Search

To authenticate to the full-text of this article, please visit this link: http://dx.doi.org/10.1111/j.1532-5415.2009.02279.x Byline: Ali A. El Solh (*[dagger][double dagger]), Morohunfolu E. Akinnusi ([dagger]), Ziad Alfarah ([dagger]), Anil Patel ([double dagger]) Keywords: nursing home acquired pneumonia; guidelines; antibiotics; mortality Abstract: OBJECTIVES: To compare the 2003 community-acquired pneumonia (CAP) guideline and the 2005 healthcare-associated pneumonia (HCAP) guideline on time to clinical stability, length of hospital stay, and mortality in nursing home patients hospitalized for pneumonia. DESIGN: Retrospective study. SETTING: Three tertiary-care hospitals. PARTICIPANTS: Three hundred thirty-four nursing home patients. MEASUREMENTS: Patients were classified according to the antibiotic regimens they received based on the 2003 CAP guideline or the 2005 HCAP guideline. Time to clinical stability, time to switch therapy, and mortality were evaluated in an intention-to-treat analysis. A multivariate survival model using propensity analysis was used to adjust for heterogeneity between the two groups. RESULTS: Of the 334 patients, 258 (77%) were treated according to the 2003 HCAP guideline. Time to clinical stability did not differ between those treated according to the 2003 CAP or the 2005 HCAP guidelines. Only the Pneumonia Severity Index (P=.006) and multilobar involvement (P=.005) were significantly associated with delay in achieving clinical stability. Adjusted in-hospital and 30-day mortality were comparable in both cohorts (odds ratio (OR)=0.87, 95% confidence interval (CI)=0.49-1.34, and OR=0.79, 95% CI=0.42-1.31, respectively), although time to switch therapy and length of stay were longer for those treated according to the 2005 HCAP guideline. CONCLUSION: In hospitalized nursing home patients with pneumonia, treatment with an antibiotic regimen according to the 2003 CAP guideline achieved comparable time to clinical stability and in-hospital and 30-day mortality with a regimen based on the 2005 HCAP guideline. Author Affiliation: (*)Veterans Affairs Western New York Healthcare System, Buffalo, New York; and ([dagger])Western New York Respiratory Research Center, Division of Pulmonary, Critical Care, and Sleep Medicine, Department of Medicine, School of Medicine and Biomedical Sciences and ([double dagger])Department of Social and Preventive Medicine, School of Public Health and Health Professions, State University of New York at Buffalo, Buffalo, New York. Article note: Address correspondence to Ali El Solh, Medical Research, Bldg. 20 (151) VISN02, VA Western New York Healthcare System, 3495 Bailey Avenue, Buffalo, NY 14215. E-mail: solh@buffalo.edu

Objective. To prepare Agency for Healthcare Research and Quality (AHRQ) for monitoring the impact of its own patient safety initiative, by exploring available outcomes data, assessing usability of measures, and estimating national trends in patient outcomes. Data Sources. Annual summary data on incidence of Joint Commission Sentinel Events, MEDMARX medication error events, and MDS measures of falls and pressure ulcers in nursing home residents. HCUP National Inpatient Sample (NIS) administrative claims data. Methods. Description and assessment of published summary data on selected safety measures. Analysis of selected Patient Safety Indicators (PSIs) and Utah- Missouri adverse event measures using HCUP NIS claims data (1994-2003). Principal Findings. Interpretation of safety outcome trends requires close attention to the characteristics of underlying data sources and measures. Encounter-based measures have been affected by changes in definitions and ICD-9 coding, as well as by changes in the structure of administrative datasets like HCUP NIS. Historical trends are mixed for the safety outcome measures reviewed, with some measures showing improvement, others deterioration, and still others remaining fairly stable. Conclusions. Constructing national trends of safety outcomes is difficult because of limitations in available data sources and measures. Tracking growth in the adoption of safe practices could offer an important strategy for complementing existing safety measurement capabilities. Key Words. Patient safety, measurement, national trends, epidemiology, methods, One of the key components of any effort to track national trends in patient safety involves measuring patient outcomes. Ultimately, outcomes at the clinical level can be understood as the [...]

Objective. To assess longitudinally whether a change in registered nurse (RN) staffing and skill mix leads to a change in nursing home resident outcomes while controlling for the potential endogeneity of staffing. Data Sources. Minimum Data Set (MDS) nursing home resident assessment data from five states merged with Online Survey Certification and Reporting (OSCAR) data from 1996 through 2000. Study Design. Resident-level longitudinal analysis with facility fixed effects and instrumental variables. Outcomes studied are incidence of pressure sores and urinary tract infections. RN staffing was measured as the care hours per resident-day and skill mix was measured as RN staffing hours as a proportion of total staffing hours. Data Extraction Method. We use all quarterly MDS assessments that fall within 120 days of an annual OSCAR data point, resulting in 399,206 resident-level observations. Principal Findings. Controlling for endogeneity of staffing increases the estimated impact of staffing on outcomes in nursing homes. Greater RN staffing significantly decreases the likelihood of both adverse outcomes. Increasing skill mix only reduces the incidence of urinary tract infections. Conclusions. Research that fails to account for endogeneity of the staffing-outcomes relationship may underestimate the benefit from increased RN staffing. Increases in RN staffing are likely to reduce adverse outcomes in some nursing homes. More research using a broader array of instruments and a national sample would be beneficial. Key Words. Staffing, nursing homes, quality of care, instrumental variables, endogeneity, The quality of care in nursing homes has long been an issue of concern to policy makers and the public. Following the Institute of Medicine report that documented serious abuses [...]

In July last year an old friend who now lives in Brighton mentioned that he and his mother, Margaret, were visiting Harlow for the first time in many years. He [...]

BACKGROUND. Lower respiratory infections (LRI) are an important cause of morbidity, mortality, and hospitalization of nursing home residents, yet treatment recommendations have primarily been based on the minority who are [...]

Background. Little is known about the factors that predict whether nursing home residents with lower respiratory infection (LKI) will do well or poorly, although this information is critically important when making treatment decisions. Methods. Using nursing home and hospital medical records we performed a case-control study to identify risk factors for death from LRI among residents of a 110-bed, midwestern community nursing home. Three experienced geriatricians aided in the development of an operational definition of an LRI, In a 3 1/2 -year period, we identified 26 cases in which the patients died from LRI and 66 control episodes in which the patients recovered from LRI. Results. Compared with those who survived, those who died were 14 times more likely to be totally dependent with respect to activities of daily living (ADL) than the group of patients least ALD-dependent (odds ratio [OR] = 14; 95% confidence interval [95% CI] = 2.85 to 68.87). After adjusting for ADL, mortality was significantly decreased when a broad-spectrum oral antibiotic (trimethoprim-sulfamethoxazole, cefaclor, amoxicillin-clavulanate, or ciprofloxacin) was used as the initial therapy (OR = .14; 95% CI = .02 to .81). Conclusions. Better functional status and initial therapy with broad-spectrum oral antibiotics were strong predictors of surviving an LRI in this population of nursing home patients. The antibiotic effect may be a treatment effect or the consequence of underlying factors leading physicians to select particular antibiotics; however, it appears possible to identify low-risk persons who do not require the aggressive treatment and hospitalization that is often recommended for these patients. An approach to the treatment of nursing home LRI is suggested. As many as 12% of hospitalizations among community nursing home residents result from lower respiratory infections (LRI).[1] Estimates of pneumonia incidence in some veterans Administration (VA) chronic care nursing units have been over one per patient-year.[2,3] Death rates from lower respiratory infection as high as 82.6 death per 1000 patient-years have been reported.[4] Given the increasing size of the geriatric population,[5] if current patterns of hospital utilization continue, nursing home lower respiratory infections will consume even greater amounts of scarce medical resources. Little is known, however, concerning which nursing home residents with LRI will do poorly with or without hospital care. Review articles uniformly recommend that almost all nursing home residents with pneumonia should be hospitalized or at least treated with broad-spectrum parenteral antibiotics.[6-9] The studies of LRI in nursing homes that have been cited in support of such assertions evaluated only those nursing home patients who were hospitalized.[10-12] Hospitalized patients are probably a more severely ill group since those who respond to therapy in the nursing home are not admitted for hospital care. In our experience, physicians commonly manage LRIs with oral antibiotics at the nursing home and achieve good outcomes. If clinicians could identify those nursing home residents likely to die from an LRI, they would be faced with a clear choice of administering either aggressive inpatient therapy, which might or might not be successful, or palliative care in the nursing home with a high probability of death. Conversely, if clinicians could identify persons at low risk for adverse outcomes even when treated within the nursing home, this would support the choice of nursing home management for selected residents who develop LRI. We are aware of only two studies of the predictors of LRI outcomes in the elderly. Both concern pneumonia in hospitalized patients.[13,14] We used a case-control design to examine risk factors for mortality from lower respiratory infection in community nursing home patients. We hypothesized that inability to perform activities of daily living (ADL), poor nutritional status, extreme old age, and cognitive impairment would all predispose a patient to dying from LRI. Furthermore, because we expected that more aggressive antibiotic therapy would be associated with more severe illness, we hypothesized that patients who received more aggressive antibiotic therapy would be more likely to die. MetHODS We compared potential risk factors for mortality that were present in nursing home residents who died of lower respiratory infections (cases with risk factors present in episodes where the person survived a lower respiratory infection (controls). Study cases and controls were identified from illnesses that occurred between January 1, 1986, and June 30, 1989, among residents of a 110-bed church-affiliated skilled and intermediate care facility in southeastern Michigan. Because mortality was the end point, some persons were included in both the case and control groups if they had experienced more than one episode of LRI while in the nursing home during the 3 1/2 year period. Otherwise, a substantial portion of the person-time during which individuals were at risk for mortality (time during an LRI episode) would have been excluded for sampling controls.[15] Data were collected by nursing home and hospital medical record review, including nurses&apos; notes, physician progress notes, flow sheets, medication records, laboratory and x-ray film reports, and admission and discharge records. Cases (deaths from LRI) were identified from hospital transfers and nursing home deaths. Controls were identified from hospitalizations that did not end in death and from episodes of antibiotic therapy that were not associated with death. Definition of Lower Respiratory Infection Because there is no standard definition of an LRI in a nursing home patient,[16] an operational definition appropriate to the nursing home setting was developed. First, we formulated a tentative definition of an LRI based on the Centers for Disease Control (CDC) hospital nosocomial infection surveillance definition, which requires evidence of acute respiratory illness, exclusion of other conditions with similar signs and symptoms, and purulent sputum production or a "suggestive chest x-ray."[17] A panel of three geriatricians with extensive experience in the care of nursing home patients reviewed 23 nursing home medical records, four of which were included twice. Each panel member was asked to evaluate the probability that an episode was an LRI on a Likert scale of 1 to 6, ranging from "highly likely" to "highly unlikely." Based on the responses of the panel, we reformulated our original definition of an LRI to arrive at our final definition. There was good agreement between the final definition and the classification of LRI by the raters. If a mean score by the panel of less than 3 (3 = slightly likely to have been an LRI, 2 = moderately likely, 1 = highly likely) was counted as identifying an episode of LRI, agreement between the panel assignment and the final definition was present for 20 of 23 episodes (Cohen&apos;s kappa,[18] a measure of chance-corrected agreement, of .74). Following the CDC definition, we defined an LRI to be present for a hospitalized nursing home resident if there was evidence of an acute infection with either purulent sputum production or a suggestive infiltrate on chest radiograph. In residents not hospitalized, one of the following was required for the patient to be included in the study: 1. An infiltrate suggestive of pneumonia on chest radiograph associated with acute respiratory symptoms or with an acute status change (diminished eating, diminished activity, worse cognitive function, temperature above 37.5 [degrees] C [99.5 [degrees] F], or less than 5 days between the onset of respiratory symptoms and the initiation of antibiotic therapy); or 2. Production of abnormal appearing sputum associated with an acute status change; or 3. Abnormal pulmonary auscultatory findings associated with an acute status change including fever or, in the absence of fever, findings that were not symmetrical (for example, rales heard only in lower left lung). Our initial screening of nursing home records sought any episodes that could have possibly been considered a respiratory infection. Because of our interest in helping physicians make better treatment decisions, individuals deliberately not treated with antibiotics because of planned treatment limitation were excluded. In cases where an individual had multiple courses of antibiotic therapy, we required that a patient had taken no antibiotics for at least 7 days before an episode was considered distinct. Using these rules, we identified 156 potential case and control episode from the records of 98 nursing home patients. Five of these episodes were excluded because they were considered to have been a continuation of another episode or because the outcome could not be determined. Using the final definition, 92 of the 151 episodes were classified as an LRI. Seven of the 59 excluded episodes involved a hospitalization where review of the hospital record did not support the diagnosis of LRI. Of the remaining 52 episodes, in only three instances was a chest radiograph taken (all negative), and in only 5 instances (none associated with an acute status change) was there a report of abnormal sputum production. Some of these 52 episodes lacked evidence of an acute status change (16 episodes; many of these were probably upper respiratory infections), some lacked evidence of abnormal pulmonary findings (25 additional episodes), and some lacked sufficient specific evidence of an LRI (9 episodes). The 92 episodes that met our definition of LRI occurred in 65 persons: 48 had one episode, 10 had two episodes, 5 had three episodes, I had four episodes and 1 had five episodes. Sixteen of the episodes included in the study involved hospitalization. Reliability of Data Abstraction While initially reviewing records, information was also abstracted for analysis. The abstractor (P.C.) was blinded to the purpose of the study other than it involved lower respiratory infections. The abstractor performed a second review of 10 episodes to establish the intrarater reliability of data abstraction. For ordinal and interval variables, 90% of correlation coefficients were above .9 (range .82 to 1.00). In particular, for the number of ADL dependencies, the Spearman correlation coefficient was .99. For categorical variables, there was complete agreement 67% of the time, with no more than two disagreements out of ten in the other cases ([kappa] = 0.57 to 1.0). Exposure Variables We evaluated ADL dependency, cognitive impairment, low weight or body mass index, decubitus ulcer, age, and nursing home diagnoses before the episode as potential predictors of mortality. Cognitive status and ADL dependency were ascertained as close as possible to 7 days before the beginning of antibiotic treatment. Data for this assessment were obtained from nurses&apos; notes and flow sheets. Subjects were classified in regard to ADL according to a modified Katz scale (simply counting the number of dependencies).[19] Coma, confusion, disorientation, dementia, and agitation were considered abnormal cognitive statuses. The patient&apos;s weight at the time of the LRI as well as the patient&apos;s weight 6 and 12 months prior to the LRI was recorded if available. A body mass index (weight/height[.sup.2]) was calculated based on recorded height (usually at admission) and most recent weight. This index provides a measure of adiposity; a low index suggests a very thin and possibly malnourished individual.[20] A decubitus ulcer was defined as a reported stage 1 (area of persistent erythema) or greater decubitus ulcer at the time of antibiotic theory. We classified initial antibiotic use into three mutually exclusive categories: (1) traditional oral antibiotics: penicillin, ampicillin, amoxicillin, erythromycin, tetracycline, cephalexin; (2) broad-spectrum oral antibiotics: trimethoprim-sulfamethoxazole, cefaclor, amoxicillin/clavulanate, ciprofloxacin; and (3) parenteral antibiotics: gentamicin or parenteral cephalosporin. Traditional oral antibiotics are used as the reference stratum for comparing the effect of each of the other two groups. Statistical Analysis Crude odds ratios of death from LRI and 95% confidence intervals were computed for each potential risk factor of interest, and adjusted odds ratios and confidence intervals were computed for potential confounders using the Mantel-Haenzsel technique.[21] As a further evaluation of confounding, multivariate logistic regression was used. Ordinal variables were assessed with a test for trend. The unit of analysis was an episode of infection; thus, some persons were included in both the case and control groups. Because so few episodes involved hospitalization, no separate analyses were performed of whether risk factors deffered in those hospitalized compared with those who received care exclusively in the nursing home. Statistical computations were performed using the following software packages: SAS,[22] Systat,[23] and dEPID,[24] Woolf/Taylor Series confidence intervals have been reported for crude odds ratios. Results Nursing home residents with LRI were a very elderly population with a mean age of 86.3 years (SD = 6.4 years). Twenty-six (28%) of the 92 episodes of LRI ended in death. Five (31%) of the 16 episodes of LRI involving hospitalization ended in death. All subjects were white, and 73 (79%) of the episodes occurred in women, while 19 (21%) occurred in men. As shown in Figure 1, ADL score was strongly related to outcome; death occurred in only 6.7% of the episodes in which the individual was independent or mildly dependent (zero to two dependencies) compared with 50% of episodes in which the individual was totally dependent, scoring six dependencies (Mantel-Haenzel test for trend, [CHI][.sup.2] = 14.2, P < .001). Sixty-two percent of deaths from LRI occurred in the most dependent group, which accounted for only 35% of the episodes. Compared with those who survived, those who died were 14 times more likely to be in the most dependent group than in the least dependent group (OR = 14.95% CI = 2.85 to 68.87). Furthermore, compared with those who survived, those who died were 5.1 times more likely to be in the intermediate dependency group than in the least dependent group (OR = 5.1; 95% CI = 0.98 to 26.43). As age increased, so did the risk of death (Table 1), though with borderline statistical significance [CHI][.sup.2] = 3.68, P = .055). Compared with those who survived, however, those who died were 2.7 times as likely to be older than 90 years (OR = 2.7; 95% CI = 1.03 to 6.96). Body mass index was also strongly associated with death: 48% of the episodes involving a patient in the lowest quartile of the body mass index ended in death while only 14% of the episodes involving a patient in the highest quartile ended in death ([CHI][.sup.2] = 4.87, P = .027). The odds ratio for death from LRI for a patient in the lowest quartile, compared with the highest, was 5.76 (95% CI = 1.30 to 25.51). Nursing home residents who died were 2.7 times as likely to have a decubitus ulcer as those who survived an LRI (OR = 2.72; 95% CI = 1.03 to 7.23). Feeding tubes, oxygen therapy, report of choking, and Foley catheters were all uncommon and, therefore, a larger study would be necessary to adequately evaluate these potential risk factors. Sex, impaired cognitive status, and designation of limited treatment ("do not hospitalize" order) were not significantly related to death from LRI when crude odds ratios were calculated. Presence of a "do not hospitalize" order, however, varied as a risk factor for death from LRI depending on the age of the patient. Those who were less than 90 years of age, were less likely to die if they had a "do not hospitalize" order than were those without such an order (OR = .3; 95% CI = 0.07 to 1.19). Conversely, those aged 90 years or older were more likely to die if they had a "do not hospitalize" order (OR = 4.2; 95% CI = 0.84 to 21.05). Though neither of these odds ratios is statistically significant at the .05 level, the Breslow-Day test for homogeneity of the odds ratio provides strong statistical evidence that the odds ratios are different ([CHI][.sup.2] = 6.33; P = .012). Similarly, the association between death from an LRI and a subject&apos;s sex differs depending on the ADL score. In those totally dependent (six ADL dependencies), women were more likely to die of an LRI than men (OR = 4.2; 95% CI = 0.70 to 25.26); but among those with fewer than six ADL dependencies, men were more likely to die than women (OR = .24; 95% CI = 0.06 to 1.09). The Breslow-Day test again suggests significantly different odds ratios between the strata ([CHI][.sup.2] = 6.347; P = .012). For episodes in which the patient had a fever, the mean initial temperature was equal in case and control patients (38.2 [degrees] C [101.2 [degrees] F, and 101.1 [degrees] respectively]), but the mean value for the maximum temperature was slightly higher in the case study group (39.2 [degrees] C [102.6 [degrees] F], compared with patients in the control group 38.7 [degrees] C [101.6 [degrees] F], P = .005). In 21 of the 92 episodes, however, no information was available for either of these variables. The type of antibiotics prescribed at the beginning of an episode shows a striking relationship to the outcome. Parenteral antibiotics were associated with a poorer outcome (Table 1); however, the result was not statistically significant and was markedly attenuated when controlling for ADL with the Mantel-Haenzsel technique (OR = 1.1; 95% CI = 0.22 to 5.77). In contrast, when similarly controlling for ADL, initial use of broad-spectrum oral antibiotics was associated with much lower odds of dying than use of traditional oral antibiotics (OR = 0.14; 95% CI = 0.02 to 0.81). Among residents not hospitalized, none of the 13 episodes involving initial use of broad-spectrum oral antibiotics and fewer than six ADL dependencies resulted in death. One individual was started on broad-spectrum oral antibiotics and then hospitalized. That person survived the episode. [Table Data 1 Omitted] We also examined whether using only one episode per person would alter the results. Though there are some differences, odds ratios are similar for one episode as compared with all episodes. To further examine the relationships of the predictor variables, we used multivariate logistics regression. Neither body mass index nor the presence of a decubitus ulcer was a significant predictor when controlling for ADL with logistic regression. These findings are consistent with results from stratified analysis, which showed that when controlling for ADL, dichotomous classifications of age, boy mass index, and decubitus ulcer exhibited attenuated odds ratios and a loss of statistical efficiency. The logistic regression model supported the very strong relationship between initial treatment with a brood-spectrum oral antibiotic and survival; the adjusted odds ratio when compared with treatment with traditional oral antibiotics was 0.09 (95% CI = 0.01 to 0.54). Initial treatment with parenteral antibiotics was a weak predictor and not statistically significant. The model also reiterated how much better the outcomes were for those in the most independent ADL group than those in the intermediate or fully dependent groups. Because of the small number of patients in each ADL classification, confidence intervals were very wide. Discussion Previous nursing home studies of lower respiratory infection either have been purely descriptive or have been focused on causative factors and microbial pathogens rather than on outcome.[16] The ability to assess the risk of adverse outcome is crucial, however, to a treatment strategy for acutely ill nursing home residents with suspected LRI. Patients who are likely to do well may be candidates for nursing home therapy. Those who are likely to do poorly may be candidates for hospital admission if their preferences and their overall condition dictate aggressive treatment. Aside from decisions to limit treatment, there are multiple potential reasons for not hospitalizing a nursing home resident, including cost, precipitation of acute confusion, and the increased possibility of iatrogenesis in the hospital environment.[25] This study involved only white patients in a rural nursing home; therefore, the results should be generalized with caution. Our findings suggest, however, that a low-risk population can be identified in which patients with LRI can be appropriately treated with oral antibiotic therapy administered in the nursing home. The study by Zweig et al[13] of elderly hospitalized patients identified the following as significant independent predictors of mortality: impaired consciousness, tachypnea, lower than normal body temperature, leukocyte count above 20,000 mm[3], and cyanosis. Starczewski and co-workers[14] identified with mortality in those over 70 years of age at a general hospital in the United Kingdom. We were not able to adequately test many of these associations with our data. We did not find an association between cognitive status and outcome, but cognitive status was not consistently reported in nursing home records. Also, we examined cognitive status just before an episode as opposed to Zweig et al, who examined cognitive status at the time of hospitalization. Based on the present study, Zweig&apos;s hospital study,[13] and suggestions from Bentley&apos;s review of pneumonia in the elderly,[7] factors that might turn out to be high-risk indicators in elderly nursing home patients with pneumonia include: (1) very dependent ADL status; (2) low body temperature; (3) depressed level of consciousness; (4) a leukocyte count of less than 5000 or greater than 20,000 mm[3]; and (5) evidence of respiratory distress (cyanosis, respiratory rate greater than 30 breaths per minute, hypoxemia). Additionally, very high body temperature (eg, above 39.4 [degrees] C [103 [degrees] F]) is often associated with severe illness, though we are aware of no solid evidence linking it to mortality. These factors have never been tested jointly in any study and will likely not all turn out to be independently important; nonetheless, they may serve as basic guidelines for clinicians in the absence of better information. In the current study, episodes where broad-spectrum oral antibiotics were the initial therapy had very low mortality, and this relationship was even stronger when controlling for ADL status. Our data do not identify how clinicians made their specific decisions about therapy. Therefore, we do not know whether this represents a treatment effect or a possible selection bias in which less severely ill individuals might have been treated with broad-spectrum oral antibiotics. Whether the result of a treatment effect or good clinical judgment, or both, the clinician&apos;s treatment decisions were supported by the positive outcomes. This is particularly interesting when compared with the lack of a relationship between the iniial use of parenteral antibiotics and LRI mortality when controlling for ADL status. We hypothesized that persons who were more severely ill would be more likely to receive parenteral antibiotics, and therefore those who were initially treated with parenteral antibiotics would have pooper outcomes. The absence of any evident relationship when controlling for ADL reemphasizes the importance for ADL status in identifying those persons likely to have the poorest outcomes. Previous work has suggested that poor ADL status is a predictor of mortality in nursing home patients,[26] and Alvarez et al[2] found that a poor Karnofsky performance index (a global measure of ability to function) was a predictor of lower respiratory infections. However, to the best our our knowledge, this is the first study that relates ADL status to the outcome of LRI. Malnutrition has been linked to nursing home mortality.[27] Our data suggest that very thin persons are at significantly higher risk of dying from LRI; however, in multivariate analysis, body mass index did not remain an important predictor. Similarly, ADL status was a better predictor of mortality than was age. Age and body mass index might remain significant independent factors when controlling for ADL in a larger study. The findings relating to subject&apos;s sex and treatment limitation are not obviously explained. In this population, person younger than 90 years with limited treatment orders were less likely to die than those younger than 90 years without limited treatment orders. We also found that women with severe ADL impairment were more likely to die than severely impaired men, while men with absent to moderate ADL impairment were more likely to die than women of similar ADL status. These findings are intriguing but need to be replicated before much significance is attached to them. The limitations of retrospective medical record review studies are well known. There is no way to ensure uniformity in what was recorded. Lacking adequate data to apply standard hospital LRI definitions, we went through a complex process to develop a nursing home LRI definition to establish eligibility for the study. Although the definition has not been previously used, it makes clinical sense (has face validity) and fits well with mean ratings by the panel. There was, however, significant variability in ratings between individual panel members. Clearly, having complete data to uniformly apply a more standard definition would have been preferable. Furthermore, using medical records as a source for data could create a spurious association if one of the exposures of interest is selectively underreported in case or control patients. Nonetheless, as a basis of generating hypotheses for prospective studies, this type of investigation has merit, and the strengh of the association of death with poor ADL status is unlikely to be wholly spurious. In fact, a strength of this study is that ADL status is apt to be better recorded in nursing home records than on hospital records. For both nursing home and hospital settings, future studies involving prospective data collection are needed to more precisely examine outcome predictors. The episode as the unit of analysis raises the question of whether observations are truly independent; however, it is appropriate to consider the same person as a case and a control at different time, particularly in studies with mortality as the end point.[15] In support of our approach, when we analyzed only the last episode for each individual, we obtained substantially similar results. Despite the limitation of our study, we did demonstrate that, at least in this nursing home, there was a significant subgroup of residents with LRI that did very well with treatment in the nursing home with broadspectrum oral antibiotics. The results also support the usefulness of simple, available clinical data in predicting outcome from LRI. If this approach can be extended and confirmed, it will provide a powerful clinical and research tool. The ability to predict outcomes on the basis of data available at the beginning of an illness could significantly aid in decision making, in the stratification of persons by expected outcome in future studies, and in the outcomebased assessment of quality of care in nursing homes. Acknowledgments Lynn Hostetler helped with initial literature searching. Jim Peggs, MD, Alan Dengiz, MD, and John Santinga, MD, volunteered their time to review nursing home records in formulating the definition of lower respiratory infection. The administration and staff of the Chelsea United Methodist Retirement Homes, Particularly the medical records coordinators, provided invaluable assistance. Similarly, the Chelsea Community Hospital medical records department was very helpful. Dan Gorenflo, PhD, performed some of the statistical analyses. Mindy Smith, MD, MS, and Barbara Reed, MD, MSPH, provided helpful comments. Mary Anne Hartness assisted with manuscript preparation.@ References [1.] Irvine PW, Van Buren N, Crossley K. Causes for hospitalization of nursing home residents: the role of infection. J Am Geriatr Soc 1984; 32:103-7. [2.] Alvarez S, Shell CG, Woolley TW, et al, Nosocomial infections in long-term facilities. J Gerontol 1988; 43:M9-17. [3.] Farber BF, Brennen C, Puntereri AJ, et al. A prospective study of nosocomial infections in a chronic care facility. Am Geriatr Soc 1984; 32:499-502. [4.] Nicolle LE, McIntyre M, Zacharias H, et al. Twelve-month surveillance of infections in institutionalized elderly men. J Am Geriatr Soc 1984;32:513-9. [5.] Guralnik JM, FitzSimmons SC. Aging in America: a demographic perspective. Cardiol Clin 1986; 4:175-83. [6.] Norman DC, Castle SC, Cantrell M. Infections in the nursing home. J Am Geriatr Soc 1987; 35L796-805. [7.] Bentley DW, Bacterial pnuemonia in the elderly: clinical features, diagnosis, etiology, and treatment. Gerontology 1984; 30:297-307. [8.] Niederman MS, Fein AM. Pneumonia in the elderly. Clin Geriatr Med 1986; 2:241-68. [9.] Yoshikawa TT. Pneumonia, UIT, and decubiti in the nursing home: optimal management. Geriatrics 1989; 44(10):32-4, 37-40, 43. [10.] Marrle TJ, Durant H, Kwan C. Nursing home-acquired pnuemonia: a case-control study. J Am Geriatr Soc 1986; 34:697-702. [11.] Garb JL, Brown B, Garb JR, et al. Differences in etiology of pneumonias in nursing home and community patients. JAMA 1978; 240:2169-72. [12.] Peterson PK, Stein D, Guay DRP, et al. Prospective study of lower respiratory tract infections in an extended-care nursing home program: potential role of oral ciprofloxacin. Am J Med 1988; 85: 164-71. [13.] Zweig S, Lawhorne L, Post R. Factors predicting mortality in rural elderly hospitalized for pneumonia. J Fam Pract 1990; 30: 153-9. [14.] Starczewski AR, Allen SC, Vargas E, Lyle M. Clinical prognostic indices of fatality in elderly patients admitted to hospital with acute pneumonia. Age Ageing 1988; 17:181-6. [15.] Rothman KH. Modern epidemiology. Boston: Little, Brown, 1986:66. [16.] Crossley KB, Thurn JR, Nursing home-acquired pneumonia. Semin Respir Infect 1989; 4:64-72. [17.] Garner JS, Bennett JV, Scheckler WE. et al. Surveillance of nosocomial infections. In: Brachman PS and Eickhoff TC, eds. Proceedings of the international conference on nosocomial infections. Baltimore: Waverly Press, 1971; 277-81. [18.] Fleiss JL. Statistical methods for rates and proportions, 2nd ed. New York: John Wiley & Sons, 1981. [19.] Katz S, Ford AB, Moskowitz RW. Studies of illness in the aged. The index of ADL: a standardized measure of biological and psychosocial function. JAMA 1963; 185:914-9. [20.] Garrow JS, Webster J. Quetelet&apos;s index (W/H2) as a measure of fatness. Int J Obes 1985; 9:147-5. [21.] Schlesselman JJ. Case control studies. New York: Oxford University Press, 1982. [22.] SAS Institute Inc. SAS user&apos;s guide: statistics. Version 5 edition. Cary, NC: SAS Institute Inc, 1985. [23.] Wilkinson, L. SYSTAT: the system for statistics. Evanston, Ill: SYSTAT, Inc, 1988. [24.] Sullivan KM, Foster DA. dEPID: a program for stratified and standardized analysis. Version 2.1 Atlanta, GA, and Ann Arbor, Mich, 1987. [25.] Rubenstein LZ, Ouslander JG, Wieland D. Dynamics and clinical implications of the nursing home-hospital interface. Clin Geriatr Med 1988; 4:471-91. [26.] Lichtenstein MJ, Federspiel CF, Schaffner W. Factors associated with early demise in nursing home residents: a case control study. J Am Geriatr Soc 1985;33:315-9. [27.] Rudman D, Feller AG. Protein-calorie undernutrition in the nursing home. J Am Geriatr Soc 1989; 37:173-83.

The 'Alzheimerization' of dementia research is discussed . A study on survival to late-stage dementia in Dutch nursing home patients touches on important issues related to clinical diagnosis of dem entia. Future studies should confirm diagnoses based on stages established by H. Braak et al. Not all Alzheimer's patients are demented.

A retrospective cohort study with development and validation cohorts was carried out to identify factors associated with one -year mortality in newly admitted and long-stay ( one year+ ) nursing home residents and to use them to create a tool for estimating risk l evels for one-year mortality. This was done by linking Minimum Data Set (MDS) information with data from the National Death Index . It was found that MDS data can be used to identify major factors associated with one-year mortality in the group studied.

An observational analysis of a cohort of patients has been carried out followed by a prospective follow-up to investigate the determinants of survival to late dementia in Dutch nursing home patients. It was found that one of seven nursing home patients with dementia survived to late dementia. Such information is useful in giving information about a prognosis to relatives of a patient. Some patients lived 25 years after disease onset. Most who reached late dementia were women with A lzheimer's disease, and for most of them the phase lasted one year or less. Dementia shortens life expectancy.

Nursing home facility risk factors for infection and hospitalization are discussed with focus on findings related to the importance of registered nurse turnover, administration, and social factors. A stratified random sample of 59 nursing homes across Maryland was carried out for this study. The positive relationship between hospitalization for infection and for-profit ownership and chain affiliation could indicate policies not to treat acute illnesses inhouse exist.

* Carman W, Elder A, Wallace L, et al. Effects of influenza vaccination of health-care workers on mortality of elderly people in long-term care. Lancet 2000; 355:93-7 CLINICAL QUESTION Does [...]

The nutritional status of nursing home residents typically is poor and often results in negative health outcomes such as increased illness, disability, and psychological complaints, as well as unintended weight [...]

PURPOSE: There is limited information regarding the outcome and survival of institutionalized patients who present with acute respiratory failure. We investigated the inpatient outcome and one year mortality of nursing [...]

There has been much speculation on the risk of mortality associated with sleep apnea. We followed-up 233 elderly patients in nursing homes, 70 percent of whom had five or more [...]

Antipsychotic appear to increase the risk of death in patients with Alzherimer's disease, especially if they were taken for more than 12 months, according to data from a randomized controlled [...]

PURPOSE: This study was conducted to evaluate the incidence, identify the risk factors, and assess the prognosis of elderly institutionalized patients who develop fecal incontinence. PATIENTS AND METHODS: We enrolled 1,186 patients 60 years of age and older living in long-term care facilities who did not have fecal incontinence. We assessed their medical history, treatment, mobility, and cognitive function. Patients were followed up for 10 months to determine the incidence of fecal incontinence, defined as at least one involuntary loss of feces. Independent risk factors associated with fecal incontinence were identified using Cox proportional hazards models. The prognosis of incontinent patients was assessed by comparing their survival rate with that in the continent patients. RESULTS: Fecal incontinence occurred in 234 patients (20%), and was usually associated with acute diarrhea or fecal impaction. We identified five risk factors for the development of fecal incontinence: a history of urinary incontinence (rate ratio [RR]: 2.0, 95% confidence interval [CI] 1.5 to 2.6); neurological disease (RR: 1.9, 95% CI 1.0 to 3.4); poor mobility (RR: 1.7, 95% CI 1.2 to 2.4); severe cognitive decline (RR: 1.4, 95% CI 1.1 to 1.9); and age older than 70 years (RR: 1.7, 95% CI 1.0 to 2.8). Tenmonth mortality in the 89 patients with long-term ([is greater than or equal to] 8 days) incontinence was 26%, significantly greater than that observed in the continent group (6.7%) or in the 145 patients with transient incontinence (10%). CONCLUSIONS: Long-lasting or permanent fecal incontinence was associated with increased mortality, suggesting that this symptom is a marker of poor health in older patients. Actions that improve mobility might help prevent fecal incontinence in elderly patients. Am J Med. 1999;106:185-190. [C] 1999 by Excerpta Medica, Inc.

Background: Congestive heart failure is a major cause of mortality and morbidity in the elderly but the disease impact on the oldest and sickest population has not been defined. Objectives: To review the mortality and hospital readmission rate of institutionalized elderly persons with congestive heart failure and to examine the relation of baseline characteristics to subsequent clinical outcomes. Methods: This was a retrospective analysis based on chart review of 231 residents of the Philadelphia (Pa) Geriatric Center (63 congregate housing tenants and 168 nursing home residents) 80 years and older, hospitalized with congestive heart failure from 1989 to 1995. Patients' demographic data and clinical, electrocardiographic, and echocardiographic findings were obtained from their initial (index) hospitalization records. Subsequent outcomes were obtained from their outpatient (nursing home or office) records. Results: Thirteen percent died during the index hospitalization but the total mortality during the follow-up period was 87%. One hundred forty-six patients (63%) died in the first year with a mean [+ or -] SD survival of 4 [+ or -] 4 months and a readmission rate of 3.9 per patient-year. Eighty-five patients survived the first year with a readmission rate of 1.2 per patient-year and 54 patients subsequently died, with a mean [+ or -] SD survival of 28 [+ or -] 12 months. The first-year decedents and survivors were comparable in sex, age, medical history, and electrocardiographic findings. However, patients who died in the first year, compared with survivors, were more likely to be nursing home residents (81% vs 59%), have New York Heart Association class IV heart failure (54% vs 32%), have impaired left ventricular function by echocardiogram (53% vs 32%), and have renal insufficiency (32% vs 11%). Conclusions: Very elderly persons with congestive heart failure had a guarded long-term prognosis. Nursing home residency, class IV heart failure, impaired left ventricular function, and renal insufficiency were associated with higher risk for early death and repetitive hospitalizations. Arch Intern Med. 1998;158:2464-2468

Advance directives in nursing homes, which consist primarily of do not resuscitate (DNR) orders, are more prevalent among whites, those with severe cognitive and functional impairment, and the very old, [...]