A Systematic Review of the Psychological Risk Factors Associated with Near Fatal Asthma or Fatal AsthmaAlvarez G.G.a · FitzGerald J.M.b
aOttawa Health Research Institute of the University of Ottawa, and Respirology Division, Ottawa Hospital, Ottawa, and bCenter for Clinical Epidemiology and Evaluation of the University of British Columbia, and Respirology Division, Vancouver General Hospital, Vancouver, Canada Corresponding Author
Psychological factors such as anxiety, depressive disorders and/or personality disorders may predispose patients with asthma to near fatal asthma (NFA) or fatal asthma (FA). NFA is defined by an asthma exacerbation resulting in respiratory arrest requiring mechanical ventilation or a pCO2 ≥45 mm Hg. Most studies have used the case-control study design. Several studies analyzing the effects of psychological factors on the risk of NFA or FA have shown conflicting results. We reviewed all of the literature found by the systematic search done of psychological factors on the risk NFA or FA. A MEDLINE search identified 423 articles between 1960 and March 2006. Seven case-controlled studies were identified following strict applications of the inclusion and exclusion criteria. Due to the significant heterogeneity in the measurement of the psychological factors, a summary statistic was not calculated. The trial characteristics were tabulated and qualitative trends were observed to explain the heterogeneity in the results of the studies. Recommendations on future studies in the field are outlined in detail. Following a systematic assessment of all published studies, we cannot conclude that psychological factors increase the risk of NFA and FA.
Copyright © 2006 S. Karger AG, Basel
Approximately 300 million people suffer from asthma in the world . It is estimated that asthma accounts for 1 in every 250 deaths worldwide . There are more than 5,000 deaths reported annually in the USA and 100,000 deaths estimated yearly throughout the world [1, 2].
Psychological factors such as anxiety disorders, depressive disorders, coping mechanisms and abnormal family dynamics of asthmatics may be associated with a higher risk of near fatal asthma (NFA) or fatal asthma (FA). Several hypotheses may explain the link between psychological factors and NFA or FA.
Psychological factors may predispose patients directly to an episode of NFA or FA. Emotional factors can be responsible for aggravating or maintaining asthma . Emotionally induced wheezing has been reported in several studies . Psychological factors may also be confused with worsening asthma resulting in a vicious circle. For example, worsening of anxiety may be confused with worsening asthma and high levels of β-agonist may be inhaled which then results in further worsening of the anxiety . Breathlessness is a common somatic accompaniment of anxiety . Furthermore, the converse may be true, asthma may increase the risk of developing an anxiety disorder . Psychological risk factors may predispose asthmatics to improper asthma management which results in poor asthma control predisposing them to NFA or FA. Ability to comply with asthma medications is influenced by the psychological interaction between asthmatic children and their families . Many studies done in asthmatic children have identified family dynamics as an important risk factor for poor asthma control. Disturbed family cohesion had a negative correlation with peak expiratory flow rates or hyperresponsiveness in asthmatic children . Family interaction in families of asthmatic children may tend to be rigid, overprotecting and may avoid conflict . Family dynamics are sometimes altered negatively following an NFA episode . Some of the psychological defenses or coping styles used by asthmatics include a ‘hopeless dependency’ on physicians and hospital services or they may have inappropriate excessive independence where they are in denial of their illness . Increasing denial is an adaptive mechanism to cope with chronic illnesses yet retain a normal social agenda . Denial of asthma severity could certainly result in patients not following appropriate action plans or a delay in seeking medical attention which could result in a NFA/FA episode (fig. 1).
|Fig. 1. Hypothetical links between psychological factors and NFA or FA.|
Identification of high-risk patients according to their specific psychological profile could allow healthcare teams to intervene earlier and prevent NFA or FA from occurring. Severe asthmatics could get a psychological assessment to determine if any pathology exists. Psychiatric therapy and treatment could become part of the intervention strategy.
We present a systematic review of all the available literature on the subject with consideration of possible sources of heterogeneity between studies and conclude by a series of recommendations regarding future studies. We qualitatively reviewed the existing studies noting methodological shortcomings in hopes that such a critique may inform further research studies.
The exposure is defined as a psychological abnormality. The outcome is NFA defined as an asthma exacerbation resulting in respiratory arrest requiring mechanical ventilation or a pCO2 >45 mm Hg [11, 12]. NFA and subjects who had FA were presumed to be part of the same pathophysiological spectrum [13, 14]. NFA is considered by many to be a good surrogate marker of FA [11,15,16,17,18,19]. This is true because if patients with NFA did not receive medical attention they would inevitably become a fatal case of asthma . Furthermore, these two groups share many characteristics thus allowing NFA episodes to be a proxy measure for FA . The population is defined as persons diagnosed with asthma as defined by the ATS (American Thoracic Society) standards .
A literature search in MEDLINE (1960–March 2006 inclusive) was conducted by one of the investigators (G.G.A.). The following search strategy was used in PubMed: (‘psychology’ [Subheading] OR ‘Mood Disorders’ [MeSH] OR (‘Dependency (Psychology)’ [MeSH] OR ‘Codependency (Psychology)’ [MeSH])) AND (‘Fatal Outcome’ [MeSH] AND ‘Death’ [MeSH] OR ‘fatal outcome’ [All Fields] OR ‘life threatening’ [Text Word] OR ‘near fatal’ [All Fields]) AND (‘asthma’ [MeSH Terms] OR asthma [Text Word]). We also searched citations from published reviews, the original articles, expert opinion and our own extensive bibliography. The literature search was cross-checked by a university librarian at The Harvard School of Public Health to ensure reproducibility and that no other new citations were identified.
The inclusion criteria included: (1) NFA defined as an asthma exacerbation resulting in respiratory arrest requiring mechanical ventilation or a pCO2 ≥45 mm Hg or asthma resulting in death (FA); (2) the number of cases (NFA and/or FA) and controls reported; (3) explicit reporting of risk factors; (4) adult and pediatric cases, and (5) all study types. The exclusion criteria included: (1) case series since they do not contain controls; (2) studies that included COPD patients; (3) studies that contained only patients over the age of 65 years were excluded to minimize COPD overlap, and (4) studies in a language other than English. One of us (G.G.A.) identified all potentially relevant articles from the abstracts yielded via the comprehensive search strategy. Any study that used human subjects that included the terms NFA or FA or SLTA (severe life-threatening asthma) or asthma and the word death in the abstract was examined in detail. Using a standardized data extraction sheet that was developed for the purpose of the study, each of the identified potentially relevant articles had the data abstracted by two observers independently (G.G.A., J.M.F.). Data extraction sheets were filled out for each article and then entered into an Excel spreadsheet. Differences were resolved by consensus. Appendix 1 lists the reasons for exclusion.
The data in this systematic review were not summarized in a quantitative manner due to the following three reasons: (1) Significant heterogeneity was detected in the measurement of the exposure. Three different categories were identified: (a) psychiatric illness assessed by reviewing medical records; (b) psychometric tests measuring personality, anxiety and depression, and (c) family members were interviewed by proxy for those patients that endured FA. (2) Significant heterogeneity was detected in the measurement of outcome as a result of different definitions for NFA. (3) Lastly, several studies did not provide enough data to calculate weighted mean differences since only p values were provided. We present a systematic review that will qualitatively focus on key components of design and differences found rather than quantitative aggregate scores. A formal meta-analysis was not done since summation of individual trials would likely have resulted in misleading conclusions. We followed the MOOSE Group proposal for reporting observational studies in this systematic review .
We identified 423 potentially relevant articles from several comprehensive searches. Review of the abstracts of these articles yielded 47 potentially relevant articles that met the study inclusion criteria. Of these, we excluded 40 articles (Appendix 1); 18 were excluded because they studied risk factors for NFA and FA other than psychological factors; 6 articles were case series and did not contain controls; 3 compared NFA to FA; 3 studied severe asthmatics but not NFA or FA; 1 included COPD patients; 1 published the same results twice; 8 did not meet the NFA definition in the inclusion criteria, and 1 was an editorial. Seven case-controlled studies met the inclusion and exclusion criteria. No other observational type studies were discovered (table 1).
|Table 1. Summary of case-control studies of psychological factors associated with NFA/FA|
Some studies assessed psychological factors (the exposure) by examining the medical records [25,26,27]. The limitation in this method is that patient’s medical records reflect their medical encounters that may or may not have been associated with their psychological profile. Furthermore, even if the psychological profile was assessed, if the healthcare team did not write it in the chart the exposure could not be detected. An interview with the patient is a more precise manner in obtaining the data. Some studies interviewed patients but only asked them yes or no questions about anxiety or depression or personality disorders. The limitation in this method is its tendency to result in bias. Patients tend to interpret conditions such as anxiety differently and their assessment of themselves may not be accurate. A more robust manner to assess a patient’s psychological profile is to actually measure it with standardized psychometric questionnaires. Several of the studies used such scales or scoring methods depending on the psychological factor of interest. The MMPI score is a personality inventory. The HAD score measures the level of anxiety and depression, specific for the distress in physically ill subjects, and state and anxiety assessed on visual analogue scale. The Zung and Hamilton scores measure anxiety and depression. Future studies should delineate clearly what psychological parameters will be studied and use validated scores such as these to obtain more objective and reproducible measures of psychological problems. Studies using a standardized psychometric measurement will produce more reproducible, comparable and accurate reflection of psychopathology. Although we did not develop a quality score because they may lack validity , studies that used standardized scales were considered to be of higher quality. Another important point brought out by the studies is the timing of the interview. The timing of the interview should ideally occur prior to the discharge of the patient following the NFA episode. Some studies collected data many months after the episode had occurred which could introduce recall bias or the patient’s psychological factor may have improved since the event.
Different studies used various definitions for NFA. Several studies recommend the definition of intubation resulting in mechanical ventilation or a pCO2 >45 mm Hg during a hospitalization for asthma. Some studies used pCO2 >50 mm Hg which would have missed asthmatics that were between 45 and 49 mm Hg. Several studies included a pH of >7.2 that would imply an elevated pCO2. Other studies included the term alteration or loss of consciousness [14, 22, 23, 26]. Using this term (e.g., respiratory failure OR alteration or loss of consciousness) introduces a problem since patients may have been included for alteration or loss of consciousness alone and not for NFA specifically . One study  used FA as the outcome, however the ability of the investigators to adequately extract information on the patients’ psychological profiles was limited. The investigators had to ask family members to answer questions by proxy. This practice is not ideal since it introduces observer bias since family members may or may not be aware of the psychological problems faced by the patient. Instead we propose using NFA as a surrogate or proxy for FA as interviews with patients with NFA can capture the psychological abnormalities experienced by the patients with more precision. Furthermore, NFA is known to be a good proxy for FA [11,15,16,17,18,19]. Amongst studies that used a standard scale to measure psychological profiles [14,22,23,24] and showed an association between psychological factors and NFA or FA, a more precise definition of NFA was used. Amongst studies that used a standard scale and did not find an association between the outcome and exposure, a less precise definition of NFA was used [22, 23]. The studies that did not use a precise definition may have diluted the effect of the exposure amongst individuals that may not have had NFA (table 2).
|Table 2. NFA and FA outcome definitions|
Controls were fairly heterogeneous between studies. Most matched on age, gender and date of event, but many matched on different variables including race, severity, baseline FEV1. Most of the studies compared cases to community controls, one of them  compared cases to hospital and community controls separately and another  compared cases to only hospital controls.
The evidence for psychological risk factors associated with NFA/FA is equivocal. Three studies by Yellowlees et al. , Barboni et al.  and Ernst et al.  were positive and three by Boulet et al. , Strunk et al.  and Rea et al.  were negative. The seventh paper by Kolbe et al.  found that psychopathology existed more frequently amongst all admitted patients including those with NFA when compared to community controls, however they did not find a difference between NFA/FA and admitted patients with acute asthma with no NFA.
A systematic review of all published studies was not able to conclude that psychological factors increase the risk of NFA and FA. Due to the significant heterogeneity in the measurement of the psychological factors, a summary statistic was not calculated. Instead, the trial characteristics were tabulated and qualitative trends were observed to explain the heterogeneity in the results of the studies.
Two major factors that can significantly affect the quality, validity and bias of case-control studies are selection of the control group and exposure status . We identified the manner in which each control group was chosen, the matching factor and the control type (community vs. hospital) in each study (table 3) in order to compare them. Choice of controls is paramount in these studies since different results were obtained when different controls were used. For instance, Kolbe et al.  showed that psychological factors played an important role in the risk of being admitted with asthma but did not show a difference between those that were admitted to hospital and those that had NFA or FA. The conclusion of this study was that psychological factors increased the risk of admission but did not necessarily increase the risk of a NFA or FA episode. Controls should be matched for index date, age, gender and explicitly categorized as a community or hospital control. It may be more instructive to use community controls to better illustrate the effect of psychological risk factors on asthmatics.
|Table 3. Details of controls in each study|
The exposure to psychological risk factors needs to be measured with an objective validated psychometric test. Although validated psychometric tests exist, not all studies used the same tests. Studies using such tests were generally of higher quality than those that did not. The testing or questionnaire must be given in a reasonable period of time preferably within close proximity to the NFA event to minimize bias. A review of the medical records is not ideal since it is fraught with recall bias as well as recording bias. Cases need a consistent definition of NFA which should include: (1) mechanical ventilation; (2) pCO2 >45 mm Hg, and (3) cardiopulmonary arrest should not on its own include altered level of consciousness as it may introduce bias in the definition since patients with level of consciousness alone may in fact not have NFA. The American and Canadian Thoracic Societies should come up with a standardized definition of NFA.
Asthma severity is an issue in many of the studies in our systematic review in that the cases may have had more severe disease than the controls. Most patients were matched for age and gender but few were matched for asthma severity. Only Strunk et al.  attempted to control for severity, however it is not clear what method was used. Ernst et al.  attempted to further adjust for asthma severity in a study of the association of inhaled β-agonist use and NFA . They showed that after attempting to adjust for confounding by severity, the overall effect of the use of a β-agonist and the risk of NFA was unchanged, suggesting that confounding by severity was less of a problem when studying NFA. Furthermore, NFA or FA may occur to asthmatic patients belonging to the whole spectrum of disease, from mild to severe asthma  providing further evidence that asthma severity may not be an important confounder in these studies. It is not only confined to patients with severe disease but can affect patients with mild disease as well . It is very difficult to categorize patients with asthma in a manner which is reproducible. In the 2005 Global Initiative on Asthma (GINA), asthma severity is categorized into four groups using an evidenced-based approach: intermittent, mild persistent, moderate persistent, and severe. Unfortunately, this has not translated into a clean tool to be used by epidemiologists around the world because it still remains a challenge to categorize the severity of asthma in a way that will be easily reproduced given the dynamic and complex nature of the disease. The document states that any patients at any level of severity – even intermittent asthma – can have severe attacks.
The quality of the systematic review was enhanced by independent assessment of trials using a standardized data extraction sheet that was used independently by two investigators. An independent librarian (A.G.) tested the reproducibility of the literature search. Bias may have been introduced since the non-published data and non-English papers were not included in the systematic review. Publication bias was not assessed given that methodological differences and heterogeneity were found in the analysis of the trials.
A randomized controlled study would be impossible to carry out to answer the clinical question being posed since patients cannot be given psychological abnormalities. Furthermore, a prospective cohort study, although not impossible, would be challenging and costly since NFA or FA is a rare event and many person-years of patient follow-up would have to be accrued to discern the issue. A well-designed large case-controlled study could further research in this field. Another design that has not been used to study psychological risk factors in NFA is the case cross-over design . The case cross-over design is used to identify transient risks that are associated with intermittent exposures. In order to find out if individuals were experiencing a psychological event or life stressor prior to the NFA episode, this design would be useful and cost-efficient. In this type of design the comparison is between different exposure periods (hazard period and non-hazard period) within individuals instead of between individuals. Incident cases of NFA could be identified by a national or provincial registry. Cases could be identified by intensive care unit medical admission records identifying anyone who was mechanically ventilated or had a pCO2 >45 mm Hg as a result of an asthma attack. The cases would then be interviewed a week after the episode to ensure accurate recall of the events preceding the NFA episode. The week before the NFA episode can be used as the hazard period for the matched analysis. Exposure to a psychological risk factor or stressor like a life event could be measured using a validated scale proposed by Tennant and Andrews  which allows the significance of life events to be scaled. Possible problems with this design include the fact many patients decompensate psychologically following a NFA episode  and it may be difficult to ascertain how the patient was feeling prior to the event, however if the scale is applied promptly after the event this effect may be minimized.
Recommendations include: (1) Use of standardized psychometric tests to assess exposure. (2) Interviews should be done within a close time frame from the NFA event. (3) Use of a standardized definition of NFA. (4) Studies should not include COPD patients as this is a different pathophysiological process. Excluding patients older than age 35 would essentially eliminate the risk of disease misclassification and allow for a firm diagnosis of obstructive lung disease due to asthma . (5) Consider using the case cross-over study design.
In summary: The systematic review presented is intended to generate discussion using the current literature and explain some of the heterogeneity observed amongst the current case-control studies. A formal meta-analysis was not done therefore a summary static was not calculated since it would have been misleading given the heterogeneity of the studies. There remains much to be studied in this field, however at this time we cannot definitively conclude that psychological disorders increase the risk of NFA and FA.
The author would like to thank Anna Getselman, Assistant Director for Reference and Education Services at the Countway Library of Medicine, for cross-checking our search findings and to Emily Levitan, PhD, candidate and Teaching Assistant at Harvard University for her advice on the project.
Dr. Gonzalo G. Alvarez
Ottawa Health Research Institute of the University of Ottawa, and
Respirology Division, Ottawa Hospital, General Campus
501 Smyth Rd, Box 211, Rm 1835B, K1H 8L6 Ottawa (Canada)
Tel. +1 613 737 8899/ext. 78198, Fax +1 613 739 6266, E-Mail firstname.lastname@example.org
Received: June 28, 2006
Accepted after revision: September 28, 2006
Published online: November 30, 2006
Number of Print Pages : 9
Number of Figures : 1, Number of Tables : 3, Number of References : 69
Respiration (International Journal of Thoracic Medicine)
Vol. 74, No. 2, Year 2007 (Cover Date: February 2007)
Journal Editor: Bolliger, C.T. (Cape Town)
ISSN: 0025–7931 (print), 1423–0356 (Online)
For additional information: http://www.karger.com/RES