Readmission rates following heart failure: a scoping review of sex and gender based considerations.

Background Although hospital readmission for heart failure (HF) is an issue for both men and women, little is known about differences in readmission rates by sex. Consequently, strategies to optimize readmission reduction programs and care strategies for women and men remain unclear. Our study aims were: (1) to identify studies examining readmission rates according to sex, and (2) to provide a qualitative overview of possible considerations for the impact of sex or gender. Methods We conducted a scoping review using the Arksey and O’Malley framework to include full text articles published between 2002 and 2017 drawn from multiple databases (MEDLINE, EMBASE), grey literature (i.e. National Technical information, Duck Duck Go), and expert consultation. Eligible articles included an index heart failure episode, readmission rates, and sex/gender-based analysis. Results The search generated 5887 articles, of which 746 underwent full abstract text consideration for eligibility. Of 164 eligible articles, 34 studies addressed the primary outcome, 103 studies considered sex differences as a secondary outcome and 25 studies stratified data for sex. Good inter-rater agreement was reached: 83% title/abstract; 88% full text; kappa: 0.69 (95%CI: 0.53–0.85). Twelve of 34 studies reported higher heart failure readmission rates for men and six studies reported higher heart failure readmission rates for women. Using non composite endpoints, five studies reported higher HF readmission rates for men compared to three studies reporting higher HF readmission rates for women. Overall, there was heterogeneity between studies when examined by sex, but one observation emerged that was related to the timing of readmissions. Readmission rates for men were higher when follow-up duration was longer than 1 year. Women were more likely to experience higher readmission rates than men when time to event was less than 1 year. Conclusions Future studies should consider different time horizons in their designs and avoid the use of composite measures, such as readmission rates combined with mortality, which are highly skewed by sex. Co-interventions and targeted post-discharge approaches with attention to sex would be of benefit to the HF patient population.


Brief summary
There has been increased attention on reducing hospital readmission rates. In this scoping review, we found notable variation exists in studies with sex stratified analyses for HF patients readmitted to hospital. Few studies were prospective in design, and results could be conflated by reporting of composite measures and time to event. Co-interventions and targeted post-discharge approaches need to be revisited for men and women.

Background
The rise in hospital readmissions is a global concern, placing considerable burden on patients, treatment costs, and hospital resources [1]. In the United States, the 30-day readmission rate for those with heart failure (HF) increased from 17 to 20% between the years 1993 and 2006 [2]. The number of people living with HF is increasing, and age/sex-standardized prevalence of the condition has been relatively stable over time [3]. Current patterns of hospital readmission are often associated with organizational factors, such as a length of stay, and clinical factors, such as age and comorbidities [4]. Improved quality of care at patient intake is also associated with a lower probability of readmission [5,6]. Strategies to reduce readmission rates have shifted from hospitalbased to more patient-centred strategies, such as telemonitoring, which may benefit patients by facilitating their access to health care services [7,8]. Public reporting and financial incentives have been trialed by governments with the intent to reduce hospitalization rates [9,10].
Heart failure is a growing problem with similar prevalence in men and women [11][12][13]. However, information on processes, quality of care, health status outcomes, or other patient care experiences has not been explored in the context of sex and gender. Heart failure with preserved ejection fraction (HFpEF) is more prevalent in women [13]. The lifetime risk of heart failure is 15% for women and 11% for men for those without a history of myocardial infarction at age 40 [8]. A few authors have focused on sex differences in heart failure [14][15][16][17] but no study to date has examined this in relation to readmission rates. This article draws on a scoping review protocol to better understand current patterns of readmission and the interpretation of observed patterns in relation to sex. Our aim was to examine studies reporting a higher heart failure readmission rate for either women or men, and to provide a qualitative overview of the possible considerations for the impact of sex and gender on this outcome.

Study protocol
A scoping review protocol was developed using the methodological framework proposed by Arksey and O'Malley [18]. In addition to the aforementioned methods, we also used the Joanna Briggs Institute Methodology for scoping reviews [19] and further refined the process using recommendations put forth by Levac et al. [20] This scoping review is related to the COACH trial (clinicaltrials.gov NCT02674438-Last accessed October 2, 2017). A consultation of experts was included to inform the search for additional articles of interest.

Information sources and literature search
The search strategy was developed in consultation with the research team and was peer-reviewed by an expert librarian using the PRESS peer-review of electronic search strategy checklist [21] (Appendix). Multiple databases were searched, from 2002 up to May 29, 2017, including MEDLINE, EMBASE, PubMed, CINAHL, and Web of Science. Experts were then consulted up until October 16, 2017 for additional articles of interest (Table 1). Keywords and combinations of mesh terms were used to narrow the search strategy: "heart failure" OR "systolic dysfunction" OR "diastolic dysfunction" OR "heart ventricular failure" OR "left ventricular dysfunction" OR "cardiac failure" OR "cardiac decompensation" OR "heart decompensation" was combined with Readmission* or readmit* or rehospital* and sex, gender or male/female. Additional filters narrowed the results to those papers published in the past 15 years and limited to humans. The database search was supplemented by a manual search of related references in the literature drawn from the Sex and Gender-Specific Medicine (SGSM) PubMed database (Texas Tech University).

Eligibility criteria
We included reports, both published and unpublished, that were primary research and in English. Further consideration for inclusion was whether the results disaggregated for sex or gender in readmission. Articles were excluded if they did not include HF as a first episode, enrolled exclusively a pediatric population (or those less than 18 years of age), was not full text, or published more than 15 years prior to the time the search was performed.

Study selection process
The screening criteria were established a priori and calibrated amongst the team (DSL, RM, PR, AHK, CP) with a pilot-test by screening title and abstract on a random sample of 220 articles and 100 full text articles. After an acceptable inter-rater agreement was established at 80%, pairs of reviewers (CP, AHK) screened the pool of potentially eligible articles.

Data items and data abstraction process
We employed a 'descriptive-analytical' method, a narrative study design, which applies a common framework to all the included studies [20]. A sample of articles was read, and a data abstraction chart was developed. Charting is a technique for synthesizing and interpreting qualitative data by sifting and sorting materials by key issues and themes [19,22]. We charted data into categories of study design, objectives (primary, other), study population, setting, sample estimation, endpoints, and statistical methods for sex group differences. Charting was stratified by sex for sample size, age, sample lost to follow up, event-free survival, mortality rate, all-cause and heart failure readmission rates, non-cardiovascular readmission rates, and any other potentially important findings. Length of hospital stay, time to event, and hazard ratios for readmission were also extracted. The chart was calibrated by the team (WW, RM, PR, DL, AHK, CP) on a sample of 12 included articles (11 from the sex and gender-specific medicine PubMed database and 1 article from the grey literature). Two reviewers used a standard form to decide which resources contained a primary objective evaluating sex differences in patients with heart failure following readmission.
When more than one outcome was reported in an article, we used a hierarchical selection process: (a) authors' explicit declaration of primary objective, (b) the outcome used to calculate the sample size, (c) authors' attribution of importance to the outcome in their description of the results, or (d) the outcome that appeared first in the methods section [23]. We used the category "Other" to classify resources that looked at patient populations presenting with a first episode of heart failure, if sex differences following readmission was specified as a secondary outcome measure or other (Fig. 1). Since studies with composite outcomes can mask associations that might be directionally opposite for death versus readmission, we determined whether composite outcomes were used in combination with mortality. We used the category "Background" to classify resources that contained information about readmission rates for men and women, but the authors did not perform sex-stratified analyses. Disagreements were resolved by discussion and other authors were involved to assist with ambiguous cases.

Definition of sex and gender
Although the number of publications that are relevant to sex/gender in the field of HF has grown over the years (Table 1), we found that there is still much confusion over terminology. For consistency, in this review "sex" will refer to a set of biological attributes in human participants that are associated with physical and physiological features [24]. The term "gender" will refer to socially constructed roles, relationships, behaviours, and powers as defined by the Government of Canada [24]. However, the data charted are reported per verbatim as used by the authors of the full text resources. Retrospective n = 25 Year

Article types
The electronic database search generated 5887 resources. Six additional potential records were identified by a hand search of references from the included articles, the grey literature, and by consulting a group of experts in the field. After duplicates were removed and resources were screened for eligibility, 36 sources remained, comprising 34 journal articles and 2 dissertations (Fig. 1). Good inter-rater agreement for a pair of reviewers was reached at each stage of the eligibility assessment (83% title/abstract; 88% full text; kappa = 0.  Table 1). Twenty-five of the studies were retrospective and 11 were prospective in study design. Most of the studies were from North America including four from Canada. One hundred three sources were classified as "Other" and 25 sources were considered "Background".

Clinical aims
A broad range of research questions was addressed. Twenty-one study objectives (34%) focused on predictors, risk or prognostic factors, 15 (24%) presented a query into the clinical profiles or etiology of patients readmitted with an index diagnosis of heart failure. Sex differences were explicitly stated only in 12 (19% study objectives). Environmental or external factors such as timing or setting were the least frequently addressed [objectives: 1 (2%) and 2 (3%)], respectively.

Definition of heart failure
Notably for defining heart failure, none of the included studies considered sex differences in HF presentation or sex-specific characteristics. A large proportion of studies in this scoping review defined HF using the World Health Organization's International Classification of Diseases code: 14 (35%) studies. The remaining 6 (16%) studies used the Framingham study clinical criteria, 4 (11%) studies, adopted the criteria defined by the European Society of Cardiology, 7 (18%) papers defined HF with clinical examinations, 3 (7%) studies, used specific regional hospital admission codes or registry database codes, in 1 (2%) study, HF was defined as impaired cardiac output, and in 3 (7%) studies, HF was not defined. In addition, 4 (10%) studies reported using multiple diagnostic criteria. HF definitions are summarized in Table 2.

Outcomes reported
Some studies reported readmission outcomes separately while others examined the composite of death or readmission to hospital. Six studies reported that women had higher rates of readmission or experienced the composite outcome more frequently than men (Table 3). Among these 6 studies, 4 had a mean follow-up duration of 3 months or less. Amongst studies in which non-composite outcomes were used, we found 3 that reported higher heart failure readmission rates in women when compared to men [25,26,38] (Table 4). One study reported cardiovascular event-free survival (as defined by survival free from HF admission, acute myocardial infarction, coronary revascularization, valvular surgery, or heart transplantation) and found significantly lower risk in women [38]. Twelve studies reported higher readmission or composite outcomes in men. Among these, 6 studies demonstrated higher all-cause readmission rates [27,29,30,40,53,58] (Table 5) and 4 studies reported higher hazard ratios for composite events [28,31,39,50] (Table 6). In a prospective study, Nieminen et al. [53] reported men having higher rates of all-cause readmission at 1 year compared to a 3-month endpoint. In a Canadian retrospective study, Howlett et al. [28] reported that men were more likely to die than women following rehospitalization at 12 months using the composite measure of death and/or hospital readmission. In total, 5 of 12 studies finding higher readmission rates (not combined with other endpoints) in men examined follow-up durations of 12 months or more.
Eighteen studies reported no sex differences in readmission rates between men and women ( Table 7).

Non-HF cardiovascular causes of readmission
The observed sex differences in readmission rates could be potentially explained by several factors including patient characteristics and differences in processes of care. Vader [46] reported risk factors for post-discharge readmission or death in patients treated for acute heart failure, including male sex, non-use of angiotensin-converting enzyme inhibitors (ACEI) or angiotensin receptor blockers (ARB), lower baseline sodium, non-white race, lower systolic blood pressure at discharge or day 7, increased length of stay, and depression. In addition, both Vader [46] and Howie-Esquivel [42] reported a two-fold increase in cardiac readmission rates for women whose ethnicity was Caucasian. Older age and the comorbidities that are associated with aging were reported to be risk factors for readmission [27,30]. Comorbidities including myocardial infarction, chronic kidney disease, cancer, and diabetes mellitus were cited as predictors of increased readmission risk [27].
Some co-variates such as diabetes, valvular heart disease, blood pressure, and atrial fibrillation were sex-sensitive, and could potentially have a role in differential risk of readmissions in women and men. Diabetes and anemia were significantly more frequent in women [53]. Women more often presented with atrial arrhythmia and atrial fibrillation, and men with ventricular arrhythmias [53]. Women were more likely to present with valvular heart disease, hypertension, and preserved left ventricular function, but also less likely to be diabetic or smokers [52,53]. The presence of coronary disease and an ischemic etiology may modify HF outcomes in women and men. Specifically, Mullens [54] reported better survival rates for women with a nonischemic cause, whereas there was a trend toward worse survival in those with ischemic heart failure.
The effects of ejection fraction on sex-specific readmission rates are variable. Alla [49] reported that males had a relative risk of hospitalization for heart failure that was greater than females when the ejection fraction was reduced but not when preserved. The authors' survival analysis showed an advantage for females with HF and reduced ejection fraction [49]. In a retrospective study, Goncalves et al. [39] found that males with preserved and reduced left ventricular systolic function were more likely to die and/or be re-admitted within 6 months.
Howlett [28] found that women were more likely to receive usual care rather than specialized heart management from a clinic (52% vs. 37%). Nieminen [53] observed higher rates of readmission in men even when both sexes were treated as frequently and as long in the different types of wards (internal medicine, cardiology, intensive care unit and cardiac care unit). The initial place of presentation for both women and men was the emergency department followed by admission to the general ward [36].
There was some indication that whether women were alone, supported socially by a partner, or working could impact readmission rates. Niemminen [39] discussed that readmitted women were more often retired, living alone or in special accommodation.

Differences in interventions
Nieminen [39] described that women underwent significantly fewer invasive procedures. However, Chang [36] found no sex differences in the use of critical life-saving procedures such as renal replacement therapy, mechanical ventilation, defibrillation or cardiopulmonary resuscitation. Zsilinszka [44] also did not note any sex differences in use of dialysis or mechanical ventilation. It appeared that where differences in procedures were observed, there was some clinical discretion involved and they were not necessarily immediately critical for sustaining life. For example, invasive procedures such as coronary angiography were performed less often in women than men (1.4% vs. 2.8%, p < 0.001) [36]. Sheppard [32] also found that women underwent fewer non-invasive assessments of  [45]; Lee (2004) [37] Confirmed by a cardiologist, using standard Framingham criteria or exacerbation of a previously documented HF; Framingham criteria (including symptoms, physical examination, chest x-ray and echocardiographic findings) Vader (2016) [46] The presence of more than one symptom (dyspnea, orthopnea, edema) and 1 sign (rales on auscultation, peripheral edema, ascites, or pulmonary vascular congestion on chest radiography) in DOSE-AHF or ROSE-HF or admission to the hospital with a primary diagnosis in CARRESS HF Gevaert (2014) [47] Preserved left ventricular fraction was defined as left ventricular ejection fraction greater than or equal to 50% and heart failure with reduced ejection fraction was defined as less than 50% Chang (2014)         Males, and lower survival Table 6 Studies reporting males with significantly higher in all cause re-admissions, death and/or hospital readmission ratios. Heart failure readmission, cardiovascular readmission and mortality/even free survival hazard ratios are also reported. Studies reported by primary author and year of publication and sorted alphabetically   More studies, in general, reported that women were less likely to receive evidence-based therapies than men. Men were more often prescribed β-blockers, vasodilators, and antiplatelet agents than women and were administered higher mean doses than women [38]. Women were also less frequently prescribed ACE inhibitors [34]. Chang [48] noted that women were less likely to receive evidence-based therapies upon discharge compared to men. In contrast to the above, Zsilinskza [52] did not note any sex differences in HF therapies, such as administration of diuretics or vasodilators. A potential reason for the similar rates of some drug classes in women is that the higher prevalence of hypertension in women may have overlapping indications for HF. Therefore, increased betablocker therapy in women, for example, may reflect management of hypertension and HF with preserved ejection fraction [34].

Discussion
Findings from this scoping review confirm that while most of the studies that included sex-based analyses showed no differences in readmission rates following HF hospitalization, several factors still point to the need of targeted sex-based management for at-risk populations. The rise in hospital readmissions is a global concern and is often used as a quality benchmark for health care systems. Hospital readmission is a considerable burden on the individual from a cost perspective and the related treatment costs and hospital resources tax the health system. This review is the first to use a systematic method to describe the literature for assessing readmission rate differences in men and women with an index HF hospitalization.
We have observed that there might be an interplay between sex, timing post-discharge, and readmission rates. Studies in which women were more likely to have higher readmission rates than men tended to have durations of follow-up of less than 12 months. This finding suggests that short-term follow-up, improved self-management and early care following discharge from the hospital may be needed. Reducing readmission rates in women may necessitate consideration of sex-specific roles and supports for women who live alone. In contrast, readmission rates appeared to be higher for men in studies with longer durations of follow-up (> 12 months) [28,53].
Few studies with a sex and gender sensitive approach have been conducted prospectively. HF readmission is more typical of patients who are men with reduced ejection fraction than women [59]. Two of the retrospective studies which contained information on LVEF [39,49] indicated that the tendency for readmission was higher in men as compared to women. When the totality of studies was examined, including those with or without LVEF assessment, more studies indicated a higher readmission rate for men overall. Sex and gender differences may relate in part to compliance with pharmacological treatments. Analysis of the EuroHeart Failure Survey indicated that compliance with pharmacological therapies differed between men and women. Men may be less compliant to pharmacological treatments, following discharge after an index diagnosis of heart failure. As future next steps, education with good integrated care and follow up conversations with practitioners warrants investigation leading to improved health outcomes and reduced readmission rates among men. The greater use of invasive diagnostic procedures in men may also relate to the observed early differences in outcomes since these procedures may impact early outcomes reflecting the care received during the acute care hospitalization.
A sex specific approach to post-discharge heart failure care may require better access points of care for certain populations. For example, earlier readmission risk in women may suggest that improved postdischarge transitional care, ongoing physician followup in the near-term, home visitation by nurses, and remote patient monitoring could be particularly useful. In men, later readmissions may indicate the need for long-term follow-up, ensurance of medication compliance, and early treatment of conditions such as ischemic HF, which can lead to late readmissions downstream. Further, future studies should report sex-specific readmission and mortality outcomes separately, since the composite outcome of death or readmission may mask informative underlying patterns. Sex-based approaches to pharmacotherapy also warrant greater exploration. For example, it is known that absorption rates and metabolism of digoxin are different between men and women [14]. Thus, serum digoxin should be administered in lower doses to women to avoid toxicity because of these pharmacokinetic differences [60].

Strengths and limitations
Given the large sample of papers in our database, we focused on sex-sensitive factors which could be part of a global strategy to reduce readmission rates. Our methodology was robust as we included a consultation panel of experts, which is an optional step in the Arksey and O'Malley framework. We also consulted individually with a professional librarian, a cardiologist, and sex and gender experts. We linked the references of relevant studies when a primary research study cited either appendices, supplementary material, or pilot studies or if the same author wrote a dissertation and an abstract in conjunction with a journal article [42,61,62].
There are some limitations to our scoping review. A primary objective needed to state evaluations of sex and gender for a study to be included in this scoping review. Secondary or subgroup analyses may not result in conclusive findings without a power calculation for factors related to sex. Further, without a prespecified sample size that powers these secondary analyses, even if findings are positive, may not be clinically relevant. We also found that sex and gender terminology was often confused which could conflate study findings and may not clearly demonstrate how the interrelated nature of these concepts may impact men and women. The time spent in observation units by hospitals and the use of emergency department might be explored in future studies. This scoping review, while providing a description of the research on readmission rates as it relates to sex and gender to date, is limited in its ability to demonstrate trends over time. Finally, the observations of the potential interaction between sex-specific readmission risk and time horizon can only be considered hypothesis generating in this review, and indeed many studies showed no difference. This interaction between sex and time horizon may need to be confirmed in different jurisdictions, accounting for the competing risks of death, before implementing policies directed toward reducing early vs. later-term readmissions.

Conclusion
Overall, there is an increase in reporting of sex and gender differentiated data; however, we found that in most studies, this was not explicitly stated in the primary objectives. We found more papers reported that men with heart failure had significantly higher readmission rates compared to women with heart failure. The effect of sex on readmission may have been dependent on follow-up duration, with longer follow-up duration favoring higher readmission rates among men. Readmission reduction programs could include targeted educational approaches and conversations with practitioners following discharge about medication compliance and management strategies. In addition, we encourage the use of singular rather than composite measures (i.e. combination of mortality and readmission), the latter having the potential to mask important sex-specific associations. Future studies would be needed to investigate heart failure in relation to readmission rates to examine sex and gender differences and their effects over time.