11.5.1 Cancer

There is significant heterogeneity between disease courses in different types of cancer, even with narrowed definitions such as “within metastatic solid tumors” [9]. For instance, metastatic breast and prostate cancer are associated with much longer survival times as compared to metastatic lung cancer. With tissue diagnosis and staging established with the aid of consultants such as oncology, surgery, gastroenterology, and interventional radiology, there are several means of estimating an initial disease-based prognosis. One method is through the use of published studies related to a particular diagnosis, specifically from treatment trials. While this can often be challenging for physicians, there are a large number of possibly relevant treatment trials that can provide guidance on prognosis for treatment versus supportive care at any particular point in an oncologic disease course. Of note though, most of these trials do not include patients who are hospitalized or who have poor functional status, multimorbidity, or organ dysfunction.

Using one of the performance status-based tools as described in the previous section is a foundation for hospitalists in cancer prognostication as there is often a dearth of other evidence from clinical trials in hospitalized, ill advanced cancer patients. Furthermore, in this population, performance status is consistently demonstrated to show an association with survival, although length of survival may depend on the underlying illness. For example, for patients with metastatic cancers with relatively good prognosis and treatment options, such as prostate or breast cancer, a KPS less than 60 correlates with a median survival of less than 6 months [10]. This is in contrast to cancers with poor prognosis, such as pancreatic or biliary cancers, which result in a median survival of less than 6 months for those with a much higher KPS score (less than 90) [10].

In addition to functional status, a range of symptoms and laboratory values have been demonstrated to correlate with survival. For instance, out of a myriad of symptoms studied, anorexia, weight loss, xerostomia, dysphagia, dyspnea, confusion, and cognitive decline have shown clear association with prognosis [4, 9]. The Palliative Prognostic Score combines a group of patient level characteristics (CPS, KPS score, symptoms, white blood cell count, lymphocyte count) into one prognostic score, recently updated and validated to include delirium as well [11].

11.5.2 Heart Failure

Like other diseases that involve chronic, progressive organ dysfunction, short-term survival is challenging to prognosticate for people with heart failure. It is clear though that the need for hospitalization in this population is associated with high mortality rates despite recent advances in heart failure management. In a study of 2.5 million Medicare beneficiaries hospitalized with heart failure, there was a reduction in in-hospital mortality from 5% to 4% between 2001 and 2005, although mortality at 30 days, 180 days, and 1 year remained unchanged at 11%, 26%, and 37%, respectively [12]. The prognosis is worse after each subsequent hospitalization for heart failure. The median survival in one study of older patients admitted for heart failure declined from 2.4 years in those with one hospitalization to 0.6 years for those with four hospitalizations [13]. The prognosis only worsens for the oldest old who have a median survival of 1 year after just a single hospitalization and a median survival of 6 months after two hospitalizations [13].

In addition to hospitalizations, there are several other markers of poor prognosis. These include patient demographic factors, heart failure severity, comorbid diseases, physical examination findings, and laboratory values including cachexia, hyponatremia, anemia, and NYHA Class. Two important prognostic tools to help clinicians estimate prognosis in heart failure include the EFFECT model and the Seattle Heart Failure Model. The EFFECT model was developed from a cohort of 2624 patients who presented with heart failure between 1999 and 2001. The factors used to stratify risk of death include age, comorbidities, and physiologic variables (respiratory rate, systolic pressure) and laboratory findings (blood urea nitrogen, serum sodium concentration, and hemoglobin) at the time of hospital presentation. Thirty-day-mortality risks range from 0.4% in the lowest risk group to 59% in the highest risk group. An online calculator for the EFFECT model can be found at www.ccort.ca/CHFriskmodel.aspx. The Seattle Heart Failure Model is a freely available although lengthy online calculator (http://depts.washington.edu/shfm/) that was developed and validated among outpatients participating in clinical trials, observational studies, and clinical registries. Caution is warranted though when generalizing the results to hospitalized patients or those with other major comorbidities such as renal failure, dementia, or cancer.

11.5.3 Advanced Dementia and Cognitive Decline

Individuals with advanced dementia typically have a prolonged period of severe functional disability, and during this time they are at risk of acute events such as pneumonias and urinary tract infections that are markers of very poor short-term survival. For instance, the 6-month mortality rates exceeded 50% in one study of individuals with advanced dementia who are admitted to the hospital with either pneumonia or a hip fracture.

Hospice eligibility criteria for dementia require that individuals need to meet or exceed stage 7a on the Functional Assessment Stage (FAST) scale (Table 11.4) and must have at least one complication from their dementia (aspiration, upper urinary tract infection, sepsis, multiple stage 3–4 ulcers, persistent fever, weight loss >10% within 6 months). However, studies have shown that these criteria fail to accurately predict 6-month survival in those with advanced disease. The Advanced Dementia Prognostic Tool (ADEPT), which can be found online at ePrognosis.org, can help identify nursing home residents with advanced dementia at high risk of death within 6 months. Compared to the hospice eligibility criteria, the ADEPT had greater predictive value of 6-month prognosis.

11.5.4 Chronic Liver Disease

Two of the most commonly used prognostic models used in chronic liver disease are the Child–Turcotte–Pugh (CTP) score and the Model for End-Stage Liver Disease (MELD). The CTP uses five variables (serum bilirubin, serum albumin, prothrombin time, ascites, and encephalopathy) to categorize patients into one of three classes (A, B, or C). The largest drawback to the CTP is the subjectivity in grading ascites and encephalopathy, and its limited ability to stage the severity along a broad continuum seen in liver disease as individuals are only placed into three categories. MELD is a prospectively developed and validated continuous scoring system that calculates the severity of chronic liver disease across a broader spectrum than the CTP. It uses a patient’s laboratory values for serum bilirubin, serum creatinine, and the international normalized ratio (INR) to predict survival. Mortality for patients awaiting liver transplantation is noted in Table 11.5. Of note, there are some conditions, such as hepatocellular carcinoma (HCC) and hepatopulmonary syndrome, in which the calculated MELD score overestimates survival. An online MELD calculator can be found at http://www.mayoclinic.org/meld/mayomodel6.html.