Fig. 5.1
Factors contributing to health care disparities. Size of font reflects perceived relative importance of factor (Reproduced from Haider et~al. [35])
5.3.3.2 Hospital Volume
Surgical outcomes are directly affected by hospital volume. This is especially true of high risk and complex operations, where outcomes are better at higher volume centers. Many studies consistently demonstrate that black and Hispanic patients receive surgical care at centers with low volume and high mortality. For example, Konety et~al. demonstrated that black patients were more likely to have surgery in low volume hospitals in their large study of Medicare coronary artery bypass grafting patients [37]. A recent evaluation of National Trauma Data Bank suggest that minority trauma patients cluster at low volume centers, where outcomes are equally bad for all racial groups [16]. Conversely, minority patients treated at high volume, urban academic centers participating in the American College of Surgeons National Surgical Quality Improvement Program have short-term mortality and morbidity similar to white patients [38]. Therefore, improving quality of care and regionalization of complex surgical care may help reduce racial disparities.
5.3.3.3 Patient Case-Mix
Aggregate patient demographics are known to influence hospital outcomes. For example, Breslin and colleagues demonstrate that hospitals with large minority populations had worse outcomes for all patients in their evaluation of breast and colon cancer [39]. These factors have also been found at “safety net” hospitals that serve predominantly minority trauma patients [40]. Similarly, Rhoads et~al. also describe hospitals with greater than 40 % Medicaid patients have a higher 30-mortality after colon surgery [41]. Therefore, racial and socioeconomic disparities may partially be attributed to patients seeking care at low quality, ill-resourced centers. With the enforcement of public sector pay-for-performance, these hospitals may be further marginalized unless supported to initiate quality improvements.
5.3.3.4 Vulnerable Populations and Participation in Clinical Research
Patient-centered clinical research provides an avenue to develop and implement tailored therapeutic interventions. However, due to conservative inclusion criteria, vulnerable populations are often excluded from clinical trials. Consequently, patients with comorbidities, racial minorities, and the uninsured often receive care that is non-adherent to established guidelines. For example, Rafaie et~al. demonstrate that such populations are less likely to receive guideline-recommended cancer care and are under-represented in cancer clinical trials [42]. To guarantee generalizability and adherence of these guidelines, an equitable representation of all patient demographics must be ensured.
5.4 How We Can Solve This Problem?
Disparities in surgical outcomes are a public health concern and as such a public health approach must be enacted to overcome this challenge. Below we describe and put in to context a six step strategy for addressing the disparities issue:
1.
Generate Awareness of the Problem
Recognizing the problem is an essential harbinger for change. Disparities have been well-documented for multiple operative procedures across different patient groups. However, most accounts offer a broader outlook and seldom explore the intricacies of the issue. Therefore, a systems approach must be adopted in trying to discover the worst-hit sub-groups to prioritize future course of action. Identification of these problem areas are vital in generating public awareness, a demand for change and securing funding to guide further explorations.
2.
Elucidate their Underlying Mechanism
A mechanistic approach attributes a cause to a problem, following its appropriate identification. This allows for a clearer understanding of a dynamic network of factors and their interactions with the environment. As described in the previous section, many of the factors leading to surgical outcomes disparities have already been identified. While some are more completely understood than others, future work must focus on identifying the relative strengths of each in driving these disparities. For example, a hospital quality effect may contribute more towards overall disparities than SES. This elucidation will certainly prioritize the design of future interventions.
3.
Create Solutions and Interventions
Solutions and interventions are more simply envisaged once a problem and its causal determinants are well-understood. For example, it is known that low quality of care (mechanism) results in surgical outcomes disparities (problem). One logical solution, therefore, may be to improve quality of care. This is exemplified in the mitigation of racial disparities at hospitals participating in National Surgical Quality Improvement Program. Similarly, separate solutions must be devised to overcome mechanism-specific challenges.
4.
Evaluating these Interventions
The central dogma of health systems improvement is in the continuous, precise measurement of structures, processes and outcomes of care. Any intervention must stand the rigors of science, ethics and economy, and offer fairly sustainable solutions. Like many surgical outcomes improvement programs, a disparity mitigation initiative must establish a quantifiable metric or a set of metrics, to accurately assess program efficacy. These could range from simple rates of surgical utilization and in-hospital outcomes, to the more complex composite measures of disparities.
5.
Advocating their Implementation
Advocacy is critical in delivering science to the masses. On multiple occasions, strong advocacy has defeated even the most compelling scientific evidence. Therefore, any intervention to reduce surgical outcomes disparities must be advocated for at all forums to drive public and political opinion; a demand for change must be created together with robust science to ensure appropriate program support and implementation.
6.
Dynamic monitoring of their effectiveness
As much as being the last step, continuous program effectiveness monitoring completes the cycle of the envisioned dynamic change. All successful outcomes improvement programs have this critical step built-in as it ensures that the program remains sustainable, effective and malleable to change.
We believe that disparities research is still in between phase 2 and 3; we still need to create credible solutions. Any surgeon who is able to create effective programs that truly decrease surgical inequities, whether they be directed at racial issues or other problems such as access or age will certainly attain the highest degree of academic success and more importantly – personal fulfillment.
Box 5.1: Six Steps to Solving a Public Health Problem
1.
Generate Awareness of the Problem
2.
Elucidate their Underlying Mechanism
3.
Create Solutions and Interventions
4.
Evaluating these Interventions
5.
Advocating their Implementation
6.
Dynamic monitoring of their effectiveness
5.4.1 Pitfalls in Conducting Disparities Research
5.4.1.1 Not Accounting for Missing Data
Most outcomes disparities research is conducted using administrative datasets, which have well known limitations. Inadequately accounting for missing data, for example, by simple exclusion of cases may introduce substantial bias and invalidate study results. In recent years, many different techniques have been developed to adequately handle missing data. In recent years, multiple imputation in particular has gained relevance and has been validated in the National Trauma Data Bank to handle missing physiology data. Careful choice and use of techniques to handle missing data will yield more meaningful results.
5.4.1.2 Not Accounting for Heterogeneity
Due to inherent heterogeneity in patient sub-groups, disparities affect patients differentially. For example, insured patients have varying outcomes depending on their exact payer-profile. Similarly, Asians constitute a heterogeneous group with Chinese, Japanese, Filipino, Hawaiian and other ethnic sub-groups having different disparities. Although super-stratification of all patient sub-groups is impractical, heterogeneous patient case-mix must be respected analytically, whenever possible.
5.4.1.3 Not Accounting for Center Effects
Many studies disregard the effects of correlated patient effects within hospitals. For example, studies may fail to account for the fact that patients will often have similar intra-hospital outcomes than inter-hospital ones. Disregarding this will result in the default assumption that patients are equally likely to receive treatment at any hospital. This is especially cause for concern in the context of known effects of clustered patient outcomes in driving racial disparities. Several techniques, including hierarchical modeling, offer solutions to this problem.
5.4.1.4 Not Properly Risk Adjusting
The importance of appropriate risk-adjustment is often overlooked. While some studies use an overly complicated model including all possible covariates, others pursue a more conservative approach and often disregard potentially important factors. Covariates used to risk-adjust must be carefully titered against model performance metrics to ensure adequate discrimination and calibration, while satisfying underlying model assumptions.
5.4.1.5 Thinking That All Patients in the Same Race/Ethnicity Group Are the Same
As pointed out above there is a significant interaction between race/ethnicity, socio-economics and even cultural norms. This leads to differences in access and a variety of other important confounder within what may be classified as a homogenous group. For example, the very affluent Cuban community in Miami may have very different health care access and baseline indicators when compared to a largely migrant, Spanish speaking farm working community in the southwest. Therefore lumping both groups together under the term “Hispanic” in a study investigating disparities will lead to biased results. We need to remain cognizant of this heterogeneity within seemingly homogenous groups.
5.4.1.6 Self-Identification Bias
Studies demonstrate that an observer who does not know the patient may misclassify an individual’s race/ethnicity group by up to 30 % if they rely on simple observation. In most administrative datasets, it’s a hospital registrar or other data acquisition person who enters the patient’s race/ethnicity in the dataset and they do not always ask questions that may be seem to be obvious like race or gender. An individual who may appear black may actually be Hispanic. The same goes for gender, especially when patients may outwardly appear to be of one gender but self-identify with another. This is especially true in circumstances such as trauma where people are unable to ask questions from patients and family is not readily available. This is difficult to control for in retrospective analyses, but should be acknowledged and needs to be addressed in studies pursuing prospective data collection.
5.4.2 Final Comment
As a country we have made substantial progress in breaking down barriers and creating opportunities for all Americans. One of the major reasons for this has been our collective ability to reflect on our past, learn from mistakes and work towards a more perfect nation. We still have a long way to go towards creating a healthcare system that treats everyone the same and produces equal outcomes for all its users- but we are making progress. To be successful as an academic surgeon one must not be overwhelmed by the enormity of the task at hand, but rather seize the opportunity and choose to make a real difference. If enough of us truly devote our attention to this important national issue, we can hopefully reduce the following quote from its current reality to a mere memory:
Of all the forms of inequality, injustice in health care is the most shocking and inhumane.Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
