This study first evaluated elven-year trends in first-time stroke with comorbidity, and the effect of comorbidity on mortality, length of stay, and hospital cost from 2010 to 2020 in Tianjin, north of China. We have three main foundings: the age group of first stroke patients was concentrated from 55 to 84 years old during the eleven years, and the population of the old elderly (≥ 85 years old) was increasing with the deepening of aging society; there was a decreased trend of those with no and moderate comorbidity, and an increased tendency of patients with severe and very severe comorbidity; those with severe and very severe comorbidity had higher in-hospital and 7-day mortality, longer LOS and more heavy economic burden, especially in the patients aged 55 to 65 years.
We observed that tendency of comorbidity increasing and aging populations was in line with the study in a developed country of Denmark 8, 9. While a more serious aging trend in Denmark was observed with the predominance of first-stroke people aged ≥ 70 years (almost 63%), and patients aged 65 years old was the dominant for those with ischemic stroke. Therefore, it was important to pay attention to impact of comorbidity on stroke in an aging society. Although less attention was paid to comorbidity, there were several previous studies in other countries devoted to comorbidity. Higher CCI scores were generally associated with worse function outcome at hospital discharge and greater 1-year mortality of stroke13, 17. Some national studies with a large sample size concluded that comorbidity was a strong prognosis predicted factor for not only short-term prognosis, but also 5-year mortality regardless of stroke subtype8, 18.
A cohort study in Australia divided 776 stroke patients into high and low CCI scores group and found that a higher CCI score as a risk factor increased in-hospital mortality, LOS, and inpatient cost10, which was consistent with our study. Different from those, the mean age in our study was younger (69.3 years vs 80.1 years), LOS were longer (14 days vs 5.44 days), and mortality in heavy comorbidity burden was lower (19.5% vs 22.1%). The reason for differences may be from the different regions, degree of social aging, national medical development levels, medical insurance policies, and sample size. Compared to the developing countries, Australia had the deeper degree of social aging and better health-care systems. Furthermore, patients aged 80 years old or older had more comorbidities and higher mortality than in those younger than 80 years19, which further supported that our mortality rate is slightly lower than Australia's. However, different sample size may be contributed to the results differences, with 5988 patients in our patients larger than 776 population in theirs.
First stroke inpatient mortality in our study was lower than the national study in our country based on community and a sample size of 0.5 million adults (4.2% vs 11%)3. For one thing, the latter study based on the big data had more regional diversity, younger population (59.3 vs 69.3 years), and higher proportion of hemorrhage stroke than ours (18% vs 4%). For another thing, the national data estimated the 28-day mortality, while in-hospital mortality in ours. Post-hospital death events may result in the increased mortality. Besides, the big data supported that mortality of hemorrhage stroke was higher than that of ischemic stroke, with the ratio of 11% higher than 3% in our study3. A nationwide inpatient data from America also reported that more comorbidities and older age were independently associated with in-hospital mortality20. Other studies found that women were related to the increased risk of in-hospital death20, 21. In our study, there was no significance in the multivariable analysis, the differences of results may attribute to regional, racial differences, and the different sample size.
We further analyzed the relationship between comorbidity and in-hospital mortality by seasonal stratification and found that patients with severe and very severe comorbidity had higher risk than those without comorbidity no matter in which season, which was in line with a published study22. Another 5-year hospital-based study on connection between season and stroke reported that stroke case-fatality rate was the highest in the winter especially in aged ≥ 65 years23. While the seasonality of 7-day mortality was never seen in our study, we made an assumption that there was a time lag effect of mortality. It is reported that pneumonia had a higher prevalence in winter24, and recent infection increased the mortality of stroke25, which may explain the phenomenon of seasonal difference in our study with the older patients with exist of higher proportion of pneumonia.
After exploring in-hospital mortality associated with individual comorbidities in patients with first stroke, we found that patients with pneumonia occupied first place (HR, 15.06, 95% CI, 10.08–22.50, P<0.001), followed by moderate to severe renal (HR, 2.47, 95% CI, 1.50–4.08, P<0.001) and moderate to severe liver disease (HR, 2.58, 95% CI, 1.32–5.08, P = 0.006). The pneumonia may result from dysphagia leading to aspiration pneumonia, acroparalysis leading to long time of stay in bed and hypostatic pneumonia, and climate change in different season leading to respiratory infection. An England study showed that the aspiration pneumonia had a higher short-term mortality than those without aspiration pneumonia26. It was validated effective and practicable to perform an early dysphagia screening by neurologist, speech–language therapists, or well-trained nurses27. The results of several studies were consistent with our findings for association between in-hospital mortality and kidney dysfunction on admission28, 29 and liver dysfunction30. Therefore, these results remind us that patients with moderate to severe renal and liver dysfunction on admission and dysphagia need to be given targeted intervention strategies to improve their prognosis on discharge, especially the reasons resulting in pneumonia.
Different from previous studies31–33, inpatient cost had no difference between ischemic and hemorrhage stroke in our study. This may be because conservative medical treatment without surgery in our neurology department. The older with heavy comorbidity burden tended to spend more money and experience longer LOS, which may be caused by that the older needed to pay money and time for the treatment of comorbidities and complications, such as pneumonia, abnormal renal and liver function. And our study demonstrated above hypothesis from a different angle that hospital cost become more higher with the increasing comorbidity burden from 2010 to 2020. Considering the clinical and economic impact among patients with first stroke with different comorbidity categories, especially in the elder with heavy comorbidity burden, the clinical physicians should systematically summarize the impact of age, sex, primary stroke disease, and comorbidity burden calculated by Charlson's comorbidity index, and propose management strategies aimed at reducing adverse effects as well as social and economic costs34.
Several limitations of our study should be acknowledged. First, we conducted the analysis on our admission data according to ICD codes. It is possible that the exist of coding errors or omissions of diagnoses and complications resulted inaccurate classification of comorbidity. Secondly, some important covariates was unavailable, such as body mass index (BMI), personal history (smoking and drinking), subtypes and severity of stroke, thrombolytic therapy, drugs for comorbidities and laboratory results. Thirdly, to facilitate the analysis of in-hospital mortality, we assumed that discharged patients were still alive during the study period35. Fourth, although the average LOS was 14 days according to the limitation of medical insurance policy, a longer hospital stay was still required among patients with heavy comorbidity burden. Finally, there was a selection bias considering that our study was a retrospective study at a single center, which may limit the generalisability of our findings. Despite these limitations, our study has its own strength and important implications. The results of a large sample from a comprehensive hospital are representative in Tianjin,and in north of China to some extent due to the similar climate, diet and lifestyle.