Open Collections

UBC Faculty Research and Publications

Post-hospital syndrome in adults with asthma: a case-crossover study Sadatsafavi, Mohsen; Lynd, Larry D; FitzGerald, J M Dec 23, 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-13223_2013_Article_404.pdf [ 308.8kB ]
JSON: 52383-1.0223603.json
JSON-LD: 52383-1.0223603-ld.json
RDF/XML (Pretty): 52383-1.0223603-rdf.xml
RDF/JSON: 52383-1.0223603-rdf.json
Turtle: 52383-1.0223603-turtle.txt
N-Triples: 52383-1.0223603-rdf-ntriples.txt
Original Record: 52383-1.0223603-source.json
Full Text

Full Text

RESEARCH Open AccessPost-hospital syndrome in adults with asthma:a case-crossover studyMohsen Sadatsafavi1,2*, Larry D Lynd2,3 and J Mark FitzGerald1AbstractBackground: Post-hospital syndrome refers to the period of generalized risk of adverse health outcomes amongpatients who are recently discharged from hospital. This period is associated with a short-term increased risk ofreadmission which may not be related to the original condition. While the majority of studies of post-hospitalsyndrome have focused on all-cause readmissions, whether and to what extent such a phenomenon exists withindiscrete medical conditions is not yet known.Objective: To investigate whether the risk of admission due to asthma is increased in individuals who aredischarged following any-cause hospital admission.Methods: Using administrative health data for the period 1997 to 2007 from the province of British Columbia, Canada,we created a cohort of adults with asthma. Using a case-crossover design, we assessed the association between dischargefrom a hospital (exposure) within 30 days before an asthma-related hospitalization (the outcome), using two 30-daycontrol periods within the same subject. Conditional logistic regression was performed to calculate the relative risk (RR)of the outcome in association with exposure. We performed several sensitivity and subgroup analyses.Results: The final cohort included 3,852 patients experiencing 6,333 instances of the outcome. Mean age at the time ofthe outcome was 43.7 (SD 14.2), 69.0% of such outcomes belonged to females. The RR of the outcome within the next30 days of a previous any-cause discharge was 1.40 (95% CI 1.22 - 1.59). However, the association was mainly caused bydischarge from asthma-related admission [RR = 1.99 (95% CI 1.65 - 2.39)]. The RR associated with non-asthma-relateddischarge was 0.88 (95% CI 0.74 - 1.04) and was not statistically significant. Similar results were obtained in arange of sensitivity analyses.Discussion: Our results indicate that in patients with asthma, the 30-day risk of asthma-related admission is increasedafter an episode of asthma-related hospitalization, but not after an episode of non-asthma-related hospitalization.IntroductionPost-hospital syndrome refers to the acquired, transientperiod of generalized increased risk for a broad range ofhealth conditions after discharge from hospital [1]. Animportant manifestation of this syndrome is the highrate of readmission in the critical 30 days after discharge,which is not necessarily due to the same condition. Arecent study showed that nearly 20% of patients coveredby a US national insurance program (Medicare, consistingof individuals aged 65 and older as well as those withdisabilities or end stage renal disease) discharged froma hospital had another acute medical problem within thesubsequent 30 days that necessitated re-hospitalization[2]. Reasons for readmissions often include heart failure,pneumonia, COPD, infection, gastrointestinal conditions,mental illness, metabolic derangements, and trauma[2]. Various etiological reasons are postulated for thisphenomenon, including disturbances of sleep, nutri-tional issues, pain and discomfort, and psychologicalconfusion [1].While the typical post-hospital syndrome affectsmostly elderly patients with co-morbid conditions, thepotential etiologic factors exist regardless of age. Inaddition, potential causal factors might have differentialeffects across different medical conditions. As such,* Correspondence: msafavi@mail.ubc.ca1Department of Medicine, Institute for Heart and Lung Health, The Universityof British Columbia, 7th Floor, 828 West 10th Avenue, Research Pavilion,Vancouver V5Z 1 M9, Vancouver, BC, Canada2Collaboration for Outcomes Research and Evaluation, Faculty of PharmaceuticalSciences, the University of British Columbia, Vancouver, CanadaFull list of author information is available at the end of the articleALLERGY, ASTHMA & CLINICAL IMMUNOLOGY© 2013 Sadatsafavi et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of theCreative Commons Attribution License (, which permits unrestricted use,distribution, and reproduction in any medium, provided the original work is properly cited. The Creative Commons PublicDomain Dedication waiver ( applies to the data made available in thisarticle, unless otherwise stated.Sadatsafavi et al. Allergy, Asthma & Clinical Immunology 2013, 9:49 post-hospital syndrome within the realm of spe-cific diseases can be informative from a patho-physiologicalperspective, and is also important from a clinical perspec-tive as it can help risk-stratify individuals at the time ofdischarge from hospital.To our knowledge, no previous study has examinedthe risk of an asthma-related hospitalization after dischargefrom a previous hospitalization. Any hospitalization inde-pendent of the cause might affect the risk of a subsequentasthma exacerbation in complex ways. Some of the postu-lated factors for the post-hospital syndrome can increasethe risk of worsening of asthma symptoms and asthmaexacerbations. For example, psychological stress is knownto be associated with asthma attacks [3], and changes inthe immune system and malnutrition are all associatedwith an increased risk of the worsening of asthma [4].In addition, medications that patients receive duringadmission (e.g., non-steroidal anti-inflammatory drugs orbeta-blockers) might cause drug-induced exacerbations[5]. On the other hand, patients with asthma who arehospitalized for other reasons will most likely also bemanaged for their asthma during the inpatient period,and at times receive potent anti-inflammatory medicationswhich can reduce the risk of an asthma attack. Combiningall these factors, there seems to be different potentialmechanisms for post-hospital syndrome affecting therisk of an asthma-related admission that has not hithertobeen evaluated.Using administrative health data from the provinceof British Columbia, Canada, we set out to evaluatewhether individuals with asthma are at higher risk of anepisode of asthma-related hospitalization after dischargefrom hospitalization due to any cause. We also evaluatedsuch a risk across different subgroups of patients, andevaluated the risk according to the cause of the originaladmission as being asthma-related, non-asthma respiratory-related, or non-respiratory-related.MethodsDataRecords of the utilization of all billed services for thefiscal years 1997 to 2007 were obtained from the BritishColumbia Ministry of Health ( [6]. This study was approved by University ofBritish Columbia-Providence Health Care Research EthicsBoard (#H08-01287). No consent was required as thedata consisted of anonymized health records released toinvestigators in accordance with the Provincial Freedomof Information and Protection of Privacy Act. We hadaccess to consolidation files [7] and all records of inpatient[8] and outpatient [9] encounters, as well as medicationdispensations (the PharmaNET database [10]) during thisperiod. Hospitalization and outpatient service use recordscontain encounter dates and International Classification ofDiseases (ICD, 9th and 10th revisions) codes for the reasonfor the encounter. For hospitalization records, up to 25ICD codes are recorded, one of which is designated as the‘most responsible’ diagnosis; i.e., the diagnosis responsiblefor the greatest portion of the patient’s stay in the hospital.The length of stay in the hospital as well as admission type(urgent versus elective) is also available. The medicationdispensation database includes variables such as theunique drug identifier and date of dispensation [11].Study cohortAdults (age 18 years and older) were considered as havingasthma if during a rolling time window of 12 months theyfilled prescriptions for at least three asthma-related medi-cations (list of such medications is available in Additionalfile 1). The date of the first of the three prescriptions wasconsidered the cohort entry date. The date of the lastresource use of any type was considered the exit date.DesignWe chose a case-crossover design for this study in whicheach individual subject acts as their own control [12]. Inthe study of the association of transient exposures withacute outcomes, such as this study, the case-crossoverdesign is an attractive option as it inherently removes thepotential biasing effects of unmeasured, time-invariantconfounding factors [13]. Adjusting for such unmeasuredconfounding factors is particularly important in thepresent context as both the exposure and outcome in thisstudy are hospitalization events, and many factors mightaffect the overall person-specific rate of hospitalization(e.g., the patient’s and care provider’s threshold forinpatient care, the availability of hospital beds in thelocal health area, and co-morbid conditions). This designis similar to the classic case–control design, with the maindifference that the case and control periods belong tothe same subject, albeit at different times. In a sensitivityanalysis we also performed a conventional nested case–control study to evaluate the robustness of the results tothe design specifications. A schematic illustration of thecase-crossover design is provided in Figure 1.ExposureThe primary exposure in this analysis was dischargefrom an episode of hospital admission with at least onefull day of stay, regardless of the cause. We further cate-gorized such admissions to be due to asthma-relatedversus non-asthma-related as well as respiratory-relatedversus non-respiratory-related. Asthma-related admissionswere those with the most responsible diagnosis beingfor asthma (ICD-9 493.xx, ICD-10 J45/J46). Respiratory-related admissions were those with the most responsiblediagnosis being for a respiratory condition (ICD-9 codes460–519, ICD-10 codes Jxx).Sadatsafavi et al. Allergy, Asthma & Clinical Immunology 2013, 9:49 Page 2 of 8 outcome of interest was a non-elective admissionto hospital with the main diagnosis being asthma (asdescribed above). A national chart review of the data forthe 2005–2006 fiscal year showed that the main diagnosisof asthma in a discharge record had a sensitivity of 87%(95% CI 79%–95%) and a positive predictive value of 90%(95% CI 85%–95%) [14]. In addition, restricting the studypopulation to those who satisfied a case definition ofasthma means that the subset of hospitalizations includedin the analysis were even more likely to be truly asthma-related. In line with the general definition of post-hospitalsyndrome, readmissions that counted towards the out-come did not include elective admissions or emergencyroom visits that did not result in inpatient admissions.Case and control time windowsThe 30-day period immediately before each asthma-related hospitalization was considered as the case timewindow (see Figure 1). For each case time window, weconsidered up to two control time windows endingexactly 364 days before and after the index date. Thechoice of the timing of control time windows was toadjust for the effect of seasonality as well as day of theweek as such factors might be potentially correlatedwith the risk of hospitalizations. We specifically avoidedusing control time windows that are adjacent to case timewindows because first, the individual is protected fromexperiencing a hospitalization event in the time periodimmediately after the case time window (due to thelength-of-stay of the index hospitalization associatedwith the case time window); second, the time adjacent tothe case time window would belong to different monthand potentially season which is an important factor affect-ing the risk of asthma-related hospitalization.Eligible control windows were those that fit entirelywithin the interval between the entry and exit date ofthe individual. Control time windows that did not satisfysuch criteria were removed. Case time windows for whichboth control time windows were excluded were alsoexcluded as they could not contribute to the statisticalinference. Each individual could contribute several casesand associated control time windows.AnalysisWe calculated the rates of the occurrence of exposurein both the case and control time windows. In doingso, and in line with basic principles, we weighted eachcontrol time window according to the reciprocal of thenumber of control time windows for the correspondingcase time window [15]. In the main analysis, using condi-tional logistic regression we calculated the adjusted relativerisk (RR) of the asthma-related hospitalization (outcome)in association with exposure. We controlled for potentiallytime-varying measures of asthma severity (number ofasthma-related admissions, outpatient service use, medi-cation dispensation, as well the number of dispensedcanisters of short-acting beta-agonists (SABA), inhaledcorticosteroids (ICS), and combined ICS and long-actingbeta-agonists (ICS + LABA)) and general measures ofco-morbidity (Charlson co-morbidity index [16], totalnumber of admissions, outpatient services use andmedication dispensations), all measured in the 180 daysprior to the (Case and control) time window (Figure 1).Robust variance estimators were used for inference toaccount for within-subject clustering of events (asoutcomes that belong to the same person cannot beconsidered as independent observations) [17].Subgroup, sensitivity, and alternative analysesSubgroup analysis involved separately performing theanalysis by sex and age groups. We performed severalsensitivity analyses. These included an unadjusted analysisas well as performing the analysis using a conventionalnested case–control design. The nested case controldesign followed the same principles as the main studyEntry dateOutcomedateControl time windowControl time window364 days 364daysExit date30 days 30 days 30daysCase time window180 daysCovariate assertion window180 daysCovariate assertion window180 daysCovariate assertion windowFigure 1 Schematic illustration of the cohort construction and analysis type. The arrow from left to right represents the timeline of anindividual within the data. The vertical arrow shows an outcome date (admission to hospital due to asthma). The immediate 30-day prior to thisdate constitutes the case time window. For each subject, up to two control time windows of the same length were also selected, each 364 daysbefore and after the start of the case time window. The presence of discharge from a hospitalization in the case and control-time windowsdefines the exposure.Sadatsafavi et al. Allergy, Asthma & Clinical Immunology 2013, 9:49 Page 3 of 8 definition, inclusion criteria, case and control timewindows) with the difference being that the control timewindows were selected from other individuals in therisk set, and no 364-day-interval rule was applied. For agiven case time window, the risk set was defined asthose individuals who were at the risk of experiencingthe outcome (that is, the follow-up day of the indexcase they was between the entry and exit date of thecontrols) and had the same sex, year of birth, and similarentry date (within 180 days) as the case. The nested-casecontrol analysis was further adjusted for all covariates thatwere controlled for in the main case-crossover design.Other sensitivity analyses included choosing control timewindows at different distances from the case time window.We performed two alternative analyses exploring theassociation between discharge from hospital and 60-day and 90-day risk of asthma-related admission. Finally,to evaluate the potential impact of treatment in thepost-discharge period in the association between theexposure and outcome, we performed an additionalsensitivity analysis in which the regression model wasfurther controlled for whether the individual received anycontroller medication (systemic corticosteroids, inhaledcorticosteroids with or without long-acting beta-agonists,or leukotriene receptor antagonists) during the exposure(case or control) time window.ResultsThe study cohort constituted 178,192 individuals withasthma among whom there were 6,333 asthma-relatedhospitalizations (outcome) experienced by 3,852 uniqueindividuals. The associated case time windows for theseevents were matched to 10,737 control time windows(1.70 control time windows per event). Table 1 providesthe basic demographics characteristics of the individualsin the final analysis.Results of the conditional logistic regression are providedin Table 2. In 10.8% of case time windows there was adischarge from hospital. In comparison, in 7.9% of controltime windows there was a discharge from hospital,resulting in a RR of asthma-related re-hospitalizationfollowing a hospital admission from any cause of 1.40(95% CI 1.22 - 1.59), P < 0.001. Nevertheless, such anincreased risk of the outcome was mainly due to theasthma-related discharge, with an RR of 1.99 (95% CI1.65 - 2.39). Non-asthma-related discharges were notassociated with 30-day asthma-related readmissions(RR = 0.88 (95% CI 0.74 - 1.04), P = 0.14). 81% of allrespiratory-related discharges were due to asthma. Assuch, and expectedly, they were associated with anincreased risk of the outcome (RR = 1.89 (95% CI 1.60 -2.22)). Respiratory-related, non-asthma discharges werenot associated with the risk of the outcome (RR = 1.23(95% CI 0.90 - 1.68), P = 0.20). Non-respiratory-relatedTable 2 Association between exposure (discharge from hospital) and outcome (asthma-related admission in thenext 30 days)Exposure Frequency Adjusted RR P-valueCase window (95% CI)†Control windowN = 6,333 N = 10,737Any discharge 687 (10.8%) 849 (7.9%) 1.40 (1.22 – 1.59) <0.001*Asthma-related discharge 442 (7.0%) 389 (3.6%) 1.99 (1.65 – 2.39) <0.001*Non-asthma-related discharge 273 (4.3%) 495 (4.6%) 0.88 (0.74 – 1.04) 0.141Respiratory-related discharge 523 (8.3%) 497 (4.6%) 1.89 (1.60 – 2.22) <0.001*Respiratory-related, non-asthma discharge 89 (1.4%) 124 (1.2%) 1.23 (0.90 – 1.68) 0.199Non-respiratory-related discharge 189 (3.0%) 381 (3.5%) 0.78 (0.63 – 0.96) 0.017**Significant at 0.05 level.†All RRs are adjusted for the following variables estimated in the 180 days prior to the start of the time window: number of asthma-related hospital admissions,outpatient services use, medication dispensations, number of dispensations of short-acting beta-agonists, inhaled corticosteroids, and combined inhaled corticosteroidsand long-acting beta-agonists, as well as total number of hospital admissions, outpatient services use, medication dispensations, and the Charlson co-morbidity index.Table 1 Characteristics of individuals in the final data setNumber of outcomes per personMean (S D) 1.64 (1.81)[1,2,3,4+] [2804,575,212,261]Entry time(In years since 1/1/1997) 1.5 (2.2)Event time(In years since 1/1/1997) 5.8 (3.0)Event time(Since entry) 4.3 (2.8)Age at entry dateMean (SD) 39.4 (14.1)Age at outcome dateMean (SD) 43.7 (14.2)SexFemale 69.0%Male 31.0%Sadatsafavi et al. Allergy, Asthma & Clinical Immunology 2013, 9:49 Page 4 of 8 were associated with a lower risk of the out-come (RR = 0.78 (95% CI 0.63 - 0.96)).Subgroup analysesResults of the subgroup analysis are provided in Figure 2.No obvious trend could be observed for the RRs acrosssex and age groups, although it appears the RRs forasthma-related or respiratory-related outcomes werehigher among women than men, whereas the non-asthma-related and non-respiratory-related RRs werereciprocally lower among women compared with men.The negative association between non-respiratory-relateddischarges and asthma admissions disappeared in severalsubgroups but persisted among women and individuals35–54 years old.Sensitivity and alternative analysesResults of the sensitivity and alternative analyses are pro-vided in Figure 3. For the most part the overall directionand significance of the associations remained the same,with one exception: the negative association between non-respiratory-related discharges and asthma-related admis-sions disappeared in all sensitivity and alternative analyses.Overall, in 31.6% of case time windows and 29.1% ofcontrol time windows a controller medication was dis-pensed (P < 0.001 for difference). In exposed time windows(time windows with a hospital discharge), this value was39.9% (43.9% when the discharge was asthma-relatedand 35.7% when it was non-asthma-related, all P < 0.001 forthe difference compared with unexposed time windows).However, in the sensitivity analysis that was adjusted forthe use of controller medication, the RR of the exposureonly slightly changed compared with the main analysis[RR = 1.38 (95% CI 1.21 - 1.57)].DiscussionWe used population-based administrative health dataof an entire geographic region to investigate whetherdischarge from hospital alters the short-term risk of ad-mission due to asthma. As expected, we found thatasthma-related and respiratory-related discharges aresignificant predictors of asthma-related readmission, butnon-asthma-related discharges were not associated withthe risk of asthma-related readmissions. Other findings ofthe study remained the same in several sensitivity analyses.Our overall conclusion is that non-respiratory-relatedAny discharge1.40 (1.22-1.59)1.31 (1.12-1.53)1.62 (1.27-2.05)1.74 (1.35-2.26)1.21 (1.02-1.44)1.56 (1.16-2.09)Asthma-related discharge1.99 (1.65-2.39)2.10 (1.66-2.67)1.73 (1.30-2.30)2.43 (1.71-3.44)1.69 (1.33-2.14)2.52 (1.60-3.96)Non-asthma-related discharge0.88 (0.74-1.04)0.78 (0.64-0.95)1.22 (0.85-1.73)1.00 (0.69-1.44)0.79 (0.63-1.00)1.00 (0.70-1.43)Respiratory-related discharge1.89 (1.60-2.22)1.93 (1.57-2.37)1.78 (1.36-2.33)2.41 (1.77-3.29)1.64 (1.33-2.02)2.09 (1.41-3.08)Respiratory-related, non-asthma discharge1.23 (0.90-1.68)1.13 (0.78-1.64)1.49 (0.85-2.60)2.00(0.87 -4.59)1.17 (0.78-1.75)1.01 (0.52-1.94)None-respiratory-related discharge0.78 (0.63-0.96)0.70 (0.56-0.89)1.05 (0.68-1.62)0.83 (0.54-1.30)0.66 (0.50-0.88)1.03 (0.69-1.52)Figure 2 Subgroup analysis*. *All RRs are adjusted for the followingvariables estimated in the 180 days prior to the start of the timewindow: number of asthma-related hospital admissions, outpatientservices use, medication dispensations, number of dispensations ofshort-acting beta-agonists, inhaled corticosteroids, and combinedinhaled corticosteroids and long-acting beta-agonists, as well as totalnumber of hospital admissions, outpatient services use, medicationdispensations, and the Charlson co-morbidity index.Sadatsafavi et al. Allergy, Asthma & Clinical Immunology 2013, 9:49 Page 5 of 8 do not appear to alter the short-term risk ofsevere asthma exacerbation requiring admission, whereasan episode of asthma-related discharge is associated withan elevated rate of readmission. In addition to being achance finding, a reduced risk of asthma-related admissionafter an episode of non-asthma-related discharge mightindicate that not only the underlying risk of asthmaexacerbation is not affected by an episode of inpatientcare, but such a discharge might result in a higherthreshold for readmitting the patient. In addition, aperiod of inpatient care independent of asthma mightprompt a review of asthma management, thus reducingthe risk of asthma exacerbation in the post-dischargeperiod. Evidence for this pattern was observed in ourdata, as dispensation of controller medications weremore likely to occur in time windows with, comparedwith those without, a hospital discharge record.To our knowledge, the association between dischargefrom hospital and the short-term risk of admission dueto asthma has not previously been investigated. Otherinvestigators have assessed the rate of readmission afterdischarge from a previous asthma-related hospitalization,and factors altering such rates [6,18-21]. But these studieshave not attempted to contrast the risk in the post-discharge period with control periods, and thus have notbeen able to show any change in the risk. In addition,many of such studies have been based on longer-termfollow-up periods, and the associations reflect the impactExposure RR (95% CI)*Any discharge1.40 (1.22 - 1.59)1.59 (1.40 - 1.80)1.20 (1.03 - 1.40)1.42 (1.27 - 1.57)1.21 (1.10 - 1.34)Asthma-related discharge 1.99 (1.65 - 2.39)2.32 (1.95 - 2.77)1.53 (1.24 - 1.89)2.01 (1.73 - 2.33)1.66 (1.46 - 1.90)Non-asthma-related discharge 0.88 (0.74 - 1.04)0.95 (0.81 - 1.11)0.92 (0.74 - 1.14)0.94 (0.84 - 1.06)0.90 (0.80 - 1.01)Respiratory-related discharge 1.89 (1.60 - 2.22)2.12 (1.82 - 2.48)1.52 (1.26 - 1.85)1.83 (1.60 - 2.10)1.54 (1.36 - 1.74)Respiratory-related,non-asthma discharge1.23 (0.90 - 1.68)1.22 (0.91 - 1.63)1.46 (0.92 - 2.33)1.06 (0.84 - 1.33)1.02 (0.82 - 1.26)Non-respiratory-related discharge 0.78 (0.63 - 0.96)0.86 (0.71 - 1.03)0.80 (0.63 - 1.03)0.92 (0.80 - 1.05)0.88 (0.77 - 1.00)Figure 3 Results of the sensitivity and subgroup analyses. *All RRs, except for the unadjusted case-cross-over, are adjusted for the followingvariables estimated in the 180 days prior to the start of the time window: number of asthma-related hospital admissions, outpatient services use,medication dispensations, number of dispensations of short-acting beta-agonists, inhaled corticosteroids, and combined inhaled corticosteroidsand long-acting beta-agonists, as well as total number of hospital admissions, outpatient services use, medication dispensations, and the Charlsonco-morbidity index.Sadatsafavi et al. Allergy, Asthma & Clinical Immunology 2013, 9:49 Page 6 of 8 post-discharge care rather than the short-term effect ofprevious hospitalization [6,18].In addition to relying on the data of an entire geo-graphic region that is free from selection bias, the choiceof the case-crossover design gives weight to the validityof our findings. Many factors, not necessarily capturedin measures of asthma severity and co-morbidity, canalter the risk of hospitalization for an individual (e.g.,threshold for hospitalization in the individual’s localhealth setting, availability of hospital beds, to name afew) thus causing a spurious association between theexposure and outcome. As such, we believe the case-crossover design, in which such factor are relativelyconstant within a patient, is a robust design and themain results are less affected by such biases than the resultsof the nested case control study (although the overallresults were similar).The limitation of our study should be acknowledged.The identification of asthma was based on resource userecords. However, the case definition of asthma, used byseveral other investigators, in combination with a recordof asthma-related hospitalization, must have created avery specific sample of asthma patients. Nonetheless, weacknowledge that the diagnostic accuracy of asthma-related hospitalization in the extreme of age groupsmight not be optimal. Further, the risk of asthma-relatedreadmissions might well be affected by the cause of theprevious hospitalization. We decided not to subdivide theexposure any further than asthma-related and respiratory-related conditions as we were concerned multiple-comparison issues would make the interpretation of theresults difficult. Additionally, the risk of asthma-relatedadmission might be a function of events occurring withinthe prior hospitalization, such as whether appropriatecare for asthma was provided at discharge, the dischargemedications for asthma, and so on. The observed patternof medication dispensation indicated that around the timeof discharge, whether asthma-related or not, patients weremore likely to fill prescriptions for asthma controllermedications, but in the sensitivity analysis that adjustedfor this pattern, no major changes in the findings wereobserved. This suggests that dispensation of controllermedications did not play a major role in the observedfindings. Unfortunately, as in many administrative healthdatabases, medication records during inpatient time werenot captured in our data, and in the 30-day time windowindividuals are most likely taking the medications theyreceived during inpatient time and upon discharge; assuch, dispensation records could not have been reliablyinterrogated for such associations.While we confirmed the previous findings of anelevated risk of readmission after an asthma-relatedhospitalization, our study indicates that the risk of anasthma-related hospitalization is not increased afterdischarge from non-asthma-related admission. This isin line with the general belief that the significance ofpost-hospital syndrome is primarily related to certainco-morbidities of chronic diseases in the older population,and is not a general period of increased risk affecting aninherently inflammatory condition such as asthma. Thesefindings can be of relevance from a policy and clinicalperspectives with regard to the managements and recom-mendations patients with asthma receive upon dischargefrom hospital. Lack of association between non-asthma-related admissions with a subsequent risk of an asthma-related hospitalizations means in patients with knownasthma who are hospitalized due to non-asthma reasons,care providers need to focus on other health conditionsthat are known to cause readmission. Future research isrequired to associate the risk of readmission with suchfactors as the level of asthma care during admission,outpatient care immediately after discharge, and provisionof asthma controller medications in the post-dischargeperiod as potentially relevant and modifiable factors deter-mining the risk of asthma-related readmission.Additional fileAdditional file 1: Table S1. List of asthma-related medications.Competing interestsThe authors declare that they have no competing interests.Authors’ contributionMS proposed the overall research question. MS, JMF, and LL conceived thedesign of the study and participated in planning the detailed analysis plan.LL helped with the acquisition of the data. MS performed the analyses andwrote the first version. All authors read and approved the final manuscript.FundingThis study is part of the project “Platform for Outcomes Research andTranslation in Asthma and aLlergy (PORTAL)” funded by AllerGen NationalCenter of Excellence. None of the sponsors played a role in the study design,data analysis, interpretation or publication of the results.Author details1Department of Medicine, Institute for Heart and Lung Health, The Universityof British Columbia, 7th Floor, 828 West 10th Avenue, Research Pavilion,Vancouver V5Z 1 M9, Vancouver, BC, Canada. 2Collaboration for OutcomesResearch and Evaluation, Faculty of Pharmaceutical Sciences, the University ofBritish Columbia, Vancouver, Canada. 3Centre for Health Evaluation andOutcome Sciences, the University of British Columbia, Vancouver, Canada.Received: 25 September 2013 Accepted: 30 November 2013Published: 23 December 2013References1. Krumholz HM: Post-hospital syndrome–an acquired, transient conditionof generalized risk. N Engl J Med 2013, 368(2):100–2.2. Jencks SF, Williams MV, Coleman EA: Rehospitalizations among patients inthe Medicare fee-for-service program. N Engl J Med 2009, 360(14):1418–28.3. Sandberg S, Paton JY, Ahola S, McCann DC, McGuinness D, Hillary CR, et al:The role of acute and chronic stress in asthma attacks in children.Lancet 2000, 356(9234):982–7.4. Greene LS: Asthma and oxidant stress: nutritional, environmental, andgenetic risk factors. J Am Coll Nutr 1995, 14(4):317–24.Sadatsafavi et al. Allergy, Asthma & Clinical Immunology 2013, 9:49 Page 7 of 8 Ben-Noun L: Drug-induced respiratory disorders: incidence, preventionand management. Drug Saf Int J Med Toxicol Drug Exp 2000, 23(2):143–64.6. Minkovitz CS, Andrews JS, Serwint JR: Rehospitalization of children withasthma. Arch Pediatr Adolesc Med 1999, 153(7):727–30.7. British Columbia Ministry of Health: Consolidation File (MSP Registration &Premium Billing). V2. [Internet]. Population Data BC; 2012. British Columbia Ministry of Health: Discharge Abstract Database (HospitalSeparations). V2. [Internet]. Population Data BC; 2012. British Columbia Ministry of Health: Medical Services Plan (MSP) PaymentInformation File. V2. Data Extract. [Internet]. Population Data BC; 2012. British Columbia Ministry of Health: PharmaNet. V2. Data Extract. DataStewardship Committee [Internet]. British Columbia Ministry of Health. Ministry of Health: PharmaNet [Internet]. [cited 2012 Mar 29]. Maclure M: The case-crossover design: a method for studying transienteffects on the risk of acute events. Am J Epidemiol 1991, 133(2):144–53.13. Delaney Chris JA, Suissa S: The case-crossover study design in pharmacoe-pidemiology. Stat Methods Med Res 2009, 18(1):53–65.14. Canadian Institute for Health Information: CIHI Data Quality Study of the2005–2006 Discharge Abstract Database [Internet]. Ottawa, Ontario; 2009. Schlesselman JJ, Stolley PD: Case–control studies: design, conduct, analysis.Oxford University Press; 1982.16. Charlson M, Szatrowski TP, Peterson J, Gold J: Validation of a combinedcomorbidity index. J Clin Epidemiol 1994, 47(11):1245–51.17. Kelly PJ, Lim LL: Survival analysis for recurrent event data: an applicationto childhood infectious diseases. Stat Med 2000, 19(1):13–33.18. Bisgaard H, Møller H: Changes in risk of hospital readmission amongasthmatic children in Denmark, 1978–93. BMJ 1999, 319(7204):229–30.19. Ather S, Chung KD, Gregory P, Demissie K: The association betweenhospital readmission and insurance provider among adults with asthma.J Asthma Off J Assoc Care Asthma 2004, 41(7):709–13.20. Reznik M, Hailpern SM, Ozuah PO: Predictors of early hospital readmissionfor asthma among inner-city children. J Asthma Off J Assoc Care Asthma2006, 43(1):37–40.21. McCaul KA, Wakefield MA, Roder DM, Ruffin RE, Heard AR, Alpers JH, et al:Trends in hospital readmission for asthma: has the Australian NationalAsthma Campaign had an effect? Med J Aust 2000, 172(2):62–6.doi:10.1186/1710-1492-9-49Cite this article as: Sadatsafavi et al.: Post-hospital syndrome in adultswith asthma: a case-crossover study. Allergy, Asthma & ClinicalImmunology 2013 9:49.Submit your next manuscript to BioMed Centraland take full advantage of: • Convenient online submission• Thorough peer review• No space constraints or color figure charges• Immediate publication on acceptance• Inclusion in PubMed, CAS, Scopus and Google Scholar• Research which is freely available for redistributionSubmit your manuscript at et al. Allergy, Asthma & Clinical Immunology 2013, 9:49 Page 8 of 8


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items