This study provides novel evidence about the presence and demographic and socioeconomic correlates of chronic morbidity in the elderly population of transitional Kosovo. Older age and inability to access medical care were the most consistent correlates of chronic morbidity and/or multimorbidity in this study population.
In line with other studies using similar methods for assessing chronic diseases (i.e. self-reported data) [23–26], CVD (including hypertension) was the most prevalent disease among elderly individuals in this sample. Thus, a similar prevalence of CVD has been reported among older people in the region, with a prevalence of 58% reported in Albania  and Serbia , and slightly over 50% in Macedonia . Conversely, a lower prevalence varying from 28% in the Netherlands to 41% in Finland was reported by the FINE study which, however, used diagnosed rather than self-assessed measurement of chronic conditions .
One out of five individuals in this Kosovo sample reported diabetes, which resembles results from Albania (19%) , Germany (17%)  and USA (22.7%) . Studies measuring diagnosed diabetes report a prevalence from 15% (in USA)  to 9% (Italy) and 6% (the Netherlands) . The prevalence of cancer in our study was quite low compared to other countries in the region: 2-3% in Macedonia , 3% in Italy (27), about 4% in Southern Germany , 8% in Finland , and 19% among the American older people . These differences could be partly explained by different methods for assessing the presence of chronic conditions (self-reported vs. diagnosed data) used in different studies, even though it has been argued that self-reports and health care records provide quite comparable estimates for diabetes, but are less concordant for chronic heart disease or other conditions [25, 29]. For example, the beyond chance agreement index (Kappa statistic) has been reported at 0.90  and 0.80  for diabetes, 0.67  and 0.40  for hypertension, but lower for other chronic conditions. Methodological issues aside, a plausible reason for the discrepancies in diabetes prevalence among the elderly people between Kosovo and developed countries such as e.g. the Netherlands could be found in the epidemiology of diabetes which suggests an increasing risk with age, lower education and socioeconomic status [30, 31] – factors which were all more prevalent in the Kosovo sample  compared with the study populations researched elsewhere . Furthermore, another possible explanation for the particularly low prevalence of self-reported cancer in our study might come from a recent survey among cancer patients in Albanian settings, which found that most cancer patients seek medical help only in advanced stages of the illness and the cultural context is largely against diagnosis disclosure .
According to a recent systematic review , the prevalence of multimorbidity among the elderly, defined as the concomitant presence of ≥2 chronic conditions, ranges from 55%-98%, whereas in our study we noted a prevalence of 45%. The discrepancies might be due to different age-groups included in different studies, differences in the number of chronic conditions investigated, differences in the study settings and, as mentioned earlier, different means of assessing the presence of diseases. Furthermore, evidence shows that the prevalence of multimorbidity depends on the study population (e.g. population-based samples vs. primary care users, implying a higher prevalence in the later study population), the nature and the number of chronic conditions included [34, 35]. Indeed, it has been convincingly shown that the prevalence of multimorbity increases as the number of disease items included in the questionnaire increases [34, 35]. However, we tried to overcome this limitation by introducing the following option: “Please mention any other type of chronic diseases not mentioned above”.
Another potential source of variability between studies in estimating morbidity and multimorbidity pertains to education. As the education attainment progresses, the knowledge and understanding of health and disease changes too, leading thus to potentially different reporting. On the other hand, the awareness of people regarding health issues has been rising in general leading to more frequent doctor visits and diagnoses, especially in developed countries. Also, people now can talk more openly about their health problems and this implies greater willingness to report such problems when under study .
In general, the literature reports that morbidity and multimorbidity is significantly higher among older people, women and individuals of a low socioeconomic status [24, 33, 34, 36]. Our findings are in concordance with the international literature as regards the association with sex and age, with women and the oldest-old reporting higher rates of multimorbidity.
Conversely, the inverse association with education was significant in crude analysis only. However, this resembles prior reports from studies conducted elsewhere, which have pointed out not significant relationships between education and the number of chronic conditions and multimorbidity in multivariable-adjusted models [37–40]. Thus, a study including individuals aged ≥18 years reported a non-significant association between education and multimorbidity , whereas a large cohort study among elderly people aged 50–75 years old reported that, upon multivariable-adjustment, the association of multimorbidity with education weakened in men, whereas in women it was not statistically significant .
Morbidity and multimorbidity among the elderly deserves special attention based on previous research which shows that, for certain diseases affecting the heart, lungs and circulatory apparatus, the presence of one or more chronic health conditions is significantly associated with a higher risk of death . Two longitudinal studies reported that persons with poor self-reported health had an early mortality risk and late mortality risk of about three times higher compared to individuals with good health status [42, 43].
About half of the elderly subjects in this study perceived themselves as poor. This might be an indicator of the difficult situation of the elderly population in Kosovo. A prior report including this very study population in Kosovo indicated that the self-perceived poverty was significantly higher among women, those without any formal schooling, urban residents and among the elderly people living alone .
We found significant associations of self-reported poverty with the number of chronic conditions: the higher the poverty level, the higher the proportion of multiple diseases (Spearman’s correlation coefficient = 0.212, P = 0.01; not shown in the tables). Indeed, evidence shows that even after controlling for a number of factors, poverty remains a strong predictor of adults’ health . Education and poverty seem to be part of a vicious circle: low education, which is greatly influenced by unfavourable family circumstances during childhood, might be closely linked to a lower income during adulthood favouring persistent poverty which in turn contributes to poor health outcomes later in life . Since the objective and subjective measures of poverty have been reported to correlate with each-other , self-perceived poverty might explain a part of unfavourable health outcomes among Kosovo elderly people, too. Yet, self-perceived poverty and well-being depend on many factors other than income .
Some of the socioeconomic and demographic determinants of chronic morbidity and multimorbidity among the elderly have been studied extensively, but little is known about other risk factors of multimorbidity including genetic, biological, lifestyle and environmental factors . Another under researched factor which could affect the health status of old people is elderly abuse, which includes “abandonment, emotional abuse, financial or material exploitation, neglect, physical abuse, and sexual abuse of the elderly” . Although a considerable number of studies have highlighted the situation of elderly abuse across different populations, very little evidence is available regarding the prevention of elderly abuse  and how this may affect the health status of older people. Elderly abuse sets an additional heavy barrier on the shoulders of older people: besides co-living with the ageing process and physical limitations that it entails, older people have to cope with the community abuse, which might further deteriorate their health status. Elderly people in Kosovo are a marginalized part of the population  which might imply the existence of elderly abuse. Future investigations should take into account this aspect when assessing the complexity of factors associated with morbidity of this community [36, 48].
In our study, access to medical care was a significant and consistent predictor of both the presence and number of chronic conditions. The access and use of health services depends not only on the need for care, but also on predisposing characteristics (demographic factors, health beliefs) and enabling resources such as the availability of health personnel and health facilities, means of transport, or health insurance . The overwhelming majority of Kosovo elderly people who couldn’t access medical care in this study (almost 90%) pointed to the economic barriers as the main reason for this inability. This is a reflection of the unclear situation of the elderly in Kosovo and the ongoing reforms in the health sector. Although protection of the rights of vulnerable groups and ensuring quality of care is one of the priorities of health reforms in Kosovo, the health system lags behind its optimal state. The health insurance system seems unable to function with half of the population unemployed and a high informality rate [3, 50]. People aged ≥65 years in Kosovo rely on the social security pension (which is quite low and not sufficient to meet their everyday needs) and remittances from their close family working abroad . Furthermore, Kosovo is in urgent need of deep reforms as the armed conflict left the country with a very inefficient health system characterized by a lack of trained personnel and disparities in health force distribution. These factors lead to variations in access to primary care, corruption and informal payments, which are all reflected in unfavourable child and adult health indicators. In this context, the continuous reforming of the health sector has brought up a complex configuration of the stakeholders operating in the health system which contributes to unequal access to health care. The primary health care is still overlooked by health policies which often favour “high-tech” clinical medicine . Furthermore, the private health sector has been expanded rapidly, but private facilities are unaffordable for the elderly . The main barrier to access care is the cost of services, despite the fact that basic health services are supposed to be covered for all citizens. Under-the-table payments put a heavy burden on the shoulders of the poor. Ultimately, the reforms have resulted in lower access to health care for the poorer groups of the society . Under these conditions, little attention is paid to the growing community of the elderly people in Kosovo  which, combined with the inadequacy of financial resources, the economic insecurity and the unclear and unstable development of the health sector, pose a serious barrier for elderly people to access medical care.
As stated by the Centre on Social Disparities in Health , in order to increase the chances of good health one needs to adopt a healthy lifestyle and have access to proper medical care. In a broader context, there is a need to promote a healthier living and working conditions. This should be supported by economic development, reducing poverty and enhancing education .
Our study has several limitations including its cross-sectional design and the differential reporting of chronic diseases among elderly people. We cannot exclude the possibility of reporting bias; however, we do not have sound reasons to assume differential reporting of chronic diseases for the elderly people’s categories differing in demographic and socioeconomic characteristics. More importantly, findings of our study should be interpreted with caution, since the observed associations from cross-sectional studies are not assumed to be causal.