Introduction
Cases of emergency hospital admissions and overnight hospitalisation are increasing significantly in developed countries, including Canada (Galenkamp, Deeg, de Jongh, Kardaun, & Huisman, Reference Galenkamp, Deeg, de Jongh, Kardaun and Huisman2016; Hallgren et al., Reference Hallgren, Fransson, Kåreholt, Reynolds, Pedersen and Dahl Aslan2016; McCusker et al., Reference McCusker, Ionescu-Ittu, Ciampi, Vadeboncoeur, Roberge, Larouche and Pineault2007). Increased life expectancy and reduced birth rates have contributed to the rise in the proportion of seniors (persons aged 65 years and older) in Canada, as well as in other developed countries (Moore & Rosenberg, Reference Moore and Rosenberg2001; Rosenberg & Moore, Reference Rosenberg and Moore1997). Canada’s ageing population comes with vital health, social, and economic consequences. Moore, Rosenberg, and Fitzgibbon (Reference Moore, Rosenberg and Fitzgibbon1999) noted that the growth in the older population comes with a change in health services and support. Advanced ageing is associated with susceptibility to degenerative chronic health conditions such as hypertension, diabetes, cancer, and musculoskeletal disorder (Bähler, Huber, Brüngger, & Reich, Reference Bähler, Huber, Brüngger and Reich2015; Šteinmiller, Routasalo, & Suominen, Reference Šteinmiller, Routasalo and Suominen2015).
Research suggests that about 20 per cent of Canadian seniors have been diagnosed with hypertension while another 20 per cent are at risk of developing hypertension (Campbell et al., Reference Campbell, Young, Drouin, Legowski, Adams, Farrell and Tobe2012). A recent report indicates almost half of Canada’s population will develop cancer in their lifetime, and a quarter will die of the disease (Canadian Cancer Society’s Advisory Committee on Cancer Statistics, 2017). Economically, the rise in the number of chronic health conditions is also associated with high medical care expenditures (Bähler et al., Reference Bähler, Huber, Brüngger and Reich2015; Bail et al., Reference Bail, Goss, Draper, Berry, Karmel and Gibson2015; Lehnert et al., Reference Lehnert, Heider, Leicht, Heinrich, Corrieri, Luppa and König2011; McCabe et al., Reference McCabe, McElroy, Cournane, Byrne, O’Riordan, Fitzgerald and Silke2017). In 2010, hypertension-associated medical care in Canada was estimated to be $13.9 billion – accounting for 10.2 per cent of the country’s total health care expenditure; this cost is expected to increase to $20 billion by 2020 (Weaver et al., Reference Weaver, Clement, Campbell, James, Klarenbach, Hemmelgarn and McBrien2015).
Community-dwelling seniors constitute a significant proportion of persons who use emergency care services (Lyon, Lancaster, Taylor, Dowrick, & Chellaswamy, Reference Lyon, Lancaster, Taylor, Dowrick and Chellaswamy2007; McCusker et al., Reference McCusker, Ionescu-Ittu, Ciampi, Vadeboncoeur, Roberge, Larouche and Pineault2007). Comparatively, seniors have a higher rate of unplanned hospital admission than younger persons. Although seniors in developed countries constitute a relatively smaller proportion of the population, they often tend to have higher acute bed occupancy and emergency admissions (Ennis et al., Reference Ennis, Larson, Grothaus, Helfrich, Balch and Phelan2014; Lyon et al., Reference Lyon, Lancaster, Taylor, Dowrick and Chellaswamy2007). They are also more likely to have a return visit for treatment after their initial visit (Galvin et al., Reference Galvin, Gilleit, Wallace, Cousins, Bolmer, Rainer and Fahey2017; Lee et al., Reference Lee, Schwindt, Langevin, Moghabghab, Alibhai, Kiss and Naglie2008; Lyon et al., Reference Lyon, Lancaster, Taylor, Dowrick and Chellaswamy2007). Seniors using emergency health care facilities for unplanned visits are also highly vulnerable to hospital-acquired infections (Cookson & Laudicella, Reference Cookson and Laudicella2011; Wallace et al., Reference Wallace, Stuart, Vaughan, Bennett, Fahey and Smith2014). Studies show that chronic health conditions together with the burden of multi-morbidity among seniors increase their risk of overnight hospital admission and the duration of hospitalisation (Bähler et al., Reference Bähler, Huber, Brüngger and Reich2015; Galenkamp et al., Reference Galenkamp, Deeg, de Jongh, Kardaun and Huisman2016). Despite seniors’ vulnerability and risk of hospital-acquired infections, existing data shows that older persons in Canada have longer wait times at emergency departments (36.3 hours) compared to their younger counterparts (31.4 hours) (Canadian Institute for Health Information, n.d.).
The literature also indicates that socio-economically vulnerable persons have higher rates of hospitalisation (Clay, Roth, Safford, Sawyer, & Allman, Reference Clay, Roth, Safford, Sawyer and Allman2011; Landi et al., Reference Landi, Onder, Cesari, Barillaro, Lattanzio, Carbonin and Bernabei2004). Some researchers are of the view that, beyond medical conditions, socio-economic factors are vital determinants of the rate of hospitalisation or emergency visits. Evidence in the literature suggests that socio-economically vulnerable persons often lack the resources to make better health and health care decisions (Buckles, Hagemann, Malamud, Morrill, & Wozniak, Reference Buckles, Hagemann, Malamud, Morrill and Wozniak2016; Clarke & Latham, Reference Clarke and Latham2014). Socio-economically vulnerable persons may not take required treatment or seek diagnosis necessary for early detection of possible hospitalisation conditions (Landi et al., Reference Landi, Onder, Cesari, Barillaro, Lattanzio, Carbonin and Bernabei2004). Although there are considerable differences among the older population, seniors are relatively considered socio-economically vulnerable compared to younger people. Their socio-economic vulnerability stems from the loss of wage income and work-related health benefits due to retirement. For instance, in Canada, a significant proportion of seniors do not have adequate access to preventive oral health compared to the younger population (Amegbor & Rosenberg, Reference Amegbor and Rosenberg2018; McNally, Reference McNally2005; Thompson, Cooney, Lawrence, Ravaghi, & Quiñonez, Reference Thompson, Cooney, Lawrence, Ravaghi and Quiñonez2014). Age, gender, and place of residence have been identified as social factors that significantly predict the risk of hospitalisation among the population (Hallgren et al., Reference Hallgren, Fransson, Kåreholt, Reynolds, Pedersen and Dahl Aslan2016; Neufeld, Viau, Hirdes, & Warry, Reference Neufeld, Viau, Hirdes and Warry2016).
Notwithstanding this prevailing knowledge, there are limited studies on the effect chronic health conditions and socio-economic factors have on overnight admission and length of stay among Canadian seniors. Existing studies focused on specific aspects of the country’s older adult population: rural residents in Ontario (Neufeld et al., Reference Neufeld, Viau, Hirdes and Warry2016), residents in Hamilton receiving home care (Smith et al., Reference Smith, Chan Carusone, Willison, Babineau, Smith, Abernathy and Loeb2005), and users of emergency care services in Quebec (McCusker et al., Reference McCusker, Ionescu-Ittu, Ciampi, Vadeboncoeur, Roberge, Larouche and Pineault2007).The target population of these three studies differ considerably; consequently, it is difficult to appreciate the effect of chronic health conditions and socio-economic status on overnight hospital stay among Canadian seniors from these existing studies. In view of the gap in research, we undertook this study to examine the influence of age-associated chronic health conditions and socio-economic factors on the risk of hospitalisation and longer hospital stays among community-dwelling Canadian seniors. The findings will further current discourse on ageing in the community, chronic morbidity, and risk of hospitalisation in Canada. Insights from the findings can inform health care policies and programmes for seniors, as well as improve community-based support and hospital-based services for the country’s senior population.
Data and Methods
The study used data from the 2014 Canadian Community Health Survey (CCHS) annual component. The CCHS is a statistically representative cross-sectional survey of residents in Canada aged 12 years and older. The survey excludes persons living on reserves and other Aboriginal settlements; full-time members of the Canadian Forces, institutionalised persons, and persons living in the Quebec health regions of Région du Nunavik and Région des Terres-Cries-de-la-Baie-James. The excluded population represent 3 per cent of the Canadian population aged 12 years and older. The CCHS, which collects a sample of 65,000 respondents annually, uses a multi-stage sampling of dwellings selected from an area frame or clusters stratified by the 110 health regions (HR). A sample is allocated to each province according to their population size and the number of HRs in the province. The survey collects information related to health status, health care utilisation, health determinants, and socio-demographic details. The sample we used in the study includes all CCHS respondents aged 65 years or older. Statistics Canada, government policy documents, and existing studies (McNally, Reference McNally2005; Rosenberg & Moore, Reference Rosenberg and Moore1997; Yao & MacEntee, Reference Yao and MacEntee2014) classify persons in the selected age group as seniors – that is, elderly persons aged 65 years or older.
Measures
We used two variables from the CCHS as outcomes or dependent variables in our study. The first dependent variable was derived from the CCHS question, “In the past 12 months, have you been a patient overnight in a hospital, nursing home or convalescent home?”, with the following response categories, coded and labelled as “1 = Yes” and “0 = No”. The second dependent variable was a total count of the number of nights a respondent stayed as a patient in a hospital, nursing home, or convalescent home. It was derived from the follow-up question “For how many nights in the past 12 months?” The minimum count was zero corresponding to the number of respondents who were not patients overnight in a hospital, nursing home, or convalescent home, as well as patients who were discharged the same day after receiving emergency care.
Our study adopted the Andersen (Reference Andersen1995) behavioural model of health utilisation as a conceptual framework. We used two sets of independent variables as predictors of overnight hospital admission. The first group of predictors were measures of chronic conditions, which the CCHS collects as self-reported data. These chronic health conditions are conceptualised as need factors, which necessitate the use of health care services. We used eight chronic medical conditions in our study analysis as determinants of overnight admission and length of stay. The chronic conditions we used were hypertension, cancer, fibromyalgia, chronic obstructive pulmonary disease, diabetes, heart diseases, stroke, and bowel disorder. These conditions are not peculiar only to older persons; however, they are generally considered as senescent-associated conditions (Divo, Martinez, & Mannino, Reference Divo, Martinez and Mannino2014).
The second set of predictor variables were age, sex, the highest level of education, and total household income per annum. Age, sex, and education are conceived as predisposing factors, whereas household income is conceptualised as an enabling factor. Age was categorised into four groups: 65–69 years, 70–74 years, 75–79 years, and 80 years or older. Education had four response categories: less than secondary, secondary school, some post-secondary, and post-secondary level. Existing studies point to the influence of socio-economic factors such as age, sex, education, and income on health and health behaviours (Andersen, Reference Andersen1995; Hallgren et al., Reference Hallgren, Fransson, Kåreholt, Reynolds, Pedersen and Dahl Aslan2016; Lehnert et al., Reference Lehnert, Heider, Leicht, Heinrich, Corrieri, Luppa and König2011). For instance, education is hypothesised as promoting good health through its influence on health behaviours (Dupre, Reference Dupre2007; Lynch & von Hippel, Reference Lynch and von Hippel2016). Income, on the other hand, can foster better health through its enabling influence on access and use of preventive health care services (Marmot, Friel, Bell, Houweling, & Taylor, Reference Marmot, Friel, Bell, Houweling and Taylor2008; Prag, Mills, & Wittek, Reference Prag, Mills and Wittek2013).
Analysis
We performed univariate analysis to examine the proportional distribution of the study variables. The descriptive statistics report calculated percentages for categorical variables such as chronic conditions, socio-economic factors, and overnight hospital stay. Also, the mean length of overnight hospital stay was calculated for the second dependent which, as noted, is a count variable. Multivariate regression analyses were used to examine the relationship between the study’s predictor variables, overnight admission and length of overnight hospital stay; using an incremental approach. The first level examines the relationship between chronic conditions (need factors) and the outcomes (overnight hospital stay and length of stay); the second level included socio-economic (predisposing and enabling) factors in the list of predictors.
The study’s first dependent variable was dichotomous, and results of the descriptive statistics show that it is not evenly distributed. Considering this, the complementary log-log function was employed in the logistic analysis instead of logit or probit analysis. Like the logit and the probit functions, the complementary log-log function fits maximum likelihood models with dichotomous dependent variables coded as “0” and “not 0” (Martuzzi & Elliott, Reference Martuzzi and Elliott1998; StataCorp, 2013). However, the complementary log-log function relaxes symmetrical distribution assumption of both logit and probit models (Hedeker, Reference Hedeker, de Leeuw and Erik2008; StataCorp, 2013). That is, it relates to the Gompertz distribution in which the distribution is flexible and can be skewed to the right and to the left (StataCorp, 2013). This logistic function is ideal because the probability of an outcome occurring is very small or very large (Davies, Cifaldi, Segurado, & Weisman, Reference Davies, Cifaldi, Segurado and Weisman2009) as observed in the first dependent variable.
The second dependent variable was a count of days stayed at the hospital. A scan of the second dependent variable shows that the variable does not have a normal distribution; thus, an ordinary least square (OLS) regression is inappropriate for analysis. Generally, a Poisson regression is used in such situations; however, a summary test of the variables shows our count outcome suffers from over-dispersion created by an excess zero. The variable’s variance is almost 18 times its mean (1.134). Ignoring the excess zero may lead to bias in the estimated parameters and standard error in the regression model (Zuur, Leno, Walker, Saveliev, & Smith, Reference Zuur, Leno, Walker, Saveliev and Smith2009). A zero-inflated negative binomial (ZINB) regression analysis was employed to examine the length of overnight hospital stay. ZINB generates two separate models. The first model is a negative binomial model that predicts respondents in “not certain zeros” – that is, respondents who are not always certain to have zero days’ overnight hospital stay.
The second model is a logistic model that predicts respondents in “certain zeros”, or respondents who will always have zero days’ overnight hospital stay. The negative binomial part is a model of the number of days of overnight hospital stay, whereas the logit part models a probability of not having an overnight hospital stay. In both parts of the ZINB model, the study reports the exponentiated coefficients. The exponentiated coefficients of the negative binomial part are synonymous to the incidence-rate ratio (IRR) and that of the logit part of the model is the odds ratio (OR). Cases with missing information for our study variables were generally low (between 0.06% for household income and 2.13% for educational level). Thus, cases with missing values for the study variables were excluded. The multivariate analyses were weighted using the master weight provided by Statistics Canada in the CCHS. This weight was adjusted to account for the use of a subsample: that is, respondents aged 65 years and older. The data were prepared and analysed using IBM SPSS Statistics (version 24) and STATA (version 14.2MP) by StataCorp (College Station, TX).
Results
Table 1 presents the descriptive summary of the characteristics of the study participants. A clear majority (86.75%) of Canadian seniors did not experience overnight hospital admission. The average length of stay for overnight admission among those admitted was 1.103 days while the range for overnight admission stay was 0 days to 31 days. Hypertension (48.69%) and arthritis (45.65%) were relatively common chronic health conditions among seniors surveyed. With respect to hypertension and cancer, 8.62 per cent and 16.01 per cent of sampled seniors indicated they had previously been diagnosed with hypertension and cancer respectively; however, they did not indicate they had these conditions currently. Approximately, a third (32.66%) of sampled seniors were between the ages of 65 and 69. Most sampled seniors were females (58.47%), had a post-secondary level of education (45.47%), and had an annual household income of $20,000 to $39,999 (36.22%).
Table 1: Descriptive summary of study variables (n = 18,491)
a Mean number of days of hospitalisation
Determinants of OHA – Multivariate Results
Table 2 presents the findings of the first multivariate analysis. Need variables defined in this study as chronic health conditions (Model 1) had statistically significant relationships with overnight admission among community-dwelling Canadian seniors. Seniors with chronic health conditions or with a history of having had chronic health issues were more likely to have had overnight hospital stays compared to seniors who did not have or never had chronic health conditions. Seniors who indicated they have hypertension and those who had previously been diagnosed with hypertension were 1.188 times and 1.244 times (p < .01) more likely to have had overnight admissions compared to their counterparts who did not have cases or a history of hypertension. Similarly, seniors with cancer and those previously diagnosed with cancer were 2.109 times and 1.404 times (p < .01) more likely to have had an overnight admission compared to those without cancer or no history of a cancer diagnosis. These observed relationships were still statistically significant after controlling for socio-economic status (Model 2). There were no significant changes in the odds ratios of seniors with chronic health conditions or with a previous history of chronic health reporting overnight admission compared to their healthier counterparts. For instance, the odds ratio of seniors with hypertension and those with a previous history of hypertension reporting overnight admission were 2.060 times (compared to 2.109 times in Model 1) and 1.193 times (compared to 1.244 times in Model 1).
Table 2: Overnight stay at hospital
Note. Robust standard error in parentheses; ***p < .01. **p < .05. *p < .1.
With respect to predisposing and enabling factors, the findings demonstrate that socio-economically vulnerable seniors were more likely to have had overnight admissions compared to seniors of higher socio-economic status. Age-wise, the result shows an increase in the odds ratio of reporting overnight admission with an increase in age group. Seniors in the age groups 70 to 74 years, 75 to 79 years, and 80 years or more were 1.101 times (p < .1), 1.220 times (p < .01), and 1.396 times (p < .01) more likely to have had an overnight stay respectively, compared to seniors in the 65-to-69-year-old age group. Female seniors were less likely to have had overnight admission compared to their male counterparts (OR = 0.906, p < .05). Canadian seniors with an educational status below secondary education were more likely to have had overnight admission compared to those with a post-secondary level of education (OR = 1.086, p < .1). Seniors living in lower income–earning households were more likely to have had an overnight admission compared to those living in a household with annual income of $80,000 or more.
The odds ratio of overnight admissions increased with a decrease in annual household income; the size of the beta coefficient estimate increased by ~30 per cent (Model 2, Table 2). The results of the model statistics show that both Model 1 and Model 2 fit well for determining overnight admission among seniors in Canada as indicated by the Wald’s statistics. The pseudo R-squared estimates indicate that in Model 1 the chronic health conditions explain approximately seven per cent of the variance in overnight admission. In Model 2, chronic health conditions and socio-economic factors explained about eight per cent of the variance in overnight admission. The Akaike’s Information Criterion (AIC) and Bayesian Information Criterion (BIC) estimates of the models indicate that Model 2 is statistically preferable to Model 1.
Determinants of Length of Overnight Hospital Stay – Multivariate Results
Table 3 shows the result of the ZINB analysis. The first component displays the result of the negative binomial model that determines the frequency of overnight admission stay. An incidence-rate ratio above 1 indicates a factor increase in the number of days a senior spent as an overnight patient whereas an incidence-rate ratio below 1 indicates a factor decrease in the number of days a senior spent as an overnight patient. The second component (the logistic model) shows the likelihood of not having overnight hospital admission. An OR of above 1 indicates a greater likelihood of not having overnight admission whereas an OR below 1 shows a lesser likelihood of not having an overnight admission.
Table 3: Days of overnight stay at hospital
Note. IRR = incidence rate-ratio, NB = negative binomial, OR = odds ratio; Robust standard error in parentheses; ***p < .01. **p < .05. *p < .1.
In the negative binomial part, the results show that hypertension, cancer, diabetes, stroke, and bowel disorder – as need factors – were significantly associated with the risk of longer overnight hospital stays. Canadian seniors who had a previously diagnosed case of hypertension compared to those without (or with no history of) hypertension were 1.250 times (p < .05) more likely to have a longer overnight hospital stay. Seniors with cancer compared to those without cancer or with no history of cancer were 1.153 times (p < .1) more likely to have a longer overnight hospital stay. Canadian seniors with diabetes, stroke, and bowel disorder were 1.143 times (p < .05), 1.424 times (p < .01), and 1.170 times (p < .1) more likely to have longer overnight hospital stays respectively, compared to their counterparts who did not have these chronic conditions. In the logistic regression part of Model 1, the findings indicate that all chronic health conditions examined in this study were significantly associated with the probability of having zero days overnight hospital stay. In other words, these seniors are more likely to be hospitalised for a day or more. Canadian seniors with chronic health conditions and previously diagnosed cases of chronic health were less likely to have zero days of overnight hospital stay. For instance, the OR of seniors with stroke and diabetes having no overnight admission stay after visiting the hospital were 0.415 times (p < .01) and 0.853 times (p < .05) less.
In Model 2, the observed relationships did not change when socio-economic factors were introduced as co-variates (Table 3). The only notable change was fibromyalgia which became significantly associated with a greater risk of longer overnight hospital stays (IRR = 1.251, p < .1). Similar to the observed relationship in Model 1 (the logistic part), seniors with chronic health conditions and those with previous history of chronic health were less likely to have zero days of overnight hospital stay compared to their healthier counterparts. The findings in Model 2 also show that age as a predisposing factor and household income as an enabling factor were significantly associated with frequency of overnight hospital stay (negative binomial part). Seniors in the age groups of 75 to 79 years and 80 years or older were 1.317 times (p < .01) and 1.556 times (p < .01) greater to have longer overnight hospital stays. With respect to household income, seniors living in households with annual income of $20,000 to $39,999 and less than $20,000 were 1.266 times (p < .05) and 1.340 times (p < .01) greater to have longer overnight hospital stays compared to those living in households with an annual income of $80,000 or more.
The logistics part shows that age, sex, and income were significantly associated with the likelihood of having zero days of overnight hospital stay. Female seniors compared to males were 1.115 times (p < .1) more likely to have zero days of overnight hospital stay, given that the other variables in the model held constant. Seniors living in households with annual income less than $20,000 were 0.687 times (p < .01) less likely to have zero days of overnight hospital stay, given that the other variables in the model held constant (Table 3). The results of Wald statistics show that both Model 1 and Model 2 fit well for determining the length of overnight stay among seniors in Canada; however, the AIC estimates of the models indicate that Model 2 is statistically preferable to Model 1, given the former’s low AIC value.
Discussion
This study sought to examine the effect of chronic health conditions and socio-economic status on overnight admission and length of stay. The prevalence of overnight admission among community-dwelling Canadian seniors was 13.25 per cent. The average length of stay was approximately a day (1.103 days). As previously noted, the research on hospitalisation and the risk of longer hospital stay among community-dwelling seniors is limited; thus, relating these figures to existing evidence is nearly impossible. Smith et al. (Reference Smith, Chan Carusone, Willison, Babineau, Smith, Abernathy and Loeb2005) in their study reported that, of seniors receiving home care in Hamilton, Ontario, 70 per cent and 50 per cent were likely to have been hospitalised and to visit the emergency department respectively. The higher proportion of hospitalisation and emergency department visit in their study can be attributed to its focus on seniors receiving home care services. This group of seniors are usually frail, increasing their risk of hospitalisation. Thus, it is difficult to compare the findings of the two studies to due to the difference in the target population.
The findings of our study show that chronic health conditions as need factors are significantly associated with the likelihood of community-dwelling seniors being admitted overnight in a hospital. All chronic health conditions examined in this study remained significantly associated with overnight admission after adjustment for socio-economic factors. This finding is consistent with evidence in existing studies on chronic health conditions and the hospitalisation. The findings of these studies show that the incidence of overnight or unplanned hospital admission is higher among seniors with chronic health conditions (Gibson, Segal, & McDermott, Reference Gibson, Segal and McDermott2013; Takahashi et al., Reference Takahashi, Ryu, Hathcock, Olson, Bielinski, Cerhan and Juhn2016). Age-associated chronic health conditions – such as hypertension, cancer, stroke, and other cognitive impairments – are associated with both higher health care utilisation and the risk of unplanned hospital admission. Evidence in studies in the United Kingdom and Canada suggests that seniors with chronic health conditions have a higher likelihood of hospital admission than their peers without such health problems (Lyon et al., Reference Lyon, Lancaster, Taylor, Dowrick and Chellaswamy2007; Neufeld et al., Reference Neufeld, Viau, Hirdes and Warry2016; Smith et al., Reference Smith, Chan Carusone, Willison, Babineau, Smith, Abernathy and Loeb2005). Smith et al. (Reference Smith, Chan Carusone, Willison, Babineau, Smith, Abernathy and Loeb2005) in their study observed that community-dwelling seniors with congestive heart failure and cancer receiving home care were 6.38 times and 4.17 times, respectively, more likely to be hospitalised compared to those who did not have such conditions. In a more general senior population study in the United Kingdom, the odds ratio for hospitalisation among seniors with heart problems was 1.440 times (Lyon et al., Reference Lyon, Lancaster, Taylor, Dowrick and Chellaswamy2007).
Although all forms of chronic health conditions increase the likelihood of overnight hospital admission, the risk of longer hospital stays is not significantly associated with all chronic conditions, which is a major strength of our study. Cardiovascular conditions, such as hypertension, cancer, diabetes, and stroke are consistently associated with the risk of longer hospital stays. Other chronic conditions, such as bowel disorder and fibromyalgia, only became significantly associated with longer hospital stays by controlling or not controlling for socio-economic factors. The evidence suggests that promoting a healthy lifestyle and active ageing that minimises the risk of cardiovascular disease can lead to reducing the risk of longer hospital stay among community-dwelling Canadians.
Another interesting finding in this study was that older persons previously diagnosed with hypertension had a greater risk of longer hospital stay. However, the risk for those currently diagnosed with hypertension was not statistically significant (Table 3). The increased risk of longer hospital stay may be attributed to non-adherence by seniors either to treatment or to prescribed lifestyle for those previously diagnosed with hypertension. Evidence in existing studies shows that compliance with anti-hypertensive treatment is generally high among patients newly diagnosed although adherence wanes over time (Mazzaglia et al., Reference Mazzaglia, Mantovani, Sturkenboom, Filippi, Trifirò, Cricelli and Caputi2005; Van Wijk, Klungel, Heerdink, & de Boer, Reference Van Wijk, Klungel, Heerdink and de Boer2005).
Socio-economic factors are important predictors of health and health outcomes. Persons of lower socio-economic status have an increased risk of poor health outcomes and increased likelihood of hospital admission (Clay et al., Reference Clay, Roth, Safford, Sawyer and Allman2011; Landi et al., Reference Landi, Onder, Cesari, Barillaro, Lattanzio, Carbonin and Bernabei2004). The study shows that socio-economically vulnerable persons such as the very old (persons aged 80 years and above), less educated, and seniors in lower income households have a higher odds ratio of overnight hospital stay (Table 2). This finding is consistent with previous studies in overnight admission and health care utilisation. Socio-economic challenges create poor health and health-care-seeking behaviour. Studies in the United Kingdom and Ireland report variations in hospital admission across places due to differences in local socio-economic status (Cournane et al., Reference Cournane, Conway, Byrne, O’Riordan, Coveney and Silke2017; O’Cathain et al., Reference O’Cathain, Knowles, Maheswaran, Pearson, Turner, Hirst and Nicholl2013).
Limited financial resources and health insurance coverage can significantly impede the ability of seniors to seek timely care, thus increasing their risk to being admitted to emergency care or unplanned overnight admission (Marmot et al., Reference Marmot, Friel, Bell, Houweling and Taylor2008; Prag et al., Reference Prag, Mills and Wittek2013). Formal education influences the ability to adopt new information, promote healthy behaviour, increase wealth level, and individual economic returns (Dupre, Reference Dupre2007; Lynch & von Hippel, Reference Lynch and von Hippel2016). Insight from existing studies show that persons with higher education engage in healthier behaviour, including less smoking and a more-active lifestyle (exercising), reducing their mortality and morbidity risk significantly compared to less-educated individuals (Buckles et al., Reference Buckles, Hagemann, Malamud, Morrill and Wozniak2016; Cutler & Lleras-Muney, Reference Cutler and Lleras-Muney2010; Jürges, Reinhold, & Salm, Reference Jürges, Reinhold and Salm2011). Thus, we postulate that less-educated Canadian seniors may not have the socio-economic and behavioural benefits of formal education, thereby contributing to their risk of ill health (both physical and psychosocial), which would invariably increase their odds ratio of incurring overnight admission.
The results, however, of multivariate ZINB analysis demonstrate that only age and income are associated with an increased risk of longer hospital stay. The risk of longer of hospital stays was significant among seniors aged 75 years and older and those living in lower-income households (Table 3). Our study shows that among community-dwelling seniors in Canada, ageing and economic deprivation may significantly contribute to longer periods of hospitalisation. Although Canada’s universal health care coverage program provides non-fee-paying access to physician and hospital-based services, the evidence suggests persons of lower income households have less contact with general practitioners (Asada & Kephart, Reference Asada and Kephart2007). Less-frequent visits to a general practitioner means that early signs of chronic conditions could go undetected, thus raising the risk of complications and leading to a prolonged hospital admission or stay. The results of this study and others on health care utilisation in Canada suggest improving the socio-economic status of community-dwelling seniors can significantly reduce the risk of hospital admission and prolonged hospital stays.
Study Strengths and Limitations
To the best of our knowledge, this study is the first to investigate the effect of chronic health and socio-economic status on overnight hospitalisation among seniors in Canada nationwide. Unlike previous studies in the country that focused on a specific location (city, rural, or a province), and a smaller subsample of seniors, our study uses a national database, Canadian Community Health Survey (CCHS), to examine a population-wide effect of chronic health and socio-economic status on overnight hospitalisation. It also offers insight into the effect of specific age-associated chronic health conditions on overnight hospitalisation among seniors in Canada.
The findings of this study should be considered in the context of its methodological and analytical limitations. First, causal inferences cannot be made about the effect of chronic health conditions and socio-economic factors on overnight admission and length of overnight hospital stay. As noted previously, the CCHS is a cross-sectional survey; accordingly, we can only infer an association between the study’s predictor variables (chronic health conditions and socio-economic status) and overnight hospital admission. Future research on this topic using longitudinal data could improve additional insight on the causative effect of chronic health conditions and socio-economic status on overnight hospital admission, as well as length of stay.
A second limitation of this study is that measures of chronic health and socio-economic status were self-reported accounts. Consequently, there might be an issue of reported status and inaccurate recall of past events. Self-reported health surveys such as the CCHS provide standard data on socio-economic and health status; also, they are reported to be reliable in large-scale surveys and for health research. Also, the outcome variable is defined broadly in this study, given that the data for the study are secondary data from the CCHS. It is difficult to determine whether there will be differences in the predictors depending on what type of overnight hospitalisation occurred – in-patient acute hospital stay or nursing and convalescent stays.
Conclusion
The findings of this study show that both chronic health conditions and socio-economic status influence the overnight admission among Canadian seniors. The study demonstrates that cardiovascular diseases, age, and lower levels of household income are both predisposing and enabling factors respectively, and can increase the susceptibility of community-dwelling seniors to longer periods of hospital stay. To reduce unplanned hospital admission and the risk of longer hospital stay among community-dwelling seniors, effort must be directed towards promoting active and healthy ageing, as well as improving socio-economic well-being. Inevitably, such health and socio-economic interventions will significantly reduce the government expenditure on health-care-associated prolonged hospital stays and contribute to the well-being of Canadian seniors ageing in the community.
