Publications
Health Research Working Paper Series
Hospitalizations for Ambulatory Care Sensitive Conditions (ACSC): The factors that matter
Introduction, findings, and conclusions
Archived Content
Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.
Introduction
Hospital admissions for ambulatory care sensitive conditions (ACSC) such as diabetes, heart disease and respiratory conditions may be indicators of problems with access to primary care. These are conditions for which there is a theoretical relationship between the risk of hospitalization and lack of ambulatory or primary care.1,2 It is commonly thought that timely and effective primary care can prevent the onset of complications, reduce the risk of acute episodes, and prevent hospitalizations.3,4 Thus, ACSC- related hospitalizations are commonly referred to as avoidable hospitalizations. In Canada, as in other countries, ACSC-related hospitalizations are an indirect measure of access to primary care and the capacity of the system to manage these conditions.5
While chronic disease management and ACSC-related hospitalizations are theoretically linked, the evidence for this association is limited. Much of the data are based on ecological studies of the relationship between service availability (for example, number of primary care practitioners) and ACSC-related hospitalizations. Moreover, the results of this research are mixed: some studies show no association,6,7 and others, the expected negative association, with lower admission rates in areas with more physicians.8,9
Evidence about the effect of the type of primary care and patient experiences is also limited. Lower rates of ACSC-related admissions, for example, have been found among Medicare enrolees in managed care programs versus fee-for-service, and in areas with community health centres and rural health clinics.10,11 A study of patients in an Italian hospital identified a negative relationship between the use of services, satisfaction with primary care, and the likelihood of experiencing an ACSC-related hospitalization.12
The role of patient characteristics in ACSC-related admissions has also been examined. Males and older people have been shown to be more likely than females and younger people to experience such admissions.5,13 Not surprisingly, ACSC-related hospitalizations are more prevalent among people with poorer health and co-morbidities13,14 and among individuals in low-income areas.15,16,17 However, much of this information is derived from administrative data, and therefore, limits the patient characteristics that can be assessed. The few studies that have considered a more comprehensive set of patient factors have often been based on surveys of small samples of patients conducted in a single institution or geographic area, thereby compromising the generalizability of the results.
This analysis, by contrast, uses a population-based approach to study a broad set of factors associated with ACSC-related hospitalizations. The unique feature of this study is the nationally representative sample of people most at risk—those with at least one ACSC (asthma, diabetes, emphysema/COPD, epilepsy, heart disease or high blood pressure). The study uses health survey data linked to hospital administrative data. The linked dataset provides a unique opportunity to consider a comprehensive set of characteristics of people who have been admitted to hospital: socio-economic, health status and lifestyle factors, and importantly, their access to primary care. Understanding the role of these factors may shed light on how primary care services may reduce the risk of these "avoidable" hospital admissions.
Findings
Rate of ACSC-related hospitalizations among "at risk" individuals
An estimated 4.2 million Canadians aged 12 to 74 have been diagnosed with at least one ACSC. Among this population, 46% reported having been diagnosed with high blood pressure, 43% heart disease, 36% diabetes, 30% asthma, and 16% COPD.
Among this "at risk" group, 161,000 (3.8%) had at least one ACSC-related hospitalization over a four-year period. The most common admissions were for COPD (26%), diabetes (20%), angina (19%), and heart failure and pulmonary edema (16%).
People with an ACSC-related hospitalization constituted just 0.4% of the population aged 12 to 74, but represented about 6% of all hospitalized individuals and used nearly 11% of all hospital days (Figure 1). Consequently, understanding who they are is important.
Figure 1Characteristics of individuals admitted for an ACSC-related hospitalization
Older and sicker
More than half (52%) of those who experienced an ACSC-related hospitalization were male, compared with about 47% of those who were hospitalized for another reason or not hospitalized at all during the four-year period (Table 1). Over half (53%) of those with an ACSC-related hospitalization were aged 61 or older, compared with 45% of those hospitalized for other reasons, and 23% who were not admitted to hospital. Given these differences in the demographic characteristics of the three groups, the remaining comparative analyses are age/sex adjusted.
Table 1Individuals who had an ACSC-related hospitalization were almost twice as likely to be separated or divorced (14%) than were those hospitalized for other reasons (8%) or not hospitalized at all (7%).
As expected, individuals who experienced an ACSC-related hospitalization reported poorer health across several measures, including fair/poor health (56%) and severe disability (37%). The odds of experiencing an ACSC-related admission were up to 10 times higher for those in fair/poor health, compared with those in excellent health (Appendix Tables B and C).
Individuals who had an ACSC-related hospitalization were more likely to have two or more co-morbid conditions, compared with the other groups: 50% versus 39% and 29%. The odds that people with multiple co-morbid conditions would experience an ACSC-related hospitalization were up to 4.5 times the odds for those with no co-morbidities (Appendix Tables B and C).
Almost one in five individuals who had an ACSC-related hospitalization reported being depressed, compared with one in ten who did not experience a hospital admission (18% versus 10%). They were less likely to report usually being free of pain (54%) than were the comparison groups (64% and 77%).
Lower socio-economic status
More than a third of individuals (34%) with an ACSC-related hospitalization were in the lowest household income quintile, compared with 24% of those hospitalized for other reasons and 16% of those not hospitalized (Table 2). Individuals in the lower-middle and lowest income groups had two to four times the odds of experiencing an ACSC-related hospitalization, compared with those in the highest income group (Appendix Tables B and C) .
About one person in five (18.2%) who had an ACSC-related hospital admission lived in a household where the highest level of education was less than secondary graduation; the corresponding figures were 16.8% and 12.8% for non-avoidable and no hospitalization, respectively. The odds that people in lower-education households would experience an ACSC-related hospitalization were two to four times the odds for residents of households in which at least one member was a postsecondary graduate (Appendix Tables B and C).
Lifestyle
One in three individuals who had an ACSC-related hospitalization were daily smokers, compared with 21% and 18% of the comparison groups (Table 3). Regardless of their smoking category (daily, occasional or former), smokers were more likely to experience such an admission. The odds of experiencing an ACSC-related hospitalization were two to three times higher among former and daily smokers compared with non-smokers (Appendix Tables B and C).
Weight also appears to matter. Underweight individuals aged 12-74 were up to three times more likely to experience an ACSC-related hospitalization than were normal-weight individuals (5.2% versus 1.5%). Being overweight appears to have a protective effect. Approximately 27% of those who experienced an ACSC-related hospitalization were overweight, compared with more than 30% of the comparison groups. On the other hand, 60% of the people who had ACSC-related hospitalizations reported being physically inactive, compared with about 50% of those who experienced no hospitalization at all (Table 3).
Access to primary and specialist care
If ACSC-related hospitalizations indicate inadequate primary care, it might be expected that people who had such hospitalizations would report less access to primary care services. The results from the linked data, however, suggest otherwise.
Individuals who had an ACSC-related hospitalization were, in fact, more likely than the comparison groups to report that they had access to a regular medical doctor (98% versus 96% and 94%) (Table 4). They were also more likely to be frequent users of care, with almost 70% reporting four or more occasions when they accessed their primary care physician; this compares with 50% of those who had no hospitalization. Similarly, they were more likely to have reported four or more specialist consultations (24% versus 17% and 9%). As well, they were more likely to have been hospitalized at least once in the past year (41% versus 19% and 9%) (Table 4).
Despite greater access to and use of services, individuals who had an ACSC-related hospitalization were also more likely to report an unmet health care need (20%) than were those who experienced no hospitalization at all (15%). (Table 4)
Which factors matter more? Results of multivariate analysis
The descriptive results of the analysis of the linked data confirm that patient characteristics such as health status, socio-economic status and risk factors, as well as access to health care, are associated with ACSC-related hospitalizations. But which factors matter more? And do the results differ for men and women? To answer these questions, we looked at a range of factors simultaneously in gender specific multivariate regression models to identify which factors were most associated with an ACSC-related hospitalization relative to those who were not admitted to hospital over the four year follow-up period. We began with those factors found to be significantly associated with an ACSC-related hospitalization in the previous analysis and applied a stepwise regression approach allowing the selection a parsimonious sets of factors. The results are presented in Table 5.
Age, disability and comorbid conditions matter
Older age remained significantly associated with ACSC-related hospitalizations. Men aged 60 or older experienced a 3.5-fold increase in the odds of experiencing such an admission, compared with those aged 21 to 40. The results for women were similar, with an odds ratio of 2.4 (Table 5).
Disability and co-morbid conditions also remained important, but the results varied by gender. Men with severe disability were at higher risk of an ACSC-related admission (OR=2.96). For women, multiple co-morbidities matter: women with two or more chronic conditions had more than four times the odds of an ACSC-related hospitalization (OR=4.41) (Table 5).
Low income matters
The association between low income and ACSC-related hospitalizations persisted for only men when adjusting for other factors. The odds of experiencing an ACSC-related hospitalization were three times as high for men in the lowest income quintiles, compared with those in the highest quintile (OR=3.25). Men in the lower-middle income also had significantly high odds of such a hospitalization (Table 5).
The odds that separated/divorced men would experience an ACSC-related hospitalization were twice those of married men (OR=2.05).
Association with smoking and weight varies by gender
Smoking continued to be significant, even when other factors were taken into account. Among men, all categories of smokers, former, daily and occasional experienced a higher odds of an ACSC-related hospitalization however, only the results for former smoker were statistically significant (OR=2.2). Among women, the odds were significantly high for daily smokers (OR=1.8) (Table 5).
Associations with BMI (body mass index) also differed by gender. Among men, being overweight had a protective effect (OR=0.59) against ACSC-related hospitalizations. Among women, the odds of an ACSC-related hospitalization were significantly high (OR=5.87) for those who were underweight, compared with normal-weight women. Being overweight and obese were also associated with an elevated risk of experiencing such an admission but the results were not statistically significant. (Table 5).
Among women, lack of physical activity was associated with higher odds of an ACSC-related hospitalization (OR=1.6), but the level of physical activity did not appear to matter for men.
Access to primary care not significant
Men who used more specialist and hospital services had significantly high odds of an ACS-related hospitalization. Men and women who were hospitalized at least once in the previous year were at higher odds (OR=3.13; OR=4.3 respectively) of experiencing such an event.
Access to primary care services (regular medical doctor and visits with family doctors) was not significantly associated with ACSC-related hospitalizations after adjusting for other factors.
"Risk" profiles: Who is at greatest risk of an ACSC-related hospitalization
With the results of the multivariate regression analyses, profiles of individuals at varying risk of experiencing an ACSC-related hospitalization can be constructed. Using the gender-specific regression models, the predicted probability of experiencing an ACSC-related hospitalization was calculated for all individuals in the study sample. Based on these probabilities, individuals were grouped into risk quintiles, ranging from "low" risk (Quintile 1) to "high" risk (Quintile 5). The members of each group are described in terms of the most prevalent characteristics determined to be significantly associated with ACSC-related hospitalizations in the previous regression analysis (Tables 6 and 7; Appendix Tables D to G).
The average predicted probability of experiencing an ACSC-related hospitalization ranged from 0.4% (Quintile 1) to 15.1% (Quintile 5) among women and between 0.4% (Quintile 1) and 21.6% (Quintile 5) among men.
As expected, the predicted probabilities are linearly related to the proportion of individuals who actually experienced an ACSC-related hospitalization in each risk quintile, providing evidence of the robustness of the models. The majority of ACSC-related hospitalizations were among people in the highest risk group – estimated to be 39,000 hospitalizations among women and 46,000 among men (Figure 2).
Results for women
For women in the highest risk group (Quintile 5), the probability of experiencing an ACSC-related hospitalization ranged from 6% to 50% (Table 6). This group was further subdivided into three groups as they have different risk profiles. Women with a 6% to 9% probability of an ACSC-related hospitalization were, on average 55 years of age and in the lowest to lower-middle income groups. They had 2 or more co-morbid chronic conditions and were generally inactive.
Women at the highest risk of experiencing an ACSC-related hospitalization (i.e. >50%) were on average 64 years of age and primarily from the lowest income quintile. Most had two or more co-morbidities, and most were either former or daily smokers. These women also tended to be underweight and generally reported being inactive. They were higher users of specialist services, and most experienced at least one hospitalization in the previous 12 months.
Results for men
Among men in the highest risk group (Quintile 5), the probability of an ACSC-related hospitalization varied from less than 10% to 75%. This high-risk group was divided into three subgroups (Table 7). Those with a probability of between 9% to 25% were on average 60 years of age, in the lowest to lower middle income groups, and were more likely to be married or formerly married. Men in this group reported a range of disability levels from mild to severe. They were generally former or daily smokers.
Men at the greatest risk of experiencing an ACSC-related hospitalization (50+%) were on average 67 years of age and primarily from the lowest income quintile and were married or formerly married (i.e. widowed or divorced). Most reported severe disabilities, and identified themselves as former smokers. Unlike men in other risk groups, high-risk men were tended to have used specialists services four or more times in the previous year, and in most cases, had experienced at least one hospitalization.
Discussion
This study is the first national-level population-based analysis of patient and primary care factors associated with ACSC-related or "avoidable" hospitalizations in Canada. A unique feature of the study is the focus on individuals most at risk—those with at least one ACS chronic condition. The linked health survey and hospital data provide an opportunity to better understand the extent to which patients' characteristics and measures of access to primary care play a role in why some individuals with chronic conditions experience ACSC-related hospital admissions, while others do not. The risk profiles help to focus on individuals at highest risk.
Overall, patient factors played a key role in ACSC-related hospitalizations, while access to a primary care appeared to matter less, at least in terms of service volume. Findings in the literature about the relationship between primary care services and ACSC-related admissions are mixed. The results of this study provide some evidence that greater access to primary care services does not necessarily reduce the risk of experiencing an ACSC-related hospitalization. In fact, people who experienced an ACSC-related hospitalization were more likely to have a regular medical doctor and to be high users of primary care services. However, when other factors were taken into account, access to primary care did not appear to matter.
While the study results do not appear to support increasing the volume of primary care services to reduce the risk of ACSC-related hospitalizations, the results cannot address the issue of appropriate use of services nor the potential role of quality of care. Given the lack of information about disease severity, the study results cannot speak to whether patients are actually receiving an appropriate level of service to meet their needs. Furthermore, data limitations also restricted the possibility of examining whether the type and quality of primary care services accessed by individuals with chronic conditions matter for ACSC-related admissions. It may be that while more services do not appear to reduce the risk of ACSC-related admissions, the type of services (access to multi-disease care, prevention programs) may matter.
The results about patient characteristics may offer some insight about where and how primary care can play a role in reducing what maybe be truly avoidable hospitalizations. The risk profiles provide clear evidence that a range of patient factors matter for ACSC-related hospitalizations, and that those at highest risk have multiple issues related to that risk. There are three potential ways this information can be used to avoid these hospitalizations: 1) to identify individuals at high risk; 2) to identify the type of primary care services that may be required to meet the needs of this patient population; and 3) to identify situations/solutions that may be beyond the scope of primary care.
Identifying "high-risk" patients
The risk profiles suggest that individuals at highest risk often share key characteristics, including poor health, and as a result, higher use of specialists and hospital services. For both men and women, experiencing a hospital admission in the previous year was associated with significantly higher odds of an ACSC-related hospitalization, even when other potentially confounding factors were taken into account.
Patterns of service use can provide valuable information to identify those who may be at greater risk of ACSC-related hospitalizations. Currently, health care use information is utilized by health care providers in the UK and the US to identify patients at high risk of short-term hospitalization. These individuals are provided with intense ambulatory/primary care to address critical needs and avoid adverse events.. Recently in Canada, use of emergency room services was identified as a key predictor of early death and unplanned readmissions among hospitalized individuals.23 Information on service use, in conjunction with other known risk factors (co-morbidities, smoking) may help primary care providers identify those most at risk of an ACSC-related admission, and who may benefit from more intense ambulatory care to avoid future hospitalizations related to their chronic conditions.
The role of primary care
Findings about the individual characteristics associated with increased odds of ACSC-related hospitalizations can be used by providers to better understand how primary care services may meet the needs of these people. The results clearly point to the role of disability (men) and co-morbid conditions (women) in increasing the odds of an ACSC-related hospitalization. The results also provide new evidence about the role of lifestyle indicators, such as smoking and physical activity (women). The risk profiles suggest that those at highest risk tend to have more than one of these factors.
More and more people have multiple chronic conditions. In Canada, approximately 30% of individuals with chronic conditions have at least two co-morbid conditions (the percentage rises above 50% among those with diabetes). Many of these chronic conditions have been linked to a handful of risk factors such as smoking.24 Considerable efforts have been made to improve the delivery of primary care to Canadians with chronic conditions. Collaboratives such as those established in British Columbia and Newfoundland are designed to provide a range of services, including both disease management and preventive services, to improve the health of patients with chronic conditions such as diabetes.25 The results of this study provide further evidence of the value of these efforts in addressing the needs of people with multiple chronic conditions and risk factors that place them at greater risk of complications and hospitalization.26
Beyond primary care…
While some risk factors identified in this study are amenable to primary care intervention, others, for example low income, may not be. This analysis adds evidence derived at the individual level regarding the association between socio-economic status and ACSC-related hospitalization, which to date had largely been derived from ecological studies.15,27 Much more information is needed about the specific role income plays in increasing individuals risk of an ACSC-related hospitalization to determine whether the response lies within the primary care system. In a publicly funded health care system that provides services based on need, not ability to pay, it is not clear why lower-income individuals are more likely to experience ACSC-related hospitalizations.
A possible explanation is that income is actually a marker for other important factors such as disease severity. Considerable evidence links low income and poor health status: individuals in lower-income households are more likely than people in more affluent households to have multiple chronic conditions, poorer health and an increased need for healthcare services. With the data available for this study, it was not possible to explicitly adjust for disease severity. Nonetheless, the association with household income persisted when proxy measures such as disability status and use of health care services were taken into account.
Alternatively, the association with household income may represent a more systematic disadvantage that compromises individuals' access to needed services, thereby potentially placing them at higher risk of adverse events. Growing evidence points to differential access to and use of health care by income including among those with ACS chronic conditions.28-30 The health care needed by people with ACS chronic conditions often extends beyond publicly provided services. These individuals often require pharmaceuticals and monitoring devices (for example, glucose monitoring) that may entail out-of-pocket expenditures. Public insurance is available for lower-income groups to alleviate this economic burden, but some people may be "slipping through the cracks." In fact, even when insurance status is taken into account, lower-income groups are generally less likely to use services that require out–of-pocket expenditures.31 It may be that low- income individuals with ACS chronic conditions have limited access to the full range of health care services required to monitor and manage their conditions, thus placing them at higher risk of adverse events such hospitalizations.
Similarly, meeting the needs of underweight women, who are at a significantly higher risk of an ACSC-related hospitalization than normal-weight women, may extend beyond the primary care system. A number of reasons could account for their high odds of ACSC-related hospitalizations. Being underweight is often associated with poor health and may be a marker for disease severity. The results of the risk profile indicate that underweight women most at risk for an ACSC-related hospitalization were also more likely to have multiple co-morbidities, and thus, likely poorer overall health.
Being underweight is also a result of nutritional risk, particularly among community dwelling elderly women in lower-income groups. Good nutrition and maintenance of healthy weight are generally associated with better health and prolonged life among the elderly.32,33 Poor nutrition is common among community-dwelling seniors, particularly those in lower income groups with rates estimated to be as high as 40% to 50%.34,35 Factors such as poor health due to chronic conditions, restricted budget and living or eating alone have all been found to contribute to nutritional risk among community dwelling elders36,37 and the consequences can be serious, including decreased quality of life and premature mortality.38,39
It is possible that the links between underweight and low income and ACSC-related admissions among women in this study may, in part, be related to poor nutrition. Although causality cannot be determined based on the available data, poor nutrition may be related to overall declines in health among an already vulnerable population. The combination of poor health and poor nutrition may place women with chronic conditions at higher risk of complications, and therefore, at higher risk of hospitalization. If this is, indeed, the case, the question is whether primary health care has a role in improving the nutrition of these women to avoid these hospitalizations. In one Canadian study, encouragement from health professionals was identified as an important facilitator in improving the dietary habits of women at nutritional risk.40 Health care providers may also play a part in screening elderly patients to identify those who may be in the early stages of nutritional risk. Primary care models may also provide some of the secondary prevention programs, such as nutritional counselling.41
In Canada, community-based services have traditionally been key in supporting the nutritional needs of community-dwelling elders at risk of poor nutrition. Programs such as meals-based delivery services, nutrition education, and cooking classes are provided by various groups including municipal governments and charitable organizations. Healthy eating was recently identified as one of five focus areas in the Healthy Aging in Canada brief prepared for federal, provincial and territorial departments42 Yet despite this attention, policies at the community level to meet the nutritional needs of the elderly are lacking. While some policies identify the needs of this population, few plans are in place to promote this agenda at the national level. The authors call on dieticians at the public health, community and government levels to advocate on behalf of this patient population to publicize the need for community-based nutritional services.43
Conclusions
ACSC-related hospitalizations indicate potential problems in the delivery of primary care, and are, therefore, of interest to policy- and decision-makers. It is presumed that timely and effective primary care can prevent the onset of complications, control acute episodes, and thereby, avoid hospitalizations. While most of the research has focussed on the role of primary care service, emerging evidence suggests that patient factors, such as socio-economic status, are also important. The results of this study provide new data about the role of factors, such as co-morbidities, low income and lifestyle, in placing individuals at risk of an ACSC-related hospitalization. While some of these factors may be addressed by the primary care system, other solutions may lie beyond its scope.
- Date modified: