heating, road transport, industry and coal-fired power plants contribute to high levels of air pollution. We performed a time-series analysis on effects of particulate air pollution (PM2.5) on respiratory health of children and adolescents, using hospital admission and ambulatory visit numbers from the pediatric university clinic. From 2018 until 2020, daily mean concentrations of PM2.5 ranged between 2.41 and 161.03 µg/m³. On average, there were 6.7 ambulatory visits per day with lower numbers on weekends and during the first COVID-19 wave in 2020. An increase in PM2.5 led to an immediate increase in visit numbers that lasted over several days. Averaged over a full week, this amounted to about a 1% increase per 10 µg/m³. There were, on average, 1.7 hospital admissions per day. Two and three days after a rise in air pollution, there was also a rise in admission numbers, followed by a decline during the consecutive days. This might indicate that the wards were overstressed because of high admission numbers and restricted additional admissions.

Particulate Air Pollution and Primary Care Visits in Kosovo: A Time-Series Approach

“…Here, we report the results of a time-series study on the effects of daily variation in particulate matter (PM2.5) concentration on the number of primary care visits in the municipality of Pristina, Kosovo. In our previous study [ 12 ], we demonstrated that PM2.5 data from the central monitoring station in Pristina display a strong temporal correlation with all of the other monitoring stations in the inhabited areas of Kosovo. Therefore, these data were deemed representative for the whole country.…”

“…We built general additive models (GAM), family: Poisson, in the same way as described in our previous study [ 12 ] following standard procedures [ 15 , 16 , 17 , 18 , 19 , 20 , 21 ]: the number of knots of the spline for the temporal variation was selected based on the minimal residual autocorrelation. The best lag for the temperature and for the temperature on consecutive days was chosen based on the Akaike Information Criterion (AIC) [ 22 ].…”

“…At first, we examined air pollution effects after different lags separately. Pending confirmation by visual inspection of our a priori hypothesis that the lag structure resembled a third-degree polynomial indicating some “harvesting” effect [ 12 ], a third-degree polynomial distributed lag model accounting for seven days (lag 0–lag 6) was implemented.…”

“…Hospital admission data are used to broaden the representativeness of the time-series study [ 11 ] with similar effect estimates as for mortality. We have recently performed such a study on hospital admissions, analyzing admissions to the pediatric clinic in Pristina, Kosovo [ 12 ]. Air pollution effects on the hospital admission data also strongly depend on the type of admission [ 13 ], which is not always available in the register data.…”

“…Air pollution effects on the hospital admission data also strongly depend on the type of admission [ 13 ], which is not always available in the register data. As we have discussed in our paper [ 12 ], the source population of a single hospital is not well defined and therefore, a clear interpretation of the results regarding population health is difficult. Routine data from primary care reported by resident doctors would overcome some of these difficulties, as these data would cover the whole population.…”