HEALTH STATUS DATA TO OPTIMIZE PATIENT MANAGEMENT
BACKGROUND OF THE INVENTION
FIELD OF THE INVENTION The invention pertains to methods and systems in the field of computerized heath care management. More specifically, one system evaluates health status information for individuals to assess a statistical likelihood of the individuals experiencing a health outcome that is related to a specific disease or medical condition.
DESCRIPTION OF THE RELATED ART A goal of delivering medical care is to allow patients to live longer and to optimize their health-related quality of life. Even so, it is very difficult to achieve this goal in practice, partially because it is prohibitively expensive to use clinical screening techniques that can identify those patients who stand the most to benefit from treatment. Patients with the greatest potential to benefit from medical intervention are those who are severely compromised by the progression of disease, or those having a declining health status indicated by deteriorating symptoms, function and quality of life. Such patients are generally described as having a relatively poor health care-related quality of life in that the health status has diminished to a point where the patients recognize that they are significantly impaired by disease. In the case of cardiac patients, persons having a poor health care-related quality of life are statistically at the highest risk of subsequent mortality, as discussed in Green et al. (2000) Journal of the American College of Cardiology 35: 1245-55 and Spertus et al. (2000) Journal of the American College of Cardiology 35: 541).
By way of example, in the treatment of patients with blocked coronary arteries, the benefits of coronary revascularization have been most clearly demonstrated among those patients who are classified as those having the highest mortality rates. (Yusuf, et al (1994) Lancet, 334:563-70). Consequently, there is a need to identify patients with coronary artery disease (CAD) who are at high risk for future cardiac events.
Health status questionnaires may be useful in identifying patients who are not well from an overall heath perspective at the time of assessment. In the most general terms of methodology, these health surveys contain questions addressing health status issues that are believed to be related to specific diseases or medical conditions. People provide answers to these questions, and may be evaluated or assessed according to their answers. The information thus obtained has little use as a prognostic tool, though it may have value as a diagnostic aid or in a medical history showing progression of the disease.
In the realm of cardiovascular disease, widely accepted examples of health status surveys include both the Seattle Angina Questionnaii-e (SAQ) disclosed in Spertus et al (1995), Journal of the American College of Cardiology, 25: 1245-55) and the Kansas City Cardiomyopathy Questionnaire (Green CP, et al. (2000) Journal of the American College of Cardiology 35: 1245-55). These surveys provide valid, reliable and sensitive measures of disease-specific health-related quality of life. The SAQ and the Kansas City Cardiomyopathy Questionnaire described above are hereby incorporated by reference to the same extent as though fully replicated herein. These questionnaires contain specific questions related to cardiac health status.
A physician's history for patients with coronary artery disease usually addresses symptomology including angina, physical limitations, and quality of life; however, the scoring and interpretation of formal health status questionnaires is usually too burdensome to impose upon physicians in most clinical settings. While there is a general appreciation that symptomology may be important in diagnosis and may require medical intervention, health status monitoring information is not regarded as a valid predictive tool. A criticism of health status or quality of life questionnaires is that the data are
"soft." Researchers often consider that the data provided by patients are less reliable than that obtained by clinicians or physiologic tests. Statistical categorization of a patient's health status has no direct nexus to an actual likelihood of any, particular patient encountering an adverse health outcome, e.g., mortality. On the other hand, current techniques for assessing risk of coronary disease, e.g., nuclear stress tests, echocardiography, and angiography, are not cost-effective means of screening large numbers of patients to identify those with the greatest underlying risk. This may be
particularly important among patients with established cardiac disease for whom the disease may progress insidiously and for whom routine annual stress testing is expensive. A great many persons may be aware that they have a deteriorating health status that arises from an undiagnosed or untreated disease or medical condition. This happens because results from health status questionnaires are not generally useful or quantifiable as a prognostic or descriptive diagnostic tool that relates health status information to health outcome, and because conventional assessment techniques are too burdensome.
Expert systems are computerized systems designed to solve problems by processing information that is specific to the problem domain. The processing of information usually corresponds to rules or procedures that are applied by human experts to solve similar problems; however, expert systems may use additional input that cannot be fully considered by human experts and they may be capable of providing outcomes beyond those which people can provide. One example of an expert system in the health care field is described in United States Patent No.
5,517,405 to McAndrew et al. This system prompts users with questions directed towards a particular medical condition or symptom. The system consults a database or library to dynamically generate queries on the basis of user responses. Ultimately, the system provides a diagnosis together with a recommendation for a medical procedure or other treatment.
Another example of an expert system in the health care field is described in United States Patent No. 6,385,589 to Trusheim et al. The system accepts input from a variety of data sources that target the identification of a "medical event." Data sources towards this end include laboratory data confirming a disease diagnosis, pharmacy benefits manager data, hospital admissions data, physician records, home health care data, and health insurance data. The system also provides member characterization data, which is compiled using member surveys, care provider data, insurance claim data, and historical health care data. The system uses a combination of member characteristics and medical event data to identify risk situations, such as diabetes, hypertension, dementia and no caregiver support, unmedicated heart conditions, and discharge of elderly patients with no caregiver support. While member survey responses may be used in making these risk assessments, there is no
mention of using the survey responses to identify the nature and extent of the risk in terms of a statistical risk assessment. Thus, there may be diagnosed a condition where an elderly person has been discharged with no caregiver, but there is no assessment of whether the risk of being so discharged represents a mild risk or a severe one. The duty of a proactive follow-up after identification of a risk situation is left to a medical care provider who assesses the patient. Alternatively, the patient may be notified that he or she has been discharged without a caregiver and advised as to health plan options.
There remains a need for an automated system that analyzes health status information by comparing the health status against a statistically validated study to make an expert description or proactive recommendation in the form of a risk assessment.
SUMMARY The method and apparatus of the invention overcomes the problems outlined above and advances the art by demonstrating that health status information may be used as a prognostic or descriptive tool that uses statistical information from a study population to evaluate new survey responses and assign a likelihood of a person encountering a health outcome. For example, the health status of a cardiac patient may be assessed through use of the SAQ, and the patient's SAQ score may be compared to statistical information to assess the patient's likelihood of one year mortality.
One embodiment pertains to a method for evaluating health status information to identify a health outcome for a person. The method begins with a step of querying a person with health status questions related to a particular disease or medical condition. The person provides personal responses, which are received in response to the health status survey questions. The personal responses are evaluated through use of statistically validated study results that associate health status survey responses from a study population with actual adverse outcomes in a test group of persons. This evaluation produces a likelihood of the person encountering an outcome resulting from the particular disease or medical condition.
Prognostic results may be enhanced by screening or preselecting persons on the basis with commonalities with the test study group. For example, where the test study group was clinically diagnosed as having a particular medical condition, e.g., ACS or CHF, prognostic results are enhanced by pre-screening a large population to select persons who also have this condition. Still, this screening step is not required, and some uses for the concepts disclosed herein include use of health status questionnaires to persons who have not been clinically screened.
Examples of widely accepted health status surveys include the SAQ and the Kansas City Cardiomyopathy Questionnaire. The SAQ measures health status information that is relevant to ACS. The Kansas City Cardiomyopathy Questionnaire measures health status information that is relevant to decompensated heart failure. Other health survey may be used so long as information from the survey may be statistically correlated to an outcome.
Evaluating the health status surveys often comprises calculating a score on the basis of the personal responses and assigning the score to a predetermined range. For example, the score may be assigned to a one of a plurality of ranges each indicating a relatively greater likelihood of encountering an adverse outcome, such as mortality, hospitalization, poor quality of life, or even sociometric outcomes like divorce or poverty. These outcomes may be assessed as to their likelihood over an interval of time. Additional data, such as demographic or clinical data, may be used to correct prognostic likelihoods for factors that are unrelated to health status. The statistically validated study results used in evaluating the person's likelihood of encountering an outcome may, for example in the case of measuring cardiac health status, include data from the test group of persons regarding physical limitation, mortality, acute subsequent coronary syndrome, angina stability, symptom frequency, treatment satisfaction, quality of life, and self -efficacy scores.
In querying persons to elicit new responses for submission to the statistical model, it may be desirable to use a health status questionnaire that contains a plurality of questions in domains regarding at least one of a physical limitation, symptom stability, symptom frequency, treatment satisfaction and quality of life. Other domains may include, for example, physical limitation, symptom frequency, symptom stability, social function, self -efficacy and quality of life.
One example of the statistically validated model includes an algorithm that relates scores from the test group data to outcomes on the basis of stratified ranges of health status information scores indicating relative severity of the disease or medical condition. These may be used to determine projected odds of the person encountering an outcome over a period of time.
In another embodiment, the statistically validated model may include an algorithm representing a plot of health status domain scores on multiple axes. A spatial range is identified as one that links a confluence of scores having a special relationship to odds of the test group population in that spatial range encountering an outcome or representing an adverse constellation of disease characteristics.
Evaluating the person's responses may entail calculating a personal score on the basis of the personal responses, determining that the personal score falls within the spatial range, and assigning to the personal score the odds of encountering the outcome for the test group that falls within the spatial range. More generally, evaluating the personal scores may include calculating a personal score on the basis of the personal responses. The personal score is processed through use of a statistically-based algorithm that is inverted to assign odds of the person encountering an outcome.
The personal responses may be statistically linked with outcomes from a diagnostic database, e.g., an angiography or echocardiography database that associates diagnoses in a test study group with outcomes from treatments or procedures. By way of example, the recommendation may consider outcomes form one or more test study populations in recommending a pharmaceutical change in dosage or agent, use of a medical device, and/or a medical procedure. The foregoing methodology may be implemented in a computer system that is programmed to perform the method. The computer system may contain a user interface configured for presenting health status questions to the person and for receiving personal responses to the health status questions from the person. Data storage for the system may include a stored form of statistically validated survey results that associate the survey responses to health status questions with actual outcomes in a test study group of persons. A processor may be communicatively connected to the user interface and the stored form of statistically validated survey
results. The processor may be configured with program instructions for evaluating the personal responses by using the stored form of statistically validated survey results to assess the person's likelihood of encountering an outcome related to a particular disease or medical condition. In some embodiments, the methodology may be employed in the evaluation and decision making process for treatment plan in post-cardiac event patients. Currently, no such method for the post-coronary event patient is described or available.
The database may in another aspect provide a method by which the post- coronary or other clinical patient, or his attending health care professional, may create a decision matrix that can be used to consider and select the most appropriate revascularization or other post-coronary event intervention procedure for the patient. This provides the patient with a set of options that is specifically tailored for that patient, and provides the patient with an assessment of the relative benefits and disadvantages associated with selecting one or another of the options being presented. In another aspect, a computer based software system is devised so as to indicate relative risk associated with the selection of a particular revascularization protocol, given the specific health status of a given patient considering potential options post-heart attack. There may also be provided a computer based method for selecting an appropriate post-cardiac clinical regimen for an individual patient having had a cardiovascular disease episode. The method entails storing group data in a database that contains group responses to a questionnaire. The questionnaire has a plurality of questions regarding quality of life and demographic information from a plurality of patients who have survived a coronary event post 1 year. By way of example, a first group of the plurality of the patients may have received a post-coronary event revascularization procedure, and a second group of the plurality of patients have not received post-coronary event revascularization procedure. A comparison may be selected form the group such that the demographic statistics for selected members of the group are similar to those of the individual patient. A questionnaire may be presented to the individual patient, which the patient completes to provide patient responses to the questionnaire. The responses provide an individual patient quality of
life assessment and demographic data profile, which may be matched to the group responses in the database. The methodology then entails performing a statistical analysis on the individual patient quality of life profile, wherein the statistical analysis indicates to the individual patient an appropriate post-cardiac clinical regimen. A revascularization procedure selected by the patient may include, for example, a coronary artery bypass grafting (CABG) procedure and/or a Percutaneous Coronary Intervention (PCI) procedure. The questionnaire may include, for example, questions to assess symptoms such as frequency of angina, relative physical limitations, satisfaction level with current medical treatment, and perceived daily living limitations. Statistical processing can include computing a statistical average or deviation of data associated with similarly situated people who have undergone a clinical option. Clinical options may include surgical procedures, dietary adjustments, herbal remedies, medication, such as anti-thrombotic medications, or no treatment.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a process schematic diagram that illustrates the creation of a statistical model through the use of data collection and statistical processing, together with inversion and prognostic use of the statistical model in evaluating new responses; Fig. 2 shows a prognostic statistical model including Kaplan-Meir calculation results that demonstrate increasing mortality over time on the basis of SAQ scores that have been stratified in quartiles of relative disease severity;
Fig. 3 shows a prognostic statistical model including a bar graph that relates SAQ physical limitation scores to one year mortality in the test study group;
Fig. 4 shows a prognostic statistical model including a bar graph that relates SAQ angina frequency scores to one year mortality in the test study group;
Fig. 5 shows a prognostic statistical model including a bar graph that relates SAQ angina stability scores to one year mortality in the test study group;
Fig. 6 shows a prognostic statistical model including a bar graph' that relates SAQ quality of life scores to one year mortality in the test study group; Fig. 7 shows a descriptive statistical model including a plot of scores from two different health status domains including SAQ Angina Frequency Score and SAQ
Physical Limitation Score, such that spatial regions of the plot are associated with extremely high risk of mortality;
Fig. 8 shows a descriptive statistical model including a plot of scores from two different health status domains including KCCQ Summary Score and KCCQ Self- Efficacy Domain Score; such that spatial regions of the plot are associated with extremely high risk of mortality;
Fig. 9 shows a descriptive statistical model including a plot of scores from two different health status domains including SAQ Angina Frequency Score and SAQ Disease Perception/Quality of Life Score, such that spatial regions of the plot are associated with patients experiencing frequent symptoms that is having an extremely adverse effect upon their quality of life
Fig. 10 shows a medical evaluation system, in accord with one embodiment of the invention;
Fig. 11 shows a flow chart illustrating an operation of the system of Fig. 10; Figs. 12A and 12B are comparative graphs showing that angina frequency may vary depending upon the type of clinical procedure a patient has undergone and the demographic background of the patient.
Fig. 13 shows comparative statistics including complications and angina frequency according to different therapeutic options; Fig. 14 is a system schematic diagram illustrating various system components that may be used to assist patients and medical workers in selecting a clinical procedure based upon a desired patient goal or outcome; diagram illustrating data and process flow according to one embodiment;
Fig. 15 shows additional aspects that may be included in the system of Fig. 14 according to one embodiment;
Fig. 16 is a flow diagram of process steps to obtain and to store data according to one embodiment;
Fig. 17 is a flow diagram of process steps to obtain stored medical data according to one embodiment; Fig. 18 is a flow chart illustrating additional detail with respect to Fig. 17;
Fig. 19 is a flow chart illustrating additional detail with respect to Fig. 17;
Fig. 20 is a flow chart illustrating additional detail with respect to Fig. 17; and
Fig. 21 is a flow chart illustrating additional detail with respect to Fig. 17;
DETAILED DESCRIPTION Traditional use of health status information has been limited to clinical evaluation or similar studies indicating a present status in the progression of a disease, in quantifying the burden of disease, and disease-related impact on health related quality of life. Experts may use these data in making clinical assessments; however, the prognostic significance of a health status score is traditionally unknown. Use of this information in prognostic fashion, according to the description below, facilitates systematic results that describe a patient's health status from new perspectives. Accordingly, physicians and patients can make better-informed decisions as to medical intervention, and the risks of intervention can now be quantitatively balanced against what the patient might expect if intervention does not occur.
The following discussion of processing health status information illustrates a particular embodiment in the field of cardiac health status information by way of example, not by limitation. The principles described below may be generally followed to produce analogous results in any field of health care for any disease or conditions.
Fig. 1 is a block diagram illustrating a general process 100 for practicing the aforementioned concepts. The process 100 may be implemented through the use of program instructions on a single computer, in a distributed processing environment, or with human intervention at respective illustrated steps. Step 102 entails the creation or identification of a survey designed to procure health status information, and possibly other information that may sometimes overlap with health status information, such as demographic information and clinical information. Health status information means information that focuses upon a disease or medical condition as measured by symptomology, physical effects (e.g., limited range of motion or lifting capability), or mental effects of the disease (e.g., depression or anxiety). The health status information may also monitor the effects of treatment. In these contexts, demographic data means descriptive information that could be used to characterize or classify a person in a statistically meamngful way, for example, age, smoking habits, income, geographic location, marital status, race, gender, military service, or drinking
habits. Clinical information may include medical history information about the person or diagnostic test results from that person.
Particularly useful clinical information includes the identification of adverse outcomes, which are defined as a medical event that people normally wish to avoid. Adverse outcomes may include, for example, hospitalizations, death, onset of mental depression, and surgery. The clinical data may also include positive outcomes, such as five year cancer survival, cancer remission, or the effectiveness of a class of therapeutic modality in treating a disease.
A variety of health status questionnaires are available and generally known to medical practitioners. Many of these questionnaires have been validated by direct or indirect means to show that patient responses to the questionnaires bear some relationship in describing a present disease or medical condition. Pre-existing questionnaires may be selected for use as the survey in step 102. This manner of selection accelerates the time required for completion of step 102 A survey may be created in step 102 using research that identifies symptoms and descriptions of a medical disease or condition. This research is preferably completed by a medical expert as to the disease or medical condition. The questions may be prepared to present survey respondents with a variety of selectable options that are each assigned a score on a relative scale that indicates a relative severity of the disease or medical condition. The overall survey may be scored on the basis of accumulated scores from some or all answers to the questions. The questions may be scored on the basis of the overall score or subsets of questions addressing domains or categories of disease symptomology, for example, domains of physical limitation, severity of symptomology, frequency of symptomology, quality of life, and satisfaction with treatment. The creation of health status surveys for purposes of using them according to the various instrumentalities and embodiments of the invention does not fundamentally differ from traditional techniques of producing these surveys or questionnaires.
In step 104, the survey that is created or identified in step 102 may be administered to a test study group of persons who provide responses 106.
Administration may occur by using written or electronic instrumentalities. Additional data including clinical or demographic data may optionally be obtained from
additional responses 106 or from electronic data storage 108, such as the database of a health insurance company, medical informatics company, medical hospital, or governmental agency.
The data-gathering step 104 may proceed over a period of time, such that the survey responses 106 may continually update throughout this period, with the initial response forming a baseline. The survey responses are scored by a suitable scoring system, and the data is subjected to statistical processing in steps 110 and 112. Univariate statistics may be calculated in step 110 to identify the most significant predictors of future health status. The statistical results may relate health status information from the responses 106 to adverse outcomes and/or positive outcomes in the clinical data. The health status information may be used to stratify the outcomes into ranges. Still further, the statistical information may be related to outcomes over a period of time. Demographic data and clinical data may be used to correct outcomes for factors that are not directly related to the medical condition that is the focus of the health status survey, such as by correcting overall mortality in a cardiac study for mortality arising from diabetes.
In step 114, the results of statistical calculations from steps 110 and 112 may be subjected to expert review, for example, review by medical and statistical experts in the field. The lessons gleaned from this review may be represented by program logic and stored in a rules base 116. Statistical correlations or algorithms may also be created for use in prognostic modeling. A particularly useful technique, for example, is the Kaplan-Meir method of relating a parameter to an adverse outcome over time, a plurality of Kaplan-Meir curves may, for example, represent stratified groupings of health status information. The statistical results may also be related to health status information, demographic data, and/or clinical data through use of artificial intelligence or self-training algorithms, such as neural networks or adaptive filters. The result of step may produce an inverted statistical model or rules that may be used to assess probabilities of outcomes on the basis of health status information. In step 118, a person (such as a patient or other individual) may be administered a questionnaire that is identical, or at least statistically identical, to the survey that was created or identified in step 102. Additional surveys may be administered or other data sources may be used to obtain, as completely as possible
from the person, an identical set of data in comparison to the data that was gathered from individuals in the test study group in step 104. Step 120 may include submitting the personal input data from step 118 for processing through the inverted model derived from step 114. The result of the evaluation modeling step 120 may be to assess the probabilities of outcomes on the basis of the person's health status information.
The results from step 120 may be categorized in step 122, e.g., by the use of delimiting values or rules-based groupings health status scores, to select persons who are at a relatively greater or lower risk of having a particular outcome. In step 124, a recommendation may be generated, for example, that the person seek out a physician for treatment, or that the person may wish to consider one medical procedure over another.
The SAQ is a well-known and widely-accepted health status questionnaire providing a disease-specific health status measure for patients with coronary artery disease; however, the questionnaire has not heretofore been shown to be a valid predictive tool. There will now be demonstrated, by way of example and not of limitation, a methodology according to the process 100 for relating health status measures to adverse outcomes, such as mortality, hospitalization and poor quality of life.
EXAMPLE 1
SURVEY DATA COLLECTION Example 1 demonstrates data collection techniques for gathering data that were used as steps 102 and 104 of Fig. 1. In some cases, an expert devised questions to obtain data possibly having some relevancy in assessing prognostic outcomes. In other cases, a well-accepted study may be selected for this purpose.
A literature search identified the SAQ as a leading survey pertaining to health status of acute coronary syndromes (ACS) or coronary artery disease (CAD). A total of 5,558 Veteran's Administration patients reporting coronary artery disease were administered the SAQ. Patient responses to the SAQ were evaluated by conventional SAQ assessment techniques to provide scores assessing physical limitation characteristics according to traditional risk variables including physical limitation,
angina stability, angina frequency, CAD-specific treatment satisfaction and quality of life. The patients were also asked to provide demographic information and medical history data. Selection of the SAQ for this example was made because the questionnaire is generally accepted to be a valid indicator of the status it was designed to measure, namely, angina stability, angina frequency and quality of life. SAQ selection was also made because the SAQ is generally accepted as a valid clinical tool by the medical profession. The SAQ is a self-administered, disease specific measure for patients with coronary artery disease. The questionnaire is valid, reproducible, and sensitive to clinical change. SAQ responses were obtained from a test study group of cardiac patients. The
SAQ were scored to quantify the patients in a plurality of domains. The domains included physical limitations due to angina, the frequency and recent change in patient symptoms, patient satisfaction with treatment, and the degree to which patients perceive that disease has impacted their quality of life. The scores were transformed assess the scores on a distribution across a range of 0 to 100, according to conventional SAQ methodology where higher scores indicate better status function, e.g., less physical limitation, less angina and better quality of life. The normalized scores from the test study group were divided into quartile ranges of increasing severity for ease of interpretation. The SAQ domains of particular study included physical limitation, angina, and better quality of life.
Demographic or clinical data not directly linked to health status information was shown to have statistical significance in using health status data to assess outcomes including morbidity and ACS hospital admission. Data was collected from the survey respondents who were asked to provide information including: • gender;
• race;
• medical comorbidities arising from or related to previous hospitalization, cancer, chronic heartburn, depression, drug abuse, kidney problems, liver disease (yellow jaundice or hepatitis), post traumatic stress disorder, seizures or convulsions, peptic ulcer disease, and thyroid disease;
• cardiac risk factors including diabetes, hypertension, and smoking history; and
• cardiovascular disease severity factors indicated by stroke, congestive heart failure (CHF), previous hospitalization for ACS, prior coronary revascularization procedure, and prior myocardial infarction (MI).
Additional demographic data was obtained using standard questionnaires that were also scored, such as the SF-36, as described in McHorney et al., The MOS-36 Item Short-form Health Survey (SF-36): II Psychometric and clinical test of validity in measuring physical and mental health constructs, Med. Care. 1193; 31:247-63. Outcomes were tracked for each patient during a two-year course of study.
The primary outcome was one-year mortality. Mortality was tracked using information provided by the Veteran's Administration. A second outcome tracked during the two-year course of the study was hospital admission for ACS.
EXAMPLE 2
STATISTICAL PROCESSING OF THE SURVEY DATA COLLECTION
The data collected in Example 1 was subjected to statistical processing, e.g., as generally shown in steps 110 and 112 of Fig. 1 to create a model relating the respective adverse outcomes to scores on the SAQ. All analyses were performed using commercially available statistical computer programs, namely, SAS™ Version 6.12 (a trademark of SAS Institute, Inc. of Gary North Carolina) and S-Plus 2000™ for Windows (a trademark of Statistical Sciences, Inc., of Seattle Washington).
Statistical curves for all cause mortality were calculated using Kaplan-Meir analysis.
The data were stratified by quartiles of the initial or baseline SAQ scores. Log-rank tests were used to compare data from the respective groups to establish that patients with lower scores initially died sooner and more frequently than patients reporting higher scores. Predictive models were developed for one year mortality and ACS hospitalization, by logistic regression.
Three models were constructed for each of the adverse outcomes including mortality and hospitalization. The first model included only SAQ domain scores. The second model included only demographic data and clinical variables. In each case, the models were developed by incorporating predictors with significant
univariate associations with outcome (probability value less than or equal to 0.1) and using backward elimination until all remaining factors were significant at the 0.1 level. Age was entered as a continuous predictor, and the remaining independent variables were categorical. A final model was developed by combining significant variables from the first two models to estimate the incremental prognostic value of the SAQ score above demographic and clinical characteristics alone. For all modes reported in Table 1, probability values and estimated odds ratios are presented.
Model calibration was assessed by comparing predicted and observed outcomes by deciles of predicted risk. Goodness of fit was determined by the Hosmer-Leslow test. Discriminatory power was evaluated by use of c-statistics and compared by Mann- Whitney tests, as described in DeLong et al., Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics, 1988; 44; 837-845. Model calibration and calibration were validated on 200 bootstrap resamples of the original data set, as described in Harrell et al, Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors, Stat. Med. 1996; 15:361-387.
In fewer than 0.3% of the cases, baseline and clinical variables were missing from the input data. The missing baseline and clinical data was imputed to the mean or by random sampling of data. Nineteen percent of patients had missing SAQ scores because of incomplete questionnaires. The one year mortality for these patients was higher (7.5% versus 5.3) and the ACS hospital admission was lower (2.3% versus 3.6) than patients who filled out completed questionnaires. Primary analyses were performed on those with complete SAQ data. Supporting analyses were performed to determine the impact of missing data were conducted with multiple imputation techniques that conditioned upon the available SAQ and SF-36 scores, according to techniques described in Little, etal, Statistical Analysis with Missing Data, New York, NY, Wiley; 1987. S-plus subroutines developed by Frank Harrell available for this purpose as Harrell, Hmisc S-Plus function library in lib.stat.cmu.edu; 1999.
EXAMPLE 3
UNTVARIATE CALCULATION RESULTS
The results of statistical calculations in Example 2 were inverted to provide a model for prognostic use. Table 1 provides the baseline characteristics and univariate odds ratios for one-year mortality and hospitalization. The mean SF-36 PCS was 30.8±10.5, indicating that these patients scored almost two standard deviations below the mean United States population in terms of general physical health status. Over the one year of observation, there were 238 deaths (5.3% of the test population) and 154 ACS hospital admissions (3.6% of the test population).
Table 1 describes the relative odds of death at one year by demographic characteristics, comorbidities, and SAQ scores. The most powerful predictive parameters were age, presence of congestive heart failure, and health status. In Table 1, the "Odds Ratio" for "One- Year Mortality" is calculated as the actual mortality divided by the "Minimal" mortality. The low p-values show that the reported modes are statistically valid predictors because the p-values are less than an arbitrary delimiting value, in this case less than 0.1. Thus, the "Odds Ratio" for the "Minimal" quartile is understood to be 1.0. A person in the "Severe" quartile is 6.2 times more likely to die than a person in the "Minimal" quartile. Table 1 provides a model for prognostic use because Table 1 may be used to relate a person's health status, demographic, or clinical information, with an outcome likelihood in the domain of mortality or ACS hospitalization. This is done, for example, by assessing the person's SAQ score as to physical limitation and entering Table 1 to determine the likelihood. This modeling capability may, for example, be performed through program instructions that apply an expert set of rules in reporting from a database.
Fig. 2 shows the Kaplan-Meir survival curves over time for the SAQ physical limitation domain, which was broken out into quartiles representing severe (0-24), moderate (25-49), mild (50-74), and minimal (75 to 100). Separation of the curves, especially for those having the worst function, begins immediately and continues over the two-year period of observation. The mortality rates shown in Fig. 2 are the actual mortality rates, as opposed to the odds ratio shown in Table 1. Linear regression, such as a least squares technique, may be used to calculate an algorithm for each of the stratified domains to obtain a prognostic model on the basis of SAQ score. As shown in Table 1, SAQ scores were divided into domains including physical limitation, angina stability, angina frequency, and quality of life.
Specifically, the SAQ was scored in a different manner for each domain to assess these parameters. The domains were divided into SAQ ranges. Table 1 also describes the odds of hospital admission for ACS. The strongest predictor of admission was a previous ACS-related hospitalization. The next most powerful predictors were SAQ angina frequency, physical limitation, and quality of life domains. For example, those reporting severe angina were 3.1% more likely to be admitted to the hospital for ACS than those reporting minimal angina.
TABLE 1
BASELINE CHARACTERISTICS AND UNIVARIATE ODDS RAΗOS
One- Year Mortality One- Year ACS
Admission
Odds Ratio Odds Ratio
(95% CI) (95% CI) n = 4,484 p-value p-value
Age (years)* 67 ± 10 1.6 <0.001 1.0 0.895 (1-4, 1.9) (0.8, 1.2)
Male 4405 (98.3%) 1.4 0.58 N/A 0.11 (0.5, 5.6)
Caucasian 3697 (85.6%) 1.1 0.71 1.0 0.97 (0.7, 1.6) (0.7, 1.6)
Prior hospitalization (any) 720 (16.1%) 1.7 0.001 4.6 <0.001 (1.2, 2.3) (3.3, 6.4)
Prior hospitalization for ACS 148 (3.3%) 1.6 0.12 10.3 <0.001
(0.8, 2.8) (6.6, 15.7)
Cancer 612 (13.7%) 1.9 <0.001 0.9 0.77 (1.4, 2.6) (0.6, 1.5)
GERD 1059 (23.6%) 0.8 0.20 1.1 0.53 (0.6, 1.1) (0.8, 1.6)
Congestive Heart Failure (CHF) 887 (19.8%) 2.4 <0.001 1.5 0.02 (1.8, 3.2) (1.1, 2.2)
Depression 1280 (28.6%) 1.0 0.76 1.1 0.69 (0.8, 1.4) (0.8, 1.5)
Diabetes 1283 (28.6%) 1.4 0.01 1.4 0.05 (1.1, 1.9) (1.0, 1.9)
Drug abuse 91 (2.0%) 0.8 0.70 0.3 0.21 (0.2, 2.0) (0.0, 1.4)
MI 2355 (52.5%) 1.4 0.02 1.6 0.004 (1.1, 1.8) (1.2, 2.3)
Hypertension 2853 (63.6%) 0.8 0.11 1.0 0.95 (0.6, 1.1) (0.7, 1.4)
Smoking status 0.12 0.24
Never 686 (15.6%) - -
Past 2721 (62.0%) 1.3 1.3 (0.8, 1.9) (0.8, 2.3)
Current 980 (22.3%) 1.6 1.6 (1.0, 2.5) (0.9, 2.9)
Chronic renal insufficiency 694 (15.5%) 1.6 0.005 1.1 0.55 (1.1, 2.1) (0.7, 1.7)
Liver disease 320 (7.1%) 1.3 0.30 0.6 0.22 (0.8, 2.0) (0.3, 1.2)
Post-traumatic stress disorder 516 (11.5%) 1.1 0.74 1.3 0.27 (0.7, 1.6) (0.8, 2.0)
Seizures/convulsions 187 (4.2%) 1.0 0.98 0.4 0.16 (0.5, 1.8) (0.1, 1.2)
Peptic ulcer disease 984 (21.9%) 1.2 0.28 1.1 0.64 (0.9, 1.6) (0.7, 1.6)
Stroke 742 (16.6%) 1.6 0.002 1.4 0.07 (1.2, 2.2) (1.0, 2.1)
Thyroid disease 196 (4.4%) 1.6 0.07 0.6 0.31 (0.9, 2.7) (0.2, 1.4)
Prior PTCA or CABG 2138 (47.7%) 1.2 0.16 1.9 <0.001 (0.9, 1.6) (1.4, 2.6)
SAQ Physical Limitation <0.001 <0.001
Minimal 856 (19.1%) - -
Mild 1239 (27.6%) 1.6 1.6 (0.9, 2.8) (0.9, 3.0)
Moderate 1728 (38.5%) 2.5 2.7 (1.5, 4.2) (1.6, 5.0)
Severe 661 (14.7%) 6.2 2.8 (3.7, 10.3) (1.5, 5.3)
SAQ Angina Stability <0.001 0.05
Much better 831 (18.5%) - -
Slightly better 439 (9.8%) 1.3 1.7 (0.8, 2.2) (0.8, 3.5)
Unchanged 2339 (52.2%) 1.2 1.9 (0.8, 1.8) (1.2, 3.4)
Slightly worse 576 (12.9%) 1.1 2.5 (0.7, 1.9) (1.4, 4.8)
Much worse 299 (6.7%) 2.9 2.3 (1.8, 4.8) (1.1, 4.9)
SAQ Angina Frequency <0.001 <0.001
Minimal 2437 (54.4%) - -
Mild 1268 (28.3%) 1.0 1.6 (0.7, 1.4) (1.1, 2.3)
Moderate 539 (12.0%) 1.8 2.6 (1.3, 2.6) (1.6, 4.0)
Severe 240 (5.4%) 2.6 3.1 (1.6, 4.0) (1.7, 5.3)
SAQ Quality of Life 0.002 <0.001
Excellent 1791 (39.9%) - -
Good 1461 (32.6%) 1.3 2.1 (0.9, 1.7) (1.4, 3.3)
Fair 916 (20.4%) 1.5 3.2 (1.1, 2.2) (2.0, 4.9)
Poor 316 (7.1%) 2.4 2.2 (1.5, 3.7) (1.1, 4.1)
* Odds ratios per +10 years
The information from Table 1 may be depicted graphically and used in a three dimensional model. For example, the Kaplan-Meir curves shown in Fig. 2 may be linked to SAQ scores through a common mortality. Accordingly, the model is able to project that a person with an initial SAQ baseline in one of the stratified rankings may cross the boundary into another stratification having a higher mortality at a future point in time. For example, Fig. 3 is a bar graph providing point-in-time graphical representation of actual mortality rates for SAQ quartile ranges of the physical limitation domain. Fig. 4 is a bar graph providing a graphical representation of actual
mortality results for the angina stability domain, which was divided into SAQ ranges including Much Worse (0-24), Slightly Worse (25-49), Unchanged (50), Slightly Better (51-75), and Much Better (76-100). Fig. 5 is a bar graph providing a graphical representation of results for the angina stability domain. Fig. 6 is a bar graph providing a graphical representation of quartile results for the quality of life domain. In this particular study group, the strongest health status predictor was SAQ physical limitation. Angina frequency and quality of life domains all had a significant trend toward lower mortality with higher patient function, fewer symptoms, and better quality of life. The angina stability SAQ scores measure increased angina. Those patients reporting increased angina over the past month had a one year mortality rate of 11.4% compared with 4.9% for the rest of the study population. An expert-defined rules base may concentrate upon making prognostic assessments of outcome likelihood on the basis of the strongest predictors, or on the basis of a combination of the strongest predictors. For example, a combination of predictors may be used to enhance overall results through use of a multivariate prognostic model that is produced using the strongest predictors, as shown in Example 4.
EXAMPLE 4
MULΗVARIABLE PROGNOSTIC MODELS As shown above, physical limitation and angina frequency were the best SAQ- based predictors for both one year mortality and ACS hospital admission. These predictors were studied in combination with demographic and clinical data, as shown in Tables 2 and 3. The effect of the multivariable analysis was, for example, to correct the univariate results shown in Table 1 for demographic and clinical factors that are unrelated to ACS, and vice versa. For example, the odds ratio for the "Severe" range of the "Physical Limitation" domain is shown as being 6.0 in the univariate approach and 4.0 when corrected for the contribution of other factors, such as cancer, age, previous hospitalization, and CHF. In addition to adjusting the magnitude of the scores in a minor way, these multivariable results confirmed that SAQ scores in the physical limitation and angina frequency domains are valid predictors of future mortality and ACS hospitalization.
TABLE 2
PROGNOSTIC MODELS FOR ONE- YEAR MORTALITY
Model 1 Model 2 Model 3
(SAQ) (Clinical) (Combined)
Odds Ratio Odds Ratio Odds Ratio
(95% CI) (95% CI) (95% CI) p-value p-value p-value
Physical Limitation < 0.001 < 0.001
Minimal - -
Mild 1.7 1.5 (1.0, 3.0) (0.9, 2.7)
Moderate 2.6 2.0 (1.6, 4.4) (1.2, 3.5)
Severe 6.0 4.0 (3.6, 10.6) (2.4, 7.2)
Angina Frequency 0.068 0.078
Minimal - -
Mild 0.8 0.8 (0.5, 1.1) (0.6, 1.2)
Moderate 1.1 1.2 (0.8, 1.6) (0.8, 1.8)
Severe 1.4 1.6 (0.9, 2.2) (0.9, 2.5)
Age (per +10 years) 1.6 < 0.001 1.6 < 0.001
(1 A 1.9) (1.3, 1.8)
CHF 2.3 < 0.001 1.9 < 0.001
(1.7, 3.0) (1.4, 2.5)
Cancer 1.5 0.015 1.5 0.031
(1.1, 2.1) (1.0, 2.0)
Prior hospitalization 1.5 0.024 1.4 0.065
(1.1, 2.0) (1.0, 1.9)
Diabetes 1.3 0.059 1.3 0.129
(1.0, 1.7) (0.9, 1.7)
Stroke 1.3 0.092 1.2 0.396
(1.0, 1.8) (0.8, 1.6) c-statistic
Original dataset 0.66 0.69 0.72
Validated 0.66 0.67 0.70
Hosmer-Lemeshow p-value 0.57 0.16 0.26
TABLE 3
PROGNOSTIC MODELS FOR ONE-YEAR ACS ADMISSION
Model 1 Model 2 Model 3
(SAQ) (Clinical) (Combined)
Odds Ratio Odds Ratio Odds Ratio
(95% CI) (95% CI) (95% CI)
' p-value P -value p-value
Physical Limitation 0.030 0.102
Minimal - -
Mild 1.4 1.5 (0.7, 2.6) (0.8, 2.9)
Moderate 2.1 2.0 (1.2, 3.8) (1.1, 3.8)
Severe 1.9 1.8 (1.0, 3.7) (0.9, 3.6)
Angina Frequency 0.005 0.016
Minimal - -
Mild 1.4 1.4 (0.9, 2.0) (0.9, 2.0)
Moderate 2.1 2.0 (1.3, 3.3) (1.2, 3.2)
Severe 2.4 2.2 (1.3, 4.3) (1.2, 4.2)
Prior hosp (ACS) 11.7 < 0.001 10.6 < 0.001 (7.4, 18.1) (6.7, 16.7)
Prior hosp (other) 2.8 < 0.001 2.7 < 0.001 (1.9, 4.2) (1.8, 4.0)
Prior PTCA/CABG 1.6 0.005 1.6 0.004
(1.2, 2.3) (1.2, 2.3) c-statistic
Original dataset 0.63 0.69 0.73
Validated 0.62 0.66 0.70
Hosmer-Lemeshow p-value 0.32 0.87 0.80
EXAMPLE 5
PERSONAL INPUT AND EVALUATION MODELING The models shown in Tables 1-3 are used as inverted models in the sense that they were created in a forward manner from the analysis of vast amounts of statistical data, but they may be used in a reverse manner by:
• accepting personal input data;
• calculating a SAQ score; and
• entering the tables to determine whether the person has a relatively greater risk of encountering an adverse outcome according to any particular modality.
In a hypothetical example, John Doe is hospitalized for a mild form of cancer. As a routine matter to pass time while undergoing treatment, a doctor administers the SAQ. John dutifully provides a response, which is scanned and scored electronically in accord with the personal input step 118 of Fig. 1. The score places John in a severe range for physical limitation and a severe range for angina frequency. Accordingly, in the evaluation step 120 an electronic system accesses the model shown in Table 2 to show that John has a 4X greater chance dying in the next year than does someone with minimal physical limitation, and a 1.6X greater chance of dying than does someone with minimal angina frequency. The system can also consult Fig. 3 to ascertain that his overall one year mortality rate, as someone with severe physical limitation, is about 12%.
EXAMPLE 6
CATEGORIZATION AND RECOMMENDATION In Example 5, John may be categorized as a man, as a man who has cancer, as a man who has severe physical limitation, and as a person with severe angina. As a person with cancer, John may stand a greater chance of dying from cancer than from CAD over the next year. Due to his particular cancer and the treatment employed, it may not be practicable to implement treatment for his CAD, or only a few treatments may be workable. Other options may be affected by his age in combination with his
cancer. These categorizations may be assisted by demographic or clinical data obtained for his personal input.
On the basis of categorization data and input from the expert-defined rules base 116, a medical recommendation in step 124 may be rendered according to the categorization.
EXAMPLE 7
DISEASE MANAGEMENT PROGRAM FIG. 7 demonstrates the application of the data from Figs. 2-6 into a disease management model. Fig. 7 provides an example of the general principle of leveraging the health status scores of populations of patients to identify high risk individuals. SAQ angina frequency scores (abscissa) are plotted against SAQ physical limitation scores (ordinate) for the same person to form a distribution. Those patients in the lower left quadrant have frequent angina (SAQ angina frequency scores <50) and marked physical limitation (SAQ Physical Limitation <50). Statistical calculations show that those patients in the lower left quadrant have a 20.4% chance of dying over the next 2 years. Those in the upper left quadrant have a 12.4% chance of dying while those in both quadrants on the right side of the graph have a 4.5% risk.
A healthcare system managing a population of patients may use such a categorization technique, i.e., plotting different personal score domains from the same health status survey against one another, stratifying the data on the basis of spatial location (e.g., the lower left quadrant), and calculating the odds of an outcome on the basis of this stratification. In Fig. 7, the technique has produced an astonishing predictor indicating a 20.4% chance of death. This information would certainly be of value to a person who stands, for example, a 1% chance of dying while undergoing an elective cardiac procedure. Prospectively, a person showing an SAQ score in the lower left quadrant would benefit most from a cardiac consultation, a healthcare provider or a health maintenance organization might choose to use these statistics to manage a healthcare plan with scarce resources in such a way that persons who will benefit the most from services with limited availability receive the services on a preferential basis.
Similarly, Fig. 8 shows a similar model for patients with congestive heart failure using cross-plotted domains from the Kansas City Cardiomyopathy Questionnaire. An overall summary score (abscissa) is plotted against the self- efficacy score (ordinate). Differing rates of hospital admission and death are statistically correlated to spatial domains, as in Fig. 8. Admission rates differ from 64% for persons who plot on the left hand half to 31% and 13% for persons in the respective right hand side quartiles.
Fig. 9 is a model formed as plot of SAQ angina frequency versus SAQ quality of life scores. The shaded region 900 identifies patients who have angina occurring approximately once-a-week of greater and who are moderately to extremely limited in their quality of life because of the angina. On the basis of subjectively defined rules from the expert-defined rules base 116, these patients may warrant careful evaluation for opportunities to improve their symptom control.
By way of example, Fig. 10 shows a medical evaluation system 1000 that is configured for predicting or describing patient health outcome in a person with sub- marginal cardiac function. Medical evaluation system 1000 may include processor 1002, graphical user interface 1004, and a data storage unit 1006. The graphical user interface 1004 may be configured for presenting questionnaires to the person. Graphical user interface 1004 may receive the responses from the person for subsequent processing by processor 1002 and storage on the data storage unit 1006, e.g., in a logically defined personal database storage component 1008. Graphical user interface 1004, by way of example, could be administered over the internet, locally through a touch screen, via computer keyboard, by voice recognition technology, or through a mouse-driven interface. In medical system 1000, processor 1002 is communicatively connected to interface 1004 for scoring the responses and evaluating the responses. Evaluation may be performed using the expert-defined rules base 116 and information from a group study model 1010 that is produced using the methodologies shown and described in Examples 1-7. Processor 1002 reports the evaluation results to a system user through interface 1004, or by another means such as a printer or electronic messaging.
A feature of the test study group model 1010 is that it may contain information from a statistically validated survey, i.e., one having statistical correlation indicators indicating that measured predictive parameters are closely related to a measured outcome. For example, validation may be proven through the use of delimiting metrics, such as p values less than a delimiting value of 0.1 (see, e.g., Tables 1-3). The group study model 1008 may be accessed to provide a prognostic ranking or other measure of health outcomes. For example, processor 1002 could use models represented by any of Figs. 2-9 or Tables 1-3 to identify SAQ scores in domains where the scores indicate sub-marginal cardiac function. New responses in the personal data 1010 may be scored and input for comparison against the group study model and the expert-defined rules base 116 to assess the respondent's odds of encountering an outcome.
A user of medical system 1000 may be able to report from the system by interactively adjusting pre-defined delimiters or parameters. For example, the test study group data 1010 might be accessed to calculate a Kaplan-Meir curve like Fig. 1, except stratification might include an additional 30% SAQ range and new responses could be evaluated and filtered on the basis of this new range. Alternatively, the shaded region 900 in Fig. 9 might be manipulated as a graphical user interface through the use of drag and drop functions to produce a new spatial range that triggers calculation and reporting of outcome statistics that are attributed to that spatial range. Processor 1002 may also rank health status scores to predict patient health outcome of an individual patient. For example, where the patient has sub-marginal cardiac function, the questionnaire may be the SAQ, the Kansas City Cardiomyopathy Questionnaire (KCCQ), or any other questionnaire that includes a disease-specific health status measure for patients with coronary artery disease or congestive heart failure. Alternatively, the health status measure may be any type of health status measure, such as patient-assessed satisfaction with quality of life.
The test study group data 1010 may also include, for example, a reporting network of other administrative data, such as the date of last visit with a cardiovascular specialist or a list of specific cardiovascular medications that the patient may or may not be taking. A patient's current treatment may be automatically reported to the personal database 1008 by other databases (not shown) that are linked
to storage unit 1006. For example, a pharmaceutical database describing what therapies have been prescribed may allow relationships between categories of scores so that opportunities for further treatment and treatment efficacy may be linked to health status information. Other diagnostic databases, such as those for angiography or echocardiography, can be used to suggest certain treatments. For example, outcomes from these diagnostic database may be linked with health status information to suggest an increase in dosage for a certain pharmaceutical agent; the use of a device, such as a stent; or a procedure including angiography or bypass surgery. Further statistical linkage of health status data to actual health outcomes associated with the recommended treatments may provide statistical optimization that improves health outcomes for a group of patients that is studied.
FIG. 11 shows a flow chart illustrating operation 1100 of system 1000, in accord with one method of the invention. Operation 1000 commences with mode selection 1102, which may entail user selection of a health status mode for a particular disease or medical condition from among a plurality of such diseases or conditions, e.g., ACS, diabetes, or cancer. Step 1104 entails processor 1102 retrieving a model group that may comprise a selected health status questionnaire, program instruction for implementing the rules base 116, and the group model 1010 for the selected health status mode. The graphical user interface 1004 receives responses from a person in an interview step 1106, which may also entail retrieval of information, such as demographic and clinical information, from other databases (not shown). Processor 1002 may receive transmitted personal responses from graphical user interface 1004, store the responses in the personal database 1008, and process the responses. Processing may include scoring 1112 to obtain scores, followed by evaluation modeling, categorization 122, and recommendation, as described above in the context of Fig. 1.
The foregoing discussion emphasizes methodology for using patient self- assessment data to monitor cardiac health, with overtones on how this data might be used in treatment programs, e.g., by changing medication in response to a changing self-assessment score. In additional contexts, the self assessment data, perhaps additionally including patient satisfaction surveys, may be statistically processed to
relate demographics and/or and patient satisfaction with having undergone a particular procedure or treatment. By way of example, Fig. 12A is a graph of angina frequency post-PCI (Percutaneous Coronary Intervention) (diamond is economically burdened patient; square is not economically burdened patient). Patients characterized as not economically burdened demonstrate a higher frequency of angina after PCI revascularization procedure. Fig. 12B is a graph of angina frequency in patients post -CABG revascularization. Patients characterized as economically burdened had about the same frequency of angina post CABG as did the patient population characterized as economically burdened. These results show that clinical patient outcomes may vary by the type of procedure and demographic background of the patient. FIG. 13 is a graph demonstrating the frequency of angina and statistical likelihood of complications in a post-coronary attack patient having had bypass surgery, as compared to patients having had angioplasty. The complications monitored in these patients were death (0.9%), stroke (1.5%), readmit (2%) and PTCA (less than 1%). FIG. 2 also shows a graph demonstrating the frequency of angina in a post-coronary event patient that had an angioplasty. The same complications were monitored in these patients, with a reported frequency of 0.1% death, 0.02% stroke, 30% readmission to the hospital, and 20% Re PTCA
FIG. 14 shows computer system 1400 configured adapted for use in a medical or clinical health care facility to identify an appropriate post-cardiac event regimen according to a desired patient outcome. An individual patient considering options for pre or post-coronary event treatment, in accord with one embodiment, may review statistical information as show in Figs. 12 and 13. The patient may complete a survey that provides computer system 1400 with the patient's demographic data, lifestyle, and goals. Computer system 1400 processes this information, categorizes the patient, and provides a comparison of outcomes from similarly situated persons who have previously undergone procedures of the types being contemplated. The report is useful in assisting the patient and medical workers in selecting a clinical procedure that is best for the patient according to the desired outcome, adjusted for demographics.
Computer system 1400 may include processor 1402, computer memory 1404, and storage unit 1406. In computer system 1400, processor 1402 is communicatively
connected to computer memory 1404 and to storage unit 1406 for operating in accord a mode of selecting a clinical procedure with assistance of statistics pertaining to other outcomes that may be used as a predictive tool. In one embodiment, computer system 1400 is configured for identifying an individual patient's disease state and demographics of the individual patient comprising age, sex, economic burden, Uving situation, social support, employment status, type of employment, and education level, disease severity, symptoms and clinical setting.
Computer system 1400 may, for example, assess a set of health status parameters from the individual patient to provide a first data assessment profile and identify the projected health outcome desired by the individual patient based upon said individual patient's preferences and goals. Computer system 1400 may also assess a set of health status parameters from a population of patients having similar demographics to said individual patient, the population of patients having received different treatments, in providing a library of specific projected health outcomes for each different treatment.
Upon assessing the health status parameters from the population of patients, Computer system 1400 may select preferred outcomes from the library of specific projected health outcomes that similarly coincide with preferences and goals of the individual patient, present the preferred outcomes to the patient, and select the a clinical treatment that has a projected health outcome that is most similar to the individual patient's desired projected health outcome for the patient based on the preferred outcomes. In one embodiment, software 1403 is configured for operatively controlling computer system 1400 and may initially reside in storage unit 1406. Upon initializing computer system 1400, software 1403 may be loaded in computer memory 1404. Processor 1402 may then software run 1403.
FIG. 15 shows medical system 1500 configured for configured identifying an appropriate post-cardiac event regimen for an individual patient considering options for post-coronary event treatment, in accord with one embodiment of the invention. Medical system 1500 may include processor 1504, storage unit 1506, and interface 1508. In medical system 1500, storage unit 1506 is configured for configured for storing group data in a database. The group data may comprise responses to a questionnaire having a plurality of questions regarding quality of life and
demographic information. The response may be derived from a plurality of patients having survived a coronary event. A first group of the patients have received a post- coronary event revascularization procedure. A second group of the patients had not received a the post-coronary event revascularization procedure. Demographics of the first and second groups of patients may be similar to an those of the individual patient. In medical system 1500, interface 1508 is configured for receiving responses to a the questions from an the individual patient. In medical system 1500, processor 1504 is communicatively connected to interface 1508 and to storage unit 1506 for performing statistical analysis on the responses from the plurality of patients and from the individual patient. A comparison of the statistical analysis of the responses from the group of patients and from the individual patient may provide a basis upon which the individual patient may select a post-cardiac event treatment appropriate to the individual patient preferences and goals of the individual patient.
With further regard to FIGS. 13 and 15, those skilled in the art should appreciate that storage unit 1506 and storage unit 1406 may illustratively represent the same storage memory and/or one or a combination of storage unit 1406 and computer memory 1404 within computer system 1400. Processor 1402 may incorporate functionality including processor 1504, for example.
FIG. 16 shows a flow chart illustrating operation 1600 of medical system 1500 (shown in FIG. 15), in accord with one methodology. Operation 1600 commences in step 1602. Processor 1504 identifies an individual patient's disease state and demographics of the individual patient comprising age, sex, economic burden, living situation, social support, employment status, type of employment, and education level, disease severity, symptoms and clinical setting, in step 1604. Processor 1504 assesses a set of health status parameters of the individual patient, in step 1606. Processor 1504 identifies the projected health outcome desired by the individual patient based upon said individual patient's preferences and goals, in step 1608. Processor 1504 assesses a set of health status parameters from a population of patients having similar demographics to the individual patient to provide a library of specific projected health outcomes for each different treatment, in step 1610. Processor 1504 selects preferred outcomes from the library of specific projected health outcomes that similarly coincide with preferences and goals of the individual patient, in step 1612.
Interface 1508 presents the preferred outcomes to the patient, in step 1614. Processor 1504 selects a clinical treatment having a projected health outcome that is most similar to the individual patient's desired projected health outcome for the patient based on the preferred outcomes, in step 1616. Operation 1600 ends in step 1618. Instructions that perform the operation discussed in FIG. 16 may be stored in storage media or computer memory. The instructions may be retrieved and executed by processor 1504. Some examples of instructions include software, program code, and firmware. Some examples of storage media include memory devices, tapes, disks, integrated circuits, and servers. The instructions are operational when executed by processor 1504 to direct processor 1504 to operate in accord with the invention. Those skilled in the art are familiar with instructions and storage media.
FIG. 17 shows a flow chart illustrating operation 1100 of medical system 1500 (shown in FIG. 15), in accord with one methodology. Operation 1700 commences in step 1702. Processor 1504 identifies the a disease state and demographics of the individual patient, in step 1704. The demographics may include age, sex, economic burden, living situation, social support, employment status, type of employment, and education level of the individual patient comprising age, sex, economic burden, living situation, social support, employment status, type of employment, and education level, disease severity, symptoms and clinical setting. Processor 1504 assesses a set of set of health status parameters from the patient to provide a first data assessment profile, in step 1705. Processor 1504 assesses a set of set of health status parameters from a first population of patients to provide a first reference data assessment profile, in step 1710. Processor 1504 assesses a set of set of health status parameters from a second population of patients to provide a second reference data assessment profile, in step 1712. Processor 1504 assesses a set of set of health status parameters from a third population of patients to provide a third reference data assessment profile, in step 1708. The first, second, and third populations may have similar demographics as the individual patient and differing treatments and/or revascularization procedures. Processor 1504 projects the patient's survival and quality of life probability of the individual patient from any number of clinical options. For example, first, second, and third reference data assessment profiles grouped by type of revascularization procedure may be statistically profiled by a processing algorithm to respectively
provide a first, second, and third projected post-procedural outcomes for the patient prospectively based upon statistics from other patients who have undergone the respective vascularization procedures and/or treatments, in steps 1714, 1716, and 1718. The revascularization procedures may include a coronary artery bypass grafting (CABG) procedure and assessing a set of health status parameters from a second population having had a Percutaneous Coronary Intervention (PCI) procedure. The treatment may include, for example, an anti-coronary disease medication, diet modification, herbal remedy, or and other non-surgical intervention procedure. Processor 1504 the compares the first projected post-procedural outcome projected post-first outcome to the post-second projected post-procedural outcome, in step 1720 procedure outcome. Processor 1504 selects the an appropriate revascularization procedure for the individual patient by in response to the step of comparing the projected post-first to said projected post-second outcome, in step 1722. Operation 1700 ends in step 1724. FIG. 18 shows a flow chart illustrating step 1705 of operation 1700, in accord with one methodology. Step 1705commences through entry point 1801, designating a specific processing algorithm, for example, one operating off of expert rules or a self- training algorithm such as a neural network that associates outcomes with clinical procedures. Processor 1504 assess the individual patient's data, in step 1802. Step 1705 exits through exit point 1803.
FIG. 19 shows a flow chart illustrating step 1710 of operation 1700, in accord with one methodology. Step 1710 commences through entry point 1901. Processor 1504 may collect the first data assessment profile of the first group of patients, in step 1902. Processor 1504 may select the CABG revascularization procedure, in step 1905. Step 1710 exits through exit point 1903.
FIG. 20 shows a flow chart illustrating step 1712 of operation 1700, in accord with one methodology. Step 1712 commences through entry point 2001. Processor 1504 may collect the second data assessment profile of the second group of patients, in step 2002. Processor 1504 may select the PCI revascularization procedure, in step 2004. Step 1712 exits through exit point 2003.
FIG. 21 shows a flow chart illustrating step 1720 of operation 1700, in accord with one methodology. Step 1720 commences through entry point 2101. Processor
1504 may perform a statistical analysis on the individual patient quality of life profile, in step 2102. The from the subset of the group data statistical analysis may indicate to the individual patient an appropriate post-cardiac clinical regimen. Processor 1504 may compute a statistical average that indicates the appropriate post-cardiac clinical regimen based on the statistical analysis, in step 2104. Processor 1504 may compute a statistical deviation about the average, in step 2106. Step 1720 exits through exit point 2103.
It will be appreciated that patient outcomes following clinical procedures selected with the assistance of system 1200 may be monitored by the medical evaluation system 300, for example, using the Seattle Angina Questionnaire (SAQ) and the Kansas City Cardiomyopathy Questionnaire (KCCQ). The patient self assessment data from these questionnaires may be used as feedback into system 1200 to facilitate better selections of clinical procedures, e.g., to assess the statistical likelihood of accomplishing the patient's desired goals from the clinical procedure. By way of example, one procedure versus another may have a different level of mortality risk that might be incurred for relatively small benefit. The patient would be able to assess whether the benefit of the riskier procedure is worth taking that risk. The patient might also be told, for example, how what percentage of patients in their position of selecting a particular clinical procedure, with analogous goals, were ultimately satisfied with the selected procedure in terms of accomplishing the intended goals, or what percentage in hindsight would select a different procedure.
The foregoing discussion is intended to illustrate the concepts of the invention by way of example with emphasis upon the preferred embodiments and instrumentalities. Accordingly, the disclosed embodiments and instrumentalities are not exhaustive of all options or mannerisms for practicing the disclosed principles of the invention. The inventor hereby states bis intention to rely upon the Doctrine of Equivalents in protecting the full scope and spirit of the invention