The impact of behavioural risk factors on communicable diseases: a systematic review of reviews

A systematic search was performed across multiple databases through ProQuest covering the 10-year period 28th October 2010 to 28th October 2020. Preliminary scans of the literature reviews and discussion between members of the research team were used to aid selection of common behavioural risk factors and communicable diseases. Thus, alcohol use, smoking, physical inactivity, obesity, illicit drug use and poor diet were chosen as behavioural risk factors, covering some of the most common behavioural contributors to NCDs [15]. The same process identified: Tuberculosis (TB), acquired immune deficiency syndrome (AIDS), human immunodeficiency virus (HIV), viral hepatitis, COVID-19, severe acute respiratory syndrome (SARS), middle-east respiratory syndrome (MERS), pneumonia, influenza, and meningitis as communicable diseases (including infections and pathogens) feasible for review. These communicable diseases were broadly consistent with some of the most prevalent disease/infection categories reported in the global burden of disease study for HICs (excluding those categories predominantly affecting specific groups (e.g. maternal, neonatal) or where a component may be associated with non-infectious causes (e.g. diarrheal) [9]) and with previous outbreaks or epidemics involving HICs [16]. In addition, it was intended to include diseases that arose from both bacterial and viral pathogens, with a range of transmission types, e.g. airborne, droplet, fomite, blood-borne and contact. Combinations of search terms were developed based on these key risk factors and diseases. Search results were restricted to English language and peer reviewed systematic reviews and meta-analyses. Whilst this strategy restricted literature to that which qualified for inclusion in systematic reviews, it allowed for the inclusion of multiple behavioural risk factors and communicable diseases at the same time. The search was restricted to a 10-year period, allowing coverage of a broad range of behavioural risk factors and communicable diseases, yet limiting the literature to a manageable volume. The full search strategy is available in Supplementary file 1. Searches also included poor housing conditions as a risk factor given the impact of housing conditions on respiratory disease [17], but this study focuses specifically on behavioural risk factors.

To identify eligible studies, the titles and abstracts of studies retrieved were screened by two reviewers, with a sample of 15% screened independently by both reviewers and achieving 98.5% agreement (SH, NJ). Discrepancies were resolved between reviewers. For full text screening, ten reviews (7%) were initially screened by three reviewers (SH, NJ, SW), with results later discussed and discrepancies resolved. Following this, the full text screening was divided across reviewers and any reviews that were not a clear exclude/include (20%) were discussed and agreed between reviewers. This meant that, across all reviews screened by full text, 27% were discussed and agreed by more than one reviewer. Where full texts could not be accessed, authors were contacted to request the full text.

Systematic reviews and meta-analyses of observational studies (including cohort and case-control studies) that examined the association between an identified behavioural risk factor and the contraction or outcomes of an identified communicable disease (including infection or related pathogens) were included in the review. Since there are no single definitions of the selected risk factors across the literature, all reviews that focused on an identified risk factor were included regardless of the definition used in the review (definitions are provided in Tables 1 and 2 and in the results section). An aim of the review was to explore risks associated with illicit drug use in general. However, it was recognised that injection drug use can be a mechanism of transmission for some pathogens relating to included communicable diseases (e.g. HIV, HCV). Studies that focused specifically on injection drug use were therefore included alongside those focusing on drug use more generally. Studies were excluded if they: were not a systematic review or meta-analysis; did not examine the association between an selected behavioural risk factor and the contraction or outcomes of an selected communicable disease; included only selected specialist sub-populations (e.g. sex workers, prisoners), or included sub-populations relating to a risk factor (e.g. people who inject drugs) without a general population comparison; or included behavioural risk factors or communicable diseases not relevant to HICs (see above). No restrictions were made for the age of participants included. A flow chart demonstrating the selection process is presented in Fig. 1.

Data were extracted by three reviewers (SH, NJ, SW) into a standardised, pre-piloted form. Each extraction was duplicated across reviewers and discrepancies resolved through discussion. Information extracted from the studies included: title, authors, abstract, behavioural risk factor(s) studied, communicable disease(s) studied (including infection or related pathogens), research question, geographical restrictions, population characteristics, number of reviews included in the systematic review or meta-analysis, main findings (including odds ratios (OR), relative risks (RR) or rate ratios where available), proposed mechanisms of association and conclusions related to identified behavioural risk factor(s) and communicable disease(s). Where information on the number of reviews included for each risk factor was not reported, the corresponding author of the paper was contacted for additional information.

Due to the variety of different communicable diseases, risk factor definitions, outcome measures, and methods of reporting in the included studies, as well as the challenges of conducting meta-analysis for observational studies [71], findings were not combined statistically through meta-analysis. Instead, a narrative synthesis of the findings was constructed [72], and effect size ranges reported for each behavioural risk factor. Key information that would have enabled calculation of a common effect size was often not available. To calculate these effect size ranges, it was assumed that ORs, RRs and rate ratios were approximately equivalent, a method suggested for umbrella reviews in these circumstances [73]. In addition, where studies reported reduced risk of a communicable disease with a health behaviour (e.g. physical activity, no alcohol drinking), an inverse OR (1/OR) for the corresponding risk behaviour was reported. Findings were structured according to the identified behavioural risk factors and their association with a) contraction of the identified communicable diseases, and b) experiencing more severe outcomes from these communicable diseases. Some study conclusions were amended for readability, to aid understanding. Further, where study conclusions were not relevant to the current research question, information was extracted from results sections and amended for readability (see Supplementary file 2).

The methodological quality of included studies was assessed using the Overview Quality Assessment Questionnaire (OQAQ); a frequently used, validated tool for assessing the methodological quality of systematic reviews [74]. Methodological quality assessment was carried out by three researchers (NJ, SH, SW), with any discrepancies resolved through discussion. Assessment ratings are available in Supplementary file 2.

Overall, 50 out of 53 reviews (94%) concluded that at least one of the behavioural risk factors studied increased the risk of contracting or having more severe outcomes of a communicable disease. Across all reviews, effect sizes, where calculated, ranged from 0.83 to 8.22 (Figs. 2 and 3; Tables 1 and 2). Nineteen out of 21 reviews (90%) concluded that at least one of the behavioural risk factors studied increased the risk of contracting a communicable disease (Table 1). Across all contraction reviews, effect sizes, where calculated, ranged from 1.03 to 8.22 (Fig. 2). Thirty-two out of 35 reviews (91%) concluded that at least one of the behavioural risk factors studied increased the likelihood of having more severe outcomes from a communicable disease (Table 2). Across all outcome reviews, effect sizes, where calculated, ranged from 0.83 to 3.96 (Fig. 3).

Seventeen reviews included alcohol as a risk factor for a communicable disease, with a range of definitions used: any alcohol consumption [19, 23, 41, 43, 44, 46, 47, 49]; any alcohol use or higher amounts [18, 21]; binge drinking or alcohol misuse [20]; alcohol misuse [40]; alcohol use disorder (AUD) [22, 45]; alcoholism [39, 42]; or current/history of excess use [48]. The majority of reviews reported an increased risk of contraction (5/6 reviews; Table 1) and more severe outcomes (10/12 reviews; Table 2). Across all alcohol reviews, effect sizes, where calculated, ranged from 1.83–8.22 for contraction (Fig. 2) and 1.45–2.47 for severe outcomes (Fig. 3). Alcohol use (any use, higher amounts, binge drinking or AUD) was reported to increase the risk of contracting TB [18], HIV [20], pneumonia [21, 22] and invasive pneumococcal diseases (IPD) [23]. One review did not draw a conclusion, but reported mixed findings for the association between alcohol consumption and contraction of TB [19]. Alcohol use (any use, misuse, current/history of excess use, alcoholism or AUD) was reported to increase the risk of having more severe outcomes from TB [39‐44], HIV [45, 46], and HCV [48, 49]. One review reported mixed findings and made no clear conclusion about the association of alcohol consumption and TB outcomes [19], and one review reported variable results among studies examining the association between alcohol consumption and the progression of HIV [47].

Fifteen reviews examined the association between illicit drug use and communicable disease contraction or outcomes, with a wide range of definitions used: drug abuse [19, 41]; illicit drug use [25]; regular/problem cocaine use [50]; recent drug use [51]; and injection drug use [20, 24, 26‐29, 39, 48, 52, 53]. The majority of reviews reported an increased risk of contraction (8/9 reviews; Table 1) and more severe outcomes (6/8 reviews; Table 2). Across all drug use reviews, effect sizes, where calculated (for more severe outcomes only), ranged from 1.01–3.96 (Fig. 3). Both injection drug use and illicit drug use were reported to increase the risk of contracting TB [24] and HIV [20, 24, 25], whilst the prevalence of HCV was found to be higher among people who inject drugs (PWID) compared to general population or community groups [26‐29]. One review did not draw a conclusion but reported mixed findings for the association between drug abuse and TB contraction [19]. Both injecting drug use and drug use/abuse were reported to increase the risk of having more severe outcomes from TB [39, 41], HIV [50] and HCV [48, 51]. Furthermore, among those with HBV, the prevalence of hepatitis D (HDV; co-infection with HDV is considered a more severe form of viral hepatitis) was substantially higher for PWID compared to a mixed population with no risk factors [53]. One review did not draw a conclusion but reported mixed findings for the association between drug abuse and TB outcomes [19], and one review concluded that treatment outcomes for HCV were similar between people who currently were and were not injecting drugs [52].

One systematic review was identified examining the association between physical activity and communicable disease contraction or outcomes. This study reported an association between increased prolonged, moderate aerobic exercise and reduced influenza-related mortality, and improved immunocompetence [38].

Ten reviews focused on the relationship between obesity [30, 31, 54‐58, 60], or overweight and obesity [32, 59], and communicable disease risk. The majority of reviews reported an increased risk of contraction (3/3 reviews; Table 1) and more severe outcomes (8/9 reviews; Table 2). Across all obesity reviews, effect sizes, where calculated, ranged from 1.03–1.33 for contraction (Fig. 2) and 0.83–2.74 for severe outcomes (Fig. 3). Obesity was reported to increase the risk of contracting influenza [30] and pneumonia [31, 32]. Obesity was reported to increase the risk of having more severe outcomes from influenza [30, 54‐56] and COVID-19 [58‐60]. One review concluded that most studies showed some degree of association between higher body mass index (BMI) and a worse clinical presentation of COVID-19 and the need for hospitalisation. This review suggested that obesity seemed to predict poor clinical evolution in patients with COVID-19, but that studies in the review had limited methodological quality [57]. However, one review, which concluded that obesity increased the risk of contracting pneumonia, also found that obese individuals had a lower mortality risk from pneumonia [32].

Eighteen reviews examined the association between smoking (current, past or both) and communicable disease contraction or outcomes. The majority of reviews reported some evidence of an increased risk of contraction (3/3 reviews; Table 1) and more severe outcomes (13/15 reviews; Table 2). Across all smoking reviews, effect sizes, where calculated, ranged from 2.17–2.31 for contraction (Fig. 2) and 0.94–3.46 for severe outcomes (Fig. 3). Smoking was reported to increase the risk of contracting HIV [20], pneumonia [33] and invasive pneumococcal disease (IPD) [23]. Further, smoking was reported to increase the risk of having more severe outcomes from TB [19, 41‐43, 61, 62], influenza [56], and COVID-19 [63‐66, 68]. Two reviews reported no associations between smoking and more severe outcomes from communicable diseases, including death from TB [39] and TB treatment outcomes [40]. One review reported that active smoking may increase the risk of severe COVID-19, but found the result was heavily influenced by one study [67].

Six reviews focused on second-hand smoke as a risk factor for a communicable disease. The majority of reviews reported some evidence of an increased risk of contraction (4/5 reviews; Table 1) and more severe outcomes (1/1 review; Table 2). Across all second-hand smoking reviews, effect sizes, where calculated, ranged from 1.11 to 2.80 for contraction (Fig. 2), and the one effect size calculated for severe outcomes was 1.52 (Fig. 3). One review suggested that second-hand smoke exposure increased the risk of TB infection and disease [34]. The remaining four reviews reported at least some evidence of second-hand smoke exposure increasing the risk of contracting a communicable disease, including TB [35, 36], pneumonia (among those aged 65+ only) [33], and IBD (invasive meningococcal disease [IMD] only) [37]. Second-hand smoke exposure was reported to increase the risk of severe outcomes from acute lower respiratory infections (ALRIs), including pneumonia [69].

Only two reviews were identified that examined the association between poor diet and communicable disease outcomes, and no reviews examining the association between poor diet and communicable disease contraction were identified. One review found that vitamin D status may influence the course of HIV disease [70]. The second review reported that a high intake of polyunsaturated fatty acids was associated with non-response to HCV antiviral therapy [49].