Chapter 7 Population Health Approach

7.1 Introduction

In public health we see community health through a socioecological lens focused on complex adaptive systems and lifecourse social networks. We are called upon to lead teams and organizations in making high stakes decisions in the setting of complex environments, limited information, multiple objectives, competing trade-offs, uncertainty, and time constraints. The emerging field of population health data science provides an integrative, dynamic approach to “data-driven decision making” that applies to community and health care systems. Our approach to population health has been heavily influenced by public health epidemiology.

First we start with some key definitions:

Health is a state of complete physical, mental and social well-being and not merely the absence of disease or infirmity (World Health Organization 1948).

Public health is what we, as a society, do collectively to assure the conditions in which people can be healthy (Institute of Medicine 1988).

Population health is an systems24 framework for studying and improving the health of populations through collective action and learning (Aragón 2016).

Data science is the art and science of transforming data into actionable knowledge (Aragón 2016).

Population health data science is the art and science of transforming data into actionable knowledge to improve health (Aragón 2016).

Our definition of population health differs from the mainstream definition of population health as “the health outcomes of a group of individuals, including the distribution of such outcomes within the group” (Kindig and Stoddart 2003). Our definition of population health emphasizes the following:

  • complex adaptive social and environmental systems
  • lifecourse health development
  • intergenerational transmission of risks (social, epigenetic)

Our definition supports modern approaches to community health and health systems transformation (Halfon et al. 2014). In general, in public health we have four population health goals.

  1. Protecting and promoting health and equity.
  2. Transforming people and place
  3. Ensuring a healthy planet
  4. Achieving health equity

To make progress on these goals we focus on changing

  • Policies: economic, social, ecological, and health
  • Systems: economic, social, ecological, and health
  • Enviroments: social, physical, and ecological

The basis for change is decision making; hence, the need for transforming data into actionable knowledge. We start by review data from an epidemiologic perspective.

7.2 Epidemiologic approach

Epidemiology is a basic science of public health, and is commonly defined as

The study of the distribution and determinants of health-related states or events in specified populations, and the application of this study to the control of health problems (Last 2000).

Operationally, the epidemiologic approach, from problem definition to public health action is summarized here:

  • Epidemiologic measures
  • Public health surveillance
  • Field investigations (e.g., outbreaks)
  • Descriptive and analytic epidemiology
  • Program and project evaluation
  • Causal and statistical inference
  • Evidence-based public health action

7.3 Epidemiologic analyses for 2-by-2 tables

7.3.1 Cohort studies with risk data or prevalence data

Exposed Unexposed Total
Cases \(a\) \(b\) \(M_{1}\)
Noncases \(c\) \(d\) \(M_{0}\)
Total at risk \(N_{1}\) \(N_{0}\) \(T_i\)

7.3.1.1 Risk difference

7.3.1.2 Risk ratio

7.3.1.3 Odds ratio

7.4 Epidemiologic analyses for stratified 2-by-2 tables

7.4.1 Cohort studies with binomial (risk or prevalence) data

Notation
Exposed Unexposed Total
Cases \(a_i\) \(b_i\) \(M_{1i}\)
Noncases \(c_i\) \(d_i\) \(M_{0i}\)
Total at risk \(N_{1i}\) \(N_{0i}\) \(T_i\)

7.4.1.1 Pooled risk difference with confidence limits

\[ RD_{MH}=\frac{\sum_i {\frac{a_i N_{0i}-b_i N_{1i}}{T_i}}} {\sum_i {\frac{N_{1i} N_{0i}}{T_i}}} \]

\[ \mathrm{Var}(RD_{MH}) = \frac{\sum_i \left(\frac{N_{1i} N_{0i}}{T_i}\right)^2 \left[ \frac{a_i c_i}{N_{1i}^2 (N_{1i}-1)} + \frac{b_i d_i}{N_{0i}^2 (N_{0i}-1)} \right]} {\left( \sum_i \frac{N_{1i} N_{0i}}{T_i} \right)^2} \]

rd.mh <- function(x, conf.level = 0.95) {
    ai <- x[1, 1, ]  # exposed cases
    bi <- x[1, 2, ]  # unexposed cases
    ci <- x[2, 1, ]  # exposed noncases
    di <- x[2, 2, ]  # unexposed noncases
    N0i <- bi + di  # total exposed (at risk)
    N1i <- ai + ci  # total unexposed
    M0i <- ci + di  # total cases
    M1i <- ai + bi  # total noncases
    Ti <- ai + bi + ci + di  # stratum total
    RDmh <- (sum((ai * N0i - bi * N1i)/(Ti)))/(sum((N1i * N0i)/(Ti)))
    numer <- ((N1i * N0i)/(Ti))^2 * (((ai * ci)/(N1i^2 * (N1i - 1))) + ((bi * 
        di)/(N0i^2 * (N0i - 1))))
    denom <- (sum((N1i * N0i)/(Ti)))^2
    Z <- qnorm((1 + conf.level)/2)
    SE.RDmh <- sqrt(numer/denom)
    LL <- RDmh - Z * SE.RDmh
    UL <- RDmh + Z * SE.RDmh
    list(data = x, risk.diff = RDmh, conf.int = c(LL, UL))
}

7.4.1.2 Pooled risk ratio with confidence limits

\[ RR_{MH}=\frac{\sum_i {\frac{a_i N_{0i}}{T_i}}} {\sum_i {\frac{b_i N_{1i}}{T_i}}} \]

\[ \mathrm{Var}[\log(RR_{MH})]= \frac{\sum_i \left(\frac{M_{1i} N_{1i} N_{0i}}{T_i^2} - \frac{a_i b_i}{T_i}\right)} {\left( \sum_i \frac{a_i N_{0i}}{T_i}\right) \left( \sum_i \frac{b_i N_{1i}}{T_i}\right)} \]

rr.mh <- function(x, conf.level = 0.95) {
    ai <- x[1, 1, ]  # exposed cases
    bi <- x[1, 2, ]  # unexposed cases
    ci <- x[2, 1, ]  # exposed noncases
    di <- x[2, 2, ]  # unexposed noncases
    N0i <- bi + di  # total exposed (at risk)
    N1i <- ai + ci  # total unexposed
    M0i <- ci + di  # total cases
    M1i <- ai + bi  # total noncases
    Ti <- ai + bi + ci + di  # stratum total
    RRmh <- sum(ai * N0i/Ti)/sum(bi * N1i/Ti)
    numer <- sum(M1i * N1i * N0i/Ti^2 - ai * bi/Ti)
    denom <- sum(ai * N0i/Ti) * sum(bi * N1i/Ti)
    Z <- qnorm((1 + conf.level)/2)
    SE.logRRmh <- sqrt(numer/denom)
    LL <- exp(log(RRmh) - Z * SE.logRRmh)
    UL <- exp(log(RRmh) + Z * SE.logRRmh)
    list(data = x, risk.ratio = RRmh, conf.int = c(LL, UL))
}

7.4.1.3 Odds ratio

Bibliography

World Health Organization. 1948. “WHO Definition of Health.” World Health Organization; World Health Organization.

Institute of Medicine. 1988. The Future of Public Health. National Academy Press. http://www.nap.edu/catalog/1091.html.

Aragón, Tomás J. 2016. “Population Health Data Science: The Art and Science of Transforming Data into Actionable Knowledge to Improve Health.” Home page. http://www.phdata.science.

Kindig, David, and Greg Stoddart. 2003. “What Is Population Health?” Am J Public Health 93 (3). Department of Population Health Sciences, University of Wisconsin-Madison School of Medicine, Madison, USA. dakindig@facstaff.wisc.edu: 380–83.

Halfon, Neal, Kandyce Larson, Michael Lu, Ericka Tullis, and Shirley Russ. 2014. “Lifecourse Health Development: Past, Present and Future.” Matern Child Health J 18 (2): 344–65. doi:10.1007/s10995-013-1346-2.

Last, John. 2000. A Dictionary of Epidemiology. 4th ed. New York: Oxford University Press.

  1. By “systems” we mean a lifecourse, complex adaptive socioecological system↩