INR targets and site-level anticoagulation control: results from the Veterans AffaiRs Study to Improve Anticoagulation (VARIA)

BACKGROUND

Not all clinicians target the same International Normalized Ratio (INR) for patients with a guideline-recommended target range of 2-3. A patient's mean INR value suggests the INR that was actually targeted. We hypothesized that sites would vary by mean INR, and that sites of care with mean values nearest to 2.5 would achieve better anticoagulation control, as measured by per cent time in therapeutic range (TTR).

OBJECTIVES

To examine variations among sites in mean INR and the relationship with anticoagulation control in an integrated system of care.

PATIENTS/METHODS

We studied 103,897 patients receiving oral anticoagulation with an expected INR target between 2 and 3 at 100 Veterans Health Administration (VA) sites from 1 October 2006 to 30 September 2008. Key site-level variables were: proportion near 2.5 (that is, percentage of patients with mean INR between 2.3 and 2.7) and mean risk-adjusted TTR.

RESULTS

Site mean INR ranged from 2.22 to 2.89; proportion near 2.5, from 30 to 64%. Sites' proportions of patients near 2.5, below 2.3 and above 2.7 were consistent from year to year. A 10 percentage point increase in the proportion near 2.5 predicted a 3.8 percentage point increase in risk-adjusted TTR (P < 0.001).

CONCLUSIONS

Proportion of patients with mean INR near 2.5 is a site-level 'signature' of care and an implicit measure of targeted INR. This proportion varies by site and is strongly associated with site-level TTR. Our study suggests that sites wishing to improve TTR, and thereby improve patient outcomes, should avoid the explicit or implicit pursuit of non-standard INR targets.

Patient characteristics associated with oral anticoagulation control: results of the Veterans AffaiRs Study to Improve Anticoagulation (VARIA)

BACKGROUND

In patients receiving oral anticoagulation, improved control can reduce adverse outcomes such as stroke and major hemorrhage. However, little is known about patient-level predictors of anticoagulation control.

OBJECTIVES

To identify patient-level predictors of oral anticoagulation control in the outpatient setting.

PATIENTS/METHODS

We studied 124,619 patients who received oral anticoagulation from the Veterans Health Administration from October 2006 to September 2008. The outcome was anticoagulation control, summarized using percentage of time in therapeutic International Normalized Ratio range (TTR). Data were divided into inception (first 6 months of therapy; 39,447 patients) and experienced (any time thereafter; 104,505 patients). Patient-level predictors of TTR were examined by multivariable regression.

RESULTS

Mean TTRs were 48% for inception management and 61% for experienced management. During inception, important predictors of TTR included hospitalizations (the expected TTR was 7.3% lower for those with two or more hospitalizations than for the non-hospitalized), receipt of more medications (16 or more medications predicted a 4.3% lower than for patients with 0-7 medications), alcohol abuse (-4.6%), cancer (-3.1%), and bipolar disorder (-2.9%). During the experienced period, important predictors of TTR included hospitalizations (four or more hospitalizations predicted 9.4% lower TTR), more medications (16 or more medications predicted 5.1% lower TTR), alcohol abuse (-5.4%), female sex (- 2.9%), cancer (-2.7%), dementia (-2.6%), non-alcohol substance abuse (-2.4%), and chronic liver disease (-2.3%).

CONCLUSIONS

Some patients receiving oral anticoagulation therapy are more challenging to maintain within the therapeutic range than others. Our findings can be used to identify patients who require closer attention or innovative management strategies to maximize benefit and minimize harm from oral anticoagulation therapy.

The epidemiology and management of severe hypertension

Hypertension guidelines stress that patients with severe hypertension (systolic blood pressure (BP)⩾180 or diastolic BP⩾110 mm Hg) require multiple drugs to achieve control and should have close follow-up to prevent adverse outcomes. However, little is known about the epidemiology or actual management of these patients. We retrospectively studied 59 207 veterans with hypertension. Patients were categorized based on their highest average BP over an 18-month period (1 July 1999 to 31 December 2000) as controlled (<140/90 mm Hg), mild (140–159/90–99 mm Hg), moderate (160–179/100–109 mm Hg) and severe hypertension. We examined severe hypertension prevalence, pattern, duration, associated patient characteristics, time to subsequent visit, percentage of visits with a medication increase, and final BP control and antihypertensive medication adequacy. Twenty-three per cent had ⩾1 visit with severe hypertension, 42% of whom had at least two such visits; median day with severe hypertension was 80 (range 1–548). These subjects were significantly older, more likely black, and with more comorbidities than other hypertension subjects. Medication increases occurred at 20% of visits with mild hypertension compared to 40% with severe hypertension; P<0.05). At study end, 76% of patients with severe hypertension remained uncontrolled; severe hypertension subjects with uncontrolled BP were less likely to be on adequate therapy than those with controlled BP (43.7 vs 45.4%). Among hypertensive veterans, severe hypertension episodes are common. Many subjects had relatively prolonged elevations, with older, sicker subjects at highest risk. Although, follow-up times are shorter and antihypertensive medication use greater in severe hypertension subjects, they are still not being managed aggressively enough. Interventions to improve providers' management of these high-risk patients are needed.

Warfarin dose management affects INR control

BACKGROUND

Little is known about how patterns of warfarin dose management contribute to percentage time in the therapeutic International Normalized Ratio (INR) range (TTR).

OBJECTIVES

To quantify the contribution of warfarin dose management to TTR and to define an optimal dose management strategy.

PATIENTS/METHODS

We enrolled 3961 patients receiving warfarin from 94 community-based clinics. We derived and validated a model for the probability of a warfarin dose change under various conditions. For each patient, we computed an observed minus expected (O - E) score, comparing the number of dose changes predicted by our model to the number of changes observed. We examined the ability of O - E scores to predict TTR, and simulated various dose management strategies in the context of our model.

RESULTS

Patients were observed for a mean of 15.2 months. Patients who deviated the least from the predicted number of dose changes achieved the best INR control (mean TTR 70.1% unadjusted); patients with greater deviations had lower TTR (65.8% and 62.0% for fewer and more dose changes respectively, Bonferroni-adjusted P < 0.05/3 for both comparisons). On average, clinicians in our study changed the dose when the INR was 1.8 or lower/3.2 or higher (mean TTR: 68%); optimal management would have been to change the dose when the INR was 1.7 or lower/3.3 or higher (predicted TTR: 74%).

CONCLUSIONS

Our observational study suggests that INR control could be improved considerably by changing the warfarin dose only when the INR is 1.7 or lower/3.3 or higher. This should be confirmed in a randomized trial.

Trends in antiepileptic drug prescribing for older patients with new-onset epilepsy: 2000-2004

BACKGROUND

Newer antiepileptic drugs (AEDs) have been shown to be equally efficacious as older seizure medications but with fewer neurotoxic and systemic side effects in the elderly. A growing body of clinical recommendations based on systematic literature review and expert opinion advocate the use of the newer agents and avoidance of phenobarbital and phenytoin. This study sought to determine if changes in practice occurred between 2000 and 2004--a time during which evidence and recommendations became increasingly available.

METHODS

National data from the Veterans Health Administration (VA; inpatient, outpatient, pharmacy) from 1998 to 2004 and Medicare data (1999-2004) were used to identify patients 66 years and older with new-onset epilepsy. Initial AED was the first AED received from the VA. AEDs were categorized into four groups: phenobarbital, phenytoin, standard (carbamazepine, valproate), and new (gabapentin, lamotrigine, levetiracetam, oxcarbazepine, topiramate).

RESULTS

We found a small reduction in use of phenytoin (70.6% to 66.1%) and phenobarbital (3.2% to 1.9%). Use of new AEDs increased significantly from 12.9% to 19.8%, due primarily to use of lamotrigine, levetiracetam, and topiramate.

CONCLUSIONS

Despite a growing list of clinical recommendations and guidelines, phenytoin was the most commonly used antiepileptic drug, and there was little change in its use for elderly patients over 5 years. Research further exploring physician and health care system factors associated with change (or lack thereof) will provide better insight into the impact of clinical recommendations on practice.

What constitutes high quality of care for adults with epilepsy?

BACKGROUND

Providers are increasingly being held accountable for the quality of care provided. While quality indicators have been used to benchmark the quality of care for a number of other disease states, no such measures are available for evaluating the quality of care provided to adults with epilepsy. In order to assess and improve quality of care, it is critical to develop valid quality indicators. Our objective is to describe the development of quality indicators for evaluating care of adults with epilepsy. As most care is provided in primary and general neurology care, we focused our assessment of quality on care within primary care and general neurology clinics.

METHODS

We reviewed existing national clinical guidelines and systematic reviews of the literature to develop an initial list of quality indicators; supplemented the list with indicators derived from patient focus groups; and convened a 10-member expert panel to rate the appropriateness, reliability, and necessity of each quality indicator.

RESULTS

From the original 37 evidence-based and 10 patient-based quality indicators, the panel identified 24 evidence-based and 5 patient-based indicators as appropriate indicators of quality. Of these, the panel identified 9 that were not necessary for high quality care.

CONCLUSION

There is, at best, a poor understanding of the quality of care provided for adults with epilepsy. These indicators, developed based on published evidence, expert opinion, and patient perceptions, provide a basis to assess and improve the quality of care for this population.

Clinical practice guidelines in the nursing home

Clinical practice guidelines are an important tool for improving quality of care. This study determined whether and how guidelines are being used in nursing homes. We surveyed staff at 36 Department of Veterans Affairs (VA) nursing homes. Employees were asked whether they were familiar with guidelines as well as whether 5 specific guidelines had been read, were available, and had been adopted. Among 1065 respondents (60% of those surveyed), 79% reported familiarity with guidelines. The proportion of staff at a facility reporting adoption was generally less than 50%. Those nursing homes in which a high percentage of the staff reported adoption of one guideline were more likely to have adopted other guidelines. However, staff were not more likely to report adoption of a specific guideline when the nurse manager stated that it was adopted. We conclude that staff at VA nursing homes are familiar with guidelines. Guideline adoption at individual nursing homes, however, is not a systematic process involving the entire staff.

Are pressure ulcers preventable? A survey of experts

OBJECTIVE

There is considerable debate regarding whether pressure ulcers can truly be prevented in nursing homes. New pressure ulcers are often taken as a sign of negligence that can lead to a lawsuit. This study sought to determine expert opinion regarding the preventability of pressure ulcers, the resources available to nursing homes for prevention, and the role of negligence lawsuits in pressure ulcer care.

DESIGN

Survey mailed to a convenience sample of 98 experts in the field of pressure ulcer care. The survey contained 36 questions, most based on a 5-point Likert scale from "strongly agree"to"strongly disagree." Several questions asked respondents to rank items.

RESULTS

Sixty-five of 92 surveys were completed (6 were returned but not completed) for a response rate of 71%. Sixty-two percent of respondents disagreed with the statement that all pressure ulcers are preventable. Only 5% said that nursing homes have adequate resources to prevent all pressure ulcers. Although most respondents disagreed that pressure ulcers are necessarily a sign of neglect and that nursing homes should be sued when a resident develops a pressure ulcer, 38% agreed with the concept that lawsuits are an appropriate way to stimulate improvement in nursing home care.

CONCLUSION

The results of this survey demonstrated divergent expert opinion on whether pressure ulcers are preventable. The role of regulations and litigation in pressure ulcer prevention needs to be further defined.

Deriving a risk-adjustment model for pressure ulcer development using the Minimum Data Set

OBJECTIVE

To use the Minimum Data Set (MDS) to derive a risk-adjustment model for pressure ulcer development that may be used in assessing the quality of nursing home care.

DESIGN

Perspective observational study using MDS data from 1997.

SETTING

A large, for-profit, nursing home chain.

PARTICIPANTS

Our unit of analysis was 39,649 observations made on 14,607 nursing home residents who were without a stage 2 or larger pressure ulcer on an index assessment.

MEASUREMENTS

Pressure ulcer status was determined at an outcome assessment approximately 90 days after an index assessment. Potential predictors of pressure ulcer development were examined for bivariate associations, contributing to the development of a multivariate logistic regression model.

RESULTS

A stage 2 or larger pressure ulcer developed in 2.3% of the observations. Seventeen resident characteristics were found to be associated with pressure ulcer development. These included dependence in mobility and transferring, diabetes mellitus, peripheral vascular disease, urinary incontinence, lower body mass index, and end-stage disease. A risk-adjustment model based on these characteristics was well calibrated and able to discriminate among residents with different levels of risk for ulcer development (model c-statistic = 0.73).

CONCLUSION

A clinically credible risk-adjustment model with good performance properties can be developed using the MDS. This model may be useful in profiling nursing homes on their rate of pressure ulcer development.

Evaluation of a risk-adjustment model for pressure ulcer development using the Minimum Data Set

OBJECTIVE

To validate a previously derived risk-adjustment model for pressure ulcer development in a separate sample of nursing home residents and to determine the extent to which use of this model affects judgments of nursing home performance.

DESIGN

Retrospective observational study using Minimum Data Set (MDS) data from 1998.

SETTING

A large, for-profit, nursing home chain.

PARTICIPANTS

Twenty-nine thousand and forty observations were made on 13,457 nursing home residents who were without a pressure ulcer on an index assessment.

MEASUREMENTS

We used logistic regression in our validation sample to calculate new coefficients for the 17 previously identified predictors of pressure ulcer development. Coefficients from this new sample were compared with those previously derived. Expected rates of pressure ulcer development were determined for 108 nursing homes. Unadjusted and risk-adjusted rates of pressure ulcer development from these homes were also calculated and outlier identification using these two approaches was compared.

RESULTS

Predictors of pressure ulcer development in the derivation sample generally showed similar effects in the validation sample. The model c-statistic was also unchanged at 0.73, but it was not calibrated as well in the validation sample. On applying the model to the nursing homes, expected rates of ulcer development ranged from 1.1% to 3.2% (P <.001). The observed rates ranged from 0% to 12.1% (P <.001). There were 12 outliers using unadjusted rates and 15 using adjusted performance. Ten nursing homes were identified as outliers using both approaches.

CONCLUSIONS

Our MDS risk-adjustment model for pressure ulcer development performed well in this new sample. Nursing homes differ significantly in their expected rates of pressure ulcer development. Outlier identification also differs depending on whether unadjusted or risk-adjusted performance is evaluated.

Evaluating diagnosis-based case-mix measures: how well do they apply to the VA population?

BACKGROUND

Diagnosis-based case-mix measures are increasingly used for provider profiling, resource allocation, and capitation rate setting. Measures developed in one setting may not adequately capture the disease burden in other settings.

OBJECTIVES

To examine the feasibility of adapting two such measures, Adjusted Clinical Groups (ACGs) and Diagnostic Cost Groups (DCGs), to the Department of Veterans Affairs (VA) population.

RESEARCH DESIGN

A 60% random sample of veterans who used health care services during FY 1997 was obtained from VA inpatient and outpatient administrative databases. A split-sample technique was used to obtain a 40% sample (n = 1,046,803) for development and a 20% sample (n = 524,461) for validation.

METHODS

Concurrent ACG and DCG risk adjustment models, using 1997 diagnoses and demographics to predict FY 1997 utilization (ambulatory provider encounters, and service days-the sum of a patient's inpatient and outpatient visit days), were fitted and cross-validated.

RESULTS

Patients were classified into groupings that indicated a population with multiple psychiatric and medical diseases. Model R-squares explained between 6% and 32% of the variation in service utilization. Although reparameterized models did better in predicting utilization than models with external weights, none of the models was adequate in characterizing the entire population. For predicting service days, DCGs were superior to ACGs in most categories, whereas ACGs did better at discriminating among veterans who had the lowest utilization.

CONCLUSIONS

Although "off-the-shelf" case-mix measures perform moderately well when applied to another setting, modifications may be required to accurately characterize a population's disease burden with respect to the resource needs of all patients.

Health care reorganization and quality of care: unintended effects on pressure ulcer prevention

BACKGROUND

Health care reorganizations, with a change in focus from inpatient to outpatient care, are becoming increasingly frequent. Little is known regarding how reorganizations may affect risk-adjusted outcomes for those programs, usually inpatient, that lose resources as a result of the change in organizational focus.

OBJECTIVES

To determine changes in risk-adjusted rates of pressure ulcer development over an 8-year period, the final 3 of which were characterized by a significant reorganization of the health care system.

DESIGN

This was an observational study that used an existing database.

SUBJECTS

Subjects were residents of Department of Veterans Affairs long-term care units between 1990 and 1997 who were without a pressure ulcer at an index assessment.

MEASURES

The study examined risk-adjusted rates of pressure ulcer development, and proportions of new ulcers that were severe (stages 3 or 4) were calculated for successive 6-month periods.

RESULTS

Between 1990 and 1994, risk-adjusted rates of pressure ulcer development declined significantly, by 27%. However, beginning in 1995, rates began to increase, and in 1997 they were similar to those in 1990. The proportion of new ulcers that were severe increased significantly over time (P = 0.01).

CONCLUSIONS

The reorganization of the VA that began in 1995, with its emphasis on outpatient care, was associated with an increase in rates of pressure ulcer development. This highlights the need to carefully monitor the quality of care in programs that may be losing resources as a result of the reorganization.

Are we improving the quality of nursing home care: the case of pressure ulcers

BACKGROUND

There are widespread concerns regarding the quality of nursing home care and whether care is improving. We evaluated a large provider of nursing home care to determine whether risk-adjusted rates of pressure ulcer development have changed.

METHODS

We used the Minimum Data Set to study National HealthCare Corporation nursing homes from 1991 through 1995. Rates of pressure ulcer development were calculated for successive 6-month periods by determining the proportion of residents initially ulcer-free having a stage 2 or larger pressure ulcer on subsequent assessments. Rates were risk-adjusted for patient characteristics. The proportion of new ulcers that were deep (stages 3 or 4) were also calculated.

RESULTS

We examined risk-adjusted rates of pressure ulcer development based on 144,379 observations of 30,510 residents at 107 nursing homes. The number of observations per 6-month period ranged from 11,041 to 15,805. Between 1991 and 1995, there was a significant (P<.05) rate decline of more than 25%. Additionally, the proportion of new ulcers that were stages 3 or 4 declined from 30 to 22% (P<.01).

CONCLUSIONS

Nursing homes showed significant improvement in the quality of pressure ulcer preventive care from 1991 to 1995.

Geriatric syndromes as outcome measures of hospital care: can administrative data be used?

OBJECTIVE

To determine how often hospital administrative databases capture the occurrence of two common geriatric syndromes, pressure ulcers and incontinence.

DESIGN

Retrospective comparison of a nursing home and hospital database.

SETTING

Department of Veterans Affairs (VA) hospitals.

PARTICIPANTS

All patients between 1992 and 1996 discharged from VA acute medical care and admitted to a VA nursing home.

MEASUREMENTS

The presence of incontinence or a pressure ulcer (stage 2 or larger) on admission to the nursing home was determined. Hospital discharge diagnoses were then reviewed to determine whether these conditions were recorded. The effect of ulcer stage, total number of discharge diagnoses, and temporal trends on the recording of these conditions in discharge diagnoses was also noted.

RESULTS

There were 17,004 admissions to nursing homes from acute care in 1996; 12.7% had a pressure ulcer and 43.4% were incontinent. Among these patients with a pressure ulcer, the hospital discharge diagnosis listed an ulcer in 30.8% of cases, and incontinence was included correctly as a discharge diagnosis in 3.4%. While deeper pressure ulcers were more likely to be recorded than superficial ulcers (P < .01), nearly 50% of stage 4 ulcers were not listed among hospital discharge diagnoses. Patients with more discharge diagnoses were more likely to record both conditions correctly. From 1992 to 1996, small but significant (P = .001) improvements were noted in the correct recording of these geriatric syndromes as discharge diagnoses.

CONCLUSIONS

The occurrence of pressure ulcers and incontinence cannot be determined from hospital administrative databases and should not be used as outcomes when measuring quality of care among hospitalized patients.

Functional status outcomes for assessment of quality in long-term care

OBJECTIVE

Although decline in functional status has been recommended as a quality indicator in long-term care, studies examining its use provide no consensus on which definition of functional status outcome is the most appropriate to use for quality assessment. We examined whether different definitions of decline in functional status affect judgments of quality of care provided in Department of Veterans Affairs (VA) long-term care facilities.

METHODS

Six measures of functional status outcome that are prominent in the literature were considered. The sample consisted of 15 409 individuals who resided in VA long-term care facilities at any time from 4/1/95 to 10/1/95. Activities of daily living variables were used to generate measures of functional status. Differences between residents' baseline and semi-annual assessments were considered and facility performance using the various definitions of functional status were described.

RESULTS

The percentage of residents seen as declining in functional status ranged from 7.7% to 31.5%, depending upon the definition applied. The definition of functional status also affected rankings, z-scores, and 'outlier' status for facilities.

CONCLUSION

Judgments of facility performance are sensitive to how outcome measures are defined. Careful selection of an appropriate definition of functional status outcome is needed when assessing quality in long-term care.

Pressure ulcer development in the VA: characteristics of nursing homes providing best care

This study identifies structural characteristics of VA nursing homes that are associated with the best patient outcomes. We evaluated risk-adjusted rates of pressure ulcer development in VA nursing homes and related these rates to facility size, staffing patterns, teaching nursing home status, and rural versus urban locale. Higher rates of pressure ulcer development were seen among urban teaching nursing homes and among nursing homes associated with both larger and smaller VA hospitals. Staffing patterns had a complex association with pressure ulcer development, and smaller nursing home staffs were not clearly associated with higher rates. For multivariate modeling, only hospital size and staffing remained significant independent predictors of pressure ulcer development. These results emphasize that while structural characteristics of VA nursing homes can provide insights about care, improving the quality of care in this setting will require a much greater understanding of how nursing homes are organized to meet patient needs.

Inadequate management of blood pressure in a hypertensive population

BACKGROUND

Many patients with hypertension have inadequate control of their blood pressure. Improving the treatment of hypertension requires an understanding of the ways in which physicians manage this condition and a means of assessing the efficacy of this care.

METHODS

We examined the care of 800 hypertensive men at five Department of Veterans Affairs sites in New England over a two-year period. Their mean (+/-SD) age was 65.5+/-9.1 years, and the average duration of hypertension was 12.6+/-5.3 years. We used recursive partitioning to assess the probability that antihypertensive therapy would be increased at a given clinic visit using several variables. We then used these predictions to define the intensity of treatment for each patient during the study period, and we examined the associations between the intensity of treatment and the degree of control of blood pressure.

RESULTS

Approximately 40 percent of the patients had a blood pressure of > or =160/90 mm Hg despite an average of more than six hypertension-related visits per year. Increases in therapy occurred during 6.7 percent of visits. Characteristics associated with an increase in antihypertensive therapy included increased levels of both systolic and diastolic blood pressure at that visit (but not previous visits), a previous change in therapy, the presence of coronary artery disease, and a scheduled visit. Patients who had more intensive therapy had significantly (P<0.01) better control of blood pressure. During the two-year period, systolic blood pressure declined by 6.3 mm Hg among patients with the most intensive treatment, but increased by 4.8 mm Hg among the patients with the least intensive treatment.

CONCLUSIONS

In a selected population of older men, blood pressure was poorly controlled in many. Those who received more intensive medical therapy had better control. Many physicians are not aggressive enough in their approach to hypertension.

Mortality associated with an influenza outbreak on a dementia care unit

The purpose of this study was to describe an episode of increased mortality, whose cause was initially unknown. This retrospective cohort investigation was conducted on a dementia special care unit of a Department of Veterans Affairs facility, with more than 75% of residents clinically diagnosed with dementia of the Alzheimer type. One hundred five residents residing in the facility during February 1995 were included as subjects. A cluster of deaths occurred, triggering the investigation. Ultimately, 21 deaths (three times greater than any previous month in the past 5 years) occurred during the 1-month period. Measures included the presence of clinical influenza-like illness based on signs, serology, and autopsy results. Of the 105 residents, 45 (42.8%) met the clinical definition for influenza-like illness. Eight autopsies were performed, and the causes of death consisting of bronchopneumonia in seven and aspiration pneumonia in one were compatible with influenza. There were no differences among those who died from those who lived with regards to age, race, gender, clinical influenza-like illness, vaccination status, diagnosis of Alzheimer disease, or duration of dementia (all p > or = 0.2). However, those who died were at a higher risk of dying due to a greater number of coexisting conditions (p < 0.01). Also, overall the groups differed in Mini-Mental State Examination and Bedford Alzheimer Nursing Scale scores with those who died being more impaired (p < 0.01). Thus, the presentation of influenza-like illness can be subtle in onset, underappreciated in this population, and not recognized until excess mortality, which affects the most frail, is noted. Care providers need to be vigilant during the winter months for the presence of influenza.

Problems in assessing diabetes control in an ambulatory setting

Problems in using medical records to assess outcomes of diabetes care have not been well defined. We reviewed the medical records of 288 patients with diabetes receiving ambulatory care over a 2-year period. We determined the availability of different tests of glycemic control and described site performance as the percentage of patients with a blood glucose exceeding either 180 or 240 mg/dl. Glycosylated hemoglobin determinations were performed in only 26.7% of patients. A blood glucose was available in 208 patients (72.2%) during a 6-month outcome period. For almost 50% of the sample, the glucose was greater than 180 mg/dl, whereas in 20% it exceeded 240 mg/dl. Judgments of whether sites differed in performance depended on how control was defined. Using a single glucose determination and a threshold of 180 mg/dl, similar fractions of patients were poorly controlled at each site (51.2 versus 45.0 versus 47.0%) (P = 0.75). At 240 mg/dl, although, one site performed much worse than the other two (14.6 versus 16.7 versus 31.8%) (P = 0.02). These results highlight difficulties in defining the outcome measure when using medical records to evaluate quality of care.

Reducing random variation in reported rates of pressure ulcer development

OBJECTIVES

The authors evaluated methods of reporting on rates of pressure ulcer development in long-term care to identify approaches that lead to more stable estimates of actual performance.

METHODS

Performance measures for facilities that adequately adjust for both random variation and casemix should be relatively stable from one time period to the next. The authors calculated facility rates of pressure ulcer development over eight consecutive time periods and correlated measures over time using different reporting methods including z-scores, combining rates from several time periods, and limiting analyses to large facilities. Results were compared with a Monte Carlo simulation.

RESULTS

Observed facility rates of pressure ulcer development varied considerably over time. The average correlation coefficient across seven time comparisons for observed rates was 0.17. Reporting performance as a z-score or limiting the analyses to large facilities increased the correlation. Combining two time periods was effective only when used with one of these other approaches. The correlation coefficient based on a simulation using only large facilities was 0.51.

CONCLUSIONS

Random variation affects reported rates of pressure ulcer development. Using only large facilities and combining two time periods limits the effects of random variation and results in more stable estimates of performance. When describing performance, management must consider tradeoffs between having more accurate data, the frequency with which data are provided, and whether it is given to all providers.

Profiling outcomes of ambulatory care: casemix affects perceived performance

OBJECTIVES

The authors explored the role of casemix adjustment when profiling outcomes of ambulatory care.

METHODS

The authors reviewed the medical records of 656 patients with hypertension, diabetes, or chronic obstructive pulmonary disease (COPD) receiving care at one of three Department of Veterans Affairs medical centers. Outcomes included measures of physiological control for hypertension and diabetes, and of exacerbations for COPD. Predictors of poor outcomes, including physical examination findings, symptoms, and comorbidities, were identified and entered into regression models. Observed minus expected performance was described for each site, both before and after casemix adjustment.

RESULTS

Risk-adjustment models were developed that were clinically plausible and had good performance properties. Differences existed among the three sites in the severity of the patients being cared for. For example, the percentage of patients expected to have poor blood pressure control were 35% at site 1, 37% at site 2, and 44% at site 3 (P < 0.01). Casemix-adjusted measures of performance were different from unadjusted measures. Sites that were outliers (P < 0.05) with one approach had observed performance no different from expected with another approach.

CONCLUSIONS

Casemix adjustment models can be developed for outpatient medical conditions. Sites differ in the severity of patients they treat, and adjusting for these differences can alter judgments of site performance. Casemix adjustment is necessary when profiling outpatient medical conditions.

Evaluating and improving pressure ulcer care: the VA experience with administrative data

BACKGROUND

A number of state initiatives are using databases originally developed for nursing home reimbursements to assess the quality of care. Since 1991 the Department of Veterans Affairs (VA; Washington, DC) has been using a long term care administrative database to calculate facility-specific rates of pressure ulcer development. This information is disseminated to all 140 long term care facilities as part of a quality assessment and improvement program.

DATA ON PRESSURE ULCER DEVELOPMENT

Assessments are performed on all long term care residents on April 1 and October 1, as well as at the time of admission or transfer to a long term care unit. Approximately 18,000 long term care residents are evaluated in each six-month period; the VA rate of pressure ulcer development is approximately 3.5%. Reports of the rates of pressure ulcer development are then disseminated to all facilities, generally within two months of the assessment date.

IMPLICATIONS FOR OTHER QUALITY IMPROVEMENT EFFORTS

The VA's more than five years' experience in using administrative data to assess outcomes for long term care highlights several important issues that should be considered when using outcome measures based on administrative data. These include the importance of carefully selecting the outcome measure, the need to consider the structure of the database, the role of case-mix adjustment, strategies for reporting rates to small facilities, and methods for information dissemination.

CONCLUSION

Attention to these issues will help ensure that results from administrative databases lead to improvements in the quality of care.

Effect of pressure ulcers on the survival of long-term care residents

BACKGROUND

Past studies have emphasized that patients with pressure ulcers are at high risk of dying. However, it remains unclear whether this increased risk is related to the ulcer or to coexisting conditions. In this study we examined the independent effect of pressure ulcers on the survival of long-term care residents.

METHODS

We evaluated all 19,981 long-term care residents institutionalized in Department of Veterans Affairs (VA) long-term care facilities as of April 1, 1993. Baseline resident characteristics and survival status were obtained by merging data from five existing VA data bases. Survival experience over a 6-month period was described using a proportional hazards model.

RESULTS

Pressure ulcers were present in 1,539 (7.7%) long-term care residents. Residents with pressure ulcers had a relative risk of 2.37 (95% CI = 2.13, 2.64) for dying as compared to those without ulcers. After adjusting for 16 other measures of clinical and functional status, the relative risk associated with pressure ulcers decreased to 1.45 (95% CI = 1.30, 1.65). No increased risk of death was noted for residents with deeper ulcers.

CONCLUSIONS

Pressure ulcers are a significant marker for long-term care residents at risk of dying. After adjusting for clinical and functional status, however, the independent risk associated with pressure ulcers declines considerably. The fact that larger ulcers are not associated with greater risk suggests that other unmeasured clinical conditions may also be contributing to the increased mortality associated with pressure ulcers.

Predictors of pressure ulcer healing among long-term care residents

OBJECTIVES

To identify predictors of pressure ulcer healing among long-term care residents.

DESIGN

A retrospective cohort study.

SETTING

Department of Veterans Affairs (VA) long-term care facilities.

PARTICIPANTS

All long-term care residents with a pressure ulcer on April 1, 1993, who remained institutionalized as of October 1, 1993. Patients and pressure ulcer status were identified from the Patient Assessment File, a VA administrative database.

MEASUREMENTS

Pressure ulcers were considered healed if patients were without an ulcer on October 1, 1993. Predictors of pressure ulcer healing were selected from among patient characteristics in the Patient Assessment File.

RESULTS

Pressure ulcers were present in 7.7% of the long-term care residents institutionalized as of April 1, 1993. Among the 819 pressure ulcer patients remaining institutionalized as of October 1, 1993, ulcers had healed in 442 (54.0%). Seventy-two percent of patients with Stage 2 ulcers were ulcer-free at 6 months, compared with 45.2% of patients with Stage 3 ulcers and 30.6% of those with Stage 4 ulcers (P < .001). Significant (P < .05) independent predictors of healing included pressure ulcer size (Odds ratio (OR) = 5.2 for Stage 2 ulcers, OR = 1.5 for Stage 3 ulcers), older age (OR = 1.5), and receiving rehabilitation services (OR = 1.3 for each additional type of therapy). Both immobility (OR = .3) and incontinence (OR = .7) were associated with ulcers not healing.

CONCLUSIONS

Most Stage 2 pressure ulcers, and many larger ulcers encountered in long-term care settings will heal. Baseline patient characteristics are important predictors of healing. Interventions may then be targeted at patients whose ulcers are unlikely to heal, and observed facility performance may be compared with expected outcomes.

Variations in the management of acute myocardial infarction. Importance of clinical measures of disease severity

OBJECTIVE

To determine the extent to which resource use for patients hospitalized with acute myocardial infarction varies with clinical status, and to see if an observed difference in resource use between two states can be explained by clinically detailed risk adjustment.

DESIGN

Retrospective review of the clinical characteristics and resource use of 342 patients hospitalised in two states with acute myocardial infarction.

DATA SOURCES

Merged data from three sources: a large, existing research database used in developing the Medicare Mortality Predictor Score, clinical data abstracted from medical charts specifically for this study, and Medicare Parts A and B claims records.

PATIENTS

A probability sample of Medicare patients hospitalized in 1986 with a diagnosis of acute myocardial infarction and residing in either Wisconsin or Washington state; patients dying within 30 days are oversampled.

MEASUREMENTS AND MAIN RESULTS

Although patients were clinically similar in the two states, there were systematic differences in resource use. Patients in Wisconsin spent more than one extra day in the intensive care unit (ICU) (2.8 vs 1.7) as well as more than one extra non-ICU day in the hospital (8.0 vs 6.5) than patients in Washington. Patients in Wisconsin were also more likely to receive an echocardiogram (35.6% vs 15.8%), nuclear ventriculogram (12.8% vs 4.1%), exercise tolerance test (21.5% vs 3.4), and Holter monitoring (5.4% vs 0%). (All p < .01.) Differences in utilization were greater for patients at lower risk of dying. The average cost of care was 20.8% higher in Wisconsin (p = .01); risk adjustment for clinical and other factors reduced this difference to 11.8%, but did not eliminate it (p = .04).

CONCLUSIONS

Patients with acute myocardial infarction vary in resource use as a function of clinical factors present at admission and occurring during the hospital stay; comparisons that do not take account of these factors may not discriminate well between providers who care for sicker patients and those who are inefficient. The greater use of resources for patients in Wisconsin is at least partially explained by differences in clinical characteristics that are not presently captured in administrative data.

Rating long-term care facilities on pressure ulcer development: importance of case-mix adjustment

OBJECTIVE

To determine the importance of case-mix adjustment in interpreting differences in rates of pressure ulcer development in Department of Veterans Affairs long- term care facilities.

DESIGN

A sample assembled from the Patient Assessment File, a Veterans Affairs administrative database, was used to derive predictors of pressure ulcer development; the resulting model was validated in a separate sample. Facility-level rates of pressure ulcer development, both unadjusted and adjusted for case mix using the predictive model, were compared.

SETTING

Department of Veterans Affairs long-term care facilities.

PATIENTS

The derivation sample consisted of 31 150 intermediate medicine and nursing home residents who were initially free of pressure ulcers and were institutionalized between October 1991 and April 1993. The validation sample consisted of 17 946 residents institutionalized from April 1993 to October 1993.

MEASUREMENT

Development of a stage 2 or greater pressure ulcer.

RESULTS

11 factors predicted pressure ulcer development. Validated performance properties of the resulting model were good. Model-predicted rates of pressure ulcer development at individual long-term care facilities varied from 1.9% to 6.3%, and observed rates ranged from 0% to 10.9%. Case-mix-adjusted rates and ranks of facilities differed considerably from unadjusted ratings. For example, among five facilities that were identified as high outliers on the basis of unadjusted rates, two remained as outliers after adjustment for case mix.

CONCLUSIONS

Long-term care facilities differ in case mix. Adjustments for case mix result in different judgments about facility performance and should be used when facility incidence rates are compared.

Evaluating pressure ulcer occurrence in long-term care: pitfalls in interpreting administrative data

Administrative databases for long-term care frequently collect information on fixed dates of the calendar year, rather than for entire episodes of care. Patients discharged or dying prior to an evaluation date are lost to follow-up. We used one such database, the VA Patient Assessment File, to examine pressure ulcer occurrence in long-term care. Clinical studies have established that most pressure ulcers develop during the first several weeks following admission. In these data, however, pressure ulcer development was less common in patients assessed within 2 months following admission, as compared to those examined at 3 to 6 months. This finding appears to be related to the selective discharge of patients, which makes these patient populations noncomparable. These results highlight that care must be exercised when interpreting results obtained from such administrative data.

Pressure ulcers: the Minimum Data Set and the Resident Assessment Protocol

As Americans age and live longer with chronic disabilities, more elderly citizens will reside in nursing homes. The nationally mandated Minimum Data Set (MDS) and Resident Assessment Protocol (RAP) were designed to help plan for the care of the frail elderly residing in nursing homes. One of the areas specifically targeted was pressure ulcers. This study was designed to describe the prevalence, incidence, and current management practices relating to pressure ulcers in nursing homes based on MDS data. In addition, a cross-sectional investigation was performed to examine the MDS items found in the pressure ulcer RAP. The MDS data for 2,011 nursing home residents, aged 60 or older, who lived in 270 facilities from 10 states were evaluated. The results found a prevalence of 11.2% for Stage II-IV lesions and a 6-month incidence of 6.2%. Logistic regression analysis determined that dependence in transfer or mobility, being bedfast, having diabetes mellitus, and having had a pressure ulcer in the past were significantly associated with a Stage II-IV pressure ulcer. The MDS and its associated RAP for pressure ulcers provide the data to begin devising a care plan for patients with pressure ulcers.

The limited value of routine laboratory assessments in severely impaired nursing home residents

OBJECTIVE

To determine the usefulness of a routine, comprehensive battery of laboratory tests in a severely impaired elderly nursing home population.

DESIGN

Prospective observational survey.

SETTING

Skilled nursing facility wards of a geriatric and extended care veterans hospital.

PATIENTS

Consecutive sample of 108 veterans with severe cognitive and functional impairments, who had been hospitalized at least 6 months.

MAIN OUTCOME MEASURES

Proportions of tests categorized as screening, monitoring, follow-up, or diagnostic; frequency of abnormal test results, interventions warranted and performed on the basis of these abnormalities, and beneficial or adverse effects.

RESULTS

Of 6771 individual nondiagnostic tests performed, 17.2% yielded abnormal results; of these, 33.3% were new. However, only 0.2% of tests resulted in patient benefit. Of 989 panels performed, 31.0% contained at least one abnormality, but only 1.0% of panels (10 patients) yielded any benefit. Overall usefulness was related to the purpose of the testing, with 31.5% of screening tests yielding abnormalities, compared with 45.5%, 78.2%, and 68.7% of monitoring, follow-up, and diagnostic panels, respectively (P < .05 for each compared with screening panels). None of the screening panels detected an abnormality that led to patient benefit, compared with 1.0%, 1.4%, and 3.0% of monitoring, follow-up, and diagnostic panels.

CONCLUSIONS

Routine comprehensive laboratory panels may not be warranted in the most severely impaired elderly patients in long-term care settings. Discontinuing true screening tests and limiting testing strictly to monitoring, follow-up, or diagnostic purposes could minimize the costs of laboratory assessment without losing its potential benefits.

Do-not-resuscitate orders at a chronic care hospital

Do-not-resuscitate (DNR) orders have become an accepted part of medical practice. While these orders have been extensively evaluated in acute care hospitals, little is known about their use in the long-term care setting. We reviewed the medical records of all admissions to a chronic care hospital over a 13-month period, collecting data on selected patient characteristics, use of DNR orders, and patient outcomes during the 6-week period following admission. Fifty-eight of the 301 patients (19.3%) had a DNR order written. Patients' families were involved predominantly in the DNR decision in 73% of the cases while patients themselves were involved in only 18%. Physicians made the decision unilaterally in 6% of the cases. Patients' functional status rather than specific diagnoses predicted the use of DNR orders. Patients with DNR orders were twice as likely to receive new intravenous therapies than patients without those orders (71% vs 33%, P less than 0.01) and four times as likely to die (38% vs 9%, P less than 0.01). They were no more likely to be transferred emergently to an acute care hospital (5% vs 9%, P greater than 0.2). We conclude that DNR orders are not infrequently used, and physicians rarely make the decision unilaterally. Patients with DNR orders have a high likelihood of dying and are infrequently transferred to acute care facilities.

The short-term outcome of pressure sores

Patients with pressure sores have been observed to have a poor prognosis. The short-term outcome of pressure sores at a long-term care hospital was therefore evaluated. Medical records on the 301 admissions to this hospital over a 13-month period were reviewed. One hundred patients (33%) had a pressure sore present on admission. Using ordinary therapies, 79% of these pressure sores improved and 40% completely healed during the 6-week follow-up period. Remaining bed- or chair-bound was the sole patient characteristic associated with a failure of the pressure sore to improve. Mortality rates were significantly increased in patients with a pressure sore present on admission (relative risk [RR] = 1.9), in patients who developed a new sore (RR = 3.1), and in patients in whom the pressure sore failed to improve (RR = 3.3). However, the pressure sores did not appear to be the direct cause of this increased mortality. These data suggest that the majority of pressure sores encountered at a long-term care hospital can be successfully managed in this setting. Although patients with pressure sores have an increased mortality rate, this is most likely due to coexisting medical conditions.

Risk factors for pressure sores. A comparison of cross-sectional and cohort-derived data

The purpose of this study was to identify prospectively risk factors for pressure sores and to compare these results with a cross-sectional analysis in the same population. Medical records on all admissions to a chronic care hospital over a 13-month period were reviewed. Data on potential risk factors were abstracted from the initial history, physical examination, nursing assessment, and laboratory studies. Pressure sore status on admission and at three weeks was determined from a standardized from completed on all patients with a score. The cross-sectional analysis was performed by comparing patients with and without a pressure sore at the time of admission. The cohort analysis used patients initially without a pressure sore and monitored for a new sore at three weeks. Factors associated with pressure sores on univariate testing were entered into a stepwise logistic regression model. One hundred of the 301 admissions presented with a pressure sore. Factors significantly associated with the presence of a sore were altered level of consciousness (OR = 4.1), bed- or chair-bound (OR = 2.4), impaired nutritional intake (OR = 1.9), and hypoalbuminemia (OR = 1.8 for 10 mg/mL decrease). Of the 185 patients without a pressure sore, 20 (10.8%) developed a sore. Factors significantly associated with the development of a new pressure sore were a history of cerebrovascular accident (OR = 5.0), bed- or chair-bound (OR = 3.8), and impaired nutritional intake (OR = 2.8). Neither urinary nor fecal incontinence, nor the presence of hypoalbuminemia, was associated with sore development. We have prospectively identified risk factors for pressure sores.(ABSTRACT TRUNCATED AT 250 WORDS)

The use of follow-up chest roentgenograms among hospitalized patients

Follow-up chest roentgenograms are a commonly performed test. We prospectively evaluated their diagnostic and therapeutic influence at a tertiary care teaching hospital. When a follow-up chest roentgenogram was ordered, physicians indicated their reason for ordering the test, the likelihood that the roentgenogram would show changes, and expected alterations in therapy. After the roentgenogram was obtained, physicians described the help provided by the roentgenogram and what changes in therapy were performed. Using receiver operating characteristic curves, we have shown that physicians have difficulty in predicting which roentgenograms will show important changes. Unexpected findings are frequent (25.4%) and highly valued by the physician. Fifty-seven percent of these roentgenograms had a definite or possible influence on patient treatment. Further studies are indicated to define when follow-up chest roentgenograms are likely to be of benefit.