• comorbidity;
  • case mix;
  • quality of life;
  • prostate cancer;
  • outcomes


  1. Top of page
  2. Abstract


Among the most pressing challenges that face physicians who care for men with prostate cancer is evaluating the patient's potential for benefiting from treatment. Because prostate cancer often follows an indolent course, the presence and severity of comorbidities may influence the decision to treat the patient aggressively. The authors adapted the Total Illness Burden Index (TIBI) for use in decision-making among men with prostate cancer at the time of the visit.


An observational study was performed of 2894 participants in the Cancer of the Prostate Strategic Urologic Research Endeavor, a national disease registry of men with prostate cancer, to examine how well the adapted TIBI for prostate cancer (TIBI-CaP) predicted mortality over the subsequent 3.5 years and health-related quality of life over the subsequent 6 months.


The men who had the highest global TIBI-CaP scores were 13 times more likely to die of causes other than prostate cancer over a 3.5-year period than the men who had the lowest scores (hazard ratio, 13.1, 95% confidence interval, 6.3–27.4) after controlling for age, education, income, and race/ethnicity. Patients who had the highest TIBI-CaP scores had 44% mortality compared with 4.9% mortality for patients who had the lowest scores. Demographic variables explained 16% of the variance in future physical function; TIBI-CaP scores explained an additional 19% of the variance.


The TIBI-CaP, a patient-reported measure of comorbidity, identified patients at high risk for nonprostate cancer mortality. It predicted both mortality and future quality of life. The TIBI-CaP may aid physicians and patients in making appropriate treatment decisions. Cancer 2007. © 2007 American Cancer Society.

Among the most pressing challenges that face physicians who care for men with prostate cancer is evaluating, at the time of the visit, the individual patient's potential of benefiting from treatment. Prostate cancer often follows an indolent course. Men with early-stage prostate cancer may live for many years, even without active treatment. Accurate assessment of comorbid conditions, which, alone or in combination, could lead to early mortality and poor quality of life totally apart from the prostate cancer, may influence the decision to treat the patient aggressively.1–4 If a man has a high risk of mortality from coexistent disease, then he, his family, and the treating physicians must balance the reduced benefit of aggressive treatment against the rigors of aggressive treatment, including increased complications of surgery5 and longer term complications, such as incontinence and impotence.6 Whitmore articulated what has become an axiom in urology: For men in whom cure is possible, it may not be necessary; yet, for men in whom cure is necessary, it may not be possible.7

Currently, data sources available to the urologist, the radiation oncologist, or the primary care physician to summarize the impact of competing comorbidities that may influence decision-making are not optimal. Diagnoses often do not represent the severity and, thus, the prognosis of the condition and, furthermore, often are not readily available at the time of decision-making. Summary measures of patient-reported functional status have been shown to predict mortality but are based on variables with which clinicians are less comfortable than the review of systems variables that they collect routinely.8–10 Most measures of comorbidities are available only after treatment and have been used primarily to adjust group differences.10–16

Physicians often feel handicapped by the lack of empirical support for forecasting life expectancy, even among patients with terminal cancer.3, 17 Urologists and radiation oncologists often resort to the 10-year rule, an implicit and unstandardized assessment of comorbidity and age.18 Applying this rule, a patient who is expected to live for many years is offered aggressive treatment, whereas a man who is expected to die of other causes over the subsequent years is counseled that his best option is watchful waiting.

We adapted a patient-reported measure of comorbidity, the Total Illness Burden Index (TIBI),19, 20 for use in men with prostate cancer. The TIBI stratifies patients into risk groups based not only the presence but also on the severity of self-reported symptoms and past events. Because it can be completed and scored in the office, it can be available to physicians at the time of treatment, without the time delays and problems with completion rates that attend other approaches. In an earlier pilot test, the prostate-specific TIBI (TIBI-CaP) was associated with self-reported quality of life among men with prostate cancer at a single point in time.21 In the current study, we used the TIBI-CaP to predict mortality over the 3.5 years after study enrollment to determine whether it could be used as an adjunct to clinical decision-making for prostate cancer management. We also assessed the validity of the TIBI-CaP for predicting health status over a 6-month period.


  1. Top of page
  2. Abstract


Men were sampled from the Cancer of the Prostate Strategic Urologic Research Endeavor (CaPSURE), a longitudinal observational study of men with biopsy-proven prostate cancer.22, 23 The CaPSURE project included 40 urology practices across the United States (34 community-based, 3 academic, and 3 Veterans Affairs). Patients in this project were sampled without regard to age or treatment and were followed from enrollment through death or withdrawal from the study. Details of the study population and design have been reported elsewhere.22, 23 Physicians participating in CaPSURE provided data on patients' diagnostic work-up, treatments, and clinical follow-up. Participating patients reported demographic information at enrollment. Survey-based measures of patients' health status and health care resources used were included on questionnaires that were administered every 6 months after enrollment in CaPSURE.

In the fall of 2002, the TIBI-CaP was sent to the 4635 active CaPSURE participants. Of these, 4635 men, 3409 returned the questionnaires (74%). A global TIBI-CaP score was computed only when data were available for each of the 11 subdimensions. We eliminated 595 men for whom no items were completed for at least 1 subdimension. The 2894 men for whom adequate data to compute a global TIBI-CaP score were available constituted the final analytic sample. The study was approved by the Committee on Human Research at the University of California-San Francisco.



We modified the original version of the TIBI,19, 20 adding and deleting selected items to represent prostate cancer as the index condition based on results from the pilot study.21 The final TIBI-CaP instrument that was used in this study included 84 items in 11 subdimensions: pulmonary disease, heart disease, stroke and neurologic disease, gastrointestinal conditions, other cancers (excluding prostate), arthritis, foot and leg conditions, eye and vision conditions, hearing problems, hypertension, and diabetes.

Using the original TIBI scoring algorithm that assigned points based on severity of illness, we derived scores for each of the 11 subdimensions, with higher scores corresponding to greater severity of comorbidities. For example, an individual with shortness of breath at rest all or most of the time, cough with heavy sputum production, and continuous wheezing was classified in the highest severity category in the pulmonary disease dimension. To weight the subdimensions differentially, those that were considered to have the greatest clinical impact on illness burden were stratified, based on clinical judgment, into 4 severity levels (0–3 points), including subdimensions with intermediate impact in 3 severity levels (0–2 points) and subdimensions with the least impact in 2 severity levels (0–1). The severity weights for each subdimension were summed to create the TIBI-CaP global score, which could range from 0 to 23. Scores in our sample ranged from 0 to 18 with a mean score of 3.5 (standard deviation, 2.6) and a median score of 3.

The amount of missing data was small—no items had >10% missing, and 78 of 84 items had <4% missing. For those subdimensions with ≥1 item(s) missing, the subdimension score was computed on the completed items based on the scoring rules. For presentation purposes, we grouped TIBI-CaP scores into 5 aggregate levels with 3-point intervals from the least severe (scores of 0–2) to the most severe (scores >12).

Risk of prostate recurrence

For this analysis, patients were classified according to their pretreatment clinical risk using a modification of the classification reported by D'Amico et al, which groups patients into those with a high risk, intermediate risk, and low risk of developing recurrent prostate cancer.24

Health status

We measured patients' health status with the self-reported Medical Outcomes Study 36-Item Health Survey (SF-36), an instrument that has been tested widely among men with prostate cancer.25 In this article, we report data for 2 of the subscales of the SF-36 most closely related to chronic disease severity: physical functioning and role limitations caused by physical health problems (role-physical).25


Patient deaths reported to CaPSURE by next-of-kin or study physicians triggered requests for the state death certificate. If prostate cancer was listed as the primary or underlying cause of death, then the death was considered to be caused by prostate cancer. For mortality analyses, survival was calculated for patients who died as the number of days from the TIBI-CaP questionnaire to date of death or, for patients who remained alive, as the number of days from the TIBI-CaP questionnaire to last study contact. Our primary outcome was death not caused by prostate cancer; therefore, patients whose deaths were caused by prostate cancer were censored at their date of death.

Statistical Analysis

Kaplan-Meier analysis was used to evaluate the association between TIBI-CaP scores and mortality. Cox proportional-hazards regression was used to control for potential confounding variables. Results of the Cox models are presented as hazard ratios (HRs) and 95% confidence intervals (95% CIs).

We used Pearson correlation coefficients and analyses of variance to compare each TIBI-CaP subdimension with the participant's scores on the physical functioning and role-physical subscales of the SF-36. For each SF-36 subscale score that was considered separately as a dependent variable, we estimated 2 general linear models: a restricted model using patient demographics (age, education, income, race/ethnicity) alone and a full model using the TIBI-CaP with demographics. We used the explained variance (R2) between the full and restricted models to assess the relative contribution of the TIBI-CaP plus the demographic variables versus demographic variables alone in explaining variance in the subscales of the SF-36. All analyses were performed in version 9.1 of SAS (Cary, NC).


  1. Top of page
  2. Abstract

For the final analytic sample, the mean age was 69.3 years. Seventy-five percent of respondents were between ages 60 years and 80 years, most were Caucasian, and most had at least a high school education (see Table 1). Nearly 25% had been diagnosed with prostate cancer during the previous year, whereas 40% had been diagnosed between 1 year and 2 years previously. With respect to tumor characteristics, 39% of patients had low-risk disease, 38% had intermediate-risk disease, and 23% had high-risk disease. The mean number of years since the diagnosis of prostate cancer was 3.7 years. The mean age and the demographic characteristics of the 1741 nonrespondents (data not shown) were comparable to those of the final analytic sample, with the exception of the proportion of white patients (83% versus 93%, respectively; P < .0001) and the proportion of men with incomes <$20,000 per year (20% versus 12%, respectively; P < .0001).

Table 1. Characteristics of the Sample (n = 2894)
Sample characteristicMean (SD)
  1. SD indicates standard deviation.

Age, y69.3 (8.5)
Percentage nonwhite6.8
Percentage with less than a high school education12.1
Percentage household income <$20,000/y11.8
Percentage in D'Amico tumor risk category
 Low risk39.2
 Intermediate risk38.2
 High risk22.6
Years since prostate cancer diagnosis3.7 (3.5)


Of the 2894 participants who were followed for 3.5 years after administration of the TIBI-CaP questionnaire, 135 men died from causes other than prostate cancer (median 1.4 years; range, from 18 days to 3 years and 2 months). Of these 135 men, >60% died from cardiovascular or pulmonary disease or from other cancers and infections; the remaining patients died from a wide variety of causes. These patients constituted the group of interest when making treatment decisions. Twenty-six patients died of prostate cancer or its complications and, thus, were censored at their time of death. In the Cox proportionate-hazards models (Table 2), TIBI-CaP scores were related significantly to mortality after controlling for age, education, income, and race/ethnicity (HR, 13.1; 95% CI, 6.3–27.4). Men with global TIBI-CaP scores ≥12 were 13 times more likely than men with scores of 0 to 2 to die in this period from causes other than prostate cancer. Figure 1 shows the Kaplan-Meier survival graph for the men who died of nonprostate cancer causes. By 3.5 years, 56% (95% CI, 12–85%) of the men with global TIBI-CaP scores ≥12 had survived compared with 95% of the men with scores of 0 to 2 (P < .0001, log rank test). These mortality rates were not caused by cardiac disease or other cancers alone (data not shown).

thumbnail image

Figure 1. Kaplan-Meier analysis of mortality according to the Total Illness Burden Index for Prostate Cancer (TIBI-CaP) score. Of the 2894 participants who were followed for 3.5 years after administration of the TIBI-CaP questionnaire, 135 died from causes other than prostate cancer.

Download figure to PowerPoint

Table 2. Mortality Rates and Hazard Ratios for Aggregate Total Illness Burden Index for Prostate Cancer Scores (N = 2984 Participants)
TIBI-CaP score*No. of participantsNo. of deathsSurvival rate at 3.5 year (95% CI), %HR (95% CI)§
  • TIBI-CaP indicates Total Illness Burden Index for Prostate Cancer; 95% CI, 95% confidence interval; HR, hazard ratio.

  • *

    Scores range from the least severe (0–2) to the most severe (12+) aggregate comorbidities.

  • Deaths from causes other than prostate cancer.

  • The 3.5-year survival rate was calculated by using life table analysis. The median follow-up for survivors was 2.7 years.

  • §

    The HR was calculated by using a Cox proportional-hazards model adjusted for age, education, income, and race/ethnicity.

  • ||

    Reference group for comparisons.

0–211772495.1 (85.5–98.4)||
3–511334890.4 (80.8–95.3)1.6 (1.0–2.7)
6–84273889.8 (83.3–93.9)3.1 (1.8–5.2)
9–111141484.3 (67.0–93.0)4.7 (2.4–9.1)
≥12431156.1 (12.1–85.4)13.1 (6.3–27.4)

Health-Related Quality of Life

Among the 2607 men who completed the health status questionnaire 6 months after the administration of the TIBI-CaP, as the aggregate TIBI-CaP level increased, mean scores in both physical functioning and role-physical limitations caused by physical function decreased (see Table 3). Each increasing category of the aggregate TIBI-CaP had a corresponding, statistically significant decrease in levels of physical function and role-physical scores compared with the adjacent category.

Table 3. Relation of Aggregate Total Illness Burden Index for Prostate Cancer Scores to Medical Outcomes Study 36-Item Health Survey Subdimensions: Physical Function and Role Physical (N = 2607 Participants)*
TIBI-CaP scoreNo. of participantsMean (SD)
SF-36 physical functionSF-36 role physical
  • TIBI-CaP indicates Total Illness Burden Index for Prostate Cancer; SF-36, Medical Outcomes Study 36-Item Health Survey; SD, standard deviation.

  • *

    This sample reflects the deletion of 287 participants who did not complete the SF-36.

  • The mean (SD) of SF-36 score for each TIBI category is shown; scores range from 0 to 100 with higher scores indicating better function.

  • Scores range from the least severe (0–2) to the most severe (≥12) comorbidities.

0–2107491.3 (13.2)89.4 (24.7)
3–5102780.0 (21.3)71.2 (38.2)
6–837865.2 (25.9)46.8 (42.2)
9–119551.1 (26.8)30.7 (39.3)
≥123335.2 (18.5)10.2 (22.8)

With the TIBI-CaP treated as a continuous variable, we examined its contribution to variation in the subdimensions of the SF-36 compared with sociodemographic characteristics alone (see Table 4). Compared with the regression model that included sociodemographic characteristics alone, addition of the TIBI-CaP to the model explained significantly more variation in both functional status measures.

Table 4. Relation Between the Total Illness Burden Index for Prostate Cancer, Sociodemographic Characteristics, and Medical Outcomes Study 36-Item Health Survey Subdimensions (N = 2607 Participants)*
Regression modelVariableSF-36 physical functionSF-36 role physicalR2F value
R2F value
  • TIBI-CaP indicates Total Illness Burden Index for Prostate Cancer; SF-36 indicates Medical Outcomes Study 36-Item Health Survey; R2, explained variance.

  • *

    Table entries are variance explained (R2) in each of the SF-36 subdimensions separately by variables listed in regression Models 1 and 2.

  • Values from F tests comparing Model 2 with Model 1.

  • P < .001.

1Age, education, income, race/ethncity0.16NA0.11NA
2Age, education, income, race/ethnicity, TIBI-CaP0.35756.20.28585.4


  1. Top of page
  2. Abstract

Accurate assessment of noncancer morbidity is crucial to the selection of therapy for men with localized prostate cancer. To develop a more clinically useful and systematic method for assessing prognosis that takes into account a broad representation of the individual's medical conditions and their severity, we adapted and validated a new instrument, the TIBI-CaP. Because this instrument is self-reported, it can be applied in clinical settings among patients who are facing real-time treatment decisions for prostate cancer. It can be scored from data that are collected from patients at the time of the office visit rather than from often incomplete and post-hoc medical records or claims data, and can be made available to clinicians at the time of treatment decisions.

After modifying and refining the TIBI for use with prostate cancer as the index condition (TIBI-CaP), we tested the instrument in this large validation study among patients who were drawn from a national sample of men with prostate cancer of all stages. The final TIBI-CaP instrument predicted both 3.5-year mortality and health status measured 6 months after its administration. Physicians treating patients in Category 4, who had an average short-term mortality of >15%, or in the highest category, with an average mortality of nearly 44%, may make different decisions than those for patients in the lowest levels. The current analysis supports prior research showing that comorbidity has a significant relation to mortality in patients with prostate cancer.26, 27

Various comorbidity measures10–16 have been used by researchers to adjust for case mix and to predict mortality from competing causes in patients with prostate cancer. Although they may be reliable and valid for use in the research setting, they use post-hoc data sources (claims data or retrospective chart review) and include only the frequency of diagnoses, not the severity of conditions. Because individuals with prostate cancer face potential mortality from many other causes,28, 29 responsibility falls on the treating physician to estimate and balance these competing risks when providing advice about how and whether to treat the cancer. This study, in a relatively broad sample of men with prostate cancer, provides evidence for the validity of patient-reported severity of comorbid illness to aid physicians and patients in clinical decisions.

The TIBI-CaP can make the traditionally implicit process of estimating the impact of comorbidity explicit. It is based on symptoms, such as graded levels of shortness of breath, which have prognostic value independent of the exact underlying diagnosis. The aggregation of severity across conditions is unique to this measure. For example, a patient with shortness of breath while at rest, a productive cough, and a recent hospital visit for lung problems is likely to have severe respiratory disease regardless of their specific underlying diagnosis. This approach also relies on historic information, such as myocardial infarction, which patients can remember accurately. The combination of variables overcomes the potential unreliability of patient report of any 1 symptom or event.

TIBI-CaP scores were associated closely with scores on the SF-36 subscales that best represented the physical dimension of health-related quality of life, physical functioning, and role limitations caused by physical health problems as well as the other dimensions of physical function (data not shown). For each level increase in the aggregate TIBI-CaP score and each of its dimensions, SF-36 scores were significantly and meaningfully worse. It has been demonstrated that quality-of-life scores predict clinically important decrements in health among men with prostate cancer.30 The TIBI-CaP explained more than twice the variance that was explained by traditional sociodemographic variables.

This current study was limited by the preponderance of nonminority, relatively well-educated men in the CaPSURE database. The findings cannot be generalized beyond the types of practices studied in this research.

In addition, the observation period was only approximately 3.5 years. Although the predictive capability of the TIBI-CaP may be expected to increase as the cumulative effect of mild and moderate diseases become more burdensome over time, longer follow-up will be needed for study. Third, although the instrument is easy to complete, its length and self-administration may burden some patients unduly. The instrument may be able to be shortened; however, in its current form, which includes 84 questions, a patient can complete the questionnaire in <15 minutes while sitting in the waiting room. The evaluation of the instrument in both a test set21 and a validation set, both of which comprised a broad clinical scope of men with prostate cancer, strengthen our results.

The TIBI-CaP was developed for physicians in office practice. It may be completed by the patient or, for patients with cognitive impairment, by the patient's relatives or partner, either at home or in the waiting room before a visit in which treatment decisions will be discussed. It can be scored, like the SF-36, in an automated fashion, and the physician can discuss the placement of the patient in a prognostic category, inform the patient about the chance of misclassification, answer questions about which subdimensions are driving the prognosis, and assist the patient and their family in coming to the optimal solution for the patient's circumstances.

Future work should include shortening the instrument and testing its value in clinical settings prior to the treatment decision compared with a thorough patient history and physical examination. The TIBI-CaP may help guide physicians' clinical judgment when caring for men with prostate cancer.


  1. Top of page
  2. Abstract