A preprint of our work on multiple-cause mortality is now available.

Peters, P. A.; Klym, M.; Lopez Barrios, M.; McGaughey, T. Cause-Related Mortality in Canada by Income Quintile: Examining the Impact of Multiple Causes before and after the COVID-19 Pandemic. Preprints 2024, 2024092342.

Abstract

Mortality rates are fundamental to understanding the overall health of a population. However, statistics are usually reported using the primary underlying cause of death, overlooking potentially relevant contributing causes listed on death certificates. This paper presents indicators for multiple cause-related mortality in Canada from 2000 ā€“ 2022. Deaths from the Canadian Vital Statistics Database (2000 ā€“ 2022) were merged with multiple cause files and classified into 136 cause of death groupings. Summary statistics for multiple causes were calculated, including the Standardized Ratio of Multiple to Underlying (SRMU), which is also calculated by neighbourhood income quintile. Age-Standardized Mortality Rates (ASMR) were calculated for the underlying cause of death (ASMRUC) and compared to rates including any mention of each respective cause (ASMRAM). These were then compared to ASMRs based on a contributing-cause weighting scheme (ASMRW). The average number of causes reported on death certificates has increased from 2.79 in 2002 to 3.19 in 2021. Those in the lowest income quintiles have a higher average number of causes (3.31 in 2021) compared to those in the highest income quintile (3.09 in 2021). When employing multiple cause weighting strategies, the rank order of age-standardized mortality rates is significantly elevated for conditions including renal failure, hypertension, pneumonia, septicemia, arterial fibrillation, and artery diseases. Multiple cause-of-death approaches provide further insight into the patterns of mortality and highlight conditions that become leading causes using weighted approaches. This provides evidence to support efforts to address these conditions. There are also differences in multiple causes of death reporting by income quintile which warrants further investigation.

Keywords

multiple causes of death; comorbidity; mortality; mortality determinants; COVID-19; chronic diseases; population health; demography

The post Preprint: Cause-Related Mortality in Canada by Income Quintile appeared first on Determinants.

Poster Images

COVID-19 Deaths in Canada

Authors: Michel Lopez Barrios, Tomoko McGaughey, Morgan Klym, and Paul Peters

Presented at: Population Association of America Conference, April 13, 2023.

The analysis of COVID-19 mortality using a single cause represents a significant limitation to understanding the disease. In this study, the relationship between the underlying cause of COVID-19 with its comorbidities provides important inputs for future comparisons.

COVID-19 has driven higher mortality since March 2020, resulting in an increasingly large cumulative increase of deaths. Since October 2021, COVID-19 as a contributing factor to excess deaths has decreased. In the first year of the pandemic, COVID-19 was coded largely as the underlying cause (UC) of death. After the 1st year, COVID-19 was increasingly coded as a contributing cause (CC) of death.

COVID-19 and Contributing Causes

We compared COVID-19 deaths from 2020 (Alpha) to 2021 (Beta, Delta & Gamma variants). For deaths where the UC was COVID-19, the mean number of CCs increased between 2020 and 2021. The increase in mean number of CCs was seen across the age groups and for both sexes.

Explanations may include:

• A shift in who died from COVID-19
• A shift in coding practices
• Differing impact of COVID-19 strains over the pandemic

Next steps include examining deaths linked to hospital discharge abstracts to better ascertain reasons for the increase.

COVID-19 and Coding Mortality

The Standardized Ratio of Multiple to Underlying Causes is the age-standardised rate for any mention compared to the age-standardised rate when the cause is the UC.

• <2 indicates the cause more often a UC
• 2 indicates equal representation as UC & CC
• 2 indicates the cause is more often a CC

Marginal Change

We used Poisson regression to estimate the marginal change in the average number of CCs for each group of UCs

• The mean number of CCs was higher for all causes in 2021 (3.33), possibly because of excess of mortality in 2020 (3.26).
• Females had a lower mean number of CC (3.25) than males (3.34)
• COVID-19 has a lower mean number of contributing causes (3.48) compared to other infectious diseases (3.50)
• The mean contributing causes of death when COVID-19 was a UC increased from 3.48 in 2020 to 3.67 in 2021
• The mean number of CCs was lowest for single individuals (3.23) compared to married (3.28)
• Those from low-income areas had higher mean number of CC (3.33) compared to those in the highest income areas (3.19)

Changes in COVID-19 Coding

• Large increase in coding of COVID-19 as a contributing cause between 2020 and 2021 in Ontario, Alberta, and British Columbia.
• Small decrease in coding of COVID-19 as a contributing cause between 2020 and 2021 in Quebec.
• Public health measures carried out in Canada have not only reduced the mortality associated with COVID-19 as an UC but have alsoĀ changed the structure of associated causes.

Data Sources

We accessed the Canadian Vital Statistics Death Database and Multiple Cause Database through Statistics Canadaā€™s Research Data Centre (RDC). Data accessed through the RDC, which follows strict disclosure protocols in accordance with the Statistics Acts, is exempt from approval by the Research Ethics Board (REB) as per Tri-Council Policy Statement: Ethical conduct for research involving humans (TCPS2) article 2.2 (a).

Funding Sources

Funding for this research was provided by Social Sciences and Humanities Research Council of Canada Insight Grant #435-2021-1052.

The post Deaths from COVID-19 in Canada: a multiple cause of death approach appeared first on Determinants.

In a recent commentary published in the I argue that we need to broaden the narrative on rural health. In this commentary, I claim that in order to be sensitive and relevant, rural health research must change the perspective through which it views rural people and rural communities. First, rural communities are dynamic and not static, and change is a normal condition in small villages as much as it is elsewhere. Second, rural communities should be approached through a lends of resilience rather than disadvantage. Third, rural does not exist only in contrast to the urban. Fourth, there are implications of these theoretical shifts for health systems, health research, healthcare, and the methods we use to study these structures. This brief post expands on the implication for this perspective on rural health to how we understand rural health systems.

These ideas were first brought forth by Professor Dean Carson and we previously summarized here in an infographic presented at one of our Free Range workshops. It’s also available to download as a PDF.

Given the complex and dynamic nature of rural communities, how should health research be conducted, health services provided, and health policy developed? Anecdotal evidence suggests that rural communities are in a perpetual state of demolition, with a reduction in services, closure of hospitals, loss of physicians, etc. As a population declines (or fails to grow), how does a government continue to support a 20-bed hospital facility? With limited opportunity for families or spouses, how do small villages retain a family physician? How are prenatal and postnatal care provided when there may only be 2-3 births per year? These questions cannot be answered from an urban-normative lens, where uniform policies and programs are applied to each community.

For health researchers and practitioners, addressing the above points may require getting out of oneā€™s comfort zone. For health policymakers, this may mean that policies need to be more flexible and less prescriptive, with allowance for adaptation and failure, as promoted in complex systems and antifragile research for eHealth implementation in Northern Sweden.

For quantitative researchers (my own domain), this shift may require involving rural residents in research design, analysis, and interpretation, and spending time living and working in the communities under study. There may be instances where the quantitative data is insufficient or where findings donā€™t match with theory. In these cases, the lived experience of rural residents and the ā€˜thickā€™ knowledge that comes from understanding the communities under study can assist.

One logical extension when considering this perspective of rural communities and health systems is that parts of the system canā€™t be viewed in isolation. One canā€™t examine hospitalization rates without considering access to primary care, which canā€™t be considered without looking at recruitment and retention, which requires an understanding of community resilience. Similarly, hospitalization rates are influenced by the social demography of the underlying population, which in rural communities may be related to colonization of indigenous populations, regional economic trends, internal migration patterns, and even shifts in international immigration.

To address this, our research lab and international collaborators have made this type of rural research a priority. We explicitly integrate a diversity of perspectives from multiple locations. Weā€™ve done this through our funded research projects and published books and require this for our trainee scholars through our Free Range program.

Over the coming months, weā€™ll be presenting some research examples and personal perspectives in this blog series. The hope is that this will provide a resource for other scholars who share our perspectives, and provide a challenge to us (and others) who may disagree.

Until later,

Paul

The post Rural Health Systems and Rural Research appeared first on Determinants.

The objective of this study was to identify seasonal variations in mental healthā€“related hospitalizations among children, adolescents, and adults using administrative health data. Hospital admission records from January 2004 to March, 2014 were sourced from the New Brunswick Discharge Abstract Database. Seasonality was measured using a cosinor model to estimate the phase, amplitude, and peak of seasonal variations in psychiatric admissions over the 12-month period from January through December. A general linear model using a Poisson distribution was used to calculate rate ratios that measured the significance of monthly variations in psychiatric admissions. We adjusted for the average number of days per month and provincial population counts using offsets in both models. Between 2004 and 2014, there were 57,730 mental healthā€“related hospital admissions by 41,690 patients. Psychiatric admissions by children and adolescents (aged 3ā€“19 years) increased from 44 admissions per 100,000 in 2004 to 51 admissions per 100,000 in 2014. Rates of psychiatric admissions among children, adolescents, and adults exhibited seasonality confirmed through the detection of a statistically significant sinusoidal pattern (p < .025). The highest rates of child and adolescent admissions were in February (phase = 2, amplitude = 4.4), whereas adult admissions to the hospital peaked in early May (phase = 5.3, amplitude = 9.7). These findings lend support to existing efforts to identify the complex array of individual, family, community, and environmental factors that influence the likelihood of acute care utilization for management of psychiatric disorders.

Slaunwhite, A. K., Ronis, S. T., Peters, P. A., & Miller, D. (2018). Seasonal variations in psychiatric admissions to hospital. Canadian Psychology/Psychologie Canadienne, doi:http://dx.doi.org.proxy.library.carleton.ca/10.1037/cap0000156

The post Seasonal variation in psychiatric admissions to hospital appeared first on Determinants.

This article is the the latest from our international research team examining small villages in the developed world. For this research, we combined three case studies we had conducted in Australia, Canada, and Sweden and used these to confront the notion of rural “decline.” Small rural villages are often characterized by ‘decline’ in literature, media, and policy. However, this characterization reflects a value judgement of what is actually occurring in these places. Using a Dirt Research methodology, we show that there these places are anything but declining. There is change, sometimes accompanied by population loss, but assuming that this is all negative is simplistic.

Abstract

The purpose of this paper is to confront the notion of ā€œdeclineā€ at the village level by illustrating a more immersive approach to sociological and demographic research within rural and remote communities. The research uses case studies of three villages in Australia, Canada, and Sweden, all of which have been labeled as ā€œdeclining villages,ā€ typified by population loss, an aging population, high rates of youth outmigration, and loss of businesses and services. This paper argues that focusing solely on quantitative indicators of demographic change provides a narrow view of rural village trajectories and ignores subtle processes of local adaptation that are hidden from quantitative data sets. Our research integrates quantitative data from the ā€œoutsideā€ with qualitative data from the ā€œinside,ā€ including visual ethnography, to develop a more balanced perspective on how villages have been changing and what change could mean locally. These objectives are accomplished by revisiting a Dirt Research methodology applicable to a broad range of research into rural and remote villages. ()

The post My Village Is Dying? Integrating Methods from the Inside Out appeared first on Determinants.

This was recently published in the , in their special issue on the health of New Brunswickers. In this article, I explored how much we can understand from core population health indicators, and the relationship between these and population change at the sub-regional level.

Abstract: This paper considers population change in small areas of New Brunswick in relation to community-level population health indicators. Population forecasts developed via regionally constrained models are presented in conjunction with health indicators data for health zones and health council communities. Temporal and spatial variation in health indicators is discussed in relation to where population growth is forecasted to occur, and where there will be continued population decline. The results show that population growth is forecasted only for the areas surrounding Fredericton, Moncton, and Saint John, with most other areas predicted to experience continued population decline, even if the province as a whole does not. These forecasts are within a context where communities with declining population are also those with lower values across a range of non-medical determinants of health. These results present an opportunity to focus provincial strategies based on the combination of observed trends of population change and geographic inequalities of population health.

The post Population Change and Population Health: A Spatio-temporal Analysis of New Brunswick Communities appeared first on Determinants.

Ambient PM_2.5, O₃, and NO₂ Exposures and Associations with Mortality over 16 Years of Follow-Up in the Canadian Census Health and Environment Cohort (CanCHEC)

Background: Few studies examining the associations between long-term exposure to ambient air pollution and mortality have considered multiple pollutants when assessing changes in exposure due to residential mobility during follow-up.

Objective: We investigated associations between cause-specific mortality and ambient concentrations of fine particulate matter (ā‰¤ 2.5 Ī¼m; PM_2.5), ozone (O₃), and nitrogen dioxide (NO₂) in a national cohort of about 2.5 million Canadians.

Methods: We assigned estimates of annual concentrations of these pollutants to the residential postal codes of subjects for each year during 16 years of follow-up. Historical tax data allowed us to track subjectsā€™ residential postal code annually. We estimated hazard ratios (HRs) for each pollutant separately and adjusted for the other pollutants. We also estimated the product of the three HRs as a measure of the cumulative association with mortality for several causes of death for an increment of the mean minus the 5th percentile of each pollutant: 5.0 Ī¼g/m³ for PM_2.5, 9.5 ppb for O₃, and 8.1 ppb for NO₂.

Results: PM_2.5, O₃, and NO₂ were associated with nonaccidental and cause-specific mortality in single-pollutant models. Exposure to PM_2.5 alone was not sufficient to fully characterize the toxicity of the atmospheric mix or to fully explain the risk of mortality associated with exposure to ambient pollution. Assuming additive associations, the estimated HR for nonaccidental mortality corresponding to a change in exposure from the mean to the 5th percentile for all three pollutants together was 1.075 (95% CI: 1.067, 1.084). Accounting for residential mobility had only a limited impact on the association between mortality and PM_2.5 and O₃, but increased associations with NO₂.

Conclusions: In this large, national-level cohort, we found positive associations between several common causes of death and exposure to PM_2.5, O₃, and NO₂.

Dan L. Crouse,^1,2 Paul A. Peters,² Perry Hystad,³ Jeffrey R. Brook,^4,5 Aaron van Donkelaar,⁶ Randall V. Martin,⁶ Paul J. Villeneuve,⁷ Michael Jerrett,⁸ Mark S. Goldberg,^9,10 C. Arden Pope III,¹¹ Michael Brauer,¹² Robert D. Brook,¹³ Alain Robichaud,¹⁴ Richard Menard,¹⁴ and Richard T. Burnett¹

The post Multi-pollutant exposure and mortality appeared first on Determinants.

Background: The risk of lung cancer has been shown to be related to socioeconomic status (SES). Because the Canadian Cancer Registry does not contain socioeconomic data, the 1991 Canadian Census Cohort was used to study social determinants of lung cancer risk in the general Canadian population.

Data and methods: This study examines incidence rates of lung cancer and histologic subtypes by educational attainment, income and occupation in a broadly representative sample of Canadians aged 25 or older. Data for the 1991 Canadian Census Cohort were analyzed. The cohort comprised 2,734,835 individuals, among whom 215,700 new cancer cases were diagnosed from 1991 through 2003. Age-standardized incidence rates were calculated by age, sex, and SES using the direct method. Rate ratios, rate differences, and excess incidence were also calculated.

Results: An inverse risk between lung cancer incidence and educational attainment, income and occupation emerged among men and women, and a stepped
negative gradient in RRs was evident for all SES variables and age groups. If all cohort members had experienced the rate of those with a university degree, lung cancer incidence would have been 56% lower in men and 55% lower in women. If all cohort members had experienced the incidence rate of those in the highest income quintile, incidence would have been 33% lower in men and 25% lower in women. If all cohort members had experienced the rate of those in managerial occupations, incidence would have been 54% lower in men and 44% lower in women.

Interpretation: A negative gradient in lung cancer risk was evident for all SES variables studied.

Keywords: Age-standardized incidence rates, longitudinal study, lung cancer, record linkage, social determinants of health, socioeconomic inequalities

Debjani Mitra, Amanda Shaw, Michael Tjepkema and Paul Peters

The post Social determinants of lung cancer incidence in Canada: a 13-year prospective study appeared first on Determinants.